Bopopia, a new monotypic genus of Gesneriaceae (Gesnerioideae, Coronanthereae) from New Caledonia

. A new genus of Gesneriaceae, Bopopia Munzinger & J.R.Morel gen. nov., is described from New Caledonia. The genus is based on B. parviflora Munzinger & J.R.Morel gen. et sp. nov., a new species collected during an expedition on Mt Katalupaik, in the North Province of New Caledonia’s main island. Originally considered as a species of Coronanthera , our phylogenetic analysis – including 19 species within Coronanthereae and two individuals of B. parviflora gen. et sp. nov., and using three molecular markers (nuclear rDNA ITS, and chloroplast regions trnL-trnF and trnE-trnT ) – showed that the new species is not close to Coronanthera in subtribe Coronantherinae, but belongs to subtribe Negriinae where it is sister to Depanthus . From that genus Bopopia gen. nov. differs in floral symmetry (zygomorphic vs actinomorphic) and the number of stamens (4 vs 5). From the other genera of Negriinae the new genus differs in the white corolla and its indeterminate thyrse with 3 to 5 levels of branching. The morphological circumscription of the subtribe Negriinae is amended to include Bopopia gen. nov. Two keys are provided, one to the subtribes in the tribe Coronanthereae, and one to the genera in subtribe Negriinae. Following the IUCN Red List categories and criteria, the conservation status of B. parviflora gen. et sp. nov. is provisionally assessed as Endangered (EN).


Introduction
New Caledonia is a biodiversity hotspot with an estimated autochthonous flora of 3409 species of flowering plants, of which nearly 75% are endemic (Morat et al. 2012;Munzinger et al. 2020). Nevertheless, this biodiversity is still poorly known: on average, one new species is described each month from the New Caledonian archipelago (Gâteblé et al. 2018). To complete the knowledge of this biodiversity, the research program ʻLa Planète Revisitéeʼ (http://www.laplaneterevisitee.org/fr) organized several expeditions between 2016 and 2018 in various unexplored and difficult-to-access areas of New Caledonia. In 2017, the southern face of the isolated Mount Katalupaik, a ʻno data areaʼ of the archipelago (see map in Birnbaum et al. 2015), was inventoried from 300-900 m altitude. During this expedition, 340 species were collected, 13 being supposedly new to science, nine could be assigned to specimens already present in herbaria, while four were collected for the first time ).
One of these potentially new species was a relatively abundant shrub in this locality and is represented by the specimens Bruy et al. 1139 and Munzinger et al. 7980. The plants were flowering at the time of the collection in October. Several characters suggested placement in the family Gesneriaceae Rich. & Juss. ex DC., including shrubby habit, opposite simple petiolate leaves without stipules, pair-flowered cymes, pentamerous zygomorphic flowers, gamosepalous calyx, gamopetalous corolla, four stamens with connate anthers, presence of hypogynous disk, syncarpous unilocular gynoecium of two carpels, one style, and two stigmas. The new species was initially considered to belong to the genus Coronanthera Vieill. ex C.B.Clarke (Gesnerioideae-Coronanthereae-Coronantherinae; Woo et al. 2011;Weber et al. 2013), because of its shrubby habit, slightly serrate leaves with similar shape and dimensions (blade and petiole), small zygomorphic corolla with four connate stamens, and annular nectary adnate to the ovary.
In order to test the identity of the New Caledonian specimens, and the generic and subtribal position of the species, a molecular phylogenetic study was performed. The combination of molecular and morphological data, consistent with the geographical distribution, confirms that this is a new species and genus. This methodology has been shown to be particularly efficient in the circumscription of genera and species, including the discovery of a considerable number of new genera in the Gesneriaceae (e.g., Araújo et al. 2010;Clark et al. 2010;Wei et al. 2010;Middleton & Möller 2012).

Sampling
Specimens of this new species were collected in the North Province of New Caledonia, on Mt Katalupaik, in October 2017 by David Bruy, Jérôme Munzinger and Marc Pignal. The plants were cut with clippers poles and pressed in the field, then dried at the base camp, notes and photos were taken in the field, and the flowers were preserved in alcohol. Vouchers were deposited at MPU, NOU and P (Appendix 1); acronyms of herbaria follow Index Herbariorum (Thiers continuously updated).

