Semenovia gyirongensis (Apiaceae), a new species from Xizang, China

Abstract Based on morphology and molecular data, a new species Semenovia gyirongensis Q.Y.Xiao & X.J.He, from Gyirong County, Xizang, China, is described and illustrated. It is morphologically most similar to S. malcolmii (Hemsley & Pearson) Pimenov, but differs in its cylindric much-branched root, intensively branching long underground caudex with distinct nodes, narrowly ovate to ovate terminal leaf lobes, oblong bracts with obtuse-rounded or cuneate apex.


Introduction
Semenovia Regel & Herder (Apiaceae, tribe Tordylieae), an endemic Asiatic genus, occurs in southwest, central and east Asia, with its center of diversity in the Pamir mountains (Shen 1992;Pimenov and Leonov 1993;Pu and Watson 2005;Ukrainskaja 2015). Most species of Semenovia are narrow endemics and grow mainly in the mid-elevation to highland areas of mountains (Ukrainskaja et al. 2013;Ukrainskaja 2015). The latest revision of Semenovia was conducted by Ukrainskaja et al. (2013), who recognized 29 species. There are 6 species of Semenovia in China, two of which are endemic to Qinghai-Tibetan Plateau (QTP) (Ukrainskaja et al. 2013). Semenovia is a perennial herb with pinnate leaves, entire or branched caudex, unequal (outer ones are larger) or subequal outer and inner petals, small bracts and bracteoles, well developed styles, thinly and narrowly winged marginal ribs, filiform vittae, solitary vittae per vallecular and two (rarely four) on commissural surface (Regel and Herder 1866;Mandenova 1959;Alava 1987;Pu and Watson 2005). Caudex states (underground, overground or emergent; unbranched or branched; short or long branches) are regarded as the most important diagnostic characters within the genus Semenovia (Ukrainskaja et al. 2013;Ukrainskaja 2015).
During examining specimens of Semenovia, we encountered one collection (Z. Y. stored in HNWP,KUN and PE), which was collected from Gyirong County, Xizang, China and was unable to identify as any described species. In August 2016, we carried out field investigation to the exact locality and gathered both flowering and fruiting plant from the natural population. After thoroughly consulting relevant literatures (e.g. Mandenova 1959;Alava 1987;Vinogradova and Kamelin 1986;Ukrainskaja et al. 2013;Ukrainskaja 2015) and herbarium specimens, as well as comparing this taxon with all described species within the genus, we come to the conclusion that the specimens from Gyirong represent a hitherto undescribed species. Herein a new name Semenovia gyirongensis is proposed, and detailed descriptions and comments of this new species, as well as comparisons with its morphologically similar species are given.
Sampling was conducted from type localities of S. gyirongensis (Gyirong County, Xizang) and S. malcolmii (Shuanghu, Nyima County, Xizang) during 2015-2016. Photographs in the field were made using a Nikon D7100 camera. The measurements of the morphological features were conducted using a vernier caliper. Mericarps were photographed using stereomicroscope Nikon SMZ 25 (Japan). Fruits from formalde-hyde-acetic acid-alcohol (FAA) preserved material were used in the anatomical study. Pollen was examined from anthers collected directly in the field. The pollen grains were mounted on clean aluminum stubs using conducting carbon adhesive tabs, coated and then scanned with a JSM-7500F scanning electron microscope (SEM). General terminologies for this study followed Kljuykov et al. (2004). Voucher specimens were deposited in the herbarium of Natural History Museum of Sichuan University (SZ).

