Sipaneopsis (Sipaneeae, Ixoroideae): A unique flowering process in the family Rubiaceae and its taxonomic significance

The process of anthesis of Sipaneopsis is unique within the Rubiaceae, and is here described and illustrated in detail for the first time. During the initial stage of anthesis the flowers are small, with the corolla lobes already open and erect before tube elongation, and the corolla lobe basal appendages are appressed against each other, forming a convex structure at the corolla mouth obstructing the entrance of visitors and pollinators. At the final stage of anthesis, the corolla lobes reflex and become perpendicular to the tube, and their basal appendages become erect, not touching each other, allowing visitors and pollinators to access the corolla tube and the nectar disk. Neobertiera and Sipaneopsis are unique within the tribe Sipaneeae in having indehiscent dry fruits and recent molecular phylogenies positioned them as sister taxa. Sipaneopsis is distinguished from Neobertiera in having flowers consistently homostylous (vs. distylous or rarely tristylous in Neobertiera ), stamens inserted at the middle of the corolla tube (vs. at variable positions), and five triangular appendages at the base of each corolla lobe (vs. corolla lobes thickened at base, without appendages). The differences in flower morphology and process of anthesis between the two genera (which most likely influence their different pollination syndromes) provide an important set of significant taxonomic and diagnostic characters that can be used to distinguish them.


Introduction
Members of the Sipaneeae have either been included in the tribe Rondeletieae (Verdcourt 1958;Steyermark 1974;Delprete 1999) or treated as a separate tribe (Bremekamp 1934, Robbrecht 1988 and positioned in subfamily Cinchonoideae, based on morphological characters. The molecular phylogenies of Rova et al. (2002) supported the Sipaneeae as a separate tribe, and positioned it within the subfamily Ixoroideae instead. Delprete & Cortés-B. (2004), produced the first phylogenetic study focused on the Sipaneeae, using trnL-F and ITS sequence data in which it was shown that the tribe and several of its genera are monophyletic. While continuing a taxonomic revision of the tribe, new observations were made on two genera unique within the tribe by having indehiscent, dry fruits, namely Neobertiera Wernham (1917: 169) and Sipaneopsis Steyermark (1967: 284), which, in the molecular phylogenies of Delprete & Cortés-B. (2004), were positioned as sister taxa. Sipaneopsis has been traditionally distinguished from Neobertiera by stamens inserted at about the middle of the corolla tube (vs. at variable positions in Neobertiera), five triangular appendages at the base of the corolla lobes (vs. corolla lobes thickened at the base, without appendages), and corolla lobes either imbricate or left-contorted in bud (vs. always left-contorted). However, the taxonomic significance of these appendages has previously been questioned. In addition, the two genera occupy different ecological niches, and have different geographical distributions. Sipaneopsis, as currently delimited (Kirkbride 1980;Delprete & Steyermark 2004;Delprete, 2015), is a genus of seven species endemic to white sand savannas, growing in sunny exposures (heliophilous), occurring mostly in the lowlands (although with a few collections from higher altitudes) of the Guiana Shield, ranging from Colombia, southern Venezuela to northern Brazil, North of the Amazon River. A few specimens collected in Guyana (Henkel et al. 1449, McDowell & Stobey 3853, and Mutchnick et al. 156) were incorrectly identified as Sipaneopsis cururuensis Kirkbride (1980: 113, figs. 19, 24); these specimens were recently re-identified as Neobertiera pakaraimensis Delprete (2015: 126, figs. 2, 5); therefore, S. cururuensis is the only species of the genus known to occur south of the Amazon River, as it is only known by the type collection from the Upper Tapajos River Basin (Cururu River, Erereri [River, ca. 7°37'S, 57°37'W], based on Egler & Raimundo 1265 [IAN, MG, US]). On the other hand, Neobertiera, as currently circumscribed (Delprete, 2015), is a genus of four species growing on lateritic or alluvial soils, in forest undercanopy (umbrophilous), often close to slowly running water courses, endemic to the Guianas, with one in Guyana, one in Guyana and Suriname, and two species apparently restricted to French Guiana.
As a result of a recent morphological analysis of the four species currently recognized in Neobertiera, Delprete (2015) discovered that all of them are distylous (rarely tristylous), while the flowers of Sipaneopsis are consistently homostylous. Field observations and a detail analysis of herbarium specimens revealed additional morphological characters of both taxa, and a newly reported process of anthesis in Sipaneopsis, which represent a set of significant characters that can be used to distinguish the two taxa.

