Cytogeography of Callisia section Cuthbertia (Commelinaceae)

Abstract Determining the distribution of cytotypes across the geographic distribution of polyploid complexes can provide valuable information about the evolution of biodiversity. Here, the phytogeography of cytotypes in section Cuthbertia (Small, 1903) Hunt, 1986 is investigated. A total of 436 voucher specimens was georeferenced; 133 new specimens were collected. Based on flow cytometry data, DNA content of all cytotypes in section Cuthbertia was estimated. Utilizing chromosome counts and flow cytometric analysis, cytotype distribution maps were generated. Two disjunct groups of populations of diploid Callisia graminea (Small, 1903) Tucker, 1989 were discovered; tetraploid C. graminea ranges broadly from the coastal plain of North Carolina through central Florida. One hexaploid C. graminea individual was recorded in South Carolina, and numerous individuals of hexaploid C. graminea were found in central Florida. Diploid C. ornata (Small, 1933) Tucker, 1989 occurs in eastern Florida; previously unknown tetraploid and hexaploid populations of C. ornata were discovered in western and central Florida, respectively. Diploid C. rosea (Ventenat, 1800) Hunt, 1986 occurs in Georgia and the Carolinas, with populations occurring on both sides of the Fall Line. The cytotype and species distributions in Callisia are complex, and these results provide hypotheses, to be tested with morphological and molecular data, about the origins of the polyploid cytotypes.


Introduction
Polyploidy (whole-genome duplication) is a speciation mechanism that is a major evolutionary force; in fact, all angiosperms have undergone at least one ancient polyploidy event (Jiao et al. 2011, Amborella Genome Project 2013, and polyploidy has been a key driver of angiosperm diversity (De Bodt et al. 2005, Soltis and Soltis 2016, Tank et al. 2015. Polyploids are classified in two major categories: allopolyploids and autopolyploids. Allopolyploids are by far the more studied form and arise via hybridization between species, whereas autopolyploids originate from the multiplication of genomes within a single species. An autopolyploid is frequently considered as a cytotype within a species along with its diploid progenitor, as in Galax urceolata (Poiret, 1804) Brummitt, 1972(Baldwin 1941, Stebbins 1950, Chamerion angustifolium (Linnaeus, 1753) Holub, 1972(Mosquin 1967, Heuchera grossulariifolia Rydberg, 1900(Wolf et al. 1990, and Vaccinium corymbosum Linnaeus, 1753 (Camp 1945, Krebs andHancock 1989). However, autotetraploids are occasionally recognized as species distinct from their diploid parent, such as Zea perennis (Hitchcock, 1922) Reeves & Mangelsdorf, 1942(Iltis et al. 1979, Tiffin and Gaut 2001 and Tolmiea menziesii Torrey & Gray, 1840 ). Lumping diploid progenitors with their multiple derivative cytotypes into a single species may mask evolutionary lineages and grossly underestimate biodiversity ).
To gain a better assessment of biodiversity and to guide conservation efforts for species of interest, data on both evolutionary and life-history characteristics are needed. Callisia section Cuthbertia (Commelinaceae) from the southeastern U.S.A. comprises a polyploid complex, with species of conservation concern, but the extent of polyploidy and the geographic distribution of cytotype diversity are unknown.
Although earlier studies (e.g., Giles 1942Giles , 1943 provided the general pattern of species distributions and cytotypic diversity, the extent of cytotypic variation within and among species has not been examined in detail. Additional sampling of both populations and species is required to understand the extent and distribution of cytological variation in this clade. In this study, numerous new field collections were made, and known populations of Callisia section Cuthbertia were revisited; with the use of both traditional chromosome counts and flow cytometry, the ploidy of samples spanning the entire range of Callisia section Cuthbertia was investigated. Distribution maps of cytotypes and species were generated based on the cytological data obtained here, enabling future studies of phylogeny and polyploid origins in Callisia section Cuthbertia.

