Complete mitogenome of endemic plum-headed parakeet Psittacula cyanocephala – characterization and phylogenetic analysis

Psittacula cyanocephala is an endemic parakeet from the Indian sub-continent that is widespread in the illegal bird trade. Previous studies on Psittacula parakeets have highlighted taxonomic ambiguities, warranting studies to resolve the issues. Since the mitochondrial genome provides useful information concerning the species evolution and phylogenetics, we sequenced the complete mitogenome of P. cyanocephala using NGS, validated 38.86% of the mitogenome using Sanger Sequencing and compared it with other available whole mitogenomes of Psittacula. The complete mitogenome of the species was 16814 bp in length with 54.08% AT composition. P. cyanocephala mitogenome comprises of 13 protein-coding genes, 2 rRNAs and 22 tRNAs. P. cyanocephala mitogenome organization was consistent with other Psittacula mitogenomes. Comparative codon usage analysis indicated the role of natural selection on Psittacula mitogenomes. Strong purifying selection pressure was observed maximum on nad1 and nad4l genes. The mitochondrial control region of all Psittacula species displayed the ancestral avian CR gene order. Phylogenetic analyses revealed the Psittacula genus as paraphyletic nature, containing at least 4 groups of species within the same genus, suggesting its taxonomic reconsideration. Our results provide useful information for developing forensic tests to control the illegal trade of the species and scientific basis for phylogenetic revision of the genus Psittacula.


Introduction
Psittaciformes is a highly speciose order of class Aves containing 86 genera and 362 distinct species [1]. These intriguing birds are gifted with highly expanded brains [2], superior cognitive ability [3] and vocal communication skills [4]. These birds have been kept as pets since ancient times because of their beauty, charm, long life span and astonishing ability to imitate many sounds, including human speech [5]. Unfortunately, this has led to widespread increase in parrot poaching and organized illegal trade of these species the world over, making them the most threatened bird species in the world [6,7]. Plum-headed parakeet Psittacula cyanocephala, belonging to the genus Psittacula (consisting of 16 long tailed parakeet species of which 13 are extant and 3 extinct) [8], is endemic to Indian sub-continent [9]. This parakeet belongs to one of those bird species which have been widely impacted by illegal live bird trade [10]. Though systematic population assessment of P. cyanocephala has never been conducted, it is suspected that, its population is dwindling which can be attributed to ongoing poaching for its illegal trade and habitat destruction [11]. Besides, it is documented that the rapidly changing climate does contribute to the loss of endemic species [12]. Hence, the management and conservation of endemic species such as P. cyanocephala would provide an umbrella of protection for various species assemblages within its endemic range [13]. However, to adopt and devise a healthy conservation and management plan, we require complete/ more information regarding behaviour, genetic, physiological, and ecological understanding of P. cyanocephala. P. cyanocephala bears close morphological resemblance to Psittacula roseata in terms of distinctive head plumage and genetic proximity [8,14]. A number of studies on evolutionary, morphological, and systematic aspects of the Psittacula genus have highlighted taxonomic ambiguities such as the paraphyletic nature of certain species and cladding of the genus Tanygnathus within the genus Psittacula [8,[14][15][16]. Recently, Braun and coworkers reconstructed the Maximum Likelihood (ML) based phylogenetic tree of Psittacula and other closely related species using a HKY model of two genetic markers (mitochondrial cob and nuclear rag1 gene) and observed that the Asian genus Psittacula is paraphyletic [8]. To create monophyletic genera, Braun and coworkers proposed the recognition of genus Himalayapsitta Braun, 2016 for P. himalayana, P. finschii, P. roseata, and P. cyanocephala [8]. However, the suggested taxonomic revision needs more confirmation using complete mitochondrial genome data.
Mitochondrial genomes are extremely useful in phylogeny and population research of avian taxa because of their inherent properties like small genome size, very less or absence of recombination frequency, maternal inheritance along with highly conserved gene content and evolutionary rate [17][18][19]. Also, mitochondrial genomes are comparatively more conserved than nuclear genomes during transition events in the evolutionary context, especially in birds [20]. Hence, such unique architecture, organization as well as evolutionary behavior render mitochondrial genomes the ability to carry phylogenetic information more consistently in comparison to nuclear markers [20]. Birds, however, are particularly noteworthy because their mitochondrial genomes are characterized by a gene order different from that in the majority of vertebrate mitogenomes due to rearrangement near the control region [21]. Although, the mitochondrial genome evolves about 10 times faster than the single-copy fraction of the nuclear genome [22], mutations in mitochondrial DNA are largely point mutations with only a few insertions or deletions. Furthermore, it has been reported that complete mitogenomes retain more information than a single gene regarding the evolutionary history of a taxon and provide consistent results compared to nuclear genes [20,23]. This reduces the effect of homoplasy and frequent stochastic errors in phylogenetic studies [23]. Considering the usefulness of complete mitogenome in phylogenetic and evolutionary studies, we attempted to decode the complete mitogenome of P. cyanocephala using Next Generation Sequencing (NGS). Furthermore, we validated NGS data using Sanger sequencing, which is considered as a gold standard for DNA sequencing, for select protein-coding genes to rule out any possible sequencing error.
We successfully decoded the complete mitochondrial genome sequence of P. cyanocephala for the first time. The newly sequenced mitogenome was validated and compared with other available mitogenomes of related genera to 1) know structural and genomic features of P. cyanocephala mitogenome and its relatedness with other available Psittacula mitogenomes, 2) identify codon usage and selection pressure on the protein-coding genes (PCGs) using bioinformatics and evolutionary analysis across the numerous parrot genera, and 3) build and elucidate comprehensive phylogeny relationships using all the available mitogenomes including that of P. cyanocephala.