DNA extraction, amplification and sequencing
Total genomic DNA was extracted from silica gel-dried leaves and herbarium materials according to the protocols of plant genomic DNA kit (Tiangen Biotech, Beijing, China). The internal transcribed spacer (ITS) and external transcribed spacer (ETS) of nuclear ribosomal DNA (nrDNA) were used for phylogenetic inference based on the previous study (Logacheva et al. 2010). The primer pairs ITS4 / ITS5 (White et al. 1990) and 18S-ETS (Baldwin and Markos 1998) / Umb-ETS (Logacheva et al. 2010) were used to amplify the ITS and ETS regions, respectively. Amplification was carried out in a 30µL volume with 2 µL plant total DNA, 10 µL ddH 2 O, 1.5 µL forward primer, 1.5 µL reverse primer and 15 µL 2 × Taq MasterMix (cwbio, Beijing, China). PCR cycling profile included a denaturing step at 95 °C for 4 min, followed by 35 cycles of 45 s at 95 °C, annealing at 54 °C for 45 s and extension at 72 °C for 1 min, with a final extension for 10 min at 72°C. Sequencing (both directions) was carried out using the amplification primers on an ABI 3730 sequencer at the Beijing Genomics Institute (BGI) in Beijing, China. All newly reported sequences were deposited in GenBank and accession numbers along with sample codes and localities were given in Suppl. material 1: Table S1.
Sequence alignment and phylogenetic analysis 62 accessions were obtained from GenBank for the nrDNA ITS and ETS, and 4 were newly sequenced for this study (Suppl. material 1: Table S1), representing 56 species from 17 genera of tribe Tordylieae (plus the new species S. gyirongensis, a total of 22 species of Semenovia were included) and 2 species of Conium. Sequence data for the ITS 5.8S region were excluded from the analysis because they were unavailable for many previously published taxa. Conium maculatum L. and Conium sphaerocarpum Hilliard & Burtt were selected as outgroups (Ajani et al. 2008;Banasiak et al. 2013).
SeqMan (Burland 2000) was used to edit DNA sequences and obtain consensus sequences. DNA sequences were aligned with ClustalX ver. 2.1 (Larkin et al. 2007) and then adjusted manually using MEGA7 (Kumar et al. 2016). Topological incongruence the partition between ITS and ETS was tested using the incongruence length difference (ILD) test (Farris et al. 1994) implemented in PAUP* version 4.0b10 (Swofford 2003). The two markers were then combined and analyzed using Bayesian Inference (BI), Maximum Likelihood (ML), and Maximum Parsimony (MP). Pairwise nucleotide differences of unambiguously aligned positions were determined from the distance matrix option in PAUP* (Swofford 2003). The BI analysis was performed in MrBayes version 3.2 (Ronquist et al. 2012). MrModeltest version 2.2 (Nylander 2004) was used to select a best-model (GTR+G) of nucleotide substitution. Four simultaneous runs were performed using Markov chain Monte Carlo (MCMC) simulations for 20 million generations, starting from a random tree and sampling one tree every 1000 generations. The convergence and effective sample size (ESS) of each replicate were checked using Tracer v. 1.6.0 (Rambaut et al. 2013). The first 25% of obtained trees were discarded as burn-in and the remaining were used to calculate a 50% majority-rule consensus topology and posterior probability (PP) values. For the ML analysis, phylogenetic reconstruction was performed using RAxML-HPC BlackBox ver. 8.2.10 under the GTR+G nucleotide substitution model and 1000 rapid bootstraps on the CIPRES Science Gateway ver. 3.3 (Miller et al. 2010). The MP tree was obtained using the programs PAUP* version 4.0b10. Heuristic searches were replicated 1000 times with random taxon addition sequences, tree bisection-reconnection (TBR) branch swapping, and setting the maximum number of trees to 10,000. Bootstrap values were calculated from 1,000,000 replicate analyses using 'fast' stepwise-addition of taxa and only those values compatible with the majority-rule consensus tree were recorded.
These characters of S. gyirongensis allow for easy discrimination from morphologically similar species S. malcolmii (fusiform unbranched roots, unbranched to muchbranched, short overground or emergent caudex, without distinct nodes, linear to narrowly long-ovate terminal leaf lobes and linear to narrowly ovate bracts, apex acute,     (Alava 1987;Ukrainskaja et al. 2013).

