Comparative plastomes of five Psittacanthus species: genome organization, structural features, and patterns of pseudogenization and gene loss

Abstract

The evolution of heterotrophic lifestyle entails varying degrees of plastome degradation. Yet, the evolutionary trajectory of plastome degradation associated with parasitism remains poorly explored in hemiparasites. We sequenced, assembled, and annotated the complete plastomes of five species of Psittacanthus mistletoes. In addition, publicly available plastomes of 58 species in Loranthaceae were obtained and re-annotated for phylogenetic and comparative analyses. We used a comparative phylogenetic approach to evaluate whether patterns of pseudogenization and gene loss differ among lineages of hemiparasites in Loranthaceae. Gene order was highly conserved, with higher sequence similarity and structural conservation between closely related Psittacanthus species but with considerable plastome size variation (from 121 238 to 125 427 bp). The expansion and contraction at the borders of inverted repeats (IRs) and intergenic regions variation greatly contribute to size variations among Psittacanthus plastomes. Phylogenetic analysis of plastomes of 60 species in Loranthaceae including 5 Psittacanthus species of the previously unsampled tribe Psittacantheae was largely congruent with previous phylogenetic studies. The loss of most of the ndh complex (10 out of 11 genes), rpl32, rps15, and rps16 genes, was identified in all studied Psittacanthus species. Also, the loss and pseudogenization of rpl33 and rpl36 genes in Psittacanthus were uncommon in other Loranthaceae species. The structural variation uncovered in Psittacanthus plastomes reveals that, despite high synteny, significant size variation exists among species. This variation can be attributed to processes such as variations in the length of intergenic regions and the expansion/contraction of IR borders, traits that have been comparatively understudied in earlier Loranthaceae works.

Despite the high conservation of gene order (synteny) or collinearity observed across different Psittacanthus chloroplast genomes (plastomes), significant genome size variation exist among species. The genome size differences between species likely result from processes such as expansions or contractions in intergenic regions and inverted repeat borders. Furthermore, our comparative-phylogenetic analysis of plastomes from hemiparasitic plants revealed distinctive and convergent degradation patterns in Psittacanthus, including gene loss and pseudogenization, which differ from patterns observed in plastomes of other hemiparasites in Loranthaceae.

Introduction

Land plants’ chloroplast genomes (plastomes) have highly conserved structures and organization, and multiple copies, meaning that the target genes are expressed at high levels (Raubeson and Jansen 2005, Verma and Daniel 2007). Thousands of complete chloroplast genomes have been sequenced, characterized, and used as a source of molecular markers (Li et al. 2020, 2022) and barcoding identification (Li et al. 2021a, Jiang et al. 2023) since the first successfully sequenced complete chloroplast genomes from the liverwort Marchantia polymorpha L. and the angiosperm Nicotiana tabacum L. (Ohyama et al. 1986, Shinozaki et al. 1986). These genomes have enhanced our understanding of plant evolution (Civáň et al. 2014, Ruhfel et al. 2014, Zhao et al. 2019) and have made significant contributions to phylogenetics (Carbonell-Caballero et al. 2015, Li et al. 2021b, Tyszka et al. 2023).

In general, the chloroplast genome comprises a single circular molecule with a quadripartite structure, including two copies of an inverted repeat (IR) region that separate the large and small single-copy (LSC and SSC) regions (Bock 2007). Typically, angiosperm chloroplast genomes range from 115 to 165 kb in size, harbouring between 101 and 118 unique genes that include ∼80 protein-coding genes (PCGs), ∼30 transfer RNA (tRNA) genes, and four ribosomal RNA (rRNA) genes (Bock 2007, Qu et al. 2023). Chloroplast genomes are highly conserved in their organization, gene order, and gene content. However, in plants that have lost the ability to photosynthesize, such as some parasitic species, chloroplast genomes show structural changes, gene loss, and high rates of pseudogenization (Wolfe et al. 1992a, Bromham et al. 2013, Petersen et al. 2015, Frailey et al. 2018, Schneider et al. 2018, Zhang et al. 2022, Banerjee and Stefanović 2023, Sánchez-Puerta et al. 2023).

Parasitism in angiosperms is considered to have at least 12 independent origins, and most parasitic species (c. 4528 species) are included in the family Orobanchaceae and the order Santalales (Nickrent 2020). Since the first complete chloroplast genome sequenced from a parasitic plant Epifagus virginiana (L.) W.P.C. Barton (Wolfe et al. 1992b), and despite their large known diversity, <10% of chloroplast genomes have been sequenced from parasitic plants. Santalales contains the largest number of genera (179) and species (2428) among the 12 parasitic plant lineages (Nickrent 2020). Also, this order contains the widest array of nutritional modes, including autotrophic nonparasites (13 genera/71 species), holoparasites (17/45), and hemiparasites (149/2312) (Nickrent 2020), as well as a variety of plant habits such as trees, shrubs, annual and perennial herbs, and aerial parasites. This makes Santalales an important model for the study of the evolution of parasitism in plants and its effects on genome reorganization. However, only 226 verified chloroplast genomes have been reported from 94 species in Santalales [c. 4% of the species in the order; estimates obtained from data available in GenBank (accessed 29 May 2024)], which highlights the need to continue generating genomic resources for the most diverse group of parasitic plants.

While holoparasitic plants obtain all nutrients from their host and show a complete loss of photosynthetic ability, hemiparasitic plants possess varying rates of photosynthetic ability (Westwood et al. 2010). Early comparative studies of complete chloroplast genomes from Santalales [(one Osyris shrub species (Santalaceae) and three hemiparasitic Viscum species (Viscaceae) (Petersen et al. 2015)] reported that chloroplast genomes were 10%–22% reduced in size relative to the chloroplast genome of the nonparasitic Vitis (Vitaceae), showing rearrangements in the borders of the IRs. Also, Petersen et al. (2015) found that several PCGs (matK, infA, ccsA, rpl33, and all 11 ndh genes) as well as 2 tRNA genes (trnG-UCC and trnV-UAC) have been pseudogenized or completely lost. Subsequent studies in hemiparasitic Santalales reported changes in selective pressures and gene content in plastomes usually characterized by pseudogenization or loss of the ndh complex genes, as well as of rpl16, rpl32, rpoC2, rps15, rps16, psaI, psbZ, ycf1, ycf2, and ycf15 genes. Thereby, hemiparasitic Santalales chloroplast genomes encode between 80 and 104 unique genes (Li et al. 2017, Shin and Lee 2018, Liu et al. 2019, 2021, Nie et al. 2019, Chen et al. 2020, Guo et al. 2020, 2021, Nickrent et al. 2021, Su et al. 2021, Al-Juhani et al. 2022, Darshetkar et al. 2023, Tang et al. 2024, Edlund et al. 2025).