Library preparation
For library preparation, 1.1 micrograms of extracted DNA was utilized as starting material for TruSeq DNA PCR-Free library preparation kit (Illumina Inc., USA). The DNA was fragmented using focused ultrasonicator (Covaris M220, USA) to the desired length of 350 base pairs as per the protocol recommended by the manufacturer. The size of the fragmented DNA population was checked using Fragment Analyzer (Agilent, USA) and made sure the majority of fragment sizes were in the desired range. Following TruSeq DNA PCR-Free library preparation kit protocol, clean-up of the fragmented DNA was performed using magnetic beads provided in the kit. The process of fragmentation creates overhangs at 3' and 5' ends of the input DNA. These ends were converted into blunt ends using the kit protocol. After the end repair, the appropriate library size was selected using different ratios of sample purification beads provided in the kit. The 3' end of the blunt fragments was adenylated with a single "A" nucleotide to prevent them from ligating to each other. Adapter ligation was then carried out following the kit protocol. After completion of library preparation, the mean peak size of the library prepared was checked using Fragment Analyzer. Quantification of the prepared library was carried out using QIAseq Library Quant Assay Kit (Qiagen N.V., Germany). The library was sequenced using NextSeq550 (Illumina Inc., USA) and at the end of the sequencing run, highquality paired-end reads were obtained.

Mitogenome assemblage
FastQC [http://www.bioinformatics.babraham.ac.uk/projects/fastqc/] was used to check the quality of the sequence data. At first, the de-novo assembly of the genome was attempted using SPAdes [26]. However, due to low coverage (~4.7X) of the whole genome of P. cyanocephala, successful retrieval of complete de-novo mitogenome sequence wasn't possible. Hence the reads were corrected using SPAdes and then mapped directly onto the reference genome (Psittacula roseata (NC045379.1)) using Bowtie2 [27]. A very high coverage (~90X) of the mitogenome was obtained using this approach. The final alignment was visualized using IGV software [28].

