Evidence of genetic diff erentiation and karyotype evolution of the sedges Cyperus ligularis L . and C . odoratus L . ( Cyperaceae )

Th e taxonomy of Cyperaceae is complex, with genera like Cyperus harboring species complexes. We analyzed the genetic similarity between Cyperus ligularis L. and C. odoratus L. based on DNA fi ngerprinting and cytogenetics. Signifi cative genetic diff erentiation (GST = 0.363) and low gene fl ow (Nm = 0.877) indicated a clear genetic distinction between the two species. Moreover, the clustering analysis showed two distinct genetic groups, suggesting a lack of evidence for hybridization. Th e phenogram revealed two diff erent lineages, and although all individuals of C. odoratus were collected from plots close to each other, they possessed greater genetic diversity than that observed among individuals of C. ligularis, which were sampled over a wider geographic range. Variation in chromosome number within the two species exhibited the opposite pattern, indicating greater karyotype stability in C. odoratus with 2n = 72 and 2n = 76, while the diploid number for C. ligularis varied from 2n = 66 to 88. Th e lower genetic variation in C. ligularis may be a result of the founder eff ect associated with seed dispersion and clonal reproduction. Field observations and analysis of reproductive biology should enrich the understanding of the genetic structure of the investigated populations and their role in successional processes.


Introduction
The family Cyperaceae (sedges) presents several taxonomic uncertainties with the presence of some species complexes, most of them described for Carex spp., such as Carex fl ava and Carex macrocephala species complexes (Hedrén 2004;King & Roalson 2009;Jiménez-Mejías et al. 2014), among others.For other Cyperaceae genera, there are few records of species complexes as seen in Rhynchospora (e.g., Rhynchospora glomerata) (Naczi & Moyer 2017).Cyperus, which is among the largest genera of the family, is considered complex, with most of the controversial groups inhabiting tropical regions (Simpson et al. 2003), mainly in Brazil, with about 100 species registered in varied vegetation types (Alves et al. 2015).In the most recent phylogenetic approaches, based on morphology and molecular data (Larridon et al. 2013;Reid et al. 2014), Cyperus was not recovered as a monophyletic group, being divided in the monophyletic clade "Cyperus C4" and the paraphyletic clade "Cyperus C3".Th e complexity of the genus Cyperus is also perceived at the intraspecifi c relationships level, as seen, for instance, in the C. rigens group (Reid et al. 2014).
It has been suggested that some degree of cross-species hybridization might occur between sympatric populations of the widely distributed group of articulated spikes, which comprises highly variable species as C. ligularis and C. odoratus (Luceño & Alves 1997).However, analyzing the internal anatomy of leaves of these two species, Martins & Alves (2009) did not find any evidence of hybridization.Species delimitation analyses in Cyperaceae have elucidated intraspecific relationships within difficult genera by accessing DNA fingerprinting profiles and karyotype features, as in the case of species of Carex and Schoenoplectus (Hipp et al. 2007;Esselman et al. 2012;Gizaw et al. 2016).Therefore, we relied on a genetic characterization of sympatric individuals of both C. ligularis and C. odoratus, by means of DNA Amplification Fingerprinting (DAF) and cytogenetic analyses, to access their diversity and the relationship between both species.

Plant material
For molecular analysis, young leaves were collected from 40 individuals of Cyperus odoratus L. and C. ligularis L. populations, 20 from each, in five collection plots in open spaces and on the border of an urban fragment of the Atlantic Forest fragment (Tab.1), with a minimum distance of three meters among individuals (Fig. 1).For cytogenetic analyses, young root tips of both species were collected, pre-treated with 2 mM 8-hidroxiquinolein for 4 h at 18 °C, and then fixed in ethanol: acetic acid (3:1) for 4-6 h at room temperature (ca. 25 °C).

