The Madeiran laurel forest endemic Goodyera macrophylla (Orchidaceae) is related to American orchids

Taxon sampling

DNA sequence data of Goodyera macrophylla was analysed in a phylogenetic framework including Goodyera and other genera of subtribe Goodyerinae (Appendix 1). To test if G. macrophylla is part of other groups besides Goodyera in its wide circumscription, other genera of Goodyerinae; namely Erythrodes Blume, Kreodanthus Garay and Aspidogyne Garay were also included. Pterostylis R.Br. (Pterostylidinae; sistergroup of Goodyerinae; Givnish & al. 2015Givnish T.J., Spalink D., Ames M., Lyon S.P., Hunter S.J., Zuluaga A., Iles W.J., Clements M.A., Arroyo M.T. & Leebens-Mack J. 2015. Orchid phylogenomics and multiple drivers of their extraordinary diversification. Proceedings Royal Society of London Series B Biological Sciences 282: 20151553.) was used as outgroup. Seventeen DNA sequences of G. macrophylla (Sequeira 9073, 9074, 9114, 10600), G. repens (L.) R.Br. (Thiv 6213) and G. striata Rchb.f. [García 127 (P) MNHN-P-P01019179] were newly generated in the context of this study.

Laboratory protocols and data matrices

DNA extraction, PCR and sequencing protocols followed Schüßler & al. (2019)Schüßler C., Bräuchler C., Reyes-Betancort J.A., Koch M.A. & Thiv M. 2019. Island biogeography of the Macaronesian Gesnouinia and Mediterranean Soleirolia (Parietarieae, Urticaceae) with implications for the evolution of insular woodiness. Taxon 68: 537-556.. Genetic markers were chosen following Hu & al. (2016)Hu C., Tian H., Li H., Hu A., Xing F., Bhattacharjee A., Hsu T., Kumar P. & Chung S. 2016. Phylogenetic analysis of a ‘Jewel Orchid’ Genus Goodyera (Orchidaceae) based on DNA Sequence Data from Nuclear and Plastid Regions. PloS one 11: e0150366., Chen & al. (2019)Chen S.-P., Tian H.-Z., Guan Q.-X., Zhai J.-W., Zhang G.-Q., Chen L.-J., Liu Z.-J., Lan S.-R. & Li M.-H. 2019. Molecular systematics of Goodyerinae (Cranichideae, Orchidoideae, Orchidaceae) based on multiple nuclear and plastid regions. Molecular Phylogenetics and Evolution 139: 106542. and Shin & al. (2002)Shin K.-S., Shin Y.K., Kim J.-H. & Tae K.-H. 2002. Phylogeny of the genus Goodyera (Orchidaceae; Cranichideae) in Korea based on nuclear ribosomal DNA ITS region sequences. Journal of Plant Biology 45: 182.. Accordingly, the nuclear ITS region and the plastid trnL intron, trnL-F spacer and matK coding region were used for phylogenetic reconstruction (Appendix 1). Primers were obtained from Hu & al. (2016)Hu C., Tian H., Li H., Hu A., Xing F., Bhattacharjee A., Hsu T., Kumar P. & Chung S. 2016. Phylogenetic analysis of a ‘Jewel Orchid’ Genus Goodyera (Orchidaceae) based on DNA Sequence Data from Nuclear and Plastid Regions. PloS one 11: e0150366. except for G. striata, for which we designed new matK primers: matK 225R ACCAAAAATTTCCACAGGTTCGT, matK 225F ACGAACCTGTGGAAATTTTTGGT, matK 578R TCCAGATGGATGGGATGGGG, matK 592F TGCTGGATCAAAGATGTTCCT, matK 1031F GGTCTCAACCTTATAGGATCCATAT, matK 1053R TGGATCCTATAAGGTTGAGACCA, matK 1356R TGAGGATCCGCTGTGATAACG, matK 1345F CGTTATCACAGCGGATCCTCA, matK 1850R ACCGTGCTTGCAGTTTTCAT, matK 1831F ATGAAAACTGCAAGCACGGT. These primers yielded fragments of lengths between ca. 200 and 450 bp using 55°C as annealing temperature and 1 min. as elongation time.

