Research Article |
Corresponding author: Renata Piwowarczyk ( renata.piwowarczyk@ujk.edu.pl ) Academic editor: Eberhard Fischer
© 2021 Renata Piwowarczyk, Adam C. Schneider, Grzegorz Góralski, Dagmara Kwolek, Magdalena Denysenko-Bennett, Anna Burda, Karolina Ruraż, Andrzej J. Joachimiak, Óscar Sánchez Pedraja.
This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Citation:
Piwowarczyk R, Schneider AC, Góralski G, Kwolek D, Denysenko-Bennett M, Burda A, Ruraż K, Joachimiak AJ, Pedraja ÓS (2021) Phylogeny and historical biogeography analysis support Caucasian and Mediterranean centres of origin of key holoparasitic Orobancheae (Orobanchaceae) lineages. PhytoKeys 174: 165-194. https://doi.org/10.3897/phytokeys.174.62524
|
The extensive diversity of the tribe Orobancheae, the most species-rich lineage of holoparasitic Orobanchaceae, is concentrated in the Caucasus and Mediterranean regions of the Old World. This extant diversity has inspired hypotheses that these regions are also centres of origin of its key lineages, however the ability to test hypotheses has been limited by a lack of sampling and phylogenetic information about the species, especially in the Caucasus region. First, we assessed the phylogenetic relationships of several poorly known, problematic, or newly described species and host-races of four genera of Orobancheae occurring in the Caucasus region–Cistanche, Phelypaea, Phelipanche and Orobanche–using nuclear ribosomal (ITS) and plastid (trnL–trnF) sequence data. Then we applied a probablistic dispersal-extinction-cladogenesis model of historical biogeography across a more inclusive clade of holoparasites, to explicitly test hypotheses of Orobancheae diversification and historical biogeography shifts. In sum, we sampled 548 sequences (including 196 newly generated) from 13 genera, 140 species, and 175 taxa across 44 countries. We find that the Western Asia (particularly the Caucasus) and the Mediterranean are the centre of origin for large clades of holoparasitic Orobancheae within the last 6 million years. In the Caucasus, the centres of diversity are composed both of long-branch taxa and shallow, recently diversified clades, while Orobancheae diversity in the Mediterranean appears to represent mainly recent diversification.
Biodiversity hotspot, chronogram, Cistanche, divergence time, historical biogeography, Orobanche, Phelipanche, Phelypaea
The tribe Orobancheae is the oldest and most species-rich of the three lineages of holoparasites comprising the cosmopolitan family Orobanchaceae, with a crown age dating to the mid-Miocene (
The Mediterranean Basin and Caucasus region of western Asia are centres of extant diversity for the two most diverse genera in the Orobancheae, Orobanche L. and Phelipanche Pomel (ca. 150 and 60 described species, respectively) (
While regions of high extant diversity for any lineage may be the result of in situ diversification, this is not necessarily the case. Thus, hypotheses of historical biogeography must be explicitly tested.
The aims of this study were two-fold. First, we sought to assess previously unknown phylogenetic relationships of Caucasian Orobancheae using nuclear ribosomal (ITS region) and plastid (trnL–trnF) DNA sequences. Second, we sought to evaluate the historical biogeography of Old World Orobancheae using a probabilistic dispersal-extinction-cladogenesis (DEC) model. In particular, we wanted to evaluate the hypothesis of Western Asia (especially the Caucasus) and the Mediterranean as potential refugia and/or centres of origin for major species-rich clade in the Orobancheae.