Molecular phylogenetic analysis
Leaves from the two collections of the new species were dried and preserved in silica gel. DNA was extracted using the mixed alkyltrimethylammonium bromide (MATAB) protocol as described in Scarcelli et al. (2006). DNA quality and quantity were then checked by optical density with the Nanodrop™ 8000 (Thermo Fisher Scientific, Waltham, Massachusetts, United States) and by agarose gel electrophoresis.
The selection of gene regions and analysis methods follows Woo et al. (2011) in order to get results that are directly comparable to their study.
In order to determine the taxonomic coverage to be included in our phylogenetic analysis, we evaluated the taxonomic proximity of our DNA sequences using blastn suite (NCBI) and relying on the query coverage and the percentage of identity. For  All these taxa are included in the tribe Coronanthereae Fritsch. Thus, the sequence matrix of Woo et al. (2011), covering the Southeast Pacific Gesneriaceae, was reduced to 35 samples corresponding to an exhaustive sampling of Coronanthereae genera. Sequences produced subsequently to Woo's work were also included in the matrix, from Coronanthera (Serrano-Serrano et al. 2017) and from Negria F.Muell (Perret et al. 2013). All GenBank sequence accessions and their corresponding vouchers can be found in Appendix 1.
Two members of the tribe Beslerieae Bartl. were selected as outgroups based on phylogenetic relationships shown in Woo et al. (2011): Gasteranthus atratus Wiehler from Ecuador and Cremosperma castroanum C.V.Morton from Peru. The 39 sequences were first aligned using MAFFT under default parameters (Katoh et al. 2017) and corrected manually to minimize substitutions and indels by eventbased criteria in order to constitute homology hypotheses (Morrison 2006). Then, we used GBlocks 0.91b (Castresana 2000) on each ITS, trnL-trnF and trnE-trnT alignments. GBlocks deletes segments of contiguous non-conserved positions, gap positions and non-conserved flanking positions. It has consequently proven very useful in increasing phylogenetic accuracy (e.g., Roalson & Roberts 2016). For ITS alignment, we used the default stringent parameters, disallowing smaller final blocks, gap positions and non-conserved flanking positions. Less stringent parameters were used (allowing gap position within the final blocks and less strict flanking positions) for trnL-trnF and trnE-trnT. The characteristics of the different DNA sequences are given in Table 1.
We analyzed the cpDNA and ITS independently (not shown) in order to assess whether or not the data could be concatenated. We used the model GTR + I + Γ for maximum likelihood (ML) analysis and the following models for Bayesian inference (BI) analysis: SYM + Γ for ITS, GTR+ I for trnL-trnF and GTR + I + Γ for trnE-trnT. The partition-homogeneity test (Farris et al. 1994) as implemented in PAUP* ver. 4.0a (Swofford 2002) was performed with 10 000 bootstrap replicates and did not find significant differences between any partitions (p = 0.8487). Therefore, a combined analysis of the DNA regions could be performed.
The research of the optimal substitution model was performed with PartitionFinder2 (Lanfear et al. 2016) using the Akaike Information Criterion (AIC; Posada & Buckley 2004). ML analysis was run on the partitioned data set in RaxML ver. 8 (Stamatakis 2014) with 100 bootstrap replicates. BI analysis was run in Mrbayes ver. 3.2.7 (Ronquist et al. 2012). All the parameters values were unlinked across partitions and estimated during the Markov chain Monte Carlo (MCMC) runs. Two independent analyses, starting from different random trees, were run, each with four Markov chains (one cold chain and three incrementally heated chains). Independent analyses were conducted with 5 000 000 generations each, with a sampling frequency of 1 tree every 1000 generations. Convergence between the two independent runs was checked using Tracer ver. 1.7.1 (Suchard et al. 2018). We discarded the first 25% of the trees as burn-in, and the retained ones were summarized in a majority rules consensus tree (Fig. 1).