For ITS and the plastid DNA markers available sequences of Goodyera species and related taxa (see taxon sampling) were downloaded from GenBank of the National Center for Biotechnology Information (NCBI). Identical and very similar sequences of the same taxon, often grouping together in a clade in a preliminary analysis, and doubtful, possibly incorrect determined sequences were not included. All newly generated sequences were deposited at GenBank, the corresponding accession numbers as well as a detailed list of the analysed samples are given in Appendix 1. DNA sequences were aligned using MAFFT v7.388 (Katoh & al. 2002Katoh K., Misawa K., Kuma K.I. & Miyata T. 2002. MAFFT: a novel method for rapid multiple sequence alignment based on fast Fourier transform. Nucleic Acids Research 30: 3059-3066.) in Geneious 11.1.5.

Data analyses and divergence time estimation

The ITS data set was 751 bp in length and included 45 species of Goodyera and 13 of Goodyerinae. In the cpDNA matrix 37 species of Goodyera and 8 other representatives of Goodyerinae were included. For non-coding parts of the cpDNA dataset several poly-A/T, very variable, repetitive regions or unique insertions were excluded from the analysis. This concerns the positions 72-96, 245-344, 412-512, 535-544, 576-599, 623-650, 721-730, 753-759, 782-790, 819-824, 1003-1016, 1069-1077, 1124-1149, 1206-1224, 1300-1305, 1341-1350, 1374-1388, 1421-1430, 3157-3180, 3301-3310, 3314-3318 of the original dataset (3400 bp) yielding a new matrix with 2929 bp.

Several methods were applied to reconstruct the phylogenetic relationships. Maximum likelihood (ML) trees were calculated using RaxML v8 (Stamatakis 2014Stamatakis A. 2014. RAxML version 8: a tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinformatics 30: 1312-1313.). Bayesian inference (BI) trees were generated using MrBayes 3.2.6 (Huelsenbeck & Ronquist 2001Huelsenbeck J.P. & Ronquist F. 2001. MRBAYES: Bayesian inference of phylogenetic trees. Bioinformatics 17: 754-755.). We used the GTR substitution model with four Gamma categories and the shape being estimated. Four runs of 20,000,000 generations with samples taken every 2,000 generations provided Effective Sample Size values > 200 in TRACER (Rambaut & Drummond 2007Rambaut A. & Drummond A.J. 2007. Tracer v1. 4. Website: https://beast.community/tracer [accessed: 15 Mar. 2020].).