PCR amplification and sequencing of the protein coding genes using Sanger sequencing
Select protein-coding genes (cox1, cox2, atp8, atp6, nad1, nad2, nd3 and cob) of P. cyanocephala mitogenome were amplified to verify the mitogenome generated using NGS data. The genes were amplified using published primers (S1 and S2 Tables) designed by our lab [29]. Desired amplicons were amplified using thermocycler as per the S1 and S2 Tables. The PCR amplified amplicons were gel purified, and cycle sequenced with Big Dye Terminator ver. 3.1 (Applied Biosystems, Foster City, CA) using Applied Biosystems Genetic Analyzer 3500 (Applied Biosystems, Foster City, CA). Difficulty was faced during cycle sequencing of certain templates due to possible secondary structure formation in respective PCR amplicons. Various PCR adjuvants such as betaine and formamide in suitable concentrations and volumes were used to sequence the templates successfully.

Gene annotation
The complete mitogenome of P. cyanocephala was annotated manually. Using MitosWebserver all the PCGs, rRNAs and tRNAs were predicted and used as a template for annotation of the newly assembled mitogenome [30,31]. The tRNA positions and secondary structures were predicted using tRNAscan-SE2.0 [32], verified with the MITOSWeb-server results and carefully annotated on the newly sequenced P. cyanocephala mitogenome. The rRNA genes were initially predicted by MitosWeb-server and the boundaries were confirmed by aligning the P. cyanocephala mitogenome with other Psittacula mitogenomes [P. alexandri (NC045378.1) [33], P. derbiana(NC042409.1) [34], P. eupatria(NC042765.1) [34], P. krameri (MN065674.1) [35] and P. roseata (NC045379.1)] from GenBank (Table 1). Protein coding regions (open reading frames (ORF)) were identified in the sequenced P. cyanocephala mitogenome by using NCBI ORF FINDER [36]. The identified regions were translated using ExPASy-Translate Tool [37] with settings for the vertebrate mitochondrial genetic code. ExPASy-Translate Tool was used to check each ORF region for a protein-coding gene (PCG) with correct start and stop codons The formula "AT-skew = (A-T)/(A+T)" and "GC-skew = (G-C)/(G+C)" was used to calculate nucleotide composition skew [38]. A circular genome map was generated with the help of the CGView Server and edited manually [39].  (Table 1). These genomes were analyzed along with the newly sequenced mitogenome of P. cyanocephala.

Codon usage and evolutionary analysis
CodonW was used for the detailed codon usage analysis [40]. This analysis includes the percentage of AT and GC, nucleotide bias at the third position of codons, frequency of optimal codons (Fop), Effective number of codons (ENc) and Relative Synonymous Codon Usage (RSCU). Overall codon and amino acid usage of the newly sequenced mitogenome was represented by roseplots. Comparative codon and amino acid usage among 45 select mitogenomes were represented by heatmaps. Roseplots and heatmaps were generated using R software [41].
Evolutionary constraints on the individual protein-coding genes in terms of selection pressure were estimated by the ratio of non-synonymous substitution to the synonymous

PLOS ONE
substitution rate (dN/dS). Maximum likelihood approach was adopted for this analysis, where dN/dS<1 indicates neutral or purifying selection and dN/dS>1 indicates positive/ Darwinian selection/ mutational pressure [42]. PAL2NAL program embedded in Phylogenetic Analysis by Maximum Likelihood (PAML) package was used for this analysis. We exploited the standalone Linux version of this server [42]. The protein-coding genes of P. cyanocephala were assessed against the protein-coding genes of other investigated genomes as mentioned earlier.
The codons used in mRNAs were aligned and were subjected to dN/dS analysis. The orthologous pair of gene sequences was given as input file and the dN/dS value for that gene was calculated.