For the initial phylogenetic analysis, we studied Caucasian species of Cistanche, Phelypaea, Phelipanche and Orobanche, mainly collected from Georgia, Armenia, Azerbaijan and Russia between 2014 and 2019. Specimens of some species were collected in other countries or taken from herbaria (B, ERCB, HMMNH, IRKU, KTC, LE, MW, herb. Ó. Sánchez Pedraja), or sequences were downloaded from GenBank. In total, 13 genera, 175 taxa representing 140 species (548 sequences, including 196 as new), from 44 countries, were analysed (see Suppl. material
Rooted Maximum Likelihood phylogenetic tree constructed using ITS sequences. Numbers near branches show ultrafast bootstrap values (values ≥ 75 are shown). The bar represents the amount of genetic change (nucleotide substitutions per site) A summary of backbone (generic) relationships, branches connecting the outgroup Lindenbergia and Boulardia are shortened to fit the figure B–E relationships of taxa within the genera Cistanche, Phelipanche, Phelypaea, and Orobanche respectively. Species names, the country of origin, host species (if available) and GenBank number are included on the phylogeny tip labels.
Material used for DNA extraction was freshly collected and silica gel-dried or was obtained from herbarium vouchers. For phylogenetic studies we used two types of sequences: nuclear ITS region (internal transcribed spacer 1, 5.8S ribosomal RNA gene, internal transcribed spacer 2, later referred to as ITS) and plastid trnL–trnF sequence (RNA-Leu (trnL) intron, the partial trnL gene, and the intergenic spacer between the trnL 3’ exon and tRNA-Phe (trnF) gene region’s plastid DNA). These two regions are commonly used for species-level phylogenetic inference, including in the Orobancheae (ITS:
Sequences were aligned with MAFFT v7.407 (
For both sequence alignments, Maximum Likelihood (Figs
Rooted Maximum Likelihood phylogenetic tree constructed using plastid trnL–trnF spacer sequences. As an outgroup, Lindenbergia sinaica was used. Numbers near branches show ultrafast bootstrap values (values ≥ 75 are shown). The bar represents the amount of genetic change (nucleotide substitutions per site) A summary of backbone (generic) relationships B Phelipanche clade C Orobanche clade. Species names, the country of origin, host species (if available) and GenBank number are included on the phylogeny tip labels.
To infer a chronogram for historical biogeography analysis of the tribe Orobancheae we used the ITS, PhyA, and PhyB Orobancheae alignments of
1. Taxonomic coverage for Orobanche and Phelipanche was expanded based on this study.
2. Taxonomic coverage for Cistanche was expanded by using sequence data submitted to GenBank by
3. Sequences for Gleadovia Gamble & Prain and Phacellanthus Siebold & Zucc. – first published by
4. The trnL–trnF plastid locus was added for most taxa based on newly generated data or pre-existing sequences (Suppl. material
5. A 637 bp region of the PhyA gene was excluded from analysis because it was poorly alignable. This region appears only in our sequences for Boschniakia himalaica Hook. f. & Thomson ex Hook. f. and Aphyllon ludovicianum (Nutt.) A.Gray but not for any other species.
6. Samples for Aphyllon californicum subspecies feudgei, grande, grayanum, and jepsonii were replaced with different samples for which both ITS and trnL–trnF sequences were available.
Sequences matrices for each gene were aligned separately using Geneious 9.1.8 (Biomatters, Auckland, New Zealand;
Each iteration consisted of 472 moves randomly scheduled from 394 possible moves. Stationarity was assessed using Tracer v.1.7.1 (
For biogeographical analysis, the global range of Orobancheae was divided into six non-overlapping regions based on physical geography and natural phytogeographic divisions (Fig.
Historical biogeography of tribe Orobancheae, reconstructed using a dispersal-extinction-cladognesis model implemented in RevBayes (maximum likelihood topology, maximum clade credibility branch lengths). Coloured circles at tips represent the current biogeographical range of each sampled taxon. Circles on each node represent the reconstructed ancestral area of the most recent common ancestor of the two daughter lineages, while circles on either side of the node show the reconstructed areas immediately following cladogenesis. Circle size is proportional to posterior probability. Each colour represents a different biogeographical region or combination of regions as indicated by the map and legend to the left of the chronogram. Tip labels for Cistanche follow nomenclature of
Ancestral geographical ranges were inferred by applying a dispersal-extinction-cladogenesis (DEC) model of historical biogeography to the maximum clade credibility (MCC) tree from the Bayesian analysis. The DEC model, also implemented in RevBayes, allows for sympatric speciation, allopatric speciation and anagenetic range expansion and contraction (
The most important results of our phylogenetic analyses clarified the position of many previously unsampled Caucasian species (Figs
Consistent with previous studies, the studied genera were each strongly supported as monophyletic (Bootstrap (BS) ≥ 90, Posterior Probability (PP) = 1.0).