To evaluate divergence times for G. macrophylla we used BEAST v2.5 (Drummond & al. 2012Drummond A.J., Suchard M.A., Xie D. & Rambaut A. 2012. Bayesian phylogenetics with BEAUti and the BEAST 1.7. Molecular Biology and Evolution 29: 1969-1973.) on the ITS and cpDNA data set. Because these data sets consisted mostly of different species of Goodyera, we used the Yule model and lognormal relaxed clock. The data sets were analysed separately because the position of G. macrophylla differed in the two data sets (see results). Assuming equal probabilities in the 95% HPD, we applied in our analyses two uniform priors following Givnish & al. (2015)Givnish T.J., Spalink D., Ames M., Lyon S.P., Hunter S.J., Zuluaga A., Iles W.J., Clements M.A., Arroyo M.T. & Leebens-Mack J. 2015. Orchid phylogenomics and multiple drivers of their extraordinary diversification. Proceedings Royal Society of London Series B Biological Sciences 282: 20151553. as secondary calibration points with 15.13-27.13 my to the most recent common ancestor (mrca) of Pristiglottis (plus other Goodyerinae genera) and all Goodyera species and 12.00-23.07 my to the mrca of Goodyera, Erythrodes and Kreodanthus (Figs. 1, 2). Several analyses have used fossils to date the origin of the orchid family. As example, Ramírez & al. (2007)Ramírez S.R., Gravendeel B., Singer R.B., Marshall C.R. & Pierce N.E. 2007. Dating the origin of the Orchidaceae from a fossil orchid with its pollinator. Nature 448: 1042-1045. dated the origin of the orchid family to the Cretaceous. Using a broad taxon sample, Givnish & al. (2015)Givnish T.J., Spalink D., Ames M., Lyon S.P., Hunter S.J., Zuluaga A., Iles W.J., Clements M.A., Arroyo M.T. & Leebens-Mack J. 2015. Orchid phylogenomics and multiple drivers of their extraordinary diversification. Proceedings Royal Society of London Series B Biological Sciences 282: 20151553. also dated the orchid family to the Cretaceous. They used a total of 17 calibration points for their family dating, among them several monocot fossils which were revised by Iles & al. (2015)Iles W.J., Smith S.Y., Gandolfo M.A. & Graham S.W. 2015. Monocot fossils suitable for molecular dating analyses. Botanical Journal of the Linnean Society 178: 346-374.. Moreover, they included some calibration points inferred from angiosperm phylogenies and the three fossil orchids from the Miocene belonging to Dendrobium Sw., Earina Lindl. (Conran et al. 2009Conran J.G., Bannister J.M. & Lee D.E. 2009. Earliest orchid macrofossils: Early Miocene Dendrobium and Earina (Orchidaceae: Epidendroideae) from New Zealand. American Journal of Botany 96: 466-474. ) and Meliorchis S.R.Ramirez, Gravend., R.B.Singer, C.R.Marshall, N.E.Pierce (Ramírez & al. 2007Ramírez S.R., Gravendeel B., Singer R.B., Marshall C.R. & Pierce N.E. 2007. Dating the origin of the Orchidaceae from a fossil orchid with its pollinator. Nature 448: 1042-1045.). According to Givnish’s & al. (2015)Givnish T.J., Spalink D., Ames M., Lyon S.P., Hunter S.J., Zuluaga A., Iles W.J., Clements M.A., Arroyo M.T. & Leebens-Mack J. 2015. Orchid phylogenomics and multiple drivers of their extraordinary diversification. Proceedings Royal Society of London Series B Biological Sciences 282: 20151553. results, within Goodyerinae, the group of Pristiglottis Cretz. & J.J.Sm and Goodyera is 20.68 my (95% HPD ca. 15.13-27.13) old and the split between Goodyera, Erythrodes and Kreodanthus is dated to 17.34 my (95% HPD ca. 12-23.07).

Biogeographic analyses

For biogeographic analyses, the following areas were coded: A: Eastern and South-Eastern Asia, B: temperate-boreal Asia, C: temperate North America, D: tropical Central America, E: Pacific region, F: Madeira (Macaronesia) and G: Europe. In a second alternative approach, temperate North America (C) and tropical Central America (D) were combined to America.

As input, the dated maximum clade credibility (MCC) consensus trees from the BEAST analyses were used after being reduced to taxa of the Goodyera subclade using Mesquite 3.70 (Maddison & Maddison 2021Maddison W. P. & Maddison D.R. 2021. Mesquite: a modular system for evolutionary analysis. Version 3.70. http://www.mesquiteproject.org. ). Different biogeographical models, i.e. dispersal-extinction-cladogenesis (DEC), Dispersal-Vicariance Analysis (DIVALIKE) and BayArea (BAYAREALIKE) were tested using the BioGeoBEARS R package (Matzke 2013aMatzke N.J. 2013a. BioGeoBEARS: BioGeography with Bayesian (and Likelihood) Evolutionary Analysis in R Scripts. Website: http://CRAN.R-project.org/package=BioGeoBEARS. [accessed: 5 Oct. 2020].; Matzke 2013bMatzke N.J. 2013b. Probabilistic historical biogeography: new models for founder-event speciation, imperfect detection, and fossils allow improved accuracy and model-testing. Frontiers of Biogeography 5: 242-248.). The inclusion of founder-event speciation (+ J) was also tested for each of these models.

