How much of the invader's genetic variability can slip between our fingers? A case study of secondary dispersal of Poa annua on King George Island (Antarctica)

Abstract We studied an invasion of Poa annua on King George Island (Maritime Antarctic). The remoteness of this location, its geographic isolation, and its limited human traffic provided an opportunity to trace the history of an invasion of the species. Poa annua was recorded for the first time at H. Arctowski Polish Antarctic Station in the austral summer of 1985/6. In 2008/9, the species was observed in a new locality at the Ecology Glacier Forefield (1.5 km from “Arctowski”). We used AFLP to analyze the genetic differences among three populations of P. annua: the two mentioned above (Station and Forefield) and the putative origin of the introduction, Warsaw (Poland). There was 38% genetic variance among the populations. Pairwise ФPT was 0.498 between the Forefield and Warsaw populations and 0.283 between Warsaw and Station. There were 15 unique bands in the Warsaw population (frequency from 6% to 100%) and one in the Station/Forefield populations (which appears in all analyzed individuals from both populations). The Δ(K) parameter indicated two groups of samples: Warsaw/Station and Forefield. As indicated by Fu's Fs statistics and an analysis of mismatch distribution, the Forefield population underwent a bottleneck and/or founder effect. The Forefield population was likely introduced by secondary dispersal from the Station population.


| INTRODUCTION
Biological invasions present interesting evolutionary problems because they are stochastic events often involving small populations that can survive rapid habitat transitions (Colautti, Alexander, Dlugosch, Keller, & Sultan, 2017;Ellstrand & Schierenbeck, 2000;Lee, 2002). New ecological conditions encountered by individuals introduced into the new habitat may differ considerably from the conditions in their primary range. Therefore, natural selection and adaptation may be the key determinants of the success of invasion at the population level (reviewed in Facon et al., 2006;Schierenbeck & Ainouche, 2006). Substantial genetic variability is expected to favor adaptation in remote territories (Facon et al., 2006;Lavergne & Molofsky, 2007;Lee, 2002;Roman & Darling, 2007), while the rapid adaptation of invaders is common and generally not limited by genetic variation (Bock et al., 2015). A common scenario in many invasions is that small founder population sizes will often lead to reduced genetic diversity, and invading populations experience large environmental perturbations, such as changes in habitat and environmental stress (Lawson Handley et al., 2011).
An invasion process is composed of four main stages (transport, colonization, establishment, and spread) that need to be overcome by a population (e.g., Beck et al., 2008;Ochocki & Miller, 2017;. By reaching the next stage, an alien species gains a new status (e.g., Beck et al., 2008;Blackburn et al., 2011;Theoharides & Dukes, 2007). However, by breaking specific barriers and continuing to the next stage of invasion, a population may incur genetic variability loss (Lawson Handley et al., 2011). Population processes during an invasion are highly dynamic (e.g., Crooks, 2005;Facon et al., 2006;Theoharides & Dukes, 2007). Any actions to control an invasion should be attuned to this varying dynamic. All authors agree that prevention and early detection of potentially invasive organisms are most effective and economic (e.g., Blackburn et al., 2011;Cacho, Spring, Pheloung, & Hester, 2006;Veitch & Clout, 2002). However, the detection of such small populations is difficult. Failure to detect these populations may lead to their growth. This may lead to a demographic explosion, making it more problematic, or even impossible, to control an invasion. The extremely harsh abiotic conditions in the Antarctic put particular pressure on alien organisms. Many alien plant propagules reach the region due to human-mediated transport (e.g., Hughes, Convey, Maslen, & Smith, 2010;Lityńska-Zając, Chwedorzewska, Olech, Korczak-Abshire, & Augustyniuk-Kram, 2012;Cuba-Díaz, Troncoso, Cordero, Finot, & Rondanelli-Reyes, 2013; for data on the broader Antarctic, see McGeoch, Shaw, Terauds, Lee, & Chown, 2015).
However, only a few of these nonindigenous species can survive even a single vegetation season in the Antarctic, reaching the status of casual alien plant (Smith, 1996;Smith & Richardson, 2011). Only one alien species, Poa pratensis L., survived for over 60 years on the Antarctic Peninsula before it was eradicated. However, this species was not able to reproduce sexually (Pertierra et al., 2017).
Population demographic development has proven to be possible in the case of one nonindigenous plant species, Poa annua L. The species was recorded in several locations in the vicinity of the research stations along the Antarctic Peninsula (see Chwedorzewska et al., 2015;Molina-Montenegro, Carrasco-Urra, Acuña-Rodríguez, Oses, & Chwedorzewska, 2014). The most numerous populations of the species have been observed since the 1985/6 austral summer at the Henryk Arctowski Polish Antarctic Station, King George Island, South Shetlands (Olech, 1996). The expansion of P. annua in the vicinity of F I G U R E 1 Location of Poa annua in the vicinity of Polish Antarctic Station Arctowski, • Station, ▲ Forefield populations "Arctowski" is well documented (Olech, 1996;Chwedorzewska 2008).
In the austral summer of 2008/09, a population with numerous individuals of P. annua was recorded in a new location, 1.5 km from the "Arctowski" on the deglaciated moraines of the Ecology Glacier ( Figure 1; Olech & Chwedorzewska, 2011). According to available historical data, one can make the hypothesis that at "Arctowski" the diaspores of P. annua originated from Poland, most likely from unsterilized soil for the greenhouse transported to the station in 1978 from the Botanical Garden in Warsaw-Powsin. This is supported by observations conducted during 2000-2001 Polish Antarctic Expedition when emergence of P. annua seedlings was observed in the greenhouse building in a box containing soil destined for incineration . A fundamental question arose, regarding the origin of this species, during our long-term eradication program which started in 2014/2015 austral summer season (Galera, Chwedorzewska, & Wódkiewicz, 2017). Therefore, our first question was whether the new population on the deglaciated moraines of the Ecology Glacier originated from "Arctowski" population due to secondary dispersal, or whether it was a new introduction from a different source. Our second question was if the Ecology Glacier population proved to originate from the "Arctowski" population and how much of the species' genetic variability had been transferred into the daughter population. The main goal of our study was to determine the level of genetic variation between the two Antarctic populations of P. annua and the Polish population which is the most probable source of primary introduction of this species at "Arctowski."