Genetic divergence and phylogenetic analysis
Divergence analysis is one of the most popular ways to estimate the cumulative differences among closely related members of a species that separated geographically leading to speciation [43] (allopatric or peripatric speciation). Pairwise base substitutions per site can provide an idea about the genomic distance and evolutionary history amongst the investigated genomes. Both species-level and genus-level divergence analysis was performed. For species-level analysis, each genome was taken as a separate entity while in genus-level analysis members of the same genus were considered as one group. For genus-level analysis members of the same genus were partitioned into a distinct group in MEGA X and average distances between the groups (i.e. generas) was calculated. The species-level analysis elucidates genetic distances at an individual level whereas the genus-level approach helps us to understand the average genetic distances between the grouped taxon from a different point of view. Thus, all considered organisms were clustered into 28 different groups which corresponds to 28 genera. The divergence among these groups was calculated using the Jukes-cantor model in MEGA X software [44]. For the construction of the phylogenetic tree, all the selected sequences were concatenated to 13 PCGs. Species across 28 genera of parrots and parakeets were used for tree construction. Emphasis was given to genus Psittacula, Tanygnathus and Eclectus during phylogenetic analysis. In total, 45 mitogenomes were concatenated to their 13 PCGs and aligned using the MUS-CLE alignment program. Phylogenetic tree analysis was carried out by Maximum likelihood (ML) and Bayesian inferences (BI) algorithms following two strategies. At the first instance, a best-fit nucleotide substitution model of TVM+F+G4 was identified for our data set using the Akaike information criterion (AIC) and Bayesian information criterion (BIC) in ModelFinder [45]. ML tree was constructed through IQ-TREE [46] version 1.6.12 using the TVM+F+G4 model with 10,000 bootstrap replicates. For BI analysis, GTR+I+G model was selected from BIC scores in ModelFinder by following previous studies [30]. Mr.Bayes [47] software was used to perform BI analysis following four independent chains running for 100000 generations, sub-sampling every 1000 generations and using a burn-in of 100 generations. FigTree [48] ver1.4.4 was used to edit the resulting phylogenetic trees. The convergence and mixing of Bayesian Markov chains was assessed by calculating the average standard deviation of split frequencies along with effective sample size of the trace respectively.

Mitogenome organization and gene arrangement
The complete mitochondrial genome of P. cyanocephala was assembled and submitted to Gen-Bank (Accession No. MT433093). Sanger sequencing of eight PCGs (cox1, cox2, atp8, atp6, nad1, nad2, nd3 and cob) showed 99-100% similarity with NGS data. We considered sequence similarity of more than 99% as robust enough for sequence verification in this study. A total of 6595 base pairs comprising the eight PCGs were Sanger sequenced, which makes 38.86% of the total P. cyanocephala mitogenome. The mitogenome was 16,814 base pairs long, which is in agreement with the average size of the complete mitogenomes (16827 bp, Table 1) of other Psittacula species. The mitogenome of P. cyanocephala is composed of 13 protein-coding genes (PCGs), 22 transfer RNAs (tRNA), 2 ribosomal RNA (rRNA) and a mitochondria control region (or D-loop) ( Table 2) reported earlier in other birds [1,30,49,50,52]. Majority of the genes including 12 PCGs, 14 tRNA, 2 rRNAs and the D-loop were located on the heavy (H or +)

PLOS ONE
strand, whereas 8 tRNAs (trn Ala , trn Cys , trn Glu , trn Asn , trn Pro , trn Gln , trn Ser (AGU) and trn Tyr and 1 PCG (nad6) were located on the light (L or -) strand (Fig 1). Earlier workers have reported similar arrangement of genes in parrots [1,49,50] and other birds [30,51,52]. The total base composition of P. cyanocephala mitogenome was A-5365 (31.9%), T-3728 (22.2%), G-2216 (13.2%) and C-5505 (32.7%). The A+T (54.08%) content was higher than the G+C content (45.92%, Table 3). High A+T content is typical of most vertebrate orders, including the members of Psittaciformes order and certain other birds [30,51,52]. The overall A-T and G-C skew were 0.1800 and -0.4259, respectively (Table 3), indicating higher A nucleotide than T nucleotide, and higher C nucleotide than G nucleotide, respectively. The A-T and G-C skew values also indicated that Psittacula species are having higher A and C content than T and G as reported earlier in certain other birds [30,51]. The skew values of A-T and G-C revealed in P. cyanocephala are nearly identical to what is reported in other species of Psittacula genus. The A-T and G-C skew values of P. alexandri, P. derbiana, P. eupatria, P. krameri and P. roseata were in the range of 0.1588 to 0.1836 and -0.4271 to -0.4151, respectively (Table 4).