Additionally, Bayesian Binary MCMC (BBM) analyses were conducted using RASP 3.2 (Ronquist & Huelsenbeck 2003Ronquist F. & Huelsenbeck J.P. 2003. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19: 1572-1574.; Yu & al. 2015Yu Y., Harris A.J., Blair C. & He X. 2015. RASP (Reconstruct Ancestral State in Phylogenies): a tool for historical biogeography. Molecular Phylogenetics and Evolution 87: 46-49.). The maximum number of ranges was set to two and nodes supported with posterior probabilities (PP) < 0.90 were excluded from analyses. The default setting of fixed state frequencies (JC) with equal among-site variation was chosen. For the MCMC settings, the default was used.

Phylogenetic relationships

Identical topologies and very similar support values were obtained using MrBayes (not shown), RAxML (not shown) and BEAST (Figs. 1, 2, 3). Accordingly, the overall patterns of phylogenetic reconstructions based on nuclear and plastid DNA largely correspond to the results of Hu & al. (2016)Hu C., Tian H., Li H., Hu A., Xing F., Bhattacharjee A., Hsu T., Kumar P. & Chung S. 2016. Phylogenetic analysis of a ‘Jewel Orchid’ Genus Goodyera (Orchidaceae) based on DNA Sequence Data from Nuclear and Plastid Regions. PloS one 11: e0150366. and Chen & al. (2019)Chen S.-P., Tian H.-Z., Guan Q.-X., Zhai J.-W., Zhang G.-Q., Chen L.-J., Liu Z.-J., Lan S.-R. & Li M.-H. 2019. Molecular systematics of Goodyerinae (Cranichideae, Orchidoideae, Orchidaceae) based on multiple nuclear and plastid regions. Molecular Phylogenetics and Evolution 139: 106542.. Within subtribe Goodyerinae, the Goodyera clade with the Goodyera and Microchilus subclades, and the Cheirostylis Blume clade with the Cheirostylis and Ludisia A.Rich. subclades were recovered. In all analyses, G. macrophylla was part of the Goodyera subclade, supporting its attribution to the genus Goodyera. It did not group with Asian or Eurasian taxa, but with two different American taxa.

For the ITS region, the BEAST analyses yielded trees with a mean log-likelihood of -5879.59 with a standard deviation of 8.95, and lower 95% HPD of -5897.17, upper 95% was HPD -5862.56. In this analysis, Goodyera macrophylla was sister to the Central American G. striata/G. brachyceras (A.Rich. & Galeotti) Garay & G.A.Romero with a PP of 1.00 (Fig. 1). These three taxa formed a clade together with the North American G. oblongifolia Raf. (PP 1.00). The ITS sequences of all four accessions of G. macrophylla were identical.

The plastid DNA analyses resulted in trees with a mean log-likelihood of -12766.71 with a standard deviation of 7.82, and lower 95% HPD of -12781.53, upper 95% was HPD -12750.91. Here, G. macrophylla grouped together with G. oblongifolia (PP 1.00; Fig. 2). Uncorrected pairwise distances among accessions of G. macrophylla varied between 0.0002 and 0.0020.

The position of the conflicting lineage including Goodyera striata and G. brachyceras in different clades was highly supported by PP in the ITS and plastid trees. We still combined the data to follow a total evidence approach. This analysis of combined data using BEAST with the same settings as for the single data sets resulted in mean log-likelihood of -18918.65 with a standard deviation of 10.56, lower 95% HPD of -18938.95 an upper 95% was HPD -18887.11. Here, G. macrophylla was also sister to G. oblongifolia, largely corroborating the results of the plastid DNA analyses (Fig. 3).

Divergence time estimation

In the ITS data set the mrca of Goodyera-Erythrodes-Kreodanthus was dated to 16.75 my (mean), 16.46 my (median), 12.04-21.84 my (95% HPD), the split between Pristiglottis and Goodyera to 19.25 my (mean), 18.71 my (median), 15.13-25.02 my (95% HPD), the mrca of G. macrophylla and G. oblongifolia to 6.09 my (mean), 5.88 my (median), 2.93-9.89 my (95% HPD) and the split between G. macrophylla and G. striata/G. brachyceras to 4.33 my (mean), 4.13 my (median), 1.61-7.33 my (95% HPD). The crown node of G. macrophylla is 0.7 my (mean), 0.58 my (median), 0.04-1.7 my (95% HPD) (Fig. 1).