ITS (trnL–trnF data was not available) trees show that Cistanche armena (K. Koch) M.V. Agab. (samples from two different hosts, Alhagi Gagnebin and Salsola L.) is closely related to C. deserticola Ma and C. salsa (C.A. Mey.) Beck (BS = 100, PP = 1.00), and with the later one it has sometimes been confused (Fig.
The three species from genus Phelypaea, P. tournefortii Desf. and P. coccinea (M. Bieb.) Poir. are clearly separated (BS = 100, PP = 1.00), however P. boissieri (Reut.) Stapf, first sequenced for this study, seems to be very similar to P. coccinea. Amplification of trnL–trnF in Phelypaea samples was successful only in the case of P. coccinea, so the above analysis was based only on ITS (Fig.
Based on ITS data P. zangezuri is separated from the clade of P. caesia (Rchb.) Soják (BS = 97, PP = 0.90) and the clade containing remain Phelipanche species (BS = 98, PP = 0.85). By contrast, trnL–trnF trees do not indicate separation of P. zangezuri and P. caesia. Rather, samples of P. arenaria form a sister clade to these two species, and together form a well-supported lineage (BS = 98, PP = 1.00) separated from the rest of Phelipanche (BS = 95, PP = 0.96) (Figs
Our results showed the relationship of samples from different parts of the range of disjunctive species, such as P. portoilicitana (A. Pujadas & M.B. Crespo) Carlón et al. and P. cernua. Whereas trnL–trnF sequences of P. cernua places samples from Armenia and Spain are grouped in the same clade (BS = 98, PP = 0.94), on the ITS tree, the European samples are separated from Caucasian sample which is in the same clade as P. sevanensis, P. schultzii and P. heldreichii (BS = 99, PP = 0.99). Also, P. portoilicitana, both on ITS and trnL–trnF trees, show differences between samples from Armenia and Spain (Figs
Orobanche gamosepala Reut. is genetically distinct (BS = 100, PP = 1.00) from O. anatolica Boiss. & Reut. ex Reut./O. colorata K. Koch and together these species are grouped in sister clade to the rest of Orobanche species (ITS: BS = 99, PP = 1.00, trnL–trnF: BS = 100, PP = 1.00) (Figs
ITS sequence data indicates that O. cicerbitae (Uhlich & Rätzel) Tzvelev is not closely related to O. flava Mart. ex F.W. Schultz, however on the trnL–trnF trees O. cicerbitae from Georgia and Azerbaijan forms a common clade with O. flava from Georgia (BS = 98, PP = 0.97), whereas Central European samples of O. flava are distant (Fig.
ITS sequences (Fig.