Material and Methods
This study was based on examinations of individuals in their natural environment, herbarium specimens, samples preserved in 60% ethanol, and digital images. Field observations of Sipaneopsis maguirei Steyermark (1967: 287) were made in populations on white-sand savannas in southern Venezuela, State of Amazonas, near Minicia Nuevo village (P.G. Delprete et al. 7451,7452,7453). Field observations of Neobertiera glomerata Delprete (2015: 124) were made in populations of two localities in French Guiana, one in the forests of the coastal region, along the trail to Bagne des Annamites (P.G. Delprete & G. Quenette 11860) and the other in the forests of the Upper Oyapock River Basin, near Tampak village (P.G. Delprete et al. 12310). Herbarium specimens were studied either by visiting and/or through loans from BBS, BM, CAY, F, IAN, K, MG, MO, NY, P, U, UB, and US herbaria, and were examined using a dissecting microscope. Details about collection locality, collectors, and herbaria of deposit of specimens of Neobertiera and Sipaneopsis analyzed for this study are available in Appendix 1.
After an exhaustive search of specialized literature (see references cited above) it was concluded that the process of anthesis of Sipaneopsis is unique within the Rubiaceae, and is here described and illustrated (Figures 1−3 (Figs. 3G, H), and therefore allowing visitors and pollinators to access the corolla tube and nectar disk. In addition, flowers of Sipaneopsis are protandrous, which means that after the lobes have become perpendicular to the corolla tube, on the first day the anthers reach full maturity and dehisce, releasing pollen, while the stigmatic branches are still appressed to each other and not receptive (Fig. 3G); on the following day the anthers are no longer releasing pollen, the stigmatic branches spread apart and are reflexed, and the stigmatic surfaces are exposed and become receptive (not illustrated).

Comparison of flower morphology and process of anthesis in Neobertiera and Sipaneopsis and their taxonomic significance
Neobertiera and Sipaneopsis are unique within tribe Sipaneeae in having indehiscent, dry fruits, while the remaining genera of the tribe have septicidal or loculicidal capsules. In the molecular phylogenies of Delprete & Cortés-B. (2004) these genera were found to be sister taxa on a strongly supported clade. However, these molecular phylogenies only included a small portion of Sipaneeae taxa, and these two genera were represented by only one species each. Therefore, more species need to be added to the analysis to test the monophyly of all the genera in the tribe and to further clarify their evolutionary relationships. In addition, all the species of Neobertiera are protandrous, meaning that on the first day of anthesis the anthers open and release pollen, and the stigmatic branches are appressed to each other and not  receptive; on the second day the anthers are no longer releasing pollen, while the stigmatic branches spread apart, are reflexed, and become receptive (pers. obs.). Additional features that can be useful to differentiate Neobertiera from Sipaneopsis are that in the former the corolla lobes are opened and reflexed only after the corolla tube is completely elongate (as in the rest of the family), while in the latter the lobes are already open (although not fully expanded) before corolla tube elongation, and its peculiar flowering process, which is described above. The differences in flower morphology and process of anthesis between the two genera (which most likely influence their different pollination syndromes) provide a significant set of taxonomic and diagnostic characters that can be used to distinguish them.
The taxonomic significance of the corolla lobe basal appendages of Sipaneopsis has sometimes been questioned (e.g., Delprete & Cortés-B. 2004), and became even more dubious when this genus was retrieved as sister taxon with Neobertiera in the molecular phylogenies of Delprete & Cortés-B. (2004). However, the corolla lobe appendages of Sipaneopsis are here confirmed to represent an important taxonomic and diagnostic character, when associated with the unique process of anthesis of this genus. Observations of pickled specimens using the dissecting microscope revealed that the lobe appendages consist of a triangular structure of the same nature as the corolla lobes (i.e., parenchymatic tissue), covered by dense tufts of thick hairs, forming tough triangular units. The evolutionary advantage of the peculiar flower morphology and process of anthesis of Sipaneopsis remains poorly understood. It was observed only in populations of S. maguirei in southern Venezuela, and no observations focusing on potential pollinators were made, therefore, its pollination syndrome remains to be studied. Considering that Sipaneopsis is a genus restricted to white-sand savannas located in remote localities of the Amazon basin, this kind of studies are difficult to do. Nevertheless, future trips to these localities are planned, and the unique pollination syndrome of this genus will hopefully be comprehended.

Acknowledgments
I would like to thank Gwenaël Quenette and Juliana Costa da Silva, who helped during field observations, producing digital images, and processing herbarium and pickled specimens of Neobertiera from a population growing along the trail to the Bagne des Annamites, French Guiana. Funds for field work in French Guiana, during which I had the occasion to study an additional population of Neobertiera in the surroundings of the Camopi village, upper Oyapock River basin, were provided by the LABEX CEBA (Laboratory of Excellence-Center for the Study of Biodiversity in Amazonia; grant managed by the Agence Nationale de la Recherche, ANR-10-LABX-0025); during this expedition, I was accompanied by Guillaume Odonne (CNRS), Damien Davy (CNRS), and Sophie Gonzalez (CAY), who are here acknowledged for help during field work. I also wish to thank Ana Claudia Araujo (K) and Luis Alvárez (TFAV), who helped during field observations and processing herbarium and pickled specimens of Sipaneopsis maguirei from a population growing in white-sand savannas in the municipality of San Fernando de Atabapo, State of Amazonas, Venezuela. I am also grateful to the directors and curators of BM, BRG, CAY, COL, F, IAN, L, K, MG, MO, NY, P, RB, U, UB, US and VEN herbaria, for loans of material and/or for providing working space during my visits. Finally, I wish to express my gratitude to the anonymous reviewers and the Associate Editor for their useful comments and valuable corrections.