In this study, a total of eight individuals from two species, C. filiformis and C. larsenii, were sampled. Fresh stems of wild vines were collected from China and Puerto Rico, and quickly dried with silica gel (Table 1). Voucher specimens were deposited in the herbarium of Guangxi Normal University. The specimens were identified by Yun-Hong Tan (Xishuangbanna Tropical Botanical Garden, CAS) and Yu Song (Guangxi Normal University). Furthermore, plastid genome sequences of related taxa of Lauraceae were downloaded from Lauraceae Chloroplast Genome Database (LCGDB) (https://lcgdb.wordpress.com) and GenBank of NCBI (https://www.ncbi.nlm.nih.gov) and a total of 61 taxa from 27 genera of Laurales were included. In addition, the partial trnK intron, including matK gene sequences, of eight newly sequenced and 83 individuals of Cassytha obtained from the NCBI database were used to conduct phylogenetic analyses (Figure 1; Supplementary Figure S1).

Whole-genomic DNA was extracted from the stem tissues using the CTAB method (Doyle and Doyle, 1987). Libraries were constructed with fragments of approximately 300 bp according to the manufacturer’s protocol (Illumina, San Diego, CA, United States). All samples were sequenced using the Illumina HiSeq 2,500 at Kunming Institute of Botany, Chinese Academy of Sciences (KIB, CAS). The plastid genomes were de novo assembled using the GetOrganelle pipeline under default settings (Jin et al., 2020; Dong et al., 2022a). The assembly quality of all plastid genomes was checked with Bandage software (Wick et al., 2015).

The newly assembled plastid genomes were annotated using the GeSeq application, followed by manual verification using the Geneious software (Kearse et al., 2012; Tillich et al., 2017). The annotation sequences and their corresponding information were then submitted to Genbank and assigned accession numbers OR766688 to OR766695. Finally, the physical map of the annotated plastid genomes was drawn using the online Chloroplot program (https://irscope.shinyapps.io/Chloroplot/).

In order to compare the divergence within eight newly assembled plastid genomes, the online mVISTA program in Shuffle-LAGAN mode (https://genome.lbl.gov/vista/mvista/submit.shtml) was employed. Additionally, the plastid genome sequence’s nucleotide diversity (Pi) was estimated using DnaSP, with a step size of 200 bp and a window length of 600 bp for sliding window analysis (Rozas et al., 2017). The number and position of Indel and single-nucleotide polymorphism (SNP) events were determined by manual statistics and analyzed in the aligned eight plastid genome sequences of two Cassytha species.

To determine the phylogenetic relationships within Cassytha and its relatives, the eight newly assembled plastid genomes were compared to the other 53 Laurales species, with Illigera celebica (LAU00199) and I. grandiflora (LAU00198) selected as outgroups. The sequences were aligned using MAFFT (Katoh et al., 2019) and manually adjusted in BioEdit (Hall et al., 2011). Maximum likelihood (ML) analysis was conducted using IQ-tree v2, and the best-fit model was determined using ModelFinder (Kalyaanamoorthy et al., 2017), with the GTR + F + I + G4 model and a bootstrap value of 1,000 (Minh et al., 2020). Then, the trnK and ITS sequences of the eight sequenced Cassytha individuals were intercepted and aligned with available trnK and ITS sequences from GenBank. A previous study showed C. melantha was basal to other Cassytha taxa based on extensive sampling (Kokubugata et al., 2012). Consequently, C. melantha was used as an outgroup in phylogenetic analyses. The best-fit DNA substitution models were chosen as TPM3uf + I + G (trnK data matrix) and TIM3+I (ITS data matrix) in jmodeltest v.2.1 (Darriba et al., 2012). Bayesian inference (BI) was performed for ten million generations, sampling every 1,000 generations in MrBayes v.3.2, with independent Markov chain Monte Carlo (MCMC) chains (Ronquist et al., 2012). The first 25% of the trees were discarded as burn-in, and the remaining trees were used to generate a majority-rule consensus tree. The MCMC output was examined and the effective sample size (ESS) values were above 200. Finally, the generated trees were visualized and adjusted using FigTree software (https://tree.bio.ed.ac.uk).

The pan-tropical C. filiformis species are the most widely studied compared with other Cassytha species, but the morphology of C. larsenii has not been fully reported to date. To this end, the morphological characters of two Cassytha species were observed by stereoscopic microscope (SM) and scanning electron microscope (SEM). The following morphological characters were focused on: haustoria, stems, petals, and fruits. For stereoscopic microscope observation, the haustoria, stems, and flowers were placed upright on a flat, wet tissue paper and observed with Leica S8 APO, LAS v 4.8 collecting photographs. SEM observations of the materials proceeded in ZEISS EVO LS10 scanning electron microscope at the Public Technology Service Center, Xishuangbanna Tropical Botanical Garden, Chinese Academy of Sciences.