The phylogenetic position of Caucasian endemic species with unclear affinity has also been presented, in particular those previously classified in inappropriate subsections, such as O. schelkownikovii Tzvel., O. grossheimii Novopokr., O. raddeana Beck, and O. laxissima Rätzel & Uhlich (Figs
Little within-species variation is shown among the samples from different host species taken from the following species: O. laxissima, O. alba Stephan ex Willd., O. bartlingii Griseb., O. caryophyllacea Sm., O. cicerbitae, O. gracilis Sm., O. centaurina Bertol., O. minor Sm., O. owerinii (Beck) Beck, O. raddeana, O. schelkovnikovii, P. cilicica, P. coelestis (Reut.) Soják, P. purpurea (Jacq.) Soják and P. coccinea (Figs
We find negligible support (PP < 0.4) for any single hypothesis ancestral range of lineages older than 6 million years. However, most diversification in the Orobancheae has happened relatively recently (Tables
Divergence times with credible intervals (95% highest probability density (HPD)) and inferred historical biogeography of selected clades. Biogeographical regions defined in Methods and Figure
Clade | Crown Age (Ma) | Biogeography | ||
---|---|---|---|---|
Mean | 95% HPD | Region | Posterior Prob. | |
Cistanche sect. Heterocalyx | 3.0 | 2.0–4.2 | Central Asia | 0.28 (0.42 for clade excluding C. deserticola) |
Western Asia | 0.14 (0.27) | |||
Phelypaea | 2.4 | 1.4–3.7 | Western Asia | 0.46 |
Western Asia + Med/Europe | 0.11 | |||
Phelipanche | 3.9 | 2.8–5.5 | Western Asia | 0.41 |
Europe/Mediterranean | 0.09 | |||
Both | 0.11 | |||
Phelipanche clade P1 | 1.8 | 1.1–2.4 | Western Asia | 0.38 |
Western Asia + Med/Europe | 0.32 | |||
Phelipanche clade P2 | 0.753 | 0.52–1.1 | Western Asia | 0.58 |
Western Asia + Europe/Med | 0.34 | |||
Orobanche clade O1 | 0.44 | 0.26–0.67 | Europe/Mediterranean | 0.93 |
Orobanche clade O2 | 0.72 | 0.46–1.0 | Europe/Mediterranean | 0.99 |
Orobanche clade O1+O2 | 0.81 | 0.52–1.1 | Western Asia + Europe/Med | 0.50 |
Europe/Mediterranean | 0.24 | |||
Orobanche clade O3 | 1.27 | 0.75–1.8 | Western Asia | 0.93 |
Orobanche clade O4 | 0.75 | 0.40–1.1 | Europe/Mediterranean + Western Asia | n/a |
+ Central Asia | 0.12 | |||
+ East Asia | 0.10 | |||
+ Both | 0.27 |
Divergence times of species or clades endemic or nearly endemic to the Caucasus region.
Species or clade, or paraphyletic groupa | Taxa | Divergence timea (Ma) | 95% HPD |
---|---|---|---|
Clade | Phelypaea coccinea + P. tournefortii | 2.4 | 1.4–3.7 |
Clade | Orobanche anatolica + O. colorata + O. gamosepala | 5.4 | 3.5–7.6 |
Paraphyletic | Orobanche arpica, O. cicerbitae on Caucasalia, O. cicerbitae, O. inulae, O. mlokosiewiczii, O. cicerbitae on Pojarkovia (+ widespread O. krylowii) | 1.3 | 0.75–1.8 |
Species | Orobanche zajaciorum | 1.7 | 1.2–2.3 |
Species | Orobanche raddeana | 2.2 | 1.7–3.3 |
Species | Orobanche grossheimii | 1.2 | 0.74–1.5 |
Clade | Orobanche laxissima + O. owerinii + O. transcaucasica | 0.08 | 0.02–0.15 |
Species | Orobanche javakhetica | 2.4 | 1.6–3.2 |
Species | Orobanche kurdica | 0.08 | 0.0002–0.23 |
Species | Orobanche schelkovnikovii | 0.50 | 0.30– 0.69 |
Species | Phelipanche bungeana | 1.48 | 0.93–2.0 |
Clade | Phelipanche coelestis + Phelipanche “on Astrodaucus” | 0.15 | 0.05–0.26 |
Species | Phelipanche hajastanica | 0.28 | 0.16–0.40 |
Clade | Phelipanche heldreichii + P. sevanensis | 0.19 | 0.08–0.32 |
Species | Phelipanche “on Artemisia” | 1.1 | 0.52–1.8 |
Species | Phelipanche “on Genista” | 0.20 | 0.10–0.31 |
Species | Phelipanche pulchella | 0.41 | 0.21–0.61 |
Species | Phelipanche zangezuri | 0.68 | 0.33–1.1 |
The phylogeny of Cistanche appears to be structured by geography, with clades of species endemic to particular areas. For example, we find weak support for a Central Asian ancestor of Cistanche sect. Heterocalyx sensu Ataei, non Beck (composed of C. salsa, C. bamianica Ataei ined. (
We found support that the clade of species C. algeriensis Ataei ined. (
We found moderate support for a Western Asia origin of Phelipanche (PP = 0.41) approximately 2.8–5.5 million years ago, with alternative biogeographical hypotheses much more weakly supported (Table
Similarly, we find it most probable that the most widespread and often weedy species of Phelipanche had direct stem ancestors limited in range to Western Asia. These include Phelipanche arenaria (Borkh.) Pomel (PP = 0.71), P. caesia (PP = 0.5), P. ramosa (L.) Pomel (PP = 0.90), and P. aegyptiaca (PP = 0.40, with the next most probable origin as Europe/the Mediterranean, PP = 0.22).
