The North American Great Lakes region, defined here as the area within the Great Lakes drainage basin, was completely glaciated during the most recent periods of Pleistocene glaciation (Dorr and Eschman 1970; Williams et al. 1998). In geologic terms, therefore, the contemporary flora of the region is young, having assembled only within the last 10,000 to 15,000 years. Most of the native vascular flora—ca. 2,500 species (Peirson 2010)—consists of species that migrated into the region following glaciation and that inhabit basic vegetation formations within the region. However, approximately 60 endemic vascular plant taxa have been described from the region. Unlike most widespread members of the flora, many of these endemics are restricted to ecologically specialized habitats, often along the shores of the Great Lakes (Peirson 2010).

The three best known Great Lakes shoreline endemics are Iris lacustris Nutt. (dwarf lake iris), Solidago houghtonii Torr. & A. Gray (Houghton’s goldenrod), and Cirsium pitcheri Torr. & A. Gray (Pitcher’s thistle), all of which are federally-listed threatened species. Iris lacustris and S. houghtonii both have geographic distributions that are tightly centered around the Straits of Mackinac in northern Michigan, a general distribution that is shared by a number of Great Lakes endemics. Iris lacustris is restricted to calcareous shores and alvar habitats of the Niagara Escarpment from eastern Wisconsin and northern Michigan to the Bruce Peninsula in Ontario (Guire and Voss 1963; Trick and Fewless 1984), while S. houghtonii is likewise essentially confined to sandy and rocky shores from northern Michigan along the northern edge of Lake Huron to the Bruce Peninsula (Guire and Voss 1963; Morton 1979; Morton and Venn 2000; Laureto and Barkman 2011). The distribution of Cirsium pitcheri, while still encompassing the Straits of Mackinac, is broader than those of the previous two species. Pitcher’s thistle inhabits beaches and active sand dunes from the southern shores of Lake Michigan north through Lake Huron, with several populations along the Lake Superior shoreline as well (Guire and Voss 1963; Voss 1996). Taken together, these three endemics illustrate the characteristic ecogeographic patterns found in a number of Great Lakes shoreline endemics (Peirson 2010).

Another striking Greats Lakes shoreline endemic is Gillman’s goldenrod or dune goldenrod, treated here at the species level as Solidago gillmanii (A. Gray) Steele (following Semple and Peirson 2013). Like Cirsium pitcheri, this species is a characteristic component of sand dune vegetation along the shores of Lake Michigan and northern Lake Huron. Gray (1882) originally described this goldenrod (at the time as S. humilis Pursh var. gillmanii A. Gray), as occurring on “sand hillocks on the shores of Lakes Superior and Michigan.” Voss (1996) likewise included dune goldenrods along the southern shore of Lake Superior in S. gillmanii (at the time as S. simplex Kunth var. gillmanii (A. Gray) Ringius), even using a photograph of a particularly robust Lake Superior plant from Chippewa County, Michigan in his Plate 7F. However, during fieldwork to sample populations of S. gillmanii near Deer Park, Luce County, Michigan, as part of a broader phylogeographic study of the species and other members of Solidago subsect. Humiles (Rydb.) Semple, it became apparent that these Lake Superior dune goldenrods did not fit well within S. gillmanii from the dunes of Lakes Michigan and Huron. While the plants shared the overall form and apparent sand dune adaptations of S. gillmanii, the Lake Superior plants had conspicuous pubescence on both the leaves and the stems, a characteristic not found in S. gillmanii. They also lacked the glutinous (sticky) leaves and capitula (flower heads) of S. gillmanii. It was clear that these dune goldenrods warranted a closer look.

The major objectives of this study were (1) to use field and herbarium studies to examine the Lake Superior dune goldenrod and describe its distribution and ecology; (2) to compare this dune goldenrod to other goldenrods in the Great Lakes region and to assess its taxonomic placement; (3) to use flow cytometry to determine its ploidy; and (4) to use the information thereby collected to place it more broadly in the context of the endemic flora of the region.

Study of herbarium specimens from dune systems along the southern shore of Lake Superior supported the preliminary conclusion based on initial field observations that the goldenrods there differed from Solidago gillmanii and likely did not belong to Solidago subsect. Humiles, to which S. gillmanii belongs. The Lake Superior dune goldenrods shared the overall form and apparent sand dune adaptations of S. gillmanii (e.g., presence of elongate vertical rhizomes that allow survival from sand burial) but differed noticeably in their pubescent stems and foliage. Vegetative pubescence (outside of the floral arrays) is not found in S. gillmanii or the other members of subsect. Humiles. Subsequent fieldwork along the southern shore of Lake Superior confirmed the earlier observations. Plants in these populations were consistently pubescent (rarely only sparsely so) and were also not noticeably glutinous. Sticky leaves, stems, and capitula have been used as defining characteristics of members of subsect. Humiles sensu lato and are readily apparent in S. gillmanii in the field.

