Parrots (Order: Psittaciformes) are a diverse clade that is easily distinguishable from other birds. Despite the clear characters that define Psittaciformes (hooked bills, zygodactylous feet, and plumage that is often predominantly green or red), relative morphological uniformity among parrots has made taxonomic classification a fraught endeavor for over a century. Parrot systematics were propelled forward when DNA sequencing data shed insights into higher- and species-level relationships. However, despite these significant advances, major gaps in taxon sampling and uncertainty in relationships remained due to inferring phylogenetic relationships with short fragments of DNA. Recent work using genome-wide molecular markers with nearly complete parrot species-level sampling has brought clarity to many of the remaining outstanding questions on taxonomic relationships. Here, we build on this work by including four additional species to present a taxonomic revision of Psittaciformes better aligned with its evolutionary tree. We infer maximum likelihood and time-calibrated phylogenies for parrots, present accounts for 106 genera, compare how our findings relate to previous work, and highlight future areas of research. The family-group nomenclature we propose reflects deep evolutionary divergences with diagnosable synapomorphies that are commensurate across comparable ranks in psittaciform clades. We erect three new family-group names at the rank of tribe (Brotogerini Smith, Thom, and Joseph, 2024; Neophemini Schodde, Mason, Smith, Thom, and Joseph, 2024; Bolbopsittacini Smith, Thom, and Joseph, 2024). We elevate one tribe to subfamily rank for the cacatuid genus Probosciger and we restrict usage of the recently introduced tribe Touitini to its type genus Touit. At shallower taxonomic scales, recognition of more rather than fewer genera addresses issues of paraphyly or high discordance in morphological and genomic characters at those levels. We support many reinstatements of older generic names advocated in recent decades, and we further reinstate five valid, available generic names not widely used in recent literature if at all (Licmetis, Gymnopsittacus, Clarkona, Suavipsitta, Cardeos). We advocate the retention of Vini Lesson, 1833, over Coriphilus Wagler, 1832, based on preliminary examination showing substantially more frequent usage of the former. We redraw generic limits in some other cases (e.g., Bolborhynchus parrotlets and allies) and this includes recognizing fewer genera than recently proposed for the Psittacula sensu lato ringneck parakeets. Our revised classification of parrots addresses many longstanding taxonomic questions including those that have arisen through the acquisition of genetic data. It provides context for the temporal origins of psittaciform clades and the taxonomic and phenotypic diversification throughout their evolutionary history. We hope that it will be a benchmark guiding further taxonomic study as well as for downstream analyses in many other fields.
INTRODUCTION
FRONTISPIECE.
Double-eyed Fig Parrots (Cyclopsitta diophthalma) in Cairns, Queensland, Australia. In this work, genera in fig parrots have been revised to better reflect their phylogenetic relationships. Photo by Daniel J. Field.
There has been little doubt about which birds are, and which birds are not, parrots.
—Sibley and Ahlquist (1990)
The overall morphological sameness which makes the order so distinct also makes further subdivision difficult.”
—Smith (1975)
However, despite [their] superficial variation, [parrots] are really a very homogeneous group and […] this presents problems to systematists.
—Forshaw (1989)
These three quotes illustrate the challenge that parrots have posed to systematists since the 1750s. Any bird is easily seen to be a parrot or not based on their hooked bills, which may more accurately be described as “short, blunt bill with a down-curved upper mandible [maxilla] fitting neatly over a broad, upturned lower mandible” (Forshaw and Knight, 2010). Among less readily visible traits, which were reviewed thoroughly by Sibley and Ahlquist (1990), only parrots have feather pigments known as psittacofulvins. Krukenberg (1882) identified that red to yellow colors in parrots are due to these unique biochrome pigments that he named psittacofulvins. This sets parrots apart from other birds, which use diet-derived carotenoids to produce pigment-based red. Moreover, the green plumage of many parrots is also unique, an interaction between feather structure and pigments such as psittacofulvins (reviewed in Berg and Bennett, 2010). Recognition of the nearly 400 species of parrots, which occur on all continents except Antarctica, as comprising the Order Psittaciformes is unquestioned. Determining relationships within and among subordinal groupings of parrots and how those groupings should be ranked taxonomically remains debated and unsettled, however. Several detailed reviews of the history of parrot classification have appeared (Holyoak, 1973; Smith, 1975; Sibley and Ahlquist, 1990; Forshaw and Knight, 2010). These works laid important foundations, but they were based mostly on a key criterion of similarity. Without any kind of cladistic or indeed phylogenetic framework, synapomorphies could not be distinguished from shared ancestral plesiomorphies let alone recognized as such. Accordingly, many outstanding questions of relationships had not been resolved, or indeed could not have been resolved, until well after the advent of molecular phylogenetic techniques. Here we briefly review major trends noting that this history itself mirrors that of avian higher-level systematics generally.
Brief History of Psittaciform Systematics
Initial attempts to discern relationships within and among the parrots themselves and to other birds were rooted in gross external morphology and anatomy (extensive reviews in Smith, 1975; Sibley and Ahlquist, 1972, 1990). Interpretation of this massive body of work was, as just noted, based on similarity and difference. That continued throughout the 20th century (Holyoak, 1973) and even into the 21st (Smith, 1975; Sibley and Ahlquist, 1990, Forshaw and Knight, 2010). In hindsight, we see that the challenge in implementing this approach is the general sparsity of morphological synapomorphies that can be reliably placed in a phylogenetic framework. Consequently, signals of convergent trait evolution were often not disentangled from those informing evolutionary relationships. Two examples follow. One involves the early hypotheses based on bill morphology proposing a close relationship between parrots and diurnal raptors, all of which have similarly “hooked” bills. Study of the evolution of bill morphology in all birds, not just parrots, has been reinvigorated in recent years in the contexts of elaboration, innovation, and micro-evolutionary and megaevolutionary trends across the avian phylogeny generally (Cooney et al., 2017; Bright et al., 2019; Guillerme et al., 2023). Second is where the sparsity of morphological synapomorphies led to the recognition of groups now seen as non-monophyletic (e.g., Loricoloriinae; Mayr, 2008).
Systematic surveys of Psittaciformes have supported contrasting intraorder configurations ranging from 1–8 families and up to nine subfamilies (Salvadori, 1891; Reichenow, 1913; Peters, 1937; Glenny, 1957; Verheyen, 1956; Brereton, 1963). Frequently, the lories and lorikeets (Loriidae), which have a highly derived phenotype, were assigned to their own family, as were the cockatoos (Cacatuidae). The vast diversity of the remaining parrots, while often recognized as distinct subfamilies, were considered by some authors to comprise a single family, Psittacidae. Clarity with respect to higher-level relationships in Psittaciformes was not resolved until the maturation of molecular phylogenetics much later (Joseph et al., 2012). Molecular systematic studies have unambiguously resolved natural groupings, although their taxonomic ranks might be debated, and highlighted the limitations of previously assessed morphological characters for classifying major parrot groups.
In the latter part of the 20th century, other methodologies appeared. Some were very short-lived, such as behavioral studies (e.g., whether head-scratching with the bird's foot was over or under the wing; Brereton and Immelmann, 2008) and taxonomy of ectoparasitic lice living in the feathers of parrots as a potential source of information about relationships among their hosts (summarized in Forshaw, 1973; see also Johnson and Doña, 2024). Early methodologies exploring the genome, whether indirectly through gene products or directly through DNA, have had vast impact.
Indirect, biochemical methods of studying relationships commenced with egg-white protein electrophoresis wherein data were analyzed using qualitative “eyeballing” of similarity and difference on electrophoretic gels within given pH ranges. An example was Sibley and Ahlquist's (1972) survey of nonpasserine birds, which concluded that parrots and pigeons were closely related, an older hypothesis long since rejected. Next was allozyme electrophoresis, which flourished for a few years in the late 1970s and early 1980s, addressing relationships within major clades of birds (Sibley, 1974; Avise and Aquadro, 1982), and within and among clades of parrots too, affirming hypotheses based in morphology, e.g., that Nymphicus is a cockatoo (Adams et al., 1984), or favoring others since rejected, e.g., that Psittacella is part of the platycercine or broad-tailed parrot assemblage (Christidis et al., 1991). These methods were replaced by more direct assays of the genome through banding patterns of chromosomes and karyology and DNA-DNA hybridization, all of which were similarity- or distance-based methods applied broadly in systematics (Sibley and Ahlquist, 1985; Christidis, 1986, 1990; Sibley and Ahlquist, 1990; Kirsch et al., 2008; O'Connor et al., 2024). Eventually, more direct DNA sequence-based methods arose. Restriction-fragment analysis of mitochondrial DNA was first and then direct sequencing of up to several hundred base pairs of mitochondrial and nuclear DNA became available. The latter sometimes even focused on one gene such as spindlin to assess major divisions with parrots (de Koet and de Kloet, 2005) and the first recognition of parrots as sister to passerine birds (Hackett et al., 2008; Suh et al., 2011). These evolved into longer read methodologies and, ultimately, today's various methodologies of modern genomics, which would be used to provide far greater clarity to parrot systematics.
As reviewed in more detail in Provost et al. (2018) and Olah et al. (2021), the proliferation of molecular phylogenetic studies using mitochondrial and nuclear DNA markers over the past two decades resulted in the most comprehensive progress in understanding relationships within the Psittaciformes. Higher-level studies (Tavares et al., 2006; Wright et al., 2008; Schirtzinger et al., 2012; Rheindt et al., 2014), while less numerous, have clarified a number of outstanding questions related to clade membership and the temporal origins of parrots (e.g., Selvatti et al., 2022). Arguably more common, however, have been studies at the generic level that examined generic limits and composition, including Amazona (Russello and Amato, 2004; Eberhard and Bermingham, 2004; Ribas et al., 2007b; Urantówka et al., 2014a), Aratinga (Ribas and Miyaki, 2004), Barnardius (Joseph and Wilke, 2006), Forpus (Smith et al., 2013), Pyrrhura (Ribas et al., 2006), Pionus (Ribas et al., 2007a), Pyrilia (Ribas et al., 2005, Eberhard and Bermingham, 2005), Platycercus (Joseph et al., 2008; Shipham et al., 2015), Psittacella and Pezoporus (Joseph et al., 2011), Prioniturus (Schweizer et al., 2012), and Psittacula (Braun et al., 2019; Groombridge et al., 2004; Kundu et al., 2012; Jackson et al., 2015; Podsiadlowski et al., 2017). Notably, these included resolving the placement of extinct species such as Carolina Parakeet (Kirchman et al., 2012); Cuban Macaw (Johansson et al., 2018); and Mascarene Parakeet (Podsiadlowski et al., 2017), the last-named having been briefly sidetracked by earlier misinterpretation of data (Kundu et al., 2012). Collectively, these studies have yielded a stable backbone of major groups and resolved several contentious groups. Similarly, at shallower taxonomic scales, species-level relationships have been resolved for numerous groups and many of the papers just cited included this level of taxonomic resolution. However, a substantial number of species have remained unsampled, leaving the monophyly of some groups and thus generic limits in question. Further, the limited size and nucleotide sampling of molecular datasets has left some relationships uncertain and unstable (e.g., Androglossini). The most recent studies have taken advantage of DNA sequencing technology that allows for sampling of genome-wide markers (Shipham et al., 2019; Smith et al., 2020; Ewart et al., 2020; Huynh et al., 2023) and so provide increased accuracy in estimating evolutionary parameters and relationships.
Revising Psittaciform Systematics Using Phylogenomics
Our aim here is to review the taxonomy of the entire Order Psittaciformes, especially at the level of genus and above, and so provide a solid benchmark taxonomic foundation for downstream analyses in fields such as ecology and biogeography as well as conservation and management, given how endangered many parrots have become through illegal trade (Pires et al., 2021; Nandika et al., 2021; Tella et al., 2022, and papers therein). We build on the scalability made possible through genomic approaches, which enabled the sampling of nearly all parrot species in a single study (Smith et al., 2023) and allowed sampling of >1 million sites across the genome. We have added four additional species into analyses that were not included in Smith et al. (2023), thus filling nearly all remaining gaps in species-level sampling. The sampling included DNA from historical museum specimens of extinct and endangered taxa and of taxa now difficult to sample afresh.
We generated a time-calibrated phylogeny representing 96% of species-level diversity to inform our discussion on revising the systematics and taxonomy of Psittaciformes. We discuss each clade in the tree, the relationships within it, and how the topology differs from previous studies. The main goal of our taxonomic revision was to align the taxonomy of parrots, again emphasizing the levels of genus and above, with the comprehensive phylogenomic analysis of Smith et al. (2023). Following Joseph et al. (2020), we adopt the rationale articulated by Provost et al. (2018) to justify taxonomic changes that reflect the underlying phylogenetic signal.
Ultimately, a taxonomic classification should reflect phylogenetic pattern without necessarily putting forth an explanation of the evolutionary processes that generated those patterns. This presents a conundrum. Using some data types, such as molecular data from neutral or near-neutral loci, diagnosis of a clade and phenotypic characters may well be concordant. In contrast, nonconcordance of molecular and phenotypic markers can arise because phenotypic characters, especially morphological ones, are subject to natural and sexual selection. These distinctions are best reflected in how each of these types of data are modeled in evolutionary analyses. DNA substitutions, under neutrality, which are reflected in branch lengths, occur at the rate at which mutations occur (Bromham and Penny, 2003). In contrast, morphological evolution in comparative phylogenetic analysis is frequently modeled as a stochastic process (e.g., Brownian motion; Blomberg et al., 2020). This discrepancy can result in rapid and convergent evolution and a decoupling between morphological disparity and time. The extent of phenotypic change, therefore, is not easily predictable.
Consequently, morphological data will often be discordant with respect to phylogenetic patterns defined by neutral and nearly neutral molecular markers. This poses serious challenges for classification at various taxonomic ranks. Synapomorphies corresponding with cladogenetic events may not always be apparent. For example, consider the speciose but phenotypically coherent parrot genus Pyrrhura, comprising ∼23 species, each of which is easily assignable to Pyrrhura. It is a monophyletic assemblage originating some seven million years ago (Mya). Recognition of membership of all its species in Pyrrhura is unquestioned, to our knowledge, owing to the relative homogeneity of its member species in morphology and plumage. In contrast, another Neotropical clade of parrots contains five monotypic and, with respect to each other, phenotypically diverse genera (Diopsittaca, Ognorhynchus, Leptosittaca, Guaruba, and Thectocercus). Collectively, its crown age was six Mya. Rates of evolution appear different in these two clades given the asymmetry in recognized genera. Of course, it is well accepted that morphological disparity varies among clades. Yet the disparity between these two clades highlights that the outcomes of evolution have been very different for these groups, which, in turn, impacts the comparability of taxa at the same rank. In this paper we have encountered this problem frequently.
Where external synapomorphies have not been identified, we have pursued an approach of recognizing more rather than fewer genera and tribes. The reason is that they can be anchored in and thus reflect our phylogenomic framework, which we stress includes temporal depths of nodes. Equally, though, our decisions have been informed by morphological, ecological, and behavioral dissimilarity among those groups in light of phylogenetic relationships.
We make two further points about how we have taxonomically interpreted relationships in our tree: First is that support values are often high in phylogenomic studies because there are so many characters (Minh et al., 2020a). These high support values can indicate that there is strong genome-wide support for a relationship or that a small number of genes or sites are driving a given relationship. In the latter case, the robustness of a relationship determined by nodal support may be overturned with data from different genomic regions. In our descriptions of phylogenetic relationships there is reference to ultrafast bootstraps (Minh et al., 2013; Hoang et al., 2018), which were performed to account for the large size of the phylogenomic dataset. Importantly, ultrafast bootstraps (UFBS) have a more stringent level for accepting a node as supported (≥95%), as do local posterior probabilities (LPP) for the species trees we discuss in the text.
Second is the reliance on historical DNA from museum specimens of species for which there are no modern genetic samples. Historical DNA tends to be heavily degraded and yields more missing data than modern samples. This poses challenges in accurately estimating branch lengths and relationships (Smith et al., 2020). We do not repeat the analyses of Smith et al. (2023), whose rigorous interrogation of data signals in the parrot phylogenomic alignment was a primary focus. Instead, we highlight areas where uncertainty may linger due to data quality. Data completeness is particularly relevant in interpreting topological differences between the concatenated and species tree phylogenies. Although concatenated trees can produce erroneous relationships by not accounting for gene tree discordance, summary species tree methods like the one included by Smith et al. (2023) are particularly sensitive to gene tree estimation error due to samples with lower phylogenetic signal. For simplicity, we present the findings of only the concatenated phylogenomic tree but point out major differences against the species tree when relevant. When the text refers to a species tree, we are referring to the species tree estimated using ASTRAL-III (Zhang et al., 2018) presented in Smith et al. (2023), which summarizes gene trees assuming a coalescent framework. Collectively, through these ways of interpreting phylogenomic signal and conflict, we have provided a detailed overview of the phylogenetic relationships and taxonomy of Psittaciformes.
Genus- and Family-Level Taxonomy
We agree that genera “should ideally make sense to someone not familiar with the constituent species” (Ford, 2006). Conversely, genera should represent monophyletic groups, yet evolution does not always result in phenotypically coherent groups that are monophyletic and vice versa. Aside from understanding the direction and polarity of character evolution, the task of delimiting genera can become one of deciding where to draw the line or how much phenotypic diversity a genus should or should not include while still consistent with monophyly. Countering the difficulties of delimiting genera are the multiple hierarchical levels that the Linnaean system offers in family-group names. Together, genera and the levels of family-group taxonomic ranks offer mechanisms for taxonomically translating much of the evolutionary complexity that a group as diverse as Psittaciformes shows geographically, evolutionarily, genetically, and even phenotypically. Reflecting these complexities, phylogenetic and phylogenomic analyses of birds have consistently driven the recognition of more rather than fewer genera in birds. This can be seen not only in parrots (reviewed in Provost et al., 2018) but in other groups as well, such as the radiations of Neotropical thamnophilid ant-birds (Isler et al., 2006, 2013) and thraupid tanagers and tanager-finches (Burns et al., 2016), the Indo-West Pacific columbid fruit-pigeons (Cibois et al., 2014) and meliphagid honeyeaters (Joseph et al., 2014; McCullough et al., 2019), and African hirundinids (de Silva et al., 2018), Afro-Asian timaliid babblers (Moyle et al., 2012) and phalacrocoracid microcormorants (Kennedy et al., 2023). To the criticism that this trend generates a high number of unfamiliar genera, we reply that the number of genera per se or indeed their familiarity have little or nothing to do either with the science underpinning a classification or the merit of a classification in reflecting that science. Clearly, such critiques are of no concern to other branches of zoology, such as entomology, that also deals with highly diverse and speciose groups. Nonetheless, a way of reducing genera while recognizing phylogenetic structure is to employ subgenera within genera. We have debated the use of subgenera extensively in preparing this paper. Earlier (Provost et al., 2018), we with a colleague refrained from subdividing some genera as a concession to maintaining familiarity (e.g., retaining four species in Psephotellus) or minimizing change to others (changing Pseudeos from monotypic to ditypic not to two monotypic genera). Here we have reconsidered many such decisions based on data now available. We have considered temporal and geographical criteria with which to recognize genera especially when the estimated divergence times are greater than ∼2 million years. We now see more merit in recognizing genera rather than subgenera to reflect phylogenetic structure. The reason is that in ornithology subgenera are rarely used and where they have been used (e.g., Schodde and Mason, 1999; Shipham et al., 2015), we are aware of little or no uptake.
Species-level Taxonomy
Notably, recent years have seen many subspecies previously known from little more than a few specimens elevated to species rank. This has followed a surge in fieldwork and increased familiarity with their natural populations. It has also arisen from application of criteria such as those of Tobias et al. (2010), whose method of species delimitation has been widely criticized (e.g., Remsen, 2016; Rheindt and Ng, 2021). While subspecies suffer from empirical and philosophical limitations (Burbrink et al., 2022), our study was designed to sample recognized species with a preference for recently collected material from wild birds. Our rationale was to optimize data quality and thereby increase the resolution of phylogenetic relationships. As stated earlier, DNA sourced from historical museum specimens is often highly fragmented and more likely to be subject to gene tree estimation error. A potential consequence of our sampling design was that we did not preferentially sample the nominotypical subspecies in polytypic species. In cases where species are later found not to be monophyletic, the inferred relationships in our phylogeny may be incorrect for some species complexes (see Forpus as an example; Smith et al., 2013). To standardize our selection of species, which began in 2015, we used the International Ornithological Congress (IOC) World Bird List (e.g., Gill et al., 2024), which provided classification for all parrots globally. During data generation of this project, a number of new species resulting from split complexes were recognized by the IOC, and we did not make an attempt to include all these taxa. We further acknowledge that for some of these new species, there is considerable disagreement about their taxonomic rank among classification committees. Our taxon sampling, which is still far from complete for many polytypic species, does not inform some current species-level taxonomic proposals in the literature. We take the opportunity, however, to discuss recent decisions and pinpoint further work needed.
Where possible, we have cited names of genera (whether in full or by initial letter abbreviation) and species, e.g., B. barnardi as shorthand for Barnardius barnardi. When the topic under discussion is the appropriate taxonomic rank for a species-group epithet, i.e., species or subspecies, or to which genus such an epithet should be assigned, we have used only the species-group epithet.
MATERIALS AND METHODS
Sampling, Wet-Lab Protocols, and Bioinformatic Processing
We used genome-wide markers known as ultra-conserved elements (UCEs; Faircloth et al. 2012) from published studies and generated sequence data from four additional taxa to fill sampling gaps. Our final data matrix included 323 species across Psittaciformes from Smith et al. (2023), 58 species within the Loriini (Smith et al., 2020), Conuropsis carolinensis (Smith et al., 2021), sequence data from a published genome of Amazona vittata (Oleksyk et al., 2012) and three outgroup taxa (Caracara cheriway, Calyptomena viridis, Icterus cucullatus; Oliveros et al., 2019). The unsampled species include both extinct (Nestor productus, Cyanoramphus ulietanus, C. zealandicus, Psittacula wardi, P. exsul, Mascarinus mascarin, Vini diadema, and Ara tricolor) and extant (Tanygnathus gramineus, Loriculus camiguinensis, L. flosculus, Pyrrhura pfrimeri, and Poicephalus crassus) species. Of the four new samples, three came from historical museum specimens (Psittacara wagleri AMNH 111438; Pionus menstruus AMNH 170946; and Psephotellus chrysopterygius AMNH 623349) and one from a modern tissue sample (Aratinga maculata YPM 137433). Other extinct species are represented only in literature e.g., extinct Indian Ocean island parrots (Hume, 2007).
We followed the same wet-lab protocol as Smith et al. (2023). Extraction of DNA from modern tissue samples used the DNeasy tissue extraction kit (Qiagen, Valencia, CA). DNA from historical samples that came from the toe pads of museum specimens were extracted using a modified protocol because the DNA in these samples is highly fragmented. To compensate for this level of DNA degradation, we replaced the DNeasy filter columns with ones from the QIA-quick PCR kit to size-select for smaller DNA fragments. DNA extraction from historical samples was done in a dedicated lab for working on degraded samples to reduce contamination risk. We quantified DNA extracts using a Qubit 2.0 Fluorometer (Thermo Fisher Scientific). Library preparation of UCEs, enrichment, and Illumina sequencing were performed by RAPiD Genomics (Gainesville, FL). The Tetrapod UCE 5K probe set was used to enrich 5060 UCE loci (Faircloth et al., 2012). Sequencing was done on an HiSeq X Ten. FASTQ files are available on the Short Read Archive (SRA numbers are available in supplementary table S1; available on Dryad at doi.org/10.5061/dryad.h70rxwdrx).
We followed the same variant calling pipeline as in Smith et al. (2023) that is briefly described here. Low-quality bases and adaptor sequences were trimmed from demultiplexed FASTQ files in Illumiprocessor v1 (Faircloth, 2013; Bolger et al., 2014). Contigs were mapped to UCE probes, and an index for a genus-specific reference sequence was generated, and reads were mapped to references using BWA v0.7.13-r1126 (Li and Durbin, 2009). SAM files were converted to BAM files and sorted with SAMtools v. 1.10 (Li et al., 2009). A VCF file was produced using the mpileup function in SAMtools v. 1.10 (-C 30; -Q 20). bcftools v. 1.12 (Danecek et al., 2021) and vcfutils were used to call and filter variant sites with <5× coverage per SNP and a quality score <20. seqtk was used to convert FASTQ files to FASTA. After retaining loci that were a minimum of 100 base pairs (bp) with heterozygous sites converted to IUPAC ambiguity codes, MAFFT v. 7.455 (Katoh and Standley, 2013) was used to align loci. The concatenated alignment and the final UCE alignments for which 75% of the samples were present in each locus were produced using default settings in PHYLUCE (Faircloth, 2016).
Phylogenetic Inference and Molecular Dating
We followed the methodology of Smith et al. (2023) to infer a concatenated topology and to time calibrate the phylogeny, which is repeated here. We used IQ-TREE2 v. 2.1.3 (Minh et al., 2020b) to produce a concatenated maximum likelihood phylogeny and 1000 ultrafast bootstraps (Hoang et al., 2018). The best-fit substitution model for each gene partition was estimated on ModelFinder (Chernomor et al., 2016; Kalyaanamoorthy et al., 2017). Bootstrap trees were rooted on the outgroup, Caracara cheriway, using Phyx (Brown et al., 2017). We conducted an additional analysis on just Cacatua using the mixture across sites and trees (MAST) model in IQTREE v.2.2.2.6 (Minh et al., 2020b; Wong et al., 2024), to calculate the tree weights for the concatenated and species tree topologies using an alignment with stringent filtering where species with >1.31 deviation in parsimony informative sites were excluded (see Smith et al., 2023).