All eight newly sequenced Cassytha plastomes were assembled into single circular molecules lacking the typical inverted repeat (IR) region and quadripartite structure (Figure 2). The size of the plastomes varied from 114,215 bp in C. filiformis growing in Puerto Rico (P304) to 114,618 bp in C. filiformis growing in Guangdong (SY6130), while the sizes range was from 114,900 bp in C. larsenii growing in Yunnan (SY9917) to 114,988 bp in C. larsenii growing in Guangxi (SY6156). The GC content was similar in all eight plastomes, with a value of 37.0%. Each plastome contained a total of 107 functional genes, including 73 protein-coding genes, 30 tRNA genes, and four rRNA genes (Table 2). Notably, unlike the NADH dehydrogenase (ndh) genes found in other sequenced Lauraceae plastomes, five ndh genes including ѱndhB, ѱndhD, ѱndhE, ѱndhF, and ѱndhH are pseudogenes, and six ndh genes, including ndhA, ndhC, ndhG, ndhI, ndhJ, and ndhK are absent in all eight Cassytha plastomes. In addition, we identified three genes pafI, pafII, and pbf1 in all of these genomes.

We assessed synteny and rearrangements in the eight Cassytha plastomes and found no large-scale recombination in the gene organization after verification. However, we manually detected 24 micro-inversions, ranging from 2 to 46 bp, in the regions of the accD, accD-psaI, accD-rbcL, atpA-trnR, atpE-trnM, ccsA-psaC, clpP intron, psbA-trnH, petA-psbJ (three regions), petD-rpoA, petL-psbE, psbC-trnS, psbM-trnD, psbN-psbT, rpl32-trnL (two regions), rpoB-trnC, rps7-trnH, rps16 intron, trnG-trnR, and ycf2 (two regions) (Table 3). Palindrome sequences in pairs with lengths of 3–23 bp were identified in the flanks of these inversions. Furthermore, we detected a total of 249 indels in the Cassytha species, which were classified into 195 simple sequences repeat (SSR) indels and 54 non-SSR indels (Supplementary Table S1).

In comparison to the previous published plastome of Eusideroxylon zwageri Teijsm. & Binn. (LAU00006), which is an early divergent species in the Lauraceae family, the Cassytha plastomes have four missing segments (Figure 3). These missing segments include a 4 kb fragment containing three ndh genes (ndhC, ndhJ, and ndhK), a 16 kb fragment flanked by ndhB and ycf1, a 4 kb fragment containing three ndh genes (ndhA, ndhG, and ndhI), and a 10 kb fragment flanked by rpl2 and trnL-CAA. The missing segments with the length of 16 kb and 10 kb are located in the IRa and IRb regions of the E. zwageri plastome, respectively. The other two missing segments with a length of 4 kb are located in LSC and SSC regions of E. zwageri plastome, respectively. Additionally, when compared with the Cassytha plastome, a 1.5 kb segment containing the rpl2 gene is absent from the IRa region of the E. zwageri plastome.

The mVISTA results show that the non-coding regions of the eight plastomes have higher levels of divergence than the coding regions. There are some gaps in the intergenic spacer regions of P304 and P306 of C. filiformis (Figure 4A). A total of 734, 200, and 193 SNP markers were detected in the plastomes of both Cassytha species, C. larsenii individuals, and C. filiformis individuals, respectively. The sequence divergence levels among the plastomes of C. filiformis and C. larsenii were determined (Figure 4B). Within the two Cassytha species, these values varied from 0 to 0.0168, with a mean of 0.0034. Within C. filiformis, these values varied from 0 to 0.0033, with a mean of 0.0004. Within C. larsenii, these values varied from 0 to 0.0053, with a mean of 0.0006. The pairwise nucleotide divergence values between two of the four plastomes varied from 0.000017 to 0.000967 in C. filiformis and from 0.000139 to 0.000715 in C. larsenii. The values between the two species varied from 0.005325 to 0.005839 (Table 4). These results indicate that the differences between the two species were more than six times higher than those among individuals. Five regions, namely, trnQ-psbK, trnP-psaJ, rpl23-ycf2, ndhE-ndhH, and trnN-rrn5 were particularly highly variable between C. filiformis and C. larsenii.

