Rubus villosus Aiton, 1789

Van de Beek, Abraham & Widrlechner, Mark P., 2021, North American species of Rubus L. (Rosaceae) described from European botanical gardens (1789 - 1823), Adansonia (3) 43 (8), pp. 1789-1823 : 69-72

Rubus villosus Aiton

In Hortus Kewensis or a Catalogue of Plants Cultivated in the Royal Botanic Garden at Kew, 2: 210 ( Aiton 1789).

— Lectotype designated by Bailey (1923: 166): BM ( BM 000583294 ) ( Fig. 1 ).

R. leviculus L.H. Bailey, Gentes Herbarum 5: 390 ( Bailey 1943).

— Holotype: Clemson College , Pickens County, South Carolina, 7.VI.1934, M.A. Rice 536, BH ( Fig. 2 ).

R. missus L.H. Bailey, Gentes Herbarum 5: 344 ( Bailey 1943).

— Holotype: near Clarksville, Mecklenburg County, Virginia, 11.V.1938, L.H. Bailey 397, BH.

CONTEXT

In 1789, Aiton published a new Rubus species which grew in Kew Gardens under the name Rubus villosus ( Aiton 1789: 210) . Actually, it was Solander who originally catalogued and gave the name to this Rubus and Aiton who published it ( Blanchard 1911: 431; Stafleu & Cowan 1976: 25). A specimen of it has been conserved in the herbarium of the British Museum. Bailey (1923: 166) selected it as the type and provided drawings of it in Bailey (1898: 372) and a photo in Bailey (1923: 166). It is presently conserved asBM000583294. So, fortunately there is a type specimen to help to clarify its identity. Nevertheless, the history of this name and its various interpretations over time are complex.

J. F. GMELIN

Soon after Aiton’s publication of R. villosus, Gmelin (1791) noted that the epithet was already used by Thunberg (in Murray 1784: 475) for a species from Japan. Therein, Thunberg speculated whether R. villosus Thunb. might be identical with R. corchorifolius L. f. ( Linnaeus 1781: 263), but this does not make his name invalid (ICN, Turland et al. 2018: art. 36.1, last sentence), even now that it is generally accepted that they are indeed identical. Consequently, R. villosus Aiton is a later homonym and thus illegitimate. Realizing this, Gmelin proposed a new name for R. villosus Aiton : R. serratus J.F. Gmel. ( Gmelin 1791: 856) . Though this publication was well known, it did not play a significant role in the various debates about R. villosus Aiton. Bailey (1944: 521) even stated that the identity of R. villosus was irrelevant because it is an illegitimate name, while failing to account for its legitimate replacement. In the index of the volume of the 13TH edition of the Systema Naturae, Gmelin (1792: 1616) failed to mention the name R. serratus , but instead used R. erectus . It is unclear why Gmelin did so; this created a new, illegitimate synonym for R. serratus if it was not a mere error.

There also exists a later homonym of R. serratus J.F. Gmel. : R. serratus Boul. & Letendre in Boulay (1873: no. 5), which is a synonym of R. winteri (P.J. Müll. ex Focke) Foerster (1878: 100) .

PORTER & BAILEY

Nineteenth-century authors generally used the name R. villosus Aiton for the highbush blackberry common in the eastern parts of North America.They neglected both Thunberg’s earlier homonym and also the legitimate name, R. serratus J.F. Gmel.

Near the close of the 19TH century, Porter (1890: 15) made new investigations into this taxon and concluded that it consisted of two distinct varieties: alongside the lowland var. villosus , he acknowledged a var. montanus from mountainous regions. He elevated the latter to the species level some years later as R. montanus (Porter) Porter (1894: 120) . When he discovered that the epithet montanus had already been used for a European species, he renamed his species to R. allegheniensis ( Porter 1896: 153) .

Bailey started his critical work on Rubus in earnest only a few years later. Soon he discovered that R. villosus was not identical with the taxon for which the name had widely been used, the highbush blackberry ( Bailey 1898: 366, 367). Therefore, the latter was in need of a new name, and Bailey chose for it R. nigrobaccus L.H. Bailey (1898: 379 ; type: see below under R. sativus [L.H. Bailey] Clark). After further research, he concluded that Porter’s (1896) division of R. villosus into two species was not tenable ( Bailey 1923: 185). The mountain form, R. allegheniensis , was only a habitat modification of the lowland form, R. nigrobaccus (heterotypic synonym of R. villosus auct. non Aiton). Joining both taxa to one species, he treated them from that point forward as R. allegheniensis Porter , as it was the oldest legitimate name in his view.

