Introduction

Knowledge of the variation and genetic structure of weed populations occurring at the borders of the non-native ranges of species may be crucial to understanding the biology of these species and planning measures to prevent their invasion. Interesting subjects for this type of research are the populations of Anthoxanthum aristatum Boiss. located at the species eastern non-native range limit. A. aristatum is an annual with loose tufts, flowering from May to June and sometimes again in summer and autumn. This species is distributed naturally in West Europe and North Africa in the Mediterranean and Atlantic regions (Kästner et al. 2001). After accidental introductions to Germany during the Napoleonic period (Hegi 1909), by the late nineteenth century, it had spread into the Netherlands and all over Germany. In subsequent decades, A. aristatum spread throughout western and central Poland and migrated towards the country’s eastern border, which it finally reached in the 1990s (Ciosek and Skrzyczyńska 1997). Beyond Europe, A. aristatum has spread to North America, Australia, New Zealand (Howell and Sawyer 2006; Randall 2007), and Asia (e.g., Mito and Uesugi 2004).

Within its natural range, A. aristatum occupies habitats of differing fertility composed of mesophilic deciduous forests (Rivas-Martínez et al. 1986), fertile grasslands (Fernández-Moya et al. 2010), mountain swards and pastures (Santa-Bárbara et al. 2003), communities connected to the edges of periodic ponds (Brullo et al. 2007; Fraga i Arguimbau 2008), grapevine fields and ruderal habitats (pers. observ.). Within its non-native range, A. aristatum mainly occupies oligotrophic crop fields, sandy grasslands, roadsides, fallow lands and wastelands (Schneider et al. 1994; Ryves et al. 1996; Niemann and Zwerger 2006; Pinke et al. 2006).

In Poland, annual vernal-grass is recognized as an invasive species (Tokarska-Guzik 2005). It has the highest number of localities in western and central Poland (Zając and Zając 2001). The spread of A. aristatum in Poland and its invasion into new habitats pose an increasing threat to the country’s crops. The number of its localities is rising, which indicates that the species should be monitored. Particularly alarming is its spread to new, more fertile habitats (Skrajna and Skrzyczyńska 2007). In addition, this species can be found in sandy grasslands and along roadsides (Jackowiak et al. 1990). Less frequently, it grows in open pine, birch or pine-oak forests of post-agricultural lands, at the forest-field borders and on sandy forest roads and roadsides (Woziwoda 2010; Bomanowska et al. 2014).

A. aristatum shows considerable morphological variability and genetic differentiation in both its native and non-native ranges (Pimentel et al. 2010; Drapikowska 2013; Drapikowska et al. 2013). Its ability to colonize various non-native habitats in Poland may result from a high level of polymorphic isozyme loci and high morphological variability previously observed in A. aristatum populations from Central and western Poland (Drapikowska et al. 2012; Drapikowska 2013).

In this study, we used molecular markers to better understand the variability and genetic structure of A. aristatum populations in different habitats at the non-native range limit of the species. Molecular biology offers many types of genetic markers that can be used in population genetic studies. PCR-based Single Primer Amplification Reaction (SPAR) markers, e.g., Random Amplification of Polymorphic DNA (RAPD), Inter Simple Sequence Repeats (ISSR) and Intron-Splice Junction (ISJ), are useful because they enable simultaneous detection of multiple polymorphic loci across the entire genome and do not require knowledge of the genome to design primers. These markers have been applied to the analysis of genetic variation in many groups of plants, including grasses (Vanderpoorten and Engels 2003; Alwala et al. 2008; Buczkowska et al. 2010). This group of markers also includes the Directed Amplification of Minisatellite-region DNA (DAMD), introduced by Heath et al. (1993) and, more recently, Sequence-Related Amplified Polymorphism (SRAP) developed by Li and Quiros (2001), which target the coding regions of the genome. These markers should be employed for research addressing new hypotheses in plant systematics, biogeography, conservation and ecology (Robarts and Wolfe 2014).

The aim of this study was: (1) to determine the degree of genetic diversity of A. aristatum at the species level at the limit of its non-native range, (2) to investigate significant genetic differences among populations representing arable field and sandy grassland habitats.