The vegetative pubescence and virgate (wand-like) inflorescences of the Lake Superior dune goldenrods suggested a possible relationship with Solidago hispida Muhl. (hairy goldenrod). That species is widespread throughout eastern North America, extending as far west as the Canadian prairie provinces, in a variety of dry, often sandy or rocky habitats (Semple et al. 2017b). The typical pubescent form of S. hispida occurs throughout Michigan. In the Lake Superior region, it occurs along the shore in open, sandy woods, on lakeshore bluffs, and in rock outcrop habitats, but is almost never present on the open dunes (except occasionally at the margins in more stabilized areas). Examination of plants in the field and the herbarium revealed that typical S. hispida seems to lack the elongate vertical rhizomes that would allow for survival from sand burial in open dune habitats. While sharing vegetative pubescence and virgate inflorescences, the Lake Superior dune goldenrods differed morphologically from nearby, typical S. hispida in their larger stature, clump-forming habit of several to many stems, elongate vertical rhizomes, and in their larger capitula (ca 6.5 mm vs ca 4.5 mm long, respectively). These observations supported the idea that these dune goldenrods along the southern shore of Lake Superior represented an undescribed species that was possibly aligned with S. hispida and subsect. Erectae (G. Don) Semple & J. B. Beck.

In some Solidago complexes, ecogeographic separation and/or morphological differences (e.g., the size of the capitula) have also been associated with differences in ploidy level (as discussed in Peirson et al. 2012). Given the observed habitat and morphological differences between the dune goldenrods, hereafter referred to S. psammophila, and nearby S. hispida, this study also examined if there were differences in ploidy as well. DNA ploidy determinations from flow cytometry were obtained for 71 individuals from four populations of S. psammophila and for 26 individuals from three populations of S. hispida from the Lake Superior shore (Tables 1 and 2). Flow cytometry recovered two non-overlapping DNA ploidy groups that correspond to diploid (2x = 18) and tetraploid (4x = 36) individuals (Table 2; Figure 1). All sampled individuals of S. psammophila were found to be tetraploid, whereas individuals of S. hispida were found to be uniformly diploid. The 2x and 4x DNA ploidy groupings from this study are consistent with groupings found in previous flow cytometry studies of other Solidago species (e.g., Halverson et al. 2008; Schlaepfer et al. 2008; Peirson et al. 2012). The diploid determination for S. hispida is also consistent with previous cytological studies of the species (as summarized from Semple and Cook 2006; Semple et al. 2017b). Solidago psammophila would be the second tetraploid in the S. bicolor–S. hispida complex, after the upper Midwest endemic S. sciaphila Steele (following Semple et al. 2017b). Whether S. psammophila formed through chromosome doubling within a single S. hispida-like ancestral species (autopolyploidy) or as a result of hybridization and subsequent chromosome doubling (allopolyploidy) was not examined in this study. The narrowly circumscribed distribution along a short portion of the southern shore of Lake Superior suggests a single origin of S. psammophila.

Given that Gray’s (1882) original concept of Solidago humilis var. gillmanii was mixed and included both the Lake Michigan/Lake Huron dune plants and the Lake Superior S. psammophila, scrutiny of the original description and the type of S. gillmanii was necessary. Gray stated in his description that S. humilis var. gillmanii was “an extreme form of this variable species, with dentate even laciniate leaves and an open compound panicle; growing on sand hillocks on the shores of Lakes Superior and Michigan.” Gray, however, did not cite any collections or designate a type for var. gillmanii. As follow-up to his biosystematic study of the “S. spathulata–S. glutinosa complex,” which included S. gillmanii, Ringius (1987) reviewed the nomenclature of the group and designated a number of lectotypes, including for S. humilis var. gillmanii. Ringius’ choice of lectotype, at first glance, presented a bit of a quandary. He stated that there were three collections from “the south shore of Lake Superior … sent to A. Gray by W. Boott” and identified by Gray as var. gillmanii that could serve as candidates for the lectotype (two 1875 collections (GH, NY) and a single 1879 collection (GH)). Ultimately, Ringius designated the 1879 GH collection as lectotype and stated that it matched “the protologue in having laciniate leaf margins and an open compound panicle.” Examination of the 1875 and 1879 collections, however, revealed that the 1879 collection designated as lectotype by Ringius was not a plant from the Lake Superior shore.