To better understand trends of modifications in plastome evolution while still maintaining photosynthetic function, it would be informative to have complete chloroplast genome sequences of fully photosynthetic hemiparasites in Santalales from the New World (Nickrent and García 2009). The single complete chloroplast genome reported for the Psittacantheae tribe (Psittacanthus schiedeanus: Loranthaceae; Ibarra-Laclette et al. 2022, Morales-Saldaña et al. 2024) revealed the trnV-UAC gene for the first time in the Loranthaceae family. With at least 110 species, Psittacanthus Mart. is the most species-rich genus in Loranthaceae (Kuijt 2009, Dettke and Caires 2021). Loranthaceae contains the largest number of species in the Santalales, with over 1000 species distributed in 76 genera classified into 5 tribes: Nuytsiae, Gaiadendreae, Elytrantheae, Lorantheae, and Psittacantheae, distributed in tropical and subtropical regions of the Americas, Africa, Asia, and Australia, with a few species extending to the temperate zones in Europe and East Asia (Vidal-Russell and Nickrent 2008, Nickrent et al. 2010, Liu et al. 2018). Here, we sequenced and assembled for the first time the complete plastomes of five hemiparasitic Psittacanthus species, half of the Psittacanthus species distributed in Mexico: Psittacanthus auriculatus, Psittacanthus palmeri, Psittacanthus rhynchanthus, Psittacanthusschiedeanus, and Psittacanthus sonorae using two organelle assemblers with Illumina reads obtained from leaf samples. Our objectives were: (i) to assess structural variations and gene loss patterns among Psittacanthus plastomes; (ii) reconstructing plastome phylogeny to assess pseudogenization and gene loss in Loranthaceae using ancestral state reconstruction; and (iii) to assess the phylogenetic informativeness (PI) of plastome genes to solve conflictive phylogenetic relationships at both shallow and deep nodes in Loranthaceae. This study is the first to investigate plastome variation, phylogenetic utility, pseudogenization, and gene loss patterns in Psittacanthus and sets the stages for future studies in the genus.

Materials and methods

Taxon sampling

We collected and preserved in silica gel young leaves from individual plants of five Psittacanthus species in Mexico during the 2020–2 flowering seasons: P. auriculatus Eichler, P. palmeri (S. Watson) Barlow & Wiens, P. rhynchanthus (Benth.) Kuijt, P. schiedeanus (Cham. & Schltdl.) G. Don, and P. sonorae (S. Watson) Kuijt. This set of species comprises around 50% of the Psittacanthus species distributed in Mexico, including both range-restricted and widely distributed species, species parasitizing one or several host species, and species occurring only in one habitat type or in several habitat types. Based on previous phylogenetic studies, the sampled species are scattered along the phylogenetic tree and show diverse phylogenetic affinities among them (Ornelas et al. 2024). Therefore, we considered these five species to be ecologically and phylogenetically representative of the genus. Voucher specimens were deposited at the XAL herbarium of the Instituto de Ecología A.C. (INECOL). Geographic information of localities sampled and voucher information of the Psittacanthus species included in the study is summarized in Supplementary Table S1.

Chloroplast DNA extraction, genome sequencing, and assembly

Total genomic DNA was extracted from the preserved silica-gel-dried leaf tissue samples using a DNeasy Plant Mini kit (Qiagen, Valencia, CA, USA) following the manufacturer's protocol. The purified genomic DNA was used to prepare a paired-end (PE) Illumina sequencing library using Illumina Hi-Seq PE100 technology (Illumina, Inc., San Diego, CA, USA) following the manufacturer's guidelines.

Raw Illumina reads were first filtered using Trimmomatic v.0.38 (Bolger et al. 2014) to remove adapter sequences and low-quality reads with parameters LEADING:5 TRAILING:0 SLIDINGWINDOW:4:20 MINLEN:75. The complete plastome of each sample was de novo assembled using NOVOPlasty v.4.0 (Dierckxsens et al. 2017), with the trimmed reads and default parameters settings. However, we failed to assemble the chloroplast genomes of P. auriculatus and P. rhynchanthus. To correct this, we employed Get Organelle (Jin et al. 2020) with the settings -w 95 -R 20 -k 21,35,45,55,65,75 -F embplant_pt parameters for P. auriculatus and the parameters -w 65 -R 30 -k 35,45,55,65,75,85,95,105,115 -F embplant_pt for P. rhynchanthus. The resulting complete sequences were checked using Geneious Prime v.2023.2.1 (Kearse et al. 2012).

Genome annotation

All assembled plastomes were annotated using GeSeq (Tillich et al. 2017, Chan and Lowe 2019). Additionally, all chloroplast genomes were annotated using the Plastid Genome Annotator (PGA) with default settings (Qu et al. 2019) and using the plastomes of the following species: Elytranthe albida, Helicanthes elasticus, Loranthus odoratus, Macrosolen bibracteolatus, and Nuytsia floribunda. GenBank accession numbers and reference information of these plastomes are provided in Supplementary Table S2. The start and stop codons of each PCG were manually checked and adjusted, whereas the PCGs with one or more frameshift mutations or premature stop codons were annotated as pseudogenes. The circular genome map was drawn using the OGDRAW programme (https://chlorobox.mpimp-golm.mpg.de/OGDraw.html; Greiner et al. 2019). All newly and fully annotated plastomes were deposited in the NCBI GenBank database under the accession numbers listed in Table 1 and Supplementary Table S2.