Phylogenetic analysis
The matrix of combined nrDNA ITS and ETS data had an aligned length of 775 positions, of which 310 were parsimony informative, 283 were constant, and 182 autapomorphic characters. The results of the ILD test for those 66 accessions common to both ITS and ETS datasets revealed that these loci yield significantly different phylogenetic estimates (P = 0.001). However, numerous reports indicated that the results of an ILD test do not adequately assess data combinability (e.g. Yoder et al. 2001;Barker and Lutzoni 2002;Hipp et al. 2004). Despite the incongruence of these data, the topologies of the ITS-and ETS-derived trees did not conflict. Meanwhile, the analysis of the combined dataset using ML, MP and BI yielded similar trees and had higher MP Bootstrap values (MP-BS), ML Bootstrap values (ML-BS) and BI posterior probabilities (BI-PP). The Bayesian majority rule consensus tree based on combined analysis was presented in Fig. 3. ML-BS, MP-BS and BI-PP values were showed along the branches. Based on our reconstructed phylogeny, 5 major evolutionary clades (Cymbocarpum clade, Heracleum sensu stricto clade, Semenovia clade, Tetrataenium I clade and Tetrataenium sensu stricto clade) of tribe Tordylieae were identified (Fig. 3), which was consistent with previous works (Logacheva et al. 2010). The Semenovia clade was well supported (ML-BS 89%; MP-BS 61%; BI-PP 1.00) comprising Zosima, Semenovia and the monotypic genera Tordyliopsis, Pastinacopsis and Kandaharia and could be divided into three sub-clades (A, B, and C). The subclade B was strongly supported (ML-BS 95%; MP-BS 73%; BI-PP 1.00), but subclade A (ML-BS 54%; MP-BS <50%; BI-PP 0.87) and subclade C (ML-BS <50%; MP-BS <50%; BI-PP 0.79) were weakly supported (Fig. 3). The monotypic genera Pastinacopsis fell into sub-clade A with 6 species of Semenovia, while two species of Zosima and the monotypic genera Kandaharia intermixed within sub-clade C with the largest number of Semenovia taxa (12 species). Subclade B consisted of Tordyliopsis brunonis DC., S. gyirongensis, S. pamirica (Lipsky) Mandenova and S. thomsonii (C.B.Clarke) Mandenova (Fig. 3). Within sub-clade B, three accessions of S. gyirongensis formed a well monophyletic clade (MP-BS 100%; ML-BS 100%; BI-PP 1.00), as a sister group to S. pamirica (Fig. 3). The genus Semenovia is not monophyletic based on these phylogenies (and neither is Zosima) (Fig. 3). The circumscription of all genera within the Semenovia clade should be revised, but this is out of the scope of the present study.

Geographical distribution
Geographically, S. gyirongensis is close or adjacent to T. brunonis, S. pamirica, S. malcolmii and S. thomsonii but do not overlap (Suppl. material 1: Fig. S2). S. gyirongensis is only known from the type locality, Gyirong County, Xizang, China. T. brunonis is distributed in Bhutan, Nepal, India (Sikkim, Himachal Pradesh, Uttarakhand) and also in South Xizang, but grows in subalpine moist dwarf scrubs, among shrubs and boulders (Pu and Watson 2005;Kumar et al. 2014). S. pamirica is confined to Pamiro-Alai and Central Asia (Shishkin 1968). S. malcolmii occurs in the QTP and adjacent regions, but never in Gyirong County. S. thomsonii is in Jammu, Kashmir and in whole India (Ukrainskaja et al. 2013) (Suppl. material 1: Fig. S2).

Conclusion
Taking the morphology, molecular and geographical distribution evidences into consideration, it is thus clear that S. gyirongensis should be recognized as a new, distinct species of Semenovia. Diagnosis. Semenovia gyirongensis is most similar to S. malcolmii, but can be easily distinguished by its roots (cylindric much-branched vs. fusiform unbranched), caudex (intensively branching, long, underground, with distinct nodes vs. unbranched to much-branched, short, overground or emergent, without distinct nodes), terminal leaf lobes (narrowly ovate to ovate vs. linear to narrowly long-ovate), and bracts (oblong, apex obtuse-rounded or cuneate vs. linear to narrowly ovate, apex acute).
Fruit anatomy. Exocarp is formed by one layer of small cells. Outer mesocarp layer is of thin-walled parenchyma cells; inner mesocarp (hypendocarp) is consisted of thick-walled lignified fibrous cells. Five ridges are found on each mericarp. Vascular bundles are thin in dorsal ridges, broad in marginal ridges and commissural side. There are 4 dorsally and 2 ventrally vittae. Endoderm is located as one line under the vittae and seems to be integrated with the spermoderm. The seed is composed of endosperm and spermoderm with a thickened cell wall (Suppl. material 1: Fig. S3).
Phenology. The species was found flowering in July-September, fruiting in August-October. Distribution and habitat. S. gyirongensis is only known from the type locality, China, Xizang, Gyirong County, Woma village, near Longda (Suppl. material 1: Fig. S2). It grows on screes, rocky slopes and sandy places, at elevations between 4000 and 4150 m.
Etymology. The specific epithet is derived from the type locality, Gyirong County in Xizang, China.
Conservation status. S. gyirongensis is hitherto known only from Gyirong County (the type locality) where it usually grows on screes, rocky slopes and sandy places, locally common. In field investigation, we found that the area is subjected to overgrazing pressure and only a handful of individuals can escape from eating or trampling, ultimately blossoming and fruiting. Because of its localized distribution and grazing pressure, it should be assessed as "Vulnerable" (VU) according to the IUCN (2016).