Organization and Structure of the Pongamia Chloroplast Genome

The Pongamia cpDNA (Fig. 1) has a length of 152,968 bp; repeats IRA and IRB of 25,528 bp, a LSC region of 83,401 bp and a SSC region of 18,511 bp. This genome encodes 135 genic features, of which 94 are single copy genes and 16 are duplicated genes on the inverted repeats. The remaining nine features are comprised of two copies of the ycf2, ycf15 and ycf68 pseudogenes (a copy of each pseudogene appears on both IRA and IRB), two copies of an IR-spanning trans-spliced gene, rps12, and a partial copy of the ycf1 gene present as a pseudogene on the end of IRA (the intact ycf1 gene spans the junction between the IRB and SSC). We noted four distinct rRNA genes, 30 distinct tRNA genes and 77 distinct protein-coding genes. Four genes, infA, sprA, rpl21 and rpl22, are commonly lost from angiosperm cpDNA and these were notably absent from the Pongamia cpDNA. When the second inverted repeat is taken into consideration, a further six protein-coding genes, four rRNA and seven tRNA genes can be added to the list. This gives a total of eight rRNA genes, 37 tRNA genes and 83 protein coding genes. The genome thus consists of 51.1% protein-coding, 5.9% rRNA and 1.8% tRNA sequences. The sequence “GGAGG”, the Shine-Dalgarno sequence, was also found upstream of ten distinct protein coding genes: rbcL, psaA, psaB, psbZ, psbD, rpoC2, psbJ, psbF, psaJ and ycf2. Additionally, 12 distinct genes contain introns, of which three (clpP, ycf3 and rps12) have two introns. Of these 12, three intron-containing genes (ndhB, rpl2 and trnA) are duplicated on the IRs, taking the total to fifteen. Although rps12 is a trans-spliced gene, it also contains an intron that resides on both IRA and IRB. When aligned, the Pongamia and L. japonicus cpDNA share a pairwise identity of 82.4%. In addition, as with other cpDNA, the Pongamia cpDNA is AT rich (i.e. the GC content is 34.8%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Map of the Pongamia cpDNA (152,968 bp).

The innermost circle indicates the locations of the IRs (IRA and IRB, 25,528 bp), which separate the LSC (83,401 bp) and SSC (18,511 bp) regions. Genes on the outermost of the map are transcribed in a counter-clockwise direction and those on the inside clockwise. The graphs plotted between these circles represent the percentage synonymous (blue) and non-synonymous (red) divergence of each protein-coding gene, between Pongamia and V. radiata, and, Pongamia and L. japonicus. The red arrow indicates the position of the replication origin, oriC. Genes marked with a red asterisk are genes in the Pongamia unique inversion event (the bounds of this region have been marked out in red). Genes underlined in red are pseudogenes.

https://doi.org/10.1371/journal.pone.0051687.g001

Structurally, the Pongamia cpDNA is similar to other legume cpDNA containing a set of IRs. It contains the 50 kb inversion that is unique in the evolution of the Fabaceae [7]. In addition, the Pongamia cpDNA also contains a small inversion of approximately 6.5 kb that appears unique when compared to other species of flowering plants (Fig. 2). This inversion reverses and complements about 6.5 kb across the second junction of the 50 kb inversion, leaving psaI, accD, rps16, trnQ and trnS to be transcribed in the opposing direction and psbK and psbI to be transcribed in the reverse direction, relative to the start of the LSC. Further, no disruption to the coding or regulatory sequences of the impacted genes has taken place.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Mauve alignment of the Pongamia cpDNA LSC to representatives of the flowering plants (Arabidopsis thaliana) and Fabaceae (G. max) showing a 50 kb inversion that is a unique event in the evolution of the Fabaceae (green and gold blocks) as well as a 6.5 kb inversion that is unique to the Pongamia cpDNA (gold and blue blocks).

Plots of similarity can be found inside of each element.

https://doi.org/10.1371/journal.pone.0051687.g002

A highly AT rich region between 14,400 bp and 14,600 bp downstream of the start of the LSC was identified as the genome’s origin of replication, oriC, by comparison with currently available origin of replication sequences in the NCBI databases. Additionally, the relative coverage of Pongamia cpDNA by DNA sequence reads during assembly (Fig. S1), shows significantly lower coverage in this location (Fig. S2). Although lower coverage at this position was also identified during the re-assembly of the L. japonicus cpDNA (Fig. S3), identification of an oriC or palindromic sequence was limited (Fig. S4). Despite this, a region believed to surround the L. japonicus oriC (between positions 14,500 bp and 14,900 bp; Fig. S5), was also AT rich.

