Cytogeography of the Solidago rugosa Mill. Complex (Asteraceae: Astereae) in Eastern North America

: Chromosome numbers are reported here for the ﬁrst time from 117 individuals of Solidago rugosa and S. ﬁstulosa . Including 178 previously published reports for the two species plus S. latissimifolia , chromosome numbers have been determined from 295 individuals from 269 locations. Only diploids (2 n = 18) were found throughout the range of S. ﬁstulosa on the coastal plain in the eastern U.S.A. (44 counts). Diploids (2 n = 18) were found in the northern portion of the range of S. latissimifolia , and tetraploids (2 n = 36) and hexaploids (2 n = 54) were found in the central and southern portions of the range (nine counts in total). Diploids (2 n = 18) were found throughout the range of S. rugosa in much of eastern North America in four of the ﬁve varieties (northern var. rugosa , var. sphagnophila ; southern var. aspera and var. celtidifolia ). Tetraploids (2 n = 36) were found in all four of these varieties and exclusively in var. cronquistiana in the southern high Appalachian Mountains. Hexaploids (2 n = 54) were found in var. sphagnophila at scattered locations. One possible hexaploid in var. rugosa was found in the Allegheny Mountains. The diversity in ploidy levels was independent of the size of the range and the diversity of growing conditions among the three species of S. subsect. Venosae .


Introduction
The Solidago rugosa Mill. complex in the broad sense occurs in eastern North America and is a taxonomically difficult group of goldenrods making up Solidago subg. Pleiactila Raf. sect. Venosae (D. Don in Loudon) Nesom subsect. Venosae (G. Don in Loudon) Nesom (Semple and Beck 2021 [1]). The subsection includes three closely related species, S. fistulosa Mill., S. latissimifolia Mill., and S. rugosa, with four varieties recognized here and multiple basionyms: Solidago aestivalis Bicknell, S. aspera Ait., S. celtidifolia Small, S. edisoniana Mack., S. elliottii Torr. & A. Gray, Solidago elliottii Torr. & A. Gray var. ascendens Fern., Solidago elliottii Torr. & A. Gray var. pedicellata Fern., Solidago elliptica Ait., S. elongata Pépin, S. fistulosa Mill., S. latissimifolia Mill., Solidago mirabilis Small, S. rugosa Mill., S. rugosa var. cronquistiana Semple, S. rugosa var. glabrata Farwell, S. rugosa var. sphagnophila Graves, and S. villosa Pursh. The Solidago rugosa complex includes three to six species depending upon the classification scheme followed (Table 1). Fernald (1950 [2]) presented a treatment of members of the complex except for the much more recently described var. cronquistiana. He recognized three varieties in S. elliottii, no varieties in S. fistulosa, and five varieties in S. rugosa: var. rugosa, var. villosa (Pursh) Fern., var. aspera (Ait.) Fern., var. celtidifolia (Small) Fern., and var. sphagnophila. Earlier, Fernald (1936 [3]) dealt with the misapplication of the name S. altissima L. to members of the S. rugosa complex and acknowledged that var. sphagnophila might deserve species status as S. aestivalis. Cronquist (1947 [4], 1968 [5]) Taxonomy 2021, 1 291 recognized three species in the complex and grouped the varieties of S. rugosa into two subspecies: a southern subsp. aspera and a northern subsp. rugosa. Cronquist (1980 [6]) referred to subsp. villosa, but this was likely a typographical error, since the taxon is clearly his subsp. rugosa of both Cronquist (1968 [5]) and Gleason and Cronquist (1991 [7]). Beaudry (1960 [8]) recognized four species based on limited cytological data that indicated normally a single ploidy level for each: S. aestivalis, S. aspera, S. celtidifolia, and S. rugosa (which included f. villosa (Pursh) Beaudry); S. fistulosa was not included in his study. Uttal and Porter (1988 [9]) determined that the older name S. latissimifolia was the same species as S. elliottii and should be used instead. This paper follows the classification presented in Semple and Cook (2006 [10]) and accepts the name S. latissimifolia and is most similar to Cronquist (1968 [5]) but includes var. celtidifolia, var. cronquistiana and does not recognize subspecies within S. rugosa. Before the first author's laboratory began working on the cytology of Solidago (Semple et al. 1981 [11]), 52 chromosome counts from 27 locations had been reported for all members of the S. rugosa complex (Table A1). The first report was by Goodwin (1937 [12]) for S. rugosa var. rugosa. Most of the early counts were published by Jean Beaudry of the Université de Montréal (Beaudry and Chabot 1959 [13]; Beaudry 1960Beaudry [8], 1963, 1969 [15]). Beaudry's contributions to our understanding of the genus are significant, but he adopted a "splitters" approach to recognizing species, which he apparently thought at the time should include little or no ploidy level variation. His very small sample sizes allowed him to maintain this position.