Materials and methods

The studied populations of A. aristatum are located in western Poland, near the eastern limit of its non-native range in Central Europe (Fig. 1). The plant material used in this study consisted of 70 samples collected from 7 populations (10 samples per populations) originating from two type of habitats: periodically disturbed habitats (sandy grassland near field roadside—FR, sandy grassland adjacent arable field—GA and sandy grassland on the edge of a pine forest—GF) and regularly disturbed habitats (arable fields—AF, recently abandoned field—RF) (Table 1). From each collected plant, all leaves were placed in silica gel or in a plastic bag and frozen upon arrival at the laboratory.

Fig. 1
figure 1

Distribution of Anthoxanthum aristatum in Europe based on the authors’ investigations and numerous literature sources (incl. Valdés et al. 2009). 1—native range, 2—primary localities beyond the native range, 3—non-native range, 4—secondary localities beyond the non-native range, 5—research localities

Full size image
Table 1 Anthoxanthum aristatum sample collection sites in Western Poland
Full size table

Localities in which plant material was collected for genetic analyses were investigated in floristic and habitat (ecological) terms. In each locality, a phytosociological relevé was made using the Braun-Blanquet method (Oberdorfer et al. 2001). Next, based on the Ellenberg method (Ellenberg et al. 1992), the mean ecological indicator values for each studied area and two distinguished groups of habitats: arable field and sandy grassland, were calculated.

Total genomic DNA was extracted from approximately 1 cm2 of frozen or silica gel dried plants. In the case of frozen samples, the leaves were ground with steel beads in a Bioprep-24 Homogenizer (BioEquip) for 35 s, and the standard CTAB procedure (Murray and Thompson 1980), downscaled to fit 1.5 ml Eppendorf tubes, was used for DNA extraction. For the extraction of DNA from silica gel material, the Novabeads Plant DNA Kit (Novazym) was used. Isolated DNA was dissolved in TE buffer and stored at − 20 °C. The quality of isolated DNA was evaluated by electrophoresis in 0.8% agarose gels, and the concentration and purity of DNA samples were determined using the Epoch™ Multi-Volume Spectrophotometer System (BioTek).

A set of 23 DAMD primers (Karaca et al. 2008; Karaca and Ince 2008; Sharma et al. 2011) was screened, which enabled us to choose six primers that amplified and produced reproducible and clear banding patterns for further analysis (Table 2). A preliminary screen was carried out using an annealing temperature gradient to optimize the PCR conditions. PCR was performed in 25 µl reaction volumes containing 40 ng of genomic DNA, 0.5 µM primer (Sigma), 0.5 mg/ml of BSA (Sigma), 1.5 mM MgCl2, 200 µM of each dNTP (Novazym), 1 × PCR buffer (20 mM (NH4)SO4, 50 mM Tris-HCl, pH 9.0 at 25 °C), and 1 U of Taq Allegro polymerase (Novazym). The amplifications were carried out in a BIO-GENER Thermocycler with a heated lid under the following amplification conditions: 94 °C for 4 min followed by 45 cycles at 94 °C for 60 s, 56 °C for 90 s, and 72 °C for 2 min, and a final elongation at 72 °C for 7 min.

Table 2 Sequences of the DAMD primers used in the Anthoxanthum aristatum study, with the marker parameters calculated for each primer
Full size table

The protocol for SRAP analysis was based on Li and Quiros (2001). Initially, a total of 27 different SRAP primers were tested according to Li and Quiros (2001), Nie et al. (2014) and Zitouna et al. (2015). Finally, eight primer pairs that produced clear bands were chosen for further analysis (Table 3). These pairs were selected from the 27 combinations of 11 forward (Me1—Me11) and 12 reverse (Em1—5, 7–9, 11, 12, 14, 15) primers. PCR was performed in 25 µl reaction mixtures with a final composition of 40 ng of genomic DNA, 0.5 µM primer (Sigma), 0.5 mg/ml of BSA (Sigma), 1.5 mM MgCl2, 200 µM of each dNTP (Novazym), 1 × PCR buffer (20 mM (NH4)SO4, 50 mM Tris-HCl, pH 9.0 at 25 °C), and 1 U of Taq Allegro polymerase (Novazym). Amplifications were carried out in a BIO-GENER Thermocycler with a heated lid under the following block temperature cycle: an initial heat denaturation at 94 °C for 5 min, five cycles were run at 94 °C for 1 min, 35 °C for 1 min and 72 °C for 1 min, for denaturation, annealing and extension, respectively. Then, the annealing temperature was raised to 48 °C for another 35 cycles, followed by final extension step of 10 min at 72 °C.

