The millipede genus Cherokia Chamberlin, 1949 is a monospecific taxon, with the type species Cherokia georgiana (Bollman, 1889). The last revision of the genus was made by Hoffman (1960) where he established three subspecies. Here we used molecular phylogenetics to assess the genus and evaluate whether it is a monophyletic group, and if the subspecies are each monophyletic. We included material from literature records and three natural history collections. Newly collected samples were obtained through a citizen science project. Morphological characters underlying subspecies groups—the shape of the paranota, body size, and coloration—were evaluated. A molecular phylogeny of the genus was estimated based on DNA sequences for seven gene loci, and a species delimitation analysis was used to evaluate the status of the subspecies. The documented geographical range of Cherokia in the United States was expanded to include a newly reported state record (Virginia) and about 160 new localities compared to the previously known range. Morphological characters, which included the shape of the paranota and body size that had been historically used to establish subspecies, showed clinal variation with a direct relationship with geographical distribution and elevation, but not with phylogeny. Coloration was highly variable and did not accord with geography or phylogeny. The phylogeny recovered Cherokia as a monophyletic lineage, and the species delimitation test supported the existence of a single species. The subspecies Cherokia georgiana ducilla (Chamberlin, 1939) and Cherokia georgiana latassa Hoffman, 1960 have been synonymized with Cherokia georgiana. The molecular and morphological evidence showed that Cherokia is a monospecific genus with the sole species, Cherokia georgiana, being geographically widespread and highly variable in its morphology.

Citizen science, DNA barcoding, morphology, phylogenetics, subspecies

The family Xystodesmidae (Polydesmida) includes 539 species with a center of diversity concentrated in the Appalachian Mountains (Means et al. 2021a, b; Hennen et al. 2022). Within the family Xystodesmidae, the Appalachian genus Cherokia Chamberlin, 1949 (Fig. 1) was named after the Cherokee, an indigenous group of peoples in the southeastern United States. This monospecific genus in the xystodesmid tribe Rhysodesmini was erected by Chamberlin (1949) for the species Fontaria georgiana Bollman, 1889 as its type species. After its description, various authors proposed multiple species that were all subsequently synonymized with the type species Cherokia georgiana (Bollman, 1889) based on gonopod morphology (Causey 1950; Hoffman 1950; Chamberlin and Hoffman 1958). However, all the above-mentioned authors pointed out the considerable color, body size, and shape variation in millipedes of the genus Cherokia.

Figure 1. 

Cherokia georgiana (Bollman, 1889), the wrinkled flat-backed millipede. Dorsal view of the whole body of specimen MPE04539 (Male, GA – White Co.) deposited in the Virginia Tech Insect Collection (VTEC). The image shows the more common coloration for the species and the prominent wrinkles of the cuticle.

Prior to Hoffman’s (1960) revision, no one had carried out a comprehensive synthesis of this genus. He (Hoffman 1960) proposed Cherokia as a monospecific genus, with the sole species Cherokia georgiana divided into three subspecies: Cherokia georgiana georgiana (Bollman, 1889), Cherokia georgiana ducilla (Chamberlin, 1939), and Cherokia georgiana latassa Hoffman, 1960. Hoffman (1960) described in detail the morphological variation and geographical distribution of Cherokia. He also differentiated the three subspecies from each other based on morphological features that included the position of the scapulora and the ratio of the body length versus its width. The scapulora is a term defined by Hoffman (1960: 231) as: “from the Latin “scapula,” a shoulder, and “ora,” the rim of a shield”. The scapulora in C. g. latassa is found in a marginal position, which separates it from C. g. georgiana and C. g. ducilla that have a submarginally located scapulora (Fig. 2A, B). The subspecies, C. g. georgiana and C. g. ducilla, were differentiated from each other based on the ratio of the previously mentioned body measurements (i.e., body length versus its width) (Hoffman 1960).

Figure 2. 

Position of the scapulorae (Sc) A strictly marginal B submarginal. Measurements of the 12^th body ring C metazonite width D metazonite length E paranota extension. Adapted from Hoffman (1960).

Hoffman (1960) confronted various problems during his revision of the genus Cherokia. The first one, he explained, was the fact that “despite the diversity of body form, color pattern, and morphological details which occurs in the genus, the male gonopods remain essentially similar” (Hoffman 1960: 227). Although some variation in the solenomere shape in specimens in the North Carolina mountains was observed by Hoffman (1960), the character was not consistent and did not correlate with geographical distribution or subspecies differentiation. Additionally, the same author expressed a struggle with confidently assigning all individuals to one of the subspecies. For this reason, Hoffman (1960) proposed an intermediate form, termed an “intergrade” between C. g. georgiana and C. g. ducilla. These intergrades were documented within a wide geographical band (~30 km) in western North Carolina between the distributions of C. g. georgiana and C. g. ducilla.

