Papers Mich. Acad. Sci. Arts & Letters 5: 77-94 (1925)
THE WILD ROSES OF THE MACKINAC REGION OF MICHIGAN*
EILEEN WHITEHEAD ERLANSON
|* Paper from the
Department of Botany
of the University of Michigan, No. 233.
DURING June and July, 1024, extensive collections of wild roses were made, chiefly in the northern parts of Emmet and Cheboygan counties, Michigan, by C. O. Erlanson. In August the writer joined him, and after a week in the same region we went over the Strait of Mackinac, and made further observations and collections in southeastern Mackinac County during the first half of September. Since then several hundred loaned herbarium specimens have been examined, including good series of wild roses from Green Bay, Wisconsin, from around Duluth, and from the northwest shore of Lake Superior, Ontario. These collections show that many of the numerous forms found in the Mackinac region of Michigan are widespread throughout the northwestern Great Lakes shore; they are, therefore, not of merely local interest.
The northwestern Great Lakes lie in the region of northeastern hardwoods, with scattered areas of northeastern pine forest and of spruce-fir forest. All three types of vegetation are included in the three Michigan counties visited last summer. The jack-pine plains of the south part of Cheboygan County yielded some extremely interesting roses, including Rosa palustriformis Rydberg. A large part of the beech-maple area, since being lumbered and burned over, is an arid waste of aspens where an occasional plant of Rosa blanda Ait. is seen. Roses do not grow in deep shade and are, therefore, absent from the coniferous woods except for a few weak, straggling plants scattered along the margins or in the pathways. Single plants for Rosa acicularis Lindl. were seen now and again in such situations, but they seldom bore any flowers. The rose flora is chiefly characteristic of the sandy beaches and of the beach thickets along the lake shores. It consists mainly of the two frequent species, Rosa blanda Ait. and Rosa acicularis Lindl. Rosa palustris Marsh. (R. carolina L. 1762, not L. 1753) forms occasional large thickets in the swampy flood-plains of streams, on stream banks or in the moist soil of a fosse near a lake shore throughout the region.
No exotic species were observed last summer, but Dodge (13) reported Rosa rubiginosa L. from Mackinac Island and Mackinac County. The University Herbarium contains a specimen of Rosa pimpinellifolia L. (R. spinosissima L. in part) from Emmet County, collected along a roadside west of Mackinaw City by Dr. J. H. Ehlers in 1916. Dr. Ehlers also collected an unusual form on the railway bank north of Pellston, Cheboygan County, in 1920. This specimen is low and herbaceous, with the stem densely bristly, terminating in a corymb of flowers. It evidently belongs to the group of Rosa suffulta Greene and may have been carried in by a railway train. It was hunted for last summer, but without success.
In the field I was immediately struck by the apparently "intergrading forms" between R. acicularis and Rosa Bourgeauiana Crépin, and, to a lesser extent, between R. acicularis and R. blanda. It is probable that hybridization has occurred in the past and is still freely taking place. Occasional groups of plants were observed, evidently clones, which exhibited a large percentage of small green abortive hips in August, a possible indication of hybridization. These hips fall off easily and would be very apt to be missing on herbarium specimens. Many of the anomalous forms of R. blanda which appeared to resemble R. acicularis in certain characters bore full crops of hips crowded with achenes. There is no reason to suppose that many of the natural hybrids are self-sterile. On the contrary the F2 generations from oft repeated crosses between the species of Rosa which occupy the same habitat and flower at the same season are no doubt partly responsible for the series of forms found.
