Disparate origins for endemic bird taxa from the ‘Gondwana Rainforests’ of Central Eastern Australia

Subtropical and temperate rainforests of Central Eastern Australia are some of the largest remaining fragments of their kind globally. The biota of these rainforests appears to comprise two broad biogeographical elements: a more ancient (Miocene or older) and typically upland temperate (‘Gondwanan’) element and a younger (Plio-Pleistocene) lowland tropical element. We present the first phylogenetic synthesis of the spatiotemporal origins for the eight bird taxa endemic to Central Eastern Australian Rainforests. At least five of these eight focal taxa show Plio-Pleistocene divergences from their respective northern sister taxa, consistent with origins driven by recent expansion and contraction of lowland rainforest. In contrast, two more strictly upland species, the rufous scrub-bird ( Atrichornis rufescens ) and the logrunner ( Orthonyx temminckii ), diverged from their nearest living relatives during the Miocene, suggesting potentially longer histories of persistence and more temperate origins. Finally, we did not recover reciprocal monophyly in mitogenomes from the two extant lyrebirds, Albert’s lyrebird ( Menura alberti ) and the superb lyrebird ( Menura novaehollandiae ). The disparate divergence ages recovered among all eight taxa are consistent with the biota of the Central Eastern Australian Rainforests comprising isolates either of younger age and tropical lowland origins or of older age and temperate upland origins.


INTRODUCTION
Since the Miocene climatic optimum, rainforests in Australia have contracted and fragmented (Bryant & Krosch, 2016), and Plio-Pleistocene glacial cycles have driven further expansion and retraction of rainforest habitats (e.g. Hugall et al., 2002). Today, the rainforests of Australia are concentrated in < 1% of the land area of the continent, mainly along its eastern seaboard. They nonetheless host exceptional species richness and phylogenetic endemism (Stork et al., 2008;Byrne et al., 2011). The tropical rainforests in the Wet Tropics region of north-east Queensland ( Fig. 1) have been recognized as a natural laboratory for understanding the processes underpinning isolation, persistence and speciation in rainforest biotas (e.g. Singhal et al., 2018). However, fewer studies have focused on the subtropical Central Eastern Australian Rainforests (CEAR) in south-eastern Queensland and north-eastern New South Wales (Hugall et al., 2003). These borderlands are home to some of the largest remaining areas of subtropical and temperate rainforest in the world, and they too host a distinct assemblage of endemic species (Weber et al., 2014).
The dynamics and history of diversification of organisms from the CEAR are more poorly understood than those of the Wet Tropics rainforest further north. Understanding these issues for all rainforests and their associated biotas is a central focus in Australian biogeography (Byrne et al., 2011;Bryant & Krosch, 2016). Two broad biogeographical elements have previously been identified in Australian rainforest biota. One element comprises deeply divergent relict lineages, which often date to the Miocene (e.g. Hugall et al., 2008;Skinner et al., 2013;Skipwith et al., 2019;Tallowin et al., 2020) or even earlier (Ponniah & Hughes, 2004;Rix & Harvey, 2012;Moreau et al., 2015). Many but not all of this first group are descended from predominantly temperate and/or mesic Australian lineages. They often have long evolutionary histories in Southern Hemisphere (Gondwanan) landmasses (Sniderman & Jordan, 2011;Yap et al., 2018), and biogeographical affinities with montane New Guinean rainforests have also been noted (Schodde & Calaby, 1972;Rix & Harvey, 2010). The second biogeographical element comprises groups for which genetic and fossil data suggest much more recent connectivity across eastern Australia. They are often associated with range expansion and contraction during the Plio-Pleistocene (Nicholls & Austin, 2005;Macqueen et al., 2011). This group is therefore often linked to recent colonization from tropical regions to the north (e.g. Queensland Wet Tropics and/or New Guinea; see Irestedt et al., 2017;Ericson et al., 2020). These younger lineages typically inhabit lowland and upland rainforest dominated by floristic lineages that have colonized Australia from Malesia (Yap et al., 2018;Kooyman et al., 2019;Brambach et al., 2020;Joyce et al., 2020).