After 1960, some authors have indirectly mentioned Cherokia in tribal revisions (Hoffman 1978) and checklists (Shelley 1980, 2000; Marek et al. 2014). More recent research on the family Xystodesmidae, using a synthesis of morphological and molecular characteristics, has confidently placed the genus Cherokia within the family Xystodesmidae and subfamily Rhysodesminae Brolemann, 1916 as sister to the genus Pleuroloma Rafinesque, 1820 (Means and Marek 2017; Means et al. 2021b). As a result of field collections for this recent work, a large number of Cherokia individuals were collected from throughout the eastern United States, and within its range. These recent results combined with materials from natural history collections from the mid-1900s up to now, point to an even greater diversity than initially uncovered.

Here we used natural history collections in combination with new material sampled from nearly 200 locations within the range of Cherokia. These new samples, specially prepared for preservation of DNA, provided the basis to estimate an evolutionary history using molecular phylogenetics and address the status of the three subspecies within Cherokia georgiana.

The previously reported geographical range of Cherokia sensu Hoffman (1960), included six states, 43 counties and 93 localities. Here we report the presence of Cherokia in a seventh state (Virginia) and 53 new counties, for a total of 160 new localities where specimens of the genus have been collected. In prior systematic analyses of the millipede family Xystodesmidae, Cherokia was represented by three individuals, sequenced for six genes (Means et al. 2021b). Here, to address species boundaries in greater detail, we increased this number to 106 individuals sequenced for seven genes, for a total of 450 sequences and 3865 base pairs of DNA. These sequences were used to infer a phylogeny (Fig. 4). Of the seven loci amplified and sequenced for the phylogenetic reconstruction of the genus Cherokia, RNAPol2 was less successful than others in terms of amplification (presence of bands on electrophoretic gels) and sequencing (low quality reads: phred scores > 20). The presence of stop codons in RNAPol2 sequences, despite viewing in six alternative reading frames, is unexpected and may indicate that it is a recent pseudogene. However, the relatively lower success in sequencing of this locus does not appear to affect the general topology of the phylogeny.

Based on the molecular phylogeny, Cherokia is a monophyletic taxon (Fig. 4) sister to Pleuroloma. There is a clade formed by two individuals from the same locality (Monte Sano State Park, Madison Co., Alabama) that is sister to the remaining ones. Three statistically well-supported clades are present and subtended by long branches; however, the other individuals in the phylogeny are paraphyletic with respect to these clades and are not reciprocally monophyletic with them. In general, individuals from the same locality or nearby localities grouped together. Individuals from Kentucky and Virginia occur together with some individuals from Tennessee in a clade with very short branches. This block of individuals corresponds with the northeastern limit of the geographical range of the genus, and to the Cumberland Mountain Thrust Block region, a mountainous and complex region lying between the dissected Appalachian Plateau to the west and the Valley and Ridges to the east. This region also houses a clade of millipedes in the genus Brachoria Chamberlin, 1939 with similarly very shallow genetic divergences as Cherokia (Marek 2010). These shallow branches in Cherokia, as in Brachoria (Marek 2010), may represent relatively more recent and/or rapid diversification in the area, and may be due to shared mechanisms of regional diversity, or be associated with mimicry evolution in the area. Cherokia is a known participant of Müllerian mimicry in the region (Marek and Bond 2009; Marek 2010).

The morphological characters evaluated by Hoffman (1960) were reexamined with new measurements and compared to geographical variables (i.e., elevation) and the phylogeny. The measurements taken from the 12^th body ring and its inverse linear correlation with elevation showed that, in general, individuals of Cherokia with smaller body size and shorter paranota tend to be present at higher elevations than those with a larger size and longer paranota (Fig. 8). While the new measurements showed the same distribution as Hoffman (1960), the variation appears to be clinal, and not discordant variation with abrupt changes that would be expected to correspond to species boundaries. Many terrestrial invertebrate taxa show smaller body sizes at higher elevations, but the converse has also been observed (Hodkinson 2005). Smaller Cherokia at higher elevations may be associated with resource limitation as has been implicated in other terrestrial invertebrate groups (Hodkinson 2005). Alternatively, the smaller body sizes may be associated with body shape differences linked to burrowing efficiency in different leaf-litter substrates at higher elevations (e.g., there is a greater diversity and abundance of evergreen trees at higher elevations).

The results of the ABGD analysis showed a congruent pattern where genetic distances are continuously distributed and no barcode gap exists (Fig. 5A). This shows that there are no clear genetic clusters indicative of a barcode gap for distinct species or subspecies (Fig. 5B). While sampling effort may affect ABGD analyses, our dataset of 106 specimens uniformly sampled from across the distribution of the genus supports the hypothesis of a single, widespread species.

The position of the scapulora (sensu Hoffman 1960) was not a useful character, due to the difficulty of distinguishing its two states from each other (marginal and submarginal); perhaps this is due to its continuous nature, as is the case with the body size characters above (Fig. 8). During the examination of this character, we observed that the two states (marginal and submarginal) were not phylogenetically or geographically concordant. As described above (Fig. 9), the anterior margin of the paranota roughly grouped into two distinguishable shapes: sinuate or straight. To evaluate the relevance of this newly discovered character, its geographical distribution was mapped (Fig. 11).