An investigation of the cytogeography of the complex was undertaken by the Semple Astereae Lab in the 1980s and early 1990s in hopes of clarifying what role ploidy level might play in taxon delimitation. The field work to gather more data continued until 2012. Herbarium specimens, including nomenclatural types, were examined in order to more clearly understand how each of the above cited authors had applied names to their cytovouchers collected as part of the cytogeographic investigation. The overall objective was a better understanding of taxon limits within the complex, an assessment of the appropriate taxonomic ranks for these taxa, and a large cytological sampling of the entire ranges of all taxa in order to determine cytogeographic patterns in the taxa recognized.

Materials and Methods
Meiotic counts were made from pollen mother cells dissected from buds fixed in the field in Carnoy's Fixative 3:1 (absolute EtOH: glacial acetic acid) and subsequently stored under refrigeration in 70% EtOH until examined. Anthers containing pollen mother cells were dissected out of florets and squashed in 1% acetic orcein stain. Counts were made from freshly prepared slides using either a Zeiss RA Standard Microscope or a Nikon Microscope with phase contrast optics. Mitotic counts were made from root tip cells taken from transplanted wild rootstocks or from seedlings grown from fruits collected in the field. Root tips were pretreated in 0.01% colchicine or saturated paradichlorobenzene for 2-3 h, fixed in Acetic Alcohol Fixative (3:1/EtOH: glacial acetic acid), and hydrolyzed in 1 N HCl for 30 min at 60 • C before squashing. Counts were made from freshly prepared slides as above. Vouchers for all counts are deposited in WAT in MT (Thiers continuously updated [16]) with duplicates distributed to various other herbaria. Identifications were made by J.C.S.

Results
Cytovouchers for nearly all 178 previously published counts of Solidago fistulosa, S. latissimifolia, and S. rugosa were examined and identifications confirmed or corrected; these are listed in Appendix A Table A1, which includes a brief indication of geography, publication data, and indication of the name under which the counts were first published. Chromosome numbers were determined from an additional 117 individuals of the Solidago rugosa complex and are reported here for the first time in Appendix A Table A2. The distribution of all 41 diploid chromosome counts for S. fistulosa is shown in Figure 1. The distribution of the 9 diploid, tetraploid, and hexaploid chromosome counts for S. latissimifolia is shown in Figure 2. The distribution of all 242 diploid, tetraploid, and possible hexaploid counts for S. rugosa is shown in Figure 3 with ranges of each variety indicated.
Taxonomy 2021, 1, FOR PEER REVIEW 3 from freshly prepared slides using either a Zeiss RA Standard Microscope or a Nikon Microscope with phase contrast optics. Mitotic counts were made from root tip cells taken from transplanted wild rootstocks or from seedlings grown from fruits collected in the field. Root tips were pretreated in 0.01% colchicine or saturated paradichlorobenzene for 2-3 h, fixed in Acetic Alcohol Fixative (3:1/EtOH: glacial acetic acid), and hydrolyzed in 1 N HCl for 30 min at 60 °C before squashing. Counts were made from freshly prepared slides as above. Vouchers for all counts are deposited in WAT in MT (Thiers continuously updated [16]) with duplicates distributed to various other herbaria. Identifications were made by J.C.S.

