Introduction

With very few exceptions the plastid genomes (plastomes) of angiosperms typically contain 79 protein-coding genes, 30 tRNA genes and four rRNA genes (reviewed in Raubeson and Jansen 2005; Bock 2007). Broadly, genes are involved in photosynthesis or housekeeping. Plastome organization is identical in the majority of angiosperms: two single copy regions, the large (~85 kb; kilobase) and small (~15 kb) single copy regions (LSC, SSC), are separated by two large (~25 kb) inverted repeats (IR) designated IRa and IRb. The ribosomal operon lies within the IR. Plastid DNA appears to be present as isomers, differing in the orientation of the single copy regions (Palmer 1983). A few seed plant lineages have lost one copy of the IR, but the vast majority of angiosperms retain this quadripartite structure. Gene order is generally the same across angiosperms, most exceptions being inversions in the LSC as in the grasses, some legumes and sunflower. Very seldom do inversions interrupt conserved transcriptional units (Raubeson and Jansen 2005; Bock 2007).

Relatively few plastid genes have been functionally transferred to the nucleus subsequent to the diversification of angiosperms (Bock and Timmis 2008). Recent transfers or functional substitutions have been documented for only five genes thus far, but for a few of these, such as infA, there have been repeated, independent transfers (Millen et al. 2001). Certain ribosomal protein genes appear to have been lost and presumably functionally transferred independently across angiosperms (Jansen et al. 2007), but most other classes of plastid genes are seldom or never lost from photosynthetic seed plants. For example, with few exceptions, genes involved in photosynthesis, electron transport, and ATP synthesis have not been found to be missing from the plastome of a photosynthetic seed plant (Bock 2007; Magee et al. 2010). The 11 plastid-encoded NADH dehydrogenase (ndh) genes are unique in being lost or retained only as a full set.

The 11 plastid-encoded ndh genes are only commonly lost in nonphotosynthetic plants (Martin and Sabater 2010). In fact, only two photosynthetic seed plant lineages have been documented to lack these genes, Pinaceae/Gnetales (Wakasugi et al. 1994; McCoy et al. 2008; Wu et al. 2009) and a large clade within the Orchidaceae (Neyland and Urbatsch 1996; Chang et al. 2006; Wu et al. 2010). In these two lineages, as in parasitic plants, the entire suite of 11 plastid-encoded ndh genes is lost together. Both of these losses of plastid-encoded ndh genes are relatively ancient—the divergence of Pinaceae has been estimated at approximately 140 million years ago (MYA) (Wang et al. 2000), and loss of ndh genes necessarily predates this divergence, especially if it represents a synapomorphy for Pinaceae and Gnetales (Braukmann et al. 2009). The sister relationship between Pinaceae and Gnetales—the so-called “gnepine” hypothesis—is still controversial (Zhong et al. 2010). However, whether ndh gene loss in these lineages represents a single or two separate events, the loss is relatively ancient. Loss of ndh genes in orchids is likely also ancient—fossil-calibrated molecular data suggests a crown radiation of Orchidaceae 76-84 MYA (Ramirez et al. 2007). The taxonomic group confirmed to have lost ndh genes encompasses at least 70 orchid genera with over 1,000 species (Wu et al. 2010), suggesting that the loss of ndh genes is not recent.

It is not known whether Ndh function has been lost or functionally replaced in these two lineages or the genes have been functionally transferred to the nucleus; however, experimental evidence suggests that the thylakoid-bound Ndh complex is a vital intermediary of linear/cyclic electron flow and is apparently indispensable to photosynthesis under a wide range of stress conditions (Endo et al. 1999; Martin et al. 2004; Rumeau et al. 2005; Casano et al. 2001). Unfortunately, neither the composition nor the functions of the Ndh complex have been fully characterized. At least three additional subunits are located in the nuclear genome of angiosperms (Rumeau et al. 2005).