The roses of this part of Michigan show evidence of the occurrence of definite similar sets of micro-species within different species groups, as found in European rose species by Almquist (1), Harrison (16) and others. Crépin drew attention to the phenomenon in 1866 (7), calling it "parallel variation." There is reason to believe that many species groups (collective species) in North America show two series of forms, one with rounded hips and the other with elliptical or tapering hips; another pair of forms commonly occurring has one member with pubescent leaflets and the other with glabrous leaflets. This will be discussed further under the individual species. It may be that the species are subject to definite mutations which produce, in each, forms with glabrous leaflets and sepals, gland-compound teeth, and tapering hips. Mr. C. C. Deam (12) believes from his observations in the field that such characters are largely influenced by the habitat. It is, therefore, of first importance to test this hypothesis experimentally in the garden as has been done with Rubus by Brainerd and Peiterson (3). Probably both hybridization and mutation are taking place in the wild roses around the Great Lakes. I saw some indication of there being a shade form of R. acicularis var. Bourgeauiana Crépin with pubescent, glandless leaflets and simple non-glandular serrations. A specimen, without flowers, probably R. acicularis, collected by Dr. Elders in a Thuya bog near Burt Lake has thin almost glabrous leaflets without glands and with very narrow stipules. The most glandular specimen of R. acicularis, with leaflets resinous beneath, was found at Prentis Bay growing at the side of a lumber road in a fir-cedar stand where it could not receive much sunlight.
Most of these roses in the living state possess glaucous leaves. This is a characteristic difficult to detect in dried specimens.
In this account I have treated anomalous specimens under the species they most nearly resemble. I feel that it is impracticable to do anything else than name such dubious types as varieties of the most similar specific type or, in the case of the most outstanding ones, to give them binomial names. One's guess at a possible mode of origin should hardly become a part of botanical nomenclature. It is necessary that these forms be grown and tested for hybridity and that the suspected cross be made in the gardens to re-create the natural hybrids, before an authoritative nomenelatural statement of hybridity can be made.
At the end of the last century Dr. J. H. Schuette made extensive field-studies of the wild roses along the shore of Green Bay, Wisconsin, and distinguished several distinct varieties of R. blanda and of R. palustris. In his one short published note (23) he says that "separating into good species or distinguishable varieties is the best way of advancing our knowledge of this group." The confusion arising from the ever increasing number of species can be avoided by the adoption of a trinomial system as advocated by Hall and Clements (15). Recognized species are then retained and the forms or segregates are described as varieties, thus keeping relationships clear.
Writing on hybridism in the New Zealand flora, Cockayne (5) remarks that, since "a flora is essentially an instrument for the identification of any plant growing wild in the area with which it deals, hybrids should bear names and be described equally with species." This has been done in several modern European monographs. Rehder (10) is naming the known rose hybrids at the Arnold Arboretum, inserting the parent names at the head of the description. It is unfortunate that he does not adhere, in every case, to the usual convention of placing the name of the seed parent in front of that of the pollen parent. Hooker (17) pointed out long ago that it is very easy to dub a difficult specimen a hybrid. In this country, with all the conveniences of transportation, when a distinct rose specimen is found in a herbarium, an effort should be made to re-discover the form at the original locality so that it may be tested out in the garden.
What is most needed in dealing with polymorphic groups such as Rosa is some idea of the range of variation within the species, either from experimental work or from good sets of complete specimens. This would guard against what Crépin called the “bush-mania.” Intensive local studies are needed but we must also become familiar with the same forms from localities throughout their range to get a correct perspective of their relative importance. Alphonse de Candolle (4) pointed out that "le danger de descriptions faiths sur des éléments détachés de leur ensemble est au maximum quand on choisit pour études les fragments locaux d'un group complique et obscur, comme des Rubus, Rosa, Hieracium, Salix, Mentha, etc." A great weakness of the system of calling a new form "species A x species B" is that there is always the possibility of the original forms A and B being themselves hybrids — in fact studies on pollen-sterility in the genus (Cole, 6) indicate hybridity in the majority of rose species.