The prevalence of the first group of lineages in the CEAR has led to popularization of the term 'Gondwana Rainforests ' (e.g. DAWE, 2020;Kooyman et al., 2020;UNESCO, 2020). However, despite the use of that term in the names of UNESCO World Heritage and Australian National Heritage sites, we favour the name Central Eastern Australian Rainforests here for two reasons: rainforests with numerous Gondwanan-derived lineages occur across eastern Australia, and many species inhabiting these rainforests occur further north and south than those officially dedicated areas. For the latter reason, earlier literature referred to these 'Gondwana Rainforests' as the Gympie-Illawarra block (Schodde & Calaby, 1972). Recently, bushfires of unprecedented extent have threatened these rainforests, exacerbating the impacts of anthropogenic land clearance and invasive species on biodiversity (Hines et al., 2020;Kooyman et al., 2020). This highlights the importance and urgency of understanding the history of the biodiversity of these rainforests and the sensitivity of their ecosystems to climatic change.
To a d d r e s s t h e s e b i o g e o g r a p h i c a l a n d conservation-related issues, we have focused on the eight bird taxa that are endemic or nearly endemic to the CEAR (Fig. 1A-H; whether some of these taxa are species or subspecies is debated; see Table 1). To date, the phylogenetic divergences of only four of these endemic birds have been studied closely with molecular data (Norman et al., 2002;Irestedt et al., 2016Irestedt et al., , 2017Ericson et al., 2020). Six of the eight endemic birds clearly have closest relatives in rainforest habitats in north Queensland and/ or New Guinea (Table 1; Norman et al., 2002;Irestedt et al., 2016Irestedt et al., , 2017Menkhorst et al., 2017;Ericson et al., 2020). The other two, although unquestionably distinct species in highly divergent relict genera of passerines, have different patterns of relationships. The first, Albert's lyrebird (Menura alberti), has one of the most restricted ranges of any Australian bird, being confined to the Queensland-New South Wales border region (Fig. 1B). It is parapatric with respect to its morphologically distinctive sister species and only extant congener, the superb lyrebird (Menura novaehollandiae) (Higgins et al., 2001;Menkhorst et al., 2017). The second species, the rufous scrub-bird (Atrichornis rufescens), is separated by > 3000 km from its only other extant congener, the noisy scrub-bird (Atrichornis clamosus) from Western Australia (Fig. 1A). Obviously, these birds have different spatial histories, but the temporal patterns of their divergences remain unknown.
Recent advances in high-throughput DNA sequencing and the use of hybridization enrichment have made it feasible and cost-effective to retrieve genetic information from challenging substrates, such as ancient bone and museum skins, from birds (e.g. Mitchell et al., 2014a, b;Scofield et al., 2017;Wood et al., 2017;Boast et al., 2019;Cole et al., 2019;Irestedt et al., 2019;Oliver et al., 2020). These are the only types of specimens available for many birds from the CEAR. In the present study, we used these methods to generate new genetic data that, when combined with published resources, allowed us to perform a taxonomically comprehensive comparative analysis of all eight birds endemic to the CEAR. Our primary aim was to test whether each taxon is either an older (e.g. Miocene) or younger (Plio-Pleistocene) rainforest element. Secondarily, and in light of the answer to the first question, we sought to determine whether each lineage originated in temperate or tropical habitats or in situ in the subtropical region. We thus provide the first comparative phylogenetic approach to understanding the origin of these unique avifaunal elements of the iconic subtropical CEAR.

Sampling
Liver and/or toepad samples for four of our focal taxa and relevant outgroups were obtained from the Australian National Wildlife Collection (ANWC), Museums Victoria (NMV) and Queensland Museum (QM), including multiple geographically distant samples where possible (N = 28; Table 1). DNA extraction and library preparation were performed on the toepad samples at the University of Adelaide in a clean-room laboratory dedicated to molecular analysis of low-biomass museum specimens, whereas the tissue samples were processed at the South Australian Regional Facility for Molecular Ecology and Evolution (SARFMEE).