The divergence time calculations yielded the following ages for the plastid DNA: the mrca of Goodyera-Erythrodes-Kreodanthus 16.54 my (mean), 16.05 my (median), 12.17-21.79 my (95% HPD), the mrca of Pristiglottis-Goodyera 19.24 my (mean), 18.55 my (median), 15.13-25.04 my (95% HPD), the split between G. macrophylla and G. oblongifolia 6.65 my (mean), 6.44 my (median), 2.80-10.73 my (95% HPD), the crown node of G. macrophylla 0.86 my (mean), 0.79 my (median), 0.18-1.85 my (95% HPD) (Fig. 2).

The results of the combined data set yielded the following ages: for Goodyera macrophylla and G. oblongifolia 7.46 my (mean), 7.22 my (median), 3.26-12.36 my (95% HPD), the crown node of G. macrophylla 0.56 my (mean), 0.79 my (median), 0.11-1.16 my (95% HPD) (Fig. 3).

Biogeographic analyses with ITS data

According to the AICc criterion, DEC + J was the most likely model for all tested data sets. To infer the origin of G. macrophylla, the ancestral areas for the node of G. macrophylla and G. striata/G. brachyceras were reconstructed. DEC + J analyses yielded Madeira with a probability of 0.5 as ancestral area and Central America with 0.5 when coding distinguished between Central and North America (Fig. 4). When only coding America, this region received a probability of 0.98 for this node. BBM analyses resulted in Central America (0.44), Madeira (0.35), North America (0.11) as ancestral areas. Alternatively, when coding only America, this continent (America, 0.93) and Madeira-America (0.07) were revealed as ancestral areas (Fig. 5).

Biogeographic analyses with the plastid data

For the node of Goodyera macrophylla and G. oblongifolia, North America had a probability of 0.54 and Madeira of 0.35 as ancestral area using DEC + J analyses. When coding America, America and Madeira had slightly higher values of 0.64 and 0.36, respectively (Fig. 6). BBM analyses yielded east/south East Asia (0.40), North America (0.29) and Madeira (0.21) for this node (Fig. 7).

Phylogenetic relationships

The position of Goodyera macrophylla varied in phylogenetic analyses based on ITS and cpDNA data. While the Madeiran endemic is sister to Central American G. brachyceras and G. striata with North American G. oblongifolia being sister to this clade in the ITS analysis, it appears as sister group to G. oblongifolia in the cpDNA trees. These phylogenetic relationships are supported by high PP values. Incongruence between the plastid and nuclear datasets was already observed by Hu & al. (2016)Hu C., Tian H., Li H., Hu A., Xing F., Bhattacharjee A., Hsu T., Kumar P. & Chung S. 2016. Phylogenetic analysis of a ‘Jewel Orchid’ Genus Goodyera (Orchidaceae) based on DNA Sequence Data from Nuclear and Plastid Regions. PloS one 11: e0150366.. Possible explanations for this include (ancient) events of hybridisation, chloroplast capture or concerted evolution. One of these mechanisms may also account for the case of G. macrophylla, but further evidence like chromosome numbers or ploidy levels is not available. A discussion about the phylogenetic pattern of Goodyera is outside of the scope of this study which focuses on G. macrophylla. Despite the incongruence, G. macrophylla always appears as sister to an American group. Goodyera macrophylla shares rather oblong leaves with G. striata and G. brachyceras which are sometimes regarded as synonyms (Garay & Romero-González 1998Garay L.A. & Romero-González G.A. 1998. Schedulae Orchidum. Harvard Papers in Botany 3: 53-62.; POWO 2019POWO 2019. Plants of the World Online. Website: http://www.plantsoftheworldonline.org/. [accessed: 23 Sep. 2020]. ), and G. oblongifolia. Based on the present evidence, we exclude a close relationship of the Madeiran species to Asian or central European taxa, in special to G. repens and hypothesise that it is part of a Central or North American clade.