THE GROUP OF ROSA ACICULARIS
The majority of the plants of this group observed in the Mackinac region had the leaflets pubescent and glandular-granuliferous beneath, gland-compound teeth and subglobose hips. Plants corresponding to the descriptions of R. acicularis do occur, some without any glandular granules on the leaflets. After many years of work M. Crépin reduced his R. Bourgeauiana to a variety of R. acicularis, and there is no doubt that J. D. Hooker would have done so. I have also been unable to discover any character which gives a reliable specific difference between R. acicularis and R. Bourgeauiana, as diagnosed by Dr. Rydberg in the North American Flora. Even the hip-form gives a good series from long and narrow, through pyriform to subglobose; the two latter forms I have seen occurring on the same specimen. According to Cockayne the very existence of such a series is an evidence of crossing between two species. For one who tries to determine the specimens in this group with Dr. Rydberg's key (20), the situation is complicated by the fact that it is possible to find, within a small area, specimens with the hip pyriform or subglobose, leaflets glandular or non-glandular, teeth gland-compound or simple; these characters being combined in every possible way. Some of these combinations, such as elliptic hips with glandular, gland-serrulate leaflets, are rare in the Mackinac region, whereas others are common. There are also glabrous forms in this group, although these are rare. No completely glabrous-leafed form has been seen from this region, but the amount of pubescence varies widely. Some seedlings from pubescent types of Rosa acicularis at the Botanical Garden have entirely glabrous leaflets, but others are sparsely pubescent in varying degrees. The same plants must be observed at maturity before the observations can be interpreted.
Matthews (18), working on the assumption that hybridization and segregation take place in the European wild rose forms, attempted all analysis of some British species on the basis of a few separate characters used in classification. He considers these as pairs of unit characters, for example, hairiness of leaflet vs. glabrousness, biserration vs. simple teeth, and treats them as though they were simple Mendelian allelomorphs. It is interesting to find that in several cases all the possible combinations of two, three or four pairs of such characters have been found in the field and named. This author also advocates a return to the Linnean conception of species in Rosa, the various combinations of unit characters to be denoted by a purely symbolical method, until we have proved whether or not the elementary species are stable.
I tried out this method with the specimens of the R. acicularis group from the Mackinac region, choosing the characteristics used to separate the forms in Rydberg's key, omitting the presence and absence of glandular granules on the leaflets, since almost all our specimens possess them in varying degrees.
In Table I, R represents a plant possessing a round hip and r an elliptic or pyriform hip; P represents pubescent leaflets and p glabrous leaflets; S represents doubly gland-serrate leaflets and s leaflets with eglandular teeth. It is to be understood that these symbols have no genetical significance, since we know nothing of the way in which these characteristics are inherited. I found that plants corresponding to four of the eight possible combinations grew in the Mackinac region, but that much confusion exists as to their respective names, and that two others have been recognized elsewhere.
COMBINATIONS OF THREE SETS OF CHARACTERISTICS IN THE GROUP OF ROSA ACICULARIS Lindl.
|HIP: round, R; pyriform or elliptic, r
LEAFLETS: pubescent, P; glabrous, p
Gland-compound teeth, S; eglandular teeth, s
|COMBINATIONS||NAME OF FORM|
|1.||PRS*||var. Bourgeauiana Crépin (leaflets glandular beneath)|
|2.||PRs*||var. rotunda, n. var. (leaflets not conspicuously glandular)|
|3.||Prs*||var. Sayiana, nom. nov. (leaflets not conspicuously glandular)|
|4.||PrS*||var. lacorum, n. var. (leaflets conspicuously glandular)|
|5.||prS||}||R. Engelmanni (leaflets conspicuously glandular)|
|*These forms are found
growing in the Mackinac region. Rosa acicularis in its
typical form does not appear in this Table because it is not glandular-granuliferous.
ROSA ACICULARIS LINDL.— Two collections were made of plants with the formula Prs, which had no trace of glandular granules on the leaflets. These were: CHEBOYGAN COUNTY, Douglas Lake, 1924, C. O. Erlanson, No. 88; MACKINAC COUNTY, Scotty Bay, 1924, C. O. and E. W. Erlanson, No. 757 B.
