Eutrema giganteum (Brassicaceae), a new species from Sichuan, southwest China

Abstract Eutrema giganteum (Brassicaceae), a new species from Hengduan Mountains in Sichuan Province, southwest China, is described, and its relationships to the closely related E. yunnanense is discussed based on morphological, cytological, and molecular data. It is similar morphologically to E. yunnanense but is readily distinguished by having robust (vs. slender), erect (vs. decumbent), and branched (vs. mostly simple), and rather tall stems (60–110 cm vs. 20–60 cm); curved (vs. straight), smooth (vs. torulose), and shorter fruit (5–8 mm vs. 8–15 mm); and fewer ovules per ovary (1–4 vs. 6–10). All examined individuals from different populations of E. giganteum clustered into a single clade sister to E. yunnanense in phylogenetic analyses using the combined nuclear ITS and plastid DNA datasets. Our cytological studies revealed that the chromosome number of E. giganteum is 2n = 44, with a genome size of 1160 (±8) Mb, while that of E. yunnanense is 2n = 28, with a genome size of 718 (±15) Mb. Multiple lines of evidence support the recognition of E. giganteum as a distinct species well differentiated from E. yunnanense.


Introduction
Eutrema R.Br. (Brassicaceae) is an important genus that includes the model plant for salt-tolerance E. salsugineum (Pall.) Al-Shehbaz & Warwick and the economic wasabi plant E. japonicum (Miq.) Koidz. This genus was expanded to comprise 26 species (Al-Shehbaz and Warwick 2005; with 16 transferred from four previously genera, Taphrospermum C. A.Mey., Thellungiella O.E.Schulz, Neomartinella Pilger, and Platycraspedum O.E.Schulz (Al-Shehbaz and Warwick 2005;. Since then, several new species were described (Ning et al. 2005;Al-Shehbaz 2007;Gan and Li 2014;Xiao et al. 2015;Hao et al. 2015Hao et al. , 2016. During botanical expeditions to Hengduan Mountains in southwest China from 2013 to 2016, we discovered three populations of Eutrema that were strikingly unusual in having large size, big cordate leaves, and stout, erect and branched stems. Only E. yunnanense Franch. has such similar big cordate leaves (10-20×10-16 cm), but its stems are slender, decumbent, and rarely branched. Therefore, it was immediately suspected these populations may represent an undescribed new species. In order to further test this hypothesis, morphological, molecular, and cytological studies are presented here on those two species and two related species E. japonicum and E. thibeticum Franch. were conducted with herein.

Material and methods
For morphological comparisons and taxonomical treatments, we examined more than ten living individuals from each population of Eutrema giganteum (three populations) and E. yunnanense (two populations), and photos of all herbarium specimens of E. yunnanense preserved in the Chinese Virtual Herbarium (http://www.cvh.org. cn/). We followed Hu et al. (2015) and Hao et al. (2015Hao et al. ( , 2016 in examining genetic differences between two morphological groups, and three individuals were studied from each population. In order to determine the systematic position of E. giganteum, we further sampled two populations each for E. japonicum and E. thibeticum because they were phylogenetically related to E. yunnanense. We sampled one individual each of E. We extracted total DNA from silica gel-dried leaves using the modified CTAB method (Doyle and Doyle 1990). The internal transcribed spacer (ITS) and four chloroplast DNA regions (trnL-F, psbA-trnH, rbcL, matK) were amplified for phylogenetic analyses. The five pairs of primers used for amplifying and sequencing trnL-F, psbA-trnH, rbcL, matK and nuclear nrITS were the same as those used by Hu et al. (2015). PCR amplification and sequencing approaches followed Hu et al. (2015) and Hao et al. (2015). For those ITS sequences with double peaks in those possible hybrids, we further cloned them using vector pGEM-T (Promega, Madison, Wisconsin). We selected ten positive clones for sequencing with primers "sp6" and "t7". We deposited all new sequences in GenBank under the accession numbers KY969594-KY969625. We aligned DNA sequences using Clustal X (Thompson et al. 1997) and MEGA 5.10 (Tamura et al. 2011) and refined them manually. We concatenated sequences from all four cpDNA fragments into a single matrix for Maximum parsimony (MP) and Maximum likelihood (ML) analyses because of their common inheritance without obvious recombination. To evaluate the congruence of the plastid and nuclear datasets, we employed the incongruence length difference (ILD) test (Farris et al. 1995). The ILD test was carried out using the PAUP* 4.10b (Swofford 2003) with the following settings: 1000 replications, each using a heuristic search with 100 random-addition-sequence replicates and TBR branch swapping. We performed ILD tests between each pair of the cpDNA dataset, and between the nrITS dataset and the combined cpDNA dataset. The P-values smaller than 0.01 were considered to be significant incongruent (Cunningham 1997). We reconstructed phylogenetic relationships based on three datasets (nrITS, cpDNAs and combined nrITS+cpDNAs) respectively using MP analyses by PAUP* 4.10b (Swofford 2003), employing a heuristic search with 10,000 replicates and TBR branch swapping. We estimated bootstrap values (Felsenstein 1985) with 1000 replicates and 100 random-addition-sequence replicates per bootstrap replicate. Because indels may contain potential phylogenetic information (Simmons et al. 2001), we coded them using the simple code method by GapCoder (Young and Healy 2003) for phylogenetic analyses. We performed ML analyses using RAxML 7.2.6 (Stamatakis 2006) with the order: raxmlHPC -f a -s sequence. phy -n boot2 -m GTRGAMMA -x 1234 -# 1000 -n outname. We selected the GTRGAMMA model and estimated ML bootstrap analyses with 1000 replicates. We followed Hao et al. (2015) to carry out chromosome number count and genome-size determination.

