≡ Viola [unranked; “Gruppe”] Stolonosae Kupffer in Oesterr. Bot. Z. 53: 329. 1903.—Lectotype (designated here): Viola palustris L.
=Viola subg. Verbasculum Nieuwl. & Kaczm. in Amer. Midl. Naturalist 3: 213. 1914.—Type: Viola primulifolia L.
=Viola [unranked] (”Gruppe”) Vaginatae W. Becker in Beih. Bot. Centralbl., Abt. 2, 36: 29. 1918 ≡ Viola ser. Vaginatae Taken in J. Sci. N.-E. Norm. Univ., Biol. 1: 86. 1955 ≡ Viola subsect. Vaginatae (W. Becker) P.Y.Fu, Fl. Pl. Herb. Chin. Bor.-Or. 6: 91. 1977 ≡ Viola sect. Vaginatae (W. Becker) Ching J.Wang, Fl. Reipubl. Popularis Sin. 51: 85. 1991.—Type (Shenzhen Code Art. 10.8): Viola vaginata Maxim.
Description.—Rhizome perennial; bulbils absent. Lateral stems present or absent: aboveground stolons, most leaves scattered; or rarely aerial stems with leaves in apical rosette (in V. moupinensis). Stipules free or occasionally up to 1/2 adnate (in V. brachyceras), pale, greenish, or brown, (linear-lanceolate to) lanceolate to ovate, acuminate, entire or remotely denticulate-fimbriate. Lamina lanceolate to reniform, base cuneate to deeply cordate, apex rounded to acuminate, margin subentire to crenate. Corolla white or pale violet. Sepals lanceolate to ovate; appendages short or long (0.5–2 mm), rounded or dentate. Lateral petals bearded or not; bottom petal shorter than, or subequal to, the other petals (6–20 mm), apex acute to emarginate; spur short (1–5 mm) and saccate. Style at apex margined and flattened, rarely bilobate.
Diagnostic characters.—Stolons (if present) with most leaves scattered AND sepals lanceolate to ovate AND stipules usually lanceolate to ovate AND style apex margined and flattened, rarely bilobate.
Ploidy and accepted chromosome counts.—4x, 8x; 2n = 20, 24, 44, 48.
Age.—Crown node age c. 12.7 Ma [45]; stem node age 13.5 (12.2–14.0) Ma [28].
Viola adenothrix Hayata, V. binayensis Okamoto & K. Ueda, V. bissetii Maxim., V. blanda Willd., V. brachyceras Turcz., V. brevipes (M. S. Baker) Marcussen, ined., V. cochranei H. E. Ballard, V. davidii Franch., V. diamantiaca Nakai, V. epipsila Ledeb., V. fargesii H. Boissieu, V. glaucescens Oudem., V. grandisepala W. Becker, V. hultenii W. Becker, V. incognita Brainerd, V. jalapaensis W. Becker, V. javanica W. Becker, V. kjellbergii Melch., V. lanceolata L., V. macloskeyi F. E. Lloyd, V. maoershanensis Y. S. Chen & Q. E. Yang, V. mearnsii Merr., V. minuscula Greene, V. moupinensis Franch., V. nitida Y. S. Chen & Q. E. Yang, V. nuda W. Becker, V. occidentalis (A. Gray) Howell, V. palustris L., V. petelotii W. Becker ex Gagnep., V. pluviae Marcussen, H. E. Ballard & Blaxland, V. primulifolia L., V. principis Boissieu, V. renifolia A. Gray, V. rossii Hemsl., V. shikokiana Makino, V. striatella H. Boissieu, V. suecica Fr., V. thomsonii Oudem., V. vaginata Maxim., V. vittata Greene, V. yazawana Makino
Distribution.—North-temperate; one species (Viola lanceolata) in northern South America. Viola suecica (=V. achyrophora Greene, V. epipsiloides Á. Löve & D. Löve, V. epipsila subsp. repens W. Becker) is circumboreal.
Discussion.—The delimitation of this subsection is “locked” by the existence of allopolyploids between distantly related internal lineages, one of which happens to be the type of the subsection (Viola palustris). The polyploids include the North American V. blanda and V. incognita (8x) which are allopolyploids of V. renifolia or perhaps more likely V. brachyceras (4x) and a taxon within the V. primulifolia group (4x); the Amphiatlantic V. palustris (8x) which is the alloploid of V. minuscula (=V. pallens auct., non (Banks) Brainerd; 4x) and V. epipsila (4x); the Pacific American V. pluviae (8x) which is the alloploid of V. macloskeyi/occidentalis (4x) and V. suecica (4x); and presumably also the North American V. brevipes [45,93]. These five allo-octoploids are no older than 2.5–5 Ma, and their marked boreal distributions suggest they originated in response to the climate cooling and repeated glaciations in the Pleistocene [93].
Disregarding allopolyploidy, at least four informal species groups are nevertheless recognisable at the 4x level based on published phylogenetic studies (Figure 8; [45,82,86,87,287]). These include (1) a clade comprising the Chinese species V. davidii and V. grandisepala; (2) a clade of mostly hairy species occurring in eastern Asia and northern North America comprising V. principis, V. renifolia, V. yazawana, and presumably also V. adenothrix and V. brachyceras; (3) a clade of mostly large species with acuminate laminas and larger pale violet to pink corollas and broad somewhat sheathing denticulate stipules comprising the circumboreal V. epipsila-suecica complex, V. moupinensis, and most of Becker’s [1] grex Vaginatae, i.e., V. bissetii, V. diamantiaca, V. vaginata, etc.; and, finally, (4) the North American stoloniferous species comprising V. primulifolia, V. lanceolata, V. macloskeyi, V. minuscula, etc., by Marcussen et al. [45] referred to as “grex Primulifoliae”.
The group of species having a creeping, remotely noded rhizome and which was previously informally designated as the Palustres grex comprises a subset of the species in clade 3, i.e., V. epipsila and V. suecica, and their allopolyploids, i.e., V. palustris, V. pluviae, and V. brevipes, formed with species in clade 4.
Phylogenetic studies of the north-temperate species of subsect. Stolonosae [45,93] indicate that a narrow species concept coinciding with morphological-geographic units best applies to these taxa. This concept challenges in particular the traditional classification of the North American taxa into a few, broadly defined species based on lamina shape [289,290,291]. Hence, we consider V. minuscula distinct from V. macloskeyi, V. occidentalis distinct from V. primulifolia, V. vittata distinct from V. lanceolata, and V. suecica distinct from V. epipsila. Among the octoploids V. brevipes and V. pluviae are distinct from V. palustris, and V. incognita is distinct from V. blanda. Taxonomy, variation, and phylogeography in this circumboreal complex are poorly understood and require further study throughout its range. Certain characters traditionally considered diagnostic, such as leaf shape and pubescence in V. palustris, have proven variable even within single specimens [292,293].
The chromosome number 2n = 20, apparently at odds with the predominance of 2n = 24 in this subsection, has been reported several times in Viola brachyceras and also in the closely related V. yazawana, for which also 2n = 40 has been reported (cf. [61] and references therein); this number could also explain 2n = 44 (not 48) in the octoploids V. blanda and V. incognita, and possibly also in V. maoershanensis [294]. Counts of 2n = 20 outside of this species group within subsect. Stolonosae are probably errors.