Table 1.

Plastome attributes for the studied Psittacanthus species.

	Psittacanthus auriculatus	Psittacanthus palmeri	Psittacanthus schiedeanus	Psittacanthus rhynchanthus	Psittacanthus sonorae
Accession number	PP236145	PP236144	OR701826	PP579889	PP313080
Average read depth	5599.25×	1803.11×	6701.48×	7449.58×	3464×
Plastome size (bp)	121 238	122 407	122 586	120 345	125 427
Coding regions	75 427 (62.2%)	76 012 (62%)	74 573 (60.8%)	75 717 (63%)	74 391 (59.3%)
Noncoding regions	45 811 (37.7%)	46 395 (37.9%)	48 013 (39.2%)	44 628 (37%)	51 036 (40.6%)
LSC size (bp)	72 322	71 489	72 507	72 087	73 497
SSC size (bp)	7646	6080	7513	8672	6124
IR size (bp)	20 635	22 419	21 283	19 793	22 903
GC (%)	36.7	36.6	36.9	36.4	36.7
Unique PCG	66	66	65	66	65
tRNA	26	26	27	27	27
rRNA	4	4	4	4	4
Unique genes	96	96	96	97	96
Total genes	115	113	113	112	112

Chloroplast genome comparison and sequence divergence

We made comparisons using the assembled plastomes that included the previously published plastome of P. schiedeanus (Morales-Saldaña et al. 2024; Table 1; Supplementary Table S2) for several attributes, including gene number, gene content, gene order, guanine and cytosine (GC) content, as well as LSC, SSC, and IR size in Geneious Prime v.2023.2.1 (Kearse et al. 2012). The numbers of shared genes among different Psittacanthus plastomes were visualized using the heatmap2 function in ‘ggplot2’ package in R (Wickham 2016). To assess sequence divergence and to detect regions with elevated mutation rates among Psittacanthus plastomes, nucleotide diversity (π) was calculated in two datasets (PCGs and noncoding regions). Multiple sequence alignment for each dataset was performed using MAFFT v.7 (Katoh et al. 2019) and analysed independently by sliding window computation of nucleotide diversity (π) with a window size of 500 bp and a step size of 100 bp in DnaSP v.6.12.03 (Rozas et al. 2017). IRScope (https://irscope.shinyapps.io/irapp/; Amiryousefi et al. 2018) was used to analyse contractions and expansions of the repeat junctions across the five Psittacanthus plastomes. Finally, to visualize the conservation of gene order among chloroplast genomes, we generated a synteny plot using the pyGenomeViz v.0.2.1 package, employing the pgv-mmseqs mode and setting an identity threshold of 50% (https://github.com/moshi4/pyGenomeViz).

Phylogenetic analysis

To carry out the phylogenetic analyses, we employed the chloroplast genome sequences of the newly reported Psittacanthus species in this paper, as well as plastomes of Loranthaceae previously verified and available on NCBI GenBank until May 2024 (Supplementary Table S2). All downloaded plastomes were reannotated, compared, and analysed to refine the missing and dubious gene annotations. Sixty PCGs shared by all Loranthaceae plastomes were extracted and aligned individually using MAFTT v.7 (Katoh et al. 2019) and concatenated into a supermatrix. The phylogenetic tree was inferred by maximum likelihood (ML) in RaxML v.8.2.12 (Stamatakis 2014) using the best-fitting model (GTR + I + G4; Supplementary Table S3) estimated from ModelTest-NG 0.1.7 (Darriba et al. 2020) by the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and corrected Akaike Information Criterion (AICc) methods with branch support assessed using 1000 fast bootstraps. MrBayes v.3.2.7a (Ronquist and Huelsenbeck 2003) was used for the Bayesian inference analysis. The Markov chain Monte Carlo algorithm was run for 2 000 000 generations and sampled every 500 generations. The average standard deviations of split frequencies for each of the datasets were below 0.01. Trees resulting from the first 25% of generations were discarded as burn-in, and then, posterior probability (PP) values were estimated. Erythropalum scandens (Erythropalaceae, Santalales), a nonparasitic woody climber and one of the first lineages to diverge within Santalales (Nickrent et al. 2010), and Schoepfia jasminodora (Schoepfiaceae, Santalales), a root parasite and more closely related to Loranthaceae, were used as outgroups (Supplementary Table S2).

Loranthaceae informativeness profiles

We compared the performance of the 60 PCGs for resolving Loranthaceae relationships in terms of PI profiles with PhyDesign (López-Giráldez and Townsend 2011) using HyPhy substitution rates algorithm recommended for DNA sequences (Pond et al. 2005). The recovered ML tree was calibrated with an arbitrary timescale where tips were assigned to time 0 and root to 1 (Townsend 2007) and then converted to rooted ultrametric tree using the ‘chronos’ function in the ape package (Paradis et al. 2004) implemented in R Studio v.4.3.0 (R Core Team 2023). The converted relative-time ultrametric tree and alignment partitioned of concatenated 60 PCGs were used as input files in PhyDesign to calculate PI with default settings. Finally, we selected the most informative genes, and we conducted phylogenetic analysis for each gene to assess their usefulness in reconstructing the phylogenetic relationships of the Loranthaceae. Inference of each gene tree was carried out in RaxML v.8.2.12 (Stamatakis 2014) using the GTR + I + G4 model, with branch support assessed using 1000 fast bootstraps.

Gene loss and pseudogenization

On the obtained phylogenetic tree, we mapped gene loss and pseudogenization events detected in each species to trace the evolutionary trajectory of plastome degradation associated with the hemiparasitic lifestyle in Loranthaceae. To reconstruct the evolutionary pathway of Loranthaceae plastome degradation under a phylogenetic framework, we mapped gene loss and pseudogenization events detected in each species using the heatmap2 function in ‘ggplot2’ package in R (Wickham 2016). Also, we reconstructed the ancestral states for 11 PCGs, which were identified as pseudogenized or lost for two or more species using the ‘phytools’ package of R (R Core Team 2023), with equal-rates (ER) likelihood model and the ‘ace’ function (Revell 2012). We classified each PCG into three different categories: intact, pseudogene, and loss.

