Nicotiana paulineana, a new Australian species in Nicotiana section Suaveolentes

Abstract. Nicotiana is found predominantly in the Americas and Australia, but also has representatives in Africa and the Pacific Islands. All native Australian Nicotiana species belong to section Suaveolentes. The number of species in this section is uncertain and subject to revision. An example of this uncertainty is the taxonomic status of a South Australian Nicotiana accession colloquially termed ‘Corunna’. Here, we report sequences for nuclear and plastid markers for N. sp. Corunna (D.E.Symon 17088) and accessions of two other Australian species, N. burbidgeae and N. benthamiana. Phylogenetic comparison of these sequences with those of other members of Nicotiana places all three taxa in N. section Suaveolentes and shows that ‘Corunna’ represents a distinct phylogenetic lineage in a well supported clade along with N. goodspeedii, N. maritima, N. amplexicaulis and N. suaveolentes. Phenetic analysis of floral characters also supports recognition of N. sp. Corunna (D.E.Symon 17088) as a distinct species, which we describe here as Nicotiana paulineana Newbigin & P.M.Waterh., sp. nov. The enlarged molecular dataset described here contributes to a better understanding of taxonomic relationships within the section.


Introduction
The nightshade or Solanaceae family is widely distributed across all temperate and tropical continents. The family contains many of the world's most important agricultural species, such as potatoes, tomatoes, eggplants and tobacco. It also includes several important model species used in plant research, including Nicotiana benthamiana Domin and Petunia Juss. (Bombarely et al. 2012;Vandenbussche et al. 2016;Bally et al. 2018). The Solanaceae consists of~100 genera and 2800 species and relationships within the family have been the subject of repeated phylogenetic revision (D'Arcy 1979;Olmstead et al. 2008).
Nicotiana L. is one of the larger genera in the Solanaceae that includes mainly annual, non-woody plants, and includes various species commonly referred to as 'tobacco plants'. The genus Nicotiana includes 86 species in 13 sections distributed across tropical and temperate regions, with most being native to South and North America (Knapp et al. 2004). With currently~35 species, section Suaveolentes Goodsp. is a monophyletic group of ancient allopolyploid origin and the largest of the Nicotiana sections (Leitch et al. 2008;Clarkson et al. 2010). Unlike other Nicotiana sections, Suaveolentes contains no American taxa and is native to Australia (26 recognised species), the Pacific (three species) and Africa (one species). Australian Suaveolentes taxa are widespread across the continent, especially in the arid zone (Knapp et al. 2004;Chase et al. 2018a).
The number of species in section Suaveolentes is currently subject to revision (Chase et al. 2018a). Misidentification of available N. section Suaveolentes seed and herbarium material is common, with estimates ranging from 23% to at least 50% (Marks et al. 2011a;Chase et al. 2018a (Chase et al. 2018b(Chase et al. , 2018c. As well as misidentified material, there are also undescribed species in N. section Suaveolentes (Chase et al. 2018a (Chase et al. 2003;Clarkson et al. 2010).

Plant material and DNA sequencing
DNA sequences for the Nicotiana species shown in Table S2 were obtained from GenBank (www.ncbi.nlm.nih.gov/ genbank). Although N. suaveolens Lehm. and N. exigua H.-M.Wheeler are considered conspecific (Horton 1981), the sequences lodged in GenBank under these names for matK, ITS and the two GS paralogs are not identical, and thus the original species names were retained and have been treated here as independent taxa. The voucher material for these sequences should be re-examined to confirm their identities.
Twelve species of Nicotiana, including the undescribed N. sp. Corunna (D.E.Symon 17088) and N. burbidgeae, were propagated for at least two generations, confirmed to have the morphological features diagnostic of the taxa, and used for DNA extraction. Total genomic DNA was extracted from leaf tissues using the CTAB method (Clarke 2009) and amplified with a high-fidelity Taq polymerase and the primers described in Table S1 of the Supplementary material, using a touchdown cycling technique and annealing temperatures of 50-55 C. For species without existing sequence information, the nuclear DNA regions amplified were the internal transcribed spacer of rRNA (ITS), the long and short forms of the chloroplastexpressed glutamine synthetase (GSL and GSS respectively; Clarkson et al. 2010), RNA-dependent RNA polymerase 1 (RDR1; Bally et al. 2015) and alcohol dehydrogenase C locus (ADHC); and the plastid region amplified was maturase K (matK). Because nuclear and chloroplast DNA sequences were not available for N. sp. Corunna (D.E.Symon 17088) or N. burbidgeae, regions of ribosomal ITS and matK from these accessions, and also from N. benthamiana, were amplified and sequenced (Table S2). Cycle sequence reactions, performed with BigDye Terminator (ver. 3.1, Applied Biosystems) at suggested cycling conditions, were purified with an ethanol and EDTA precipitation. After purification, the amplified fragments were run on a Life Technologies 3500 Genetic Analyser at the Central Analytical Research Facility Genomics Laboratory at the Queensland University of Technology. Analysis of output chromatograms and further preliminary sequence editing was conducted using Geneious (ver. R11, Biomatters, New Zealand, see www.geneious.com/; Kearse et al. 2012).