Georeferencing
To obtain locality data for Callisia graminea, C. ornata, and C. rosea, voucher specimens were examined from the following herbaria: GA, USCH, NCU, DUKE, US, AAH, FLAS, FSU, VSC, and SFU (codes follow Thiers 2016). The locality of each specimen was georeferenced by manually incorporating the label data into the web applications ACME mapper 2.1 (Poskanzer 2001) and/or GEOLocate (Rios and Bart 2010). Additional localities were obtained from the Master's Thesis of A. Kelly (1991) and personal communications with members of the Florida Native Plant Society and photographers from Flickr. com. In all, 436 specimens were georeferenced from herbarium specimens and observation records. (See supplementary file 1: Table 1 for georeferenced data points.) The data points were used to produce a distribution map using ArcGIS 10.4 (ESRI 2016) and to locate known populations and contact zones of all three species and their cytotypes.

Collecting of specimens
The georeferenced data were used to relocate populations within the southeastern U.S.A.; additional localities were discovered by exploring similar habitats in protected areas and on private land. Collections on private land were made with permission of the land owners. Based on the georeferenced data, permits were obtained to collect in state parks, state forests, national parks, and protected areas of The Nature Conservancy and the U.S. Fish and Wildlife Service in Florida, Georgia, South Carolina, North Carolina, and Virginia (Table 1). Mature individuals were sampled in the summers of 2012, 2013, 2014, and 2015. Only known localities with collection years between 1970 and 2012 were visited, unless the locality was in a protected area. This approach was used to increase the chances of finding intact populations but meant that we were unable to resample all of Giles's (1942Giles's ( , 1943 locations. Voucher specimens were deposited at the University of Florida Herbarium (FLAS); collection numbers are provided in Table 1.
Population localities were surveyed for individuals with different growth habit and habitat; we then collected across that diversity. Contact zones between species, based on the georeferenced localities, were more intensively surveyed by searching for distinct morphological variation (habit, leaf, and flower) to increase the probability of encountering mixed cytotypes. Two to six live plants were collected per locality. Plants were removed with 15 cm of soil circumference to increase the survival rate and placed in plastic bags. At the Department of Biology, University of Florida greenhouse, plants were then potted in a soil mixture of 1:1 sand and potting soil (Pro-Mix) and were kept under natural light. During the period from December-March, the individuals of putative diploid C. graminea and C. rosea were given a four-month dormancy treatment at 4°C to mimic their natural habitat.

Chromosome counts
Two individuals per cytotype of C. graminea were used as a control for flow cytometry analysis by counting chromosome numbers using established methods (see below). Previous studies of members of Commelinaceae found that cell division in root tips occurs at high frequency during late morning to early afternoon (Faden and Suda 1980). After a series of hourly collections, 2:00 pm was determined to be the optimal time for collecting root tips of C. graminea, C. ornata, and C. rosea.
Root tips were placed in 2 mM 8-hydroxyquinoline following Soltis (1980) for 24 hours at 4°C and then fixed in a 3:1 absolute ethanol-glacial acetic acid solution for 24 hours. Root tips were then placed in 70% ethanol and stored until needed at 4°C. Digestion of the root tips and spreading of the chromosomes on slides were performed following the methods of Kato et al. (2011). Chromosomes were stained with DAPI and visualized using a Zeiss Axio Imager M2 microscope (Carl Zeiss Microscopy LLC, Thornwood, NY, U.S.A.).

Flow cytometry
Preparation of all samples for flow cytometry followed Roberts et al. (2009), in which each sample consisted of approximately 1 cm 2 of fresh leaf tissue of Callisia; 0.5 cm 2 dried leaf tissue of Vicia faba (26.9 pg) was used as an internal standard (Dolezel et al. 2007). Samples were finely chopped with a sharp single-edged razor blade in a petri dish for 2 min in 1 ml of cold lysis buffer (0.1 M citric acid, 0.5% v/v Triton X−100, 1% w/v PVP−40 in distilled water) (Hanson et al. 2005, Mavrodiev et al. 2015. After 20-30 sec of incubation on a cold brick that served as a cold chopping surface, each sample was further treated and measured based on the methods of Mavrodiev et al. (2015) on an Accuri C6 flow cytometer (BD Biosciences, San Jose, CA, U.S.A). In all, the ploidy of 300 samples was assessed in batches of 28 samples.
For the estimation of genome size, three plants of the same accession were analyzed using the Flow Cytometry Kaluza Analysis Software 1.3 (Beckman Coulter Life Sciences 2016). The relative DNA content was calculated using the ratio of the mean fluorescent peak of the sample to the mean fluorescent peak of the internal standard, multiplied by the genome size of the standard, Vicia faba (Dolezel et al. 2007).