Among our remaining four focal taxa, phylogenetic and molecular dating results have already been published for three (Table 2): the paradise riflebird (Ptiloris paradisaeus; Irestedt et al., 2017), green catbird (Ailuroedus crassirostris; Irestedt et al., 2016;Ericson et al., 2020) and regent bowerbird (Sericulus chrysocephalus;Ericson et al., 2020). For the remaining taxon, the logrunner (Orthonyx temminckii), genetic data have been published (Norman et al., 2002) but not included in a timecalibrated phylogenetic analysis. We downloaded Norman et al.'s (2002) Orthonyx data and analysed them alongside the new data produced as part of the present study.

Dna extraction, library preparation anD
Sequencing DNA was extracted from all samples using a QIAGEN (Hilden, Germany) DNeasy Blood & Tissue kit following the manufacturer's protocol. DNA from the liver samples was then fragmented to 200-300 bp using a focused ultrasonicator (Covaris: Woburn, Massachusetts, USA). Extracts from all samples were converted into DNA sequencing libraries following the protocol of Meyer & Kircher (2010), with each adapter (5′ and 3′) modified to contain a unique 7mer barcode sequence. Libraries created from the toepad samples included an additional partial uracil-DNAglycosylase treatment (Rohland et al., 2015).
At the end of the library preparation protocol, each library was amplified using polymerase chain reaction (PCR) in eight separate reactions to reduce amplification bias. Each reaction contained 1× PCR buffer, 2 mM MgSO 4 , 1 mM dNTPs, 0.4 mM primer, 0.5 U Platinum Taq High Fidelity (ThermoFisher: Waltham, Massachusetts, USA) and 3 μL DNA library template. Reactions were subjected to the following thermocycling regimen: 94 °C for 2 min; 12 or 16 cycles (for tissues or toepads, respectively) of 94 °C for 30 s, 60 °C for 30 s and 68 °C for 40 s; and a final extension of 68 °C for 10 min. Individual PCR products for each library were pooled after amplification, and purified using 1.8× AxyPrep magnetic beads (Axygen: Blackburn, Victoria, Australia).
Libraries generated from the scrub-bird and fig-parrot toepad samples were enriched for bird mitochondrial DNA using hybridization enrichment with the RNA probes described by Mitchell et al. (2014b) following the protocol outlined by Cole et al. (2019). All libraries were then subjected to a final round of PCR in five separate reactions using fusion primers to add fulllength indexed Illumina (San Diego, California, USA) sequencing adapters for sequencing (Meyer & Kircher, 2010). Each reaction contained 1× PCR buffer, 2.5 mM MgCl 2 , 1 mM dNTPs, 0.5 mM primer, 1.25 U AmpliTaq (ThermoFisher) Gold and 2 μL DNA library template. Reactions were subjected to the following thermocycling regimen: 94 °C for 6 min; seven or 15 cycles (for tissues or toepads, respectively) of 94 °C for 30 s, 60 °C for 30 s and 72 °C for 45 s; and a final extension of 72 °C for 10 min.

mapping anD mitochonDrial genome aSSembly
To create the final mitochondrial consensus sequences used for all downstream analyses, all merged reads for each library were mapped against an appropriate reference using BWA v.1.7.8 (Li & Durbin, 2009; aln -t 8 -l 1024 -n 0.04 -o 2). In some cases no appropriate reference was publicly available; therefore, we initially constructed a draft reference using the iterative mapping feature of the geneiouS reaD mapper (Kearse et al., 2016) in geneiouS v.9.1.6 (https://www. geneious.com; for details, see Supporting Information, Table S1). After mapping with BWA, reads with a mapping quality Phred score > 30 were selected and retained using the SamtoolS v.1.4 (Li et al., 2009) view command (-q 30), and duplicate reads were discarded using 'FilterUniqueSAMCons.py' (Kircher 2012). The only exception to this was the mitochondrial genome sequence we assembled from the previously published shotgun sequencing data from a superb lyrebird specimen of unknown provenance ( (Nguyen et al., 2015). In each analysis, we partitioned our data such that each partition had its own evolutionary rate but all partitions contributed to the branch lengths of a single best tree (-spp;Chernomor et al., 2016), and we assessed topological support using 1000 ultrafast bootstrap replicates (resampling within partitions; Hoang et al., 2017).