Sequence alignment and phylogenetic analyses
Gene sequences were aligned (Fig. S1 of the Supplementary material) using MUSCLE (ver. 3.8.31, see http://www.drive5. com/muscle/; Edgar 2004), followed by manual adjustments in BioEdit (ver. 7.2.6, see https://bioedit.software.informer.com/). For concatenation, the sequences were appended in the following order: ITS, matK, GSL, GSS, RDR1 and ADHC. The Akaike information criteria in ModelFinder in IQ-TREE (ver. 1.5.4, see http://www.iqtree.org/; Kalyaanamoorthy et al. 2017) and jModelTest2 (ver. 2.1.10, see https://github.com/ ddarriba/jmodeltest2/; Darriba et al. 2012) were used to estimate the best-fit substitution models. Model selection and the parameters used are described in Table S3. Phylogenetic analyses of individual and concatenated gene sequences were based on maximum likelihood implemented by a rapid and effective stochastic algorithm in IQ-TREE (Trifinopoulos et al. 2016) including partition files. Additional phylogenetic trees for concatenated data of all six genes were built through Bayesian inference as implemented in MrBayes (ver. 3.2.6, see https://github.com/ NBISweden/MrBayes/; Ronquist et al. 2012) including partition files. The final consensus trees were displayed using FigTree (ver. 1.4.3, see https://github.com/rambaut/ figtree/releases/tag/release_1_3/). Marks et al. (2011a) reported measurements of 21 floral character states from a range of N. section Suaveolentes taxa, with each measurement being based on a minimum of 10 biological replicates. Measurements for selected taxa were extracted from those reported in this paper and data matrices of floral characters were subject to principal-component analysis (PCA) using R (ver. 3.5.0, R Foundation for Statistical Computing, Vienna, Austria) and the package factoextra (ver. 1.0.3, see https://CRAN.R-project.org/package= factoextra/).

Phylogenetic analyses
A maximum-likelihood analysis ( Fig. 1) was performed using the concatenated ITS and matK sequences for N. burbidgeae, N. benthamiana and N. sp. Corunna (D.E.Symon 17088), and equivalent sequences from other Nicotiana taxa and from the Australian genus Anthocercis Labill. (Solanaceae: Anthocercideae) as the designated outgroup (Clarkson et al. 2010). The trees produced with ITS or matK sequences alone are shown in Fig. S1 and S2. For the concatenated tree, the total number of nucleotides used was 2213, of which 406 were variable. Nicotiana formed a well supported clade made up of several lineages and the branching pattern shown in Fig. 1 was consistent with previous analyses of this genus (e.g. see Clarkson et al. 2010). The branching pattern of the ITS tree was like that of the concatenated tree but the matK tree had fewer resolved nodes. All trees placed members of N. section Suaveolentes in a well supported lineage that included N. benthamiana, N. burbidgeae and N. sp. Corunna (D.E. Symon 17088). Nicotiana benthamiana was sister to N. excelsior (J.M.Black) J.M.Black (bootstrap support (BS) = 100%) and N. burbidgeae along with N. umbratica were on early diverging branches among Australian representatives of the section (Fig. 1). All trees showed that N. sp. Corunna (D.E. Symon 17088) and N. goodspeedii were on separate branches, as were N. suaveolens and its synonym N. exigua.
To further confirm N. sp. Corunna (D.E.Symon 17088) as a distinct species, additional nuclear sequences were obtained. Species in the allopolyploid N. section Suaveolentes retain both parental copies of the nuclear-encoded, chloroplastexpressed glutamine synthetase, and the paralogs, called GSL and GSS, have previously been used to determine phylogenetic relationships (Clarkson et al. 2010). As well as N. sp. Corunna (D.E.Symon 17088), GSL and GSS sequences were obtained for N. burbidgeae, N. benthamiana and N. rosulata (S.Moore) Domin and added to the existing N. section Suaveolentes sequence dataset (Table S2). Regions of a further two genes, RDR1 and ADHC, were either sequenced from these species or retrieved from GenBank. Most species in N. section Suaveolentes appear to have retained only one of the parental copies of these genes (Kelly et al. 2013;Bally et al. 2015Bally et al. , 2018. Consistent with this, there were no polymorphisms observed in the amplified ADHC and RDR1 products for any accession. The sequences of all six gene regions (ITS, matK, GSL, GSS, RDR1 and ADHC) were used to generate phylogenetic trees for each individual gene region (Fig. S1-S6) and for a concatenation of all six (Fig. 2)  Suaveolentes using different sequence datasets (Clarkson et al. 2010;Marks et al. 2011a), the patterns from the different generegion sequences had several conflicting branches. Nevertheless, in each tree, with one exception, N. sp. Corunna (D.E.Symon 17088) was distinct from other members of N. section Suaveolentes. On the basis of the ADHC sequences N. sp. Corunna (D.E.Symon 17088) was not distinct from N.burbidgeae. Using the tree generated from the concatenated sequences, among N. section Suaveolentes, the African species N. africana Merxm. and the New Caledonian species N. fragrans Hook. were sister to a well supported clade that contained all the Australian members. Most nodes in the Australian Suaveolentes clade were poorly supported and only two clades, being composed of more derived species, were well resolved. Species in these two groups generally have fewer chromosomes than do other members of the section. In Fig. 2 N. goodspeedii, N. maritima, N. exigua, N. amplexicaulis N.T. Burb. and N. suaveolens. Although both are placed in Clade A, N. suaveolens and its synonym N. exigua do not cluster together. Clade B contains N. excelsior, N. rosulata, N. rotundifolia Lindl., and N. velutina H.-M.Wheeler.