Organization and Structure of the Pongamia and L. japonicus Mitochondrial Genomes

The Pongamia mtDNA (Fig. 3) and L. japonicus mtDNA (Fig. 4) were determined to be 425,718 bp and 380,861 bp in size, respectively. The Pongamia mtDNA contains two sets of inverted repeats and two sets of direct repeats (Fig. S6). The first set of inverted repeats, IR1A and IR1B, are each 6,229 bp whilst the second set of inverted repeats, IR2A and IR2B, are each 2,274 bp. The first set of direct repeats, DR1A and DR1B, are each 13,319 bp whilst the second set of direct repeats, DR2A and DR2B, are each 3,919 bp. It should be noted that IR1B and DR2A share an overlap of 669 bp and that this overlap may be of significance when considering the secondary structure and folding of this genome. Comparatively, the L. japonicus mtDNA is structurally much simpler than that of the Pongamia mtDNA. It contains a single set of inverted repeats, IRA and IRB, each of 4,460 bp and a single set of direct repeats, DRA and DRB, each of 18,971 bp (Fig. S7). Furthermore, the Pongamia mtDNA contains the same intron containing and trans-spliced protein-coding genes that could be found in the L. japonicus mtDNA.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Map of the Pongamia mtDNA (425,718 bp).

The innermost circle indicates the locations of two sets of inverted repeats (IR1A and IR1B, 6,229 bp; IR2A and IR2B, 2,274 bp) and two sets of direct repeats (DR1A and DR1B, 13,319 bp; DR2A and DR2B, 3,919 bp). Furthermore, IR1B and DR2A share an overlap of 669 bp. Genes on the outside of the map are transcribed in a counter-clockwise direction and those on the inside clockwise. The graphs plotted between these circles represent the percentage synonymous (blue) and non-synonymous (red) divergence of each protein-coding gene, between Pongamia and V. radiata, and, Pongamia and L. japonicus. Genes underlined in red are pseudogenes.

https://doi.org/10.1371/journal.pone.0051687.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Map of the L. japonicus mtDNA (380,861 bp).

The innermost circle indicates the locations of a set of inverted repeats (IRA and IRB, 4,460 bp) and a set of direct repeats (DRA and DRB, 18,971 bp). Genes on the outermost of the map are transcribed in a counter-clockwise direction and those on the inside clockwise. The graphs plotted between these circles represent the percentage synonymous (blue) and non-synonymous (red) divergence of each protein-coding gene, between L. japonicus and V. radiata, and, L. japonicus and Pongamia. Genes underlined in red are pseudogenes.

https://doi.org/10.1371/journal.pone.0051687.g004

The Pongamia mtDNA contains 71 features, of which there are 37 protein-coding genes, three rRNA genes, 24 tRNA genes and seven pseudogenes. We found 33 distinct protein-coding genes, three distinct rRNA genes, 22 distinct tRNA genes and six distinct pseudogenes. Hence, the Pongamia mtDNA contains duplicated genes that reside in repeat regions. Three genes, trnfM, sdh3 and atp1, reside in IR1A and IR1B, and three more, rps4, nad9 and trnK, reside in DR1A and DR1B. Although no genes were found to reside in IR2A and IR2B or DR2A and DR2B, an 87 bp fragment of rpl2 was found in IR2A and IR2B. It is also worth noting that none of the 71 features found in the Pongamia mtDNA were found to reside across any of the repeat region boundaries. When looking for the list of 17 genes frequently lost during angiosperm evolution, including rpl2, eight of these were either absent from the Pongamia mtDNA (rpl10, rps2, rps11, rps13) or were found as pseudogenes in various stages of attrition (rpl2, rps7, rps19, sdh4). Surprisingly, Pongamia sdh3 did not make the list of pseudogenes. It was found translationally uninterrupted and divergence from the L. japonicus mitochondrial pseudo-orthologue was calculated at 11.3%. In contrast, Pongamia sdh4 remains a pseudogene, the open reading frame is interrupted not only by a 439 bp fragment of cpDNA, but also by over 3 kb of flanking DNA of unknown origin. Additionally, we were able to locate a single rps7 fragment as well as two rps19 fragments, suggesting the transfer of these genes to the nucleus. We propose that one fragment represents the start of the rps19 gene and the other represents the end of the rps19 gene, as these sequences do not overlap. Comparative analysis also revealed numerous short fragments with strong similarity to known functional genes. While many of these could be considered illegitimate, a partial match (90 bp) to nad6 was detected. Interestingly, this fragment is not at all similar to an orthologous fragment found in the L. japonicus mtDNA. We also calculated a gene density of 8.0% protein-coding, 1.2% rRNA and 0.4% tRNA gene sequences as well as a GC content of 45.0%.