Specimens examined and morphology
After confirmation that the collected species was indeed a member of Gesneriaceae, we examined all collections within this family from Oceania and Asia present in MPU, NOU and P, as well as all specimens without family identification from the same areas in these herbaria. In addition, we used virtual collections: Global Plant initiative (https://plants.jstor.org/), e-ReColNat infrastructure (https://www.recolnat.org/fr/), and Z herbarium (https://www.herbarien.uzh.ch/en/belegsuche.html).
Morphological descriptions were prepared using standard terminology (Harris & Harris 2001). Vegetative parts were measured directly on herbarium specimens. Details of androecium and gynoecium were measured in flowers preserved in alcohol (part of the collection Munzinger et al. 7980). The measurements are given as follows: (minimum) first quartile-third quartile (maximum) (following Munzinger et al. 2016). For the remaining measurements with fewer observations, we chose to only give the minimum and maximum values observed. The drawings of the new species are based on photographs taken in the field and of specimen parts preserved in alcohol. We applied the IUCN Red List categories and criteria (IUCN 2019) to propose a conservation assessment of the species.

Abbreviations used in figures 2, 4 and 6
AFn = axillary flower of the axis n An = axis of the n th level ASn = axis of the n th level with accessory origin FAn = front axis of the axis n, terminating by a front flower (Fn) Fn = front flower of the axis n (Weber 1982) ISU = inflorescence sub-unit PFC = pair-flowered cyme Tn = terminal flower of the axis n αn, βn = bracts on the axis n, axillary of the two axis n+1 γ = bracteole on the axis n, axillary of the front axis (or front flower) n, = γ-bracteole sensu

Molecular phylogenetic analysis
Fig . 1 shows the 50% majority-rule consensus tree obtained with the Bayesian analysis. The maximum likelihood tree has the same topology, aside from two nodes which were not supported (< 60% bootstrap value).
Subtribe Negriinae is strongly supported with a posterior probability (PP) value of 1 and a bootstrap value (BS) of 100. The separation between subtribes Coronantherinae Fritsch and Mitrariinae Hanst. is supported with a PP value of 0.9 and a BS value of 66. Monophyly for all genera represented by two or more species is strongly supported (PP = 1; BS = 100).
The new species is supported as a member of subtribe Negriinae (PP = 1; BS = 100) and as sister to the genus Depanthus S.Moore (PP = 0.8; BS = 81).

Herbarium study
We found that this taxon had never been collected before.

Diagnosis
Bopopia gen. nov. differs from other genera of Coronanthereae in its inflorescence: an axillary indeterminate thyrse with ultimate axes being pair-flowered cymes and inferior axes being indeterminate thyrses with three to five levels of branching (vs 3-flowered cymes or solitary flowers); it differs from Depanthus in floral symmetry (zygomorphy vs actinomorphy), stamen number (4 vs 5); from Negria in inserted (vs exserted) stamens, and connate (vs free) anthers; from Lenbrassia in bilobed (vs spatulate) stigma.

Etymology
The genus is named after the land and people of Bopope (Pwöpwöp), in the vicinity of Mt Katalupaik. Clarke and Depanthus glaber in its glabrous adaxial leaf surface but differs in its unique inflorescence structure within the tribe and also vegetatively in leaf measurements, blade, apex and base shape, margin (entire vs serrate), and abaxial indumentum, as well as petiole length and indumentum (Table 2).

Etymology
The specific epithet refers to the small flowers as compared to other Gesneriaceae known from New Caledonia.

Distribution
The new species is presently only known from the North Province of New Caledonia's main island Grande Terre, on the south flank of Mt Katalupaik, around 500 m a.s.l. (Fig. 7), about 17 km as the crow flies south of Hienghène.

Habitat and ecology
The plants occur in dense humid forests from low to medium elevations on volcano-sedimentary substrates (Jaffré et al. 2012). Collecting points projected on the geological map (Gouvernement de la Nouvelle-Calédonie 2019) fall in a wide area of black siltites surrounded by basalts, dolerites, undifferentiated gabbros or fine tuffs. They are quite far from serpentine veins (ultramafic) in that area; thus the species is considered to grow on non-ultramafic substrates. Individuals of B. parviflora gen. et sp. nov. were observed to be relatively abundant in the valley that was explored (~ 100 m wide). Additional surveys are needed in the surrounding valleys.

Phenology
Plants collected in flower in October 2017. The length of the flowering time and the period of fructification are currently unknown.

Conservation status
Bopopia parviflora gen. et sp. nov. is only known from one population corresponding to a single location sensu IUCN (2019), and was estimated to contain < 250 individuals. This population is not in a protected area. The forest on the southern flank of Mt Katalupaik is highly fragmented by fire (Fig. 7), and fire appears to be a recurrent threat. The conservation status of the species is therefore preliminarily assessed as Endangered [EN: D]. This assessment has been submitted to the New Caledonian Plant Red List Authority for review and validation.