Figure 11. 

Geographical distribution of Cherokia Chamberlin, 1949, showing the two types of paranota shape. Map includes specimens used for the morphological analysis and deposited at VTEC (N = 105).

The geographical distribution shows that the individuals with sinuate paranota generally tend to be located in the western part of the Appalachian region, while the individuals with a straight paranota are located in the eastern part. This separation appears to correspond to the Tennessee River Valley and the geological barrier that it represents for the genus, and other co-distributed taxa (e.g. Nannaria wilsoni species group; Hennen et al. 2022). However, in the southern part of the geographical distribution of Cherokia, especially in the state of Alabama, both shapes of the paranota overlap and no clear geographical separation was observed (Fig. 11). When this character was traced on the phylogeny of the genus, most individuals in one clade exhibited straight paranota (Fig. 4, blue), while the other clade (and two individuals from Monte Sano State Park, Alabama) possessed sinuate paranota (Fig. 4, red). One individual in the phylogeny and geographical distribution appears as an outlier for the general trend of this character (Fig. 4, GA-FLO-MPE03260*). Although a qualitative character and correlated with metazonite width (p = 0.0001), in some cases it is difficult to distinguish straight versus sinuate, and the variation appears to be clinal. In contrast with the scapulora and color characteristics, this character is largely concordant with the phylogeny, but in itself as a single character, insufficient for species or subspecies delimitation.

The coloration patterns were plotted on a map to assess concordance with the geographical distribution. Fig. 12 shows the distribution of the patterns (bimaculate, trimaculate, or striped), and the colors (red, orange or yellow). Some localities have all three types of patterns and/or colors—in contrast with Hoffman’s (1960) supposition that each coloration is geographically isolated. Nearly all possible combinations of colors and patterns were observed, but the trimaculate yellow color morph was the most common (both in frequency of individuals and geographical area). The bimaculate pattern was only observed with an orange hue (the bimaculate orange color morph, Fig. 10A–C). Fig. 12 shows that neither the pattern (bimaculate, trimaculate, striped) nor the colors (red, orange, yellow), have any clear geographical association. [Note that the number of geographical data points that were used for these maps (Fig. 11) were greater (N = 124) than the one used for the phylogenetic analysis (N = 106). Because the number of images available for the specimens actually used in the phylogeny was relatively small (N = 26) and limited the scope of inference, iNaturalist reports for Cherokia were also included in this section.] Perception of color can be affected by the observer, lighting conditions, veiling conditions, and distance, thereby adding error to the evaluation of this character (Endler 1990). In the future, a less error-prone and less human-centric technique should be implemented to obtain more accurate coloration data such as using a spectrometer and incorporating the visual systems of the predators of Cherokia (likely avian) to evaluate the coloration according to the perceivers’ eyes.

Figure 12. 

Geographical distribution of Cherokia Chamberlin, 1949 vs. coloration patterns A patterns B colors. Mapped using iNaturalist pictures reported for Cherokia (N = 124).

The use of citizen science as a tool for obtaining and analyzing data has been successfully demonstrated by various research groups. The Cornell Lab of Ornithology, for example, has developed multiple projects involving amateur ornithologists and the general public for around two decades. Data obtained from those initiatives have been published in several peer-reviewed research papers in various journals (Bonney et al. 2009). The small-scale citizen science project that was made as part of this research demonstrated that it is an effective method to obtain samples from remote and inaccessible localities, or in special situations such as the SARS-CoV-2 pandemic. Although the first response to the initiative was highly positive, follow-up contact with the interested participants was more difficult and less successful. The number of people who shipped samples back to us (N = 12) corresponds to around the 30% of the kits shipped to selected participants (N = 41). Improved communication with the participants, and a more structured timeline will be needed to increase the overall success of this initiative in future projects. Nonetheless the citizen science project offered an impactful opportunity to share the research with a broader community.

Morphological characters showed clinal variation and a direct relationship with geographical distribution and elevation, but not with the phylogeny. Coloration was highly variable and did not accord with neither geography nor phylogeny. The phylogeny recovered Cherokia as a monophyletic taxon, and the ABGD species delimitation test showed no barcode gap supporting the existence of multiple species. The molecular and morphological evidence showed that Cherokia is a monospecific genus with the sole species Cherokia georgiana being geographically widespread and highly variable in its morphology.

This research was supported by a National Science Foundation grant to P. Marek (Division of Environmental Biology, Systematics and Biodiversity Sciences # 1916368). Derek Hennen and Jackson Means helped confirm identifications of Cherokia observations on iNaturalist. We thank those who provided specimens: Roger Birkhead, Sawyer Birkhead, Chris Eaton, Barbara Graham, Christina Fizer, Sherrie White, Jessica Clay, Tracey Muise and Raegan Rainey. Robin Andrews and Bill Shear served on FVV’s thesis committee, and provided suggestions for earlier versions of the manuscript. Nesrine Akkari, Weixin Liu and Dragan Antić were reviewers who provided comments on previous versions of this paper.