Phenetic analysis of floral characters
Because DNA-based phylogenies pointed to N. sp. Corunna (D.E.Symon 17088) being a distinct species, a phenetic analysis of flowers was performed to obtain further evidence for it being a new species and to find characters that could be potentially useful in its identification. A PCA biplot (Fig. 2, inset)   N. maritima and N. velutina) that overlap its geographic distribution (Fig. 3) and with its sequence-based cladistic sister, N. amplexicaulis. Nicotiana sp. Corunna (D.E.Symon 17088) flower characters formed a cluster that was well separated from N. goodspeedii, N. velutina and N. amplexicaulis, and predominantly separated from N. maritima. Nicotiana sp. Corunna (D.E.Symon 17088) was readily distinguished from N. goodspeedii, by its shorter stamens, smaller floral tubes and calyx, and thinner corollas; it was distinguishable from N. velutina by its shorter calyx length and narrower corolla limb diameter. A PCA that uses all the species described in Marks et al. (2011a) is shown in Fig. S7.

Discussion
The  (N. maritima, N. goodspeedii and N. velutina) have distributions that intersect with N. sp. Corunna (D.E.Symon 17088); however, of the three, only N. goodspeedii has the same chromosome number (n = 16) as does N. sp. Corunna (D.E.Symon 17088) and none has flowers that are identical to those of N. sp. Corunna (D.E.Symon 17088), as demonstrated in the PCA. Nicotiana maritima has recently been recircumscribed following recognition of a new species, N. yandinga (Chase et al. 2018b), that occurs in the vicinity of N. sp. Corunna (D.E.Symon 17088) (Fig. 4A). However, N. yandinga has features, such as indumentum and a genome of 21 chromosome pairs (Chase et al. 2018b), that distinguish it from N. sp. Corunna (D.E.Symon 17088). Altogether, this provides strong evidence that N. sp. Corunna (D.E.Symon 17088) merits recognition as a new species that we name as N. paulineana, sp. nov. and describe below. This posthumously fulfils an ambition of the highly respected Australian plant taxonomist, David Symon. His obituary (Barker 2013) listed several projects Plants glabrous, erect, annual, 0.4-1 m high. Initially singlestemmed, commonly developing several branched stems. Seedlings with cotyledons 6-7 mm long. Basal leaves rosulate, petiolate, attenuate, 10-30 cm long. Eglandular trichomes with one single gland cell on the calyx. Panicles loosely decompound, major branching long, rapidly ascendant. Calyx 5-7 mm long, appressed to tube. Corolla slightly zygomorphic, lobes emarginate, spreading; corolla limb lobes pure white inside, with a green to yellow vein running down the back of each lobe; floral tube broadening above calyx, 15-25 mm long exclusive of limb, 2-3 mm wide, often purplish. Stamens all included, 4 anthers at or close to one level, near mouth of corolla, anther of 5th stamen 2-3 mm lower; filaments inserted in lower half of corolla. Capsule not constricted or thickened, no seeds retained. Seeds reinform, sinuous seeds testa, 0.9 mm long, brown. Chromosome number 16 pairs.

Notes
This species can be distinguished from Nicotiana goodspeedii and N. maritima by its narrower leaf shape and more deeply cleft corolla lobes (Fig. 3) and from N. maritima by its lack of a woolly indumentum The holotype is a flowering specimen taken from the plant used by Marks et al. (2011b) to assess the chromosome number. Additional notes on the specimen lodged at The University of Melbourne Herbarium (MELUD106464) say 'Hydroponically grown for chromosome count, actual plant used for count. . .', 'cultivated from seed supplied by David Symon (AD) with provenance', 'grown from Symon 17088'. David Symon indicated to Claire Marks that this seed lot came from Corunna Station and has the provenance described for D.E. Symon s.n. (AD 169037), a specimen grown from a wild collection from Corunna Station.

Distribution and habitat
The five known collection locations of N. paulineana are all from South Australia within a 50-km radius south of Port Augusta, and its habitat is primarily vegetated natural and   . amplexicaulis, N. velutina, N. goodspeedii, N. maritima, N. suaveolens and N. rotundifolia based on Burbidge (Burbidge 1960), Horton (Horton 1981), Chase (Chase et al. 2018b), and records in the Atlas of Living Australia (ala.org.au). Overlapping distributions are denoted by colours (black, green and purple) and, in the magnified region (A), the locations of N. yandinga and N. paulineana are shown as red and blue pins respectively.