We aligned our newly assembled lyrebird sequences with the previously published superb lyrebird sequence (AY542313) using the MUSCLE algorithm as implemented in geneiouS. We used popart (Leigh, 2015) to create a median-joining network from this alignment (Bandelt, 1999). To test the reciprocal monophyly of the two lyrebird species, we also aligned the lyrebird sequences with the mitochondrial genome of Prosthemadera novaeseelandiae (NC_029144), a passerine outgroup. We then extracted the 12 mitochondrial H-strand protein-coding genes from this alignment and performed a phylogenetic analysis using raxml v.8.2.0 (Stamatakis, 2014), comprising a maximum likelihood search for the best-scoring tree from 1000 rapid bootstrap replicates.
We also added our new lyrebird and scrub-bird data, and previously published sequence data for three species of Orthonyx, to a modified version of a previously published Passeriformes supermatrix (see Hugall & Stuart-Fox, 2012). In addition to the sequences we added, we subsampled down to representatives f o r e a c h p a s s e r i n e f a m i l y ( w i t h Pa s s e r i d a reduced further to only four representatives) and 13 loci (five nuclear protein coding, three nuclear introns and five mitochondrial protein coding). The final alignment amounted to 13 361 nucleotides (8574 nuclear and 4787 mitochondrial) for 74 taxa (including four outgroups) and contained 29.1% missing data (Supporting Information, Tables S4 and S5). We created a maximum likelihood tree from this alignment using the ultrafast bootstrap consensus in IQ-TREE (Hoang et al., 2017), based on a partitioning scheme and substitution models (Supporting Information, Table  S3) determined by moDelfinDer (as implemented in IQ-TREE; Kalyaanamoorthy et al., 2017).

molecular Dating analySeS
We performed individual molecular dating analyses corresponding to three different alignments, as follows.

Australasian frogmouths (Podargus spp.)
We subsampled the alignment used previously for our IQ-TREE analysis to retain only single representatives of each Podargus subspecies (removing samples with the highest number of missing data). The optimal partitioning scheme and nucleotide substitution models for this subsampled alignment were established using partitionfinDer. We then co-estimated the phylogeny and node ages using BEAST v.1.8.4 (Drummond & Rambaut, 2007), with a birth-death tree prior and a single lognormal relaxed clock model (including a rate multiplier parameter for each partition). We constrained the crown age of Podargus using a normal distribution (mean = 9.60 Mya; standard deviation = 1.449 Mya) to match the mean (9.60 Mya) and 95% highest posterior density (95% HPD; 6.76-12.71 Mya) for this node estimated by Oliver et al. (2020). We ran three independent Markov chain Monte Carlo (MCMC) chains of 1 000 000 generations each, sampling every 1000 generations. Convergence of parameter values was monitored using tracer v.1.6 (Rambaut & Drummond, 2014), ensuring an Effective Sample Size (ESS) > 200. The first 10% of sampled trees from each chain was discarded as burn-in. We combined the remaining trees and generated a final maximum clade credibility tree.

Fig-parrots (Cyclopsitta spp.)
We subsampled the alignment used previously for our IQ-TREE analysis to retain only single representatives of each Cyclopsitta diophthalma subspecies studied (removing samples with the highest number of missing data). The optimal partitioning scheme and nucleotide substitution models for this subsampled alignment were established using partitionfinDer. We then co-estimated the phylogeny and node ages using BEAST, with a birth-death tree prior and a single lognormal relaxed clock model (including a rate multiplier parameter for each partition). Following the results of Schweizer et al. (2011), we enforced the reciprocal monophyly of the clade comprising Cyclopsitta and Psittaculirostris and the clade comprising the remaining samples, and we constrained the age of the root of the tree according to a normal distribution (mean = 18.6 Mya; standard deviation = 3.5 Mya). Otherwise, this analysis was performed as described above for our frogmouth alignment.