Molecular data analyses
DNA extraction procedures followed the CTAB protocol as described by Weising et al. (2005), with minor modifications on the tissue amount (2 g to 1 g of fresh leaves) and the volume of the extraction buffer (10 ml to 12 ml).The DAF analysis was carried out following Caetano-Anollés et al. (1991) with modifications introduced by Winter et al. (2000) and Benko-Iseppon et al. (2003).The PCR reactions, with a final volume of 15 µL, consisted of 20 ng of DNA template, 10× PCR buffer, 2.5 mM MgCl 2 , 10 mM dNTPmix (Fermentas), 20 µM primer (Tab.S1 in supplementary material) and 1.0 U Taq polymerase (Fermentas).The DNA amplifications were performed in an Eppendorf Mastercycler Gradient thermocycler, using a program consisting of 2 min initial denaturation step (95 °C), followed by 40 cycles of 15 s at 95 °C, 1 min at 35 °C and 2 min at 72 °C, and the reactions were completed by a final extension step of 7 min at 72 °C.PCR products were separated by electrophoresis on 1.8 % agarose gel and stained with 0.5 µg/mL ethidium bromide.
The selection of the most informative primers was performed from an initial set of 50 DAF primers (Tab.S1 in supplementary material) tested on four random individuals (two of each species), using as parameters the number of amplified loci and polymorphic bands.The 10 most informative primers were used to amplify the DNA of the 40 individuals of both species.The genotyping process resulted in a binary matrix of presence (1) and absence (0) of amplified DNA fragments above 100 pb.
In order to check the relationships between the two species, the genetic dissimilarity analysis was performed according to the Jaccard's coefficient (Perrier et al. 2003) using the program DARwin 6.0 (Perrier & Jacquemoud-Collet 2006), to generate a phenogram based on the weighted neighbor-joining algorithm (2000 replications of bootstrap).The genetic differentiation between the two species (here represented by two populations) was accessed by G ST , a statistical index which is appropriate to analysis where the contribution of genetic drift among populations differences is not of interest (Holsinger & Weir 2009).Moreover, according to Wright (1931), the genetic differentiation among populations is directly related to  the migration, which was accessed through the number of migrants (N m ).The G ST and N m were calculated using the Popgene 3.2 program (Yeh & Boyle 1997) based on 1000 simulations.The genetic clusters were verified to assure whether there is evidence of interspecific hybridization between the two species.For this purpose, the software Structure 2.3.3 (Pritchard et al. 2000) was used through the Bayesian method where the K value was identified and analyzed to obtain the level of genetic admixture between the two species.The interaction between individuals was measured based on 600.000 replicates (60.000 in the burn-in stage), using the admixture model.The number of genetic clusters was inferred through the index described by Evanno et al. (2005).

Cytogenetic analyses
A total of 15 individuals of the two populations were analyzed, nine of C. ligularis and six of C. odoratus.Slides were prepared as described by Benko-Iseppon & Morawetz (2000), with modifications.Root tips were digested in an aqueous solution of 2 % cellulose (Calbiochem 21947) and 20 % pectinase (Sigma) for 90 min at 37 °C.Meristems were squashed in 45% acetic acid and frozen in liquid nitrogen to remove the coverslip.Then, slides were aged for three days at room temperature, stained with CMA (0.5 mg/mL, 1 h) and DAPI (2 µg/mL, 30 min), mounted in McIlvaine's buffer (pH 7.0): glycerol (1:1, v/v) and stored for another three days, according to Schweizer & Ambros (1994).After image capture of ca. 100 cells of each species (with a Leica DMLB epifluorescence microscope and an Olympus SP350 camera), slides were destained in ethanol:acetic acid (3:1, v/v) for 30 min, followed by immersion in ethanol for 1 h.Then, chromosome preparations were hydrolyzed in 5 N HCl at room temperature for 20 min and stained with 2 % Giemsa for 15 min and mounted with Entellan (Merck), for further image acquisition.Silver nitrate (AgNO 3 ) impregnation of interphase nuclei followed the protocol described by Hizume et al. (1980), with the modifications described in Vasconcelos et al. (2010).Thirty microliters of 50% silver nitrate were added to the preparations, which were covered with a piece of nylon mesh (24×32 cm), incubated at 37 °C for 40 min in a moisture chamber, being subsequently washed in distilled water and mounted with Entellan.For the chromosome counting and the silver nitrate impregnation approaches, at least 100 and 1000 cells were analyzed per species, respectively.