A few angiosperm lineages have been identified with highly rearranged plastomes lacking a variety of genes and introns. One of these lineages, the geranium family (Geraniaceae), contains four major genera, each with distinctly rearranged plastomes (Chumley et al. 2006; Guisinger et al. 2011). All members of the family share the loss of trnT-ggu and of two introns (rps16 and rpl16). At 217 kb, the garden geranium (Pelargonium x hortorum) has the largest and most rearranged plastid genome among angiosperms. Most of the expansion in the P. × hortorum plastid genome results from massive expansion of the IR into the single copy regions and from proliferation of complex repeats. While P. × hortorum has an IR triple the normal size for an angiosperm at 76 kb, the IR has been lost completely in Erodium texanum and reduced in Geranium palmatum (11 kb) and Monsonia speciosa (7 kb). Monsonia speciosa is the only land plant plastome in which the IR does not include the entire rRNA operon. Geranium, Monsonia, and Erodium all share the loss of the two large ycfs, ycf1 and ycf2, present in Pelargonium and in most angiosperm plastid genomes. Families of large dispersed repeats associated with rearrangement endpoints are found in the plastomes of all four Geraniaceae genera (Guisinger et al. 2011).

We chose to survey the genus Erodium (with 74 species) to improve our understanding of the extent and diversity of plastome abnormalities in a single lineage within Geraniaceae. The focus of this paper is to compare plastid genome organization of species representing the two major clades and to report the loss of plastid-encoded ndh genes in a single divergent clade of 13 species. This clade represents the most recent loss of ndh genes yet identified among photosynthetic seed plants. Because the subgenera and sections in Erodium have been found to be polyphyletic and have not yet been revised, this clade has no official taxonomic status (Fiz et al. 2006). Here we refer to the ndh-lacking clade as the long-branch clade (LBC), as it is separated from the rest of the genus by a long branch in phylogenetic reconstructions (Guisinger et al. 2008).

Materials and methods

Taxon sampling and sample preparation

Based on a molecular phylogeny of the genus (Fiz et al. 2006), E. carvifolium was chosen to represent Clade II—the published sequence of E. texanum represents Clade I (Guisinger et al. 2011)—and three additional taxa, E. chrysanthum, E. guicciardii and E. gruinum, were chosen from a group of species designated as the long-branch clade (LBC; Fig. 1). Protocols for plastid isolation have been previously described (Jansen et al. 2005). Five to 10 g of fresh young leaf tissue from a single plant was used in each plastid isolation; plants were obtained from commercial sources (Geraniaceae.com; B&T World Seeds, Paguignan, France) and cultivated in the greenhouses at the Brackenridge Field Laboratory at University of Texas at Austin. Vouchers are deposited at TEX.

Fig. 1
figure 1

Phylogeny of Erodium adapted from Fiz et al. (2006). Clades I, II, and the long-branch clade (LBC) are indicated. Numbers at nodes are Bayesian posterior probabilities as percentages. Taxa in bold represent those with complete (E. texanum, E. carvifolium) or draft plastid genome sequences. The asterisk indicates the new genome sequence of E. carvifolium reported in this paper

Full size image

Genome amplification, sequencing, and finishing

Plastid DNA was amplified using rolling circle amplification (RCA; Qiagen GmbH, Hilden, Germany) utilizing bacteriophage Phi29 polymerase and random hexamer primers (Dean et al. 2001). An EcoRI restriction digest was performed and visualized with ethidium bromide on 1% agarose gel to verify the purity and quantity of plastid DNA. Amplified DNA was either sent to the DOE Joint Genome Institute (Walnut Creek, CA) for Sanger sequencing or the W. M. Keck Center for Comparative and Functional Genomics at University of Illinois for 454 pyrosequencing.

Genomic data was assembled into contigs de novo in the native 454 assembler (Newbler) under default settings, as well as with open-source assembler MIRA (Chevreux 2009) using the “accurate” setting. Putative gene identifications in each contig were performed in DOGMA (Wyman et al. 2004), and contigs were assembled in Geneious (Drummond et al. 2009). Final annotations were made using DOGMA. A circular map was created for E. carvifolium using GenomeVx (Conant and Wolfe 2007).

Extraction and analysis of ndh pseudogenes

Due to long complex repeats including the apparent duplication of the atpB/E transcriptional unit (data not shown), the plastomes of E. chrysanthum, E. guicciardii, and E. gruinum have not yet been assembled into single contigs. To investigate the presence/absence of ndh genes, contigs for each species were converted to a custom BLAST database and queried with intact ndh genes from E. texanum and E. carvifolium.