Discussion
Gesneriaceae are a family in the order Lamiales and comprise ca 150 genera and over 3400 species worldwide (Perret et al. 2013;Woo et al. 2011;Weber et al. 2013). In New Caledonia, this family is represented by the genera Cyrtandra J.R.Forst. & G.Forst (Didymocarpoideae-Trichosporeae), Coronanthera, and Depanthus (Gesnerioideae-Coronanthereae). Cyrtandra comprises more than 650 species (Atkins et al. 2013), although only a single species is found in New Caledonia, Cyrtandra mareensis Däniker, which is endemic to Maré Island (Morat et al. 2001;Cronk et al. 2005). Coronanthera and Depanthus are part of the tribe Coronanthereae that comprises nine genera in total (Woo 2007;Woo et al. 2011;Weber et al. 2013). Depanthus is a genus endemic to New Caledonia, with two described species, D. glaber and D. pubescens Guillaumin, although Woo (2007) considers these species to be synonymous in his PhD thesis (not validly published). All known species of Coronanthera are found in the archipelago of New Caledonia except Coronanthera grandis G.W.Gillett, an endemic of the Solomon Islands (Bougainville and Ysabel islands; Gillett 1967  Herbaria are a major frontier for species discovery (Bebber et al. 2010), and indeed Woo (2007) listed nine new species of Coronanthera, based on herbaria collections, in need of formal description. Several of these were collected a long time ago and are still waiting for the attention of a taxonomic specialist. An examination of all New Caledonian Gesneriaceae in major herbaria showed that the species described here as Bopopia parviflora gen. et sp. nov. was never collected before. Thus field expeditions, such as ʻLa Planète Revisitéeʼ, are still very important to complete the inventory of New Caledonia's flora.
The molecular data place the new species into tribe Coronanthereae, but surprisingly, not in Coronanthera and subtribe Coronantherinae, but in subtribe Negriinae, here being the sister taxon to the genus Depanthus (Fig. 1). It is important to note that the placement of the new species as sister to Depanthus is only weakly (PP = 0.8) to moderately (BS = 81) supported. Consequently, additional taxon sampling and / or inclusion of additional loci (particularly nuclear) is warranted. Morphologically, the cyme structure is in general agreement with Gesneriaceae, with ultimate axes ending by pair-flowered cymes (PFC) (Weber 1982(Weber , 1995 as illustrated in Fig. 4. But, in Gesneriaceae, the front flower generally has no subtending bracteole, this latest being vestigial (Weber 2004;Haston & De Craene 2007) except in Microchirita hamosa (R.Br.) Yin Z.Wang and Sinningia bulbosa (Ker Gawl.) Wiehler (Weber 1973(Weber , 2013. In Bopopia gen. nov. all front flowers are subtended by a distinct bract (γ-bracteole) (Figs 4, 6). Its inflorescence is also strongly distinguished by the fact that one (or two) inflorescence(s) are born from the axil of the γ-bracteole of the first axis (Figs 2, 6), and not a solitary flower as classically described in the Gesneriaceae and more broadly the Lamiales (Weber 2013), with the exception of Penstemon serrulatus Menzies ex Sm. (Scrophulariaceae) which, however, develops only a single-branched cyme (Weber 1973). In Bopopia parviflora gen. et sp. nov., two axes develop from the axils of bracts α and β of the first axis (Figs 2, 6) into what we call the inflorescence sub-unit (ISU) (Fig. 2). One of these axes has an accessory origin. The material available already shows that, in addition to its molecular position in Negriinae, its inflorescence is unique, at least within the subtribe, and confirms that Bopopia gen. nov. deserves a special taxonomic rank. Further studies are needed, ideally with material at more advanced developmental stages, to define the variability and the complexity of this surprising inflorescence and to consider whether other taxonomic ranks need to be amended, including potentially the delimitation of the Gesneriaceae. Appendix 1 (continued on the next page). Voucher specimens for molecular phylogenetic analysis with GenBank accession numbers indicated. ---= sequence not available for this study; ! = voucher or its scan seen. Appendix 1 (continued). Voucher specimens for molecular phylogenetic analysis with GenBank accession numbers indicated. ---= sequence not available for this study; ! = voucher or its scan seen.