ROSA ACICULARIS var. BOURGEAUIANA Crépin.— Bull. Soc. Bot. Belg., 15: 3f). 1876. — The Formula PRS corresponds to this variety, having subglobose hips, with a very short neck, leaflets with pubescence and glandular granules beneath (varying from sparse to dense), and the teeth doubly gland-serrate. The shape of the leaflets varies greatly, as does their size. In some specimens from the Mackinac region the lateral leaflets are sessile; in others they possess distinct petiolules 1-2 mm. long. The following specimens of this were collected in the Mackinac region: EMMET COUNTY, Big Stone Bay, 1924, C. O. Erlanson, Nos. 325, 551, 572, 576, 585; CHEBOYGAN COUNTY, Black Lake, 1924, C. O. Erlanson, No. 518; MACKINAC COUNTY, Whitefish Point, Prentis Bay, 1924, C. O. and E. W. Erlanson, No. 664 (Botanical Garden, University of Michigan, No. 6005); Scotty Bay, 1924, C. O. and E. W. Erlanson, Nos. 757 A, 772.
A common form closely related to R. acicularis var. Bourgeauiana possesses almost eglandular leaflets (glandular granules if present being small and sessile) with simple non-glandular teeth. If specimens are examined in flower, when the final shape of the hypanthium is difficult to judge, there is an even chance that they will be classified under either R. acicularis or var. Bourgeauiana. In Table I this form is represented by the formula PRs, which I have named var. rotunda.
R. ACICULARIS var. rotunda, n. var.— Low bush 2.5-7 dm. high. Stem usually densely bristly with straight unequal bristles, branches also bristly. Stipules adnate, 1-4 min. wide and 1-2 cm. long, with acute spreading auricles, either glabrous, glandular-granuliferous or finely pubescent on the back, varying on the same bush. Petiole and rhachis glabrous or pubescent, usually gland-hispid and often with a few setae. Leaflets mostly 7, sometimes 5, often 9 on the turions, varying from elliptic and acute at both ends (e.g. 1.9 cm. long x 0.7 cm. wide), to oval and obtuse at both ends (e.g. 4.5 cm. long x 3 cm. wide), 1-5.3 cm. long and 0.6-2.8 cm. wide; bright green and glabrous, or nearly so above, paler, glaucous, densely pubescent to villous beneath, glandular granules, if present, small and not noticeable to the naked eye nor to the touch. Teeth simple, ciliate and eglandular, acute to ovate, ascending but convex on both margins. Flowers solitary to 3 together. Bracts conspicuous, ovate, glabrous or pubescent on the back with margins glandular-ciliate. Peduncles glabrous, rarely with a few weak hispid glands, from 1-2.3 cm. long, mostly 1-1.5 cm. long, reflexed in fruit. Hypanthium glabrous, subglobose or round elliptic, rounded at the base, neck none or very short, 1.1-1.4 cm. in diameter, 1.2-1.5 cm. High. Sepals lanceolate, caudate-attenuate, 1-1.3 cm. long, tips 0.5-2 cm. long usually about 1 cm, long, glabrous or short-pubescent on the back, usually with a few hispid glands, erect and persistent in fruit. Petals rose pink, 2-2.5 cm. long. The following specimens were collected in the Mackinac region of Michigan: CHEBOYGAN COUNTY, North Fishtail Bay, Douglas Lake, 1924, C. O. and E. W. Erlanson, No. 628 (Bot. Gard., Univ. of Mich., No. 5892), TYPE; jack-pine plains, 1924, C. O. Erlanson, Nos. 444, 405, 407; EMMET COUNTY, Big Stone Bay, 1924, C. O. Erlanson, Nos. 572 A, 575; MACKINAC COUNTY, Prentis Bay, 1920, J. H. Ehlers, No. 1401; 1924, C. O. and E. W. Erlanson, Nos. 664 A; Scotty Bay, 1924, C. O. and E. W. Erlanson, No. 757 (Bot. Gard., Univ. of Mich., No. 600G); Bois Blanc Island, 1914, C. K. Dodge. (This variety often grows in close association with var. Bourgeauiana.)