The species-rich genus Goodyera is far from being sufficiently represented in phylogenetic analyses (Pace 2020Pace M.C. 2020. A recircumscription of Goodyera (Orchidaceae), including the description of Paorchis gen. nov., and resurrection of Cionisaccus, Eucosia, and Salacistis. Brittonia 72: 257-267.). Although, all North American species are included in this study, several of the Central American species have not been sequenced for phylogenetic analysis yet and it appears relatively likely that they may form at least partly geographically defined clades. Therefore, we cannot rule out that G. macrophylla is closest related to other Meso-American Goodyera species which were not included in our analysis. Among such taxa accepted by POWO (2019)POWO 2019. Plants of the World Online. Website: http://www.plantsoftheworldonline.org/. [accessed: 23 Sep. 2020]. are G. bradeorum Schltr., G. corniculata (Rchb.f.) Ackerman, G. dolabripetala (Ames) Schltr., G. erosa (Ames & C.Schweinf.) Ames, F.T.Hubb. & C.Schweinf., G. fimbrilabia Ormerod, G. hispaniolae Dod, G. major Ames & Correll, G. micrantha Schltr., G. modesta Schltr., G. ovatilabia Schltr., G. polyphylla Ormerod, G. purpusii Ormerod, G. turialbae Schltr., G. venusta Schltr., G. zacuapanensis Ormerod. This supports Pace’s (2020)Pace M.C. 2020. A recircumscription of Goodyera (Orchidaceae), including the description of Paorchis gen. nov., and resurrection of Cionisaccus, Eucosia, and Salacistis. Brittonia 72: 257-267. view that more phylogenetic work on Goodyera is needed.

