Juncus quartinianus (Juncaceae, sect. Ozophyllum): A Neglected Species from the Horn of Africa and Its Re-Description Based on Morphological SEM Studies

Juncus quartinianus (Juncaceae sect. Ozophyllum) was described by Richard in 1851 from Ethiopia. Some authors have treated this species as a synonym of J. fontanesii and others as a synonym of J. oxycarpus. Based on morphological analyses of flowers, fruit and seeds, we propose to restore J. quartinianus as a distinct species from both these taxa. Its detailed re-description and an identification key to the morphologically similar species of Juncus sect. Ozophyllum are provided.


Introduction
The genus Juncus L. [1] is a group of widely distributed perennial or rarely annuals herbs, which is divided into two subgenera and ten sections. It contains approximately 315 species, with centres of diversity in temperate zones [2][3][4]. Section Ozophyllum Dumort. [5], the largest in the genus Juncus, consists of annuals or rhizomatous to caespitose perennials with terete stems and leaf blades that are perfectly or imperfectly septate. The section comprises about 84 species with centres of distribution in eastern North America, southwestern Europe and the Far East [2,6].
Juncus oxycarpus was described by Kunth [7]. According to the latest available monograph of the Juncaceae family [2] and Juffe [8], J. oxycarpus occurs in Africa from Southern Sudan, Ethiopia, Eritrea and Cameroon to South Africa. Plants from Eritrea, Ethiopia, Kenya and even Tanzania used to be referred to in the literature as Juncus fontanesii J. Gay ex Laharpe [9]. Kirschner et al. [2] reported that some specimens from Ethiopia and Eritrea that are classified as J. oxycarpus have longer capsules and are similar to J. fontanesii subsp. pyramidatus (Laharpe) Snogerup, which occurs in the eastern Mediterranean eastwards to Iraq, in North Africa from Tunisia eastwards, and in the Arabian Peninsula. Indeed, in the course of studying the herbarium material of Juncus oxycarpus and J. fontanesii subsp. pyramidatus, some unusual specimens from Ethiopia, Eritrea and Somalia were found. These were previously identified a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 mainly as J. oxycarpus, but differ in their capsules that are longer than the perianth and their quite different seed coat surface. Some specimens were initially classified as J. fontanesii; however, we found that they do not belong to the latter species, because J. fontanesii possesses anthers that are distinctly longer than filaments but in the studied plants from Ethiopia, Eritrea and Somalia the situation is reversed (the anthers are shorter than the filaments or rarely, the anthers and filaments are equally sized).
Thus, our detailed morphological studies show that these specimens represent a distinct species. After careful analysis of the literature and herbarium material, it was found that this species was named Juncus quartinianus A. Rich. [10]. Type specimens were collected by R. Quartin-Dillon and A. Petit in Chiré (Ethiopia) in July 1844. This taxon was distinguished according to the following features: sessile, lateral single head, lanceolate, acute, subcastaneous tepals, and the inner ones slightly longer than the outer, six stamens one-third as long as the tepals and pyramidal capsules longer than the perianth [10]. According to Buchenau [11], type specimens have erect, slender and delicate stems, two flower heads with 6-16 flowers, pale green tepals, the inner conspicuously longer than the outer, capsules about one-third longer than the perianth and seeds that are similar to those of J. articulatus L. The same author in 1890 mentioned that specimens collected by Schimper represent a form of Juncus fontanesii with greenish flowers, and the material collected by Quartin-Dillon and A. Petit is similar to both J. fontanesii and J. articulatus [12]. Baker [13] placed J. quartinianus as a synonym of J. fontanesii, which according to the author, occurred in Eritrea, Ethiopia, British East Africa (currently Kenya and Uganda), German East Africa (a part of present Tanzania), British Central Africa (present Malawi), in the Mediterranean region and western Asia. Four years later, Buchenau [14] suggested that specimens classified as J. quartinianus probably represent J. oxycarpus, but some specimens from Ethiopia, Eritrea and even Kenya were still classified as J. fontanesii. Subsequent research, however, shows that this latter taxon is restricted only to the Mediterranean region and in fact, does not extend to Ethiopia and Eritrea [15]. Therefore, according to Carter [16], specimens from the type collection of J. quartinianus differ from specimens of J. oxycarpus only by having pale-green tepals and brown fruits, and thus, "these two names must be regarded as synonyms".
The main goal of the present paper is to show that J. quartinianus is a well-defined species that occurs in Ethiopia, Eritrea and Somalia. Another purpose of our article is to indicate the differences among three taxa: J. quartinianus, J. oxycarpus and J. fontanesii subsp. pyramidatus.