Table 3 Sequences of the SRAP primers used in the Anthoxanthum aristatum study, with the marker parameters calculated for each primer
Full size table

The PCR products of both types of markers were separated in 2% agarose gels containing ethidium bromide at 150 V for 2 h at room temperature. The amplified products were visualized using the ethidium bromide staining method under UV light and were documented using the Kodak (1D v. 3.5.4) gel documentation system. Amplified bands were manually scored as present (1) or absent (0); only clear and distinct bands were scored. The Nova 100 bp DNA ladder (Novazym) was used to estimate the size of the bands. PCR was conducted in duplicate to confirm the reproducibility of the banding patterns. One negative control in each amplification was included to exclude contamination.

Statistical analysis was based on 178 loci (96 DAMDs and 82 SRAPs). The obtained binary data matrices were analysed using GenAlEx 6.5 (Peakall and Smouse 2012). The following parameters were used to estimate the genetic diversity of the studied populations: the percentage of polymorphic loci (P), number of different alleles (A), number of effective alleles (Ae), Nei’s (1973) gene diversity or the expected heterozygosity (He) and Shannon’s information index of diversity (Is). The patterns of the population structure were determined according to the Bayesian clustering method implemented in STRUCTURE v.2.3.4 (Pritchard et al. 2000; Falush et al. 2007). STRUCTURE analysis was performed using a burn-in of 10,000 generations, followed by 100,000 MCMC repetitions and 10 iterations for 1–8 clusters (K) and default settings for other parameters. The optimal number of genetic groups (K) in the dataset was determined based on the ΔK method (Evanno et al. 2005) using the web-based program Structure Harvester (Earl and von Holdt 2012). The results of STRUCTURE analysis were visualized using the CLUMPAK server (Kopelman et al. 2015).

Analysis of molecular variance (AMOVA) was hierarchically calculated to estimate the allocation of genetic variation for the following configurations: (1) among and within the studied population and (2) at three levels: among two types of habitats (sandy grassland habitats, FR, GA, GF and arable field habitats, AF, RF), among populations within habitats and within populations. The overall and pairwise genetic differentiation between populations were also estimated by AMOVA using ΦPT statistics (an analogue to FST). The statistical significance of the ΦPT statistic was tested by a permutation test with 1000 permutations. Gene flow as measured by Nm was approximated from the ΦPT statistic (Cockerham and Weir 1993). Additionally, principal coordinate analysis (PCoA) based on the Nei’s (1978) genetic distance matrix was used to illustrate the relationships between individuals and populations. The above analyses were carried out using GenAlEx 6.5. Cluster analysis for all individuals was performed using Mega 6.06 (Tamura et al. 2013) based on the Nei’s genetic distances and the Unweighted Pair Group Method with Arithmetic Mean (UPGMA).

The primer banding characteristics, such as the number of total bands (NTB), number of polymorphic bands (NPB), number of monomorphic bands (NMB) and percentage of polymorphic bands (PPB), were calculated from the binary data matrix. To evaluate the usefulness of each primer to distinguish between the studied genotypes, the following measures were computed: marker index (MI) and polymorphic information content (PIC) according to Powell et al. (1996), resolving power (RP) and mean resolving power (MRP) according to Prevost and Wilkinson (1999). The parameters of the genetic diversity of A. aristatum were calculated at the species level.