The L. japonicus mtDNA contains six fewer genic features than the Pongamia mtDNA, with three rRNA genes, 20 tRNA genes, 34 protein-coding genes and eight pseudogenes. Of the 34 protein-coding genes, 31 are unique. Hence, like the Pongamia mtDNA the L. japonicus mtDNA contains duplicated genes that reside in repeat DNA. However, the genes that can be found in the L. japonicus mtDNA are different to those found in the repeat regions of the Pongamia mtDNA. These genes, ccmFn, rpl5 and rps14, are each duplicated with single copies of each on DRA and DRB. These direct repeats also lead to partial duplication of cob. As such, whilst a complete 621 bp copy of cob is positioned across DRB, a cob pseudogene resides across the boundary of DRA. The inverted repeats also lead to partial duplication of atp6. Here, a complete 720 bp copy of atp6 resides across the IRA boundary, with an atp6 pseudogene across the IRB boundary. We also interrogated the L. japonicus mtDNA for a list of 17 genes (15 ribosomal protein and two respiratory) that are well understood to have been lost frequently during angiosperm evolution. Ten of these were either absent from L. japonicus mtDNA (rpl2, rps1, rps2, rps11, rps13) or were present as pseudogenes in various stages of attrition (rpl10, rps7, rps19, sdh3, sdh4). The rpl10 and sdh3 pseudogenes were the most intact, were both created by a single point mutation and each contain just one internal stop codon. Another gene, sdh4, is also very much intact; however, multiple frame-shift mutations have disrupted the reading frame and introduced a premature stop codon. A 333 bp fragment of rps7 was also located immediately downstream of ccmC. It contains an internal stop codon and lacks almost 40 codons from the 5′ end of the gene that are likely required for functionality. We also found eight intron-containing genes, of which three (nad1, nad2, nad5) are trans-spliced. Sensitive BLAST searches of this genome also revealed numerous short DNA fragments (<50 bp) with strong similarity to a number of genes. Another hit was to nad6, and this represents the eighth pseudogene. The L. japonicus nuclear genome BLAST databases [12] were also interrogated to confirm the absence of rps1. Our preliminary results conclude that rps1 has been transferred to the nuclear genome where a functional copy of rps1 exists. Furthermore, the genome was found to be almost 10% gene dense; it consists of 8.2% protein-coding, 1.3% rRNA and 0.4% tRNA sequences with a GC content of 45.4%.

RNA Editing, Codon Usage and Utility of Legume Organelle DNA

Codon usage and putative RNA editing sites of Pongamia and L. japonicus cpDNA and mtDNA were determined using the organellar genomes of Vigna radiata [10], [11] as references (Table S1). Using the CURE software, we were able to predict 556, 517 and 485 C-to-U RNA editing sites on gene coding sequences in the L. japonicus, Pongamia and V. radiata mtDNA, respectively. C-to-U RNA editing was predicted at the start of nad1, nad4L and rps10 as well as the stop of atp6, ccmFc and rps10 for all three species. This type of editing may also be required for the initiation of the Pongamia mttB. The mitochondrial nad5 of V. radiata is also reportedly initiated by C-to-U RNA editing; however we were unable to predict this site using CURE. During our analysis of RNA editing and codon usage it was necessary to adjust the start sites of rpl16 and mttB and stop sites of rps10 and atp6 of the V. radiata mtDNA. CURE-Chloroplast software was also used to predict 50 C-to-U RNA editing sites in the cpDNA of the same three species. Although RNA editing does not create any in-frame stop codons, psbL and ndhD often start at ‘ACG’. However, with a total of only 50 RNA editing sites, analysis of codon usage after RNA editing is limited (Table S2). As expected, the codon usage pattern of all three cpDNA was also found to be highly similar (comparison not shown).