To ascertain the phylogenetic placement of Cassytha species in relation to other members of the Lauraceae family with fully sequenced plastid genome sequences, we employed the complete plastomes of three Cassytha species to reconstruct phylogenetic relationships. We used two plastomes of Illigera species as out-groups. The phylogeny derived from the analysis of 61 complete plastid genome sequences is highly supported. Our phylogenetic analysis shows that the three Cassytha species form a sister clade to a group consisting of species belonging to the tribes Neocinnamomeae, Caryodaphnopsideae, and Laureae. The tribe Cryptocaryeae represents the next sister groups, followed by Illigera species (Figure 5A). The branch length in the maximum likelihood (ML) tree are 4.934 × 10⁻⁴ for four individuals of C. filiformis and 2.247 × 10⁻³ for four individuals of C. larsenii. To further investigate the phylogenetic relationships among the eight sequenced Cassytha individuals and other Cassytha taxa with reported barcoding data, we downloaded available trnK sequences from NCBI database. We included 91 Cassytha samples, with an outgroup accession of C. melantha, in the analysis of the data matrix with the length of 903 bp. The result of the Bayesian analysis shows that C. filiformis is sisters to C. larsenii, rather than C. ciliolata (Figure 5B). The branch lengths in the Bayesian inference (BI) tree are 1.997 × 10⁻³ for 57 individuals of C. filiformis and 5.902 × 10⁻⁴ for six individuals of C. larsenii. Finally, we downloaded available ITS sequences from GenBank and reconstructed the phylogeny consisted of 19 ITS sequences with a length of 579 bp. We used C. pubescens as an out-group (Figure 5C). The result of the Bayesian analysis shows that C. filiformis and C. larsenii individuals form two independent groups and the branch lengths are 1.204 × 10⁻² for ten individuals of C. filiformis and 6.401 × 10⁻³ for eight individuals of C. larsenii.

For C. larsenii, the scattered hairs on the haustoria were stable (Supplementary Figure S2A), and filiform hairs were observed on young stems (Supplementary Figure S2B) but not on annual and biennial stems (Supplementary Figure S2C), whereas stems with indumenta of simple, multicellular hairs were observed in two accessions of C. filiformis in China (Supplementary Figures S2G–I). For both C. larsenii and C. filiformis, the glabrous petal surfaces were observed with low magnification light microscopy (Supplementary Figures S2D, J), however, villous hairs were clearly observed on the edges of petals and petal surfaces in the electron microscopy photos (Supplementary Figure S2E, K). All members of genus Cassytha have fleshy fruits with a single seed, and the fruits of C. filiformis are oval with the mean size of 8 mm × 8 mm, while the fruits of C. larsenii are ellipsoid with the mean size of 5 mm × 7 mm. Compared with C. larsenii, C. filiformis has rounder and larger fruits (Supplementary Figure S2F, L).

This study produced eight complete plastid genomes for two species of the stem hemiparasitic genus Cassytha, which comprises nearly 23 species. The plastomes of Cassytha with the length of 114,215 ∼ 114,988 bp was significantly smaller than the other published plastomes in the family Lauraceae (Song et al., 2017b; Song et al., 2020). Three main reasons for these size differences were detected through comparative genomics analysis (Figure 3). First, one copy of the IR regions with the length of 24,717 bp was complete in E. zwageri but lost in the Cassytha plastomes with two segments, contributing almost 25 kb to the length difference. Second, the eight Cassytha plastomes have no six ndh genes including ndhA, ndhC, ndhG, ndhI, ndhJ, and ndhK. Two missing fragments with the length of 4 kb consist of three of the six genes, respectively, and their intergenic regions, which contributed around 8 kb to the length difference with Cassytha species. Third, five pseudogenes were detected in the eight Cassytha plastomes. Three ndh genes (ndhB, ndhD, and ndhE) were found to be pseudogenized, similar to C. filiformis (Wu et al., 2017). The length of ѱndhB, ѱndhD, ѱndhE, ѱndhF, and ѱndhH in C. filiformis are 1,191 bp, 428 bp, 159 bp, 302 bp, and 547 bp, respectively. In the plastome of E. zwageri, the length of the five ndh genes are 2,181 bp for ndhB, 1,508 bp for ndhD, 306 bp for ndhE, 2,229 bp for ndhF, and 1,182 bp for ndhH, which contributed around 7 kb to the length difference with Cassytha species.

Comparative genomic analysis indicated that there are 1,013 mutation sites including 24 micro-inversions, 249 indels, and 740 substitutions in the eight plastomes, which indicated that the nucleotide mutation sites in the plastomes of Cassytha species are more than that between species of Machillus (one micro-inversion, 65 indels, and 231 substitutions) and Phoebe (three micro-inversions, 73 indels, and 146 substitutions) (Song et al., 2015; Song et al., 2017a). The nucleotide variability values of the whole plastomes among the eight individuals from two Cassytha species were 0.34%, which approximates the nucleotide variability of five taxa (0.32%) (Song et al., 2016), 15 taxa (0.31%) (Liu et al., 2022), 18 taxa (0.37%) (Liu C. et al., 2021) in the tribe Laureae, and was much higher than the sequence divergence among three Alseodaphne species (0.12%) (Song et al., 2018), seven trinerved Lindera species (0.15%) (Tian et al., 2019), and seven Ocotea species (0.10%) (Trofimov et al., 2022).