Materials and Methods
In total, 56 specimens (from 18 herbarium sheets) of J. quartinianus, 487 specimens (from 259 herbarium sheets) of J. oxycarpus (S1 Appendix) and 260 (from 86 herbarium sheets) of J. fontanesii subsp. pyramidatus (S2 Appendix), from the following 21 herbaria: B, BEI, BR, E, ETH, FR, H, JE, K, L, LG, LISU, MO, POZG, PRE, S, SAM, UPS, WAG, WRSL and WU were revised. For statistical analysis, we selected a representative subset of specimens from the geographic range of these three taxa. They are marked with an asterisk ( Ã ) after the herbarium acronym in the citations of representative specimens.

Statistical analyses
The statistical analyses are based on 38 specimens of J. quartinianus, 71 specimens of J. oxycarpus and 75 specimens of J. fontanesii subsp. pyramidatus. Measurements were conducted only on mature specimens using a Nikon SMZ 800 stereoscopic microscope. We analysed 19 qualitative and quantitative characters, including two describing ratios between characters ( Table 1).
The width of the structure was measured at the widest point and the length of capsule and tepals from the base to the top. The inflorescence was measured from the base to its top. Plant height was measured from the base of the culm to the top of the inflorescence. When a character was present more than once per individual (e.g., tepal length), we measured that with the greatest value [3,6]. Each specimen was treated as an Operational Taxonomic Unit (OTU). Means and standard deviation for 17 quantitative characters were calculated. One-way analysis of variance (ANOVA) followed by Tukey's honestly significant difference (HSD) post hoc test for unequal sample sizes were used. Next, principal component analysis (PCA) based on the correlation matrix was performed for all quantitative characters. The analysis was used to examine the general variation as well as to reduce the set of characters that were correlated the strongest with the principal components. The missing data were replaced by means. Components in this analysis were extracted by a scree test. We selected the characters with the highest factor loadings in the first three components (r > 0.60). Prior to PCA, the measurements were standardised. Statistical analyses were performed using Statistica version 12 [17].

Seed micromorphology
Seed morphology and ultrastructure were studied using scanning electron microscopy (SEM) in the Laboratory of Microscopic Techniques of the Faculty of Biological Sciences, University of Wrocław, Poland. A total of 33 seeds (12 of J. quartinianus, six of J. oxycarpus and 15 of J. fontanesii subsp. pyramidatus) was examined. Seeds were mounted to the microscope stub by double-adhesive tape; coated with carbon and silver particles; and examined using a Tesla BS-300 instrument. Each seed was measured using tpsDig version 2.0 software [18]. The material used for SEM imagery was collected from the selected herbarium specimens from E, ETH, FR and K (S1 Table).

Statistical analyses
The results of the one-way ANOVA showed significant differences among 16 examined characters ( Table 2). The characters accounting for the most morphologic dissimilarity were the ratio between length of capsule and length of perianth; length of filament; ratio between length of anther and length of filament; length of capsule and capsule mucro ( Table 2). The results of Tukey's HSD test are presented in supplemental material (S2 Table). PCA performed based on 18 quantitative characters yielded three components, which accounted for 68.9% of total variation. The first component accounted for 36.16% of total variation, while the second represented 21.86% of total variation and the third 10.92% of total variation. Eleven analysed characters had high positive factor loadings (> 0.60) and five high negative loadings (< -0.60). The characters, which showed the greatest influence in the first component, were plant height (HP); length of anther (AL); length of filament (FL); ratio between length of anther and length of filament (AL/FL); length of capsule (CL); length of capsule mucro (CML); ratio between length of capsule and length of perianth (CL/TP); number of flowers in head (NFH); and number of filaments (NF). Length of outer and inner tepals (OTL and ITL); width of inner tepal (including scarious margin (ITSM)); width of inner tepal without scarious margin (ITEM); and width of capsule (CW) showed the greatest influence in the second component. Length of inflorescence (IL) and number of heads in the inflorescence (NHI) were highly correlated with the third component ( Table 3).
The scatter plot according to the first two factors revealed three neighbouring or slightly overlapping groups of specimens, which corresponded to the three taxa (Fig 1).

Seed micromorphology
The seeds of the three taxa are similar in the size, but they have different seed coat surfaces. The seeds of J. oxycarpus are longitudinally 14-20 striate (lr), with short transverse ridges (st) that do not adhere to each other, short border transverse ridges (sb) usually pronounced, and formed by single thicker ridges (about 2.0-3.5 times that of other transverse ridges (st)). The  However, their short transverse ridges (st) do not adhere or sometimes adhere to one another but, in many cases, not so tightly. Moreover, their short border transverse ridges (sb) (between longitudinal ridges) are usually faint and are formed by single, slightly thicker ridges. The seed coat surface of J. quartinianus is distinctly longitudinally (18)20-24(26) striate (lr), with short transverse ridges (st) (between longitudinal ridges), usually adhering to one another (i.e., they are very tightly arranged). The short border transverse ridges (sb) are usually pronounced (rarely faint) and formed by 2-3 adherent ridges of the same thickness like other transverse ridges (st) or are thinner (Fig 2).