Both before and after RNA editing, similarity was found between the codon usage patterns of all three mtDNA. When comparing the codon usage of cpDNA and mtDNA of Pongamia, both before and after RNA editing, most of the amino acids of the cpDNA that encode proteins had a stronger association for a particular codon or set of codons. However, when we look at the frequency of particular codons, some points of difference could be noticed. For example, when compared to the Pongamia mtDNA, the Pongamia cpDNA has a higher frequency of codons for glutamic acid. Yet, when we look at the frequency of codons for phenylalanine, for example, RNA editing increases the number of ‘UUU’ and ‘UUC’ codons on the Pongamia mtDNA by over 10%. Interestingly, the number of ‘UUU’ and ‘UUC’ codons created by RNA editing is similar to the number of like codons found on the Pongamia cpDNA. We also notice a preference of codons for leucine, and indifference for proline, arginine and serine.

Mapping transcriptomic data [13] to Pongamia organellar DNA has also provided evidence for the computationally predicted RNA editing sites (Table S3 and Table S4, respectively). This data was also used to confirm protein-coding gene splice-site junctions as well as to provide candidate chloroplast and mitochondrial genes for further gene expression studies involving leaf (Table S5) and root (Table S6) tissue, under stress conditions involving salt-water treatment.

Legume Organelle Protein-coding Gene Sequence Divergence

We measured the divergence of almost 120 organellar protein-coding genes from the three legume species (Table S7). Despite the structural fluidity of plant mtDNA and the frequent lack of structural conservation, synonymous and non-synonymous divergence of mitochondrial protein-coding genes was approximately 5.26- and 3.74-fold lower than the divergence of the chloroplast protein-coding genes, respectively. On average, synonymous divergence of the chloroplast genes of L. japonicus and Pongamia, Pongamia and V. radiata, and V. radiata and L. japonicus were calculated at 7.99%, 7.49% and 9.62%, respectively. Synonymous divergence of the protein-coding mitochondrial genes of the same groupings of species was calculated at 1.60%, 1.29% and 1.87%, respectively. These results simply indicate that chloroplast genes have a higher rate of nucleotide substitution. However, when focusing on non-synonymous mutations, of the cpDNA and mtDNA, a high degree of identity was noticed between L. japonicus – Pongamia (91.58% and 97.74%, respectively) and Pongamia – V. radiata (91.53% and 97.7%, respectively). When compared to these groups, non-synonymous identity of the cpDNA and mtDNA of the V. radiata – L. japonicus group was slightly lower at 89.91% and 97.37%, respectively. This of course suggests that the V. radiata and L. japonicus are the most divergent of the three species.

Phylogenesis and the Promiscuity of Plastid DNA

After compiling all the published organelle genome sequences of the Streptophyta, we present the relatedness of the Pongamia and L. japonicus cpDNA (Fig. S8) and mtDNA (Fig. S9). The cpDNA alignment makes obvious the separation of the Fabaceae into two groups based on the presence or absence of two copies of the inverted repeat. Whilst the mitochondrial phylogram appears to simply confirm the relatedness of the Fabaceae, it also suggests that the nuclear genomes of Glycine max, Phaseolus vulgaris, V. radiata and L. japonicus are worth investigating when attempting to study the Pongamia nuclear genomic environment.

The L. japonicus mtDNA contains six tRNA genes of chloroplast origin. Of these, there is one copy each of the aspartic acid, histidine, tryptophan, and methionine tRNA genes, and two copies of the asparagine tRNA gene. We found an identical set of tRNA genes in the Pongamia mtDNA, however, one copy of the tRNA for asparagine folds with a low Cove score, the product of which is most likely dysfunctional. BLAST was also able to identify four and ten other chloroplast-like features that appear to be present in both the Pongamia and L. japonicus mtDNA, respectively. Two of the ten fragments of cpDNA in the L. japonicus mtDNA have been incorporated into the genome’s direct repeat. Thus in total, the L. japonicus mtDNA comprised 1.52% cpDNA, whilst the Pongamia mtDNA is made up of just 0.70% cpDNA.