Results

Floristic and ecological characteristics of studied habitats

A. aristatum occupies two types of habitats at the eastern limit of its non-native range: field habitats, where regular agrotechnical treatment is conducted (P4, P5, P6, P7), and relatively stable grassland habitats that develop at the edges of coniferous forests and along forest and field roadsides (P1, P2, P3) (Table 4). This species most often occurs in winter crop fields, mainly in rye fields, and recently abandoned fallow lands. These habitats are well lit (L = 6.9), warm (T = 5.8), relatively dry to periodically moist (F = 3.9), moderately to highly acidic (R = 4.1) and poor in nitrogen (N = 3.9). A. aristatum frequently dominates in species-poor agrocoenoses that occupy this type of habitats, where it covers up to 40% of the area. More and more often, A. aristatum migrates also into habitats that are less-disturbed than crop fields, i.e., into species-poor sandy grasslands. Some of them are related to Corynephorus-Agrostis grasslands, protected by the Habitat Directive—habitat code: 2330 (Interpretation Manual of EU Habitats 2007). Ecological parameters of grassland habitats are similar to agrocoenosis parameters, although the former are even more insolated (L = 7.8), dryer (F = 3.3) and distinctly poorer in nitrogen (N = 2.7). In the conditions of much stronger competition with grassland species, including perennial plants, A. aristatum forms less dense populations that cover up to 20% of the area.

Table 4 Floristic and ecological characteristic of habitats occupied by the studied populations of Anthoxanthum aristatum
Full size table

Polymorphism of DAMD and SRAP markers

A total of 178 bands were detected in 70 examined individuals of A. aristatum, with 96 bands amplified by 6 DAMD primers and 82 by 8 SRAP primers. The parameters that characterized each DAMD and SRAP primer are listed in Tables 2 and 3, respectively. The DAMD primers produced from 9 to 22 bands (average = 16). Most bands were amplified by the primer URP13R, whereas the least efficient primer was URP9F. The resolving power of the analysed DAMD primers ranged from 3.51 to 14.54. The highest RP value was noted for the primer URP13R, whereas the lowest was recovered for the primer URP9F. The primer URP13R also had the highest values for PIC (0.31) and MI (6.82), while URP9F had the lowest values for both of these parameters. The SRAP primers amplified between 5 and 17 bands, with an average of 10.25. The highest number of bands was amplified by the primer pair Me7–Em8, and the fewest were amplified by the primer pair Me8–Em14. The resolving power of the analysed SRAP primers ranged from 2.69 (Me8–Em14) to 11.23 (Me7–Em8). The highest PIC value (0.43) was from the primer pair Me7–Em4, whereas the maximum MI value (5.95) was found for the primer pair Me7–Em8. As each individual examined by both types of these markers had its own genotype, seventy genotypes were detected.

Genetic variation within populations

Of the 96 detected DAMD bands, 67 were polymorphic in all of the studied populations. The highest number of polymorphic loci was recovered in populations P7—81.25%, P6—73.96% and P4—71.88%. The mean percentage of polymorphic loci in populations was 70.09%. The average number of alleles per locus was 1.50. The mean gene diversity within populations (He) for all loci was 0.210 and Shannon’s information index (Is) was 0.33 (Table 5). In SRAP, 4 bands were monomorphic in all of the studied populations of A. aristatum. The highest number of polymorphic loci was found in population P6—76.83%, whereas the lowest in population P5, this number was 57.32%. The mean percentage of polymorphic loci in the populations was 65.33%. The average number of alleles per locus was 1.50. The mean gene diversity within populations (He) for all loci was 0.217 and Shannon’s information index (Is) was 0.33. In general, for both types of markers, all of the statistics of genetic variation were higher in populations originating from regularly disturbed habitats (RDH) compared to populations from periodically disturbed habitats (PDH) (Table 5). However, as the Mann–Whitney U test demonstrated, the differences are not statistically significant.

Table 5 Comparative list of genetic variability across 7 populations of Anthoxanthum aristatum based on DAMD and SRAP markers
Full size table

Genetic structure of populations

The analysis performed with STRUCTURE v2.3.4 based on a Bayesian clustering method suggested the presence of two genetic groups. The optimum number of genetic clusters was estimated using the ΔK criterion from the Evanno method. The maximal value of ΔK was 2 in both DAMD and SRAP markers, indicating that the studied individuals can be divided into 2 clusters. Bar plots showed the inclusion of both genetic groups in the studied populations (Fig. 2). Individual populations differed in their share of these genetic groups; populations P1 and P2 from sandy grassland (GA, FR) predominate in the first genetic group (orange cluster), while populations P4 from recently abandoned field (RF) and P5–P7 from arable fields (AF) formed the second group (blue cluster). Depending on the marker used, population P3 from sandy grassland (GF) fell within the first (in DAMD) or second (in SRAP) genetic group (Fig. 2).