Results

Chloroplast genome content and structural comparison

The read depth of newly assembled Psittacanthus plastomes ranged from 1803.11× in P. palmeri to 7449.58× in P. rhynchanthus. The plastomes of the five Psittacanthus hemiparasitic species exhibited a typical quadripartite structure consisting of a pair of IR regions separated by a LSC and a SSC (Fig. 1). The lengths of these plastomes varied from 120 345 bp (P. rhynchanthus) to 125 427 bp (P. sonorae), exhibiting a genomic size difference of 5082 bp. The P. sonorae plastome showed the largest LSC region (73 497 bp) and P. palmeri the shortest LSC region (71 489 bp), whereas the P. rhynchanthus plastome showed the largest SSC region (8672 bp) and P. palmeri the shortest SSC region (6080 bp). The smallest IR region was observed in the P. rhynchanthus plastome (19 793 bp), whereas the largest IR region was observed in the plastome of P. sonorae (22 903 bp). The GC content was similar among all species, ranging from 36.4% to 36.9% (Table 1).

Figure 1.

Circular representation of annotated plastid genomes (plastomes) of the five species of Psittacanthus included in this study. Structural organization of the gene content ring was colour-coded based on its functional category. Only functional genes are drawn. Different categories of genes labelled with distinct colours. Genes drawn inside the circle are transcribed clockwise, and those outside are transcribed counter clockwise. The dark grey in the inner circle corresponds to the GC content across the genome, and the light grey corresponds to the AT content.

Gene order is highly conserved in all five Psittacanthus plastomes, whereas the total gene number, as well as protein-coding rRNA and tRNA genes, is slightly different (Table 1). The visualization of syntenic relationships between plastomes showed a high level of sequence similarity and structural conservation across species (Fig. 2a). The synteny visualization demonstrated that the more closely related species P. schiedeanus and P. auriculatus exhibited higher levels of synteny and similarity corresponding to the inverted region of the ycf2 gene in both species (Fig. 2a).

Figure 2.

Structure and comparison of genes in Psittacanthus plastomes. (a) Synteny plot of plastomes of the five species of Psittacanthus. The synteny plot shows the conservation of gene order among genomes. Normal links shown with grey colour, inverted links with red colour, and gene features with yellow colour. (b) Heatmap showing the distribution of the presence of, absence of, and pseudogenized genes for 60 PCGs from plastomes of the five species of Psittacanthus included in this study. (c) Sliding window analysis of nucleotide variability for coding regions among the plastomes of the five Psittacanthus species included in this study (window length = 1000 bp; step size = 500 bp). The x-axis and y-axis indicate the position of the midpoint of the nucleotide diversity of each window. (d) Sliding window analysis of nucleotide variability for noncoding regions + introns among the plastomes of the five Psittacanthus species included in this study (window length = 1000 bp; step size = 500 bp). The x-axis and y-axis indicate the position of the midpoint of the nucleotide diversity of each window. The most variable regions are labelled.

The loss of most of the NAD(P)H-dehydrogenase (ndh) complex (10 out of 11 genes), rpl32, rps15, and rps16 genes, was identified in all studied Psittacanthus species (Fig. 2b). In addition, the infA and ycf15 genes were not identified in some of the species, the former in P. sonorae and the latter in P. rhynchanthus, P. auriculatus, and P. schiedeanus (Fig. 2b). Pseudogenization for the ndhB gene was identified in all studied Psittacanthus species and infA in P. palmeri (Fig. 2b). The rpl36 gene was lost in P. sonorae, P. palmeri, and P. schiedeanus and pseudogenized in P. auriculatus and P. rhynchanthus, whereas rpl33 was identified as a pseudogene in P. sonorae, P. palmeri, P. auriculatus, and P. schiedeanus, i.e. identified as a potentially functional gene in the P. rhynchanthus plastome (Fig. 2b). Lastly, the loss of trnV-UAC occurred in all Psittacanthus species except P. schiedeanus, trnA-UGC was lost in P. sonorae and P. palmeri and pseudogenized in P. rhynchanthus and P. auriculatus, and trnK-UUU was only lost in P. palmeri.

For the studied Psittacanthus species, nucleotide diversity (π) for the protein-coding regions averaged 0.017, ranging from 0.002 (psaB and rpl2 genes) to 0.075 (ycf1 gene) (Table 2). Among the PCGs analysed, five genes (atpA, rpoC2, accD, rps8, and rpl22) exhibited π values > 0.03, whereas two genes (ccsa and ycf1) had π values > 0.04 (Fig. 2c). Compared with the LSC and SSC regions, the nucleotide diversity in the IR region showed the smallest π values. The nucleotide diversity values in noncoding regions were higher compared with those in the coding regions, with values ranging from 0.014 to 0.11; the regions with the highest diversity were rpoB-trnC, trnT-psbD, rbcL-accD, rps12-trnV, and psaC-ycf1 (Fig. 2c).

Table 2.

Summary of the characteristics of the most phylogenetically informative genes.

Gene	Aligned length (bp)	Polymorphic sites (S)	Missing data (%)	Nucleotide diversity (π)	Max value reached at
ccsA	1002	358	5.22	0.087	0.44
matK	1722	131	14.09	0.080	0.67
rpoB	3252	916	0.98	0.042	0.99
rpoC2	4429	275	8.81	0.080	0.81
ycf1	6480	42	25.64	0.127	0.40
ycf2	7308	968	7.87	0.019	0.99

Contraction and expansion of inverted repeats

The results indicated expansions/contractions throughout, with slight variations in the junctions of the IR and SSC regions and between the IR and LSC regions in the Psittacanthus plastomes (Fig. 3). The ycf1 gene was located completely within the SSC region in P. auriculatus, P. rhynchanthus, and P. schiedeanus, but in P. palmeri and P. sonorae the same gene passes through the SSC and IRa regions. The junctions between the IRb and SSC regions were located between the trnN and trnL genes in P. auriculatus and P. schiedeanus, but not in P. rhynchanthus. For closely related P. palmeri and P. sonorae, the trnL gene passes through the IRb and SSC junctions. The IRa–LSC junction was characterized by the presence of the psbA and trnH genes in all species, whereas the rpl2 gene was located through the LSC–IRb junctions in all plastomes (Fig. 3).