Next, we estimated a time-calibrated tree for the phylogenomic tree of Psittaciformes using a penalized likelihood method to estimate divergence dates to accommodate the large number of species and characters in our dataset. We dated the rooted bootstrap trees using treePL (Smith and O'Meara, 2012) from 100 ultrafast bootstrap trees, and calibrated nodes with the fossil ages of four taxa and specifying the root age using an external calibration from Jarvis et al. (2014) for the split between Falconiformes and Psittacopasserae (minimum age 57 Mya, maximum age 62 Mya). Kimball et al. (2019) and Oliveros et al. (2019) provide justifications for the fossil calibrations, except Cacatua, which was not used in those studies and whose age was reported in Boles (1993). The four taxa used to calibrate nodes are: (1) Eozygodactylus americanus (Weidig, 2010) [minimum age 51.81 Mya, maximum age 66.5 Mya for calibrating the split within Psittacopasserae (Psittaciformes and Passeriformes)]; (2) Nelepsittacus minimus (Worthy et al., 2011) [minimum age 15.9 Mya, maximum age 66.5 Mya for the split between Strigops and Nestor]; (3) an unidentified member of Cacatuinae (Boles, 1993) [minimum age 11.608 Mya, maximum age 23.03 Mya for the split between Eolophus and Callocephalon]; and (4) Suboscines indet. (Mayr and Manegold, 2006) [minimum age 27.25 Mya, maximum age 56.0 Mya for the split between oscine and suboscine passerines]. We estimated the optimal parameter settings using the prime and thorough options and randomly sampled during the cross-validation over a range of smoothing parameters (1×10-7–1×104) for 10 iterations over the 100 ultrafast bootstrapped trees. We summarized the 100 treePL trees in TreeAnnotator v.1.10.4 (Suchard et al., 2018) by setting the burnin to 0 and specifying the target tree type as Maximum Clade Credibility and node heights as Common ancestor height.
The taxon-sampling table, DNA sequence alignment, and time-calibrated phylogeny are available on Dryad (doi: 10.5061/dryad.h70rxwdrx).
RESULTS AND DISCUSSION
We inferred a time-calibrated phylogeny from a maximum likelihood tree estimated from a 1,142,292 bp alignment, constituting 3040 ultra-conserved element loci. Support values are reported from the maximum likelihood tree. We present and discuss our results in the order provided by the frameworks of figure 1 and table 1.
STRIGOPOIDEA
Strigopidae (Strigops)
Nestoridae (Nestor)
These two genera are of large to massive parrots of New Zealand. Strigops is nocturnal and flightless and extraordinarily convergent with Australia's Pezoporus parrots in its green plumage mottled black and yellow. Species of Nestor are long-billed, predominantly brownish or bronze-green parrots with red underwings or reddish underparts.
Aside from the extinct Miocene Nelepsittacus (see Worthy et al., 2011), the low-diversity superfamily Strigopoidea consists of three extant and two recently extinct, large-bodied parrots that are endemic to New Zealand and surrounding islands. It was sister to the rest of the entire parrot radiation. One species, Strigops habroptilus (Family: Strigopidae; see Savage and Digby, 2023, for orthography of the epithet habroptilus), is monotypic and flightless, and two extant species, Nestor notabilis and N. meridionalis (Family: Nestoridae), are diurnal and volant. The three species' positions are stable and well resolved in our phylogenomic tree (fig. 1). This is in agreement with previous work. The ages of Strigopoidea and the families Strigopidae and Nestoridae have varied across studies, but a consensus has emerged in more recent work. Using a calibration of the splitting of New Zealand from Gondwana to date the split between Strigopoidea and all other parrots, Wright et al. (2008) dated the crown age of Strigopoidea to 81.91 Mya and 49.84 Mya for the split of Strigopidae and Nestoridae. We stress, however, that later work (e.g., reviewed in Worthy et al., 2017) has cautioned use of the 82 Mya split of New Zealand. The opportunity probably existed until about 55 Mya for taxa to join the Zealandian terrestrial biota by overland dispersal. Subsequent studies using fossil calibrations beyond Psittaciformes or secondary calibrations have found progressively younger ages for dates within the clade. Schweizer et al. (2011) estimated that Strigopoidea, which was represented in their taxon sampling by Nestor, diverged from other parrots 58.6 Mya (95% highest posterior density [HPD]): 44.9–72). Rheindt et al. (2014) estimated that this same divergence time had a mean of 49 or 42 Ma, depending on the type of calibration. Within the Strigopoidea, the divergence of Strigopidae and Nestoridae was dated to 28–29 Mya (95% HPD: 18–38 Mya) by Rheindt et al. (2014). We found that the stem and crown splits for Strigopoidea occurred at 40 and 35 Mya (figs. 1, 2), dates that are similar to the conclusions of another phylogenomic study using a reduced number of taxa (Huang et al., 2022).
FIGURE 1.
Generic-level phylogeny of Psittaciformes. Left to right: tribes, subfamilies, families. * = unsupported nodes; all other nodes have ultrafast bootstrap values ≥ 95%. Main text discusses nomenclature of family-group names Microglossinae, Touitini, Amoropsittacini, and Androglossini. Brotogerini tr. nov., Neophemini tr. nov., and Bolbopsittacini tr. nov. are described herein.
Divergences within Nestor range from the late Miocene to early Pleistocene. We estimated the divergence of N. notabilis and N. meridionalis at 2.6 Mya (fig. 2), similar to the age determined by Rheindt et al. (2014), whereas Wood et al. (2014) dated the split to ∼5 Mya. An extinct species, Nestor productus from Norfolk Island, is known from only 16 museum specimens and went extinct in the 1840s (Holdaway et al., 2001). It has yet to be placed in a molecular phylogeny. Also extinct, the Chatham Island Kaka (Nestor chathamensis) was found to be sister to N. meridionalis and molecular dating based on 571 bp of mtDNA estimated the split between the two taxa at 1.74 Mya (Wood et al., 2014).
Although not directly related to systematics, recent efforts to produce genomic resources for the three species in Strigopoidea have served both applied and basic science purposes. For example, extensive species management of the critically endangered Strigops habroptilus has led to whole genome sequencing of all individuals in the population (N = 169), an unprecedented resource for mitigating genetic associations with disease risk and low reproductive output (Guhlin et al., 2023). Genomic data for Nestor has indicated that N. notabilis, which occurs in the alpine zone of New Zealand, exhibits the same adaptive signatures of high-elevation habitation as the forest-adapted N. meridionalis (Martini et al., 2021).
CACATUOIDEA
CACATUIDAE
Cacatuoidea has its highest diversity in Australia, but it also occurs in New Guinea, Indonesia, Solomon Islands, and the Philippines. The phylogenomic time-calibrated tree shows Cacatuoidea splitting from Psittacoidea 35.9 Mya and that the clade has a crown age of 26.4 Mya (fig. 2). Our phylogeny further supports four subfamilies, in addition to the three now conventional: (1) Calyptorhynchinae—the large black-cockatoos with distinctively colored tail panels; (2) Nymphicinae—the small-bodied, monotypic Nymphicus hollandicus; (3) Microglossinae (sometimes Proboscigerinae (see below)—the massive, large-bodied slaty black monotypic Probosciger; and (4) Cacatuinae—mid- to large-sized cockatoos that are primarily white, gray, or pink. In advocating placement of Probosciger in its own subfamily (or, alternatively, as a monotypic tribe within Cacatuinae), we acknowledge current literature (Gregory and Sangster, 2023) on whether Microgloss- or Probosciger- is the correct stem for the family-group name to which this genus should be assigned. A counter argument to Gregory and Sangster (2023) was made by Schodde (accepted) to retain Microglossinae. The issue turns on debate over a small number of usages of these names by an even smaller number of authors. Until this is settled, we use Microglossinae. Among the main complications in comparing our temporal estimates of divergence in Cacatuidae with previous ages was the placement of Nymphicinae in the phylogenomic tree. We found Nymphicus to be sister to Cacatuinae whereas previous phylogenies placed it as sister to the clade comprising Calyptorhynchinae and Cacatuinae (Wright et al., 2008; White et al., 2011).
TABLE 1
Taxonomic and nomenclatural summary of the classification of Psittaciformes, parrots, recommended in the present study for family-group and genus-group names.
See the text for discussion in current literature about use of the following names and alternatives: Microglossinae, Forpini, Amoropsittacini, Androglossini.
continued
FIGURE 2.
Species-level topology of Strigopoidea and Cacatuoidea. Support values come from the maximum likelihood tree. Nodes have ultrafast bootstrap values of ≥95% otherwise noted.
Calyptorhynchinae
Calyptorhynchus and Zanda
Calyptorhynchus and Zanda are large predominantly black cockatoos with distinctively colored, broad subterminal tail panels, predominantly red or orange-yellow in Calyptorhynchus and yellow to creamy white in Zanda. Bill color is often sexually dimorphic (less prominently so in C. lathami) and Zanda also have sexually dimorphic ear-covert patches broadly concolorous with the subterminal tail band. Vocalizations are distinctive, typically being an oddly “funereal” cry in Zanda and higher pitched somewhat wheezy cries in Calyptorhynchus.
In the Calyptorhynchinae, relationships among the five conventionally recognized species are stable and well supported across phylogenomic trees and a previously inferred species tree (White et al., 2011). We support the generic placement of the species funerea, baudinii, and latirostris in Zanda Mathews, 1913. Zanda was in synonymy with Calyptorhynchus for most of the 20th century, but its reinstatement began with Schodde's (1997) recognition of it at subgeneric rank within Calyptorhynchus. Following White et al.'s (2011) molecular study, its elevation to genus level has become conventional (Gill et al., 2023; Dickinson and Remsen, 2013). Saunders and Pickup (2023a) have revisited and reinforced support for the recognition of Zanda based on morphological and vocalization data. Divergence between Zanda and Calyptorhynchus dates to 19.2 Mya and divergences within each genus date to 0.6–2 Mya and 8.4 Mya, respectively (fig. 2).
Turning to species-level systematics, we note that an analysis of genome-wide single nucleotide polymorphisms (SNPs) (Ewart et al., 2020) found that the Red-tailed Black Cockatoo (Calyptorhynchus banksii) comprises five evolutionarily significant units, the corresponding species-group epithets of which are banskii, graptogyne, naso, escondidus, and samueli. Unsurprisingly, some uncertainties remain in relationships among these five taxa, given their shallow divergence times. Ewart et al. (2020), however, made two taxonomic recommendations: (1) synonymy of banksii and macrorhynchus in nominotypical banksii and (2) that samueli, the three isolated populations of which did not form a single monophyletic group, should be broken into eastern and central populations, samueli, which did form a single monophyletic group, and western population, escondidus, which was sister to naso and which they newly described at subspecies rank.
Saunders and Pickup (2023a) elevated the five subspecies to species rank. Key elements of their decision were morphometric analyses, the time of divergence (latter half of the Pleistocene; Ewart et al., 2020), and their allopatry. Present-day geographic interactions and gene flow between banksii and samueli in eastern Australia seem highly probable, however, and warrant further study regardless of their taxonomic status (see maps in Saunders and Pickup, 2023a). Similarly, but perhaps with a lower probability, escondidus and banksii may be in geographic and genetic contact in northwestern Australia. One species with five subspecies or five monotypic species are among equally plausible arrangements in light of the data available: five geographical units appear monophyletic, but their divergences are shallow relative to many other species and there is no extensive sympatry to affirm species status by that criterion.
The shallow divergence between Z. latirostris and Z. baudinii warrants discussion. This pair is among the most notable examples in ornithology of cryptically differentiated sibling species (White et al., 2014). Saunders (1974, 1979) elegantly showed on morphological grounds that the two taxa are distinct, primarily in bill morphology. This work led to the removal of latirostris from synonymy of baudinii and recognition of the two taxa at species rank. Next, White et al. (2011) presented a multilocus phylogenetic analysis arguing that radiation within Zanda into three species happened relatively recently, within the last 1.3 million years. White et al. (2014) pursued genetic differentiation between Z. latirostris and Z. baudinii with microsatellite data. They suggested that the differentiation between the two taxa started during anthropogenic land clearing in the 20th century. They also remarked that some gene flow is still occurring between the two but estimated that fewer than 10 individuals per generation are moving between regions separated by an anthropogenically generated barrier of unsuitable habitat.
Saunders and Pickup (2023a) presented an updated synthesis of all data concerning this pair. They disputed White et al.'s (2014) claim of differentiation within the past ∼100 years, a finding that was corroborated with our phylogenomic data that dates the split at 0.6 Mya (0.2–1) (fig. 2). Further, they found no morphological indication of hybridization and stressed the overlapping breeding ranges of the two taxa, surely here a gold standard for species-level recognition. On adding our data to these syntheses, we conclude that while population-level sampling of Z. latirostris and Z. baudinii with genomic methods now available is likely to be highly rewarding, the optimal taxonomic conclusion at this point is that they indeed represent two species.
Taxonomic subdivision of Z. funerea, in contrast, has become far more contentious since Saunders and Pickup (2023a, 2003b) proposed that the three taxa at best tentatively recognized as subspecies within it by earlier treatments (Schodde and Mason, 1997; Forshaw and Cooper, 2002) should be elevated to species rank. Our sampling has not addressed variation within Z. funerea. We are highly skeptical of Saunders and Pickup's (2023a, 2023b) proposal to elevate three questionably recognizable subspecies within Z. funerea to species rank but will address this in more detail elsewhere.
Nymphicinae
Nymphicus
Nymphicus is predominantly a small midsized gray parrot with yellow and red facial markings, white wing patches conspicuous in flight and a forward-curving crest. It is one of the world's most popular aviary birds.
Although monotypic Nymphicus has unquestionably been recognized for decades as within the Cacatuoidea and essentially a diminutive cockatoo (see Adams et al., 1984, and Schodde and Mason, 1997, for brief reviews of morphological, karyologi-cal, behavioral, and allozyme data), its precise relationships have remained unclear. Adams et al. (1984) in an allozyme study found its position ambiguous depending on the analysis, either aligning it with the Calyptorhynchus black cockatoos or the Cacatua sensu lato white cockatoos. Brown and Toft's (1999) results based on mitochondrially encoded 12S sequence data and reanalysis of allozyme data were similarly ambiguous. A consensus of their results was that it aligned with Calyptorhynchus as one of the “more basal cockatoo species.” A multilocus analysis led White et al. (2011) to place it as the sister group to the rest of the Cacatuoidea.
Our phylogenomic data strongly place Nymphicus as the sister to all non-calyptorhynchine cockatoos, and not as the sister to all cockatoos or indeed to the calyptorhynchines, diverging 24.8 Mya (16.2–30.4; fig. 2). We argue that all lines of data fully support retention of Nymphicus in a monogeneric subfamily, Nymphicinae, which was sister to the Cacatuinae. Similarly, it is usually treated as a monotypic species given its almost continentwide range and high vagility.
Microglossinae
Probosciger
Probosciger is a massive, large-bodied slaty black cockatoo, which is phylogenetically as well as phenotypically distinctive in its permanently open bill, rose-red colored facial skin, colored tongue, slaty black plumage with some fine orange-red ventral barring in young birds and finely plumed erectile but uncolored crest, whistled vocalizations, and drumming behavior in males (Schodde and Mason, 1997; Heinsohn et al., 2017, 2023).
We have discussed above why we favor placing Probosciger in its own subfamily rather than tribe and we acknowledge ongoing debate about the correct stem for the family-group name to be assigned to this subfamily.
Within Probosciger aterrimus, 3.3% divergence in the control region of the mitochondrial genome corresponds to a geographical split between populations on the Bird's Head Peninsula at the western end of New Guinea and those in the rest of New Guinea and Australia (Murphy et al., 2007). A subsequent study using SNP data and the mtDNA gene ND2 found evidence of shallow genetic structuring between Australia and New Guinea (Keighley et al., 2019).
Cacatuinae
Callocephalon, Eolophus, Lophochroa, Cacatua, and Licmetis
Callocephalon is a monotypic genus of medium-sized cockatoo, predominantly gray in plumage and having a distinctive, forward-curving crest of filamentous feathers. It is sexually dimorphic in its head plumage, red in males, gray in females. It is endemic to wetter forests and woodlands in southeastern Australia. Eolophus is also a monotypic, medium-sized cockatoo with three recognized subspecies. It is distinctively gray dorsally and pink ventrally with a backward-curving pink to whitish crest having a “split” appearance in some populations. It has bare, carunculated periophthalmic skin varying subspecifically from white to shades of pink. It is one of the most familiar of all Australian birds occurring across the continent in all but the wettest forests and woodlands. Lophochroa comprises one species with two currently recognized subspecies. Resembling in overall morphology the large Cacatua galerita, it differs in having salmon-pink underparts and a similarly pink and yellow crest, and a pale bill. It primarily inhabits arid- and semiarid woodlands. All three of these genera have distinctive vocalizations.
We discuss the taxonomic implications of the phylogenomic data in light of the two phenotypically distinct groups within Cacatua sensu lato that the data in turn mostly affirm. The first of these is Cacatua sensu stricto having mostly forward-curving, yellow or white to orange crests, broad wings, naked white to bluish ocular skin, usually concentric with the eye itself, black bills, and characteristically as well as extraordinarily harsh shrieks. The second is Licmetis, the corellas, having vestigial, backward-curving crests, slender wings, milky to leaden blue ocular skin extending below the eye in an oval shape, white bills and a distinctive yodel or crying vocalizations (see Forshaw, 1973; Schodde et al., 1979; Schodde and Mason, 1997; Forshaw and Knight, 2010).
Relationships within Cacatuinae are the most unstable and taxonomically challenging within Cacatuoidea. This is largely due to lower-quality samples for some taxa (e.g., Cacatua ophthalmica) and gene tree–species tree discordance reported in Smith et al. (2023). Phylogenetic patterns within Cacatuinae are represented by a series of well-supported and consecutive nodes, then nodes leading to each of the three monotypic genera Callocephalon, Eolophus, and Lophochroa, and a final node subtending a clade comprising white cockatoos long treated within the genus Cacatua (fig. 2). Previous multilocus work found weak evidence of Callocephalon fimbriatum and Eolophus roseicapilla as sister taxa (White et al., 2011), but we found them diverging on successive branches with 100% bootstrap support. They are each phenotypically unique, and this includes their vocalizations as well as plumage patterns. Generic-level divergences span 9.3–22.6 Mya (fig. 2). We first examine genus-level divergences before discussing species-level issues.
We fully support the retention of Callocephalon, Eolophus, and Lophochroa as monotypic genera. Notably, Lophochroa was sister to a clade comprising Cacatua sensu stricto and the Licmetis corellas, not just the former with which it has long been placed in Cacatua and with some species of which it shares the phenotypic trait of a forward-curving crest. We favor the retention of Lophochroa at generic rank because of its phylogenomic separation from Cacatua and this includes the age of its lineage coupled with its utterly distinct phenotypic traits including bicolored crest and unique vocalizations.
There was topological discordance between the concatenated and species tree topologies for Cacatua. The bulk of the discordance was within the two main clades corresponding to Cacatua sensu stricto and Licmetis, and often involved lower-quality historical samples and nodes with low support, but there are also likely cases of phylogenetic conflict. We find it puzzling that these two groups have for so long been unquestioningly placed in one genus, Cacatua Vieillot, 1817. This treatment has prevailed in Australian literature at least since RAOU (1926); it is difficult to determine why it was introduced and why it has prevailed despite the obvious phenotypic differences just outlined. Further, it has prevailed despite several demonstrations of substantial genetic divergences (Adams et al., 1984; Brown and Toft, 1999). We now address in more detail the merit of breaking of Cacatua sensu lato into these two genera. For ease of discussion, we anticipate our recognition of Licmetis at rank of genus.
One source of uncertainty in our analysis is the generic placement of haematuropygia. It is either sister to the remaining Licmetis (concatenated tree) or the clade composed of Cacatua/ Licmetis (species tree). Morphologically, haematuropygia would appear to belong in Licmetis in that it is similar in size, shape, and bill color to L. goffiniana, a taxon firmly nested within the corellas (but we note its white ocular skin concentric with the eye). From our additional MAST analysis on Cacatua we found that the concatenated phylogeny, which has a monophyletic Licmetis, was the predominant topology (tree weight: 0.87) across the alignment versus the topology with L. haematuropygia as sister to all other Licmetis and Cacatua as the minor tree (tree weight: 0.13). These results confirm the analyses of this clade by Smith et al. (2023), where there was more signal for the concatenated topology. Given haematuropygia's morphological affinities with Licmetis and the overriding phylogenomic signature that indicates it is sister to the other corellas, the best evidence supports the recognition of Cacatua and Licmetis. We do not advocate a new genus for L. haematuropygia.
There were also several cases of gene tree–species tree discordance within both groups. Relationships among the corellas (L. sanguinea, L. tenuirostris, L. pastinator, and L. goffiniana) varied among the phylogenomic trees. Similarly, there was such discordance in Cacatua sensu stricto (C. sulphurea, C. moluccensis, and C. ophthalmica). Divergence times within Cacatua and Licmetis ranged from 2.3 to 5.8 Mya and 3.4 to 8.2, respectively (fig. 2).
We are confident that the combined weight of our genomic data, uncertainties within the two strongly supported clades notwithstanding, and the phenotypic differentiation in body and wing morphologies generally, crest morphology in particular and vocalizations, all combine to provide strong support for the breakup of Cacatua sensu lato and the reinstatement of Licmetis Wagler, 1832, for the corellas. Indeed, we argue that this situation strongly parallels the basis for recognition of Calyptorhynchus and Zanda, the separation of which is now accepted and entrenched such that treating them as congeneric is considered little more than bad taxonomic habit. A similarly prevalent generic separation of sister clades is that of Platycercus and Barnardius. We acknowledge the call made by Adams et al. (1984) for detailed comparative study of any social function of crests and bare facial skin in these cockatoos to test whether contrasting combinations of these traits may have arisen through character displacement.
Phylogeographic work in Cacatuinae has been limited to a few species. Engelhard et al. (2015) found that phylogeographic structure in Eolophus roseicapilla broadly corresponds with eastern, western, and northern subspecies recognized by morphological traits. Future work should examine phylogeographic variation in Callocephalon, Cacatua, and Licmetis. In the widely distributed Australian endemic Lophochroa leadbeateri, SNP data identified contains 2–3 genetic groups corresponding with a break across the Eyrean Barrier and in central Queensland (Ewart et al., 2021). This break corresponds to the subspecies L. l. leadbeateri and L. l. mollis, but interestingly, the genetic structuring among these groups was not observed in mtDNA. Cacatua galerita and the Licmetis sanguinea complex are both of particular interest given their uncertain subspecific taxonomy and their widespread distributions.
PSITTACOIDEA
The majority of parrot diversity is within the superfamily Psittacoidea. It is distributed on multiple continents, and the majority of species are either in the family Psittaculidae or the subfamily Arinae. Higher-level relationships are generally stable across phylogenomic trees and previous phylogenies with a few exceptions discussed below. Psittacidae (Psittacinae: African parrots; Arinae: Neotropical parrots) was sister to Psittrichasidae/Psittaculidae. Psittrichasidae (Coracopseinae) of Madagascar and the Seychelles-Comoros archipelago in the southwestern Indian Ocean; Psittrichasinae of New Guinea) was sister to Psittaculidae. The families in Psittacoidea (Psittacidae, Psittrichasidae, and Psittaculidae) have estimated ages that indicate diversification within these lineages began in the early Oligocene (30.2–31.1 Mya; fig. 1). In comparison with previous time-calibrated phylogenies for Psittaciformes inferred from a small number of molecular markers (Wright et al., 2008; Rheindt et al., 2014), our time-calibrated tree found younger ages for common nodes.
Psittacidae
Psittacinae
The Psittacinae is a low-diversity radiation endemic to Africa comprising two genera, Psittacus and Poicephalus. The Psittacinae split from their sister clade, the Arinae in the early Oligocene (∼30 Mya; fig. 1) well after the proposed split between Africa and South America in the Middle Cretaceous age, about 115–125 Mya (Valencio and Vilas, 1969).
Psittacus
Psittacus are predominantly gray-plumaged parrots with red tails and prominent bare white facial skin. Along with Australian Nymphicus and Melopsittacus, they are among the world's most familiar parrots, and are among the parrots most heavily harvested for the captive bird trade (Mulliken, 1995). Their ethology has been intensely studied (e.g., Giret et al., 2009, Pepperberg et al., 2013). Psittacus was long treated as monotypic, but two species are now conventionally recognized, P. erithacus and P. timneh. In the phylogenomic tree, the sister relationship between the two was stable and dated to the early Pleistocene (1.9 Mya; fig. 3). Relationships within and among populations of the two species have also been addressed through mitochondrial DNA sequences by Melo and O'Ryan (2007).
Poicephalus
The Poicephalus parrots are a small radiation endemic to Africa. They are midsized stocky parrots, predominantly green with dark upperparts but with species-specific bands of color often on the abdominal (belly) plumage, underwings and rump. We note below (see Psephotellus of Australia) a possible case of convergent plumage evolution involving several species of Poicephalus.
Pioneering molecular work on Poicephalus (Massa et al., 2000) was based on random amplified polymorphic DNA (RAPD). While one of its findings, a sister-group relationship between P. cryptoxanthus and P. meyeri, may have reflected plesiomorphy or homoplasy, many of their results are supported broadly here.
The radiation of Poicephalus had a crown age of 7.5 Mya and the majority of the species had divergences in the Pliocene (fig. 3). The topology within Poicephalus was less robust due to low-quality samples and nodes with low support. The issues likely stemmed from the P. flavifrons and P. rueppellii samples coming from historical samples that produced limited data, making them harder to place in the tree. Poicephalus fuscicollis also had lower-quality data and did not meet more stringent filters of quality control of Smith et al. (2023); it was, however, sister to P. robustus in the phylogenomic trees. This result was consistent with previous phylogenetic work based on mtDNA but inconsistent with a phylogeny inferred from concatenated nuclear and mtDNA, which found P. robustus to be more closely related to P. gulielmi (Coetzer et al., 2015). Poicephalus crassus is known from few museum specimens and has not been sequenced. Given the close similarity in plumage in P. crassus and P. cryptoxanthus, they are likely closely related. The positions of P. rufiventris and P. senegalus varied between estimated trees, but the species tree had a better phylogenetic resolution. Recent work describes an unrecognized subspecies in P. rueppellii that differs in size and plumage color across populations in Angola (Hubers et al., 2023).
FIGURE 3.
Species-level topology of Psittacinae. Support values come from the maximum likelihood tree. Nodes have ultrafast bootstrap values of ≥95% otherwise noted.
Arinae
Subfamily Arinae is highly diverse in species richness and morphological variation, yet it is a cohesive monophyletic assemblage occurring in the Neotropical region. Revisionary work has delimited four tribes within Arinae that are all supported by our phylogenomic trees (Schodde et al., 2013). The Arinae has a crown age of 27.1 Mya and the tribes (Arini, Androglossini, Amoropsittacini, and Forpini) began diverging within a few million years (fig. 1). The crown ages of tribes were widely variable. The diverse Arini, the long/attenuate-tailed macaws and allies, had a crown age of only 14.5 Mya, whereas the largely short/round-tailed amazons and allies of Androglossini had a crown age of 25.6 Mya (fig. 1). Amoropsittacini also had a deep divergence of 21.1 Mya separating Touit and Bolborhynchus/ Psilopsiagon/Nannopsittaca (fig. 1). The Forpini diverged from the clade containing Arini/Androglossini 26.5 Mya and were estimated to have begun diversifying at around 5 Mya (figs. 1, 4). The position of the Forpini in relation to the other tribes in Arinae has varied across phylogenetic studies. In the phylogenomic tree, Forpini was separated from Arini/Androglossini by a short internode, consistent with the instability observed in this part of the tree.