Protein-coding genes
The total length of all the PCGs of P. cyanocephala was 11,400 base pairs constituting about 67.8% of the entire mitogenome. The occurrence and sequence of all the PCGs of P. cyanocephala mitogenome are following other avian mitochondrial genomes [30,51,52]. The entire length of all the PCGs translated into 3789 amino acid coding codons excluding stop codons. 11 PCGs of the P. cyanocephala mitogenome initiated with the start codon ATG, i.e. nad1, nad2, cox2, atp8, atp6, cox3, nad4l, nad4, nad5, cob, and nad6. Exceptions were cox1 and nad3 initiated with a start codon GTG and ATA, respectively. TAA was the most prevalent stop codon being punctuated in sequences of nad1, nad2, cox2, atp8, atp6, cox3, nad3, nad4l, nad4, nad5 and cob. The stop codon TAG was associated with the PCG nad6, whereas the cox1 gene had a stop codon of AGG. The TAA stop codons of nad2, cox3, nad3 and nad4 genes are partially incomplete and were completed by adding of polyadenylated tails at the 3' ends. Overlapping of sequences was observed only in PCGs of atp8, atp6 (10 nucleotides) and nad4l, nad4 (7 nucleotides). Many of the codon features mentioned here found to be similar across various avian orders [53]. However, within the genus Psittacula, start codons remained almost the same for respective genes irrespective of the species. ATG was the most frequent start codon of the PCGs in all Psittacula species investigated in this study (P. cyanocephala, P. alexandri, P. derbiana, P. eupatria, P. krameri and P. roseata). The exception of cox1 (start codon GTG) and nad3 (start codon ATA) genes, also holds across all the Psittacula species investigated so far.
In the case of stop codons, TAA was the most prevalent stop codon followed by TAG, with the cox1 gene punctuated by AGG stop codon across all the Psittacula species studied so far. The sequence overlaps of atp8, atp6 and nad4l, nad4 genes is common amongst all the Psittacula species analyzed in this study.
Codon usage analysis revealed that either adenine or cytosine was generally at the third positions of the codons. In P. cyanocephala, A3 (0.40) dominated over C3 (0.38) (Fig 2A). This was also validated with codon usage-based roseplots (Fig 2B). For other genera, except Agapornis, adenine and cytosine were dominating at the third position of codons. Agapornis showed a C3 bias at codons' third position. The amino acid usage-based heatmap along with roseplot of amino acid of P. cyanocephala (Figs 2C and 3B) showed Leucine, Isoleucine, Proline, Serine and Threonine were dominant than other amino acids in the mitogenome of P. cyanocephala. The effective number of codons (ENc) using plot analysis (Fig 2E), revealed that all the investigated mitogenomes were well below the curve of selection pressure. The ENc plot primarily reflects the mutational bias (Fig 2E). Plotting ENc against the GC3 thus provides an evidence of whether mutational pressure/natural selection is acting on the PCGs of an organism. An umbrella-line is drawn with the "expected ENc value" (value assuming only mutational pressure is acting on the considered genes) and is compared to the observed ENc values. Several previous reports have stated, if the observed ENc values exceeds the expected ENc values it will depict complete mutational pressure on the respective genes [54,55]. However, if the observed value is less than the expected value it is due to the selection pressure, lowering the effective number of codons. In the Fig 2E, all the plotted points have placed below the umbrella-line indicating the selection pressure over mutational bias on those genes. This clearly indicated that all investigated mitogenomes were translationally efficient, and natural selection was playing a crucial role on these genomes. From RSCU analysis, a set of optimal codons among the studied mitogenomes was identified. These optimal codons were-GCC  (Fig 2D and S3 Table). Heatmaps based on codon usage (Fig 3A) showed CUA (L), AUC (I), ACC (T), CUC (L), UUC (F) and ACA (T) to be the most used codons among the select mitogenomes. Amino acid usage heat-map also showed Leucine, Isoleucine and Threonine as the most-used amino acids ( Fig  3B).

