Persicariajucunda var. rotunda (Polygonaceae, Persicarieae), a distinct distylous taxa raised to specific rank

Abstract Persicariajucunda(Meisn.)Migovar.rotunda (Z.Z.Zhou & Q.Y.Sun) Bo Li was originally published in the genus Polygonum L. and treated as a variety of P.rotundum Meisn. [≡Persicariajucunda (Meisn.) Migo]. After carefully comparing the macro- and micro-morphological characteristics of the achenes, leaf epidermis and tepals and the habitat between the variety and its typical variety, we confirmed that P.jucundavar.rotunda is clearly different from P.jucunda and should not be treated as a variety, but be raised to a specific rank as P.rotunda (Z.Z.Zhou & Q.Y.Sun) Bo Li. The species is distylous and could be easily distinguished from all other Persicaria taxa by a combination of morphological characters, such as completely decumbent leafless basal branches, almost sessile leaves, linear-lanceolate with rounded leaf bases, spicate, short and dense inflorescences, slender pedicels longer than bracts and dimorphic flowers and achenes. P.rotunda is endemic to several large wetlands of eastern China and usually occurs as one of the dominant species in some plant communities.

Persicaria jucunda var. rotunda (Z.Z.Zhou & Q.Y.Sun) Bo Li (Li et al. 2013) was originally published in the genus Polygonum and treated as a variety of P. rotundum (≡Persicaria jucunda) (Zhou et al. 2007). The variety resembles P. rotundum in having glabrous stems and prostrate at base, densely spicate inflorescences, pinkish tepals, slender pedicels longer than bracts and trigonous achenes, but differs from the typical variety in having truncate and linear-lanceolate leaf blades with barely noticeable petioles (Zhou et al. 2007). However, when conducting a micro-morphological study of Chinese Persicaria species, we found that there are a number of distinct differences between the two taxa, including the shape of epidermal cells of both leaf sides, the occurrence of stomata on adaxial leaf surface, the stomatal type of abaxial leaf surface and the sculpture of achene surface. After re-examining the macro-morphology and the habitat of the two taxa, we confirmed that P. jucunda var. rotunda is clearly different from P. jucunda and should not be treated as a variety, but be raised to a specific rank as Persicaria rotunda (Z.Z.Zhou & Q.Y.Sun) Bo Li.

Materials and Methods
The field investigations were carried out from 2014 to 2019. Fresh leaf materials and flowers of both P. jucunda and P. rotunda were collected and immediately fixed in FAA solution (formalin: acetic acid: alcohol = 18:1:1). The measurement of morphological characters was conducted based on both herbarium specimens (JXAU, acronym according to Thiers 2019) and living plants by using a micrometer and a stereomicroscope. To make a morphological comparison between P. jucunda and P. rotunda, the variability of four quantitative characters (leaf length, leaf width, number of leaf lateral vein pairs, inflorescence length) was evaluated using univariate statistics (box plots) by SPSS 11.5 statistical software package (SPSS Inc., Chicago, IL, USA). To confirm the distyly in P. rotunda, the height of stigmas and anthers were measured for a single flower removed from 30 individuals per style morph. Methodology follows Chen and Zhang (2010).
For light microscopy (LM) observation of leaf epidermis, samples were taken from the mature leaves fixed in FAA solution, dissected under a OPTPro stereoscope (Chongqing Optec Instrument Co. Ltd., China), stained in a solution of 1% safranin, and cleaned in distilled water three times before being mounted in glycerine jelly. Observations and micrographs were conducted randomly from 5 prepared slides per species under LM. Terminology follows Hou (2006).

