Population structure in chicory (Cichorium intybus): A successful U.S. weed since the American revolutionary war

Abstract Plant invasions are recognized as major drivers of ecosystem change, yet the precise cause of these invasions remains unknown for many species. Frequency and modes of introductions during the first, transport and colonization, stages of the invasion process as well as phenotypic changes due to plasticity or changing genetic diversity and adaptation during later establishment and expansion stages can all influence the “success” of invasion. Here, we examine some of these factors in, and the origin of, a very successful weed, Cichorium intybus (chicory) which was introduced to North America in the 18th century and which now can be found in all 48 continental U.S. states and much of Canada. We genotyped a Eurasian collection of 11 chicory cultivars, nine native populations and a North American collection of 20 introduced wild populations which span the species range (592 individuals in total). To detect the geographic sources of North American chicory populations and to assess the genetic diversity among cultivars, native, and introduced populations, we used both a sequenced cpDNA region and 12 nuclear simple sequence repeat (SSR), microsatellite loci. Four cpDNA haplotypes were identified and revealed clear geographic subdivisions in the chicory native range and an interspecific hybrid origin of Radicchio group. Nuclear data suggested that domesticated lines deliberately introduced to North America were major contributors to extant weedy populations, although unintended sources such as seed contaminants likely also played important roles. The high private allelic richness and novel genetic groups were detected in some introduced populations, suggesting the potential for local adaptation in natural sites such as deserts and nature reserves. Our findings suggest that the current populations of weedy U.S. chicory have evolved primarily from several sources of domesticated and weedy ancestors and subsequent admixture among escaped lineages.

surrogate, a source of polysaccharide inulin, and as a leafy vegetable (Kiers, Mes, Van der Meijden, & Bachmann, 1999). Chicory is native to Eurasia and the majority of the world production and breeding is in European countries. Most of the U.S. commercially produced chicory comes from California, New York, and Ohio (www.nass.usda.gov/ Data_and_Statistics/). Chicory also became a weedy/invasive species in North America and Australia and is labeled a noxious weed in the state of Colorado. Weedy chicory can be found across North America in 48 continental states of the United States and most provinces of Canada (USDA Plants Database). Chicory was also collected in 1956 on O'ahu Island (www.hear.org/vouchers/pier/bish0000011844.htm), but is not currently reported in Hawaii. Chicory exhibits a great range of phenotypes for leaf shape, color, leaf surface, hairiness, as well as plant size and, based on greenhouse experiments with variable soil types, temperatures and climatic conditions much of the phenotypic diversity can be attributed to environmental plasticity (Gemeinholzer & Bachmann, 2005). The plasticity of this species has been discussed for more than a century and was noted by early American farmers in field observations "…the foliage [of chicory cultivars] is by no means a constant character of variety" (Kains, 1898).
AFLP and RAPD markers for chicory were developed during the last two decades (Bellamy, Vedel, & Bannerot, 1996;Kiers, Mes, van der Meijden, & Bachmann, 2000;Koch & Jung, 1997;Van Cutsem et al., 2003;Van Stallen et al., 2001). Some of these markers have been used to construct a genetic map of chicory that was based on an intraspecific F2 population derived from a cross between two inbred lines of Witloof chicory varieties (Van Stallen, Vandenbussche, Verdoodt, & De Proft, 2003). Cadalen et al. (2010) constructed a consensus genetic map for chicory after the integration of molecular marker data of two industrial chicory progenies and one Witloof chicory progeny. These genetic markers have been useful for elucidating the origins and evolutionary history of the various domesticated lines. The genetic variation of available Witloof cultivars was shown to be low using RFLP data (Bellamy, Mathieu, Vedel, & Bannerot, 1995). In contrast, radicchio cultivars are highly heterozygous and genetically heterogeneous with some lines originating from a cross between C. intybus and C. endivia (Van Stallen et al., 2001). Unlike the situation for many domesticated species, particularly inbred taxa, most of genetic variation in the radicchio cultivars is partitioned within not between accessions (Barcaccia et al., 2003). Kiaer et al. (2009) measured spontaneous gene flow among wild European and cultivated chicory. The study indicated high levels of gene flow among populations in Europe with many incidents of recent gene flow between cultivars and wild populations.
The invasion history of chicory in North American is mostly unknown although there are some fascinating anecdotal accounts. [Sichorium Intibus] has been growing here in abundance and perfection now 20 years without any cultivation after the first transplanting" (Looney, 2004). Chicory plants would start to spread all over the continental United States to the point, that by 1900s, farmers would call for a chicory control and eradication. Seeds were distributed as an impurity in both foreign and domestic grass and clover seed (Hansen, 1920).
Population genetic structure can reveal some aspects of the invasion history of a species, most notably sources and modes of introductions and hybridization events (Fitzpatrick, Fordyce, Niemiller, & Reynolds, 2012), which should provide a more complete understanding of invasive weeds and enable better management of invasions.
Considering the references to chicory, both as a grass and clover seed contaminant, and as a crop, it is very likely that the invasion history of this species in North American is complicated. Currently, the levels of diversity, likely number and sources of introduction, occurrence of hybridization and the importance of selection are all unknown factors which may have affected the invasion process of chicory in North America. In this study, we genotyped cultivars, as well as wild Eurasian and North American chicory populations in order to assess the genetic diversity of this species, and to examine evolutionary changes since chicory was introduced to the United States in the late 1700s.