[ROSA ACICULARIOIDES Schuette.— pecimens of R. acicularis var. rotunda, as well as those of R. acicularis var. Bougeauiana with leaflets more pubescent than glandular beneath, will run down to R. acicularioides by the key in the North American Flora, but they differ from the description in not having corymbose flowers. I have seen Schuette's original specimens of this form, now in the herbarium of the Field Columbian Museum, and I am convinced that this is a form of R. blanda possessing leaflets with glandular teeth. Notes by Schuette on one of his sheets of R. acicularioides state that similar specimens sent by him to Crépin and Best were placed by them in the group of R. blanda. The habit and leaflets of Schuette's specimens are not at all like those of R. acicularis. The armature of the stem is similar to that of R. blanda var. hispida, and the ripe hips are smaller and more opaque than in the forms of R. acicularis.]
|1 Rehder in establishing the name R. acicularis var. Sayi does not cite R. Sayi Schweinitz as a synonym. If he did, it would be technically justifiable to use Rehder's combination, even though it were impossible to place under it the round-hipped plants which he describes. It would likewise be in order to disregard the synonym which he does cite: R. acicularis var. Bourgeauiana. Since he does not cite Schweinitz's name, the only solution of the difficulty seems to be to make a new varietal name. Rehder's varietal name has to be interpreted by the description and by the synonymy he does give, rather than by the implied synonymy. It is here treated not as a synonym of R. Sayi Schwein., but rather of R. Bourgeauiana.|
ROSA ACICULARIS var. Sayiana, nom. nov.—R. Sayi Schweinitz in Keating's Narr. Exped. Long., 2: 388, 1824, not R. acicularis var. Sayi Rehder,1 Cycl. Am. Hort., 1555, 1902. — This variety may not be distinct from R. acicularis Lindl., but I have used it to cover any plants having the combination Prs, ellipsoid hips, pubescent leaflets with non-glandular, usually simple, teeth, also possessing at least some glandular granules on the under side of the leaflets. The leaflets of R. acicularis in the Old World are said to be "always eglandular beneath" (Crépin, 10). We have the following specimens from the Mackinac region of Michigan: CHEBOYGAN COUNTY, Douglas Lake and vicinity, 1924, C. O. Erlanson, Nos. 92, 98, 353, 355; EMMET COUNTY, Cecil Bay, 1922, J. H. Elders, No. 2160; 1023, No. 2604; C. O. Erlanson, 1924, No. 311.
ROSA ENGELMANNI S. Wats.—No specimens belonging to the R. acicularis group were found in the Mackinac region corresponding to the combination prS (Table I). Dr. Rydberg (22) states that the type of R. Engelmanni has leaflets "without pubescence, but distinctly glandular-granuliferous beneath, usually double-toothed and with gland-tipped teeth." He also insists that it is a "purely Rocky Mountain species." Several plants in the Mackinac region were observed having elliptic hips, and the leaflets pubescent and densely glandular beneath, with distinctly doubly gland-serrate teeth (9). This form I have named var. lacorum.
R. ACICULARIS var. lacorum, n. var.—R. Engelmanni S. Wats. Card. & For., 2: 376, 1889, in part but not as to type. R. acicularis var. Engelmanni Crépin; L. H. Bailey, Cycl. Am. Hort., 1555, 1902, — Low bush 3-8 dm. high. Stem densely bristly; scattered bristles on the branches. Stipules adnate, 1-5 mm. wide, 1-2.3 cm. long, with ovate, spreading auricles, usually glabrous and glandular-granuliferous on the back, the margins densely glandular-ciliate. Petioles and rhachis glandular with stipitate glands, glabrous or pubescent, rarely with a few setae. Leaflets 5-7, varying from elliptic and acute at both ends (e.g. 3.5 cm. long x 1.5 cm. wide.) to oval and obtuse (e.g. 2.6 cm. long x 2.1 cm. wide), in length from 1 to 4.5 cm., in width from 0.9 to 2.3 cm., dark green and glabrous above, paler, glaucous, sparingly to densely pubescent beneath, with numerous conspicuous stipitate, resinous glands, doubly serrate with glandular teeth. Flowers solitary to 3 together. Bracts, if present, conspicuous, ovate, glabrous and glandular-granuliferous on the back, gland-ciliate on the margin. Peduncles glabrous, or rarely with a few hispid glands, from 