Ribosomal and transfer RNA genes
Two ribosomal RNAs were packaged into the mitogenome of P. cyanocephala as in most vertebrates. The smaller 12s rRNA gene comprised of 976 base pairs with an A+T content of 50.72% and the larger 16s rRNA gene comprised of 1584 base pairs with an A+T content of 52.08% (Table 3). The rRNAs were nestled between tRNA Phe and tRNA Leu , with tRNA Val separating 12s and 16s RNA. The A+T content in 12s rRNA and 16s rRNA of other Psittacula species compared in this study were in the range of 49.8-51.13% and 51.5-52.5% respectively, similar to the nucleotide contents of the P. cyanocephala rRNAs. Birds are generally characterized by similar positioning of and high A+T content in their rRNAs, reported across many

PLOS ONE
avian orders [51,56]. As such the rRNAs of the Psittacula species compared in this study displayed similar positioning and characteristics of rRNAs typical of avian orders. As in other avian species, 22 tRNAs were identified in the mitogenome of P. cyanocephala in this study and their secondary structures were interpreted. The length of the tRNAs varied from 64 base pairs to 74 base pairs ( Table 2). The total concatenated length of all tRNAs present in the mitogenome was calculated to be 1486 base pairs. The A-T skew was calculated at 0.1821 and G-C skew at -0.2157, indicating higher A (33.8%) content than T (23.4%) and higher C (26%) content than G (16.8%) ( Table 3). The secondary structures of all the tRNAs displayed a cloverleaf model; the exception was tRNA Ser (GCU), which was missing the entire dihydrouridine arm (Fig 4). One of the key features of tRNA secondary structure is the presence of wobble base pairing. This can often substitute the Watson-crick base pairing and provide thermodynamic stability to tRNA. An extensive study on the tRNAs is crucial for the proper understanding of functional and structural features of mitogenomes. Highest wobble base pairing was found in trn Gln followed by trn Pro and trn Cys . Both trn Gln and trn Pro had three wobble base pairings each in amino acid acceptor arm, DU stem and anticodon stem. However, it was also found trn Gln had an extra wobble base pairing on TψC stem. trn Cys had wobble

Mitochondrial control region
Structurally, the D-loop of P. cyanocephala sequenced in this study consisted of 1317 base pairs similar to other birds, and located between tRNA Glu and tRNA Phe . The D-loop consists of A (26.9%), C (26.4%), G (15.2%) and T (31.5%) indicating very high A+T content (Table 3). High A+T content was also noticed in the D-loop region of the other Psittacula species and in other birds [30,51,52].
The mitochondrial D-loop is a non-coding region mostly involved in initiation, replication, and regulation of transcription related activities in vertebrate mitogenomes [1]. The D-loop though hypervariable in sequence, is conserved in very selective fashion within or across various orders of vertebrates. As vertebrate mitochondrial genome is believed to evolve under strong selection pressure, various gene orders around the mitochondrial control region (CR) are believed to be dynamically inherited, duplicated, or degenerated during genome evolution [1]. Psittaciformes is a species-rich and taxonomically confusing taxa [1] and hence, several studies targeting the mitochondrial CR region of Psittaciformes have been undertaken to answer wider phylogenetic and taxonomic queries [50]. As such a typical (or ancestral) avian gene order varies from a typical vertebrate gene order [21]. However, previous studies have identified at least 7 re-arrangements of duplicated ancestral gene orders in some of the Psittaciformes species [1].
We report that the D-loop region of P. cyanocephala mitogenome displays the ancestral avian CR gene order. This typical ancestral gene order is also displayed by other Psittacula species (P. alexandri, P. derbiana, P. eupatria, P. krameri and P. roseata) compared in this study. It is believed that the duplicated CR region provides a selective advantage to size and energy metabolism efficiency [1]. Hence, parrots with duplicated CR regions can live long while supporting large body mass such as Amazona parrots [1]. Parakeets of Psittacula genus are small to medium-sized (~40 cm) and have a lower expected life span than that of the Amazona genus parrots. Such a life history trait can be explained by the presence of the ancestral CR gene order in Psittacula genera, assuming duplications and pseudogenization have resulted avian CR rearrangements [1,51].