Geographic origin

A close putative phylogenetic relationship of Goodyera macrophylla to American taxa prompts the question on the origin of this Macaronesian laurel forest nemoral element. Most biogeographic analyses yielded North and or Central America as the source area for G. macrophylla. Given the fact that Madeira is of volcanic origin (Geldmacher & al. 2006Geldmacher J., Hoernle K., Klügel A., Wombacher F. & Berning B. 2006. Origin and geochemical evolution of the Madeira-Tore Rise (eastern North Atlantic). Journal of Geophysical Research 111: B09206.) which requires biotic dispersal from outside of the island at a certain point, colonisation from America to Madeira seems plausible in this case. The small, light seeds of Orchidaceae are easily dispersed by wind. Another orchid illustrating the dispersal ability from North America to Atlantic islands is Spiranthes romanzoffiana Cham. which is found in large parts of North America and in Ireland and Great Britain. Dueck & al. (2014)Dueck L.A., Aygoren D. & Cameron K.M. 2014. A molecular framework for understanding the phylogeny of Spiranthes (Orchidaceae), a cosmopolitan genus with a North American center of diversity. American Journal of Botany 101: 1551-1571. discussed wind transport and exozoochory by birds as possible vectors explaining this disjunction. References to biogeographical links from America to Macaronesia are rather scarce. Kondraskov & al. (2015)Kondraskov P., Schütz N., Schüßler C., de Sequeira M.M., Santos-Guerra A., Caujapé-Castells J., Jaén-Molina R., Marrero-Rodríguez Á., Koch M.A., Linder P., Kovar-Eder J. & Thiv M. 2015. Biogeography of Mediterranean Hotspot Biodiversity: Re-Evaluating the ‘Tertiary Relict’ Hypothesis of Macaronesian Laurel Forests. PloS one 10: p.e0132091. found only 6% of laurel forest elements originating from America. Accordingly, G. macrophylla is in line with other taxa showing Macaronesian-American links like Persea indica, Arbutus canariensis Veill. ex Duhamel (Hileman & al. 2001Hileman L.C., Vasey M.C. & Parker V.T. 2001. Phylogeny and biogeography of the Arbutoideae (Ericaceae): Implications for the Madrean-Tethyan hypothesis. Systematic Botany 26: 131-143.; Kondraskov & al. 2015Kondraskov P., Schütz N., Schüßler C., de Sequeira M.M., Santos-Guerra A., Caujapé-Castells J., Jaén-Molina R., Marrero-Rodríguez Á., Koch M.A., Linder P., Kovar-Eder J. & Thiv M. 2015. Biogeography of Mediterranean Hotspot Biodiversity: Re-Evaluating the ‘Tertiary Relict’ Hypothesis of Macaronesian Laurel Forests. PloS one 10: p.e0132091.), Clethra arborea Aiton (Fior & al. 2003Fior S., Karis P.O. & Anderberg A.A. 2003. Phylogeny, taxonomy, and systematic position of Clethra (Clethraceae, Ericales) with notes on biogeography: evidence from plastid and nuclear DNA sequences. International Journal of Plant Sciences 164: 997-1006.) and possibly Pericallis D. Don. (Panero & al. 1999Panero J.L., Francisco-Ortega J., Jansen R.K. & Santos-Guerra A. 1999. Molecular evidence for multiple origins of woodiness and a New World biogeographic connection of the Macaronesian Island endemic Pericallis (Asteraceae: Senecioneae). Proceedings of the National Academy of Sciences 96: 13886-13891.; but see Swenson & Manns 2003Swenson U. & Manns U. 2003. Phylogeny of Pericallis (Asteraceae): a total evidence approach reappraising the double origin of woodiness. Taxon 52: 533-548.). Among other orchids occurring on Madeira are the two endemics, Orchis mascula subsp. scopulorum (Summerh.) H.Sund. ex H.Kretzschmar, Eccarius & H.Dietr. and Dactylorhiza foliosa (Rchb.f.) Soó and the Mediterranean Neotinea maculata (Desf.) Stearn as well as Gennaria diphylla (Link) Parl. (Press & Short 1994Press J.R. & Short M. 1994. Flora of Madeira. HMSO, London.). Phylogenetic analyses indicate affinities with the European/Mediterranean flora for the first three of these four Madeiran species (the relationships of Gennaria diphylla are not resolved; Bateman et al. 2003Bateman R.M., Hollingsworth P.M., Preston J., Yi-Bo L., Pridgeon A.M. & Chase M.W. 2003. Molecular phylogenetics and evolution of Orchidinae and selected Habenariinae (Orchidaceae). Botanical Journal of the Linnean Society 142: 1-40.), stressing the exceptional biogeographic pattern of Goodyera macrophylla.

Estimated age

Our divergence time estimations yielded ca. 4 (1.3-7.2) my and 6.3 (2.45-10.63) my for the stem node of Goodyera macrophylla, respectively in the ITS and cpDNA datasets. This discrepancy is likely due to the different phylogenetic topologies which are discussed above. Still, these molecular clock calculations suggest that G. macrophylla earliest originated in the Upper Pliocene or Lower Miocene.