The labels on the 1875 GH collection indicate “s. Shor l. Superior root sent by W. Boott.” Examination of that collection, which consists of two sheets with portions of the same rosette (GH-00274538, GH-00274539), revealed sparse pubescence on the rosette leaves as well as capitula that did not appear glutinous. Similarly, examination of a digital image of the NY collection (NY-02369465) also appears to show pubescence on the rosette leaves and an overall lack of glutinosity. The pubescence on the foliage, the lack of glutinosity, and the label data indicating that the collections originated from the southern shore of Lake Superior confirm that the 1875 collections are Solidago psammophila. The label on the lectotype, the 1879 GH collection (00012486), indicates “Roots from upper Michigan by W. Boott” and “same as 1875.” Examination of the lectotype did not reveal any vegetative pubescence outside of the floral array. In addition, the capitula and leaves within the capitulescence appear to have been glutinous (e.g., they have a slightly varnished/resinous appearance). Morphologically, the 1879 collection is a characteristic, cultivated specimen of S. gillmanii, consistent with plants cultivated from Lake Michigan dune systems at the Matthaei Botanical Gardens during this and previous studies. Ringius (1987) stated that the 1879 GH collection was from the south shore of Lake Superior, presumably inferring that the notation on the label of “same as 1875” indicated that the specimen was from the same plant. Examination of the collections revealed that this cannot be correct, since the 1875 and 1879 collections represent S. psammophila and S. gillmanii, respectively. By “same as 1875” Gray presumably simply meant that he considered the collections to represent the same taxon. In addition, the locality information of “upper Michigan” on the label would indicate that the collection came from Michigan’s Upper Peninsula, not specifically that it came from the Lake Superior shore, and Gray did include “Lake Michigan” in the original description. The lectotype of S. gillmanii originated from the dunes along northern Lake Michigan/Huron in Michigan’s Upper Peninsula. Solidago gillmanii is quite common on the dunes there and absent from the Lake Superior shore, so far as is known.

Solidago psammophila J.A. Peirson, sp. nov.—-TYPE: U.S.A. Michigan: Alger Co., Sect. 7, ca. 6 miles west of Grand Marais, high dunes above Lake Superior, July 26, 1948, McVaugh 9586 (holotype: MICH!, isotypes: MT!, UMBS!).

Perennial herbs from branching vertical rhizomes or caudices. Stems 1 to ca. 15, 35–75 cm, ascending or more commonly erect (occasionally slightly decumbent at base), generally unbranched below the capitulescence, sparsely to densely hispid or short-villous proximally, occasionally appearing glabrate, especially if trichomes have been abraded by blowing sand, moderately to densely hispid to strigose in the capitulescence. Leaves alternate, simple, petiolate or sessile, sometimes stipitate glandular (but not becoming resinous or glutinous), moderately or sparsely short-pubescent to sericeous or more rarely strigose. Basal rosette and proximal stem leaves petiolate, petiole ciliate, blade ovate-oblanceolate to narrowly oblanceolate, tapering to petiole, 4–14 cm long, 0.7–3 cm wide, apex acute to obtuse or less often rounded, margin serrate. Mid and distal stem leaves sessile, lanceolate to linear, 1–4 cm long, 0.2–0.6 cm wide, reduced upward, margin entire or sparsely serrate. Capitulescence narrowly to broadly elongate-paniculiform, 7–25 cm long, 2.5–8.5 cm wide, consisting of short axillary and terminal racemiform clusters, lower branches occasionally elongated in larger plants, branches strigulose; heads few to numerous, not secund. Peduncles 3–10 mm long, strigulose; bracteoles few, linear. Involucres campanulate, 5–8.1 mm long. Phyllaries in 3–4 graduated rows, the outer ones ovate, the inner ones linear-oblong, apex acute to obtuse or rounded, often ciliate or fringed. Ray florets 8–13, strap-shaped, 2–3 mm long, 0.6–0.9 mm wide. Disc florets 8–15, corollas 3.5–5 mm long. Cypselae narrowly obconic, antrorse-strigose, ca. 3 mm long. 2n = 36 (from DNA ploidy determination). (Figure 2).

In addition to Solidago psammophila and S. gillmanii, four other goldenrods have commonly been recognized as endemic to the glaciated North American Great Lakes region. Except for S. hispida var. huronensis Semple, which is diploid, all recognized Great Lakes endemic taxa of Solidago are polyploid (Table 3). Like the majority of the broader endemic flora, these Solidago taxa have distributions centered in the northern parts of the region. They are restricted to regionally rare, non-forested habitats that are linked to present and/or past Great Lakes shorelines (Peirson 2010).

(modified with permission from Voss and Reznicek 2012)