Lyrebirds (Menura spp.), scrub-birds (Atrichornis spp.) and logrunners (Orthonyx spp.)
We estimated divergence times for passerines using the same modified supermatrix described above, because it: (1) provided a backbone of slowly evolving nuclear genes that are less affected by issues of saturation than a mitochondrial-only dataset; (2) included a range of passerine taxa, facilitating dating calibration; and (3) contained enough tips to properly inform sequence evolution and tree prior models. Molecular dating was performed using BEAST2 v.2.4.2 (Bouckaert et al., 2014), with a lognormal relaxed clock model and Yule tree prior (both using 1/X parameterization), and the molecular clock was calibrated by constraining the age of six nodes in accordance with the recent literature on Neoaves and Passeriformes (Supporting Information, Table S6). Briefly, three deep Neoaves and Passeriformes secondary calibrations were drawn from Prum et al. (2015), and three fossil calibrations were implemented following Jønsson et al. (2016) and Oliveros et al. (2019). Two independent MCMC chains of 50 000 000 generations were run using BEAST2, sampling every 5000 generations. Convergence of parameter values was monitored using tracer, ensuring ESS > 200. The first 10% of sampled trees from each chain was discarded as burn-in. We combined the remaining trees and generated a final maximum clade credibility tree.

Data availability
Mitochondrial consensus sequences produced as part of this study are available on GenBank (MW883515-MW883541). Consensus sequences, demultiplexed sequencing reads, and phylogenetic analysis files are available through figshare (DOI: 10.25909/c.5405865).

auStralaSian frogmouthS (Podargus Spp.)
Samples of Podargus ocellatus from across New Guinea and eastern Australia were reciprocally monophyletic with respect to other Podargus species [bootstrap support (BS) = 100%; Fig. 2A Fig. S2). The New Guinean Podargus o. ocellatus sample was highly differentiated from Australian samples, having a mean node age of 5.71 Mya (95% HPD = 3.22-8.29 Mya). Podargus strigoides samples were not monophyletic, with our results weakly supporting one Western Australian sample (ANWC B34665) as sister to all other Podargus strigoides and Podargus papuensis samples combined (BS = 53%). More comprehensive sampling, including nuclear DNA of Podargus strigoides from its continentwide distribution, will be necessary to resolve relationships within and between these two species.  Fig. S3). Non-monophyly of Cyclopsitta is consistent with previously proposed genus-level reclassification of the dusky-cheeked fig-parrot and its close relatives (e.g. Schnitker, 2014), highlighting the need for a detailed taxonomic reassessment of the group. A close relationship between the two C. [d.] Fig. S4).