Similar to Phelipanche we infer a Western Asian origin for ancestral Orobanche (PP = 0.43; 0.39 for Orobanche + Boulardia). Four key subclades are diagnosable by their biogeographic affinities. The first and second subclades are closely related and comprise predominantly Europe/Mediterranean species that have diversified in situ (O1 + O2 in Table
C. armena was described by
This genus includes three holoparasite species (P. coccinea, P. boissieri, and P. tournefortii) that parasitize Asteraceae hosts. Phelypaea coccinea, a parasite of Psephellus Cass. and Centaurea L., rarely Klasea Cass., occurs in the Caucasus and Crimea, while P. tournefortii, a parasite of Tanacetum L., occurs in the Caucasus and Turkey (
The phylogenetic relations of the newly described species, i.e., P. zangezuri (
Phylogenetic analysis of two species previously known mainly from the Mediterranean area and later found in the Caucasus, i.e., P. portoilicitana and P. cernua (
ITS (Fig.
The recently described O. flava subsp. cicerbitae Uhlich & Rätzel [≡ O. cicerbitae (Uhlich et Rätzel) Tzvelev] parasitising Cicerbita Wallr. and Senecio propinquus Schischk. is distantly related to O. flava, at least as far as ITS (Fig.
We confirm that the newly described O. javakhetica (
Orobanche schelkovnikovii was incorrectly included in the O. trib./Grex Galeatae sensu Beck by
According to
The newly described species O. zajaciorum (
Orobanche rapum-genistae Thuill., O. rigens Loisel. vs O. colorata/O. anatolica placed by
Orobanche gamosepala is genetically very distinct, yet nested within Orobanche, forming a clade with O. anatolica/O. colorata (O. subsect. Arcuatae) (ITS: BS = 99, PP = 1.00, trnL–trnF: BS = 100, PP = 1.00) that is sister to the clade containing all other Orobanche species (Figs
According to some authors (e.g.,
Orobanche raddeana is a Caucasian endemic parasitising on Campanulaceae (Campanula L., Asyneuma Griseb. & Schenk). The ITS tree may suggest that it is related to species from the subsect. Glandulosae (Fig.
Within the O. subsect. Inflatae Beck, O. grenieri (parasitic on mainly Lactuca L.) is clearly distinguished morphologically and phylogenetically from related species (O. cernua and O. cumana), as has already been shown (
Species from subsect. Minores subsect. Speciosae Teryokhin are highly polymorphic, especially regarding colour, inflorescence length and variability of flower, as well as range of hosts. In our research we used Caucasian samples of O. laxissima (a parasite of various tree species, i.e., Fraxinus L., Carpinus L., Punica L., Robinia L.), O. owerinii (a parasite of herbaceous hosts, i.e., Trifolium L., Vicia L.), and O. minor (samples from Chondrilla L. and Lactuca hosts) (Figs
Molecular studies do not indicate the validity of dividing species into subsect. Speciosae because the species included here are both very morphologically and genetically similar to the subsect. Minores. Similar conclusions can be used to merit the inclusion of Vitellinae Teryokhin, Hederae Teryokhin, and Camptolepides Teryokhin in separate subsections when they are clearly similar to species from the subsect. Minores and Inflatae (respectively). The results presented here suggest that the currently distinguished systematic division of Orobanche-based morphology is frequently inconsistent with the phylogenetic studies and thus needs revision, regarding both phenotypic traits and molecular analyses, for example, the heterogeneous subsect. Curvatae is clearly resolved as polyphyletic (Fig.