| Plant material
Our "Eurasian collection" consists of 11 domesticated lines and nine wild accessions obtained from a variety of sources and grown from seed in our greenhouse at University of Massachusetts, Boston  Table 2 and Figure 2 for locations and source of the collections). For the assays, we scored between 6 and 32 random plants per population, for a total of 592 individuals.

| Markers and genotyping
In this study, we used twelve microsatellite nuclear markers (Table 3) and the sequencing of one uniparentally inherited chloroplast trnL-trnF T A B L E 1 Chicory cultivars (1-11) and wild (12-20) Eurasian chicory populations

| Data analysis
Chloroplast DNA (cpDNA) fragments were sequenced and PCR prod-  Table 2 expected (H e ) heterozygosity, inbreeding coefficient (F), and the analysis of molecular variance (AMOVA) were calculated using Arlequin v. 3.5.1.3 (Excoffier, Laval, & Schneider, 2005). Significance of Φ ST values was determined via the maximum number of permutations in Arlequin 3.5. To characterize the genetic diversity at the population level and to control for sample size variation, allelic richness and private allelic richness were calculated using a rarefaction method in HP-Rare (Kalinowski, 2005). Chloroplast DNA haplotype maps were constructed using GPS visualizer (http://www.gpsvisualizer.com/).
The ancestry and the genetic composition of chicory individuals were evaluated with a Bayesian clustering method in program Structure v.
2.3.4 (Falush, Stephens, & Pritchard, 2003;Pritchard, Stephens, & Donnelly, 2000). We assumed that all loci were independent and found no evidence of linkage disequilibrium using Arlequin v. 3.5. All individuals were allowed to be products of admixture, and we used prior information about the population origin. The length of burn-in period was set to 200,000 iterations, and the number of Markov chain Monte Carlo (MCMC) steps after burn-in was 1,000,000. We conducted five independent runs with a partial data set (120 individuals-11 chicory cultivars and 9 wild chicory Eurasian chicory populations, with K set from 1 to 7), and with a complete data set (592 individuals) with K set from 1 to 10, with 10 iterations for each K in each independent run. Structure results were run through STRUCTURE HARVESTER v. 0.6.93 (Earl & vonHoldt, 2012) in order to calculate ΔK for each value of K according to Evanno, Regnaut, and Goudet (2005). The STRUCTURE HARVESTER output data were permuted with CLUMPP v. 1.1.2 (Jakobsson & Rosenberg, 2007). The final visualization of genetic data was plotted with DISTRUCT v. 1.1 (Rosenberg, 2004).