These dates could still be adjusted if any of the so far unsampled Goodyera species would appear as the closest relative of the Madeiran endemic. These cases could lead to a younger stem node age of G. macrophylla. Other MLF elements of presumably similar stem node age are Heberdenia excelsa (Aiton) DC., Picconia excelsa (Aiton) DC., Bystropogon L’Hér. sect. Canariense La-Serna, and the Pericallis hansenii (G.Kunkel) Sunding clade (Kondraskov & al. 2015Kondraskov P., Schütz N., Schüßler C., de Sequeira M.M., Santos-Guerra A., Caujapé-Castells J., Jaén-Molina R., Marrero-Rodríguez Á., Koch M.A., Linder P., Kovar-Eder J. & Thiv M. 2015. Biogeography of Mediterranean Hotspot Biodiversity: Re-Evaluating the ‘Tertiary Relict’ Hypothesis of Macaronesian Laurel Forests. PloS one 10: p.e0132091.; Schüßler 2020Schüßler C. 2020. No Tertiary relicts? A biogeographical study on the Macaronesian laurel forest species in Daucus (Apiaceae), Geranium (Geraniaceae), Gesnouinia (Urticaceae), Phyllis (Rubiaceae), Semele (Asparagaceae) and Visnea (Pentaphylacaceae). Ph.D. dissertation, University of Heidelberg, Heidelberg.). Still, the majority of investigated MLF plants is younger than Goodyera. Using a different interpretation of divergence times, e.g., favouring crown nodes ages as indicator for colonisation times (García-Verdugo & al. 2019García-Verdugo C., Caujapé-Castells J. & Sanmartín I. 2019. Colonization time on island settings: lessons from the Hawaiian and Canary Island floras. Botanical Journal of the Linnean Society 191: 155-163.), a much younger colonisation in the Pleistocene indicated by the crown node of G. macrophylla (0.04-1.85 my) would indeed be possible. In the end, stem and crown node ages represent the temporal span in which island colonisation happened. In our case this includes the Upper Miocene, Pliocene and Pleistocene. Fossils of the São Jorge flora on Madeira (Góis-Marques & al. 2017Góis-Marques C.A., MadeiraJ. & de Sequeira M.M. 2017. Inventory and review of the Mio-Pleistocene São Jorge flora (Madeira Island, Portugal): palaeoecological and biogeographical implications. Journal of Systematic Palaeontology 16: 159-177.) show that suitable conditions for laurel forests, in special the stink-laurel temperate forest association sensu Capelo & al. (2005)Capelo J., de Sequeira M.M., Jardim R., Mesquita S. & Costa J.C. 2005. The vegetation of Madeira Island (Portugal). A brief overview and excursion guide. Quercetea 7: 95-122. existed at least since 1.8 my ago. The variation of infraspecific cpDNA is in line with crown node ages, suggesting that G. macrophylla diversified on Madeira since the Pleistocene. This is also of interest for conservation purposes. Still, the level of clonality within populations is still unknown and under current investigation (Gouveia & al. in prep.).

Considering its sister group relationship to American taxa, it seems obvious that the Madeiran species does not fulfil the criteria for the relict hypothesis of Macaronesian laurel forests (Kondraskov & al. 2015Kondraskov P., Schütz N., Schüßler C., de Sequeira M.M., Santos-Guerra A., Caujapé-Castells J., Jaén-Molina R., Marrero-Rodríguez Á., Koch M.A., Linder P., Kovar-Eder J. & Thiv M. 2015. Biogeography of Mediterranean Hotspot Biodiversity: Re-Evaluating the ‘Tertiary Relict’ Hypothesis of Macaronesian Laurel Forests. PloS one 10: p.e0132091.). Originally, this hypothesis relates to Central European and Mediterranean laurel forests, which went extinct at the latest by the end of the Pliocene (Kondraskov & al. 2015Kondraskov P., Schütz N., Schüßler C., de Sequeira M.M., Santos-Guerra A., Caujapé-Castells J., Jaén-Molina R., Marrero-Rodríguez Á., Koch M.A., Linder P., Kovar-Eder J. & Thiv M. 2015. Biogeography of Mediterranean Hotspot Biodiversity: Re-Evaluating the ‘Tertiary Relict’ Hypothesis of Macaronesian Laurel Forests. PloS one 10: p.e0132091.). A sister group relationship to American taxa contradicts this assumption. Theoretically, the relict hypothesis could only be upheld, if extinction of more closely related continental European Goodyera taxa is assumed. However, since there is no evidence for this, we refrain from such explanation. Frey & Pickering’s (1975)Frey G. & Pickering C. 1975. Contribution to the knowledge of the orchids of Madeira and the Azores. Bocagiana 38: 1-20. hypothesis that G. macrophylla is a relict species of a former Atlantic vegetation, is not supported by our data. Under this scenario, closest relatives of G. macrophylla should still be present in laurel forests of the Canary Islands or the Azores, which is not the case. Only the assumption of extinction events in the Atlantic archipelagos could then uphold this hypothesis. This is theoretically possible, but in our view rather unlikely. Based on the available data, we hypothesise that G. macrophylla colonised the Madeiran laurel forest earliest in the lower Miocene/upper Pliocene from America.