Results
Cytovouchers for nearly all 178 previously published counts of Solidago fistulosa, S. latissimifolia, and S. rugosa were examined and identifications confirmed or corrected; these are listed in Appendix A Table A1, which includes a brief indication of geography, publication data, and indication of the name under which the counts were first published. Chromosome numbers were determined from an additional 117 individuals of the Solidago rugosa complex and are reported here for the first time in Appendix A Table A2. The distribution of all 41 diploid chromosome counts for S. fistulosa is shown in Figure 1. The distribution of the 9 diploid, tetraploid, and hexaploid chromosome counts for S. latissimifolia is shown in Figure 2. The distribution of all 242 diploid, tetraploid, and possible hexaploid counts for S. rugosa is shown in Figure 3 with ranges of each variety indicated.   A range of inflorescence branching patterns in S. rugosa were observed in the field. In one growth form, the inflorescence has very long arching lower branches. Type specimens and the majority of field collections of S. rugosa var. aspera and var. rugosa had large inflorescences with many long diverging branches with mostly secund branchlets with heads. Both varieties include some specimens with inflorescences with very short branches barely to slightly exceeding the subtending leaves. This is the supposedly diagnostic feature of var. villosa, but garden and greenhouse observations on transplanted individuals indicate that this trait is too unstable to be used as a diagnostic feature. It is common to find shoots of the same clone with the two extreme forms of inflorescence A range of inflorescence branching patterns in S. rugosa were observed in the field. In one growth form, the inflorescence has very long arching lower branches. Type specimens and the majority of field collections of S. rugosa var. aspera and var. rugosa had large inflorescences with many long diverging branches with mostly secund branchlets with heads. Both varieties include some specimens with inflorescences with very short branches barely to slightly exceeding the subtending leaves. This is the supposedly diagnostic feature of var. villosa, but garden and greenhouse observations on transplanted individuals indicate that this trait is too unstable to be used as a diagnostic feature. It is common to find shoots of the same clone with the two extreme forms of inflorescence branching patterns. Plants with short-branched inflorescences are scattered through the range of var. rugosa. Thus, we treat var. villosa as a synonym of var. rugosa.
The type specimens of S. rugosa var. celtidifolia have ovate upper stem leaves and inflorescences that have very short lateral branches. Throughout the range of the variety, plants often have large inflorescences with long lower branches like inflorescences in var. rugosa and var. aspera. Intermediate plants between obvious var. aspera and obvious var. celtidifolia plants were also encountered in the field.
During field work, a number of collections were found that had leaves typical of var. aspera, but an inflorescence form that was like the "var. villosa" inflorescence with short lateral branches. The indument was sparse for either of these morphs. Plants with these characteristics did not fit well into any of the five varieties recognized by Fernald and Cronquist ( Table 1). All these populations were found in the Carolinas and northern Georgia at the southern end of the Appalachian Mountains. Additional herbarium sheet collections were seen subsequently. All collections counted were tetraploid 2n = 36. These were placed in var. cronquistiana, named in honor of Art Cronquist.
The four varieties of S. rugosa recognized here occur in overlapping ranges and include many individuals that can be difficult to assign definitively to a single variety. Solidago rugosa var. sphagnophila is the most morphologically distinct with its absence of hairs on at least the lower stems and often glabrous or glabrate leaves; upper stems can be glabrous/glabrate to moderately hairy in the inflorescence. Only diploids were found and reported here, while only hexaploids were reported by Beaudry (1960 [8], 1969 [15]); tetraploids were not found in the sample of 13 individuals from eight locations in outer coastal plain cedar swamps and bogs from Massachusetts to New Jersey, and no chromosome counts from disjunct populations in Virginia and North Carolina have been reported (Figure 3).