Georeferencing and collecting
All GPS points obtained here were incorporated into a map with ARCGIS 10.4 (ESRI 2016) (Figure 1). The results show that Callisia graminea ranges from North Carolina to central Florida with an isolated population in southern Virginia. Callisia rosea occurs predominantly in South Carolina and Georgia, and C. ornata is found in central to southern Florida. Specimens were collected at 133 localities, of which 61 were known from the 436 georeferenced localities and 72 were newly discovered populations. A list of these localities is provided in Table 1, indicating the geographic origin, ploidal level with corresponding number of plants, total number of analyzed individuals, and voucher information for each sample. Illustrations of the habits of diploid C. graminea, C. ornata, and C. rosea are provided in Figure 2.

Chromosome counts
Chromosome numbers were obtained for three individuals per cytotype in C. graminea, confirming the presence of 2n = 2x = 12 (diploids; Figure 3a), 2n = 4x = 24 (tetraploids; Figure 3b), and 2n = 6x = 36 (hexaploids; Figure 3c). The diploid and tetraploid counts were obtained for plants from known locations for which previous counts were available (Giles 1942, Kelly 1991. The hexaploids were discovered while counting spreads of putatively tetraploid C. graminea from Lake County, FL (Table 1). These 2x, 4x, and 6x individuals of C. graminea were then used as references in subsequent analyses using flow cytometry.

Flow cytometry
Ploidy was estimated via flow cytometry for 300 plants of C. graminea (representing 96 populations), C. ornata (from 23 populations), and C. rosea (from 7 populations). The results and the number of individuals analyzed per population are given in Table 1. Three distinct groups of fluorescence intensities were obtained from these analyses that were congruent with chromosome counts of diploid, tetraploid, and hexaploid C. graminea. Histograms for the cytotypes of C. graminea are shown in Figure 4. Results for 26 individuals (17%) of tetraploid C. graminea had a lower fluorescence intensity (sug- gesting a smaller genome size) than the remaining 83% of tetraploid C. graminea. The ploidy of the former plants was verified by chromosome counts, and all were tetraploid.
The relative genome size of individuals of C. rosea was similar to that of diploid C. graminea (2n = 2x = 12) (see below), confirming that our samples of C. rosea are diploid, in agreement with the literature (Giles 1942). Most individuals of C. ornata (2n =2x =12) were also inferred to be diploid, as expected based on previous counts (Giles unpublished), but our analysis also revealed previously unknown tetraploid (2n = 4x = 24) and hexaploid populations (2n = 4x = 36) of C. ornata. The latter were found in Seminole State Forest, FL, where they occur in sympatry with tetraploid individuals of C. graminea. All polyploid levels were verified with chromosome counts; chromosome spreads are depicted in Figure 3.
Genome size (2C-value) of cytotypes in Callisia section Cuthbertia was estimated; data are presented in Table 2 along with previously calculated genome sizes by Hertweck (2011) and Jones and Kenton (1984).
Distribution map -Based on the flow cytometry data, the distribution of cytotypic variation among the 126 populations sampled [C. graminea (96 populations), C. ornata (23 populations), and C. rosea (7 populations)] was mapped ( Figure 5). This map shows that diploid C. graminea is restricted to two disjunct areas: one in Franklin County, VA, and the second stretching along the Fall Line from North Carolina to South Carolina. Tetraploid C. graminea has a broader distribution that runs along the     , and tetraploid C. ornata occurs in western Florida (Polk, Hillsborough, Highlands, and Lake Counties). Hexaploid C. ornata occurs in Lake and Volusia Counties in central Florida.
Diploid C. rosea occurs in the piedmont of Georgia and South Carolina with some scattered populations in the coastal plain.

Discussion
Georeferencing -Callisia section Cuthbertia consists of three species native to the southeastern U.S.A., with three ploidal levels within C. graminea and C. ornata and diploids in C. rosea. The map of the geographic distribution (Figure 1) of all georeferenced voucher specimens depicts all specimens of C. graminea, C. ornata, and C. rosea without ploidal levels, collected from 1894 until present. Callisia graminea is the most widely distributed of all species in the genus, ranging from Virginia to Florida. Callisia ornata is restricted to Florida; although one specimen was recorded from Charleston County, GA, C. ornata was not found in Georgia in this study. Callisia rosea occurs mainly in Georgia and the Carolinas, but two herbarium specimens were found from Duval County and Highlands County, FL. The localities of these two herbarium specimens of C. rosea were vague, and C. rosea was not observed in Florida in this study.