Fig. 2
figure 2

The results of STRUCTURE analysis of DAMD (a) and SRAP markers (b). Each individual of Anthoxanthum aristatum is represented by a vertical bar showing the degree of admixture. Populations are separated by black lines. The results for ΔK are shown on the left side of the bar plot

Full size image

The genetic groups defined using the Bayesian method (Fig. 2) were generally confirmed by PCoA (Fig. 3), UPGMA clustering (Fig. 4). To determine whether the genetic structure of the studied population was affected by the outlier loci, we repeated the analysis using only putative neutral loci. Removing the outlier loci from the analysis did not significantly affect the genetic structure of the populations (not shown).

Fig. 3
figure 3

A two-dimensional scatter plot of the first 2 axes of Principal Coordinates Analysis (PCoA) of Anthoxanthum aristatum individuals based on DAMD (a) and SRAP (b) markers

Full size image
Fig. 4
figure 4

UPGMA dendrogram of Anthoxanthum aristatum samples based on Nei’s genetic distances for DAMD (a) and SRAP (b) data. The studied populations are marked by different symbols. Samples: 1–10—population P1 (filled circle), 11–20—P2 (open circle), 21–30—P3 (filled square), 31–40—P4 (open square), 41–50—P5 (filled diamond), 51–60—P6 (open diamond), 61–70—P7 (filled triangle). Blue-orange bars to the right of the graphs indicate populations belonging to the groups highlighted in STRUCTURE analysis

Full size image

Analysis of molecular variance (AMOVA) conducted without the division of the A. aristatum populations into habitat types revealed that most of the genetic variation (72% and 70%) was located within populations, regardless of the type of marker used (Table 6). AMOVA with the division of populations into two types of habitats showed that 68% and 66% of total genetic variation was still within populations, while the remaining molecular variation was distributed among populations within the habitats (20% and 21%) and among habitat types (12% and 13%) in DAMD and SRAP markers, respectively (Table 6). The genetic differentiation (Φ statistic) at all hierarchical levels was statistically significant (P = 0.001) (Table 6). The gene flow Nm among all of the studied populations estimated from DAMD and SRAP markers was 1.18 and 1.06 migrants per generation, respectively.

Table 6 Hierarchical analysis of molecular variance (AMOVA) without (A) and with (B) division of the studied populations of Anthoxanthum aristatum into two types of habitats: periodically disturbed habitats and regularly disturbed habitats, based on DAMD and SRAP markers
Full size table

The pairwise genetic differentiation ΦPT between the studied populations ranged from 0.086 to 0.427 in DAMD and 0.039 to 0.428 in SRAP markers. All of the ΦPT values, except for the lowest value in SRAP, were statistically significant at P = 0.001 or P = 0.002. The highest ΦPT values (0.427 and 0.428) were between populations 3 and 5 for DAMD markers and 2 and 5 for SRAP markers (Table 7).

Table 7 Genetic differentiation between the studied populations of Anthoxanthum aristatum (ΦPT) based on DAMD (below diagonal) and SRAP markers (above diagonal)
Full size table

Discussion

Preliminary studies on the genetic diversity of A. aristatum populations occurring at the border of the species non-native range lead to several significant conclusions.

Firstly, our study showed that the polymorphism of invasive populations of A. aristatum at the eastern border of its non-native range was relatively high: P = 70.9% for DAMD and P = 65.33% for SRAP markers. Additionally, the heterozygote frequency and PIC were lower. Depending on the markers used, He was 0.210 for DAMD markers and 0.217 for SRAP, and the PIC values were 0.31 and 0.43, respectively.