| DNA extraction and AFLP assay
Total DNA was extracted with the MagAttract ® 96 DNA Plant kit (Qiagen) following the manufacturer's recommendations. To assess genetic variability, we used the amplified fragment length polymorphism (AFLP) procedure (Vos et al., 1995) with modifications (Chwedorzewska, Bednarek, Puchalski, & Krajewski, 2002), using KpnI/MseI enzymes for the digestion of 500 ng of genomic DNA.
After digestion, ligation of the appropriate adaptors was performed, followed by preselective and selective amplification steps. The selective amplification was carried out in the presence of 5′-( 32 P)-labeled primers. Eight selective primer pair combinations were used ( Table 1).
The PCR products were separated on 5% polyacrylamide gel and visualized by exposure to X-ray films at −70°C overnight. Two independent repeats of selective amplification and polyacrylamide gel electrophoresis were performed. AFLPs are dominant markers. Each amplification product (band) represents the phenotype at a single biallelic locus. Reproducible, clearly distinguishable bands were scored manually (two times by two independent persons) across all samples as either present (1) or absent (0) and recorded in the form of a binary matrix.

| Data analysis
GenAlEx 6.5 (Peakall & Smouse, 2006 was used to evaluate allele frequencies; number of bands shared among individuals with a frequency greater or equal to 5%; number of unique bands; Shannon's Information Index (I); and expected heterozygosity (H e ) for each population from binary data assuming Hardy-Weinberg equilibrium (Nei 1973, Bensch & Ĺkesson, 2005 (Excoffier, 2005;Fu 1997).
The bottleneck hypothesis was tested using the Bottleneck software (Cornuet & Luikart, 1996). The population structure was analyzed with Structure Harvester ver. 0.6.94 (Earl & Vonholdt, 2012) software set to the default parameters (Falush, Stephens, & Pritchard, 2007;Foll & Gaggiotti, 2008). The admixture model with correlated allele frequencies between populations was applied without using a priori information on population origin. Lambda (λ), the parameter of the distribution of allelic frequencies, was set to 1. A pilot study with the length of the burn-in and MCMC (Markov chain Monte Carlo) of 100,000-300,000 each was performed. Finally, 500,000 burn-ins and 500,000 iterations with 10 runs were carried out on the bioportal server (www.bioportal.uio.no) to quantify the amount of variation of the likelihood for each K. The range of possible Ks tested was 1-10. In order to determine the optimal number of clusters (K), an ad hoc statistic ΔK (Evanno, Regnaut, & Goudet, 2005) was used. Additionally, in order to investigate patterns of genetic subdivision of analyzed populations of P. annua, dendrogram using UPGMA (unweighted pair-group method with arithmetical averages) was created (STATISTICA 12.0, StatSoft Polska; Figure 2).

| DNA polymorphism
The DNA profiling of all plant samples allowed the identification of 270, 252, and 238 AFLPs generated by eight primer pair combinations, for the Warsaw, Station, and Forefield populations, respectively (Table 1). Most of the bands were present with a frequency higher than 5% within the given population ( Table 2).
The population from Poland amplified 15 unique bands, while both the Antarctic populations had only one common band, which was absent from the Warsaw population. The level of polymorphism was highest in the Warsaw population and lowest in the Forefield. Also, the Warsaw population was more heterozygous than Station and Forefield, while the Forefield shows the lowest heterozygosity. According to the Shannon's Information Index, the available markers were informative enough to proceed with further analysis (Table 3).