With species from 27 genera of Laurales, our phylogenomic analysis based on 61 plastid genomes supported a monophyletic Cassytha clade comprising species of C. filiformis, C. larsenii, and C. pubescens. Species of Beilschmiedia, Cryptocarya, Endiandra, Eusideroxylon, Potameia, Sinopora, and Syndiclis, formed the Beilschmiedia-Cryptocarya clade in the phylogeny, and the third clade including Neocinnamomeae, Caryodaphnopsideae, and Laureae species is separate from both the Beilschmiedia-Cryptocarya clade and Cassytha clade, as in previously published phylogenetic trees in the family Lauraceae (Song et al., 2017b; Song et al., 2020). The deep relationships of Cassytha taxa are separated into the following groups in our study. C. melantha, endemic to Australia, forms the first group in the phylogeny. C. glabella, endemic to Western Australia, forms the second group. The third group includes four Australia species C. muelleri, C. pubescens, C. racemose, and C. rufa. The fourth group includes two Asia species C. capillaris and C. pergracilis. And the last group includes an Africa species C. ciliolata, an Asia species C. larsenii, and the pantropical species C. filiformis. The phylogenetic placements of most groups are consistent with previously published phylogenetic relationships (Kokubugata et al., 2012), and the position of C. larsenii was firstly settled here in the way predicted from morphology.

Although C. ciliolata, C. filiformis, and C. larsenii form the same group in the phylogeny, the persistence of hairs on stems and petal surfaces was used to distinguish the Cassytha species (Kokubugata et al., 2012). In C. ciliolata, filiform hairs on stems and glabrous or rufous-hispidulous twig tips were descripted on the basis of type specimens collected from mountains near Cape Town in Africa (Stapf, 1912). In C. larsenii, absence or sparsity of hairs on stems and glabrous or rufous-hispidulous twig tips were observed in all samples in China (Kostermans, 1994). In C. filiformis, it is remarkable that glabrous stems were observed not only in samples from Ryukyus of Japan and Taiwan of China but also in samples from Luzon Island and Rota Island in the Pacific (Kokubugata et al., 2012). However, stems with indumenta of simple, multicellular hairs were observed in C. filiformis samples from Australia, China, Japan, and Malaysia (Kokubugata et al., 2012). Therefore, glabrous stem is not a taxonomic key character for identifying C. ciliolata, C. filiformis and C. larsenii. Based on the type specimens, the glabrous petal surface was descripted as a common taxonomic character for C. ciliolata, C. filiformis and C. larsenii. However, villous petal surfaces and pubescent edge of petals were observed in the electron microscopy photos of C. filiformis and C. larsenii. Thus, presence or absence of hairs on petal surfaces is not appropriate as a taxonomic key character for identifying the three species. Finally, we suggest that, presence or absence of hairs on haustoria should be treated as key taxonomic evidence to distinguish C. filiformis and C. larsenii. Also, fruit shape or size can be a well distinction between the two species. Diversity of fruits may be related to growing in various geographical environments (Table 5).

Cassytha larsenii Kosterm. was only known from Khun Yuam District, Thailand, this is the first record from China (Puer, Yunnan Province; Guangzhou, Guangdong Province; Linshui County, Hainan Province and Dongxing, Guangxi Province). It was found hosting as a hemiparasitic on the tree trunks of Fabaceae, Myrtaceae and Asteraceae at altitudes from 900 to 1900 m in the forest (Zhang et al., 2022). C. larsenii was observed flowering in June in Mojiang of Yunnan. Morphologically, it bears resemblance to C. filiformis due to hairs on petal surfaces, but differs from it in terms of the presence or absence of hairs on haustoria, shorter inflorescence, smaller flowers (0.5–0.75 mm) (Kostermans, 1994), and smaller fruits. Additionally, C. filiformis grows along coastal regions while C. larsenii is found in mountainous areas. Furthermore, strong evidence from our phylogenetic studies supports treating the new records of C. larsenii from China as a distinct species from C. filiformis. Using different analyses, in the present study we showed that the biodiversity of the genus Cassytha in China is underestimated, with more species than previously recognized (Li et al., 2008). And, evidences from plastid genome size, phylogenetics, and morphology characters suggest that at least two species partitions would require validation and formal description (Liu Z. F. et al., 2021). Our study provides important insights into the taxonomic, biodiversity, conservation biology, and phylogeographic of the genus Cassytha.