Genetic divergence and phylogenetic analysis
The evolutionary rate of the vertebrate mitochondrial genome is rapid when compared to the nuclear genome while having a relatively stable genome structure and recombination ratio [50]. Hence, mitochondrial genome sequences are widely used to infer phylogenetic relationships as they offer small stable changes over a long period in any given taxa. In this regard, whole mitochondrial genomes relay better phylogenetic information than single-gene (nuclear/mitochondrial) phylogeny. To elucidate phylogenetic relationships in parrots/parakeets with a focus on Psittacula genus, nucleotide sequences of 13 PCGs concatenated from the whole mitochondrial genomes of 28 genera belonging to 6 parrot families were used in this study ( Table 1).
The pairwise genetic divergence calculated for each individual genome and when grouped as a genus correlated well while clustering in the phylogenetic tree. The genus-level divergence analysis showed that the Psittacula genus was placed closest to Tanygnathus (0.08 units) then to Eclectus (0.09 units) and showed the highest divergence with Agaporins, Rhynchopsitta, Primolius, Cacatua and Nestor genus (0.19 units) (S4 Table). This result corroborates the previous report that Psittacula genus clusters with Tanygnathus and is closest to the Eclectus genus as a parallel-group [8,35]. The species-level (mitogenome based) divergence analysis elucidated that P. cyanocephala is closest to P. roseata (divergence of 0.06 units) than to any other Psittacula species (S5 Table). Divergence of P. cyanocephala with P. eupatria, P. krameri, P. alexandri and P. derbiana was calculated at 0.10, 0.12, 0.12 and 0.11 units respectively (S5 Table).
The phylogenetic tree obtained through both ML ( Fig 5) and BI (S2 Fig) analysis reveals the same topologies for the Psittacula genus. Species were clustered in respective clades following their genus level separation. The genus Eclectus, Psittacula, Tanygnathus and Prioniturus have originated from a nearest common parrot ancestor. P. cyanocephala and P. roseata have branched together from a common node that runs parallel to other Psittacula and Tanygnathus species. As mentioned earlier this makes a strong case for the creation of monophyletic taxa of P. cyanocephala and P. roseata, which has been indicated by previous studies on single mitochondrial (cob)/nuclear (rag-1) markers [8,14]. Another important fact evident from the clustering and divergence analysis is that Tanygnathus genus clusters within the Psittacula species, indicating that Tanygnathus and Psittacula species originated from a common ancestor.
A possible explanation for the phylogenetic incongruence of Tanygnathus and Psittacula generas reported in this study might have resulted from hybridization events i.e. introgression (admixture between genetic material of two species) and/or retention of ancestral polymorphisms because of incomplete lineage sorting during speciation (ILS) [57]. The phenomenon of introgression and ILS have been immensely appreciated in avian genomic studies and highly regarded as a pervasive mechanism of adaptive evolution, especially in birds [57]. Incongruent phylogenetic trees arising as a result of introgression/hybridization incidences are a common phenomenon. Avian orders are especially intriguing with 16% of the total bird species being affected by introgression [58]. Previous reports have attributed conflicting topologies in gene tree to introgression processes in woodpeckers [59], darwin's finches [60], flycatchers [20] and various other bird species [58]. On the other hand from a population point of view, ILS have been reported to affect the entire phylogeny of the Neoaves as well as the Palaeognathae clade, which have confounded the estimation of their respective species tree [61]. To summarize, speciation level events are notoriously hard to snapshot and hard bifurcations in gene trees difficult to distinguish against the phenomenon of introgression and/or ILS. Hence, the clustering of Tanygnathus genus within Psittacula complex may be attributed to the above mentioned population level incidences and the categorization of which is beyond the scope of our study [20,61]. Different genomic entities (coding regions, non-coding regions, mitochondrial DNA, nuclear DNA) vary structurally, hence undergo different evolutionary constraints [61]. Phylogenetic analysis associated with varying genomic loci may produce discordance in their respective phylogenetic trees [20,61]. If the genomic loci is under evolutionary constraint, the phylogenetic signal often gets masked with homoplasmy, multiple substitutions on the same site, and heterogeneous base composition [61]. In such cases whole mitogenomes presents itself as an excellent candidate for phylogenetic analysis. Ideally in absence of such evolutionary constraints (e.g. positive selection) mitochondrial DNA and Wlinked loci (both female transmitted) are least less likely to introgress as compared to nuclear DNA loci (e.g. Z-linked loci) during hybridization [20,61]. Previous studies have shown levels of mitochondrial DNA introgression to be nearly zero in avian hybrid zones, reiterating the fact that mitochondrial DNA are highly conserved across speciation events [20,61]. In the most likelihood gene trees based on whole mitochondrial genome of P. cyanocephala is expected to ditto the true species phylogeny of Psittacula genus even in incidences of introgression and ILS.
From the pairwise divergence calculations, we observed that species belonging Psittacula genus are more closely related to Tanygnathus than to other Psittacula species. For instance P. cyanocephala is more closely related to T. lucionensis (0.11 units) than to P. krameri (0.12 units). Also, the phylogenetic trees obtained through ML and BI analyses display clustering of 4 separate groups within the trees. These groups include 1) P. cyanocephala, P. roseata; 2) P. alexandri, P. derbiana; 3) T. lucionensis, and 4) P. eupatria, P. krameri. These clustering pattern are supported by perfect nodal values (100%/100%) obtained in both ML and BI trees. Ambiguity in pairwise divergence scores and nestling of Tanygnathus within Psittacula genus complex makes a strong case for its taxonomic reconsideration. Previous studies of Braun [8] and Kundu [62] used morphological markers, genetic data and molecular dating approaches to present a near complete phylogeny of Psittacula genus. The unavailability of complete mitogenomes of all Psittacula and allied species, curtailed our data set and limited the scope of our phylogenetic analysis. However the phylogenetic results obtained in this study exactly mirror the incongruences presented by previous studies [8,62]. Such genetic and evolutionary evidences weigh in support of Psittacula being is a non-monophyletic genus and warrants taxonomic reconsideration.