Although the indigenous populations of this species were studied with the help of other molecular markers, it is worth noting that their level of polymorphism is even higher. A study of Iberian populations using AFLP markers showed that the proportions of polymorphic loci were P = 87.9% (Pereira et al. 2007) and P = 87.96% (Pimentel et al. 2010), whereas the heterozygote frequency was He = 0.66 (Pereira et al. 2007). In turn, investigations using microsatellite markers showed that He ranged from 0.36 to 0.82, while the polymorphism information content (PIC) ranged from 0.33 to 0.80 (Lema-Suárez et al. 2016). Significant differences in genetic variability among populations located in the centre and at the species natural range limit have already been repeatedly demonstrated (e.g., Eckert et al. 2008; Zhang et al. 2015).

Secondly, our study of A. aristatum invasive populations at the eastern limit of the species non-native range showed a higher level of intrapopulation than interpopulation variability. AMOVA demonstrated that 72% and 70% of genetic variability occurs within populations for DAMD and SRAP markers, respectively, while only 28% and 30% results from interpopulation differences. Genetic variability of native populations of A. aristatum presents a different pattern. Investigations using AFLPs conducted in the Iberian Peninsula indicated that the interpopulation variability for this species was 77.34%, whereas the intrapopulation variability was 22.66% (Pereira et al. 2007; Pimentel et al. 2010).

Thirdly, our study demonstrated that habitat differentiation is reflected in the variation and genetic structure of A. aristatum populations occurring at the non-native range limit. These results refer to the often-discussed problem of interactions between an alien species and habitats it colonizes beyond its natural range limit (among others, Dietz and Edwards 2006; Riis et al. 2010; Erfmeier et al. 2011; Pahl et al. 2013; Tunaitienė et al. 2017). Previous studies of plants occurring in habitats differing in use intensity indicate three possible patterns of population genetic diversity: (1) plants in disturbed habitats have higher genotypic and genetic diversity than in stable habitats (Koppitz et al. 2000; Xie et al. 2001; Lambertini et al. 2008); (2) there are no significant differences between the genotypic and genetic diversity of populations occurring in habitats with different disturbance regimes (Solé et al. 2004); and (3) populations in stable habitats are more differentiated than populations in disturbed habitats (Trtikova et al. 2011; Tunaitienė et al. 2017).

In our study AMOVA with the division of populations into two types of habitats (field and non-field) shows that 68% and 66% of the total genetic variation remains within populations based on DAMD and SRAP markers, respectively, while the remaining molecular variation was distributed among populations within habitats (20% and 21%) and among habitat types (12% and 13%). Genetic differentiation (Φ statistic) at all hierarchical levels was statistically significant (P = 0.001).

In the case of A. aristatum occurring at its non-native range limit, our study demonstrated that despite a short period of adaptation, habitat differentiation reflects the variation and genetic structure of populations.

It is worth stressing that, irrespective of marker type used, two groups of populations can be distinguished according to Bayesian clustering, PCoA (Fig. 3) and cluster analysis based on Nei’s distance. According to the analyses, most groups consisted of populations from similar habitats. The first group comprises populations that inhabit relatively stable habitats (semi-natural sandy grasslands). The second group includes populations connected to field habitats that are regularly disturbed. The affiliation of the population occurring near pine forest is not clear and belongs to either the first or second group depending on the marker used (DAMD or SRAP). This may result from the instability of the habitat and the short-term, transitory character of this type of population. During colonization, A. aristatum is subject to selective pressure, which was supported by a positive alpha factor value (BayeScan analysis), probably favouring genotypes adapted to oligotrophic habitats. This is in accordance with the post-invasion evolution hypothesis (Müller-Schärer and Steinger 2004). A higher level of genetic variability was found in field populations, which may be due to two reasons: (1) a relatively longer time of occurrence in a given habitat compared to sandy grassland populations and necessity of adapting to more dynamic and competitive conditions and (2) more selective ecological barriers separating sandy grasslands from A. aristatum migration routes that permit only some ecotypes to enter.

The high genetic diversity populations at the A. aristatum non-native range limit in Central Europe and their marked diversity in arable field and sandy grassland habitats suggest that the studied species may systematically cross both geographical and ecological barriers.