Flow cytometry and genome size -Flow cytometry analysis of ploidal levels in 300 individuals from 126 populations together with 60 additional chromosome counts confirmed the presence of diploid, tetraploid, and hexaploid cytotypes of C. graminea and C. ornata. Significantly, tetraploid and hexaploid C. ornata were previously unknown. Our analysis also confirmed that C. rosea is diploid. However, Anderson and Sax (1936) and Ichikawa and Sparrow (1967) reported only tetraploids in C. rosea. This might be a misidentification of broad-leaved tetraploid C. graminea as C. rosea, as suggested by Giles (1942), who only detected diploids in C. rosea. Overall, three distinct fluorescent intensity peaks were seen in the histograms among the C. graminea and C. ornata cytotypes, with peaks for the tetraploids that are approximately twice the size of those of the diploids and for the hexaploids that are approximately three times those of the diploids. This general pattern of genome size increase in polyploids is to be expected relative to their diploid progenitors (Leitch and Bennett 2004).
It is interesting to note that 26 individuals (17%) of tetraploid C. graminea had a lower fluorescence intensity than the remaining 83%, suggesting a smaller genome size. The individuals with the smaller peak than that typical of other tetraploids were measured twice with the flow cytometer, and the results were consistent. The chromosome numbers of these samples were verified by chromosome counts, and all were tetraploid (2n = 4x = 24). Reductions in genome size in polyploids are common (Leitch and Bennett 2004), and in this study two hypotheses are possible: genome downsizing or the occurrence of multiple origins from parents having different genome sizes. Because this variation in genome size occurs among individuals within populations and because the individuals are not clustered in a single geographic area, we suggest that this variation in DNA content might be a result of genome downsizing, but this hypothesis requires further testing.
Genome size can be used, with other methods, to hypothesize putative progenitors of polyploids (e.g. Eilam et al. 2010). In diploid C. graminea the estimated 2C-value is 41.75 pg; the value for tetraploid C. graminea is 78.55 pg. According to Giles (1942), multivalent chromosome pairing was observed in tetraploid C. graminea, suggesting autopolyploidy. If tetraploid C. graminea is of autopolyploid origin, the expected DNA content would be 83.47 pg, but the observed DNA content of tetraploid C. graminea is 4.95 pg lower than the expected 2C-value. Newly formed polyploids usually possess a DNA content equal to the sum of the 2C-values of their progenitors (Bennett et al. 2000, Eilam et al. 2010. Over time, however, genome downsizing in polyploids relative to their progenitors is expected (Leitch and Bennett 2004), which seems to be the case in tetraploid relative to diploid C. graminea.
Due to the rarity of hexaploid C. graminea in South Carolina, we only calculated the 2C-value of hexaploids that occur in Florida. Hexaploid C. graminea may be of allo-or autopolyploid origin. If from allopolyploid origin, the expected 2C-value would be 127.06 pg, with diploid C. ornata (48.51 pg) and tetraploid C. graminea (78.55 pg) as the progenitors. The observed genome size of hexaploid C. graminea is 122.86 pg, which is lower than the expected value, again consistent with genome downsizing. In the case of an autopolyploid origin with tetraploid C. graminea (78.55 pg) as parent, we would expect a genome size of 117.83 pg, which is approximately 5 pg less than the observed 2C-value. Genome size data do not conclusively elucidate the origins of hexaploid C. graminea; both allo-and autopolyploidy are possible, and its origin requires further testing. However, Giles (1942) noted multivalent formation, generally indicative of autpolyploidy, in hexaploid C. graminea.
Tetraploid C. ornata has a 2C-value of 87.99 pg. It could be of autopolyploid origin with diploid C. ornata (48.51 pg) as the parent given that no other extant taxa are sympatric with it. However, the expected DNA content (97.02 pg) is at least 9 pg higher than observed; in contrast, when considering tetraploid C. ornata as a possible allopolyploid with tetraploid C. graminea (78.55 pg) and diploid C. ornata (48.51 pg) as parents (based on an unreduced gamete of the latter), the results (87.79 pg) are similar to the observed DNA content. These results therefore support allopolyploidy over autopolyploidy, yet further analyses are needed to clarify the origin of this cytotype.