DISCUSSION
We present the first comparative analysis of patterns of divergence among eight birds that are endemic to the CEAR ( Fig. 1; Table 1) and add to the body of similar data for many other groups (see Bryant & Krosch, 2016: fig. 3). Mitochondrial data for seven of the eight species (i.e. all except the Albert's lyrebird) are broadly consistent with the previously proposed dichotomy between older, Miocene upland temperate lineages and more recent, Plio-Pleistocene lowland tropical lineages (Sniderman & Jordan, 2011). Two species, the logrunner and the rufous scrub-bird, clearly fall within  (Figs S2, S4, S5) and tree files are available through figshare (DOI: 10.25909/c.5405865). Node heights represent mean estimated node ages, and bars associated with nodes represent the 95% highest posterior density (HPD) for node age estimates (for focal taxon age values, Table 2). Samples belonging to focal taxa are coloured according to their distribution. Central Eastern Australian Rainforest taxa are highlighted with bold text and green branches and node bars.  (Table  2; Figure 3), respectively. We suggest that their isolation was driven by Pleistocene cycles of rainforest expansion and contraction. Both taxa are nested within genera that primarily inhabit tropical rainforest habitats (Figure 2), whether upland or lowland, to the north in Queensland's Wet Tropics, Cape York Peninsula or New Guinea. Notably, both are also phenotypically divergent, being substantially larger in size than members of their sister lineages, and thus represent possible examples of Bergmann's rule. In addition, both are currently separated by large geographical distances from their nearest relatives (Higgins, 1999). The geographical disjunction is particularly striking in the case of the frogmouths. The CEAR plumed (Podargus Published data for the CEAR endemic green catbird and paradise riflebird show similarly shallow divergences from closely related taxa occurring further north. Divergences of these two species from their closest relatives, however, are possibly slightly older than in the frogmouths and fig-parrots (Table 2). Finally among the five younger bird lineages, the divergence of the regent bowerbird from its extant Sericulus congeners appears the oldest, with published data indicating an age of > ~3 Mya (Table 2). Nonetheless, the distribution of Sericulus spp. across subtropical rainforest of eastern Australia, and lowland and lower montane rainforest in New Guinea, again points to a predominantly Plio-Pleistocene history of range contraction and expansion.
Taken together, the results for the five most recently diverging taxa highlight that although some components of the CEAR avifauna are relictual and diverged before the Plio-Pleistocene, most endemics are likely to stem from dynamic recent histories of Plio-Pleistocene range contraction. Indeed, fossil and genetic data provide complementary evidence of recent range contraction, isolation and fragmentation across central Queensland in rainforest or wet forest vertebrates (Nicholls & Austin, 2005;Hocknull et al., 2007;Macqueen et al., 2010;Irestedt et al., 2017) and plants (Sniderman & Jordan, 2011). Fig-parrots and catbirds are more diverse in New Guinea than in Australia. This suggests that New Guinea, in addition to acting as a refuge for Australian rainforest lineages (Schodde & Calaby, 1972;Brambach et al., 2020), might also have functioned as a source of dispersal during phases of lowland tropical rainforest expansion in Australia throughout the Plio-Pleistocene. We predict that future analyses of other less phenotypically distinct subspecies of birds with ranges including but less strictly confined to the CEAR will further reinforce this inference of Plio-Pleistocene connectivity between the major tropical and subtropical rainforest blocs in eastern Australia.
In contrast, only two of the eight endemic bird species in our study present phylogenetic patterns that point unequivocally to deeper (Miocene) divergence of taxa now endemic to the CEAR. The rufous scrub-bird and its only extant congener, the noisy scrub-bird from Western Australia, diverged during the late Miocene (Table 2), before several major cooling and drying cycles in Australia (Byrne et al., 2008;Sniderman et al., 2016;Andrae et al., 2018), and the two taxa are currently separated by > 3000 km. The rufous scrub-bird is primarily restricted to upland heath and rainforest associations in the CEAR, but is typically recorded > 600 m above sea level. The noisy scrubbird is confined to small relict patches of temperate mesic heath in far south-west Western Australia (Garnett & Dutson, 2010). Given that both scrub-bird species occur in habitats away from rainforest, their distribution might be limited primarily by aspects of habitat structure, climate and/or frequency of fire. Our new divergence date estimates for all logrunners (Orthonyx spp.) likewise indicate that the species endemic to the CEAR, O. temminckii, diverged from its sister lineage (i.e. the clade of two species from the Wet Tropics and montane New Guinea) in the late Miocene ( Table 2). The depth of genetic divergences shown by scrub-birds and logrunners strongly indicates that they are Miocene relict elements of more temperate 'Gondwanan' rainforest ecosystems, mirroring patterns in other taxa, such as lizards (Skinner et al., 2013;Tallowin et al., 2020), crayfish (Ponniah & Hughes, 2004), snails (Hugall et al., 2003) and spiders (Rix & Harvey, 2012).