For discussion of the often contentious topic of extinct parrots in the Caribbean region, in particular species of the genera Ara and Amazona, we refer the reader to Williams and Steadman (2001), Olson and López (2008), Gala and Lenoble (2015), Forshaw and Knight (2017), and Oswald et al. (2023).
Forpus
Brereton (1963) first published Forpini and Amoropsittacini diagnosed in symbols and letters in a table with an accompanying legend explaining the meaning of the symbols and letters in words. Schodde et al. (2013) accepted this as validating the names. Sangster et al. (2023) nevertheless disputed it, claiming that Brereton's descriptions did not conform to the requirements of the current edition of the International Code of Zoological Nomenclature (ICZN, 1999), and replaced Brereton's names with their own. Schodde et al. (submitted) have since applied to the International Commission on Zoological Nomenclature to conserve Forpini and Amoropsittacini as published by Brereton (1963). The details of these nomenclatural arguments are to be explained fully in that paper but hinge on due and fully appropriate detail having been given by Brereton (1963). Until this is settled, we consider Forpini and Amoropsittacini as valid family-group names.
FIGURE 4.
Time-calibrated phylogeny of Forpini. Support values come from the maximum likelihood tree. Nodes have ultrafast bootstrap values of ≥95% otherwise noted.
The Forpini contains nine small, stocky-bodied, sexually dimorphic, and largely green species in the genus Forpus. Forpus is currently distributed from Mexico to Argentina, but the fossil record suggests that it occurred in the Bahamas up until at least the Late Pleistocene (Steadman and Franklin, 2020). Recent molecular (Smith et al., 2013) and morphological (Bocalini and Silveira, 2015) data were used to elevate F. spengeli of northwest Colombia and F. crassirostris of western Amazonia to species rank. Bocalini and Silveria (2015) assessed phenotypic variation within F. xanthopterygius and synonymized the remaining taxa (crassirostris, flavissimus, flavescens, and olallae) within monotypic F. xanthopterygius due to high morphological variation among localities and because plumage patterns did not conform to described subspecies limits. This finding has important implications for species limits in the genus and highlights the limits of using morphological data to assign taxonomic ranks. As recovered in previous work (Smith et al., 2013), our phylogenomic tree had F. crassirostris as sister to a clade containing all other Forpus except F. modestus and F. cyanopygius. The timing of this divergence (approximately 3.8 Mya; fig 4) was similar to the age estimated using a small number of mtDNA and nuclear DNA loci (Smith et al., 2013). This deep intraspecific divergence, which was not reflected in the high number of specimens examined by Bocalini and Silveria (2015), indicates that morphological evolution is not tracking population history and is thus not a reliable data source for species delimitation.
The opposite is true for F. passerinus. Forpus spengeli was originally treated at species rank until at least Cory (1918). To the best of our knowledge, it was then placed in F. passerinus by Peters (1937), without comment. At a later date, it was transferred to F. xanthopterygius (Forshaw, 1973). Smith et al. (2013) found that spengeli was nested within passerinus differing by only a few base pairs in mtDNA, whereas Bocalini and Silveria (2015) found spengeli to be distinct given its clearly diagnostic turquoise rump on males. They restored the taxon to species rank as F. spengeli. Bocalini and Silveria (2015) included F. passerinus in their study, but did not consider that F. spengeli could be a geographical variant within F. passerinus or, alternatively, its mtDNA may have been introgressed and/or captured by the nearest population of F. passerinus, i.e., F. passerinus cyanophanes. Forpus passerinus, which exhibits a wide variation in rump color in males, ranging from blue to green even with excluding F. spengeli. The molecular changes that result in a shift from a green to turquoise feather patch may be simple, as observed in captive Melopsittacus undulatus, in which a single base-pair change expresses tryptophan, blocking expression of yellow pigment, and so changes green feathers to blue (Cooke et al., 2017). The identical mutation was also found to produce blue feathers in captive Agapornis (Ke et al., 2024), suggesting there may be a single genetic mechanism underlying the green to blue color shift in parrots. While the phylogenomic tree here strongly shows a deep divergence between F. passerinus and F. spengeli (fig. 4), caution is warranted when interpreting this divergence in light of previously characterized shallow mtDNA divergence. The phylogenomic tree is either reflecting discordance between nuclear and mtDNA, possibly due to mitochondrial introgression, or the branch length is artificially long due to the high number of singletons in phylogenomic data. Population-level sampling is necessary to distinguish these scenarios.
FIGURE 5.
Time-calibrated phylogeny of Amoropsittacini and Touitini. Support values come from the maximum likelihood tree. Nodes have ultrafast bootstrap values of ≥95% otherwise noted.
Amoropsittacini/Touitini
We acknowledge the recent introduction of Touitini by Sangster et al. (2023) to encompass the genera Touit, Bolborhynchus, Psilopsiagon, and Nannopsittaca, and thereby replacing Amoropsittacini Brereton, 1963. As noted above, we similarly acknowledge concomitant preparation of a counterargument by Schodde et al. (submitted) to conserve the family-group name Amoropsittacidae Brereton, 1963. If accepted, the latter still allows retention of Touitini for recognition as a monotypic tribe i.e., containing only Touit. Until this is resolved, we use Touitini for Touit and Amoropsittacini for the other three genera for reasons we now outline.
There was a deep divergence that dates to 21.1 Mya (13.7–25.5) between Touit and the clade containing Bolborhynchus, Psilopsiagon, and Nannospsittaca, which in turn exhibits a basal divergence dating to 16.9 Mya (10.3–21; fig. 5). Based on these deep divergences, we advocate recognition of two tribes, one for Touit and one for Bolborhynchus, Psilopsiagon, and Nannopsittaca, this decision being driven largely by the molecular data.
Touit, Bolborhynchus, Nannopsittaca, and Psilopsiagon
The four genera in this group (Touit, Bolborhynchus, Nannopsittaca, and Psilopsiagon, whether in Amoropsittacini or Touitini as discussed above) comprise small-bodied and largely green-plumaged species occasionally broken up with traits such as black barring, small areas of other colors about the head and face, differently colored underwing coverts, or, as in one species, a crown demarcated from the upperparts and underparts. Tails vary from long and attenuate to short and blunt. Touitini is generally more morphologically uniform than Amoropsittacini. Species in Touit have small, stocky bodies with round tails, whereas Amoropsittacini shows greater variation in gross morphology, tail morphology and body size. Sexual dimorphism is also much more pronounced.
Turning to species-level relationships, we first report the relationships inferred from the concatenated tree. Touit stictopterus of central Colombia to northern Peru diverged from the remaining Touit at 10.1 Mya (5.9–13.1; fig. 5). The next divergence represents a mostly east-west break across the Andes. The clade comprising sister species T. dilectissimus and T. costaricensis diverged across the Andes at 5.5 Mya (3.1–7.6) from the other clade entirely east of the Andes (T. surdus, T. purpuratus, T. batavicus, T. melanotus, and T. huetii; fig. 5). The divergence across the Isthmus of Panama between T. dilectissimus and T. costaricensis occurred 1.1 Mya. This timeframe aligns with other humid forest birds that colonized North America (Smith et al., 2012). The first divergence in the clade east of the Andes was between T. huetii and the other taxa (fig. 5). Touit huetii is widespread across Amazonia in relatively small allopatric populations. Touit melanonotus of southern Brazil and T. batavicus of northern Colombia, Venezuela, Suriname, the Guianas, and northern Brazil are highly disjunct with respect to each other, separated by the Amazon Basin and the South American dry diagonal, including the Caatinga and Cerrado Biomes. That disjunction dates to the late Pliocene–early Pleistocene. A second large disjunction is between sister species T. purpuratus of northern Amazonian and T. surdus of southeastern Brazil. In contrast, the species tree found notable differences in relationships. However, some taxa (T. costaricensis, T. melanonotus, and T. surdus) with lower-quality samples could not be accurately placed in the species tree. The most notable difference was the placement of T. huetii, which was sister to T. purpuratus in the species tree.
The species in the other three genera occur primarily in montane regions, except Nannopsittaca dachilleae of lowland western Amazonia. The traditionally recognized genera Bolborhynchus and Nannopsittaca, however, are not monophyletic. That is, two of three nominal species within Bolborhynchus (B. ferrugineifrons and B. orbygnesius) and the two species of Psilopsiagon (P. aymara and P. aurifrons) comprise a clade that represents a radiation across the high Andes, with its species-level divergences ranging from 3.6–7.6 Mya (fig. 5). The concatenated and species trees represented here are mostly concordant. Bolborhynchus ferrugineifrons was accurately placed in the same clade in both phylogenomic trees but had a significant amount of missing data at parsimony informative sites. The order of P. aymara and P. aurifrons also differed in the species tree, but they were still paraphyletic with respect to each other, a result that we note with much caution. Although B. lineola and the two species in Nannopsittaca form a clade, they are biogeographically less cohesive. Nannopsittaca panychlora occurs disjunctly in small, restricted parts of Venezuela in the tepui region from 750 to 1850 m (Forshaw and Knight, 2010). Bolborhynchus lineola is widely distributed in montane regions from Mexico to Peru. Nannopsittaca panychlora was sister of N. dachilleae and B. lineola, and so Nannopsittaca was paraphyletic.
We consider five alternative taxonomic scenarios for addressing the paraphyly within Bolborhynchus, Nannopsittaca, and potentially Psilopsiagon (see fig. 5). They are:
Most expediently, all species could be placed in the oldest of the three generic names, Bolborhynchus Bonaparte, 1857. Previous authors have indeed treated Psilopsiagon aymara and P. aurifrons within Bolborhynchus (De Schauensee and Eisenmann, 1966, Forshaw and Knight, 2010). This would produce a genus with high interspecies morphological disparity reflected temporally in the clade dating to the mid-Miocene. Conversely, there is no precedent elsewhere within Psittaciformes for a genus having a crown age as deep as 16.9 Mya, which reflects a times-pan encompassing substantial evolutionary divergence. Recalling Ford's (2006) suggestion that a genus should make sense to someone not familiar with it, we reject this option.
Given that Bolborhynchus lineola (Cassin, 1853) = Psittacula lineola Cassin, 1853, is the type species of Bolborhynchus (and, parenthetically, of Grammopsittaca Ridgway, 1912), the two species of Nannopsittaca Ridgway, 1912, although paraphyletic with respect to each other, could be placed in Bolborhynchus with B. lineola. This is consistent with O'Neill et al.'s (1991) case that they are congeneric notwithstanding that we here advocate placing them in the same genus as B. lineola. Two other species currently in Bolborhynchus (ferrugineifrons, orbygnesius) could then be placed in Psilopsiagon with P. aymara and P. aurifrons. However, this option also produces genera with species that are morphologically disparate albeit less so than the previous option.
As option 2 but reinstating monotypic Amoropsittaca Richmond, 1915, only for aymara.
Retention of N. panychlora in monotypic Nannopsittaca, although counter to O'Neill et al.'s (1991) view that Nannopsittaca is ditypic, opens this fourth option. Notwithstanding the superficial similarity in gross morphology of the two species usually placed in Nannopsittaca (e.g., fig. 5), we find this a reasonable option for two main reasons: (1) N. panychlora diverged from N. dachilleae/B. lineola at 7.5 Mya (4.2–9.8), a depth of divergence that was similar to deeper intra- and intergeneric splits within Psittaciformes, and (2) it was the only Neotropical parrot endemic to the Pantepui region of northern South America (e.g., Mayr and Phelps, 1967). By this option, B. ferrugineifrons and B. orbygnesius and aymara and aurifrons could be placed in Psilopsiagon as already suggested, and N. dachilleae would be moved to Bolborhynchus because it was sister to the type species of that genus, B. lineola. Bolborhynchus lineola and N. dachilleae are strongly associated with bamboo albeit of different genera, i.e., Chusquea spp. and Guadua spp., respectively (O'Neill et al., 1991; del Hoyo et al., 1997; Harvey et al., 2014).
As option 4 but with aymara in Amoropsittaca.
Choosing among these options inevitably becomes subjective. We favor option 4, arguing that retention of aymara in Psilopsiagon is prudent until its position relative to the other species in that genus is clarified. Similarly, the position of aurifrons was poorly supported (UFBS = 74%). This is consistent with our generic decisions made elsewhere to convey as accurately as possible well-supported phylogenomic diversity (e.g., Licmetis vs. Cacatua; Zanda vs. Calyptorhynchus; Psephotellus vs. Clarkona; Aratinga vs. Nandayus), regardless of the patterns of phenotypic diversity among the various species. Further, it retains all three recognized genera albeit with some inevitable but minimal disruption to their usage, and it presents no conflict between classification and phylogeny. Although it places in the same genera species that are somewhat different in gross morphology, e.g., relative tail length and shape, this is common to either of the arrangements we discussed (see fig. 6). The difficulty of discerning phenotypic traits in diagnosing the three genera as so circumscribed warrants brief comment. We suggest that this likely arises from very strong selective pressures on these species for cryptic, largely green plumage (see fig. 6). This could feasibly apply to either species inhabiting and primarily foraging in lowland or upland rainforest canopies, such as dachilleae and panychlora, respectively, or to those foraging in other montane habitats such as lower, scrubby hillside vegetation and terrestrial bogs, as in most of the other species.
FIGURE 6.
Specimens from the American Museum of Natural History (AMNH) and Louisiana State University Museum of Natural Science (LSUMNS) showing proposed groupings for (left to right): Nannopsittaca panychlora AMNH 324136, Bolborhynchus (dachilleae LSUMNS 156199; lineola AMNH 820295), and Psilopsiagon (aymara AMNH 136906; aurifrons AMNH 151238; orbygnesius AMNH 168922; and ferrugineifrons AMNH 44744). Photographs: Sahid Robles Bello.
We stress that the need for recognition of these three genera and their circumscriptions as we have suggested have arisen largely from our concerns with reflecting the phylogenomic data and structure among them. In summary, we advocate the following taxonomy with phenotypic notes (fig. 6):
Psilopsiagon Ridgway, 1912: P. ferrugineifrons, P. orbygnesius, P. aurifrons, P. aymara. Predominantly green plumage with patterning that does not involve barring and moderate to long tails. Species membership in Psilopsiagon is tentative pending clearer resolution of phylogenetic relationships. Amoropsittaca Richmond, 1915 is available for aymara if warranted.
Nannopsittaca Ridgway, 1912: N. panychlora. Short-tailed, green plumage with little or no blue particularly in the crown but with a distinctive “broken” ring of yellow periophthalmic feathering. Orbital region mostly feathered (Ridgway, 1916). Bill and cere generally dark gray.
Bolborhynchus Bonaparte, 1857: B. lineola, B. dachilleae. Diagnostic phenotypic traits shared by the two species are difficult to discern beyond both being short-tailed green parrotlets. Bolborhynchus lineola has barred dorsal plumage, which we postulate to be derived, recalling that of Australian Pezoporus and Melopsittacus and arising from black feather centers on the ventral surface especially under the tail. B. dachilleae has generally plainer green plumage, which we postulate as ancestral, apart from a distinctive blue crown. In B. dachilleae, the orbital region's bare skin is whitish and the bill and cere are pinkish buff, not dark gray as in Nannopsittaca (O'Neill et al., 1991) whereas the similarly colored bill of B. lineola is often distally tipped dark. Bolborhynchus lineola and N. dachilleae are strongly associated with bamboo (see option 4 above), an ecological preference, to our knowledge that has not been observed in N. panychlora.
Androglossini and Brotogerini
The next clade to be discussed is diverse, consisting of the amazons and their allies, and has mostly been termed the Androglossini in recent literature (e.g., Joseph et al., 2012; Schodde et al., 2013). We acknowledge the current literature on the correct family-group name for this tribe. Gregory and Sangster (2023) advocate Amazonini in contrast to Joseph et al.'s (2012) and Schodde et al.'s (2013) arguments for Androglossini. Schodde (accepted) has further contested the use of Amazonini in favor of Androglossini. Until this is settled, we use Androglossini.
More immediately relevant here, then, is that within the Androglossini there was a deep divergence dating to 25.6 Mya (17.7–29.9; fig. 1). The two clades resulting from this divergence are diagnostic with respect to their tail morphology. One clade comprises the genera Myiopsitta and Brotogeris, which have longer, attenuate or pointed tails that are atypical characteristics of the tribe. The sister clade, comprising Pionopsitta, Triclaria, Pyrilia, Hapalopsittaca, Amazona, Pionus, Graydidascalus, and Alipiopsitta, has species with mostly round and proportionately short tails. The relationship is supported in the species tree, but support for the node and inclusion of both Brotogeris and Myiopsitta varied depending on how strictly the sample retention was set. Interestingly, whole genome data from an exemplar from each clade (Myiopsitta monachus and Amazona aestiva) have diploid numbers of 48 and 70 (Huang et al., 2022). This level of variation, due to the fusion of autosomal and sex chromosomes, is atypical for birds because avian karyotypes and genome sizes are generally conserved; birds usually have diploid numbers with a range of 76–82 (Kapusta and Suh, 2017; Zhang, 2018; Furo et al., 2020). The two species of Brotogeris that have been karyo-typed exhibit different diploid numbers (B. sanctithomae 72 and B. versicolurus 82 chromosomes; de Lucca et al., 1991), indicating that extensive chromosome fusion may not be a general characteristic of the clade containing Brotogeris and Myiopsitta.
We are aware of no prior family-group name to accommodate the genera Brotogeris and Myiopsitta. In accordance with articles 13.1 and 13.2 of the Code (ICZN, 1999) we therefore introduce a new family-group name at the rank of tribe for these two genera:
Brotogerini, new tribe
Zoobank registration: urn:lsid:zoobank. org:pub:3BFB5FD8-9DDC-4419-9EEFD41AFBB2E9B1
Diagnosis: In accordance with Article 13.1.1, we note that Brotogerini differs from the members of the Androglossini (in the sense used above) from which it is separated principally in the morphology of the tail, which relative to the body is long and attenuated in Brotogeris and Myiopsitta rather than short and rounded in all other genera in Androglossini. In accordance with Article 13.1.2, we note that this difference has been well illustrated and described on many occasions in the literature, citing as examples reference works on the world's parrots: e.g., Forshaw and Knight (2010); Juniper and Parr (1998).
Type genus Brotogeris Vigors, 1825, Zoological Journal 2: 400, by original designation. Type species Psittacus pyrrhopterus Latham 1801, Supplementum Indicis Ornithologici, 1801: xxii.
Taxonomic position of tribe: Subfamily Arinae of family Psittacidae.
Component genera: Brotogeris Vigors, 1825, and Myiopsitta Bonaparte, 1854.
Myiopsitta Bonaparte, 1854. Revue et magasin de zoologie pure et appliquée.(2), 6, 1854: 150. Type, by subsequent designation, Psittacus monachus Boddaert (G.R. Gray, Catalogue of the Genera and Subgenera of Birds Contained in the British Museum , 1855: 87)
Geographic distribution: Endemic from South America to Mexico.
Brotogeris
Brotogeris are smaller, largely green parrots with pointed, graduated tails varying in relative proportion to body size, however. Relationships among the eight recognized species in the genus are strongly supported and concordant across molecular phylogenies (e.g., Ribas et al., 2009). General tail morphology does not conform neatly to phylogenetic groups. The three longer-tailed species (B. tirica, B. versicolurus, and B. chiriri) are in one clade with shorter-tailed B. sanctithomae of the Amazonian floodplain. The other clade contains B. pyrrhoptera, B. jugularis, B. cyanoptera, and B. chrysoptera. The majority of the species are distributed east of the Andes in the Amazon Basin. There was a cross-Andes divergence (1.3–3.7 Mya) represented by the split between B. pyrrhoptera/B. jugularis and B. cyanoptera/B. chrysoptera. Brotogeris jugularis is the only species in Central America, so dispersal out of South America occurred over the Panamanian land bridge. No study to date has sampled across the range of B. jugularis, which spans from Mexico to Venezuela, but northward dispersal most likely occurred within the Pleistocene, the time frame after B. pyrrhoptera and B. jugularis split.
Myiopsitta
Myiopsitta are midsized parrots with pointed, attenuate tails. Uniquely among all parrots it nests not in hollows in trees, cliffs, or rock faces, but in nests of sticks constructed by the birds and placed in trees, cliffs, or in more anthropogenic environments on poles and near buildings (see Agapornis below for other parrots that construct smaller nests). They occur in southern South America and have a number of naturalized populations around the globe (Edelaar et al., 2015). Myiopsitta monachus occurs in southern Brazil through northern Argentina and Uruguay, and M. luchsi is distributed in the Andean valleys of Bolivia. Myiopsitta luchsi has long been a subspecies of M. monachus and is still treated by some as such (e.g., South American Checklist Committee) whereas others have elevated it to species rank (e.g., IOC). The rationale that has been given for species-rank recognition of M. luchsi is in its genetic and phenotypic distinctiveness, ecological divergence in occupying higher elevations, and obligatory cliff-nesting, as opposed to nest construction in trees (Russello et al., 2008). The South American Checklist Committee (proposal 503) rejected these arguments based on the limited mtDNA differentiation of luchsi from other taxa, and the lack of both a formal vocal analysis and a published description of how their nesting ecology differs. The phylogenomic molecular dating provides new temporal context for the debate on species limits in Myiopsitta, the divergence likely more than a million years (1.7 Mya: 0.8–2.6).
FIGURE 7.
Species-level time-calibrated topology of Androglossini and Brotogerini. Support values come from the maximum likelihood tree. * denotes an unsupported node where the topology of the presented time-calibrated phylogeny differs from that of the maximum likelihood tree. Nodes have ultrafast bootstrap values of ≥95% otherwise noted. Gray bars represent divergence time ranges were estimated from 100 bootstrap trees.
Androglossini
With Brotogeris and Myiopsitta now separated in Brotogerini, we note that relationships among genera that remain within Androglossini were well resolved. Pionopsitta pileata (13.3–24.1 Mya) and then Triclaria malachitacea (11.9–22.1 Mya) were deeply split from the remaining taxa in their clade (fig. 7). In the species tree from Smith et al. (2023), Triclaria malachitacea was sister to the clade containing Pyrilia and Hapalopsittaca with varying levels of support across trees with different filtering treatments (LPP = 0.6–0.98). Both of these monotypic genera are endemic to humid, rainforest habitats in Brazil's Atlantic Forest, and they are predominantly in sympatry. Notably, the Atlantic Forest has many other “relict” species, which like these two genera, are on long branches in phylogenetic analyses (e.g., Pyrrhura cruentata).
The next major divergence in the core Androglossini was the split between Pyrilia/Hapalopsittaca and the remaining genera at 11.4–21.4 Mya. The final subclade had Amazona sister to Pionus/ Graydidascalus/Alipiopsitta. Graydidascalus of Amazonian floodplains and Alipiopsitta of open cerrado woodlands in central Brazil are also monotypic genera. Alipiopsitta xanthops was formerly within Amazona, but it has been consistently placed in monotypic Alipiopsitta due to its robustly supported placement outside Amazona (Tavares et al., 2006; Schirtzinger et al., 2012). The two genera were estimated to have shared a common ancestor 4.1–10.4 Mya and diverged from Pionus 6.8–15.8 Mya (fig. 7). In the species tree Alipiopsitta xanthops was sister to Amazona/ Pionus/Graydidascalus, but the sample was dropped from more stringent alignments due to a high amount of missing data at parsimony informative sites (Smith et al., 2023).
Pionopsitta
The sole species in this genus, Pionopsitta pileata, is midsized and mostly green apart from its red coronal and facial plumage. Although it ranges into some drier forest and woodland habitats in Argentina and Paraguay, it is essentially an Atlantic Forest endemic of southeastern Brazil. Our data support the now well-established retention of this species in monotypic Pionopsitta of which it is the type species, and the removal of the other species that had been placed in Pionopsitta into Pyrilia (see Ribas et al., 2005; Eberhard and Bermingham, 2005). The distinctiveness of Pionopsitta was further reflected by its deep split from the rest of Androglossini at 20.2 Mya (13.3–24.1).
FIGURE 8.
Photographs (not to scale) of Calyptomena hosii (left photo: Dubi Shapiro) a suboscine passerine of Borneo (Brunei, Indonesia) and Triclaria malachitacea (right photo: Marcos Eugênio) of southeastern Brazil showing presumably convergent evolution in ventral coloring. See text for discussion. Photographs reproduced with permission from the photographers.
Triclaria
Triclaria malachitacea is a midsized green parrot in which males have a unique blue abdominal (belly) patch sometimes extending to the lower breast, and with a relatively long tail for Androglossini. It is another Atlantic Forest endemic found in the lower montane forests of southeastern Brazil (Forshaw and Knight, 2010). Apparently, the striking and presumably convergent evolution in the plumage of the underparts of this species and that of a similarly sexually dichromatic suboscine passerine Calyptomena hosii of Indonesia has never been noted (fig. 8). Study of the drivers of this convergence would surely be rewarding and presumably may relate to the function of blue in displays of birds inhabiting forests and the forest strata at which the blue is functional in those rainforests (see Endler, 1993).
Pyrilia
Pyrilia are midsized, stocky parrots of humid forest lowlands in Central America through the Amazon Basin. They are predominantly green, but each has a distinctively colored pattern of plumage or skin about the head. Two species remarkable for largely bare heads were formerly placed in Gypopsitta Bonaparte, 1856. Banks et al. (2008) demonstrated why Pyrilia Bonaparte, 1856, has priority over Gypopsitta Bonaparte, 1856, for this assemblage.
The phylogenomic concatenated and species trees differ in the placement of P. caica amongst the trees and with previous studies based on mitochondrial DNA and plumage characters (Ribas et al., 2005, Eberhard and Bermingham, 2005). In the phylogeny by Ribas et al. (Ribas et al., 2005), which includes subspecific sampling, P. caica was sister to P. vulturina/P. aurantiocephala. The phylogenomic species tree, however, had P. caica as sister to P. barrabandi, and they were in turn sister to P. vulturina/P. aurantiocephala. The phylogenomic concatenated tree had P. caica as sister to the clade containing P. barrabandi, P. vulturina, and P. aurantiocephala. These alternative topologies support different biogeographic sequences of divergence. The species tree sister relationship between P. caica and P. barrabandi supports an initial break across the Rio Negro, whereas the concatenated and prior mtDNA trees support alternative scenarios involving the initial divergence across the Amazon river and Andes. None of the trees support a clade (P. haematotis, P. pulchra, and P. pyrilia) distributed west of the Andes indicating there were multiple crossings over or around the Andes. The time-calibrated phylogenomic tree supported a relatively shallow temporal scale for the diversification of Pyrilia (crown age 4.5 Mya [2.6–6.5]).