Evolutionary analysis
dN/dS ratios were calculated to evaluate the effect of purifying selection via mutational pressure over protein-coding genes of P. cyanocephala. dN/dS values of all protein-coding genes were found to be less than 1 indicating the presence of purifying selection pressure on those genes (S6 Table). These values varied from 0.08 (nad1, nad4l) to 0.14 (nad6, nad5 and nad4) suggesting differential selection constrains among genes. The lowest dN/dS values of nad1 and nad4l indicated the strongest purifying selection, whereas, nad6, nad5 and nad4 were under the least selection pressure. Thus, natural selection (purifying selection pressure) is acting as one of the major indices governing the evolution of Psittaciformes.

Conclusion
We provide for the first time, the high-quality complete mitogenome of P. cyanocephala, a parakeet endemic to the Indian Subcontinent, whose existence is threatened by illegal bird trade. P. cyanocephala mitogenome consists of 37 genes and the gene arrangement shows conserved patterns like other parrots. We also report that D-loop region of P. cyanocephala mitogenome displays the ancestral avian CR gene order. Codon usage analysis revealed that third positions of the codons are dominated by either adenine or cytosine. The dN/dS results indicated that nad1 and nad4l are under the strongest purifying selection whereas nad6, nad5 and nad4 are under the lower selection pressure in P. cyanocephala. Furthermore, phylogenetic results provided concrete evidence of multiple clades/groups clustering within the Psittacula genus supporting the call for a taxonomic reconsideration of the Psittacula genus. The complete mitogenome sequence of P. cyanocephala provided here will help in future phylogenetic and phylogeography studies as well as increase our understanding of the evolutionary relationships of such endemic species. Furthermore, mitogenome data will be instrumental in designing forensic tools for improving the conservation efforts and preventing the illegal trading of parakeet species.
Supporting information S1