Finally, our mitogenomic results for the two lyrebird species present another, entirely unexpected pattern. The age of the clade comprising the lyrebirds and scrub-birds, dating to before the Miocene (28.7 Mya; Fig. 3; Oliveros et al., 2019), indicates that both genera are strikingly relictual and ancient components of the Australia avifauna. Indeed, Early Miocene fossil lyrebirds have been described further west in northern Australia (Boles, 1995). However, our observation of negligible mitochondrial differences between the two lyrebird species conflicts with the dramatic phenotypic differences between the species (Figure 4). We note that all our Albert's lyrebird samples were from males, which are phenotypically most distinct, suggesting that our unusual results are unlikely to be driven by sample misidentification as has been reported for some field observations (Higgins et al., 2001). Thus, although the genus Menura is clearly relictual and ancient, our data so far shed little light on the evolutionary origin of Albert's lyrebird. Its position in the historical biogeography of the CEAR remains enigmatic. Future studies will need to use nuclear genomic data to test alternative explanations for our observations, such as incomplete lineage sorting and recent speciation or fixation of an introgressed mitochondrial lineage following hybridization. As observed in other birds from Australia and New Guinea, species boundaries can be porous and allow for one-way flow of certain genetic markers (Joseph et al., 2019a, b), making this a plausible hypothesis for our unexpected mitochondrial results.
Based on our analyses to date, we advocate no taxonomic changes from prevalent usage among the eight endemic birds of the CEAR. The taxonomic distinctiveness of the rufous scrub-bird, in particular, is unequivocal. Miocene divergence between the two extant species of Atrichornis, coupled with the Oligocene origins of the genus (Oliveros et al., 2019), emphasizes the conservation importance of both the rufous and noisy scrub-birds. They are arguably the most endangered and evolutionarily distinctive relict taxa in the entire Australian avifauna. Although having merit as an interim measure, prevalent usage, however, maintains some ongoing inconsistency in recognizing subspecies or species rank among CEAR isolates. For example, despite splitting from their respective sister lineages over a comparable time frame, the paradise riflebird (Ptiloris paradisaeus) is ranked as a species, whereas the plumed frogmouth (Podargus o. plumiferus) is ranked as a subspecies (Table 1). In any case, both the plumed frogmouth and Coxen's fig-parrot are disjunct southern isolates deserving of recognition (Higgins, 1999), possibly as evolutionarily significant units (sensu Moritz, 1994). However, we suggest that the existing subspecies designations are still the most appropriate and will remain so at least until more extensive taxon and population sampling, including sampling of the nuclear genome, can broaden our perspective on their respective histories. There is further justification for maintaining prevalent usage of these two taxa at subspecies rank. Recognition at species rank would render their respective closest relatives in northern Australia and New Guinea paraphyletic and require more extensive taxonomic revisions. Those revisions, in turn, need to be based on more extensive population and nucleotide sampling that would involve extensive sampling of museum skins.
Finally, we highlight how our study relates to current understanding of the molecular phylogeography and divergence times in other elements of the CEAR biota. Figure 3 of Bryant & Krosch (2016) summarizes divergence times among a range of animal and plant taxa from eastern Australian mesic habitats. In their review, divergence times involving three biogeographical barriers pertinent to the isolation of CEAR taxa (Brisbane Valley Barrier, St Lawrence Gap and Burdekin Gap) fell broadly into the two main categories of divergence times that we have used to structure our study, either Plio-Pleistocene or at least Miocene. Notably, just as the rufous scrub-bird and logrunner in our study were exceptional among our focal bird lineages in having a Miocene origin, so too do other (but notably few) elements of the eastern Australian mesic biota reviewed by Bryant & Krosch (2016). We suggest that these results further emphasize the value of the framework we have adopted as a useful approach to understanding the history of the CEAR biota.