Similarity between the Pongamia and L. japonicus organellar repeat regions was also measured (Fig. S10). Additionally, the GC content of these repeats was also calculated. We found that while the IRs of the cpDNA were 41.6% GC, the LSC and SSC regions had lower GC content, at 32.2% and 28.0%, respectively. Interestingly, the Pongamia and L. japonicus mtDNA repeat elements had very similar GC content, of approximately 45%, when compared to their non-repetitive counterparts.

The SaSSY Assembler and Pipeline

De novo second-generation DNA sequencing of the Pongamia and L. japonicus organellar genomes was performed using samples of leaf-derived DNA (whole genomic DNA). We used the Illumina® GAIIx platform to generate 12,501,168 Pongamia paired-end sequences of 36 bp from genomic fragments of approximately 390 bp (Pon_03_001), as well as 29,474,558 mate-pair sequences of 75 bp with fragment lengths of approximately 3 kb (Pon_37_001). It should be noted that this large-insert library also contained a shadow (paired-end) library with a fragment length of approximately 220 bp. The same platform was also used to generate 65,930,582 L. japonicus paired-end sequences of 100 bp from genomic fragments of approximately 350 bp (LjDIMG_03_001). These three datasets have been deposited into the NCBI Short Read Archive (SRA) database under the accession number: SRA051251. Information regarding experimental methods, sample preparation and library construction can also be found under this resource.

SaSSY [4] (v0.1.1.3) is a short paired read de novo assembler written in C++ utilizing a novel set of algorithms. It was designed for the assembly of repetitive genomes, specifically plant BAC sequences. As an overlap-layout assembler, SaSSY avoids cutting reads into k-mers and building de Bruijn graphs [19]. Instead, SaSSY joins two reads by an edge if their offset (the length of the non-overlapping parts of two overlapped reads) is less than a specified maximum value. SaSSY is able to use paired read information to produce scaffolded contigs and has proved useful in a number of in-house projects. SaSSY has not been published but is freely available. For more information as well as a detailed description of the algorithm, please read the documentation available at http://sassy.mikeimelfort.com/.

We validated our assembly approach by re-assembling the L. japonicus cpDNA using the LjDIMG_03_001 paired-end dataset described above. SaSSY allows the user to alter two parameters; naïve and extension offsets which are used to determine the minimum overlap needed to assemble reads. To re-assemble the L. japonicus cpDNA, naïve and extension offsets of 2 and 6 were used respectively, as well as a readlength of 63 bp. These assembly parameters were chosen based upon recommendations in the SaSSY manual. When we aligned our re-assembly of the L. japonicus cpDNA to the published sequence [6], we found a single SNP (a ‘G’ to a ‘T’ substitution) at position 124,653. This change impacts ycf1 at the codon for the 409^th amino acid changing it from a threonine to an asparagine. Confident of the reliability of SaSSY, we assembled the L. japonicus mtDNA and the two Pongamia organellar genomes. To re-assemble the L. japonicus mtDNA, we used SaSSY with naïve and extension offsets of 8 and 12, and a readlength of 63 bp respectively. We used the same parameters and a readlength of 35 bp to assemble the Pongamia cpDNA. To assemble the Pongamia mtDNA, we used SaSSY’s default offsets of 4 and 12, however we first cut and filtered the raw data using the Perl script ‘cut_Down_ILL.pl’ included in the SaSSY package. This script first trims the reads and then cuts the resulting sequence into a number of short fragments resulting in an increased number of fragmented reads. The user must specify the trim length and the amount that the start of each new fragment is offset from the next. The 75 bp reads were processed two times using an offset of 4 bp, an “out” readlength of 35 bp and trim lengths of 36 bp and 32 bp resulting in 58,949,116 and 88,423,674, shorter fragments respectively. We then ran SaSSY twice with a readlength of 35 bp and in each run we added the ‘Pon_03_001’ dataset.