Hapalopsittaca
Hapalopsittaca is a low-diversity clade comprising four midsized species that occur in allopatry across humid forests of the Andes. Relationships among the species reflect a stepping-stone pattern across the linear distribution of the Andes. The first split, dated to 4.6 Mya (2.8–6), was between a northern clade containing H. amazonina, H. fuertesi, and H. pyrrhops from a southern lineage, H. melanotis, which occurs disjunctly along the eastern slopes of the Andes to Huánuco to Cusco, Peru, and La Paz to Cochabamba, Bolivia. The two remaining divergences occurred within the Pleistocene and resulted in the divergence of H. amazonina of the Colombian Andes from H. fuertesi of the Cordillera Central in western Colombia, and H. pyrrhops ranging from southern Ecuador to northern Peru.
Amazona
Amazona is a diverse yet easily recognizable clade of mid- to large-sized stocky birds with relatively short tails, large heads, and barred plumage. Most species are green with colorful patterning on the head but a few Caribbean species have plainer or more elaborate plumage. The concatenated and species trees are largely concordant and indicate at least three independent colonizations of the Caribbean. The main discrepancy between them arises from the relatively high number of lower-quality samples that were not accurately placed in the species tree. Our topology was generally concordant with that of Rusello and Amato (2004) with the exception of the position of two clades (1, 2) that stemmed from short internodes and were placed in our tree with high support: (1) A. festiva, A. pretrei, A tucumana, and A. vinacea; and (2) A. amazonica, A. guildingii, A. brasiliensis, and A. imperialis. The basal divergence in Amazona dated to 9.9 Mya (4.7–13.1; fig. 7). It separates a clade of smaller bodied Caribbean (A. agilis, A. vittata, A. ventralis, A. leucocephala, and A. collaria and Central American/ Mexican (A. xantholora and A. albifrons) species from all other Amazona, a phylogenetic position recovered previously by Russello and Amato (2004). This clade includes dichromatic species (A. agilis, A. xantholora, and A. albifrons), a character state that is generally lacking in Amazona. Ancient DNA and archaeological evidence show that Amazona in the Greater Antilles, specifically A. leucocephala and A. ventralis, were translocated across the Caribbean and had natural, more widespread ranges in the past (Oswald et al., 2023). Amazona leucocephala was also found to have intraisland divergences throughout the Pleistocene. The next divergence results in two sister clades whose short node now has more support (UFBS = 91% vs. BS = 53%) than that in Russello and Amato (2004). The first clade had a crown age of 9.2 Mya (4.4–12.1) and includes a heterogeneous group of species spanning from Mexico (e.g., A. viridigenalis, A. finschi) to Brazil (A. diadema, A. amazonica; fig. 7). It also represents a colonization of the Caribbean (A. imperialis and A. guildingii). Amazon guildingii is polymorphic, having yellow-brown and green plumage morphs (Forshaw and Knight, 2010). The second of the major sister clades is also geographically widespread and consists of two independent colonizations of Central America. The first was seen in the divergence within what was formerly treated as a single species A. farinosa. It supports species rank for its Central American populations now recognized as A. guatemalae with two subspecies A. g. guatemalae and A. g. virenticeps. This accords with Wenner et al.'s (2012) phylogeographic study on the A. farinosa complex in which they detected a deep divergence between Central and South American lineages (1.75–2.70 Myr split in mtDNA). Our estimate was comparable but older. Together, the two data sets support divergence that reflects independent evolution of A. farinosa and A. guatemalae irrespective of a potential contact zone, vocal differences, or morphological distinctiveness. Chesser et al. (2023) did not recognize A. guatemalae as a species, arguing that a more thorough population-level study is necessary to better understand species limits. The second divergence across the Isthmus of Panama involves the yellow-headed Amazons and the split of A. ochrocephala from A. oratrix and A. auropalliata of Central America and Mexico, and A. tresmariae of the Tres Marias Islands of the Pacific Coast of Mexico. The recognition of these lineages as species (as opposed to subspecies) was in part due to phylogeographic studies that found deep genetic divisions within the yellow-headed Amazon complex (Eberhard and Bermingham, 2004; Ribas et al., 2007b). Amazona tresmariae, as with A. guatemalae, was not recognized as a species by the North American Checklist Committee for similar reasons of inconclusive data (Chesser et al., 2023). This clade was sister to the third colonization of the Caribbean where A. barbadensis of Venezuela and neighboring islands was sister to A. versicolor of St. Lucia and A. arausiaca of Dominica. The final clade was a group of three species, A. vinacea, A. tucumana, and A. pretrei, that have a southern distribution in Bolivia, Argentina, Brazil, and Paraguay.
Pionus
Pionus comprises a small radiation mostly predominantly either green or bronze-green and blue, midsized parrots with square tails that are distributed from Mexico to Argentina and southern Brazil (Forshaw and Knight, 2010). Pionus has a crown age of 4.5 Mya (2.3–6.3), a temporal span that overlaps the estimated ages of Ribas et al. (2007a), but the phylogenomic ages are younger. There was considerable topological disagreement between the phylogenomic trees and a previously published mtDNA tree of Pionus (2007a). Immediate sister pairs (P. senilis/P. chalcopterus; P. seniloides/P. tumultuosus; and P. maximiliani/P. sordidus) are consistent across all trees, but most other nodes differ. The widely distributed Pionus menstruus was included in the present study but not in Smith et al. (2023), due to a sampling error detected prior to that work's publication. Pionus fuscus of the Guianan Shield and adjoining parts of Amazonia is exceptional in the genus in lacking green in its otherwise predominantly brown, blue and red plumage. Both of these species are in unresolved positions in the concatenated tree. Pionus fuscus was strongly supported as sister to P. senilis and P. chalcopterus in the concatenated tree in Smith et al. (2023). However, the inclusion of P. menstruus in this study destabilized the position of P. fuscus. This may be due in part to our P. menstruus coming from a historical museum specimen. The species tree had P. fuscus as sister to P. seniloides and P. tumultuosus (LPP = 1.0). Ribas et al. (2007a) had a third alternative topology, and placed P. fuscus as sister to P. sordidus and P. maximiliani, and P. menstruus as sister to P. chalcopterus and P. senilis. The disparity between the concatenated and species trees presented here was illuminating in that all nodes are supported in the species tree and all internal nodes differ. Another distinction between studies was that Ribas et al. (2007a) included samples of subspecies, which showed deep divergences within species (e.g., P. sordidus). Subsequent phylogeographic research on Pionus senilis of Mexico and Central America found genetic structure across its range that coalesces to 1.3 Mya (Rivera-Arroyo et al., 2022). Future work with more extensive genome-level sampling will be necessary to resolve the high phylogenetic discordance observed in Pionus.
Graydidascalus
Graydidascalus brachyurus occurs throughout the Amazonian floodplains and is a midsized, predominantly green stocky parrot with black lores, prominent red iris, very short tail, and unusually large bill. As with many other Amazonian birds of floodplains, it is taxonomically monotypic having no described geographic variation (Forshaw, 1973; Remsen and Parker, 1983), yet it may contain uncharacterized genetic structuring along rivers (Thom et al., 2020). Graydidascalus brachyurus shares a common ancestor with Alipiopsitta xanthops 7.8 Mya (4.1–10.4), reflecting a long period, in both taxa, during which taxonomic or geographic variation did not accumulate.
Alipiopsitta
Alipiopsitta xanthops was formerly placed in Amazona due to similarities in size, shape, and general plumage. However, phylogenetic work since the early 1990s shows that Al. xanthops was paraphyletic with respect to Amazona (Birt et al., 1992), which to our knowledge, has been confirmed by all subsequent molecular studies, including the phylogenomic data presented here. Further, comparisons of karyotypes in Arini showed high similarity between Al. xanthops, Pionus maximiliani, and Graydidascalus brachyurus, and indicated a sister relationship between Alipiopsitta and Graydidascalus (Caparroz and Duarte, 2004). That sister relationship was supported by DNA sequence data (Russello and Amato, 2004) and our phylogenomic tree. Its continued placement in Amazona in some sources (e.g., Forshaw and Knight, 2010), likely an artifact of when the book was published, is unfounded. As noted by Caparroz and Pacheco (2006), Miranda-Ribeiro (1920) erected a new genus Salvatoria for xanthops to differentiate it from “true” Amazona, but this name was preoccupied by Salvatoria McIntosh, 1885, a genus of polychaete worms. Accordingly, Alipiopsitta was erected (Caparroz and Pacheco, 2006). Alipiopsitta xanthops is also polymorphic in that its plumage has green and yellow morphs. Such plumage polymorphism, excluding sexual and age-related dimorphism, is generally rare in parrots, other examples being Charmosyna stellae, Pseudeos fuscata, Amazona ochrocephala, and A. guildingii. There is no described geographic variation in Al. xanthops. As noted, Al. xanthops and G. brachyurus shared a common ancestor 7.8 Mya (4.1–10.4).
Arini
Arini exhibits a wide range in body sizes and is a colorful group, including the largest extant parrots, the macaws. It ranges from Mexico (formerly southwestern U.S.) to Argentina. Within Arinae, the Arini is the most speciose clade. Arini also contains a number of other low-diversity genera (2–3 species). Enicognathus, Rhynchopsitta, and Pionites, which, like Deroptyus and Cyanoliseus, are all on long branches within Arini (fig. 9). The round-tailed Pionites and Deroptyus shared a common ancestor with the clade that comprises all other Arini at 14.5 Mya (7.1–19.2), which are long-tailed parrots, parakeets, and macaws (fig. 9). Depending on the threshold used to include lower quality samples, the position of Rhynchopsitta of Mexico and, formerly, Arizona varied across different phylogenomic trees. The main source of phylogenetic uncertainty for Rhynchopsitta was the position of Pyrrhura, which was separated from other clades within Arini by a short branch. In the concatenated tree, albeit with more modest ultrafast bootstrap support (81%) and in the most stringent species tree with high support, Rhynchopsitta was sister to all other Arini (except Pionites and Deroptyus; Smith et al., 2023). Previous work recovered Rhynchopsitta as part of a polytomy at the base of Arini or in an unsupported position (Tavares et al., 2006; Wright et al., 2008). Rhynchopsitta's phylogenetic position curiously also reflects one of two large biogeographic disjunctions in Arini. Its distribution in the Mexican highlands can be explained by a single colonization of Central and North America. The occurrences of all other species in Arini in Central America and Mexico are attributable to multiple independent dispersal events. In addition to the colonization of North America and the current large disjunction from South American lineages, Rhynchopsitta species are also characterized by a considerable ecological shift. The genus specializes in high-elevation pine forests that occur throughout temperate western North America to Nicaragua, a habitat type that has low psittaciform diversity and one species of which, R. terrisi, obligatorily nests in cliff faces.
FIGURE 9.
Time-calibrated topology of Arini. Support values come from the maximum likelihood tree. * denotes an unsupported node where the topology of the presented time-calibrated phylogeny differs from that of the maximum likelihood tree. Nodes have ultrafast bootstrap values of ≥95% otherwise noted. Gray bars represent divergence time ranges were estimated from 100 bootstrap trees.
The remaining taxa in Arini fall into two clades that separated between 6.4–18 Mya (fig. 9). One clade consists of the most speciose genus in Arini, Pyrrhura, which began radiating 7.1 Ma (2.2–10.6). The sister clade to Pyrrhura contains 15 genera, reflecting the high morphological disparity within this group of Arini. The basal divergence within this sister clade was that of the two temperate South American sister genera (Enicognathus and Cyanoliseus) from the remaining taxa in Arini. As with previous studies (Wright et al., 2008; Schirtzinger et al., 2012), the phylogenomic tree confirms that neither the large-bodied macaws nor the blue macaws are monophyletic (fig. 9). The large-bodied and large-billed blue macaws in Anodorhynchus diverged from the remaining Arini at an early internode. In contrast, the mostly large-bodied and colorful macaws in Ara were sister to the genus Primolius in a clade that includes all blue Cyanopsitta and Orthopsittaca (fig. 9). The clade containing Ara, Primolius, Cyanopsitta, and Orthopsittaca had a crown age of 8.1 Mya (3.5–11.4; fig. 9). Our results reaffirmed previous findings that Aratinga as traditionally recognized is not monophyletic and that instead breaking it up into three genera Aratinga, Eupsittula, and Psittacara is warranted (e.g., Remsen et al., 2013). The position of Eupsittula was unresolved in the phylogenomic tree (fig. 9). In the concatenated tree, we found that Eupsittula was sister to a clade containing Psittacara and Thectocercus, Diopsittaca, Guaruba, Leptosittaca, and Ognorhynchus with low support (UFBS <70%). In the species tree, Eupsittula was sister to Aratinga, Ara, Primolius, Cyanopsitta, and Orthopsittaca, but also with low support (LPP = 0.79). In both phylogenomic trees, the Aratinga sensu stricto species comprise the sister group to Conuropsis and in a clade with Ara, Primolius, Cyanopsitta, and Orthopsittaca (fig. 9). Also notable within Aratinga sensu stricto is that A. weddellii of western lowland Amazonia was on a relatively long branch and sister to all other Aratinga sensu stricto. It exemplifies a dilemma of whether such species should be separated generically (cf. whether among Australian parrots Psephotellus varius should be separated generically or, at least, subgenerically through genus-group name Clarkona). A genus-group name is available for A. weddellii and we discuss its reintroduction below.
The remaining taxa formerly in Aratinga comprise the genus Psittacara and are sister to a clade containing five monotypic genera (Thectocercus, Diopsittaca, Guaruba, Leptosittaca, and Ognorhynchus). These five South American taxa have a crown age of only 6.4 Mya (2.8–9.1), are phenotypically distinct, and span a wide range of ecoregions.
Deroptyus
Deroptyus comprises one species, D. accipitrinus, whose placement at or near the end of linear taxonomic sequences (e.g., Peters, 1937; Forshaw, 1973; Forshaw and Knight, 2010) reflects either uncertainty as to its relationships or a hypothesized close relationship to Amazona. The species' phenotypic uniqueness warrants comment. While dorsally its plumage is unexceptionally mostly green, uniqueness arises from the combination of the following: dark facial plumage streaked white, the yellow eye of adults, the feathers of the underparts and hind neck red or maroon with a terminal blue band, and the erectile feathers of its hind neck, which can be raised into a ruff, perhaps the only parrot to which this term could validly be applied. These combine to give the species its “hawk-headed” appearance. Its two conventionally recognized subspecies, D. a. accipitrinus and D. a. fuscifrons, occur north and south respectively of the Solimões/Amazon river system.
We have clarified here that Deroptyus is part of the earliest divergence in the Arini, as sister to Pionites on one of the two branches of that earliest divergence (fig. 9).
Pionites
Pionites has traditionally comprised two species of medium-sized, stocky, short-tailed parrots that largely replace each other on either side of the Amazon river system. Within Pionites, del Hoyo and Collar (2014) and Collar et al. (2020) advocate recognition of three species within Pionites leucogaster (P. leucogaster, P. xanthurus, and P. xanthomerius) based on plumage and leg and foot colors, including a reported leapfrog pattern in uppertail color (sensu Remsen, 1984). Collar et al. (2020) acknowledge intergradation in these traits at zones of contact between the various forms but cite no primary data. We note clear inconsistencies in at least one of these traits, leg and foot color, as assigned to photographs of wild birds in Collar et al. (2020), i.e., plainly pale colored legs and feet in birds cited by Collar et al. (2020) as P. xanthomerius, which is putatively dark legged. Although P. l. xanthomerius and P. melanocephalus palidus hybridize along the Ucayali River (Haffer, 1977; Moncrieff et al., 2021), the extent of the hybrid zone seems fairly restricted, which might indicate substantial levels of reproductive isolation between these species. Given the scant and inconsistent current evidence for elevating subspecies of P. leucogaster to species, we advocate retention of only P. leucogaster and P. melanocephalus as the two species in Pionites until more in-depth genetic and phenotypic research is conducted.
Rhynchopsitta
Rhynchopsitta comprises two predominantly green, massively billed species (R. pachyrhyncha and R. terrisi) that occur in the high-elevation pine forests of Mexico, and formerly the southwestern United States. The split between R. terrisi of the Sierra Madre Oriental and R. pachyrhyncha of the Sierra Madre Occidental was dated to 0.5 Mya (0.2–0.8). Despite their relatively recent origin, both taxa are independently evolving. We do not support proposals to treat terrisi as a subspecies of pachyrhyncha (Urantowka et al., 2014b). This young age could reflect a general pattern identified in temperate North American birds that have shallower origins in comparison with tropical species, presumably due to higher extinction rates (Smith et al., 2017). This scenario is consistent with the long branch Rhynchopsitta was placed on. Alternatively, the shallow divergence could reflect a recent colonization of the highland pine forests of Mexico, indicating there has not been enough time to accumulate a deep divergence between the taxa.
Pyrrhura
Pyrrhura are small parakeets easily recognizable by the combination of maroon upper tails, often but not always scalloped or mottled chest plumage, and similarly often differentiated ear coverts and crown and shoulder markings on the wing.
Similar to Amazona, Pyrrhura is among the most speciose genera of Neotropical parrots. We estimate that its species diversity arose via rapid radiation within 7.1 Mya (2.2–10.6; fig. 9). Relationships within Pyrrhura, however, are among the most poorly resolved across the entire parrot phylogeny. This is due to Pyrrhura's rapid radiation, the use of degraded DNA from museum skins for several species and, potentially, extensive introgression and incomplete lineage sorting (see Ribas et al., 2006; Somenzari and Silveira, 2015; Urantowka et al., 2016). Although the most stringently filtered concatenated and species trees yielded a moderately well-resolved topology, there was high discordance in species-level relationships of the trees. Despite most recognized species having been sampled except P. pfrimeri, more phylogenetic work is still needed to resolve finer-scale relationships and species limits within the genus.
Our phylogenomic tree (fig. 9) is concordant with the membership of the three main evolutionary lineages identified by Joseph (2000, 2002) and Ribas et al. (2006), but some species-level relationships within these clades vary among phylogenies. Clade 1 comprises only P. cruentata of the Atlantic Forest, Brazil. Clade 2 is the picta-leucotis complex and clade 3 comprises the remaining species. Forshaw and Knight (2010) described three major groups: species with (1) barring on breast; (2) scalloping on breasts; and (3) neither breast barring nor scalloping. These three groups are not monophyletic in the phylogenomic tree, although some phylogenetic signal is apparent in these traits. Clade 1 (P. cruentata) has neither breast barring or scalloping. The difference has been illustrated in Joseph (2000, 2002: see figs. 2, 3 in both papers). Clade 2, which was sister to P. cruentata (fig. 9), includes representatives with breast barring or scalloping. For example, clade 2 taxa include P. leucotis and P. griseipectus of eastern Brazil, which have breast barring, and P. picta and P. amazonum that have scalloping. Clade 3 also contains the majority of Forshaw and Knight's third group, species without barring or scalloping on breast. The remaining species, without barring or scalloping, do not form a monophyletic group within clade 3. Clade 3 also contains P. rupicola of western Amazonia, whose plumage Forshaw and Knight described as a special case of scalloping on the breast, and several species with barring (e.g., P. frontalis; P. devillei). Despite the presence of some phylogenetic signal in barring, scalloping, or the absence of either, the phylogeny shows this trait should not be used to define taxonomic groups.
Further genomic work to clarify relationships and species limits within Pyrrhura ideally would use fresh, wild-collected material of all taxa, many of which remain unsampled. A focus of such effort should be thorough population-level sampling across the geographical ranges of all nominal species but especially for the taxonomically complex and contentious groups, foremost of which is the P. picta and P. leucotis sensu lato (i.e., sensu Peters, 1937) complex. Here we also note others such as the P. melanura complex. Population-level sampling is especially relevant in a radiation as rapid as that of Pyrrhura. It is critical to understand how diversity within and among populations, as well as patterns of introgression and gene flow among them, can improve decisions about species limits and relationships. Concerning P. picta and P. leucotis sensu lato, we note that our data at best provide weak support for the taxa subsumed under the name P. picta by Peters (1937) as monophyletic, noting that Peters' (1937) long since entrenched view had no accompanying support. This in turn suggests the merit of earlier suggestions to revisit the breaking up of P. picta and P. leucotis (Joseph, 2000, 2002; Ribas et al., 2006). Concomitantly, it questions the persistent recognition of P. picta as polytypic and inclusion of such divergent taxa as caeruleiceps within it at subspecies rank by, for example, Gill et al. (2021) and Remsen et al. (2024). Similarly, and strictly speaking, our data offer ambiguous support for P. leucotis as Peters (1937) construed it, i.e., despite being closely related we do not find strong support for P. leucotis and P. griseipectus as sister taxa. We do nonetheless see strong support for a geographical grouping of western Amazonian taxa (roseifrons, peruviana) in our data and much weaker support for other geographical groupings such as that of northwestern South America-Central America (eisenmanni, subandina, caeruleiceps) and, potentially, northern South America (picta, emma) and central Amazonia (lucianii, amazonum). Determining whether these geographical patterns also correspond to phylogenetic and taxonomic groupings warrants specific attention in future work. Compounding the issue is the recognition of more taxa within the group in recent decades, some of which we have included but many we have not (Delgado, 1985; Ridgely and Gwynne, 1989; Joseph, 2002; Gaban-Lima and Raposo, 2016; Arndt and Wink, 2017).
Enicognathus
Enicognathus comprises two predominantly green species (Enicognathus leptorhynchus and E. ferrugineus) with dark maroon bellies, the latter presumably a recurrent plesiomorphic trait in Arini. They occur only in temperate forests and woodlands in southernmost South America and primarily along its Pacific seaboard. Interestingly, both Enicognathus, and Rhynchopsitta of Mexico, which are, respectively, among the most southerly and northerly distributed parrots in the New World, contained shallow divergences among sister species. Enicognathus leptorhynchus has an elongate maxilla that may be an adaptation to feed on the seeds of Araucaria, a conifer. We find that Enicognathus was not closely related to Pyrrhura, although it shares with many of them and indeed with Orthopsittaca manilatus (fig. 9), a distinctively dark red or maroon abdominal (belly) patch of plumage, presumably a plesiomorphic state within the Arini. Enicognathus leptorhynchus and E. ferrugineus were of recent origin and shared a common ancestor 0.6 Mya (0.2–0.9). Our data suggest no taxonomic changes to long-standing treatment of two species in both Enicognathus.
Cyanoliseus
Cyanoliseus comprises one polytypic species, C. patagonus, of southern South America. Its placement in a monotypic genus is absolutely warranted by its phenotypic uniqueness stemming from unusually brownish-olive plumage, orange belly plumage fringed yellow and its large-bodied, long-tailed morphology. Further, it obligatorily rather than opportunistically nests in holes in cliffs, one of just a few parrots to do so. The biology of the species has been extensively studied (e.g., Masello and Quillfeldt, 2004a, 2004b; Masello et al., 2006) and intraspecific genetic diversity has been addressed by Masello et al. (2011, 2015). It continues to challenge accurate phylogenetic placement. We found it to be one of several long branches diverging from very short internodes in the Arini suggesting it as possibly sister to the assemblage of macaws and their allies (e.g., Aratinga and their conure-like allies; fig. 9).
Anodorhynchus
Anodorhynchus is a genus of three species of blue macaw each with yellow bare skin about the eye and mandible. One species (A. hyacinthinus) is the largest of all extant parrots. Of the two other species, one is extinct and one is endangered and restricted in range to arid zone canyons in eastern Brazil. The extinct Anodorhynchus glaucus was sister to A. leari, diverging 2 Mya (0.7–3.1), and they in turn are sister to A. hyacinthinus, diverging 3.7 Mya (1.4–5.6). This placement of A. leari could not be confirmed in the species tree because the sample was of low quality. For a review of morphological specialization and the ecology of Anodorhynchus, see Yamashita and Valle (1993). Sick and Teixeira (1983) note the obligatory cliff-nesting of A. leari.
Eupsittula
Eupsittula consists of five mostly green small parakeets with long, graduated tails that occur from Mexico to Paraguay. The concatenated and species trees are in agreement and resolved, except for the placement of E. aurea. This species ranges from Suriname and northeastern Brazil to northwestern Argentina. The concatenated tree had South American E. aurea sister to the two Central American species, E. nana and E. canicularis, with 80% ultrafast bootstrap support (fig. 9). Eupsittula aurea and E. canicularis are the only two species in the clade that have orange foreheads. In contrast, the species tree had E. aurea as sister to E. pertinax and E. cactorum in an all South American grouping with a local posterior probability of 1. Provost et al. (2018) had E. aurea grouping with E. nana and E. canicularis, but the position of Central American E. canicularis was unresolved. Resolving relationships among these five species is important biogeographically because it will help determine when Eupsittula first dispersed into Central America, which seems the most likely direction of dispersal. All the nodes within the clade have mean estimates that fall within the past three million years (fig. 9).
Psittacara
Psittacara comprises medium- to small-sized parakeets with long, attenuated tails and are mostly green. They are remarkable among parrots and indeed all birds for the highly irregular “scattered” distribution of red in their plumage, mostly about the head, tibial feathering (thighs), underwing coverts, and particularly their contour plumage. Some species also have yellow in the underwing. Attempts to delimit species using plumage color (e.g., Arndt, 2006), particularly the extent of red plumage on the head, are now understood to produce paraphyletic species.
Relationships among the species were poorly resolved but nonetheless suggest geographically cohesive clades with varying levels of support. There are three clades of note: (1) P. mitrata and P. frontatus, species distributed from Peru to Argentina that occur up to 3500 m; (2) P. wagleri, P. finschi, P. mitratus, P. erythrogenys, and P. leucophthalmus, a widely distributed clade east and west of the Amazon; and (3) P. chloropterus, P. euops, P. holochlorus, P. brevipes, P. rubritorquis, and P. strenuus of the Caribbean, Central America, and Mexico. Given the strong geographic adjacency of species in this clade, it is most likely that the extinct P. maugei, formerly of Puerto Rico, was closely related to P. chloropterus and P. euops of Hispaniola and Cuba, respectively. The whole radiation was dated to have occurred within 0.6–6.7 Mya (fig. 9). This includes multiple colonization events of Central America dated to 2.2 and 3.1 Mya and dispersal into the Caribbean at 1–5.1 Mya (fig. 9).