| Genetic structure
Analysis of Molecular Variance revealed that all the populations differed from each other (Table 4). Nei's genetic distance and Ф PT between the studied populations were biggest in the case of the Forefield and Warsaw populations and smallest between Warsaw and Station (Table 5). The evaluation of the agglomeration analysis based on the Δ(K) parameter revealed the presence of two groups of samples. No additional structuring was observed (Figure 3), what was also visible in the dendrogram (Figure 2

| Neutrality tests and demography
Tajima's D and Fu's F ST neutrality tests revealed that Tajima's D did not show any deviation from 0, while Fu's F S was negative and significant for all populations (Table 6).
In the mismatch distribution test for demographic/spatial expansion, there were no significant SSD values, and all the samples had a very low raggedness index (Table 7).
The three tests (Sing, Standardized, and Wilcoxon) for excess heterozygosity implemented in the bottleneck software produced significant p values based on the IAM model (Table 8).

| DISCUSSION
The KpnI/MseI platform was highly efficient in differentiating the analyzed populations as pointed out I value (Table 3) introduction. In order to study such effects using dominant markers, the infinite allele model can be used to test the mutation-drift versus the bottleneck hypothesis (Tero, Aspi, Siikamäki, Jäkäläniemi, & Tuomi, 2003). As expected, the Forefield population fulfilled the bottleneck hypothesis, or more likely the founder effect. However, it is difficult to perceive the difference between the bottleneck and founder effect with the application of dominant markers. In parallel to the demographic processes, the Forefield population may have been affected by some kind of selection processes. The most probable vector responsible for the establishment of the Forefield population is wind and/or human activity. This population is located in a place isolated by hills and at a substantial distance (approximately 1.5 km, Figure 1) from the Station population (Olech & Chwedorzewska, 2011); therefore, wind dispersal seems less probable, although it cannot be totally excluded. Consequently, seeds very likely were transferred on shoes or clothing of the personnel working at ASPA 128. This is also supported by our previous study showing that a great number of propagules were associated not only with cargo but also with personal clothes and field gear (Lityńska-Zając et al., 2012). Poa annua flowers profusely  and produces numerous viable seeds under Antarctic conditions (Wódkiewicz, Galera, Giełwanowska, Chwedorzewska, & Olech, 2013 Schrad., both 2n = 2x = 14 (Heide, 2001). Polyploids occur with greater frequency among invasive plants than among angiosperms in general (Brown & Marshall, 1981;Pandit, Tan, & Bisht, 2006;Prentis, Wilson, Dormontt, Richardson, & Lowe, 2008), with many allopolyploid hybrids among them (Lee, 2002). It is considered that polyploid hybrids tend to have greater fitness, possibly because of increased heterozygosity and reduced inbreeding depression (Soltis & Soltis, 2000), which can make them better colonizers than diploids, particularly under stress conditions (Prentis et al., 2008). This is supported by the high rate of allopolyploid species in the Arctic flora (Brochmann et al., 2004).

Percentage of variability
Our results show that the Forefield population was very likely introduced from the vicinity of Arctowski. Due to a limited number of individuals, this population was influenced by a bottleneck or founder effect and strong selection pressure, with parallel expansion. A critical factor in the success of this species is the ability to adapt rapidly to new environments following introduction Wódkiewicz et al., 2014). The genetic variation is not necessary for an invasion to succeed (Dlugosch & Parker, 2008), because invasions can be followed by rapid adaptive evolution (e.g., Amsellem, Noyer, Le Bourgeois, & Hossaert-McKey, 2000;Dlugosch & Parker, 2008).
A particularly successful invasive population may originate from a former introduction by secondary dispersal (Lawson Handley et al., 2011).
Biological invasions have become regarded as "natural experiments," offering unique insights into ecological and evolutionary processes occurring in real time (Lee, 2002;Sax et al., 2007). Understanding of these processes is crucial for implementing successful management policies. In a situation where much of the international scientific community's concern is devoted to minimizing the anthropogenic impact on Antarctic ecosystems (e.g., Hughes, Pertierra, Molina-Montenegro, & Convey, 2015;McGeoch et al., 2015;Znój et al., 2017), the monitoring and eradication of even such a spatially limited invasion as in the case of P. annua on King George Island become an important conservation issue. Station. We also thank two anonymous reviewers for a detailed and throughout review, language correction suggestions, and valuable comments on our manuscript.