Hexaploid C. ornata could be of allo-or autopolyploid origin. If allopolyploid, the expected genome size would be 127.06 pg with diploid C. ornata (48.51 pg) and tetraploid C. graminea (78.55 pg) as parents. The observed DNA content is 129.73 pg, which is slightly higher than the expected 2C-value. Alternatively, it could be an allohexaploid between tetraploid C. ornata (87.99 pg) and diploid C. graminea (41.75 pg), with an expected genome size of 129.74 pg, essentially identical to the observed value. In the case of autopolyploidy, we calculated an expected 2C-value of 145.53 if the value is 3 times that of diploid C. ornata (48.51 pg), 136.5 pg if tetraploid (87.99 pg) and diploid (48.51 pg) C. ornata are considered the parents, and 131.99 pg if a reduced and unreduced gamete of tetraploid C. ornata yield the hexaploid. The latter case is closest to the observed value, suggesting either that hexaploid C. ornata is of allopolyploid origin, or if an autopolyploid, it arose via the third possible mechanism outlined above; these hypotheses require further investigation.
Based on the Plant DNA C-values Database, http://data.kew.org/cvalues/ (Bennett and Leitch 2012), recorded species of Commelinaceae have a minimum 2C-value of 5.16 pg for Commelina erecta L.1753 and a maximum of 86.7 pg for Tradescantia virginiana L. 1753. The DNA content of hexaploid C. graminea and hexaploid C. ornata are currently the highest within Commelinaceae and Commelinales (Leitch et al. 2010) with 122.86 pg and 129.73 pg, respectively. Jones and Kenton (1984) reported that the 2C-value of C. rosea is 77.3 pg, with a chromosome count of 2n = 24, consistent with tetraploidy reported by Anderson and Sax (1936) and Ichikawa and Sparrow (1967); however, as noted above, Giles (1942) only detected diploids (2n = 12) for C. rosea, consistent with our results. The closest 2C-value to 77.3 pg is the 2C-value of tetraploid C. graminea with 78. 55 pg and 2n = 24 chromosomes; tetraploid C. graminea plants with broad leaves may be misidentified as C. rosea (Giles 1942). A voucher specimen of C. rosea from Jones and Kenton (1984) was not reported, so we cannot assess if the plant material used for the DNA content analysis was identified correctly. A misidentification is likely since the genome size estimation of Hertweck (2011) is close to our values. Likewise, previous tetraploid counts (Anderson and Sax 1936, Ichikawa and Sparrow 1967, Jones and Kenton 1984 may also be for tetraploid C. graminea plants that were misidentified as C. rosea. Alternatively, there may be cryptic tetraploidy in C. rosea that we failed to detect, but given our extensive sampling, we do not believe this to be the case. Distribution -As shown in Figure 5, two isolated populations of diploid C. graminea were detected. One population is in Suffolk County, VA, and the other is in North and South Carolina. These two isolated populations may have been part of a once larger geographic range for diploid C. graminea, but due to heavy agricultural activities in this part of North Carolina, suitable habitats ranging from Johnston County to Northampton County were transformed to farmland (personal observation). This anthropogenic influence may have caused the separation of the two isolated groups of diploid C. graminea.
Tetraploid C. graminea ranges from the coastal plain of the Carolinas to central Florida, with additional populations in the Florida panhandle (Franklin County, FL). This cytotype is clearly more abundant than diploid C. graminea; it is usually found in xeric disturbed areas and exhibits a larger growth form than diploid C. graminea. These tetraploids were abundant in Bladen and southern Cumberland Counties, NC, which border the isolated locality of diploid C. graminea in North Carolina. These two areas (occupied by tetraploid and diploid plants, respectively) are separated by the city of Fayetteville, NC. Although diploid and tetraploid entities of C. graminea were reported to be geographically isolated (Bergamo 2003, Giles 1942, 1943, Kelly 1991, one tetraploid individual was found within a diploid population in Cheraw State Park, SC; this individual is morphologically similar to the surrounding diploid C. graminea. This finding supports Giles's (1942) hypothesis that tetraploid C. graminea is an autotetraploid because it occurs consistently with diploid C. graminea. This hypothesis requires testing with molecular data.