Given the number of poorly resolved nodes and nonmonophyly of traditionally recognized species, which we discuss in more detail below, Psittacara is a high priority for a more detailed study. Peters (1937) considered P. frontatus conspecific with P. wagleri, and Collar (1997) suggested that P. wagleri and P. mitrata may be conspecific. Both are inconsistent with the phylogeny (see also Collar et al., 2020). Psittacara frontatus was formerly recognized as a subspecies of P. wagleri and, based on our analyses, they are not closely related. Psittacara frontatus of the Pacific slope and central Andes of Peru and Bolivia and P. mitratus of the southern Andes, branch sequentially at the base of the Psittacara clade although their relationship had 78% ultrafast bootstrap support (fig. 9). Moreover, all nodes in the clade of South and Central American Psittacara (clade 2, above) had low support values.
The resolution of relationships within clade 3 containing taxa in Central America, Mexico, and Cuba/Hispaniola was better, but there were several key nodes that were not also strongly supported (fig. 9). The clade containing the four species of Psittacara (P. holochlorus, P. strenuus, P. rubritorquis, and P. brevipes) from Mexico and northern Central America had high support. Within this clade, only the placement of P. brevipes of Socorro Island was weakly supported as sister to P. strenuus of southern Mexico to Guatemala and P. rubritorquis of Central America. The clade also included P. holochlorus of Mexico (and extreme southeastern Texas where the origin of its populations is unclear; Uehling et al., 2019) as sister to the other three species. However, this relationship could not be confirmed because of low support for the placement of P. brevipes (fig. 9). Within clade 3, the Psittacara of Mexico and Central America are sister to the two Caribbean species, P. chloropterus of Hispaniola and P. euops of Cuba, but this relationship was poorly supported (UFBS = 61%).
Leptosittaca, Ognorhynchus, Thectocercus, Diopsittaca, and Guaruba
These five genera are all midsized parrots with long, attenuate tails. All are predominantly green or green and yellow with blue, yellow and red in localized parts of the plumage such as the bend of the wing, head and face, or underparts. They are thus a phenotypically disparate group.
The monotypic genera Thectocercus, Diopsittaca, Guaruba, Leptosittaca, and Ognorhynchus are recovered as a clade, which was sister to Psittacara. Relationships within the clade have biogeographical cohesion. The sister pair of Ognorhynchus and Leptosittaca are relatively restricted-range species endemic to midmontane parts of the northern to western Andes. Sister to them are the other three genera, which form a subclade, mainly in lowlands east of the Andes. Overall, the latter three are more widespread than Leptosittaca and Ognorhynchus, notwithstanding Guaruba being fragmented and rare. Each is as phenotypically unique as might be expected. We recommend no changes to generic taxonomy and in particular we reject synonymy of Leptosittaca within Aratinga as is occasionally proposed (e.g., Forshaw and Knight, 2010). Of the five, only Diopsittaca nobilis and Thectocercus acuticaudatus have described subspecies, but we know of no suggestions to recognize any of these at species rank. Although both species have subspecies with large disjunctions that warrant phylogeographic investigations. Maximum likelihood estimates of node ages in this clade range from 2.5–6.4 Mya (fig. 9).
Conuropsis
Conuropsis is extinct and comprises only one species, C. carolinensis, the only parrot known solely from North America since European settlement of the continent. It was predominantly green with a bold yellow neck and head but for its orange crown. We confirm the earlier discovery by Kirchman et al. (2012) that C. carolinensis has a close relationship to the Aratinga solstitialis complex. However, we placed it as sister to that group (fig. 9), along with what we recognize as Gymnopsittacus weddellii, a species not sampled in their study. We further agree with Kirchman et al. (2012) in finding that that clade is itself sister to the macaws. Our phylogenomic data further casts doubt that Conuropsis fratercula, a taxon described from a single humerus from a 16 Mya deposit in Nebraska (Wetmore, 1926), was closely related to C. carolinensis. Conuropsis fratercula is older than the phylogenomic molecular estimate of 7.1 Mya (3–10.1; fig. 9), the divergence of Conuropsis from the ancestor of Gymnopsittacus and Aratinga. This divergence date also encompasses the likely origin of Conuropsis in North America, which is well after the age of C. fratercula. Demographic modeling using whole genome data (Gelabert et al., 2020) and genome-wide markers of C. carolinensis (Smith et al., 2021) indicate that that there were not strong signatures of genetic decline prior to its extinction and that it exhibits the typical genetic signatures of population declines and expansion during the Pleistocene in eastern Nearctic birds.
Aratinga and Gymnopsittacus
Aratinga are small to medium-sized parakeets that are mostly green, but several species have extensive yellow and orange plumage. They have long, attenuated tails. We included the first genetic data for A. maculata, which was recently separated from the closely related A. solstitialis (Silveira et al., 2005; Nemésio et al., 2009). Phylogenomic results confirm previously proposed relationships from mtDNA (Ribas and Miyaki, 2004). That is, Aratinga auricapillus, A. jandaya, A. maculata, and A. solstitialis, hereafter the core A. solstitialis group, are closely similar in plumage and form a well-supported clade in the concatenated and species tree. As suspected, based on genetic distances in mtDNA within the core A. solstitialis group, their divergences from one another occurred in the past two million years. We further confirmed a sister species relationship between A. maculata and A. solstitialis, the two taxa sharing an estimated common ancestor 1 Mya (0.3–1.7; fig. 9).
Also reinforcing Ribas and Miyaki's (2004) mtDNA findings, we recovered a further species, A. nenday, as sister to the core A. solstitialis group. This species has often been placed in monotypic Nandayus, Bonaparte 1856. Its phenotypic distinctiveness is undeniable due to its black head, differently colored and sharply demarcated chest and abdominal plumage, and red confined to the tibial feathers. A close relationship between A. nenday and the core A. solstitialis group, however, was first proposed by Miranda-Ribeiro (1920) who placed them all in Nendayus [sic] Bonaparte, 1856. Most phenotypic similarities of nenday to the brightly colored core A. solstitialis group, possibly synapomorphic and thus consistent with a sister relationship, are not readily apparent in a typical museum specimen or a perched living individual. However, Silveira et al. (2005) outlined this similarity as follows: “The most remarkable characters shared by the members of the group [i.e., the core A. solstitialis group] are found in the wings and tail. Remiges are dorsally green, with middle and distal portions deep blue and black, respectively. They are black ventrally, and the greater wing coverts are mostly deep blue. Upper side of tail is mostly green with deep blue in the tip; underside is mostly black. Nandayus nenday (Nanday Parakeet) also shares the characters that otherwise diagnose the A. solstitialis group and probably belongs to the same clade” (Silveira et al., 2005). To these we can add a black bill.
The impetus to move nenday to Aratinga involved a further species, weddellii. That is, although nenday is sister to the core A. solstitialis group (Ribas and Miyaki, 2004; this study), weddellii is in turn that larger group's sister (see South American Checklist Committee [SACC] proposal 578). Retaining all these species in Aratinga is simple and expedient (e.g., Remsen et al., 2013); it even has biogeographic cohesion in that the group is distributed around the lowland periphery of most of Amazonia. It obviates the need to recognize three genera, which is necessitated if nenday, nested within the group, is assigned to monotypic Nandayus. However, we do recommend assigning weddellii to its own genus (see below), so the issue returns to one of whether to generically recognize the two sister lineages, i.e., nenday and the core A. solstitialis group. Like so many similar decisions involving sister lineages, this ultimately is arbitrary. How might we resolve it here?
The divergence between A. nenday and the core A. solstitialis group had a mean estimate of 3.2 Mya (1.1–4.8; fig. 9). This age was younger than the majority of intrageneric divergences in parrots (only the Diopsittaca and Guaruba split at 2.5 Mya [0.9–3.8] was younger), and it is comparable with that of intraspecific divergences frequently observed within Neotropical birds (Smith et al., 2017). We conclude that relative to the core A. solstitialis group, phenotypic differences of nenday have evolved rapidly while plumage similarities (Silveira et al., 2005) likely are synapomorphic. We advocate retention of nenday in Aratinga.
Further research might usefully focus on the evolution of plumage color in the core A. solstitialis group complex, which presumably uses blue structural color and psittacofulvins to produce the greens and yellows in plumage of the entire head (reviewed in Berg and Bennett, 2010), whereas nenday has a melanin-producing pathway in its dark crown. Mutations in the gene SLC45A2 in captive color morphs of Psittacula sensu lato parakeets were shown to lead to a loss of melanin and produce a yellow phenotype (Roy et al., 2024). Identifying the reverse molecular pathway, a gain in melanin production, would help clarify the taxonomic relevance of the dark crown in nenday. Qualitatively, evolution to melanin-producing plumage color was infrequent in parrots, but it has independently evolved multiple times outside of nenday.
The final species to be discussed in the genus is A. weddellii of western lowland Amazonia. In our analyses, it is strongly supported as sister to all other Aratinga and on a relatively long branch (fig. 9), corroborating previous work (Kirchman et al., 2012). The disparity in plumage and skin color and patterning between A. weddellii and all other Aratinga is striking. It is reflected in the deep split at 6.6 Mya (2.7–9.4; fig. 9), separating these two lineages. The estimated timing of this divergence is largely overlapping with the split of the extinct Conuropsis carolinensis from the whole subclade (fig. 9).
The combined phenotypic and phylogenetic distinctiveness of Aratinga weddellii strongly suggests the merit of its generic separation (cf. distinctiveness of Psephotellus varius relative to other Psephotellus spp.). A genus group name, Gymnopsittacus Miranda-Ribeiro, 1920, is available for A. weddellii. We advocate for and suggest the reintroduction of its status as a monotypic genus.
Miranda-Ribeiro (1920) designated Aratinga weddellii (see note below on orthography) as the type species and Eupsittula cactorum as what he termed a cotype species. Peters (1937) as first reviser fixed Conurus weddellii Deville, 1851, as the type species of Gymnopsittacus. Further, Miranda-Ribeiro (1920) included a third taxon, aeruginosa, in Gymnopsittacus but it has long since been recognized as a subspecies within the polytypic Eupsittula pertinax complex. Similarly, Eupsittula also holds E. cactorum.
It is beneficial to reiterate the diagnostic phenotypic traits of Gymnopsittacus as follows: rose-pink (côr de carne = color of flesh sensu Miranda-Ribeiro, 1920) colored nares and lateral gular skin along the base of the bill; frequently very pale iris, albeit possibly bicolored; large, broadly circular area of naked, pale periophthalmic skin that is proportionally larger relative to the eye than in closely related Aratinga sensu stricto but that resembles in its extent that of Thectocercus acuticaudata, for example; the highly variable dusky blue-gray coloring of the feathering of the head that results from individual feathers being brownish proximally and bluish gray distally. The overall color of the head is thus highly variable due, we posit, to combined factors of wear and perhaps individual variation; we know of no evidence that there are sexual differences in its intensity, although this may warrant study.
Lastly, we note that Miranda-Ribeiro (1920) consistently misspelled the epithet weddellii as weddelli. The orthography introduced by Deville (1851) was indeed the former, i.e., weddellii, having double d, double l, and double i.
Cyanopsitta
Cyanopsitta is a small blue macaw of northeastern Brazil. It is one of the rarest birds in the world and the subject of intense in situ and ex situ conservation management (Hammer and Watson, 2012). It was sister to a clade containing the majority of other macaws (except Anodorhynchus and Diopsittaca), diverging 8.1 Mya (3.5–11.4; fig. 9). Its phylogenetic position is stable across multilocus phylogenies (e.g., Tavares et al., 2006; Wright et al., 2008; Schirtzinger et al., 2012) and the concatenated and species trees.
Orthopsittaca
Orthopsittaca manilatus of Amazonian savannas and swamp forests is predominantly dark green but for its large patch of bare facial skin and maroon abdominal (belly) patch, thus resembling some of the more plainly plumaged species of Pyrrhura. It is diet-specialized to Mauritia palms within which it also nests (Forshaw and Knight, 2010). Despite its large geographic distribution across Amazonia, no subspecies have been described. After Cyanopsitta, it represents the next divergence in the macaws, sharing a common ancestor with Ara and Primolius, 7.4 Mya (3.1–10.4; fig. 9).
Lastly, here, we make a nomenclatural note about this species. Whitney's (1996) reinstatement of monotypic Orthopsittaca Ridgway, 1912, for this species has since become conventional. When Ridgway (1912) introduced Orthopsittaca, however, he cited Orthopsittaca manilata as the type species, implying that manilata was adjectival and feminine. Dickinson and Remsen (2013), in contrast, argued that usage should return to the masculine form manilatus of the species epithet, i.e., Orthopsittaca manilatus, noting that the name manilatus was invariable. We have been advised (see Acknowledgments) that in Boddaert's (1783: 52) naming of the bird as Psittacus manilatus, the specific epithet manilatus means “broad hand.” We are further advised that under Article 31.2.1 of ICZN (1999) Orthopsittaca manilatus would be the correct nomenclature. Patterns of usage also argue that despite a long familiarity of the feminine manilata when the species was in Ara and since Orthopsittaca was reinstated by Whitney (1996), the masculine manilatus has prevailed since Dickinson and Remsen (2013). A cursory “Catalog” search of the Biodiversity Heritage Library on 19 March 2024 returned 52 instances of Orthopsittaca manilata, only two of which were later than 2013, and eight instances of O. manilatus all of which were after 2013. Further, other global checklists (e.g., Clements et al., 2022; Gill et al., 2021, 2022, 2023, 2024) have used manilatus. Given the apparent trend toward the use of manilatus since 2013, we concede, albeit reluctantly, that usage of Orthopsittaca manilatus should remain.
Primolius
Primolius consists of three smaller, predominantly green macaws. Ara and Primolius are estimated to have shared a common ancestor 7.1 Mya (3.1–10), a depth of divergence comparable to other intergeneric splits in Arini (fig. 9). There was discordance between the phylogenomic concatenated and species trees, but relationships in both trees are highly supported. In the concatenated tree, P. couloni was sister to P. maracana and P. auricollis, whereas in the species tree, P. auricollis was sister to P. couloni and P. maracana. The topology from Provost et al. (2018) is concordant with the phylogenomic tree presented here. Despite the apparent lack of strong phenotypic patterns favoring one topology versus the other, both nodes have divergences estimated to have occurred within the Late Pliocene to Pleistocene. The case for the priority of Primolius over Propyrrhura as the correct genus for these three species has been made by Penhallurick (2001) and cf. Whitney (1996).
Ara
Ara contains the other group of extant large-bodied brightly colored macaws, and we note that at least one smaller species, A. tricolor, is extinct (Forshaw and Knight, 2017). Ara has a crown age of 6.4 Mya (2.8–9.2), and maximum likelihood divergence time estimates among species in the genus ranged from 1.7–6.4 Mya (fig. 9). Relationships within Ara differ between the concatenated and species tree and relative to previously inferred trees. All trees support three different groups, but relationships among and within them vary. One group comprises four species, the two large red (A. macao and A. chloropterus) and two large green macaws (A. militaris and A. ambiguus). A phylogenetic study on Ara using mitochondrial DNA with incomplete species-level placed the extinct A. tricolor as sister to the red and green macaws (Johansson et al., 2018). The second group is the closely related blue-and-yellow macaws (A. glaucogularis and A. ararauna), our data robustly affirming the species status of A. glaucogularis (see Ingels et al., 1981; Hesse and Duffield, 2000) and the phenotypically distinct A. rubrogenys, which was not included in previous phylogenetic studies. The third group consisted solely of A. severus. In the phylogenomic trees, A. severus was sister to all other Ara. The concatenated and species trees differed with respect to the monophyly of the large green and red macaws. Only the concatenated tree showed that the red (A. macao and A. chloropterus) and green macaws (A. militaris and A. ambiguus) were respectively monophyletic for each color group (fig. 9). Phylogeographic studies have found genetic structuring across the ranges of A. macao (Schmidt et al., 2020; Aardema et al., 2023), and A. militaris (Eberhard et al., 2015, Rivera-Ortíz et al., 2023), and no structuring in A. ambiguus. The critically endangered A. glaucogularis of Bolivia shows weak genetic differentiation between northern and southern populations (Campos et al., 2021).
Psittrichasidae
Psittrichas and Coracopsis
The family Psittrichasidae is phenotypically and biogeographically diverse but comprises a low number of species in two monotypic subfamilies Psittrichasinae for Psittrichas and Coracopseinae for Coracopsis and is consistently placed as closely related to Psittaculidae. Psittrichasidae exhibits an unusual transoceanic distribution, where the monotypic subfamily Psittrichasinae (Psittrichas fulgidus) of New Guinea was sister to small radiation in Coracopseinae (Coracopsis) found in Madagascar and adjacent islands. The approximately 7000 km gap across the Indian Ocean between the ranges of these lineages was also unusual in that the divergence date was 30.4 Mya (22.3–35.4) (figs. 1, 10), some 40 million years after proposed Gondwana vicariant events observed in other sister taxa that show this disjunct distribution (Chakrabarty et al., 2012). The historical biogeography that accompanied this astonishing pattern of phylogeny warrants comment. Selvatti et al. (2022) reject Schweizer et al.'s (2011) proposal of dispersal across exposed land in the southern Indian Ocean, the Kerguelen Plateau–Broken Ridge hypothesis. They instead provided geological data in favor of an alternative history involving dispersal through Indomalaya and then south across the now submerged microcontinent of Mauritia.
Psittrichas fulgidus has an elongate bill, is among the largest parrots, and is utterly unlike any other parrot in its distinctive black and red plumage and black facial skin (Forshaw and Knight, 2010). It occurs widely across montane forests in New Guinea and has no described subspecies.
Coracopsis comprises 2–4 grayish-brown species that occur on Madagascar, Seychelles, and Comoro Islands, one of which, C. vasa, has extraordinary reproductive biology reviewed by Ekstrom et al. (2007). The uncertainty in species limits reflects that some recognize the C. nigra subspecies sibilans and barklyi as species. In our study we included C. nigra, C. vasa, and C. barklyi and found discordance between the concatenated and species tree topologies. The species tree topology matches the mtDNA tree of Podsiadlowski et al. (Podsiadlowski et al., 2017) where C. barklyi was sister to C. nigra and C. vasa. The time-calibrated concatenated topology has C. nigra and C. barklyi diverging 4.5 Mya (1.7–6.5) and subsequently coalescing to a common ancestor with C. vasa 8.6 Mya (3.9–12.1; fig. 10).
Psittaculidae
Psittaculidae comprises five subfamilies (Psittaculinae, Psittacellinae, Platycercinae, Agapornithinae, Loriinae) that span the Old World. The high diversity of this radiation is reflected in the high number of recognized tribes and genera that are geographically cohesive. Psittaculinae (Polytelini, Psittaculini, Micropsittini) was sister to all other groups within Psittaculidae (fig. 1). Within the Psittaculinae, Micropsittini (pygmy parrots) was sister to Polytelini/Psittaculini (fig. 1). Next, Psittacellinae (tiger parrots) was sister to a clade containing Platycercinae, Loriinae, and Agapornithinae (fig. 1). The placement of Agapornithinae (lovebirds, hanging parrots, and Guaiabero) differed between phylogenomic trees. In the concatenated tree, Agapornithinae was sister to Loriinae (lorikeets, fig parrots, and Budgerigar) whereas in the species tree it grouped with Platycercinae, albeit with low support (LPP = 63%). Our results do not support monophyly of Mayr's (2008) Loricoloriinae, its key defining character of hypotarsal morphology likely indicating convergent evolution rather than relationships. For example, we confirm earlier findings (Joseph et al., 2011) that Micropsittini falls well outside the other taxa (Mayr, 2008) united in Loricoloriinae.
The supermatrix phylogeny of Provost et al. (2018) showed instability in the monophyly of Platycercinae, but phylogenomic data validated the group. The subfamilies have estimated crown ages dating from the late Oligocene to the early Miocene. The Platycercini and Pezoporini share a common ancestor estimated as the crown age of Platycercinae dated at 27.5 Mya (19.7–32.8; fig. 1).
Micropsitta
Micropsitta are the smallest parrots. They are largely green with modified, strengthened tail shafts and metatarsals having long claws, which are adaptations for foraging on tree trunks. The clade has a deep stem age of 27.6 Mya (19.8–32.6) and crown age of only 4.1 Mya (2.8–5.2) indicating that extinction was responsible for the long branch (figs. 1, 10). There was considerable discordance between the phylogenomic concatenated and species trees within Micropsitta. In the concatenated tree, M. bruijnii, which ranges from Moluccas, Indonesia, across all of New Guinea to the Solomon Islands, was sister to the other two clades. Although with low support in the concatenated tree, M. keiensis was sister to M. finschii and these two were sister to a clade in which M. geelvinkiana was in turn sister to M. meeki and M. pusio (fig. 10). In contrast, M. keiensis and then M. geelvinkiana on consecutive nodes, were sister to the remaining Micropsitta in the species tree with strong support. The discordant relationships in the species tree were likely due to gene tree estimation error caused by low quality samples (Smith et al., 2023).
Polytelini
Polytelini contains three genera (Alisterus, Aprosmictus, and Polytelis) of midsized parrots that occur in Indonesia, New Guinea, and Australia. Relationships among the genera are stable across the phylogenomic trees in this study and in previous work (Provost et al., 2018), with the exception of the placement of Polytelis alexandrae, which is discussed below. Aprosmictus and Polytelis share a common ancestor 8.6 Mya (5.3–11.3), and their ancestor split from Alisterus 9.6 Mya (6.1–12.5; fig. 10).
Alisterus
The king parrots (Alisterus) comprise three species that are largely red ventrally and green dorsally and that have long, broad tails. A streak of a lighter green in the upper wing coverts in two species is strongly UV-fluorescent (Nemesio, 2001). They occur in Australia, Indonesia, and New Guinea. Relationships among the species were consistent and highly supported across previous topologies (Provost et al., 2018) and the one presented here. At approximately 3.2 Mya (1.8–4.2) A. scapularis (Australia) diverged from the ancestor of A. amboinensis (Moluccas, western New Guinea) and A. chloropterus (remainder of New Guinea), which in turn diversified at 1.8 Mya (1.1–2.5; fig. 10).
Aprosmictus
Aprosmictus parrots are two predominantly green species with varying extents of red on the upper wing coverts. Aprosmictus erythropterus of Australia and New Guinea has a black back and, in males, the most extensive red in the upper wing coverts, relative to the other species A. jonquillaceus of Timor, Roti, and Wetar, Lesser Sunda Islands, Indonesia (Forshaw and Knight, 2010). Both species have two generally accepted but poorly differentiated subspecies. Our estimate of divergence time between A. erythropterus and A. jonquillaceus was 2.3 Mya (1.3–3.2; fig. 10).
Polytelis
Polytelis comprises three species endemic to Australia. They are long-tailed, slender-bodied parrots and have vocalizations that are as notably unique to the genus as they are similar among the three species themselves. Phenotypically disparate relative to each other, they share at least one plumage trait, a distinctively colored wing-covert patch. Notably, however, this patch is most easily visible in one species, P. swainsonii when safely viewed under a “black” light where ultraviolet pigments reflect as yellow (L.J., personal obs.). Similarly, P. swainsonii is sexually dimorphic in its coronal plumage but UV-fluorescent pigments are again only visible when safely viewed under a “black” light (L.J., personal obs.).
Previous multilocus phylogenetic studies analyzed only two of the three species, P. anthopeplus and P. alexandrae, and not P. swainsonii (Wright et al., 2008; Schweizer et al., 2010, 2011). Further, these earlier multilocus studies sampled only single individuals each of P. anthopeplus and P. alexandrae that, moreover, had been reared in captivity either in Europe or North America and presumably were descended from many generations bred in captivity. Wild-collected specimens of Polytelis alexandrae are rare in museum collections, and we know of no cryofrozen tissue samples of it from natural populations. Given that background and taxon sampling, we note that the earlier multilocus studies consistently and unexpectedly found that P. anthopeplus and P. alexandrae were not each other's closest relatives. In particular, P. alexandrae consistently was more closely related to Aprosmictus erythropterus than to P. anthopeplus (reviewed in Provost et al., 2018).
In an ongoing phylogenomic study of this unexpected result, we have sequenced all three species using multiple samples, whether toe pads of older museum specimens or cryofrozen tissue samples, from the natural ranges of all species, as well as captive-bred samples of P. alexandrae (note again that there are no cryofrozen tissue samples of wild P. alexandrae). We specifically probed the unexpected paraphyly of Polytelis. We wished to test, for example, whether it was due to the use in earlier work of captive birds in Europe and North America. Specifically, such birds could have been descended, albeit by many generations, from well-known hybridization between P. alexandrae and A. erythropterus in captivity, although there is no evidence of back-crossing after such hybridization (Forshaw, 2002; Sindel and Gill, 2003) during the 20th century. The concatenated tree found that P. alexandrae was sister to P. anthopeplus and P. swainsonii with 93% UFBS (fig. 10). In contrast, the species tree affirmed earlier multilocus work showing P. alexandrae as sister to Aprosmictus with 100% support. The divergence time estimate for P. alexandrae was approximately 7.9 Mya (4.7–11). This phylogenetic discordance could indicate several possibilities: ancient introgression occurred between between P. alexandrae and A. erythropterus (or their ancestral lineages); P. alexandrae is of hybrid origin; or P. alexandrae warrants full generic, not subgeneric, separation from Polytelis. Concerning the last alternative, we note that Spathopterus North, 1895, was erected solely for this species based on the spatu-late tip of one primary feather in males. Mathews (1912) noted, however, that the name is preoccupied by Spathoptera Audinet-Serville, 1835, in the Coleoptera (see Schodde and Mason, 1997). Northipsitta Mathews, 1912, is available for it if Spathoptera is truly unavailable. Currently, we do not favor its generic separation from Polytelis (pace Provost et al., 2018). We instead argue that ancient introgression within the past natural ranges of these birds, not in captive populations, is potentially an equally good explanation of the data bearing on this problem. We are planning a separate publication specifically to address this surprisingly intransigent conundrum in more detail and our approach to its resolution.
Psittaculini
Eclectus, Geoffroyus, Tanygnathus, and Psittinus are all genera with low species diversity that radiated across Indonesia, the Philippines, and New Guinea. Their plumage is often described as sleek. Much of the diversity is reflected in subspecies that are allopatrically distributed across islands. Generic-level relationships remain stable and more supported with phylogenomic data. The branching pattern starts with Eclectus then Geoffroyus then a clade containing Psittacula, Tanygnathus, and Psittinus.