Morphological comparison and geographical distribution
Our study of herbarium specimens and living plants demonstrated that Eutrema giganteum is a morphologically distinctive species. As shown in Fig. 1, it is a glabrous herbaceous perennial, the tallest in the genus, with robust, erect or ascending stems 60-110(-140) cm and alternate branches. Each branch is divaricate-ascending or almost perpendicular to stem and originates from the axil of cauline leaf. The fruits are narrowly oblong, 5-8 × 2-3 mm and curved, but not torulose. Eutrema giganteum is most similar to E. yunnanense, but it is readily distinguished from the latter by having stout (vs. slender), erect (vs. decumbent), and branched (vs. rarely branched) stems (60-110 cm vs. 20-60 cm; curved (vs. straight), smooth (vs. torulose), and shorter (5- According to specimen records and field investigation, Eutrema giganteum is currently known only from Hengduan Mountains in western Sichuan at elevation between 2300 and 2900 m (Fig. 2), while E. yunnanense may occurs in the southern part of Hengduan Mountain, Yunnan province, at elevation between 2500 and 3200 m. Zhou et al. (2001) reported that E. yunnanense is widely distributed in other provinces of China (for example, Anhui, Gansu, Hebei, Hubei, Hunan, Jiangsu and Zhejiang) at elevation between 400 and 3500 m. Phylogenetic and taxonomic relationships between populations of these provinces and those from Yunnan await future studies.

Genetic relationship of Eutrema giganteum with E. yunnanense, E. japonicum and E. thibeticum
Based on sequence variations of nrITS, cpDNAs, and the combined nrITS and cpDNAs (Table 2), phylogenetic analyses suggested that E. giganteum is mostly related to E. yunnanense (Figs 3, 4). In the MP analyses of nrITS sequence data, E. giganteum, one clone of E. yunnanense, E. japonicum, and E. thibeticum formed a single cluster, which together was sister to the other E. yunnanense sequences with medium support (50%-70%). In the MP analyses of the sequence variations from cpDNAs, E. giganteum and E. yunnanense formed a single cluster, then sister to E. japonicum and E. thibeticum with higher support (>70%). The P-values resulting from the ILD tests show that there is significant incongruence between the four cpDNA and nrITS when including all species (P = 0.003). After removing conflicting sequences of E. yunnanense, E. altaicum and E. verticillatum, the P value rose to 1.000, indicating that there is no significant incongruence between nrDNA and cpDNA datasets. We therefore combined them for further analyses. In the MP analyses of the combined dataset, all E. giganteum individuals formed a single cluster, sister to the cluster comprising E. yunnanense individuals with medium supports (50%-90%). The clade comprising both of them was sister to the clade formed by both E. japonicum and E. thibeticum with a high support (94%) (Fig. 3). ML analyses produced similar tree topologies to MP trees but the supports were higher than MP analyses (Fig. 4).

Chromosome number and genome size
Two populations of Eutrema giganteum from Xiling Snow Mountain and Erlang Mountain, and one population of E. yunnanense from the type locality, Cangshan Mountain, were cytologically examined. Mitotic chromosome number of E. giganteum was determined as 2n = 44 (Fig. 5), while that of E. yunnanense was 2n = 28, as the same as counted by Du and Gu (2004). Genome size of E. giganteum was determined as 1160 (±8) Mb while that of E. yunnanense was 718 (±15) Mb.
Additional specimens examined (