The Fall Line runs essentially east-west through Georgia and from southwest to northeast in the Carolinas. Diploid C. rosea occurs on both sides of the Fall Line from Georgia to North Carolina. In Fort Gordon (Richmond County, GA), diploid C. rosea occurs in sympatry with tetraploid C. graminea. Although these two species occur in sympatry, hybrids were not observed at the site.
Diploid C. ornata is endemic to Florida, and tetraploid individuals of C. ornata occur in western Florida. These individuals may be autopolyploid, with diploid C. ornata as their progenitor. The distribution map in Figure 5 clearly supports the assumption of autopolyploidy, because there are no other Callisia species recorded in the region of diploid and tetraploid C. ornata. Morphologically, tetraploid C. ornata individuals show an increased axillary branching pattern, which is less common in diploid individuals. Axillary branching is a characteristic of C. graminea. Tetraploid C. graminea and diploid C. ornata are likely parents, through the union of one reduced gamete of tetraploid C. graminea and one unreduced gamete of diploid C. ornata.
In South Carolina, one hexaploid individual of C. graminea was found growing sympatrically with multiple tetraploid individuals of C. graminea. Hexaploid C. graminea in South Carolina appeared to be rare, and in 1942 only one individual was reported by Giles (1942). These rare hexaploid individuals may be allopolyploids, with diploid C. rosea and tetraploid C. graminea as their parents or autopolyploids with tetraploid C. graminea as their progenitor. Regarding allopolyploidy, C. rosea was not found sympatrically with tetraploid C. graminea in South Carolina; however, from the map of georeferenced specimens (Figure 1), there is a significant overlap of distribution between tetraploid C. graminea and diploid C. rosea in the Carolinas. With regard to autopolyploidy, individuals may have resulted through the union of one reduced and one unreduced gamete of tetraploid C. graminea given that no other Callisia species were observed in the population.
In Lake and Hernando Counties, FL, hexaploid individuals exhibited intermediate morphological characteristics between C. graminea and C. ornata. Some populations had typical tetraploid C. graminea or diploid C. ornata characteristics (Figure 2). Two forms were distinguished based on habit: (1) hexaploid C. graminea and (2) hexaploid C. ornata. Hexaploid C. graminea and one of its possible progenitors, tetraploid C. graminea, grow in sympatry at the Seminole State Forest, and hexaploid C. ornata was found growing with tetraploid C. graminea at the entrance to Brantley Branch Rd. (Seminole State Forest). The co-occurrence of hexaploids and tetraploids suggests that the hexaploids may be of allopolyploid origin. Hexaploid C. graminea was also collected at Lake Griffin State Park, Edward Rd., Lady Lake, and Seminole State Forest, FL. In Dunns Creek State Park and Welaka State Forest, diploid C. ornata and tetraploid C. graminea occur in sympatry; however, hexaploids were not found in these contact zones.
The rare hexaploid collected in South Carolina is most likely independently evolved from the hexaploids from Florida, and this entity from South Carolina could be either an allo-or autopolyploid. If allopolyploid, one likely parent, C. rosea, only occurs in Georgia and the Carolinas; if autopolyploid, the likely parent is tetraploid C. graminea. The hexaploid entities of Florida might be allopolyploid due to the intermediate morphological characters, with diploid C. ornata and tetraploid C. graminea as progenitors.
Callisia graminea forma leucantha, which was reported near Tampa, FL, was not found, but one white-flowered tetraploid individual of C. graminea was encountered among pink-flowered individuals in each of the following three locations: Sesquicentennial State Park, SC; Chesterfield Co., SC; and Tate's Hell State Forest, FL. One white-flowered individual of diploid C. rosea was found in Heggie's Rock Preserve, Appling, GA. White flowers reflect an absence of anthocyanins, which may result from mutations in any of the genes in the anthocyanin pathway or from lack of expression of potentially functional genes (Ho andSmith 2016, Rausher 2008). In Callisia section Cuthbertia, variation in flower color is common, but there is no association between color and ploidy within or among populations. Loss of anthocyanin pigments seems to occur sporadically within this complex.
Morphological and molecular analysis is an important next step in unraveling the complex relationships among cytotypes of Callisia section Cuthbertia. This work will allow us to reveal the parentage, evolutionary history, and the evolutionary role of all cytotypes within Callisia section Cuthbertia.