Figure 3.

A phylogenetic tree of Psittacanthus along with distances between adjacent genes and junctions of the SSC, LSC, and two IR regions among plastomes of the five species of Psittacanthus included in this study. Boxes above and below the main line indicate the adjacent border genes. The figure is not to scale regarding sequence length and only shows relative changes at or near the IR/SC junctions.

Phylogenetic analysis

The final alignment of 60 PCGs shared by 62 taxa was 59 703 bp with 6538 parsimony informative sites and 7.3% of missing data. ML and Bayesian inference BI analyses generated identical tree topologies (Fig. 4), with most nodes being fully supported (BS = 100%, PP = 1.00). Our results showed that N. floribunda was strongly supported as the sister group to the remaining Loranthaceae lineages (BS = 100%, PP = 1.00/0.99; Fig. 4). Within the tribe Elytrantheae, Lysiana exocarpi was placed sister to a well-supported clade that included Elytranthe and Macrosolen with high support values (BS = 100%, PP = 1.00). However, the position of Lysiana as sister to the rest of the Elytrantheae remains uncertain, due to the absence of several genera and species in Elytrantheae in this study. Elytranthe albida and Elytranthe parasitica were nested within Macrosolen with high support value (BS = 100%, PP = 1.00). For the largest tribe Lorantheae retrieved with strong support value (BS = 100%, PP = 1.00), subtribes Ileostylinae and Loranthinae were placed sister to clade composed of subtribes Amyeminae, Emelianthinae, Tapinanthinae, Dendrophthoinae, and Scurrulinae with strong support value (BS = 100%, PP = 0.99) and subtribes Amyeminae and Dendrophthoinae as paraphyletic. Lastly, the five species of Psittacanthus (tribe Psittacantheae, subtribe Psittacanthinae) were recovered as one clade with strong support (BS = 100%, PP = 1.00) and placed sister to tribe Lorantheae with moderate support value (BS = 82%, PP = 0.99), in which P. sonorae and P. palmeri formed a subclade that was placed sister to the P. rhynchanthus and P. auriculatus + P. schiedeanus subclade (Fig. 4).

Figure 4.

Phylogenetic tree of Loranthaceae reconstructed by analysing 60 plastid PCGs and plastomes of 52 species of Loranthaceae using (a) Bayesian inference and (b) ML methods. Numbers above branches indicate bootstrap (BS) percentage/PP, with the asterisk (*) indicating full support in both analyses (BS = 100%, PP = 1.00) from 1000 fast replicates. Label annotation against each species name represents the GenBank accession number.

Loranthaceae informativeness profiles

Profiles of PI for each of the 60 PCGs showed ycf1 as the most informative locus, increasing quickly in shallow time spans and attaining the highest values at a reference time (phylogenetic depth) of 0.4 (Supplementary Fig. S2a and b). Other genes with considerable PI values were rpoC2, matK, ycf2, rpoB, and ccsA (Table 2, Supplementary Fig. S2c). PCGs with high net PI values had long lengths (>1000 bp), although other long-length genes (e.g. atpA, atpB, psbA, psbB, psbC, psbD, rbcL, and rpoA) showed low PI values. However, individual gene trees recovered from the most informative genes showed topological discrepancies in the position of the Psittacantheae and Elytrantheae tribes as well as the subtribe Ileostylinae phylogenetic position (Supplementary Fig. S3). Likewise, the phylogenetic trees based on ccsA or ycf2 showed low support for shallow nodes, whereas the tree based on ycf1 was the only one that showed similar backbone topologies to the tree recovered with the 60 PCGs, with high support values for shallow nodes, although at deeper nodes the support values were lower. Finally, the gene tree based on rpoC2 recovered the worst tree in terms of topology and support values (Supplementary Fig. S3).

Gene loss and pseudogenization

Gene loss and pseudogenization events were mapped onto a phylogenetic tree (Fig. 5). Eight ndh genes were absent in Loranthaceae, except ndhB in 56 species, ndhD in E. parasitica and L. exocarpi (tribe Elytrantheae), and ndhF in the five Amyema species (subtribe Amyeminae) (Fig. 5). Pseudogenization events were detected in 12 genes for Loranthaceae (Fig. 5). Additionally, unique pseudogenized or gene loss events were identified for psaI in Loranthus delavayi and trnP-UGC in Loranthus europaeus. The rps15 and rps16 genes, as well as the trnG-GAC gene, were also absent in all Loranthaceae chloroplast genomes (Fig. 5).

Figure 5.

Pseudogenization or loss of PCGs heatmap in the plastome of 60 species of Loranthaceae and the autotrophic nonparasitic woody climber E. scandens (Erythropalaceae) and the woody root parasite S. jasminodora (Schoepfiaceae) outgroup species (Supplementary Table S2).

The ancestral reconstruction using the ER likelihood model revealed that infA and trnA genes displayed losses and pseudogenization differently within Loranthaceae. In contrast, loss and pseudogenization of the ycf15 gene occurred in all analysed species except for Plicosepalus acaciae, Plicosepalus curviflorus, Tolypanthus maclurei, P. palmeri, and P. sonorae (Supplementary Fig. S1). The pseudogenization of rpl16 and trnI might have occurred in the stem lineage ancestor of Scurrula and Amyema, and the pseudogenization of rpl33 in Psittacanthus also occurred in Cecarria obtusifolia, but the further loss was observed in P. acaciae and P. curviflorus. Lastly, the pseudogenization or loss of the rpl36 gene was observed in Psittacanthus, but the loss was observed in Amyema preissii, P. acaciae, and Dendrophthoe pentandra (Fig. 6).

Figure 6.