The relationships within Psittaculini are mostly well resolved, but there is still scope for clarifying relationships within paraphyletic Psittacula sensu lato, and the placement of Psittinus varies between our tree and previous work (Braun et al., 2019). The 10 species of racquet-tailed parrots (Prioniturus) form a clade that was sister to all other members of Psittaculini. Phylogenomic relationships among the 10 species are largely congruent with an mtDNA tree (Schweizer et al., 2012). Within the sister clade of Prioniturus, Eclectus, then Geoffroyus, on successive branches were sister to a subclade containing Tanygnathus, Psittinus, Psittacula, and Prioniturus. All these relationships were highly supported and consistent among phylogenies.
Prioniturus
The racquet-tailed parrots (Prioniturus) comprise a small radiation in the Philippines and Indonesia. The species are largely green and stocky with rounded tails in which the central tail feathers have extended bare shafts protruding from the distal end of the tail and terminating with a racquet of feather barbs (Forshaw and Knight, 2010). An mtDNA phylogeny of Prioniturus inferred three separate clades (Schweizer et al., 2012): (1) mada and platurus; (2) waterstradti, montanus, platenae, and mindorensis; and (3) discurus, luconensis, verticalis, and flavicans). Our phylogenomic results are largely congruent but with some notable exceptions. We confirm the finding of Schweizer et al. (2012) that P. discurus is paraphyletic. All phylogenetic data showed that P. [discurus] mindorensis was highly supported as a member of clade 2, which does not include nominotypical P. d. discurus. The species recognition of P. mindorensis appears further warranted, although its phenotypic divergence is primarily in iris color (pale, not dark) and much reduced blue throughout the plumage. Within clade 2, the position of P. montanus differed between the mtDNA and phylogenomic tree. The phylogenomic tree had P. montanus as sister to P. waterstradti (UFBS = 100%), whereas the mtDNA tree places the taxon as sister to all other members of clade 2. The trees also differed with the placement of P. flavicans. In the phylogenomic tree, P. flavicans was sister to all other members of clade 3, whereas in the mtDNA tree P. flavicans was sister to only P. discurus and P. luconensis. The phylogenomic species tree places P. flavicans outside clade 3 as sister to clades 2 and 3. It is unclear to what extent the position of P. flavicans in the species tree was driven by a lower-quality sample. The P. flavicans sample was retained after more moderate data filtering, but in the most stringent approach the sample was dropped. The radiation of Prioniturus, which was estimated to have begun 8.4 Mya (4.9–11.2; fig. 10), reflects the complex relationships among islands within the Philippines and Indonesia (Brown et al., 2013).
Eclectus
Eclectus occurs from across New Guinea and its surrounding islands to the Solomon Islands in the east, far northeastern Australia in the south and Sumba in the Lesser Sundas in the west. Across its range it exhibits considerable geographic variation in size and plumage color while maintaining the oft-remarked exceptional sexual dichromatism of red and blue (female) and green (male) forms (Heinsohn et al., 2005). The phylogenetic placement of Eclectus has remained stable and the genus is unquestionably monophyletic. The remaining taxonomic questions deal with whether subspecies should be elevated to species. Using captive individuals and mtDNA, Braun et al. (2016) inferred phylogenetic relationships among six of the nine described subspecies. They found deep phylogeographic structure representing an Indonesian clade and a New Guinea/Solomon Islands clade. The available data indicate there are up to four phylogenetic species in the clade, but until genomic data from wild specimens is published, the recognition of a single species should be maintained.
Geoffroyus
Geoffroyus are midsized parrots that are largely green with contrastingly colorful head plumage in two species and a distinctive blue nuchal collar in one. They occur in Australia, Indonesia, Papua New Guinea, and the Solomon Islands. Traditionally, three species have been recognized: G. geoffroyi, G. simplex, and G. heteroclitus. Geoffroyus heteroclitus of the Bismarck Archipelago, Bougainville Island, and the Solomon Islands has a subspecies on Rennell Island (G. h. hyacinthinus) that differs in size and plumage color. Del Hoyo and Collar (2014) elevated it to species rank, but hyacinthinus was not included in our sampling. We found a deep divergence between G. geoffroyi and G. simplex/G. heteroclitus dating to 7.2 Mya (3.7–9.7; fig. 10). The subsequent split between G. simplex and G. heteroclitus was dated to 2.8 Mya (1.3–4; fig. 10). The underlying historical biogeography of Geoffroyus is unclear. We suggest that initial divergence between G. geoffroyi and the ancestor of G. simplex/G. heteroclitus likely occurred within New Guinea, and that subsequent dispersal out of New Guinea and into the Bismarck Archipelago and Solomon Islands could have been involved in the speciation of G. heteroclitus.
Psittinus, Psittacula, Himalayapsitta, and Palaeornis
Long treated in Psittacula sensu lato, the genera Psittacula, Himalayapsitta, and Palaeornis comprise medium- to large-sized long-tailed parrots that are predominantly green in plumage and often with colored or bicolored bills. They variously have distinctively colored heads and nuchal collars or, in the absence of a collar, distinctively colored facial areas and bold, black submalar stripes. Colored patches on the wing coverts vary in size from a small shoulder patch to large covering most of the greater wing coverts.
Nonmonophyly of Psittacula has been reported in several studies, which have all found Tanygnathus to be nested within it (Groombridge et al., 2004; Kundu et al., 2012; Jackson et al., 2015; Podsiadlowski et al., 2017). However, taxonomic recognition of Psittacula sensu lato has persisted despite its established nonmonophyly for two decades. Braun et al. (2019) proposed a major taxonomic revision of a subclade within Psittaculini (Tanygnathus, Psittinus, Psittacula, and Prioniturus) to recognize only monophyletic genera as follows († denoting extinct species): Himalayapsitta (himalayana, finschii, roseata, cyanocephala), Nicopsitta (columboides, calthrapae), Belocercus (longicauda), Psittacula (alexandri, derbiana), Palaeornis (†wardi, eupatria; see Hume, 2007, and Podsiadlowski et al., 2017, for placement of †Mascarinus Lesson, 1831, within this group), and Alexandrinus (krameri, †exsul, and eques including echo).
Psittinus
The mono- or ditypic genus Psittinus of continental Southeast Asia, Borneo, and Sumatra is morphologically divergent relative to Psittacula. Psittacula sensu lato are slender bodied, long-tailed birds perhaps recalling Polytelis, whereas Psittinus has a short, rounded tail and a stocky body recalling the “true parrots” of older literature such as Geoffroyus. Largely green, various populations have blue or black distributed in distinctive patches on the head and dorsal surface. The bill is bicolored, the maxilla being red and the mandible black. Traditionally treated as comprising one species with three subspecies, a recent trend has been to elevate P. c. abbotti, found on two islands of western Sumatra, to species rank on the basis of its larger size and different pattern of sexual dimorphism in plumage (del Hoyo et al., 2014; Eaton et al., 2016; Gill et al., 2023).
The phylogenomic trees show discordance between the concatenated and species tree topologies, but the trees certainly support the need for a taxonomic revision. A prominent difference between the Braun et al. (2019) tree and both the phylogenomic trees was in the placement of Psittinus. Braun et al. (2019) found that Psittinus was sister to a clade containing Tanygnathus, Psittacula longicauda, Psittacula alexandri, and Psittacula derbiana. In contrast, the phylogenomic tree showed that Psittinus was strongly supported as sister to the entire clade containing Psittacula/ Tanygnathus (fig. 10), and that pattern was also recovered in the species tree. These two basal lineages shared a common ancestor 10.5 Mya (6.2–13.7; fig. 10). The position of Psittinus does not impact the proposed generic revision of Braun et al. (2019), but it is important for understanding the evolutionary history of the clade.
For the remaining lineages, the phylogenomic tree favors fewer new monophyletic genera. It supports Himalayapsitta (roseata, cyanocephala, himalayana, and finschii) as represented by a clade stemming from the basal node in the entire Psittacula-Tanygnathus group 10 Mya (5.9–13; fig. 10). Braun et al. (2019) further proposed Nicopsitta, Palaeornis, and Alexandrinus to account for the nonmonophyly among the group more broadly. Our phylogenomic analyses place all the species that would be subsumed in the first two of these genera (calthrapae, columboides, wardi, eupatria) as well as eques in a clade for which we advocate recognition under the oldest available name, Palaeornis Vigors, 1825. The generic variation characterized by Braun et al. (2019) was reflected in the relatively deep crown age of Palaeornis at 8.2 Mya (4.8–10.8; fig. 10). Note that if the phylogenetic position of extinct Mascarinus mascarinus (Linnaeus, 1771) found by Podsiadlowski et al. (2017) is confirmed then this species would also be placed within Palaeornis, which is the older generic name. Braun et al. (2019) inferred longicauda with weak support to be sister species to derbiana/alexandri and accordingly suggested placing longicauda in a monotypic genus, Belocercus S. Muller, 1847. We have found support for that relationship among these species in the phylogenomic tree (UFBS = 94%; fig. 10). This includes caniceps, which was not sampled in Braun et al. (2019), as sister to longicauda in our tree. We dated the basal divergence of Psittacula sensu stricto (derbiana, alexandri, longicauda, caniceps) at 6.8 Mya (3.8–9.4; fig. 10). Accordingly, we favor retention of the two sister species pairs that comprise there four taxa, (longicauda, caniceps, alexandri, and derbiana) within Psittacula Cuvier, 1800 (type species is Psittacus alexandri Linnaeus, 1758; see Schodde et al., 2012; ICZN, 2014), rendering Belocercus synonymous with it. However, the species tree has longicauda and caniceps (the latter not sampled by Braun et al., 2019) as sister to Tanygnathus (LPP = 1.0). This relationship, if eventually confirmed, could be expressed by the adoption of Belocercus to avoid a paraphyletic Psittacula, but we do not recommend its adoption yet.
Braun et al. (2019) also found some cases of paraphyletic species, derbiana being nested within alexandri, and echo within krameri. Because the sampling of the entire group has relied on using captive birds, samples from wild birds will be required to verify these patterns. We do not yet advocate placing derbiana within alexandri but urge closer study.
Tanygnathus
Tanygnathus are predominantly green, round-tailed parrots with large or massive bills, and in most species, at least partly red. The four usually recognized species (see below) are distributed in the Philippines and Indonesia. The genus is nested deeply within Psittacula sensu lato, so the large bill and rounded tails of these birds are likely derived traits, but this depends on the position of Psittinus. Tanygnathus gramineus occurs only on Buru (Moluccas, Indonesia). It is monotypic and extremely rare in collections and we have been unable to sample it. The other species all show geographic variation expressed with up to six described subspecies.
Provost et al. (2018) and Braun et al. (2019) each inferred relationships for Tanygnathus, but each of those studies drew on samples of two species (T. sumatranus, T. megalorhynchos), whereas Arndt et al. (2019) and the present study included T. sumatranus. The best estimation of phylogenetic relationships among the other species based on our analyses and that of morphology (plumage, iris color) and mitochondrial DNA sequences (Arndt et al., 2019) was that T. megalorhynchos and T. lucionensis were sister species, and that T. sumatranus sensu lato was sister to both (fig. 10). This topology was not comparable to other trees. The position of T. lucionensis in the species tree was unreliable because the sample was of lower quality. Despite the lower diversity of Tanygnathus, the clade had a crown age of 6.0 Mya (3.4–7.9; fig. 10).
Arndt et al. (2019) argued on grounds of morphology (plumage, iris color) and cytochrome b sequences from mitochondrial DNA that Philippine populations of T. sumatranus should be elevated to species rank as T. everetti. We agree that further work is needed to accurately place T. e. burbidgii (Sulu Islands) and T. s. sangirensis (Talaud Islands) before species limits in T. sumatranus sensu lato are resolved.
Psittacella
The tiger parrots (Psittacella) are stocky, small to midsized with short round tails (Forshaw and Knight, 2010). The English name tiger-parrot is derived from the yellow-and-black barring in the chest (upper ventral) plumage in females of three of the four species. There are two small and two larger species, clearly forming two species pairs that were reflected in our phylogeny (fig. 11).
FIGURE 11.
Time-calibrated topology of Psittacellinae, Platycercini, Pezoporini, Neophemini, and Agapornithinae. Support values come from the maximum likelihood tree. Nodes have ultrafast bootstrap values of ≥95% otherwise noted. Gray bars represent divergence time ranges were estimated from 100 bootstrap trees.
The clade is on a deep branch, sharing a common ancestor with the clade containing the subfamilies Platycercinae, Loriinae, and Agapornithinae 20.6–33.5 Mya (fig. 1). Across this relatively long evolutionary period, there are only four species that diversified in the mountains of New Guinea. Phylogenetic relationships are stable and consistent across concatenated and species tree approaches in the phylogenomic data. Prior phylogenetic work only sampled P. brehmii and P. picta (Provost et al., 2018). Psittacella modesta and P. madaraszi are sister and have olive-brown heads with green backs. The other sister pair, P. brehmii and P. picta, have black barring on their backs. The crown age of Psittacella, 11.3 Mya (6.8–14.5; fig. 11), was among the oldest of all parrot genera, as are the species-level divergence time estimates. One subspecies of P. picta, P. p. lorentzi, has occasionally been elevated to species rank (Mayr, 1941; del Hoyo and Collar, 2014) based on plumage differences, Mayr (1941) only further noting that they “form a superspecies.” Gregory (2017) commented that they differ “significantly in plumage,” lorentzi seeming “as distinct from Painted [P. picta] as Modest [P. modesta] and Madarasz's Tiger-Parrots [P. madaraszi] are from each other.” More recently, Collar et al. (2020) imply that they will retain the two species arrangement, although at the time of our writing (11 May 2024) had not yet not employed it. We agree with Beehler and Pratt (2016) in stressing that this is premature until the nature of any contact where P. p. lorentzi and P. p. excelsa meet is explored, especially with genetic data. Beehler and Pratt (2016) noted that the likely area of contact, near the Strickland River gorge, is a barrier for a number of montane species. The details of this case, to reiterate, need further study.
We know of no proposal to separate the large- and small-bodied species pairs of Psittacella into two genera (fig. 12). Given their ages, this is not without merit. That differently sized but otherwise extraordinarily similar species with a sister-group relationship should be placed in different genera finds support elsewhere in Aves, e.g., the meliphagid genera Entomyzon and Melithreptus. In that case, there is no extant species of clearly intermediate size whereas in Psittacella the four species more smoothly intergrade in size between extremes. Further, subclades in Psittacella cannot be diagnosed by yellow-and-black barring in the chest of females because three out of four of the species have the character state. We refrain from breaking up Psittacella here primarily for those reasons.
Platycercinae Pezoporini
The Pezoporini are restricted to Australia. The phylogenomic trees confirm the results of Joseph et al. (2011) that Neopsephotus, Neophema, and Pezoporus form a clade, the Pezoporini, and that Pezoporus is monophyletic. Estimated divergence times among these three genera are deep. Neopsephotus and Neophema share a common ancestor 16.9 Mya (11.2–20.7), and this lineage coalesces with Pezoporus 23.8 Mya (16.7–28.5; figs. 1, 11). Both the concatenated and species trees inferred from phylogenomic data are topologically concordant for higher-level and the majority of species-level relationships within the Pezoporini. Our data affirm that Melopsittacus undulatus is not closely related to Pezoporini as had been thought prior to the advent of molecular studies. Lastly, here, we argue that Pezoporini warrants being broken into two tribes, monogeneric Pezoporini and a second tribe for Neophema and Neopsephotus. We name that tribe below after closer review of our data and the biology of these birds.
Pezoporus
Pezoporus contains three midsized and highly terrestrial species with varying tail lengths relative to body size. They are either nocturnal (P. occidentalis) or active at dawn and dusk (P. wallicus, P. flaviventris). They are essentially green, finely mottled black and yellow, having very few other differences in plumage (e.g., red frons absent in P. occidentalis but present in the other two; yellower belly in P. flaviventris relative to P. wallicus). Their plumage pattern is strikingly convergent with that of Strigops.
FIGURE 12.
Specimens from the American Museum of Natural History (AMNH) of the four traditionally recognized species of Psittacella to show the size difference between the sister pairs of larger (P. brehmii AMNH 293610 ♂, AMNH 339594 ♀; P. picta AMNH 419599 ♂, AMNH 622039 ♀) and smaller (P. modesta AMNH 339630 ♂, AMNH 622046 ♀; P. madaraszi AMNH 419061 ♂, AMNH 622057 ♀) species. Photographs: Sahid Robles Bello.
The estimated split of Pezoporus wallicus and P. flaviventris was 3.8 Mya (2.1–5.2; fig. 11), which is older but overlaps with the level divergence in mtDNA that was used to elevate the two to species rank (Murphy et al., 2011). The enigmatic and recently rediscovered P. occidentalis diverged from the ground parrots early in their history at 7.5 Mya (4.4–10). Shute et al. (2023) have reexamined the cranial osteology of the arid zone, nocturnal P. occidentalis. They summarize its osteological distinctiveness in the title of their paper as adaptive and also note it as an evolutionary trade-off between the need for species to have good hearing and vision in its largely nocturnal biology. Taxonomically, they raised the specter of reinstating monotypic Geopsittacus for P. occidentalis. As their work clearly suggests adaptive differentiation relative to the other two Pezoporus species, we do not advocate breaking up Pezoporus.
Neophema and Neopsephotus
The seven species long placed in Neophema are the smallest-bodied Platycercinae and have long tails. One group of four species (chrysostoma, elegans, petrophila, chrysogaster) for which the genus-group name Neonanodes is available, is predominantly yellow-green with species-specific patterns of shades of blue about the face, forehead and wing. A further pair of species (pulchella, splendida), one of which, pulchella, is the type species of Neophema, have distinctive blue faces and red in the wings or chest. Lastly, a seventh species, usually now placed in monotypic Neopsephotus, is pastel pink and blue. The clade, excluding Neopsephotus, began diversifying 6 Mya (3.4–8.1; fig. 11). We affirm the now well-established transferal of Neophema bourkii to monotypic Neopsephotus as sister to the six species we advocate leaving in Neophema. In addition to our genomic support for this, Neopsephotus bourkii is ecologically and phenotypically highly distinctive. Its plumage is uniquely pastel pink and blue, it has distinctive, musical vocalizations unlike the high-pitched tzeet like contact flight vocalizations made by the other six species, it inhabits arid zone Acacia scrubs, and is markedly crepuscular, i.e., noticeably active at dawn and dusk.
Among the six species remaining in Neophema, four are strictly or partially migratory, although the scale of movements varies. Neophema chrysostoma of southeastern Australia and the critically endangered N. chrysogaster are typical long-distance migrants crossing Bass Strait between Tasmania and the Australian mainland, N. chrysostoma migrating still further north deep into arid central Australia. As some of these movements occur at night, we reiterate the observation of Joseph et al. (2011) that crepuscular or nocturnal activity characterizes a number of species in Pezoporini as currently construed, suggesting its early origin in the tribe's history. Neophema elegans in the west of its range is partially migratory with at least a part of its population regularly migrating seasonally up to several hundred kilometers between the arid and temperate zones (Davis and Burbidge, 2008). Neophema elegans in eastern Australia moves locally with some suggestion of seasonality (Collar and Boesman, 2020) and N. petrophila certainly undergoes pronounced postbreeding dispersal and probably some seasonal movement (Baxter and Parker, 1981; Higgins, 1999). The sister species N. splendida and N. pulchella are in turn sister to the preceding four species and geographically replace each other in low woodlands in arid central-southern Australia and mesic southeastern Australia, respectively.
Wolters (1975) introduced Neophemini at tribal rank to accommodate the genera Neopsephotus and Neophema. He gave neither a description nor a reference to one as is required for new family-group names introduced after 1930. Therefore, his name is a nomen nudum and unavailable. Our colleagues R. Schodde and I. Mason have independently concluded that there is merit in recognizing Neophema and Neopsephotus at tribal or subtribal rank. We here join with them in introducing a family-group name to accommodate the genera Neopsephotus and Neophema either at tribal rank or as a subtribe within Pezoporini. We introduce a new family-group name at the rank of tribe as:
Neophemini Schodde, Mason, Smith, Thom,
and Joseph, 2024, new tribe
ZooBank registration: urn:lsid:zoobank. org:act:7156B56E-42BF-4410-AAB4-2C3A84628231
Diagnosis: Small, usually plain-plumaged parrots without transverse barring or black chevroning, e.g., dorsal feathers in Pezoporus are transversely barred black and yellow in their centers but broadly fringed with green, whereas in Neophema and Neopsephotus the dorsal feathers are plain green or dusky brown, respectively; unmarked outer rectrices with extensive yellow (or white) tips and dark bases; remiges with white to yellowish white bar on inner vanes evanescent, earliest in adult males, hardly extending on to outer vanes and usually obscure on upper surface of wing; pale “submissive” spot present in nape down; underwing coverts rich dark blue, often brighter on bend of wing; hypotarsal flexor tendon arrangement approaching ancestral condition: deep flexor tendons enclosed in separate bony canals, superficial tendon musculus flexor perforatus digiti II enclosed in a separate canal, and remaining superficial tendons external and separated in two separated shallow grooves on plantar wall; usually colonial, in small groups; foraging on ground and low shrubs and nesting in tree hollows (scrapes under rocks in Neophema petrophila); nestling begging call a husky quavering note that increases in time-frequency with age developing into bursts of brief whistles. In accordance with Article 13.1.2, we note that gross phenotypic differences have been well-illustrated and described on many occasions in the literature and we cite as examples reference works on the world's parrots (e.g., Forshaw and Knight, 2010; Juniper and Parr, 1998). This tribe comprises the genera Neophema and Neopsephotus.
Type genus: Neophema Salvadori, 1891, Catalogue of the Birds in the British Museum 20: 539, 569, by original designation. Type species: Psittacus pulchellus Shaw, 1792 = Nephema (Neophema) pulchella (Shaw, 1792).
Taxonomic position of the tribe: Subfamily Platycercinae in the Family Psittaculidae.
Component genera: Neopsephotus and Neophema.
Neopsephotus Mathews, 1912. Novitates Zoologicae 18: 279 – type species: Euphema bourkii Gould, 1841 = Neopsephotus bourkii (Gould, 1841), by monotypy.
Neophema Salvadori, 1891. Catalogue of the Birds in the British Museum 20: 539, 569 – type species: Psittacus pulchellus Shaw, 1792 = Neophema (Neophema) pulchella (Shaw, 1792), by original designation.
subgenus Neophema
subgenus Neonanodes Mathews, 1912. Novitates Zoologicae 18: 279 – type species: Psittacus chrysogaster Latham, 1790 = Neophema (Neonanodes) chrysogaster (Latham, 1790), by monotypy.
Geographical distribution: Australia and some of its offshore islands including Tasmania.
Platycercini
The broad-tailed parrots (Platycercini) consist of nine genera and 38 extant species distributed across Australasia. The clade includes the radiation of rosellas (Platycercus) in Australia and Cyanoramphus (New Zealand and islands of the South Pacific). The seven other genera contain up to a maximum of four species. Four of these genera are monotypic, i.e., Purpureicephalus, Lathamus, Psephotus, and Clarkona, the last-named here reinstated (see below). There are also four species in Cyanoramphus that went extinct upon European colonization of their islands, and Psephotellus pulcherrimus of Australia is extinct. There are two well-supported subclades in Platycercini that are repeatedly inferred across different studies, datasets, and phylogenetic methods. Intergeneric relationships are also resolved and stable. One clade comprises Barnardius, Platycercus, Purpureicephalus, Psephotus, Psephotellus, Clarkona, and Northiella (figs. 1, 11). Barnardius and Platycercus are closely related, which reflects some authors uniting them in Platycercus (e.g., Johnstone and Storr, 1998), but they diverged at 10.5 Mya (4.7–14.7) and we support retention of both (figs. 1, 11). The nonmonophyly of Psephotus sensu lato (Joseph et al., 2011) was corroborated with phylogenomic data, the type species Psephotus haematonotus not sharing a most recent common ancestor with the four species comprising Clarkona and Psephotellus. Phylogenetically between these genera are ditypic Northiella and monotypic Purpureicephalus. The other major subclade within the Platycercini (Lathamus, Prosopeia, Eunymphicus, Cyanoramphus) is more widely distributed, ranging from Australia, New Caledonia, Fiji, New Zealand, and various small islands in the South Pacific. As with previous phylogenetic work, the migratory and critically endangered Lathamus discolor was sister to Prosopeia, Eunymphicus, and Cyanoramphus (figs. 1, 11). In turn Prosopeia, which is restricted to Fiji, was sister to Eunymphicus of New Caledonia and the Loyalty Islands and Cyanoramphus. This suggests a key role for a propensity for long-distance movement in the evolution of this clade (see Joseph et al., 2011).
Our concatenated tree was concordant with previous phylogenies (Boon et al., 2008; Joseph et al., 2011) showing that Eunymphicus and Cyanoramphus are sister lineages, and that Prosopeia was sister to both. Some of the species trees had Cyanoramphus and Prosopeia as sister with strong support. However, interpreting the robustness of this alternative relationship will require new genetic data because the Eunymphicus samples were dropped in the most stringent data-retention filter.
Lathamus
The critically endangered Lathamus discolor is largely green and small bodied. In its nectarivory and associated aspects of morphology, such as slender, pointed wings and tail, and crude brush tongue, it is convergent with the lories and lorikeets (Loriini; see Gartrell et al., 2000; Kennedy and Tzaros, 2005). It is a seasonal migrant that breeds in Tasmania and winters in mainland southeastern Australia. As with the other monotypic genera in broad-tailed parrots, it is on a long branch that diverged 10.6 Mya (4.4–15) from a clade that radiated across New Zealand and nearby islands (Cyanoramphus), New Caledonia (Eunymphicus), and Fiji (Prosopeia) (fig. 11).
Prosopeia
Prosopeia (shining parrots) encompasses three species that are endemic to Fiji. They are large parrots dorsally green and ventrally red to yellow. Prosopeia likely colonized the archipelago at around 9.3 Mya (3.7–13.4), as this was the timeframe of its divergence from its common ancestor with Eunymphicus and Cyanoramphus (fig. 11). Prosopeia personata of Viti Levu diverged from P. tabuensis and P. splendens 5.9 Mya (2.3–8.9). Some authors (Forshaw and Knight, 2010) consider P. splendens of Kadavu, Ono, Viti Levu, a subspecies of P. tabuensis of Vanua Levu, Kioa, Koro, and Gau. Our time-calibrated tree has the two species diverging at 2.8 Mya (1–4.4; fig. 11).