Protein-coding genes were previously extracted from the E. chrysanthum genome assembly (Guisinger et al. 2008), suggesting that the assembly represents the complete plastid genome. The assembly is comprised of 3034 Sanger reads in three large contigs (37, 37, and 32 kb) and three small contigs (6, 3, and 2 kb) for a concatenated length of 118,538 bp. The E. guiccardii and E. gruinum datasets comprise of 27,620 and 43,067 454 Titanium pyrosequencing reads of 370 and 379 bp average length, respectively; the largest contigs obtained were 101 and 98 kb respectively. Both assemblies contain the same set of protein-coding genes previously found in E. chrysanthum (data not shown). Besides missing ndh genes, all three assemblies show the loss of the rpoC1 intron, which is present in both E. texanum and E. carvifolium.

Putative ndh pseudogenes were aligned against intact genes from E. texanum and E. carvifolium in Geneious under default settings; indels relative to the intact genes were characterized. Only indels >3 basepairs (bp) were tabulated in order to avoid counting as indels any sequencing errors due to homopolymer runs. For degraded, unalignable pseudogenes the sizes of BLAST hits from E. texanum ndh genes were tabulated.

Primers were designed to amplify an approximately 600 bp region of ψndhD containing four deletions shared among the three LBC genome taxa (ψndhD-Forward TCCGCAGGTTCCTTCATTTGT; ψndhD-Reverse TCTCCGCGAGTGTCTGGTAAC). This ψndhD region was amplified for all 13 LBC taxa to confirm the presence of pseudogenes with shared indels throughout the clade. LBC DNAs were provided by J. J. Aldasoro. Sanger sequencing of PCR products was performed on an ABI 3730 platform at The University of Texas at Austin.

Two non-ndh genes formerly transcribed with ndh genes, rps15 and psaC, were also extracted from LBC genome assemblies using BLAST in the manner described above for ndh pseudogenes. The rps15 and psaC genes were aligned with those from E carvifolium and E. texanum using MUSCLE (Edgar 2004) under default settings as implemented in Geneious.

Results

Genome organization in Erodium

The plastome of E. carvifolium is 116,934 bp and contains 75 protein-coding genes, 28 tRNA genes and the standard four rRNA genes (Fig. 2; accession number NC_015083). Gene content of E. carvifolium is nearly identical to that of E. texanum (Table 1), the only difference being the putative loss of trnK-uuu in the latter (Guisinger et al. 2011). Both exons of trnK-uuu appear intact in E. carvifolium. Relative to Arabidopsis the first and second exons of trnK-uuu in E. carvifolium both have 94% identity, compared to 83.8% and 63.9% for the E. texanum exons, respectively. The losses of the trnT-ggu gene and the introns in rpl16 and rps16 were previously described for the family Geraniaceae, and we confirm these losses in E. carvifolium. The genes ycf1, ycf2, and trnG-gcc were shown to be lost in three of the four major Geraniaceae genera (Erodium, Geranium, and Monsonia) and these genes also appear to be lost in E. carvifolium. The gene accD is absent from E. texanum, and we did not detect the gene in E. carvifolium.

Fig. 2
figure 2

Circularized gene map of the E. carvifolium plastid genome. Genes on the outside of map are transcribed in the counterclockwise direction and genes on the inside of the map are transcribed in the clockwise direction. ndh genes are in black; all others in gray. The arrow indicates IR deletion location

Full size image
Table 1 Comparison of E. texanum (clade I) and E. carvifolium (clade II) plastid genomes
Full size table

Given the high degree of genomic rearrangement found in representatives of all major Geraniaceae genera (Chumley et al. 2006; Guisinger et al. 2011), the plastome of E. carvifolium is remarkably unrearranged, displaying no unique inversions. Just two inversions separate E. carvifolium from the ancestral angiosperm gene order typified by tobacco (Raubeson and Jansen 2005), and both inversions are shared among all Geraniaceae plastomes. Among photosynthetic angiosperms E. carvifolium has the smallest reported plastid genome at 116,934 bp. Extensive rearrangement in E. texanum is associated with the proliferation of long, complex repeats in intergenic regions, contributing to the 14 kb difference in genome size despite nearly identical gene content.