Taxonomy
Phenology. Flowering was observed from July to December and fruiting from late July to January.
Distribution and habitat. Persicaira rotunda is currently known only from several wetlands in north of Jiangxi Province and southwest of Anhui Province, eastern China and mainly grows in marshy and grassy areas around Daguan Lake, Lihu Lake, Longgan Lake, Poyang Lake Qingcao Lake, Shengjin Lake, Shimeng Lake and Wuchang Lake, which are several small to large lakes located near the Yangtze River. It usually occurs as one of the dominant species of some lakeshore plant communities (Fig. 1A). Notes. As noted by Zhou et al. (2007), P. rotunda is most similar to P. jucunda (Fig.  3) in gross morphology, particularly in having uninterrupted spicate inflorescences with dense flowers and slender pedicels longer than bracts. However, the authors did not notice that both of the species are distylous, which is another important similarity between the two taxa. The distyly of P. jucunda was firstly observed and confirmed by Chen and Zhang (2010). In the present study, we confirmed that P. rotunda is also a typical distylous species. The heights of the stigmas (4.63 ± 0.191 mm vs. 2.61 ± 0.056 mm, L-morph vs. S-morph) and anthers (2.74 ± 0.092 mm vs. 4.68 ± 0.178 mm, Lmorph vs. S-morph) are reciprocal in the two morphs. However, P. rotunda is clearly different from P. jucunda, not only in some morphological traits (Fig. 4), but also in several micro-morphological characters (Table 1).
Besides the differences summarised by Zhou et al. (2007), such as leaf shape, leaf width, petiole length and stem diameter, we observed several additional morphological traits that are clearly distinct between P. rotunda and P. jucunda. The stems of P. rotunda have 6-26 basal branches which are leafless and completely decumbent with numerous fibrous roots at each node. On the upper nodes of each basal branch, there are 3-12 flowering shoots which are prostrate to erect and normally bearing leaves and inflorescences (Fig. 1C). However, the stems of P. jucunda are mostly erect or only prostrate at the base and the number of its branches are much fewer than those of P. rotunda. Persicaria rotunda also has more pairs of leaf lateral veins and much shorter inflorescences than P. jucunda (Fig. 4).   Though both of P. rotunda and P. jucunda have dimorphic flowers, the achenes of P. rotunda are also dimorphic, with the L-morph ellipsoid in shape and dark brown in colour, whereas the S-morph achene is ovoid in shape and black in colour (Fig. 5A). Additionally, the L-morph achenes of P. rotunda have larger size and more raised reticulate epidermal ornamentations than those of the S-morph (Figs. 5B-E). In contrast, the achenes of P. jucunda are homomorphic with the same smooth surfaces in both morphs (Chen and Zhang 2010). Amongst the distylous taxa reported in Persicaria, P. rotunda is, so far, the only species that shows dimorphic features on achenes.
Leaf epidermis characters have been tested to be of important taxonomic significance in Persicaria (Hou 2006, Zhu et al. 2007, Yasmin et al. 2010. When observing the leaf epidermis of P. rotunda and P. jucunda, we found that there are significant differences in the leaf epidermal micro-morphology of the two taxa (Fig. 6). In P. rotunda, the adaxial leaf epidermal cells are polygonal in shape with the straight anticlinal walls and no stomatal apparatus or gland occurs on the surface (Fig. 6A). However, the adaxial leaf epidermis of P. jucunda is covered by irregular epidermal cells with the anticlinal walls straight to curved and has mostly anisocytic or occasionally paracytic stomata and sparsely two-celled peltate glands (Fig. 6C). On the abaxial leaf epidermis, plenty of four-celled peltate and spheroidal glands, anisocytic stomata and irregular epidermal cells with the anticlinal walls curved to sinuolate were observed for P. rotunda (Fig. 6B), while in P. jucunda, no glands have been found, the stomata are paracytic and the anticlinal walls of epidermal cells are sinuolate to sinuate (Fig. 6D).
In the protologue, Zhou et al. (2007) also investigated the tepal micro-characteristics of P. rotunda and P. jucunda and listed their differences: the anticlinal walls of epidermal cells are curved to sinuolate in P. rotunda, while sinuolate to sinuate in P. jucunda; the cuticular layer has longitudinally 10-14 of sinuate striates in P. rotunda, while 12-18 straight to sinuolate striates in P. jucunda. Taking all the above morphological and micro-morphological evidence together, we think that P. rotunda represents a distinct species in Persicaria and it should not be placed under P. jucunda as a variety, but be treated as a separate species.