1 to 2.5 cm. long, generally reflexed in fruit. Hypanthium glabrous, pyriform to oblong or elliptic, usually tapering at the base and with a distinct neck, in fruit 0.8-1.4 cm. in diameter and 1.3-2.1 cm. high. Sepals lanceolate, caudate-attenuate, 0.8-1.2 cm. long, tips 0.5-1 cm. long, usually pubescent and gland-hispid on the back, but sometimes glandless or glabrous, erect and persistent in fruit. The following specimens were collected in the Mackinac region of Michigan: MACKINAC COUNTY, Scotty Bay, 1924, C. O. and E. W. Erlanson, No. 761, (Bot. Gard., Univ. of Mich., No. 6007) (TYPE); Prentis Bay, 1924, E. W. Erlanson, No. 153, (Bot. Gard., No. 6008); Mackinaw City, 1922, H. H. Bartlett, (Bot. Gard., Univ. of Mich., No. 2592); Bois Blanc Island, 1913, C. K. Dodge; EMMET COUNTY, Cecil Bay, 1916, J. H. Elders, No. 222; Big Stone Bay, 1921, J. H. Elders, No. 1675; 1924, C. O. Erlanson, Nos. 555, 576 A; CHEBOYGAN COUNTY, Douglas Lake, 1924, C. O. Erlanson, No. 88 A.
These five forms seem to be widespread in the northern central states and Canada, and, although they intergrade to some extent, they are well marked and should be looked for in the field throughout the wide range of R. acicularis. This is a circumpolar species and is apparently more frequent in North America than in Europe. In the examination of herbarium material for these forms, it is always difficult, in flowering specimens, to judge from the hypanthium in the flower the shape of the ripe hips. It would be of great value if local collectors would tag a few rose plants with numbered labels and would collect the flowers, leafy branches, ripe fruit and turions from each individual plant for distribution to the herbaria of the country.
There has always been difficulty in distinguishing some of the bristly forms of R. blanda from R. acicularis, although the ripe hips of the latter are bigger than in R. blanda and possess in life beautiful pellucid flesh. The direction and general shape of the serrations of the leaflets are apparently good characters for separating the forms of these two species, and were stressed by Crépin (10). The serrations in the R. acicularis forms are coarse, open, directed outwards, with both sides of a tooth convex; in R. blanda the serrations are directed forward, so that the outer side of each tooth tends to be convex and the inner concave. Dr. Almquist lays great stress on the characters of the leaf serrations in Rosa and says that glands are unreliable as diagnostic characters. It is important, also, to notice that in the forms of R. acicularis the lateral branches bearing flowers are short, under 10 cm. long, while in R. blanda they are almost always over 10 cm. long.
The R. acicularis forms are the first of the Mackinac region roses to come into flower. In 1924 they began to bloom about June 20 and flowering was over after the first week in July. The flowers are large and showy, with petals often 3 cm. long. By August 1, the hips were full size but green. They were ripe by mid-August.
THE GROUP OF ROSA BLANDA AIT.
R. blanda is a common species, especially along beach thickets and sandy shores. It varies a great deal, apparently with the habitat. The leaflets of all the specimens collected in the Mackinac region were found to have at least some short scattered hairs on the upper surface, as well as beneath. Plants growing in the dry sandy soil of the aspen formation and the jack-pine plains tend to be unarmed, with slender stems and small leaflets varying from pubescent to glabrate. No completely glabrous form was found. In the beach thickets and on sandy lake shores, R. blanda tends to produce thick shoots, often bristly, with leaflets densely pubescent on both sides. Sometimes the underground stems send up shoots almost at the water's edge, which flower when only 1.2 dm. high and often have 9 leaflets on most of the leaves. Glandless forms often occur, either with no glands at all, or with glands on the stipules, the sepals being glabrous.
A form of R. blanda with white flowers was collected on Douglas Lake by C. O. Erlanson, No. 396. This is R. blanda var. alba Schuette in herb. (TYPE: Allonez, Brown Co. Wisconsin, J. H. Schuette, June 16, 1895. Herb. Field Columb. Mus., sheet No. 379,239.)