Results and discussion
A total of 172 polymorphic bands were observed at intra-and interspecific levels (Tab.2).The C. odoratus population presented the highest polymorphism level, represented by 124 polymorphic loci (Tab.2).According to Hollingsworth & Ennos (2004), for a reliable estimation of genetic diversity with dominant markers, at least 50 polymorphic loci are necessary for both resolving tree topologies and differentiating populations.The authors conclude that there is a direct relationship between the consistency of the genetic diversity analysis and the number of obtained polymorphic loci.Therefore, our results for the two analyzed Cyperus species are well supported, considering the relatively high number of sampled loci, despite the low number of sampled individuals.

Genetic dissimilarity and clustering analyzes
As previously mentioned, there are evidences of natural interspecific hybridization among sympatric populations of related Cyperus species (e.g., C. iria and C. microiria; Chozin & Yasuda 1991), as suggested for C. odoratus and C. ligularis by Luceño & Alves (1997).Therefore, considering this hypothesis, we analyzed the genetic dissimilarity between these two species.
The analysis of population genetic differentiation (G ST ) and gene flow (N m ) between the sampled species revealed a clear differentiation between the two species, with G ST = 0.363 and N m = 0.877, considering that values of G ST > 0.25 (Wright 1978) and N m < 1 indicate a favorable scenario for genetic differentiation (Almeida et al. 2003).Moreover, the Bayesian clustering analysis also indicated no evidence of hybridization between the species, although some individuals of C. ligularis and C. odoratus presented subtle evidences of ancestral mixing, which also may be associated with the highly polymorphic nature of the DAF markers.Interestingly, three genetic clusters were observed, being two distinct groups between C. odoratus and C. ligularis (Fig. 2).In addition, the neighbor-joining phenogram (Fig. 3) showed two main clades clearly separating C. odoratus and C. ligularis in different clusters, confirming the ITS-based molecular phylogeny by Reid et al. (2014) and the internal anatomy analysis by Martins & Alves (2009), respectively.
Regarding the intraspecific relationships, the individuals of C. odoratus presented a higher level of differentiation when compared with individuals of C. ligularis (Tab.2).The clustering data also makes this clear, where C. odoratus presents two predominant genetic clusters (Fig. 2 -red and blue clusters) showing a higher level of genetic differentiation.Moreover, a higher level of genetic differentiation was also observed in the neighbor-joining tree (Fig. 3), once several subclades with good support were separated from each other.All specimens of C. odoratus were collected in the vicinity of two dykes in the Dois Irmãos natural reserve.In turn, the C. ligularis individuals were found in various habitats, including the same as C. odoratus, and in different heavily perturbed open areas located nearby (Figs. 1, 3).The variability observed for C. odoratus may be related to higher levels of intraspecific polymorphism due to seedling recruitment, as previously reported for C. papyrus (Triest et al. 2014).On the other hand, the lower genetic differentiation observed in C. ligularis may be related to vegetative reproduction and/or a more efficient seed dispersion, which occurs by ornithochory, mirmechory, anemochory and mainly hydrochory (Linder & Rudall 2005), thus enhancing gene flow and reducing spatial aggregation among individuals (Triest et al. 2014).Moreover, we observed on the field that C. odoratus presents smaller clusters with more spherical rhizomes with lower density and a low number of clones, whereas C. ligularis has longer and thicker rhizomes with a higher number of clonal individuals.These different rates of clonal individuals between both species are probably related to the higher genetic diversity found for C. odoratus.

