Hylocereus (A. Berger) Britton & Rose, 1909

Korotkova, Nadja, Borsch, Thomas & Arias, Salvador, 2017, A phylogenetic framework for the Hylocereeae (Cactaceae) and implications for the circumscription of the genera, Phytotaxa 327 (1), pp. 1-46 : 10-11

The Hylocereus / Selenicereus clade

Hylocereus is confirmed as monophyletic while Selenicereus forms a grade in which Hylocereus is nested ( Fig. 2B ).

Within Hylocereus , there are two highly supported clades that largely correspond to the two current sections, but neither of them is monophyletic. The clade of Hylocereus sect. Hylocereus ( Fig. 2B ) contains the type species H. triangularis Britton & Rose and corresponds to Hylocereus in its traditional circumscription, e.g. sensu Britton & Rose (1920) . This clade includes the earlier-proposed monotypic genus Wilmattea Britton & Rose [ W. minutiflora (Britton & Rose) Britton & Rose ], that was originally separated from Hylocereus due to its smaller flowers and the lack of a floral tube.

The second clade (Salmdyckia clade, Fig. 2B ) contains most of Hylocereus sect. Salmdyckia D.R. Hunt. It was transferred from Selenicereus to Hylocereus only recently by Bauer (2003a). This section was based on the genus Mediocactus Britton & Rose that was characterized by having characters of both Hylocereus and Selenicereus [see Hunt (1989) for a detailed classification history and notes on typification]. Buxbaum (1962) regarded Mediocactus to be close to Hylocereus , while Hunt (1989) suggested including it in Selenicereus and this was followed in subsequent treatments (e.g., Anderson 2001, Barthlott & Hunt 1993). The transfer of Selenicereus sect. Salmdyckia D.R. Hunt to Hylocereus was based on unpublished DNA sequence data of Robert Wallace presented at an IOS congress in 1996 and was adopted by Bauer (2003a), Taylor & Zappi (2004), and Hunt (2006). Most species comprising the Salmdyckia clade are native to South America, except H. ocamponis , which occurs in Mexico.

The Selenicereus grade is formed by three lineages and several accessions of S. inermis (Otto) Britton & Rose that do not form a clade and remain unresolved ( Fig. 2B ). They are also not found closely related to the Salmdyckia clade, contrary to the classification according to Bauer (2003a). The monotypic genus Cryptocereus Alexander [ C. anthonyanus Alexander ≡ Selenicereus anthonyanus (Alexander) D.R. Hunt ] is resolved within Selenicereus . Backeberg (1959) additionally included Werckleocereus imitans Kimnach & Hutchison in Cryptocereus and suggested a close relationship to phyllocactoid genera, i.e. the Epiphyllum alliance. These species differ from typical members of Selenicereus having flat-stemmed epiphytes with stems resembling fernleaves ( Fig. 3 ). Our data confirm the proposal by Hunt (1989) who pointed out a possible affinity with Werckleocereus Britton & Rose , as well as its inclusion in Selenicereus because it shares the spiny pericarpel.

The genera Hylocereus and Selenicereus are morphologically similar as both contain scrambling or climbing cacti with large, usually white nocturnal flowers. Britton & Rose (1920) originally defined Hylocereus to possess pericarpels covered by scales while the pericarpels in Selenicereus have hairs or spines. The transfer of Selenicereus sect. Salmdyckia to Hylocereus ( Bauer 2003a) included species with spines on the pericarpel in Hylocereus so that the original concept of Britton & Rose did not apply anymore. Bauer (2003a) had justified this transfer only by unpublished sequence data of Wallace, referred to in Taylor and Zappi (2004) and did not mention any morphological characters to substantiate his new concept of Hylocereus .

Our results confirm the finding of earlier molecular phylogenetic studies, using either plastid or nuclear DNA sequences. All of them have shown that there is no separation between Hylocereus and Selenicereus , although only few species were sampled ( Arias et al. 2005, Bárcenas et al. 2011, Cruz et al. 2016, Hernández-Hernández et al. 2011, Plume et al. 2013). The most extensive dataset generated so far (plastid rpl32-trnL, trnQ -rps16 and psbD -trnT and nuclear ITS) also found Hylocereus nested in a paraphyletic Selenicereus ( Plume et al. 2013) . The same conclusion was reached using a morphological and anatomical comparison of Hylocereus and Selenicereus ( Gómez-Hinostrosa et al. 2014) . Therefore, we merge both genera under the name Selenicereus and provide the necessary new combinations (see Taxonomic synopsis). A next level of study is required in Selenicereus concerning the delimitation of several species which remain insufficiently known, e.g. Selenicereus ocamponis / S. purpusii , S. triangularis / S. trigonus , S. vagans / S. murrillii , and S. costaricensis / S.monacanthus . These taxa require insights from extensive field studies and a combination of molecular, morphological and ecological data.

Figures

FIGURE 2A. Majority-rule consensus tree from Bayesian inference.Values above branches are Posterior Probabilities (left) and Bootstrap support values from Maximum Likelihood (10.000 replicates) calculated with RAxML GUI using the “ML + thorough bootstrap” option, values below branches are Jackknife support values (10.000) calculated in PAUP. (-- denotes ML BS support below 50, n.f.: node not found). Taxon names in the cladogram reflect hitherto accepted names (Bauer 2003a, Hunt 2006), with the exception of new names in Disocactus published by Cruz et al. (2016). The new/updated generic concepts presented here are annotated with brackets.

FIGURE 3. Convergent fern-like flattened stems in A) Epiphyllum chrysocardium, B) Weberocereus imitans, C) Epiphyllum anguliger, D) Selenicereus (Cryptocereus) anthonyanus (photos by W. Barthlott, Lotus-Salvinia.de).