A form collected Sept. 7, 1924 (Erlanson, No. 767) on the shore of Scotty Bay, Mackinac County, bore many hips, all solitary and only 7-8 mm. broad, spherical and crowned by gland-hispid erect sepals, with bases as large as the hip. Specimens of this are now at the University of Michigan Botanical Gardens (accession No. 6004).
Plants of R. blanda bearing oblong hips, tapering at one or both ends, corresponding to R. blanda glandulosa Schuette, were collected at several stations in Chehoygan County (Bot. Gard., Univ. of Mich., No. 5889). The distribution of glands on the rhachis, stipules, bracts and sepals varies a great deal, often on the same individual. This form has been found near Lake Michigan in Indiana by Deam. I do not think it is necessarily a hybrid with R. acicularis, as Dr. Rydberg believes. Some plants were found the last week in August in a shady thicket on Douglas Lake (Erlanson, No. 627) with scattered bristles on the main stem, bearing several half-ripe subglobose hips and many abortive ones which fell off easily. This I suspected to be a hybrid, but the fruit sterility might be due to the shaded situation. This also is being cultivated at Ann Arbor (Garden accession No. 5888).
Many bristly forms of R. blanda are common near the shore in Mackinac County, corresponding to R. blanda var. hispida Farwell (14) (Erlanson, No. 593) and also to R. blanda var. subgeminata Schuette (Erlanson, No. 665), with "twin-prickles intermittent and irregular."
An interesting and puzzling form grows at the edge of a small Ammophila dune on the north shore of Douglas Lake. (Erlanson, No. 637.) It is about a meter high with the habit of R. blanda and bears terminal corymbs of globose glabrous hips by the middle of August. The stem is strongly armed with stout curved prickles, only occasionally paired, while the shape and serration of the leaflets are much as in R. palustris. It approaches some of the specimens of R. carolina aculeata Schuette and might possibly be a hybrid between R. blanda and R. palustris, although the former begins to flower about July 1 and ceases to do so by the end of July, while the latter seldom begins to flower until the first week in August, or later. Schuette particularly noticed that his var. aculeata flowered from three to four weeks earlier than typical R. palustris.
ROSA PALUSTRIFORMIS RYDBERG
Specimens in flower corresponding to the description of this Species and resembling some of Schuette's specimens were collected July 20, by C. O. Erlanson, No. 456, on the jack-pine plains five miles south of the town of Indian River, Cheboygan County. Two or three plants were seen growing in slightly shaded dry sandy soil.
THE GROUP OF ROSA PALUSTRIS
ROSA PALUSTRIS MARSH. (R. carolina of authors) occurs throughout the region, but is much less frequent than R. blanda and the forms of R. acicularis. It seems to need fairly rich moist soil, a sheltered situation and adequate sunlight. It was collected from several stations in Cheboygan and Mackinac counties. The plants were nearly all four or five feet high, forming dense but small thickets.
A specimen collected in flower by C. O. Erlanson at the edge of a bog in Cheboygan County was completely unarmed except for a few bristles at the base of the main stems; its leaflets were short-pubescent above as well as densely fine-puberulent beneath. It is probably R. palustris var. inermis (Schuette), n. comb. (Schuette used the varietal name in his herbarium under Rosa carolina.)
The leaflets on all the specimens observed were only 5 or 7 to the leaf and were mostly small for the species. Two plants from Cheboygan County (Nos. 535 and 544) had leaflets varying from 0.5 to 3.5 cm. only. All our specimens have the lower surface of the leaflets densely short-pubescent all over, except in one case. Many have some pubescence on the upper side also. The specimens collected by Schuette around Green Bay resemble our specimens in leaf characters. Dr. Rydberg has suggested that there may be a northern race of R. palustris with smaller densely pubescent leaflets (21).
R. palustris is the last rose to bloom, in the Mackinac region. It begins to flower during the last few days of July and continues throughout August. The flowers are large and showy, with petals usually deep rose and often 3 cm. long.