It is this name that came into common use in the 20TH century up until now for one of the most common brambles of the eastern and central United States and eastern Canada. Bailey (1925: 288) indicated a provisional type: “ Rubus villosus, Ait. , var. montanus , mihi Pocono Summit. Monroe Co, Penns. Aug. 15/89” NY ( NYBG 5990 ) (iso-, PH[ PH 00040760]). Later , this type was formally confirmed byBailey (1944: 522).

THE IDENTITY OF R. VILLOSUS AITON

Though it was clear to Bailey that R. villosus Aiton was not identical with R. allegheniensis , it was not easy to connect it to any of the known North American species of Rubus . Bailey (1898: 339) initially thought it was identical with R. flagellaris Willd. (at that time, incorrectly called R. canadensis L., until Bailey’s discovery, after checking the type of R. canadensis , that R. canadensis is a totally different species; see Bailey 1943: 243; 1944: 473-475). Brainerd (1900: 25, 27) followed him in this initial interpretation, and Blanchard was also initially of this opinion (1906a: 148). Rydberg (1913: 472; 1915: 149) thought R. villosus to be identical with R. plicatifolius Blanchard (1906a: 149) , which Bailey later (1923: 167) supported with some hesitation. Later, Bailey (1944: 521) again supposed this identification might be right but was uncertain.

After our examination of Bailey’s lectotype of R. villosus Aiton , nine characters turned out to be critical for its identification:

a. The cane prickles are very weak.

b. The stem looks like a trailer, as Bailey (1898: 373; 1944: 521) already noted.

c. The stems are only 2 mm wide, even the mature floricane.

d. The leaves are very small; the largest terminal leaflet is only 3.7 × 2.4 cm.

e. The leaves are broader than it seems at first sight. The margins are folded so that the bases look cuneate. The original shape of the best terminal leaflet was restored by projecting the folded margin outward. By doing so, an almost elliptical leaflet appears ( Fig. 1 ).

f. The leaves are slightly hairy adaxially and rather hairy abaxially, though less than might seem from the unexpanded leaves at the tip of the primocane.

g. The inflorescence has only one flower. Of course, this could be due to unfavorable growing conditions or collecting an atypical inflorescence, but it should be accounted for in combination with the other characteristics.

h. The pedicels have some short, stipitate glands.

i. The ovaries are provided with hairs like the rest of the plant.

Mr.Fred Rumsey(BM)kindly checked the last two characteristics.

The combination of a-d excludes species of the highbush blackberries of the Canadenses (L.H. Bailey) L.H. Bailey, Arguti (Rydb.) L.H. Bailey, and Alleghenienses (L.H.Bailey)L.H. Bailey. Taken together, they point explicitly to the Procumbentes (Rydb.) L.H. Bailey.The single flower clearly supports this conclusion as well. So the question can be narrowed to which members of the Procumbentes have pedicels bearing stipitate glands, canes with small, sparse prickles, and coarsely serrate leaflets bearing soft abaxial hairs? A search through keys in Davis et al. (1968b) and Widrlechner &Riley (2017) and a review of Bailey’s treatments of potential candidates led us to the two taxa that seemed most likely:

a. R. invisus (L.H. Bailey) Britton , of which the serrature is very similar to the type of v illosus, and

b. R. leviculus L.H. Bailey (1943: 390) which s eems to correspond better, at least if R. missus L.H. Bailey (1943: 344) , with its 5-nate leaves, is included within this species ( Widrlechner 1998: 436, 439). It also bears few to many short stipitate glands on its pedicels much like R. villosus .

We should note that although R. plicatifolius was proposed as a candidate by Blanchard according to Bailey (1944: 521), by Rydberg (1913: 472; 1915: 156), and by Bailey himself ( Bailey 1923: 167; 1925: 244), it lacks stipitate glands and soft hairs on the underside of its leaves, which have an obviously plicate appearance, much like the European taxon, R. plicatus Weihe & Nees. The type of R. plicatifolius is ( Widrlechner 1998: 440, lectotype):Wells Beach Depot, York County, Maine, 14.VIII.1905, W.H. Blanchard 477, BH.

It should be noted here that Britton (1893) also cited R. villosus var. humifusus Torr. & Gray (1840: 455) , but he borrowed that epithet from Bailey to whom he explicitly and extensively referred. Bailey did not cite R. villosus var. humifusus . Thus, the type must be selected from Bailey’s specimens.