Eunymphicus
The two species of Eunymphicus are midsized green parrots that have a black mask and crest, red and yellow crowns and long tails. They occur in the New Caledonian archipelago. Boon et al. (2008) found genetic, morphological, and behavioral data supporting recognition of two species, E. cornutus of Grande Terre (New Caledonia's main island) and E. uvaeensis of Ouvéa in the Loyalty Islands. Our time-calibrated tree dates this split at 3.2 Mya (1–5.1; fig. 11).
Cyanoramphus
Cyanoramphus are small to midsized, largely green with long tails that radiated in New Zealand and surrounding islands of the South Pacific as well as eastward to the remote archipelago of the Society Islands. Cyanoramphus was estimated to have diverged from Eunymphicus of New Caledonia and the Loyalty Islands 8.2 Mya (3.2–12) and diversification within the group was estimated to have begun 5.5 Mya (2–8.2; fig. 11). Systematic relationships with Cyanoramphus have been and still are hampered by incomplete taxon and genetic sampling, a reminder that we have not sampled the extinct Cyanoramphus ulietanus and C. zealandicus. Comparison of the phylogenomic tree with mtDNA data (Rawlence et al., 2015) was limited because of the low support in our concatenated tree. These conditions are analogous to that of Pyrrhura—a rapid radiation and degraded DNA from historical specimens. The phylogenomic results confirm that classifying taxa together based on color of head plumage was not consistent with the phylogeny, an observation that was used to elevate several taxa to species rank. All trees agree that C. forbesi and C. auriceps are not sister species. Cyanoramphus forbesi was previously classified as a subspecies with C. auriceps because both have a red frontal band and yellow crown. The phylogenomic tree has C. forbesi as sister to the endangered C. cooki, of Norfolk Island, albeit with low support (UFBS = 51%). The mtDNA tree has C. cooki as sister to all other Cyanoramphus, excluding C. forbesi. Regardless, if C. cooki and C. forbesi were sister or on successive nodes, then they shared a common ancestor 4.7 Mya (1.7–7.3). The mtDNA tree and our phylogenomic trees agree that C. saisseti was sister to all other Cyanoramphus whereas species tree had the more plainly patterned C. unicolor in this position. Cyanoramphus hochstetteri, which was formerly a subspecies of C. novaezelandiae, was most closely related to C. malherbi and C. unicolor in the mtDNA and the concatenated phylogenomic tree (fig. 11).
Platycercus
Rosellas (Platycercus) are small to midsized parrots that have radiated entirely within Australia including Tasmania. They are among the most easily diagnosable of any parrot genus on plumage by their uniquely mottled upperparts and prominent mono- or bicolored cheek patches. The phylogenomic trees parallel the results from prior multilocus data in supporting two groups sometimes recognized as subgenera (Shipham et al., 2015). The nominotypical subgenus Platycercus comprises blue-cheeked species Platycercus caledonicus of Tasmania and islands of Bass Strait, and the P. elegans complex of mainland southeastern Australia (the blue-cheeked crimson rosella complex); those sometimes placed in subgenus Violania comprises species having white, bicolored blue and white, or yellow cheeks (icterotis, eximius, adscitus, and venustus). Both a multilocus (Shipham et al., 2015) and the phylogenomic tree presented here showed the placement of P. icterotis of southwestern Western Australia as unresolved or poorly supported as sister to the rest of Violania. Our analyses accord with those of Shipham et al. (2015, 2017) who showed that mtDNA of mainland P. eximius has been captured by that of nonsister species P. adscitus, which occurs to its north and with which it has a complex hybrid zone (Shipham et al., 2019); the mtDNA of P. eximius diemensis of Tasmania is “true” P. eximius mtDNA.
Notably, the P. elegans group has long been a focus of work in speciation and systematics particularly after it was proposed by Cain (1955) as an example of a ring species. Our sampling was not designed to address these issues, but genomic testing of earlier work by Joseph et al. (2008) rejecting the ring species hypothesis is long overdue.
Barnardius
The Australian ringneck parrots, Barnardius spp., are the sister group of Platycercus rosellas. The two genera are each phenotypically distinct and easily recognizable. Mostly, Barnardius are arid and semiarid zone replacements of Platycercus, with notable exceptions of broad sympatry or parapatry in southwestern Australia, the Flinders Ranges, and in and near riparian zones of some inland river systems in eastern Australia. Our molecular dating indicates divergence within Barnardius beginning at 0.9 Mya (0.4–1.4; fig. 11) was consistent with the 1.72% divergence between two mtDNA-based clades (Joseph and Wilke, 2006).
Until Schodde and Mason's (1997) review recognizing one species, most texts treated two species: Barnardius zonarius having subspecies B. z. zonarius and B. z. semitorquatus; and Barnardius barnardi more controversially having two subspecies, B. b. barnardi and B. b. macgillivrayi. Ford and Parker (1974) had noted that the phenotypically more complex plumage pattern of barnardi set it apart from the simpler patterned macgillivrayi, which they argued to be closer to the similarly patterned zonarius and semitorquatus. This was clearly supported in mtDNA phylogeographic data, which showed two clades, 1 and 2, having 1.72% net divergence between them (Joseph and Wilke, 2006). Clade 1 comprised samples of the barnardi phenotype, and clade 2 included all four subspecies (zonarius, semitorquatus, macgillivrayi, and barnardi and their phenotypic inter-grades). Mitochondrial haplotypes in clade 2 were weakly differentiated and unstructured among the phenotypic groups. The lack of reciprocally monophyletic taxa in mtDNA data of Barnardius and the multiple zones of contact including between nonsister taxa barnardi and macgillivrayi can now be reasonably attributed to post-isolation gene flow.
Given that phenotypic signals of introgression of zonarius into barnardi are known to extend well east of their primary zone of contact (Schodde and Mason, 1997; Joseph and Wilke, 2006), we chose a sample of barnardi as far as possible from zones of its contact with zonarius and macgillivrayi and from where it shows minimal phenotypic signal of introgression from other taxa. Allowing that we were examining the world's parrots and not focusing in detail on any one species complex, we hoped this would be adequate. However, mtDNA extracted from UCE reads (not reported) showed that barnardi was nested within zonarius, a result inconsistent with Joseph and Wilke (2006). We must allow for several possibilities, assuming that the shallow mtDNA divergences in our samples are not due to contamination or a laboratory mix-up. One is that genetic introgression extends even further than we had anticipated and that our sampling has been obviously inadequate to place any introgression into full context. Alternatively, there is a puzzling discrepancy between large and small mitochondrial and nuclear DNA divergences, respectively, within Barnardius. For now, we reiterate Schodde and Mason's (1997) call for a full study of introgression across Barnardius and, of course, that it be a genomic assessment. This should be done before disrupting the current, if less than ideal, taxonomic practice of recognizing one species. We predict that the species-level taxonomy will eventually settle on the merit of recognizing monotypic B. barnardi and polytypic B. zonarius, however.
Psephotus
Psephotus haematonotus is a small-sized parrot with an attenuated tail, and it occurs broadly in southeastern Australia from the Lake Eyre Basin deep in the continent's interior south and east to the coasts. It diverged at 11.1 Mya (5.2–15.3; fig. 11). Its sexual dimorphism generally recalls that of Clarkona in that males are extensively green and females are grayish olive, but details of plumage patterning differ profoundly. Psephotus haematonotus has unique, whistled vocalizations. Although previously thought to be the eastern member of a vicariant sister pair with Clarkona varia (e.g., Ford, 1974) clarification of their phylogenetic positions dispelled this.
Northiella
Northiella are small-bodied parrots, with attenuated tails, but uniquely grayish-olive plumage punctuated by distinctively yellow or red wing-covert patches, blue facial feathering, (two-toned blue in N. narethae) and their prominent red or yellow abdominal (belly) patches. Their harsh contact calls recall those of Platycercus rosellas, but they also have a unique whistled vocalization. Phylogeographic work on what had been construed as one species, N. haematogaster, revealed genetic structuring to the extent of 2.16% mtDNA sequence divergence between what are now recognized as two species (Dolman and Joseph, 2015), N. haematogaster and N. narethae. In comparison to the temporal estimate of their divergence derived by converting the mtDNA distance with a standard 2% molecular clock, our phylogenomic estimate of divergence between these taxa is about twice as old at 2 Mya (0.8–3.1; fig. 11).
Purpureicephalus
Purpureicephalus spurius is a midsized, extraordinarily colored parrot. It has a strongly demarcated maroon-red crown, greenish yellow face, purplish underparts, yellow rump, green back and a long, attenuated blue tail. It has an unusually long maxilla with which it can extract seeds from the fruits of a cooccurring eucalypt, the marri Corymbia calophylla. It occurs in forest and woodland in southwest Australia and within this area shows no geographic variation. Purpureicephalus lies on a long branch diverging from the clade containing Psephotellus and Clarkona 9.8 Mya (4.6–13.6; figs. 1, 11). Its continued recognition as a monotypic genus is warranted on genomic, phenotypic, and biogeographical grounds. Notably among platycercine parrots, it is broadly sympatric with the largest and smallest of the Barnardius and Platycercus parrots, respectively. Any close affinity with Platycercus adscitus due to its unusually purplish blue ventral coloring has long been dispelled as has Cain's (1955) suggestion that Purpureicephalus and Eunymphicus should be synonymized.
Psephotellus and Clarkona
Psephotellus comprises a small radiation (here treated as three species) of midsized parrots endemic to Australia. Its three species (Psephotellus dissimilis, P. chrysopterygius, and the now extinct P. pulcherrimus) are almost unique in the entire Order Psittaciformes in nesting in terrestrial termite mounds in cavities dug out by the birds themselves. This well-supported clade has a crown age of 6.6 Mya (3.1–9.5; fig. 11). A fourth species, varia, was sister to the other three but on a long branch, having diverged from the other three species at (4–12.2 Mya; fig. 11). This species has usually been placed in Psephotellus. Unlike the three tropical or subtropical Psephotellus species, this species is an arid and semiarid zone species that nests unremarkably in tree hollows. Further, it is phenotypically unlike the other three in having, for example, uniquely patterned head plumage and distinctive vocalizations. On these collective grounds, we advocate recognition of Clarkona Mathews, 1917, as a genus for the species usually referred to as Psephotellus varius (cf. approach to recognizing Gymnopsittacus weddellii above), but which will become Clarkona varia. Clarkona was clearly and consistently treated as feminine by Mathews (1917a) and by CSIRO (1969). For example, Mathews (1917a) referred to the subspecies he recognized in this species as follows: “Psephotus varius varius or Clarkona varia varia,” or “Psephotus varius exsul or Clarkona varia exsul,” and so on.
Discordances in patterns of plumage color and phylogeny among the three species retained in Psephotellus are notable. The two species of Australia's tropics, P. dissimilis of the Northern Territory and P. chrysopterygius of Cape York Peninsula, are extremely similar in having aqua-green underparts and prominent yellow patches on the wing coverts. They occur either side of a well-established biogeographic barrier in northeastern Australia, the Carpentarian Barrier, which can explain the distribution of many similarly distributed pairs. Unexpectedly, therefore, our results and an earlier study based on mitochondrial DNA sequence data (Christidis and Norman, 1996) found that they are not sister taxa. Instead, the two eastern Australian species, P. chrysopterygius and the now extinct P. pulcherrimus, which had a red wing-covert patch and underparts of a similarly unique, if slightly different shade, of aqua-green, have emerged as sisters. Necessarily, both studies used DNA from historical specimens for the extinct P. pulcherrimus. Our study did likewise for the highly endangered P. chrysopterygius, that of P. dissimilis, with a sample derived from fresh tissue. We are currently sequencing DNA from a fresh P. chrysopterygius sample that has since become available to test this unexpected relationship.
If the pattern of relationships recovered in our study and by Christidis and Norman (1996) is confirmed, then it would imply that P. dissimilis diverged first from the common ancestor of P. chrysopterygius and P. pulcherrimus through vicariance across the Carpentarian Barrier. Next, divergence within eastern Australia led to separation of P. pulcherrimus and P. chrysopterygius (see Bryant and Krosch, 2016, for relevant biogeography). Psephotellus pulcherrimus was extinct by 1930, but genomic analyses show no evidence of a bottleneck prior to its extinction (Irestedt et al., 2019). Given the discordance just noted between color of wing coverts and relationships, it is notable that the reduced wing-covert patch is sexually dimorphic in C. varia, yellow in males and red in females and younger males.
Lastly, we note that the plumage patterning of adult male P. dissimilis and P. chrysopterygius is, at least within the constraints of natural selection, broadly similar to that of several species of African Poicephalus most especially Poicephalus meyeri (fig. 13) in showing dark upperparts, prominent yellow about the greater wing coverts, and a similarly unusual shade of green on the underparts. Study of the drivers of this convergence in the high-intensity light, mostly tropical savannas that these species inhabit would surely be rewarding.
Agapornithinae
Agapornithinae are small green parrots with short round or wedge-shaped tails. The clade consists of the monotypic Bolbopsittacus lunulatus of the Philippines, the hanging parrots (Loriculus), which occur from India to New Guinea, and the lovebirds (Agapornis) of the Afrotropics. Agapornis and Loriculus share the behavior of carrying nesting material in their feathers (Forshaw, 1973). Relationships among the three genera are stable. Bolbopsittacus diverged from the common ancestor of Loriculus and Agapornis at 24.4 Mya (17.4–28.8), and Loriculus and Agapornis then split at 23.1 Mya (16.4–27.3; figs. 1, 11) We advocate recognition of this phylogenetic depth and structure at least through recognition of a tribe for Bolbopsittacus and we note that two family-group names are available with which to recognize two other currently monogeneric tribes for Loriculus and Agapornis. In accordance with articles 13.1 and 13.2 of the Code (ICZN, 1999) we therefore introduce a new family-group name at the rank of tribe for the genus Bolbopsittacus:
Bolbopsittacini, new tribe
ZooBank registration: urn:lsid:zoobank. org:act:6CB65343-DA33-4016-BE31-B8348E1813A8
Diagnosis: In accordance with Article 13.1.1, we note that diagnostic traits of Bolbopsittacini—the same as those of its sole genus and species, the Guaiabero Bolbopsittacus lunulatus—include the combination of the following: a small (∼15 cm) parrot with a very short tail; proportionally broad, bicolored bill that is black distally and bluish gray proximally; predominantly green in plumage but with light blue around the face and a sexually dimorphic collar uniquely in parrots located across the mantle, which in males is blue concolorous with or slightly darker than blue of the face and in females is yellow with a black terminal band; yellow rump edged with black in females of B. l. mindanensis. We note convergent similarities in plumage with Cyclopsitta diophthalma, especially in the tone of blue about the face. In accordance with Article 13.1.2, we note that these differences have been well illustrated and described on many occasions in the literature and we cite as examples reference works on the world's parrots (e.g., Forshaw and Knight, 2010; Juniper and Parr, 1998).
Type-genus Bolbopsittacus Salvadori, 1891, Catalogue of the Birds in the British Museum 20: 388, 503 – type species: Psittacus lunulatus Scopoli, 1786 = Bolbopsittacus lunulatus (Salvadori, 1891), by original designation.
Taxonomic position of tribe: Subfamily Agapornithinae of Family Psittaculidae.
Component genera: Monospecific including only Bolbopsittacus lunulatus (Scopoli, 1786).
Geographic distribution: Philippines.
FIGURE 13.
Photographs (not to scale) of an adult male Psephotellus dissimilis (left photo: David Cook) of Australia and Poicephalus meyeri (right photo: John Missing) of Africa and showing broadly similar plumage patterning that includes most unusual but similar shades of green on the underparts, dark brownish upperparts and a prominent patch of yellow about the wing coverts and bend of the wing. See text for discussion. Photographs reproduced with permission from the photographers.
Bolbopsittacus
Bolbopsittacus lunulatus is a small green parrot with pale blue facial markings and is restricted to the Philippines. Our sampling has not addressed divergence within the species, which is currently divided into three subspecies (callainipictus, intermedius, and mindanensis). None of several recent studies of cryptic diversity in Philippine birds have addressed this species either (e.g., Peterson, 2006; Lohman et al., 2010; Hosner et al., 2014, 2018). Its phylogenetic placement as sister to Loriculus and Agapornis was well supported and stable across the phylogenomic trees and previous phylogenetic hypotheses for parrots (e.g., Wright et al., 2008).
Loriculus
Hanging parrots (Loriculus) are so named from their oft-quoted but rarely reported behavior of roosting while hanging by the feet from a branch like a pteropodid bat. The behavior has been described, however, for L. galgulus in captivity (e.g., Griffiths, 1947; Buckley, 1968). We would welcome modern data on this behavior in the wild, its consistency, or whether the birds have evolved responses to any physiological challenges caused by roosting upside down. Three out of up to 15 species recognized were not sampled in this study, namely L. camiguinensis of Camigun Sur, southern Philippines (Tello et al., 2006), L. flosculus of Flores Island, Indonesia, and L. bonapartei, also part of the L. philippensis–L. camiguinensis complex and debat-ably elevated to species rank (e.g., Collar and Kirwan, 2023). Forshaw and Knight (2010) treated the relatively recently described L. camiguinensis as a subspecies of L. philippensis; certainly the taxon is diagnosable by traits of plumage and size. Two subclades within Loriculus diverged 8.7 Mya (4.7–12.2; fig. 11), concordant with variation in bill color but not the presence of a red throat patch. Clade 1 is defined by taxa having black bills and contains L. tener and L. aurantiifrons of New Guinea and surrounding islands and the Bismarck Archipelago, and L. catamene, L. sclateri, L. stigmatus, and L. amabilis of Moluccas and Sulawesi and neighboring islands (fig. 11). Clade 2, which is defined by taxa with red bills, has a deep divergence whereby L. philippensis was sister to the other taxa in the clade (L. pusillus, L. vernalis, L. galgulus, L. exilis, and L. beryllinus) from which it diverged at 7.2 Mya (3.7–10.4), although L. philippensis bonapartei notably has a black bill (Forshaw and Knight, 2010; fig. 11). Within clade 2 L. pusillus of Java and Bali, Indonesia diverged from two sister pairs. One pair contains the widely distributed L. vernalis occurring in India through Southeast Asia and neighboring archipelagoes and its geographic replacement, L. galgulus of the Malay Peninsula, Sumatra, Borneo, and surrounding islands. The other sister pair is widely disjunct geographically and comprises L. beryllinus endemic to Sri Lanka and L. exilis of Sulawesi. The estimated divergence time between L. beryllinus and L. exilis was 1.3 Mya (0.5–2.2; fig. 11), indicating that this complex biogeographic pattern is a relatively recent pattern.
Agapornis
The lovebirds (Agapornis) are small, short-tailed, stocky parrots, mostly with colored bills and species-specific color patterns of the head and chest. The nine species of lovebirds have radiated within sub-Saharan Africa and Madagascar. Pioneering analyses of their evolution were based on morphology and environmental data by Moreau (1948) and courtship and nesting behavior (Dilger, 1960, 1962, 1964). Later, Racheli (1999) and Huynh et al. (2023), respectively, presented a cladistic morphological analysis and a phylogenomic analysis based on low-coverage whole-genome sequencing. The phylogenomic concatenated tree has high support (>95% UFBS) for all nodes, and the species tree had only one node (A. nigrigenis and A. roseicollis) with lower support (LPP = 0.88). The phylogenomic trees were discordant with the placement of A. taranta and A. pullarius, but this discordance was attributable to data quality. In analyses with more stringent filters for retaining samples, A. pullarius was excluded and the position of A. taranta became stable. In the concatenated tree A. canus of Madagascar then A. swindernianus were sisters to the other lovebirds (fig. 11). The next divergence was between a clade containing A. pullarius and A. swindernianus of western and central Africa from the five remaining species. The final relationships in the group had A. roseicollis of southwestern Africa as sister to two species pairs: A. nigrigenis/A. roseicollis and A. personatus/A. fischeri. This clade was also characterized by nest building in cavities: A. roseicollis builds a cup-shaped nest whereas the other four species build domed nests (Huynh et al., 2023).
The topology of Provost et al. (2018) is concordant with the phylogenomic tree for nodes with common sampling except A. personatus and A. fischeri were found not to be sisters. The whole genome phylogeny Huynh et al. (2023) also recovered the same topology as our concatenated tree except it had lower support for nodes lilianae/ nigrigenis and personatus/fischeri. Maximum likelihood estimates of species-level divergences cover a large timeframe, ranging from 0.7 Mya to 10.8 Mya (fig. 11). We advocate retention of one genus, Agapornis, for all of these species.
Loriinae
Relationships within the Loriinae are resolved and stable. The speciose lorikeets and lories (Loriini) and monotypic Budgerigar (Melopsittacini) are sisters and split 17 Mya (8.6–22.2). Cyclopsittini diverged from the ancestor of Loriini and Melopsittacini 19.3 Mya (10.9–24.8; figs. 1, 14).
Suavipsitta, Cyclopsitta, and Psittaculirostris
The fig parrots are among the smallest of all parrots. Predominantly green (one species has orange underparts), they have richly colored and diversely patterned facial plumages characterized by shades of red and blue, sometimes black, and which are often sexually dichromatic. Psittaculirostris are the largest and have strikingly colored elongate, lanceolate facial feathers. They are frugivorous and, again almost uniquely in parrots, they excavate rather than use existing hollows for nests in rainforest trees. They range across Indonesia, New Guinea, and Australia.
Two genera, Cyclopsitta and Psittaculirostris, are currently recognized. We recognize long-running, contentious debate about whether Cyclopsitta should be used instead of Opopsitta (e.g., Mathews, 1916; Holyoak, 1970; Storr, 1973, Schodde, 1978; Schnitker, 2014) but follow the decision to use Cyclopsitta. Previous work by Mitchell et al. (2021) and phylogenomic trees show that Cyclopsitta was paraphyletic (fig. 14), and Psittaculirostris, while monophyletic, was nested within Cyclopsitta. The complexes of taxa long treated as subspecies of C. gulielmitertii and C. diophthalma have not been completely sampled for molecular data. With that caveat, C. gulielmitertii emerged as sister to a clade comprising Psittaculirostris and the similarly diverse, polytypic C. diophthalma complex. Note that this is based on the limited taxon sampling that has been done in a mitochondrial genome analysis (Mitchell et al., 2021) and in our phylogenomic analysis. In our study, as iterated before, we sampled only one individual per species. In contrast, Mitchell et al. (2021) sampled several but not every subspecies in C. diophthalma, and in C. gulielmitertii they sampled just one individual of C. gulielmitertii melanogenia. It is worth noting that the same C. gulielmitertii sample (ANWC B 56211) was used here and by Mitchell et al. (2021). Within Psittaculirostris, P. edwardsii and P. salvadorii diverged 0.8 Mya (0.4–1.6) and the ancestor of these taxa diverged from P. desmarestii 1.8 Mya (0.7–2.8; fig. 14).
A case to break up Cyclopsitta had been made on morphological and ecological grounds (Schnitker, 2014), reinforcing the molecular data that show the genus to be paraphyletic. Suavipsitta Mathews, 1917, is available for all taxa within the polytypic gulielmitertii complex (see Mathews 1917b), whether ascribed species or subspecies rank (see summary in Beehler and Pratt, 2016) in the recent trend to recognize more than one species within the gulielmitertii complex). The only alternative to not recognizing Suavipsitta would appear to be placing all these birds—including Psittaculirostris—in one genus. We argue that this would unhelpfully obscure the group's phenotypic and phylogenetic diversity. It also simply repeats the recognition of the group at tribe level in Cyclopsittini. We therefore endorse the recognition of Suavipsitta.
Satisfactory species-level taxonomies for both the C. diophthalma and S. gulielmitertii complexes await molecular studies with full taxon sampling.
Melopsittacus
Melopsittacini contains a single genus and monotypic species (Melopsittacus undulatus), a small green parakeet that occurs throughout the interior of Australia. It is well known in captivity around the world (reviewed in Olsen, 2021).
Loriini
The nectarivorous lories and lorikeets (Loriini) are a diverse group that radiated across Indonesia, the Philippines, Australia, and the South Pacific, having their highest diversity on New Guinea. The clade exhibits a wide range of body sizes and plumage colors (Merwin et al., 2020). Relationships among genera are consistent across studies (Schweizer et al., 2015; Smith et al., 2020) and the phylogenomic tree presented here (Smith et al., 2023). The poorer-quality samples in the clade containing Eos, Saudareos, and Trichoglossus, which had the highest frequency of nodes in the Loriini with <95% UFBS support, prevented verifying its placement in the species tree. In our discussion below of species-level relationships, we exclude reference to the species tree from Smith et al. (2023) given the lower quality of sequence data, thus poorly resolved gene trees, for the lorikeets. We attribute the lower information content of lorikeet sequence data relative to the other parrots because the lorikeets were the first group for which we derived data from a commercial sequencing facility. Nonetheless, the concatenated phylogenomic tree provided a robust phylogenetic hypothesis that reaffirmed previous works and added nearly all described taxa.
FIGURE 14.
Time-calibrated topology of Loriinae. Support values come from the maximum likelihood tree. Support values come from the maximum likelihood tree. Nodes have ultrafast bootstrap values of ≥95% otherwise noted. Gray bars represent divergence time ranges were estimated from 100 bootstrap trees. * denotes an unsupported node where the topology of the presented time-calibrated phylogeny differs from that of the maximum likelihood tree.