Discussion
The analysis results confirmed that J. quartinianus, J. oxycarpus and J. fontanesii subsp. pyramidatus are differentiated. In the past, this first taxon was treated originally as a synonym of J. fontanesii [13] and later as a synonym of J. oxycarpus [16]. We found in the herbarium material four sheets initially identified as J. fontanesii and then as J. oxycarpus, and the rest of the specimens were wrongly classified as this latter species. Our research indicated the diagnostic value of the characters, which distinguished these three taxa. One of the significant features providing differentiation between J. quartinianus and J. oxycarpus is ratio of the capsule and perianth length. Juncus quartinianus has capsules longer than the perianth or, extremely rarely in some flowers, only slightly longer than the perianth. Juncus oxycarpus capsules are usually equally long or shorter than the perianth or, rarely, are slightly longer than the perianth (Fig 3).
Juncus fontanesii subsp. pyramidatus has capsules longer than the perianth, but J. quartinianus can be distinguished easily from the latter taxon by oblong-ovoid, apiculate or rarely ovoid and acute capsules as well as anthers that are shorter than the filaments (or, exceptionally, equal in size or slightly longer than the filaments), whereas J. fontanesii subsp. pyramidatus produces capsules that are narrowly pyramidal to subprismatic, gradually narrowing into the rostrum, and, generally, the anthers are distinctly longer than the filaments. The main differences among the three taxa are summarised in Table 4.
The most conspicuous difference among the taxa is the seed coat surface. The sculpture of the seeds is a distinct diagnostic character that is useful for distinguishing Juncus taxa [4,19]. The J. quartinianus seed coat surface is distinctly longitudinally (18)20-24(26) striate, with short transverse ridges usually adhering to one another, and the seeds of J. oxycarpus are  longitudinally 14-20 striate, with short transverse ridges that do not adhere to each other. Thus, the seeds of J. quartinianus more resemble the seeds of J. fontanesii subsp. pyramidatus, which also are longitudinally (18)20-24(26) striate. However, the short transverse ridges of the latter seeds do not adhere or sometimes adhere to one another but, in many cases, not so tightly. Moreover, their short border transverse ridges are usually faint and are formed by single slightly thicker ridges. The seeds of J. quartinianus have short transverse ridges that usually adhere to one another, and their short border transverse ridges are usually pronounced and formed by 2-3 adherent ridges of the same thickness, like other transverse ridges, or are thinner (Fig 2, Table 1). Our detailed analysis shows that J. quartinianus should be recognised as a separate species; therefore, we propose to restore it.
Revision of the herbarium material revealed that Richard's [10] description of J. quartinianus is inaccurate, since, according to the author, this species should have a lateral single flower head and inner tepals slightly longer than the outer ones. However, some specimens from the type material (and other plants that belong to the species) have 2-5(6) heads on the top of the stem, and the tepals are equal in size in some flowers. The other distinguishing characters are the capsule shape and the ratio of its length to the perianth. Richard [10] mentioned that capsules of J. quartinianus are pyramidal, apiculate, glabrous and longer than the perianth. Indeed, plants of this species have apiculate or rarely acute fruit that is longer than the perianth; however, they are oblong-ovoid (including type specimens) and are rarely ovoid. Another character added by Richard [10] in the diagnosis is the presence of six stamens, which are 3 times shorter than the tepals, but in the description below the diagnosis, he stated that the stamens are 2 times shorter than the tepals. We found that this is a more correct feature, including for type specimens, which have a ratio of stamen to tepal length of about 1:2 (nevertheless, the proportion of 1:3 is also possible in the species). Moreover, according to Richard's diagnosis [10], perianth segments should be lanceolate, acute, subcastaneous and the inner tepals slightly longer than the outer ones. However, our study revealed that in flowers of the species (including the type specimens), there are six (or, less often, three), four or five stamens; the tepals are broadly lanceolate to ovate-lanceolate, acuminate to acute, equal or the inner longer, and all tepals are greenish-red or reddish to dark red or greenish, usually with a red apex and often with additional red stripes along the margins. The numbers in parentheses are the minimum and maximum values measured for the individual characters.
Representative specimens examined ( Ã specimens measured for statistical analyses).

Distribution and habitat
Eritrea, Ethiopia and Somalia, in wet places on sandy soil, limestone along streams, bogs near water holes, clay soil, seasonally seepage ground, the margins of small ponds, dry river banks, at 1,300-2,500 m a.s.l.

Phenology
The capsules usually mature from October to June. However, ripe fruits were also present on 13