The IR adjacent to the trnH-gug—psbA LSC junction has been cleanly deleted in E. carvifolium, leaving only 233 bp between trnH-gug and trnI-cau, formerly LSC and SSC junctions, respectively. Comparison of E. texanum and E. carvifolium reveals that movement of the IR boundary prior to its loss is not the cause of rearrangement in Erodium. The two inversions present in E. carvifolium are present in genomes of all four major genera of Geraniaceae, and the endpoints are associated with the loss of trnT-ggu and trnG-gcc (Guisinger et al. 2011). The tandem duplication of trnfM-cau between the rearranged psaA/B psbC/D units is also found in Geranium. The three additional duplications of trnfM-cau found in E. texanum are absent from E. carvifolium.

The only unique area of rearrangement in E. carvifolium is located between the rearranged psaC-D genes and a conserved cluster of three tRNAs (trnE-ucc/trnY-gua/trnD-guc). This intergenic region contains a pseudogene of rps18 flanked by two intact copies of trnG-ucc. The rps18 pseudogene shows 76% pairwise identity with the intact gene and contains eight premature stop codons. The two copies of trnG-ucc are located on opposite strands. Duplication of rps18 and trnG-ucc are not associated with further genomic rearrangement.

Loss of plastid-encoded ndh genes in Erodium

The complete plastomes of E. texanum and E. carvifolium, representing the two major clades of Erodium, both contain intact copies of all 11 plastid-encoded ndh genes (Fig. 1). The draft genome of E. chrysanthum, however, was suggested to lack intact copies of these genes (Guisinger et al. 2008). To verify this unusual loss and determine its phylogenetic distribution, two additional species were chosen for plastome sequencing from the LBC (Fig. 1): E. guicciardii and E. gruinum.

No intact open reading frames (ORFs) were found for ndh genes in E. chrysanthum, E. guicciardii, or E. gruinum. For larger ndh subunits (e.g., ndhF, ndhD, ndhJ, ndhA, and ndhB) pseudogenes were found by BLAST searches (accession numbers HQ730936–HQ730953), using intact ndh genes from E. texanum. These larger subunits could be aligned with confidence against intact genes from E. texanum and E. carvifolium, revealing that ndh pseudogenes retain high sequence identity to intact genes (~90% pairwise identity) despite disruption by copious indels inducing frameshifts (Table 2). For other, primarily shorter ndh genes, pseudogenes could not be aligned with confidence with intact ndh genes. Only short BLAST hits (<50 bp) were obtained when queried with intact genes from E. texanum (Table 3). With few exceptions, the same genes are unalignable or completely degraded in the three draft genome assemblies of E. chrysanthum, E. guicciardii and E. gruinum. Erodium gruinum contains longer pseudogenes for a few genes degraded in E. chrysanthum and E. guicciardii (Table 3); this is consistent with the sister relationship between E. chrysanthum and E. guicciardii (Fig. 1).

Table 2 Number of indels present in ndh pseudogenes that can be aligned with confidence with intact genes from E. texanum and E. carvifolium
Full size table
Table 3 Size of BLAST hits (in base pairs) returned for ndh pseudogenes that could not be aligned with confidence
Full size table

Analysis of indel events in ndh pseudogenes reveals a remarkably consistent pattern (Table 2): the great majority of indel events are deletions (86%), and a large proportion of these events (74%) is shared among all three taxa. To determine the extent of ndh gene loss in Erodium, we sequenced a disrupted, deletion-rich region of ψndhD for all 13 species in the LBC (Fig. 1). The sequenced region contained the same four deletions, three of which cause frameshifts, in all 13 taxa (Fig. 3a, b), indicating that all 13 members of this clade share the dispersed deletions associated with ndh gene loss (accession numbers HQ730926–HQ730935).

Fig. 3
figure 3

a Alignment of ndhD regions from C. macrophylla, E. carvifolium, and E. texanum with ψndhD from 3 LBC taxa. The ψndhD region amplified for all 13 LBC taxa (Fig. 3b) is marked with an arrow. b Alignment of ψndhD fragments from all 13 LBC taxa against intact ndhD genes from E. texanum and E. carvifolium. The lengths of the shared deletions, in bp, from left to right are 7, 5, 9, and 8

Full size image

Two genes formerly co-transcribed with ndh genes in an operon, rps15 and psaC, are intact in Erodium genomes lacking ndh genes. psaC encodes a subunit of photosystem I and rps15 encodes a ribosomal protein. The 5′ end of rps15 is variable in Erodium, and there are two indels unique to those taxa also lacking ndh genes (accession numbers HQ730922-HQ730923). However, psaC is highly conserved and shows no indels or nonsynonymous substitutions in the Erodium taxa examined (accession numbers HQ730924–HQ730925).