Although Cronquist (1947 [4], 1968 [5], 1980 [6]) grouped the more northern varieties into subsp. rugosa and the southern varieties into subsp. aspera, var. rugosa and var. aspera are often difficult to distinguish, with many specimens being only somewhat to moderately rugose, that is, the intervening leaf surface is elevated slightly above the plain of the supporting venation which itself is prominent on the undersurface. The difference may correlate with growing conditions more than with geographic location, with more plants in southern populations growing in seasonally drier habitats during longer growing seasons than the more northern populations. In fact, only some of the plants assigned here to var. aspera were considered to have very pronouncedly rugose leaves and these came from the drier open woods and full sun habitats encountered in the southeastern states of the U.S.A. Plants assigned to var. aspera that grew in shaded thickets with wetter soils sometimes had less rugose leaves and were close in appearance to more northern plants of var. rugosa. Beck et al. (2021 [33]) included 15 samples of S. rugosa in their phylogenomic investigation of S. ulmifolia, but did not assign the S. rugosa samples to varieties. Seven of the collections were made by J.C. Semple and were cytovouchers for diploids cited in this paper; eight were for collections by other researchers and were not cytovouchers but varieties could be assigned to five based on online images. All specimens of S. rugosa were placed in a single branch of the tree with 74% support. The single collection of var. sphagnophila from Massachusetts along with a single diploid collection of var. rugosa from New York were placed in the clade sister of the remainder of S. rugosa. The most derived branch of the S. rugosa samples included three diploid collections of var. celtidifolia from Louisiana, although two of these were biological duplicates. The remainder of the samples included a sample of var. rugosa from Ontario, samples of var. aspera from Arkansas and Mississippi, a diploid sample of var. celtidifolia from Mississippi, and samples of either var. rugosa or var. aspera from Arkansas, Missouri, and Louisiana. Thus, there was phylogenetic limited distinction among samples of vars. aspera, rugosa, sphagnophila and celtidifolia nor complete separation of northern and southern samples. Clearly, a much larger sampling of specimens of all varieties of S. rugosa from across the entire range of each variety must be undertaken before strong conclusions can be drawn about the phylogenetic separation of varieties in S. rugosa. Certainly, the limited data available do not support the recognition of northern and southern subspecies but do hint at some separation of varieties and raise the question of whether or not var. aspera really is distinct from var. rugosa.
There were some patterns to the distribution of diploids and tetraploids in S. rugosa with differences in the patterns between varieties. All 55 samples of var. rugosa from Canada were diploid. Tetraploid samples of var. rugosa were scattered among diploid samples in New England and further south with the two ploidy levels occurring in approximately equal numbers of samples in the southern portion of the range of var. rugosa. For var. rugosa, 90% of the 106 samples were diploid. No unquestioned hexaploid samples were found in var. rugosa; one possible sample reported here (Table A2) is shown in Figure 3 with a question mark because of uncertainty that the voucher and the plant counted came from the same wild plant. A second hexaploid count by Löve and Löve (1982 [22]) was reported under the name S. altissima L., but the CAN voucher of the collection number is a specimen of S. rugosa var. rugosa. Therefore, the count cannot be assigned to either species with certainty as the voucher in CAN may not be the source plant of the count. In contrast, 91% of the 66 samples of var. aspera were tetraploid with diploids scattered in the central and southern portions of the range of var. aspera from eastern Kentucky to Mississippi. In var. celtidifolia, 51% of the 37 samples were tetraploids with diploids and tetraploids occurring throughout most of the range of var. celtidifolia. Only tetraploids were found in the southern high Appalachian Mountains in var. cronquistiana. In var. sphagnophila, diploids were sampled at five locations in the middle portion of the range and hexaploids were sampled in four locations over the range, but no tetraploids were sampled, perhaps due to limited sampling of populations.
The diversity in ploidy levels in the three species of S. subsect. Venosae was independent of the size of the range and the diversity of growing conditions. Solidago latissimifolia had the narrowest and smallest range, which included diploids, tetraploids, and hexaploids, even though the sample size was small and included only acidic soils mostly in Eastern White Cedar swamps along the Atlantic coast. Solidago fistuosa included only diploids throughout its less narrow range from the sandy Atlantic Coastal plain of southern New Jersey to southern Florida and west to eastern Louisiana on the coastal plain in a mixture of wetter soil habitats. Solidago rugosa included diploids, tetraploids, and possibly hexaploids in a diversity of habitats across much of eastern North America with the ranges of the varieties being smaller and overlapping with at least one other variety. While the strictly tetraploid S. rugosa var. cronquistiana came from only higher elevation habitats in the southern Appalachian Mountains, these were not very different from the higher elevation habitats of some tetraploid samples of var. aspera. While each taxon in subsect. Venosae has a different range of distribution, the ranges of all taxa overlap with one or more other taxa. Ploidy level is thus neither correlated with a specific habitat nor associated with any particular taxon in subsect. Venosae. Herbarium abbreviations from Thiers continuously updated [13].