Ancestral state reconstruction of pseudogenization and loss of rpl33 and rpl36 genes in Loranthaceae using the ER likelihood model. The plastid genes were classified into three types (intact, pseudogenization, and loss). The pies indicate the ancestral state reconstruction. Light blue indicates a functional gene; dark blue indicates the gene becomes pseudogenized; and black indicates gene loss.

Discussion

In recent years, several studies have contributed to understanding the chloroplast genome evolution in hemiparasites of Santalales. However, these studies are focused almost exclusively on Asian (Li et al. 2017, Shin and Lee 2018, Liu et al. 2019, 2021, Nie et al. 2019, Chen et al. 2020, Al-Juhani et al. 2022, Darshetkar et al. 2023, Tang et al. 2024) and European lineages (Guo et al. 2020, 2021, Nickrent et al. 2021, Su et al. 2021, Tang et al. 2024), while mistletoes in the American Continent have received no attention. In our study, we sequenced, assembled, and annotated the chloroplast genomes of five Psittacanthus species (Psittacantheae tribe), generating valuable genomic data to enhance our understanding of genome evolution in parasitic plants.

Size and structural variation in Psittacanthus chloroplast genomes

Typically, chloroplast genomes of autotrophic angiosperms with sizes ranging from 115 to 165 kb in size show highly conserved gene composition (101–118 unique genes), as well as collinear sequences arrangement (Bock 2007, Qu et al. 2023). The Psittacanthus chloroplast genomes reported here show a reduction in size and gene content between 19.7%–22.9% and 13.4%–14.3%, respectively, relative to the chloroplast genome of the outgroup E. scandens (Erythropalaceae), a nonparasitic species placed near the base in the large Santalales order (Zhu et al. 2018). Moreover, the chloroplast genomes of Psittacanthus are significantly larger than the one of S. jasminodora (118 743 bp), a root hemiparasite in Schoepfiaceae, a more closely related family to Loranthaceae, and the Psittacanthus chloroplast genomes encode fewer genes (96–97 unique genes) compared with 112 found in S. jasminodora (Su and Hu 2016). Within the Loranthaceae family, plastome sizes exhibit significant variation. Genera such as Loranthus, Scurrula, Plicosepalus, Taxillus, and Tolypanthus possess chloroplast genomes of similar size to those reported here for Psittacanthus, whereas Cecarria, Dendrophthoe, Phyllodesmis, and Scurrula pulverulenta have smaller plastomes. In contrast, basal genera such as Elytranthe, Macrosolen, and Nuytsia have notably larger plastomes sizes, ranging from 126 621 to 139 027 bp (Li et al. 2017, Shin HW, Lee NS 2018, Al-Juhani et al. 2022, Tang et al. 2024). These results reflect distinct evolutionary trajectories in genome size and gene content among these hemiparasitic plants.

The reduction of plastome size is common phenomenon observed in parasitic plants (Petersen et al. 2015, Chen et al. 2020, Guo et al. 2020, Darshetkar et al. 2023). While hemiparasites are expected to maintain most or all their photosynthetic genes (Frailey et al. 2018), this process is highly variable across different lineages and appears to have occurred independently in multiple clades (Guo et al. 2020, Su et al. 2021). Pseudogenization/gene loss has been linked to size variation in the chloroplast genome of parasitic plants (Bungard 2004, Xiao-Ming et al. 2017, Guo et al. 2020, Liu et al. 2021, Su et al. 2021, Darshetkar et al. 2023). Even though these mechanisms might explain the size reduction observed in the Psittacanthus chloroplast genomes, when compared with E. scandens, the observed size variation among Psittacanthus species chloroplast genomes could also be explained by additional processes such as variations in the length of intergenic regions and the expansion/contraction at the borders of IRs (Petersen et al. 2015, Xiao-Ming et al. 2017). Intergenic region variation has been reported as a mechanism that mainly affects the variation in chloroplast genomes size at the intrageneric level (Xiao-Ming et al. 2017). In this context, our sequence divergence analysis revealed that the highest proportion of the intergenic regions occurred in the largest chloroplast genome (P. sonorae) with 40.6% of the total size of the plastome, whereas the lowest proportion of intergenic regions (37%) occurred in P. rhynchanthus, the smallest plastome reported for Psittacanthus. Moreover, previous studies have related a greater decrease in intergenic regions with an increasing degree of parasitism and with improved energy efficiency and nutrition utilization (McNeal et al. 2007, Wu et al. 2009, Petersen et al. 2015), as well as reinforced fitness to the partial heterotrophic lifestyle (Guo et al. 2020).

Inverted repeats and rearrangement patterns

The expansion/contraction at the borders of IRs are major factors contributing to plastome size variation (e.g. Yao et al. 2015, Wang et al. 2017). We found that IR boundary shifts are accompanied by variation in plastome size in Psittacanthus. More closely related species in Psittacanthus showed similar IR boundary shifts where the P. auriculatus/P. schiedeanus/P. rhynchanthus clade showed the ycf1 gene located exclusively within the SSC region. In contrast, for the P. sonorae/P. palmeri clade, the ycf1 gene was located crossing both the SSC and IRa regions, a trait shared with nonparasitic E. scandens (Zhu et al. 2018), suggesting that a reduction of the IR regions is occurring in the Psittacanthus chloroplast genomes. The IR shifts are common in hemiparasitic Santalales and have been associated with the functional loss of ndh genes (Shin and Lee 2018, Guo et al. 2020, Edlund et al. 2025). However, rearrangements in plastid genomes are not exclusive to parasitic plants, so they might not be connected necessarily to a photosynthesis-driven conservative evolution (Krause 2011).