Discussion

Plastome organization in Erodium

Despite encoding almost the same number of genes, genome organization is strikingly different in E. texanum (clade I) and E. carvifolium (clade II). The four published Geraniaceae plastomes representing each major genus, P. × hortorum, M. speciosa, G. palmatum, and E. texanum are all highly rearranged and show extreme variation in IR size, from 76 kb in Pelargonium to IR loss in Erodium. In addition to inversions, expansion and contraction of the IR has been proposed as a mechanism of genomic rearrangement (Chumley et al. 2006). However, comparison of E. carvifolium and the highly rearranged E. texanum, both lacking the IR, suggests that the deletion of the IR occurred before the divergence of the two major clades, and that no rearrangements occurred concurrent with the deletion event. In E. carvifolium, the IR deletion site lies between two tRNAs, leaving less than 250 bp between former single copy region junctions. The E. carvifolium genome, with a ‘clean’ deletion of the IR but no other associated rearrangements (despite multiple rearrangements in related species) is reminiscent of the situation in legumes, in Medicago relative to Pisum (Palmer et al. 1988).

The paucity of rearrangements in E. carvifolium is striking given the extensive rearrangement evident in other published Geraniaceae plastid genomes (Chumley et al. 2006; Guisinger et al. 2011). The two inversions that distinguish E. carvifolium from the ancestral angiosperm gene order are common to all four major genera and thus a synapomorphy of Geraniaceae. Associated with these inversions are the loss of trnT-ggu and the duplication of trnfM-cau, raising the possibility of illegitimate recombination between tRNAs, a mechanism that has been suggested in at least two other angiosperm groups (Hiratsuka et al. 1989; Haberle et al. 2008). The loss of an IR is a derived character in Erodium, and the lack of other unique rearrangements allows us to hypothesize an ancestral gene order for Geraniaceae that resembles E. carvifolium prior to the deletion of the IR. Having a model for the organization of the simplest possible Geraniaceae plastome should greatly assist in reconstructing the rearrangement history of each genus. This model suggests that the rearrangement in each genus has occurred independently, with the exception of two inversions shared by all genera. It also suggests that expansion and contraction of the IR was not a force in genomic rearrangement in all genera, because loss of the IR preceded any unique genomic rearrangements in Erodium. Finally, the conservation of the S10 operon in E. carvifolium suggests that its fragmentation in E. texanum and G. palmatum represents two independent events (Guisinger et al. 2011).

Phylogenetic distribution and timing of ndh gene loss

The loss of ndh genes in the long-branch clade of Erodium (Fig. 1) is the most recent and phylogenetically restricted among photosynthetic seed plants. The presence of intact ndh genes in both major clades of Erodium has been established with the publication of E. texanum (Guisinger et al. 2011) and in E. carvifolium in this paper; the absence of intact ndh genes has been confirmed in the long-branch clade for all 13 species by plastid genome sequencing (3 taxa) or targeted PCR (10 taxa) (Fig. 3a, b). Except for a difference in chromosome number—n = 8 or 9 in the LBC compared with n = 10 in the rest of the genus—no morphology, life history or other trait appears to distinguish this clade from other members of the genus (Fiz et al. 2006). Indeed, homoplasy in leaf morphology caused the LBC to be grouped into the polyphyletic section Absinthoidea (Guittonneau 1990; Fiz et al. 2006). Alignment of pseudogenes from three LBC taxa with intact ndh genes from E. texanum and E. carvifolium reveals that the great majority of indels present in pseudogenes are shared by all three LBC taxa (74%). This consistency is a strong indication that loss of plastid-encoded ndh genes occurred prior to the diversification of the clade.