Using the phylogenomic tree of the Loriini, Joseph et al. (2020) reclassified generic limits within the clade to align with phylogenetic history. This revision resulted in the splitting of several nonmonophyletic genera (Trichoglossus; Charmosyna; Psitteuteles). There are three main clades in the Loriini: (1) the monotypic and small-bodied Oreopsittacus; (2) includes diminutive to small lorikeets with long tails: Charmosyna, Vini, Charminetta, Hypocharmosyna, Charmosynopsis, Synorhacma, and Charmosynoides; and (3) species with a broad range in body size (small to large) and tails varying from short and square to long and attenuated: Neopsittacus, Glossopsitta, Lorius, Parvipsitta, Psitteuteles, Pseudeos, Chalcopsitta, Glossoptilus, Eos, Trichoglossus, and Saudareos. Clade 1 was sister to clades 2 and 3. The origin of the deepest clades within Loriini occurred at 14.2 Mya (6.0–19.5) and 12.5 Mya (4.8–17.5; fig. 14). Within clade 2, the topology forms a ladder with consecutive divergences of Charminetta, Hypocharmosyna, and Charmosynopsis. The remaining relationships were sister genera, Synorhacma and Charmosyna, and Vini and Charmosynoides, respectively (fig. 14). Maximum likelihood divergence dates among genera within clade 2 ranged from 6.1 to 10.4 Mya (fig. 14). Within clade 3, Neopsittacus then Lorius were sister to the remaining members of the group. The phylogenetic placement of Neopsittacus as sister to the diverse clade containing Lorius, Saudareos, Eos, Trichoglossus, Glossopsitta, Pseudeos, Chalcopsitta, Psitteuteles, and Glossoptilus was stable across phylogenies. Parvipsitta and Psitteuteles, and Pseudeos and Chalcopsitta were both sister pairs. The remaining members of clade 3 included Glossoptilus, which was sister to a group containing Glossopsitta, Eos, Trichoglossus, and Saudareos. The high generic diversity of clade 3 was estimated to have originated within a range of 5 to 9.7 Mya based on maximum likelihood divergence dates (fig. 14).
Oreopsittacus
Oreopsittacus is monotypic with geographic variation across the New Guinea highlands. Oreopsittacus arfaki is a small green lorikeet, with a long tail, and a distinct facial pattern of purple cheek patches and a broken “spotted” white malar stripe. As sister to all other lorikeets, it is on a long branch stemming from the basal divergence of the radiation dating to 14.2 Mya (6–19.5; fig. 14). Despite the significance of its evolutionary position in the lorikeet phylogeny, it is unclear how Oreopsittacus may inform the evolution of traits in the clade. Body size and plumage evolution are labile in lorikeets (Merwin et al., 2020). From a biogeographic perspective Oreopsittacus helps reinforce the hypothesis that lorikeets originated in New Guinea.
Charminetta, Hypocharmosyna, Charmosynopsis, Synorhacma, and Charmosyna
Charmosyna was split into five additional genera to reflect phylogenetic relationships (Joseph et al., 2020). For simplicity, we will first discuss each of these newly circumscribed or reinstated genera (Charminetta, Hypocharmosyna, Charmosynopsis, and Synorhacma) and the current membership of Charmosyna followed by a section on Vini and Charmosynoides. The largest species in this clade are Charmosyna papou and C. stellae, but most taxa in this clade are small bodied with long tails.
Charminetta comprises one New Guinean species, C. wilhelminae. It is a very small lorikeet with distinctive red underwings and rump and a blue hindcrown, presumably a plesiomorphic trait in lorikeets similar to the maroon belly plumage of Arini.
Hypocharmosyna comprises two very similar species with sexually dichromatic elongated ear-covert feathers (blue in males, yellow in females) and red flanks or sides of the underparts. They occur on New Guinea and satellite islands.
Charmosynopsis comprises two phenotypically disparate species, C. pulchella of New Guinea, largely red below and green above; and the other, C. toxopei, of Buru in Maluku, Indonesia, almost entirely green but for its blue frond and forecrown.
Synorhacma of New Guinea although generally green shows yellow striations on its underparts, a patterning that is most unusual among all parrots, and uniquely in lorikeets, a bicolored bill.
Charmosyna, which formerly was a large paraphyletic assemblage, now comprises just three New Guinean species that are largely red below and green above with blue hindcrowns posteriorly edged with black. Two of the species, C. stellae and C. papou, have extraordinarily long central tail feathers. The third species, C. josefinae, resembles the other two but with nonelongated central tail feathers.
The phylogenetic pattern of elevational preference is interesting in that there have been multiple colonizations of the highlands and lowlands, where the number and directionality of the transition would be dependent on the ancestral state that is assigned. The sister to all remaining taxa in clade 2 was the smallest lorikeet Charminetta wilhelminae, which is a monotypic genus of the New Guinea highlands with no recognized geographic variation. There are then four subclades that reflect transitions between elevations. Hypocharmosyna (H. rubronotata and H. placentis) occur in the lowlands of New Guinea and surrounding islands. The widely distributed Hypocharmosyna placentis has deep phylogeographic structure that includes both mainland and insular forms (Smith et al., 2020; Joseph et al., 2020). More detailed populational-level sampling would be required to fully understand the evolutionary history of this taxon. Vini and Charmosynoides primarily occur in lower elevation habitats, but it is worth noting that montane conditions can be present at low elevations on Pacific islands (e.g., see V. palmarum in Gaua Island, Vanuatu; Andersen et al., 2017). Charmosynopsis toxopei occurs on Buru Island from 600–1000 m (Forshaw and Knight, 2010), Indonesia, and Charmosynopsis pulchella is another endemic species of the New Guinea highlands. Monotypic Synorhacma containing only S. multistriata occurs on the southern slopes of the central range in western New Guinea, occurring up to 1800 m (Forshaw and Knight, 2010).
Charmosyna sensu stricto now contains three species that occur in the highlands of New Guinea. Charmosyna papou and C. stellae have distinctly elongated central tail feathers and were taxonomically recognized as two species prior to molecular evidence. The phylogenomic data affirms the placement of subspecies within these species (Smith et al., 2020; Joseph et al., 2020): wahnesi and goliathina (UFBS <70%) was placed within C. stellae, and papou was rendered monotypic. The Charmosyna radiation was dated from 5.2 Mya (1.1–8.6; fig. 14).
Vini and Charmosynoides
The expanded Vini of Joseph et al. (2020) encompasses a radiation of diversely colored, small-bodied birds. We acknowledge informal discussion of the case to replace Vini Lesson, 1833, with Coriphilus Wagler, 1832. In contrast to the case of Orthopsittaca manilatus discussed above, we cautiously predict that Vini will warrant retention on the grounds of much more frequent usage, and therefore, stability. For example, in a search of the Biodiversity Heritage Library ( https://www.biodiversitylibrary.org; accessed 8 January 2024) for references to “Vini Lesson” and “Coriphilus Wagler,” we found 1871 vs. 223 full-text and 31 vs. 0 catalog references, respectively, citing these names, i.e., usage of Vini has demonstrably been more frequent than that of Coriphilus. Retention of Vini will promote stability and minimize disruption.
Vini as now construed exhibits high morphological diversity as it includes species formerly assigned to Vini, Charmosyna, and Phigys. It includes taxa with either short rounded tails (e.g., V. solitarius) through longer attenuated tails (e.g., V. meeki), species that are predominantly green with minimal red about the face and remarkably colored species with much red or blue, or both, in their plumage.
Charmosynoides was erected for the Solomon Islands endemic C. margarethae. Although C. margarethae certainly resembles the red and green species now placed in Charmosyna sensu stricto, traits such as its broad yellow pectoral band and its biogeographic distinctiveness were highlighted by placement in a monotypic genus. Charmosynoides and Vini are sister taxa that shared a common ancestor 7.7 Mya (2.7–11.2; fig. 14).
The more basal nodes in the radiation gave rise to mostly green birds, whereas more brilliantly and multicolored species (e.g., V. peruviana; V. ultramarina) stem from more derived nodes. This pattern of plumage evolution mainly follows a west-east expansion from Melanesia into Polynesia that occurred within the past 6.1 Mya (2.1–8.9; fig. 14). A still unresolved systematic puzzle within this clade is the placement of extinct V. diadema, which is known from a single specimen at the Paris Natural History Museum. Vini diadema was not included in Smith et al. (2020), but Joseph et al. (2020) advocated for its placement in Vini given its phenotypic similarity and geographic proximity to other small-bodied, green Vini.
Neopsittacus
Neopsittacus comprises two very similar, larger New Guinean lorikeets that, while dorsally green like most lorikeets, have extensive red underparts and underwings and yellow streaked facial plumage. They are placed on a long branch that coalesces with the other members of clade 3, 9.7 Mya (3.2–14.2; fig. 14). Smith et al. (2020) and Joseph et al. (2020) sampled two of three subspecies for each group that exhibits relatively shallow geographic structuring across their range.
Lorius
Lorius are midsized, stocky lories, mostly red with green wings and tail, and often with a dark crown. The six species form three sister pairs that are geographically coherent, a finding recovered in previous work using a small multilocus dataset (Schweizer et al., 2015). The three sister species pairs conform to a longitudinal geographic pattern of diversification and are: the westernmost pair, L. domicella and L. garrulus of Moluccas, Indonesia, was sister to the other two pairs, geographically central L. lory and L. hypoinochrous of New Guinea and surrounding islands; and easternmost L. albidinucha and L. chlorocercus of New Ireland and the Solomon Islands. The genus is absent from geographically intervening islands. The sister pairs Lorius albidinucha/L. chlorocercus and L. lory/L. hypoinochrous form a well-supported clade. Lorius had a crown age 5.3 Mya (1.7–8.2) in the tree presented here (fig. 14), and intraspecific variation continued to diversify into the late Pleistocene (Smith et al., 2020). Future systematic work on the group should resolve relationships among subspecies and assess whether any warrant species status. Smith et al. (2020) and Joseph et al. (2020) showed that species with geographic variation were monophyletic, but relationships often had lower support.
Parvipsitta and Psitteuteles
Parvipsitta comprises two Australian species, one of which (P. pusilla) is green with some red about the bill. As such, it closely resembles a number of other species now in newly circumscribed Vini, thus indicating the likely plesiomorphic nature of this predominantly green plumage with some facial red. Parvipsitta porphyrocephala in contrast has a more complex pattern of bluish underparts and purple crown, orange ear-coverts, and red underwing. It is the only lorikeet to range extensively into semiarid and arid zone habitats.
Psitteuteles now comprises just one species, Ps. versicolor, of tropical northern Australia. Predominantly green, it has a mauve chest and red crown, striated underparts, and a prominent naked, white periophthalmic ring.
Parvipsitta and Psitteuteles are deeply divergent lineages subject to previous reclassification (Schweizer et al., 2015). Although the distinction between Parvipsitta and Psitteuteles might be considered arbitrary, we reiterate the conclusions of Joseph et al. (2020) that placing all three species in Psitteuteles would form a very highly heterogeneous and differentiated group. Parvipsitta and Psitteuteles diverged from each other in the Pliocene at 7.3 Mya (2.1–11) and Pa. pusilla and Pa. porphyrocephala share a common ancestor dating back to 6.1 Mya (1.8–9.2; fig. 14). The three species are distinct in plumage, vocalization, and have nonoverlapping ranges in semiarid or mesic woodlands in Australia. For these reasons as well as to minimize nomenclatural change, we advocate their retention in Parvipsitta (cf. Pseudeos fuscata and Ps. cardinalis below). Psitteuteles versicolor of northern Australia, Pa. pusilla of eastern Australia, and Pa. porphyrocephala of southern and western Australia each exhibit no known geographic variation, a pattern observed in numerous highly vagile lorikeets that do not exhibit disjointed ranges caused by topographic or oceanic barriers. Network analysis of mtDNA also indicated that geographic genetic structure was lacking in Pa. porphyrocephala (Dolman and Joseph, 2015).
Chalcopsitta, Pseudeos, and Cardeos
Chalcopsitta are large lories with midlength tails and atypical colors. The three species in Chalcopsitta (atra, scintillata, and duivenbodei) occur in the New Guinea lowlands and some adjacent islands. We follow Beehler and Pratt (2016) in using the emended spelling of the epithet scintillata despite its introduction as Psittacus sintillatus. Chalcopsitta scintillata has generally green body plumage that is heavily striated yellowish, similar to Synorhacma multistriata and Psitteuteles versicolor. Apart from a red forecrown in adults, however, its head plumage is largely dark to black. Apart from the black crown of some Lorius and melanic individuals of Charmosyna stellae, this is not observed in the Loriini outside Chalcopsitta. Even more distinct is the nearly entirely black C. atra (save for yellow and red undertail in some populations) and the brown, yellow, and purple C. duivenbodei (which has very prominent yellow underwings). Relationships among the three species are consistent and well supported (fig. 14). Chalcopsitta duivenbodei was sister to C. atra and C. scintillata. The crown age of Chalcopsitta is 3.7 Mya (0.9–6.0; fig. 14).
Pseudeos and Cardeos both comprise single species and have been recently treated as congeneric (Joseph et al., 2020). We will argue below that separation into two genera can be justified. They are both characterized by extensive red or orange in their plumage.
Prior molecular work showed that Pseudeos cardinalis, long placed without comment in Chalcopsitta, is more closely related to Pseudeos fuscata than to the three species of its former genus, Chalcopsitta (Schweizer et al., 2015). Our phylogenomic work confirms this result, but the species tree showed Pseudeos cardinalis as sister to Chalcopsitta, and P. fuscata as sister to the whole clade (fig. 14). The phylogenetic discordance can be explained in part by the short internodal distance spanning the mean divergence between Chalcopsitta and Pseudeos at just 0.6 Mya with almost largely overlapping temporal ranges from the bootstrapped trees. The split between Chalcopsitta and Pseudeos was estimated at 6.7 Mya (2–10.3) and the divergence within Pseudeos was 6.1 Mya (1.9–9.5; fig. 14).
Concerning plumage traits in these five species, we stress the caveat of not giving undue taxonomic significance to similarities and differences. A general pattern of evolution in parrots, especially “radiations” of closely related species, is that relatively high diversity can evolve quickly under a range of evolutionary forces (Merwin et al., 2020). With that proviso, we make the following observations.
Pseudeos show atypical plumage patterns and indeed colors not only for lorikeets but for parrots generally (e.g., the peculiar light greenish rump color in Pseudeos fuscata unique in all Psittaciformes except perhaps a similar but not necessarily homologous color in Poicephalus meyeri, Psephotellus dissimilis, and P. chrysopterygius). The more uniformly “cardinal” red species cardinalis, recently transferred to Pseudeos but whose generic status we investigate here, is distinct from all species within the Chalcopsitta/Pseudeos clade, most strikingly from its closest relative, P. fuscata. The most common plumage in P. fuscata is a combination of brown, black, and orange, but it has a polymorphic plumage in which yellow largely replaces orange. Research into the pigment(s) underlying these colors in the two species of Pseudeos would be most interesting, e.g., are they all due to modified expression of psittacofulvins, the pigment underlying pigment-based red-to-yellow hues in parrots (McGraw and Nogare, 2004)? Striated plumage patterning is absent in Pseudeos whereas in Chalcopsitta all three species share unique nuchal streaking. Further, among all five of these species only C. scintillata is ventrally streaked, the ventral barring in the other four presumably an ancestral trait. Figure 15 illustrates these traits.
Concerning other characters, we note that in Pseudeos (including cardinalis) bill color is orange, or orange and black, whereas in Chalcopsitta it is black. The species cardinalis has an apparently autapomorphic trait of bare yellow skin around the base of the mandible. This skin of course is dark in museum specimens (fig. 15). Gross body size is disparate within both Chalcopsitta and Pseudeos sensu lato, with P. fuscata and C. scintillata the smaller species in their respective genus and the other three (atra, duivenbodei, and cardinalis) similarly large (see fig. 15). Lastly here, we note that P. cardinalis is endemic to a major center of avian endemism, the Solomon Islands.
In contrast to our earlier position (Provost et al., 2018; Joseph et al., 2020), we here acknowledge the collective weight of available genomic data coupled with biogeographic evidence and phenotypic data (notwithstanding the lability of color evolution in lorikeets) that warrants placement of cardinalis in monotypic Cardeos Verheyen, 1956. We acknowledge that the temporal divergence between Cardeos cardinalis and Pseudeos fuscata closely matches that between the two species of Parvipsitta, but we do not advocate generic separation of the latter. Very different biogeographical patterns and contexts apply in the two groups and we argue that the generic taxonomy should capture this.
Glossoptilus
Glossoptilus was reinstated for the New Guinean species G. goldiei once the paraphyly of Psitteuteles sensu lato was clarified. A small green lorikeet of the New Guinea highlands, it is distinguished by a full red crown and violet-purple facial plumage, and dark streaked underparts. It has no described geographic variation. Glossoptilus goldiei diverged from the clade containing Glossopsitta, Eos, Saudareos, and Trichoglossus 6.9 Mya (2.1–10.3; fig. 14).
Trichoglossus
Trichoglossus lorikeets often have clearly discrete colors across their heads, wings, chest, and bellies. A differently colored nuchal collar is also present in many forms. Narrow, dark terminal barring is on underparts' plumage.
FIGURE 15.
Specimens from the American Museum of Natural History (AMNH) of the species assigned to Chalcopsitta (atra AMNH 616728; scintillata AMNH 425529; duivenbodei AMNH 339804), Pseudeos fuscata AMNH 339121, and Cardeos cardinalis AMNH 217110. Ventral, side, and dorsal images from left to right. See text for discussion. Photographs: Sahid Robles Bello.
The reconfigured Trichoglossus sensu Joseph et al. (2020) brought morphological homogeneity to the clade in that it now comprises species of similar size and gross morphology and that are also biogeographically cohesive, albeit occurring over a large geographic area. Despite improved circumscription of Trichoglossus, its extraordinary color variation still presents challenges in understanding species limits within it. Most of the current uncertainty lies with the Rainbow Lorikeet (Trichoglossus haematodus) complex, long considered one of the most polytypic bird species. Recent taxonomic revisions have led to recognition of 10 species, mostly from elevating variants of typically “rainbow” plumaged T. haematodus to species rank (haematodus, rosenbergii, moluccanus, rubritorquis, capistratus, and forsteni), as well as the yellow-billed and yellow-tailed but otherwise almost uniformly maroon-plumaged species usually treated at species rank (rubiginosus) and taxa that are mostly green-plumaged birds (chlorolepidotus, euteles, and weberi). Based on subspecific sampling and phylogenomic data it is unclear whether these currently recognized species limits define monophyletic groups (Smith et al., 2020; Joseph et al., 2020). Both rubiginosus and chlorolepidotus appear to be outside the main Rainbow Lorikeet radiation, although their exact relationships to each other and Trichoglossus was not stable. For this reason, we refrain from using Oenopsittacus Reichenbach, 1913, of which T. rubiginosus is the type species for either or both of these species. We are currently employing an approach that uses whole-genome data and population-level sampling to further attempt to differentiate relationships and species limits in Trichoglossus. The extraordinary Trichoglossus radiation has a crown age 4.6 Ma (1.2–7.4; fig. 14).
Glossopsitta
Glossopsitta is a monotypic genus of small green lorikeet with a red stripe of plumage through the eye and blue crown. G. concinna of mainland southeastern Australia and Tasmania shares a common ancestor with its sister lineage, Trichoglossus at 4.8 Mya (1.3–7.8; figs. 1, 14). Figure 16 indicates that the difference of more blue in the crown of mainland populations likely is valid for adult males and that while closer study is certainly warranted, we see merit to continued recognition of G. c. didimus.
Saudareos
Saudareos are smaller bodied, largely green lorikeets with long, attenuated tails. They are phenotypically disparate, perhaps reflecting evolutionary reshuffling of a number of plesiomorphic traits among them, e.g., distinctive ear-coverts, well-demarcated chest and belly plumage with narrow, dark terminal barring. This combined with their similarity to either the brightly colored (Trichoglossus haematodus) or mostly green Trichoglossus (T. euteles) caused confusion in premolecular classification. The taxonomic uncertainty extended beyond Trichoglossus. Saudareos iris was previously placed in Psitteuteles, which is now considered a paraphyletic assemblage of small lorikeets (Smith et al., 2020). Support for the Saudareos and Eos sister relationship varied across phylogenomic studies, which come from the same dataset but the sequence alignments varied slightly (phylogenomic tree presented here UFBS = 73%; subspecies tree from Smith et al., 2020, UFBS = 100%). The most northerly distributed lorikeet, S. johnstoniae, occurs in the mountains of Mindanao in the southern Philippines and diverged from other members of the genus 4.5 Mya (1.2–7.3; fig. 14). The following nodes reflect the divergence among and potentially within Indonesian archipelagos. Saudareos iris of Timor and Wetar Islands, Lesser Sundas, diverged from S. flavoviridis and S. ornata 3.9 Mya (1–6.9; fig. 14). Saudareos flavoviridis and S. ornata shared a common ancestor 3.4 Mya (0.8–6.3) and both occur on Sulawesi and neighboring islands. The species currently exhibit partial elevational segregation on Sulawesi, which may have contributed to the initial isolation. We note that in our earlier study (Joseph et al., 2020) Saudareos flavoviridis meyeri was closer to S. ornata than to other S. flavoviridis. Although we have not sampled to this extent in this study, we acknowledge the elevation of meyeri to species rank elsewhere (del Hoyo and Collar, 2014; Gill et al., 2023).
FIGURE 16.
Dorsal view of specimens of Glossopsitta concinna from the Australian National Wildlife Collection (ANWC) showing variation within and between sexes and within and between mainland southeastern Australia versus Tasmania. Note the bluer coronal color in mainland males. Registration numbers from the bird collection at ANWC are shown. Photograph: Gordon Gullock.
Eos
Eos comprises a fairly uniform clade of midsize lories with rounded tails and predominantly red plumage with blue and black patches, whose radiation dated to 3.7 Mya (1–6.3) occurred entirely in Indonesia; members of this clade are sister to Saudareos (fig. 14), the formerly Indonesian species of Trichoglossus. Relationships within Eos were poorly resolved (fig. 14) in our tree, likely due to poorer-quality samples. This lack of resolution limits what can be inferred about spatial history of the group. Smith et al. (2020) included subspecific sampling of all but two subspecies, and they found similar results of overall low support. It is of note that the Eos squamata squamata used in Smith et al. (2020) was mislabeled and is Eos squamata riciniata. Eos will require higher-quality genomic data to verify phylogenetic relationships.
CONCLUSION
If we have succeeded in providing a new benchmark in the taxonomy of the entire Psittaciformes, then we will have built a robust framework for downstream analyses of parrots in diverse fields such as ecology, ethology, physiology, conservation, and biogeography to name a few.
Several themes for further work have emerged. Apart from a few nomenclatural issues, we are confident that family-level systematics of parrots has largely stabilized. At the level of genera, however, challenges remain in resolving interrelationships, especially where very short internodes reflecting rapid evolutionary divergences have been detected. Relationships within the Arini are a clear example. This pattern was present in early molecular studies and continues to be observed in the phylogenomic dataset presented here. Genus-level taxonomy of parrots remains a challenging exercise. The relatively conserved psittaciform body plan along with plumage color that is subject to different degrees of natural and sexual selection still are at the root of remaining uncertainty. Study of the environmental and genomic drivers of the evolutionary forces that have produced similarly patterned taxa that are not closely related (Strigops vs. Pezoporus; Touit costaricensis and T. dilectissimus vs. Cyclopsitta diophthalma) and dissimilar taxa that are closely related (e.g., the species of Aratinga) would be rewarding. We hope that debates arising from our work on whether to recognize genera will be few. We predict that any that do arise ultimately will be grounded either in the non-predictability of morphological evolution or the complexities of introgression and, we hope, will be resolvable through the phylogenetic framework we have generated. If nothing else, we hope we have contributed substantially to understanding relationships.
Researchers should continue probing diversity within and between what are currently recognized as species to clarify species limits. Of course, much remains to be done in sampling geographic and intraspecific taxonomic diversity especially where species limits are debated or simply still need clarification (e.g., Pionites leucogaster, Psittacella picta, Barnardius zonarius, Trichoglossus haematodus sensu lato, Psittinus cyanurus, Tanygnathus sumatranus sensu lato). We hope these studies will be accompanied by disciplined discussion of rates of phenotypic evolution coupled with research into underlying patterns of gene flow and shared versus unique genetic diversity.
Further, we have often noted the lower quality of some of our data as likely due to the necessity in many cases of using historical museum specimens for DNA samples. That necessity will be difficult to overcome, but the use of genomic methodologies in achieving all of this work should be seen as de rigeur.
ACKNOWLEDGMENTS
We thank Joseph M. Forshaw and illustrators William T. Cooper, Frank Knight, and John C. Yrizarry for their works on the parrots of the world and Australia in particular. Words cannot express the depth of the inspiration they have given us. We express our thanks to several colleagues for their interest in this work and various suggestions they have made as the work has materialized. In particular, we thank V. Remsen, J. Croxall, R. Schodde, E. Tenorio, R. Voss, and an anonymous reviewer for comments on the entire text, and E. Dickinson, M. Gosselin, and S. Gregory for nomenclatural advice concerning Orthopsittaca manilatus. For permission to use their photographs we thank David Cook, Marcos Eugênio, John Messing, and Dubi Shapiro. Sahid Robles Mello and Gordon Gullock kindly took photographs of specimens in the American Museum of Natural History and the Australian National Wildlife Collection, respectively. We also thank the following institutions and people for providing samples: Academy of Natural Sciences of Drexel University (N. Rice, J. Weckstein), American Museum of Natural History (P. Sweet, T. Trombone, G. Amato), Carnegie Museum of Natural History (S. Rogers), Cornell University Museum of Vertebrates (C. Dardia), CSIRO Australian National Wildlife Collection (R. Palmer, C. Wilson, R. Schodde), Delaware Museum of Natural History (J. Woods), Field Museum of Natural History (B. Marks, J. Bates, S. Hackett), Florida Museum of Natural History (A. Kratter, D. Steadman), Louisiana State University Museum of Natural Science (R. Brumfield, D. Dittmann, S. Cardiff, V. Remsen, F. Sheldon, B. Faircloth, N. Mason), Museum of Comparative Zoology, Harvard University (J. Trimble, S. Edwards), Museum of Southwestern Biology (A. Johnson, C. Witt, M. Andersen), Museums Victoria (J. Sumner), National Museums Scotland (R. McGowan), National Museum of Natural History, Smithsonian Institution (B. Schmidt, H. James, G. Graves), Natural History Museum of Denmark (P. Hosner, K. Jønsson), Natural History Museum of Los Angeles County (K. Garrett, A. Shultz), Royal Ontario Museum (M. Peck), San Diego Natural History Museum (P. Unitt), San Diego State University (K. Burns), S. Claramunt, University of Kansas Biodiversity Institute (M. Robbins, R. Moyle), University of Washington Burke Museum (S. Birks, R. Faucett, J. Klicka), Western Australian Museum (R. Bray, R. Johnstone), Western Foundation of Vertebrate Zoology (R. Corado), Wildlife Conservation Society, and Yale Peabody Museum (K. Zyskowski, R. Prum). For reviewing the submitted paper we are indebted to James V. Remsen, John Croxall, Richard Schodde, and an anonymous reviewer. We thank Mary Knight for the polish she applied through her diligent copyediting. This study was funded in part by National Science Foundation awards to BTS (DEB-1655736 and DBI- 2029955).
Copyright © American Museum of Natural History 2024