concluSion
Our findings further illustrate the complex history and varying composition of all the Australian rainforest biota. Our results are consistent with the existence of two broad historical elements in the endemic bird fauna of the CEAR, and particularly highlight the prevalence of more recent divergences among lowland biogeographical elements in shaping the modern endemism of the ecosystem. Our results regarding fig-parrots and frogmouths point to Pleistocene connectivity among the CEAR habitats (see also Joseph et al., 1993), and potentially also New Guinea to the north (Schweizer et al., 2015). At least for birds, older relicts having Miocene affinities are a less dominant component of the endemic biota. We also highlighted the unexpected and possibly more complex, multifaceted origins and/or demographic history of Albert's lyrebird from the CEAR. Across all taxa, however, there remains a striking signal of relictual diversity regardless of the timescale of isolation. This pattern highlights the overall sensitivity of rainforest systems to climatic shifts and associated habitat contractions. It underscores the looming conservation issues posed by the combination of a legacy of past clearance and an impending future of unprecedented rapid climatic change. These challenges have been brought into sharp relief by recent fires that decimated large areas of habitat for many of our focal species (Boer et al., 2020;Kooyman et al., 2020;Ward et al., 2020).
the Queensland Museum for access to specimens. We also thank Per Ericson for providing information about catbird and bowerbird node ages. Laboratory facilities were provided by the University of Adelaide's Australian Centre for Ancient DNA (ACAD) and the South Australian Regional Facility for Molecular Ecology and Evolution (SARFMEE). DNA sequencing services were provided by the Australian Cancer Research Foundation (ACRF) Cancer Genomics Facility, SA Pathology, Adelaide. We are grateful to two anonymous reviewers, whose constructive feedback improved our manuscript. Project funding was provided to P.

SUPPORTING INFORMATION
Additional Supporting Information may be found in the online version of this article at the publisher's web-site: Table S1. Details of draft references constructed using geneiouS reaD mapper. Table S2. Details of mitochondrial genome consensus sequences generated from genetic data produced in this study. Table S3. Partitions and substitution models used for phylogenetic analyses. Table S4. Passerine family-level supermatrix data information. Table S5. Passerine family-level supermatrix gene partition information. Table S6. Node age constraints used for BEAST analysis of passerine supermatrix. Figure S1. IQ-TREE results for frogmouths, with ultrafast (+NNI) bootstrap node support values (as percentages). The scale is in number of substitutions per site. Specimen numbers are given in bold for new sequences generated as part of this study (Table 1; Supporting Information, Table S2). GenBank accessions are provided for published data. Figure S2. BEAST maximum clade credibility (MCC) tree for frogmouths, with 95% highest posterior density (95% HPD) bars for node ages, posterior probability node support values (above node bars) and mean node ages (below node bars). Specimen numbers are given in bold for new sequences generated as part of this study (Table  1; Supporting Information, Table S2). GenBank accessions are provided for published data. Figure S3. IQ-TREE results for parrots, with ultrafast (+NNI) bootstrap node support values (as percentages). The scale is in number of substitutions per site. Specimen numbers are given in bold for new sequences generated as part of this study (Table 1; Supporting Information, Table S2). GenBank accessions are provided for published data. Figure S4. BEAST maximum clade credibility (MCC) tree for parrots, with 95% highest posterior density (95% HPD) bars for node ages, posterior probability node support values (black) and mean node ages (blue). Specimen numbers are given in bold for new sequences generated as part of this study (Table 1; Supporting Information,  Table S2). GenBank accessions are provided for published data. Figure S5. Passerine family-level supermatrix phylogeny. Left, IQ-TREE results with ultrafast (+NNI) bootstrap node support values (as percentages). Right, BEAST maximum clade credibility (MCC) tree with posterior probability node support values and 95% HPD bars for node ages. Key Menura, Atrichornis and Orthonyx lineages are highlighted in purple. Taxon labels include information on the number of nuclear (n) and mitochondrial (m) loci (Supporting Information, Table S4). Menuridae_8n5m is a composite of NCBI sequences for southern Australian Menura novaehollandiae. Figure S6. raxml results for lyrebirds, with bootstrap node support values (as percentages). The scale is in number of substitutions per site. Specimen numbers are given in bold for new sequences generated as part of this study (Table 1; Supporting Information, Table S2). GenBank/NCBI accessions are provided for published data.