Karyotype analysis
Both species presented an intrapopulational variation in chromosome numbers, as it has been reported for other Cyperaceae genera (see Hipp 2007;Roalson 2008;Escudero et al. 2010).For C. odoratus, two chromosome numbers were observed: 2n = 72 (in 67 % of the cells) and 2n = 76 (in 33 % of the cells).On the other hand, C. ligularis showed a higher chromosome number variation, in which the diploid number ranged from 2n = 66 to 88, with 2n = 82 as the most frequent   Contrasting with the significant amount of published data on the karyological features of some genera of Cyperaceae, such as Carex, Eleocharis and Rhynchospora (Vanzela & Guerra 2000;Roalson 2008;Silva et al. 2010;Cabral et al. 2014), very little is known about the chromatin organization in Cyperus species.In fact, to the best of our knowledge, this is the first time that heterochromatic regions are evidenced in chromosomes of Cyperus species by means of fluorochrome banding.Consequently, it is hard to know if the coincident number of nucleolar organizer regions (NORs), which were evidenced by both CMA/DAPI and silver nitrate staining, indicates some conserved feature within the genus, as reported for other groups of Cyperaceae (Silva et al. 2010).Thus, even sharing the same number of active NORs, the different ranges of chromosome number variation among sympatric individuals of C. odoratus and C. ligularis reinforces the assumption of a clear separation of the two species.
The high degree of chromosome number variation in Cyperaceae members is widely attributed to the occurrence of holokinetic chromosomes in all members of the family, as well as in Juncaceae, its sister group.Thus, the presence of diffuse kinetochores may facilitate a faster karyotypic evolution by means of chromosome fission/fusion and/or translocations (Escudero et al. 2010;Chung et al. 2011).As demonstrated for Carex species and herein observed in the sampled populations, it seems that there is no correlation between geographic range, chromosome number variation and genetic diversity in these species with holokinetic chromosomes (see Escudero et al. 2010;Chung et al. 2011).

Final considerations
The natural interspecific hybridization is not rare in Cyperaceae, where species of several genera as Carex, Schoenoplectus and Isolepis present species with hybrid origin (Pedersen et al. 2016;Yano et al. 2010;2016).Although the morphology-based suggestion of hybridization between C. odoratus and C. liguraris by Luceño & Alves (1997), our molecular markers and cytogenetic analyses showed no evidence of hybridization.
Regarding the genetic diversity, C. odoratus surprisingly presents higher levels than C. ligularis, which is likely a result of a long period of evolution (Sheng et al. 2006) where the genetic diversity could be preserved by local vegetative regrowth of individuals.In the case of C. ligularis, the dispersal syndrome of the seeds, mainly by hydrochory and related with the colonization ability under various environmental conditions, seems to play a significant role in keeping the gene flow among individuals from different localities.However, the chromosome numbers showed a different picture, with C. odoratus less variable than C. ligularis, probably indicating a faster karyotypic evolution in the last species.
These results should be associated with additional field observations and specific analysis of reproductive biology, in order to enrich the knowledge concerning the dynamics and structure of the studied populations, as well as the role of these species in the successional processes and environmental restoration.

Figure 2 .
Figure 2. Bayesian clustering analysis based on Structure 2.3.3. A. Value of Delta K in accordance with the method described by Evanno et al. (2005).From the results obtained, 10 DAF primers were used, and 172 polymorphic bands were obtained for Cyperus odoratus and C. ligularis.Note that the peak occurred in K = 3. B. Ancestry of each individual in either of the two groups plotted by the software Structure version 2.3.4,using 10 DAF markers and 172 polymorphic bands for C. odoratus (individuals 1 -20) and C. ligularis (individuals 21 -40).Each vertical bar represents one of 40 individuals analyzed, and its genetic patrimony is noted through different colors (red, green and blue).Bar length is proportional to the ancestry values inferred in each group for each individual.

Figure 3 .
Figure 3. Neighbor-joining clustering analysis among 40 individuals of Cyperus odoratus and C. ligularis (Jaccard's coefficient and bootstrap of 2000 replicates).Collection plots: AP = Açude da Prata; API = Distrito de Apipucos; EP = Estrada dos Passarinhos; FM = Fábrica da Macaxeira; IB = Ilha do Bananal.Bootstrap values below 50 were not evidenced.one (40 % of the cells) (Tab.S2, Fig. S1 in supplementary material).For both species, the CMA/DAPI banding always revealed two pairs of CMA + /DAPI -signals (Fig. 4), which were associated with the two satellited chromosome pairs, also revealed by the silver nitrate impregnation (S Vasconcelos unpubl.res.).Contrasting with the significant amount of published data on the karyological features of some genera of

Table 2 .
Number of polymorphic loci at the intra-and interspecific levels, considering the ten selected DAF primers applied in Cyperus ligularis and C. odoratus.