In order to further scaffold and complete the genome assemblies, it was necessary to determine the origins of the contigs; as either chloroplast-, mitochondrial-, or nuclear-like DNA sequence. We did this by comparison with the nucleotide sequence database (v4) at the NCBI using BLAST+ (v2.2.24) with default search parameters. We also interrogated our datasets with these contigs using MegaBLAST [20] (v2.2.21): `megablast -a 24 -F F -W 12`. For each read in a ‘hit’, we applied a minimum alignment filter of 60 to the L. japonicus and Pongamia mate-pair libraries and a filter of 30 to the Pongamia paired-end library using Perl (v5.10.1). Perl script was also used to generate insert-size distribution and coverage statistics which were in turn used to visually assess and validate the structural integrity of these assemblies allowing re-assembly as necessary. All single base assembly errors and small indels were corrected by mapping the reads back onto the assemblies using BWA [21] (v0.6.1) and SAMtools [22] (v0.1.18). Tablet [23] (v1.12.03.26) was used for viewing the mapping files. An in-house Perl script with the Bio::DB::Sam module was also used to refine the assembly. Geneious Pro [24] (v5.4.6) was then used to circularize and orientate the sequences, as well as to correct annotation errors. We also generated large contigs with Velvet [25] (v1.2.03) and mapped these with high accuracy to their respective genomes to confirm our SaSSY assemblies.

Gene Discovery and Annotation

The Pongamia cpDNA was annotated using DOGMA [26]. Nucleotide and CDS databases of flowering plant mtDNA were compiled, and the mtDNA assembled from this study were then interrogated against these databases to find protein and RNA genes using BLASTN and BLASTX. Blast2GO [27] was also used to find homologous sequences and confirm functional annotation. We considered annotation with a mean similarity greater or equal to 50% and an e-value smaller than 1E⁻²⁰ with five or more hits. We manually curated these annotation databases, removing any ambiguous, uninformative or inaccurate entries (including those classed as hypothetical or unclassified). To identify tRNA genes, tRNAscan-SE [28] (v1.12) and the available web server [29] (http://lowelab.ucsc.edu/tRNAscan-SE) were utilized. To predict the bulk of mitochondrial and chloroplast RNA editing (cytidine to uridine) sites the computational tools CURE [30] (http://59.67.33.228/biosrv/cure/, v1.1) and CURE-Chloroplast [31] (http://bioinfo.au.tsinghua.edu.cn/software/pure, v1.0) were used, respectively. The four available transcriptomic datasets (described above) were combined and re-mapped against each of the 110 unique coding-sequences for potential RNA editing sites. This was done using our in-house Perl script with the Bio::DB::Sam module. Repeat regions as well as the designated chloroplast origin were identified with the nucleotide sequence database (v4) at the NCBI using BLASTN with default parameters. Genome sequences and annotations for the Pongamia chloroplast and mitochondrion, as well as the mitochondrion of L. japonicus, can be found in the GenBank data library under accession numbers JN673818, JN872550 and JN872551, respectively. All GenBank files contain predicted RNA editing sites. Genome map construction was performed with OGDraw [32] and Circos [33].

Transcriptomic Mapping and Analysis

Four transcriptomic datasets were downloaded from the NCBI SRA under the accession number SRA046342. Each dataset was produced using the Illumina GAIIx platform and each consists of 24 million 75 bp paired-end reads taken from pooled samples of leaf or root tissue at two hours, four hours and eight hours under application of either fresh- or salt-water. These datasets were used to confirm predicted splice-site junctions of the Pongamia organellar whole genome sequences using TopHat [34] (v2.0.0). They were then re-mapped using BWA to a list of 110 unique protein-coding genes (deduced using methodology discussed above) taken from the two Pongamia organellar genomes. Using SAMtools, the mapping files were normalized to Reads Per Kilobase of exon model per Million mapped reads (RPKM) [35]. Unfortunately these datasets were not produced with any biological replication, thus statistical testing for the significance of the salt treatment was limited.

Phylogenesis and Divergence

The cpDNA of nearly 190 species of Streptophyta and the mtDNA of 36 species in the same phylum were aligned using the progressiveMauve [36] (v2.3.1) algorithm. The alignment output was used to create three rectangular phylograms with TreeGraph [37] (v2.0.47). Whilst the mitochondrial phylogram shows all 36 species, a subset of 23 cpDNA (the most closely related species with respect to Pongamia) were taken to improve tree readability. Mauve [38] (v2.3.1) was also used to analyse genome alignment output. Divergence percentages were calculated: 100% minus pairwise percent identity. ClustalW2 [39] (v2.1) output was parsed using Perl script to calculate pairwise percent identity. This was computed in the same way that Geneious Pro calculates pairwise percent identity: by looking at all pairs of bases at each column position and scoring a hit, divided by the total number of pairs, for each column when they are identical.