We observed a high level of collinearity among the chloroplast genomes of the species of Psittacanthus included in our study. This observation suggests that no significant structural changes in the plastomes have occurred throughout their evolutionary history, as reported for chloroplast genomes of phylogenetically distant genera from Loranthaceae (Su et al. 2021, Darshetkar et al. 2023, Tang et al. 2024) and for other Santalales (Liu et al. 2021, Edlund et al. 2025). However, the lack of major structural rearrangements contrasts with major structural rearrangements reported in other orders of hemiparasitic plants (Zhang et al. 2022, Chen et al. 2024). Unlike other hemiparasitic plant orders (e.g. Lamiales), which plastomes exhibit major structural rearrangements with ≤36% GC content (Wicke et al. 2013, 2016, Zhang et al. 2020a), hemiparasites in Santalales show remarkably high plastome synteny. Nevertheless, cross-order comparisons between parasitic lineages (e.g. Santalales vs. Lamiales) require caution, as observed differences may arise from lineage-specific evolutionary dynamics rather than parasitic adaptations alone. Altogether, these results reveal that the chloroplast genomes for these Psittacanthus species are highly conserved structurally and highly similar in terms of gene content compared with earlier Loranthaceae lineages.

Phylogenetic relationships and phylogenetic informativeness

The tree topologies from ML and BI phylogenetic analyses of Loranthaceae based on chloroplast genomes of 60 species including five Psittacanthus species of the previously unsampled tribe Psittacantheae were largely congruent with previous studies (Vidal-Russell and Nickrent 2008, Liu et al. 2018, Nickrent et al. 2010, Su et al. 2021, Tang et al. 2024). The sister relationships between the facultative root parasite N. floribunda (Nuytsieae tribe) and the rest of the obligate stem parasites Loranthaceae tribes and between the tribe Elytrantheae and the other Loranthaceae tribes were recovered with strong support. However, the genera Macrosolen and Elytranthe in the tribe Elytrantheae and Helixanthera in the tribe Lorantheae are recovered as paraphyletic (see also Vidal-Russell and Nickrent 2008, Liu et al. 2018, Tang et al. 2024). Helixanthera parasitica was placed sister to the subtribe Scurrulinae clade (Scurrula and Taxillus), which is consistent with previous findings (Liu et al. 2018, Su et al. 2021, Tang et al. 2024 ). Within the tribe Elytrantheae, L. exocarpi was recovered as sister to a well-supported clade that included both Elytranthe and Macrosolen species, which is consistent with previous studies that have proposed these morphologically similar genera to be treated taxonomically as congeners (Barlow 1997, Liu et al. 2018, Tang et al. 2024). For the largest tribe Lorantheae, subtribes Ileostylinae and Loranthinae were placed sister to clade comprising Amyeminae, Emelianthinae, Tapinanthinae, Dendrophthoinae, and Scurrulinae with strong support value, though Amyeminae and Dendrophthoinae were retrieved as paraphyletic (see also Tang et al. 2024). In addition, our study placed the clade with plastomes of five Psittacanthus species, previously unsampled members of the tribe Psittacantheae, sister to members of the tribe Lorantheae (see also Vidal-Russell and Nickrent 2008). Despite the robust sampling at the generic level in previous phylogenetic studies of Loranthaceae (e.g. Vidal-Russell and Nickrent 2008, Grímsson et al. 2017, Liu et al. 2018, Ortiz-Rodriguez et al. 2018), the limited number of chloroplast markers employed produced low node support values along the backbone of the reported trees, polytomies, and topological incongruences in some genera. Studies using chloroplast genome sequence data have shown well-resolved clades, but increased taxonomic sampling is still needed to evaluate the monophyly of most subtribes in Loranthaceae (Guo et al. 2020, Nickrent et al. 2021, Su et al. 2021, Tang et al. 2024).

Whereas previous studies have not assessed the utility of individual plastome genes for phylogenetic inferences in Loranthaceae, our analysis identified the ycf1 gene as particularly useful. Among the most informative PCGs, ycf1 produced tree topologies comparable to those from full plastome analysis (60 genes), despite its high mutation rate and nucleotide diversity (consistent with Su et al. 2021). Although too long for standard PCR, ycf1 shows promise for resolving shallow-level relationships. In contrast, the IR-located ycf2 gene showed low nucleotide diversity but strong support for deep nodes. Combining these two markers may help resolve phylogenetic relationships at both shallow and deep levels in Loranthaceae.

The loss of genes in Loranthaceae

Based on the obtained phylogenetic tree (see Fig. 5), our results show that the loss/pseudogenization of the tRNA gene (trnG-GAC), two ribosomal protein genes (rps15 and rps16), and 10 ndh genes (ndhA, ndhC–K) are plesiomorphic traits in Loranthaceae. The loss of ndh genes is common in hemiparasitic Santalales, including Loranthaceae, except for the ndhB gene (Guo et al. 2020, Nickrent et al. 2021, Su et al. 2021, Darshetkar et al. 2023, Edlund et al. 2025), which in most species including Psittacanthus has been pseudogenized. Two explanations for the loss of ndh genes have been proposed: (i) the function of the ndh gene is related to hydric stress conditions (Horváth et al. 2000, Martín and Sabater 2010, Chen et al. 2020) and (ii) its role in photosynthesis is often dispensable and even selected against in some plant lineages (Lin et al. 2017, Frailey et al. 2018). Edlund et al. (2025) found one functional loss of the 11 ndh genes in Santalales and signs of relaxed selection in autotroph lineages, suggesting that PGR5/PGRL1 protein complex might act as an alternative pathway for photosystem I cyclic electron transport. Although the pseudogenization and eventual loss of the ndh genes have been proposed as early genomic changes associated with the transition to a parasitic lifestyle (Shin and Lee 2018, Darshetkar et al. 2023), establishing a direct link between ndh degradation and hemiparasitism remains challenging. First, there is insufficient physiological evidence demonstrating that the loss of these genes impairs function in hemiparasitic species. Second, no studies have investigated whether ndh genes have been functionally transferred to the nuclear or mitochondrial genome. Moreover, the absence of ndh genes is not exclusive to parasitic plants, multiple nonparasitic lineages, including gymnosperms (Pinaceae; Wakasugi et al. 1994; Gnetales; Braukmann et al. 2009) and angiosperms such as orchids (Jheng et al. 2012, Lin et al. 2017) and cacti (Sanderson et al. 2015), as well as species of Alismatales (Iles et al. 2013, Peredo et al. 2013), Geraniales (Blazier et al. 2011), and Ranunculales (Sun et al. 2017), seem not to suffer negative impacts. Thus, the ndh complex might be of limited biological significance for contemporary plants (Ruhlman et al. 2015, Lin et al. 2017).