| Chloroplast markers
Two random samples from each population were sequenced at trnL-trnF locus. We detected four cpDNA haplotypes (

| Nuclear markers
Twelve assayed microsatellite loci were polymorphic, and markers amplified in all 592 individuals. The number of alleles per locus ranged from 8 to 26 (Table 3)  Four genetic groups for Eurasian chicory and cultivars were resolved by the microsatellite data analysis and individuals in several populations showed evidence of admixture ( Figure 1). We conducted a partial STRUCTURE analysis for just the 120 sampled plants of the 11 CC and 9 EU populations, and the number of clusters (K) was varied from one-seven. The highest likelihood in the partial analysis was obtained when K was set to four, and also the maximal ΔK occurred at

| DISCUSSION
The present study is a first broad genetic survey of North American chicory placed into a global framework of the species' natural history.
The genetic diversity for the 12 SSR markers was high (H e = 0.61), similar to another North American non-native taxon in the Asteraceae (Eriksen et al., 2014) & Brzyski, 2015). cpDNA variation in populations is generally low, but it serves as a useful tool for monitoring seed dispersal and maternal contributions (Ennos, Sinclair, Hu, & Langdon, 1999;Wallace et al., 2011).
In chicory, cpDNA haplotype diversity was low as expected, and the different native and domesticated sources of seed generally possessed a single cpDNA haplotype. Significantly, we detected all three haplotypes from native European collections within North America. Together, chloroplast and nuclear data provided evidence of multiple introductions and admixture; three different cpDNA haplotypes suggest different seed sources and unique genotypes, high nuclear genetic diversity, and high intergroup gene flow suggest hybridization and recombination.
Domestication of wild plants is considered a long process that starts with human selection. Breeding and cultivation of these plants terminates in a fixation of favored morphological and genetic differences distinguishing a domesticate from its wild progenitor (Pickersgill, 2007). A subset of weeds and invasive plants has evolved in the reverse direction from domesticated ancestors by at least two different pathways. In California where no wild relative existed, weedy rye appears to have evolved directly from the escaped crop (Burger, Lee, & Ellstrand, 2006). A different pathway is represented by Europe's weed beet (Desplanque et al., 1999) and root chicories were shown to be in closely related clusters using AFLP data analysis (Kiers et al., 2000). It was suggested that Witloof chicory is derived from the Magdeburger root chicory type (Bellamy et al., 1996) and our microsatellite data showed Witloof to be an extraction of the more diverse Magdeburger root chicory, supporting this hypothesis. The Radicchio accessions are cultivated for their leaves and showed wide genetic variability. The literature lists a hybrid origin of the Radicchio type Chioggia as a result of crosses between chicory and an endive cultivar (Barcaccia et al., 2003), but this hypothesis could not be either confirmed or rejected by previous genetic studies (Kiers et al., 2000;McDade, 1997). The presence of endive cpDNA haplotype detected in Radicchio "Variegata Di Chioggia" (It1) accession supports the interspecific hybrid origin of Radicchio. While a high genetic diversity is not necessarily a prerequisite for successful habitat colonization (Ward, Gaskin, & Wilson, 2008), invaders with greater genetic diversity may be able to adapt more readily to new environments. Analysis of allelic diversity of microsatellite data revealed no reduction of genetic diversity in wild, introduced populations versus domesticated lines. Bottlenecks were not expected, given the outcrossing breeding system, and the dispersal history in North America spanning over two hundred years since its introduction.

Chicory was introduced into
Recent genetic evidence implies that many large-scale plant colorizations were associated with multiple introductions, and a bottleneck would be inferred only if introduced populations contain fewer rare alleles than expected (Peery et al., 2012). Furthermore, simulations indicate that even moderate gene flow can mitigate the detection of genetic bottlenecks using traditional methods (Fitzpatrick et al., 2012), and extensive admixture was evident in our STRUCTURE analysis.
Our results are consistent with previous studies in European chicory which showed high levels of gene flow between cultivated and wild chicory accessions (Kiaer et al., 2009;Van Cutsem et al., 2003).
As a group, the domesticated accessions in this study were as similar to wild Eurasian accessions (F ST = 0.0254) as they were to wild NA accessions (F ST = 0.0238). Conversely, the wild accessions of Eurasia and wild NA accessions showed nearly twice the F ST value (0.0442). This suggests that North American chicory evolved primarily from introduced domesticated lines over the past two hundred years. However, some NA populations (NV and CO) seemed to have a closer affinity to wild accessions from EU and other populations in NA seemed relatively distinct from the major domesticated lines surveyed in this study. This together with the high private allelic richness found in the invaded region indicate additional introductions and sources of diversity as well as the possibility of local adaptation to the new environments.
Our NA collection of populations did cover a broad spectrum of habitats with different histories. One of the sampled populations was collected along the roadsides leading to Tufton Farm, the location of the first recorded chicory planting in the United States by Thomas Jefferson.