A new check of the type and other specimens of R. missus confirmed its identity with R. leviculus . The 5-foliolate leaves of the type are coincidental; other collections of R. missus display 3-foliolate leaves, and R. leviculus can also occasionally bear 5-foliolate leaves.

Based on Widrlechner (2013) and an examination of more than 80 specimens of these two taxa, key differences between R. invisus and R. leviculus for eight traits are listed in Table 1. In addition, we measured two additional characters from 58 specimens bearing 1-flowered inflorescences: the length of the shortest 1-flowered inflorescence and the cane diameter at the corresponding node ( Table 1). Based on the data presented in Table 1, the type of R. villosus at BM corresponds best to R. leviculus with few of the characteristic details of R. invisus . We conclude that it is identical with R. leviculus .

Insupportofthisfinding,wediscoveredaspecimen(G00653637) in the De Candolle herbarium of the Geneva Botanical Garden (G-DC)with the label:“ Rubus villosus .Sol./Amer.septentr./ in herb. folia quin./ Kew” ( Fig. 3 ), representing a sample that was sent to De Candolle from Kew. It still has Solander’s name as the author. It is a trailer with ternate leaves, which has no flowers, but the label also mentions 5-foliolate leaves. It is a typical example of R. leviculus .

NINETEENTH- CENTURY HOMONYMS

After Aiton, the name R.villosus was twice applied improperly to other taxa during the 19TH century. The first instance was R. villosus Lasch (1833: 297) . The specimen in Lasch’s herbarium is R. umbrosus (Weihe) Arrh. ( Arrhenius 1840: 94) , but because he, in the addenda to his original publication ( Lasch 1833: 315), stated that R. radula Weihe (1824: 152) is identical, his name is a nomen superfluum for R. radula .

The other example was R. villosus Mérat , which Mérat (1843: 440) applied to a species that Vaillant (1727: 174) described with the phrase name ‘ Rubus montanus repens, sarmentis rotundis, spinis minutissimis munitis, foliis rotundis, utrinque lanatis, superne cinereis, inferne candicantibus, flore albo, fructu nigro, parvo’.Vaillant in turn borrowed this description from Micheli in Tilli (1723: 149).

The validating description is by Micheli, and the type must be selected from his specimens.Usually original specimens for such old names are no longer extant, but Mrs.Chiara Nepi(Botanical SectionNaturalHistoryMuseumUniversityofFlorence,FI)found, after a long search, a specimen with its label corresponding to the protologue.It consists of two inflorescences and some primocane leaves, though without a piece of the stem itself,mounted on three sheets. It is wonderful to find such an old specimen consisting of more parts and including primocane leaves.

From the description, one could get the impression that it describes R. aetnicus Cupani exWeston (1770: 258).The specimen in Florence, however, shows that it is a hybrid of that species with R. ulmifolius Schott (1818: 42) . The petiole is not furrowed; the inflorescence is wider than normal; and the leaves are too large for R. aetnicus . Field research in middle Italy conducted by the senior author (cf. Beek 2014: 178) over several years showed that hybrids of R. aetnicus and R. ulmifolius are abundant in that region in great variation, and the sample fits into that swarm.

Micheli’s plant must be considered part of that group and, by consequence, its correct name is R. collinus DC ( De Candolle 1815: 545) . Unfortunately, due to the late validation of Micheli’s description, this name is both a later homonym and a later synonym. Nevertheless, it was validly published, and, thus, Micheli’s plant is the type of R. villosus Mérat :

R.villosus Mérat , lectotype (here designated): Micheli nr.32 “Rovo del Monte di S. Giusto/di foglie utrinque lanate”, FI[FI050742, FI050743, FI050744], 3 sheets ( Fig. 4 A-C).

N.B.: None of these 19TH-century homonyms has been in common use.

Figures

FIG. 1. — Leaves of type of Rubus villosus Aiton unfolded (editing Abraham van de Beek). Scale bar: 1 cm.

FIG. 2. — Rubus leviculus L.H. Bailey type (M.A. Rice 536, BH). Scale bar: 2 cm.

FIG. 3. — Rubus villosus Aiton in G, G-DC(G00653637). Scale bar: 2 cm.

FIG. 4A. — Rubus villosus Mérat type (1), Micheli nr. 32 (FI[FI050742]).