In this part of Michigan R. palustris flowers late while R. blanda is early. Further south the former flowers earlier and often overlaps the flowering period of R. blanda along the southern parts of the latter's range. This reverses the state of affairs mentioned by Dr. Harrison (16) where the late spring in Scotland brings the flowering periods of R. mollis and R. pimpinellifolia together, giving ample chances for hybridization. Unless there are certain early-flowering races of R. palustris in the northern part of its range, as Schuette's specimens seem to show, there is not much chance for hybrids of R. palustris and R. blanda to occur in the Mackinac region.
Dr. Almquist has found that the most important characters to use in classifying roses are the color and consistency of the leaves, the form of the leaflets, and the shape and direction of the teeth. The latter seems to be a reliable character in separating R. palustris from all the related forms, as was pointed out by Crépin (8).
Counts made of the number of teeth per leaflet in rose specimens from the Mackinac region showed that in R. blanda they vary from 15 to 40, the mean being 29, In R. palustris the number of teeth per leaflet varied from 25 to 70, the mean being 52. The leaf serrations of the anomalous prickly form of R. blanda (No. 637) from Douglas Lake numbered from 30 to 55 per leaflet, the mean being 42.
ROSA CAROLINA L., 1753
Only one group of plants doubtfully referred to this species was found. These are on a damp wooded hank at the edge of Douglas Lake on North Fishtail Bay. The plants are about a meter high, unarmed except for straight, slender, paired, infrastipular prickles on the branches. On August 18 they bore small, green, subglobose hips, 8-10 mm. in diameter, sparingly gland-hispid, with the sepals spreading. Leaflets 5 to 7, 2 to 5 cm. long, with short scattered pubescence beneath (Bot. Gard., Univ. of Mich., No. 5891).
This colony is not typical R. carolina, but resembles some of the forms of this species collected by Schuette in Wisconsin.
THE GROUP OF ROSA SUFFULTA
The specimens collected July, 1920, near Pellston by Dr. Ehlers, seem to be a form of R. suffulta Greene, with densely glandular rhachis and bracts. The specimens are in flower and possess an elongated hypanthium as in the type known as R. Bushii. R. suffulta has not been observed again in the region and may have been a chance introduction. Some member of the same group was reported as R. arkansana by Beal (2) as having been collected at Harbor Springs, Emmet County, by C. F. Wheeler, but I have not seen his specimens.
KEY TO THE ROSES FOUND OR TO BE EXPECTED IN THE MACKINAC REGION
|Achenes parietal, hips usually glabrous, sepals persistent in fruit.|
|No paired infra-stipular prickles|
|Hips subglobose||R. blanda|
|Hips pyriform||R. blanda var. glandulosa|
|Flowers in corymbs; flowering laterals 10 cm. long or more|
|Leaflets simple-toothed||R. blanda var. hispida|
|Leaflets glandular double-toothed||R. acicularioides|
|Flowers solitary to 3 together; flowering laterals under 10 cm. long|
|Hips pyriform or elliptic|
|Leaflets pubescent beneath, teeth simple|
|Leaflets not glandular-granuliferous beneath||R. acicularis|
|Leaflets glandular-granuliferous beneath||R. acicularis var. Sayiana|
|Leaflets pubescent and resinous-glandular beneath; teeth gland-compound||R. acicularis var. lacorum|
|Leaflets pubescent beneath, teeth simple||R. acicularis var. rotunda|
|Leaflets pubescent and glandular beneath, teeth gland-compound||R. acicularis var. Bourgeauiana|
|Paired infra-stipular prickles on stem and branches||R. palustriformis|
|Achenes basal, or mainly so in R. palustris, hypanthium and pedicels gland-hispid, sepals deciduous in fruit|
|Paired infrastipular prickles straight||R. carolina|
|Paired infrastipular prickles curved||R. palustris|
|Plant unarmed||R. palustris var. inermis|
Specimens of all the roses of the Mackinac region cited by the writer have boon deposited in the Herbarium of the University of Michigan. In conclusion I wish to thank Professor H. H. Bartlett for his helpful suggestions and encouragement during this work.
UNIVERSITY OF MICHIGAN