We found strong support for Western Asia as the centre of origin for large subclades of Phelipanche, Orobanche, and Cistanche (Table
The broader floristic and geological history of the Caucasus and high mountain region does provide some clues to the processes that its status as a centre of extant diversity, a centre of origin for large portions of this diversity, and potentially a region of mixed endemism for holoparasitic Orobancheae. The Caucasus has an unusually high proportion of endemic and relict species for a continental, non-tropical region (
However, we also found a number of very recent diversification events in Orobancheae, pointing to recent in situ speciation as a complementary mechanism that explains the high levels of endemism in this region (Table
By contrast, the biodiversity of Orobanche and Cistanche that evolved in Europe and especially in the Mediterranean Basin appears to have done so more recently than that in West Asia, although we cannot confidently infer ancestral states of lineages greater than 5 million years (Table
We conclude with a cautionary note that we were not able to exhaustively sample the Orobancheae, in particular certain species of Orobanche, such as O. sect. Kotschyinae Teryokhin from the Middle East and western and central Asia. The addition of certain other lineages, such as species in O. subsect. Coerulescentes Teryokhin would likely strengthen the importance of diversification in East Asia. Finally, our results within Cistanche are sensitive to changing taxonomic concepts.
The authors thank the curators of herbaria and other people who kindly made their data and samples of the species available. Will Freyman provided computational resources.
This work was partially financed by the National Geographic grant GEFNE 192-16 (2017), Polish State Committee for Scientific Research (KBN grant no. NN303357733 (2008–2009), NN303551939 (2010–2013), the Research Projects of the Jan Kochanowski University in Kielce 612419 (2014–2017) and SMGR.20.208-615 (2020) for Renata Piwowarczyk, and statutory research funds (K/DSC/002930) of the Institute of Botany, Faculty of Biology, Jagiellonian University, Kraków, Poland.
The DNA sequence data generated and analysed during this work are available in the GenBank repository [https://www.ncbi.nlm.nih.gov/genbank/]. Alignments for the biogeography data generated or analysed during this study are available in Suppl. material
Table S1
Data type: List of taxa and sequences analysed
Explanation note: List of taxa and sequences analysed (* sequences obtained from GenBank).
Figure S1
Data type: phylogenetic tree
Explanation note: Rooted Bayesian phylogenetic consensus tree inferred from ITS sequences (outgroup: Lindenbergia sinaica). Numbers near branches show Bayesian posterior probabilities ≥ 0.75. The bar represents the amount of genetic change (nucleotide substitutions per site). Species names, the country of origin, host species (if available) and GenBank accession number are included on the phylogeny tip labels.
Figure S2
Data type: phylogenetic tree
Explanation note: Rooted Bayesian phylogenetic consensus tree constructed using plastid trnL–trnF spacer sequences (outgroup: Lindenbergia sinaica). Numbers near branches show Bayesian posterior probabilities ≥ 0.75. The bar represents the amount of genetic change (nucleotide substitutions per site). Species names, the country of origin, host species (if available) and GenBank accession number are included on the phylogeny tip labels.
Data S1
Data type: DNA sequence alignment (fasta format)
Explanation note: Sequence matrix used for biogeography analysis.
Data S2
Data type: tree file
Explanation note: Historical biogeography reconstruction (maximum clade credibility tree).