While the loss of the rpl32 gene appears to have occurred during the evolutionary transition from root parasitism to stem parasitism, comprehensive testing of this hypothesis requires analysis of all the facultative root-parasitic lineages (N. floribunda, Atkinsonia ligustrina, and Gaiadendron punctatum). The loss/pseudogenization of plastid ribosomal protein-encoding genes has been detected not only in parasitic plants (McNeal et al. 2007, Su et al. 2018), but also in a wide range of autotrophic angiosperms (Sabir et al. 2014, Park et al. 2015, Zhang et al. 2020b, Silva et al. 2021). Thereby, the loss of rpl32, rps15, and rps16 genes may also be related to the migration of genes to the nucleus and not necessarily to the evolution of parasitism (Park et al. 2018, Dobrogojski et al. 2020, Guo et al. 2020, Yang et al. 2021, Darshetkar et al. 2023). Interestingly, the lineage-specific loss/pseudogenization of ribosomal protein genes (rpl16, rpl33, and rpl36) may reflect evolutionary parallelism, as these independent events occurred in closely related taxa sharing a common ancestor with functional copies of these genes (see Figs 5 and 6; Tang et al. 2024). Particularly, the loss/pseudogenization of the rpl33 gene has been reported in many Santalales (Chen et al. 2020) but is uncommon in Loranthaceae, suggesting that the degradation of the rpl33 gene has occurred several times independently. Although the rps15, rpl33, and rpl36 genes are not essential for chloroplast gene translation, the remaining genes play essential roles (Tiller and Bock 2014). Hence, the loss of these plastid ribosomal protein-encoding genes in Loranthaceae might be compensated for by other plastid rpl/rps genes or by nuclear-encoded rpl/rps genes.

Although pseudogenization of psbZ, matK, rpoC2, ycf1, and ycf2 genes has been previously reported in Loranthaceae (Tang et al. 2024), our dual annotation approach (combining Geseq and PGA annotators) successfully identified these genes with enhanced confidence. This is consistent with previous reports in which transcriptional evidence was found for the rpoC2, ycf1, and ycf2 genes from P. schiedeanus by RNAseq (Ibarra-Laclette et al. 2022). These results suggest that these genes have undergone size reduction processes but continued to conserve the start and stop codons, and they also do not show intermediate stop codons. The loss/pseudogenization of the infA and ycf15 genes in other Loranthaceae species were also observed in Psittacanthus. Both genes are characterized by showing anomalous distribution patterns in angiosperms and have been related to nuclear transfer events (Millen et al. 2001, Shi et al. 2013). Our phylogenetic analysis of plastomes showed that both the infA and ycf15 genes are lost/pseudogenized several times along the evolution of Loranthaceae, which suggests multiple independent transfers to the nucleus (Millen et al. 2001). Lastly, the loss/pseudogenization of the trnA-UGC and trnV-UAC genes in Psittacanthus (except P. schiedeanus) is also observed in other Loranthaceae species as well as other Santalales (Chen et al. 2020, Darshetkar et al. 2023). Both genes have been considered essential for cell viability, so mechanisms such as the tRNA import from the cytosol to the plastids were proposed to replace the essential of these tRNA losses (Alkatib et al. 2012, Dalla Costa et al. 2022).

Conclusion

Chloroplast genomes are currently the turning point of comparative studies in hemiparasitic plants. It is now essential to generate genomic resources for diverse lineages and then contrast different hypotheses about diverse evolutionary trajectories of hemiparasitic plastome degradation. Our results show that the plastomes of analysed species possess traits uncommon within Loranthaceae, including the loss or pseudogenization of rpl33 and rpl36 genes. The pseudogenization of the ndhB gene occurred in the plastomes of the five species of Psittacanthus included in this study. Contrary to previous studies, our findings reveal that despite the high synteny observed in Psittacanthus plastomes, significant size variation exists among species. This variation can be attributed to processes such as variations in the length of intergenic regions and the expansion/contraction of IR borders, traits that have been comparatively understudied in earlier Loranthaceae works. The present study establishes an important precedent to understanding plastome evolution in the New World mistletoes. Although our study demonstrates the utility of plastid genes for phylogenetics of Loranthaceae, some conflictive clades remain unsolved. These discrepancies could potentially be resolved using genomic data across Loranthaceae in the American Continent, particularly from the nuclear genome.

Author contributions

Saddan Morales-Saldaña (Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Visualization, Writing–original draft, Writing–review & editing), Andrea I. Barraza-Ochoa (Data analysis, Writing–review & editing), Emanuel Villafán (Data curation, Software, Writing–review & editing), Antonio Acini Vásquez-Aguilar (Data curation, Writing–review & editing), Santiago Ramírez-Barahona (Investigation, Writing–review & editing), Enrique Ibarra-Laclette (Data curation, Data analysis, Methodology, Resources, Software, Supervision, Writing–review & editing), and Juan Francisco Ornelas (Conceptualization, Data analysis, Funding acquisition, Investigation, Project administration, Resources, Supervision, Validation, Visualization, Writing-original draft, Writing–review & editing). All authors approved the final manuscript.

Supplementary data

Supplementary data is available at AoB Plants online.

Funding

This study was supported by two grants from the Consejo Nacional de Ciencia y Tecnología (CONACyT), Mexico, under grant numbers 155686 and A1-S-26134 and research funds from the Instituto de Ecología, A.C. (20030/10563), awarded to J.F.O. S.M.-S. was supported by a postdoctoral fellowship (‘Estancias Posdoctorales por México' programme, I1200/311/2023) and A.I.B.-O. by a doctoral scholarship (1140508) from the Consejo Nacional de Humanidades, Ciencias, y Tecnologías (CONAHCyT).

Data availability

The data underlying this article are available in its supplementary material. Fully annotated plastomes were deposited in the NCBI GenBank database under the accession numbers listed in Table 1 and Supplementary Table S2.