Parkinson et al. (2005) used the cox1 mitochondrial marker to date the divergence of the major clades in Geraniaceae. The inferred divergence times for Erodium (from Geranium) and the LBC (from the rest of Erodium clade I) are at approximately 22 MYA and 7 MYA, respectively. Another estimate using plastid markers indicated slightly earlier divergence times for Erodium and the LBC (from the rest of clade I) at approximately 15.45 MYA and 5 MYA respectively (Fiz et al. 2008). These divergence time estimates are extremely recent compared with the age of the other two known losses of ndh genes from photosynthetic seed plants. The loss of ndh genes in Pinaceae/Gnetales, if indeed this represents a single loss and a synapomorphy for the clade proposed under the “gnepine” hypothesis (Bowe et al. 2000; Chaw et al. 2000), is ancient. Within gymnosperms, divergence of Pinaceae has been estimated at 140 MYA (Wang et al. 2000), and loss of ndh genes necessarily predates this divergence estimate if it represents a synapomorphy for Pinaceae and Gnetales (Braukmann et al. 2009).

Loss of ndh genes in orchids is likely also ancient compared to that in Erodium. An estimate based on fossil-calibrated molecular data suggests a crown radiation of Orchidaceae 76-84 MYA (Ramirez et al. 2007). The taxonomic group recently confirmed to have lost ndh genes encompasses 70 orchid genera and over 1,000 species, suggesting that the loss of ndh genes is not recent (Wu et al. 2010). Orchidaceae is among the largest angiosperm families with ~30,000 species (Atwood 1986), and the full extent of ndh gene loss in Orchidaceae has not been determined. For example, many members of the subfamily Epidendroideae (Brassavola, Meiracyllium, Cattlea, Epidendrum, Encyclia, Stanhopea, Phalaenopsis, and Oncidium), the largest subfamily with more than 10,000 species, have deletions in ndhF inducing frameshifts. These data suggest that the loss of plastid-encoded ndh genes may be a synapomorphy for much of Orchidaceae (Neyland and Urbatsch 1996; Chang et al. 2006; Wu et al. 2010).

Fate of ndh genes: loss, replacement, or transfer?

It is unknown whether Ndh function has been lost entirely in these three seed plant lineages—in gymnosperms, orchids, and Geraniaceae—or whether the genes have been functionally transferred to the nucleus or otherwise replaced. No common features such as parasitism, mycotrophism, or epiphytic lifestyle are associated with ndh gene loss in Erodium. This discovery of a recent loss of ndh genes in Erodium presents an opportunity to investigate changes in photosynthetic function through comparative biochemistry between Erodium species with and without plastid-encoded ndh genes.

The chlorophyll fluorescence assay used to detect ndh mutants in tobacco (Horváth et al. 2000) can be used as a first step to assess the presence or absence of Ndh function in LBC species. Detection of Ndh function in LBC taxa could indicate an unprecedented recent burst of gene transfer from the plastid genome to the nuclear genome. Functional replacement of the Ndh complex by a nuclear gene or genes is another possibility; for example, in grasses, the multisubunit acetyl-Co carboxylase complex encoded by the plastid gene accD and nuclear-encoded subunits was replaced by a single-subunit complex of bacterial origin (Konishi et al. 1996). A failure to detect Ndh function in LBC taxa will suggest that regulation of electron flow between the two photosystems occurs through an alternative mechanism. Experimental data indicates that the Ndh complex is indispensable to the photosynthetic stress response (Endo et al. 1999; Martin et al. 2004; Rumeau et al. 2007; Casano et al. 2001). However, the consistent pattern of deletions found among LBC ndh pseudogenes suggests that loss of gene function occurred prior to the diversification of the clade >5MYA, and the subsequent diversification and persistence of this clade suggest that photosynthetic stress response has not been greatly compromised. Because the electron donor and functions of the Ndh complex are still unresolved it is difficult to evaluate the relative likelihood of gene loss, functional replacement, and functional gene transfer of ndh genes in Erodium.

Conclusion

The causal relationships among the evolutionary rate acceleration, indels, and loss of ndh genes in the LBC remain unclear. It is possible that the loss or functional transfer of ndh genes was promoted by the acceleration in the rate of evolution, if the rate in the plastome surpassed that in the nucleus, as has been recently suggested by Magee et al. (2010) for a hypervariable region in legume plastomes. A full evolutionary rate analysis of the genes from each functional class across eight additional Erodium species and monotypic sister species California macrophylla is underway to disentangle the relative roles of evolutionary rate accelerations, gene losses, indel frequency, and genomic rearrangements in Erodium plastome evolution. Finally, the recentness and restricted distribution of ndh gene loss in Erodium raises the possibility of finding additional recent losses among angiosperms, perhaps within lineages known to display plastome rearrangements.