By testing biogeographical hypotheses, e.g. a central-marginal gradient, we evaluated demographic characteristics (i.e. population density, population size), reproductive success and genetic composition along our study transect. Regarding the demographic characterisation of our studied populations,
Poa badensis followed the hypothesised gradient (Sagarin and Gaines
2002; Abeli et al.
2014). In Germany, the species is classified as ‘vulnerable’ and regionally ‘endangered’, e.g. in Rhineland-Palatinate where only a few populations exist (see chapters below; Korneck et al.
1996). Although we have no concrete information on the total number of existing populations in Austria and Hungary, the species is more common there (Niklfeld
2013; Bartha et al.
2015). Hence, we conclude that the density of populations in our study regions is indeed declining towards the species (north)westernmost distribution limit. Moreover, we analysed the individual-richest populations in Hungary, represented by hundreds of individuals. The smallest populations within our study transect, most contained less than 200 individuals (Tables
1,
2), were found in Eastern Austria. However, the individual-poorest population analysed was again from Western Germany (i.e. Eckelsheim, n < 30).
On the contrary, our results did not indicate the expected reduced reproductive success of the (north)westernmost populations (cf. Abeli et al.
2014). The values of performance parameters were highest or at least equally high as values from the continuous part of the distribution range (i.e. Central Hungary; Table
2), as also described for
Stipa capillata (Wagner et al.
2011b) and
Lychnis viscaria (Lammi et al.
1999). Nevertheless, our performance parameters include some uncertainty, since they reflect the species’ reproductive success in situ (i.e. in the wild; cf. Wagner et al.
2011b) and may therefore include some maternal effects (Gutterman
2000).
Regarding the genetic composition of our study species, we observed a regional genetic impoverishment of (but see next Discussion chapter) and an increasing genetic differentiation among the (north)westernmost populations, as expected for a central-marginal gradient (Sagarin and Gaines
2002; Eckert et al.
2008; Sexton et al.
2009), but also partially for ‘stable’ rear edge populations (rather than for leading edge populations; Hampe and Petit
2005). Regional population differentiation was by far the highest in Germany. Pannonian populations from Austria and Hungary showed a lower population differentiation (Online Resource 3), which should also be found when further populations from the centre of the continuous Pannonian range are analysed, because gene flow is obviously high in this region. Both marker systems (AFLP & cpDNA) revealed a clear distinctiveness of the (north)westernmost exclave. Such a strong genetic distinctiveness and/or high population differentiation of peripheral populations has been observed, for example, in two other Poaceae species,
Stipa capillata (Hensen et al.
2010; Wagner et al.
2011a) and
Stipa pennata (Wagner et al.
2012), as well as in our previous study on
Linum flavum (Plenk et al.
2017), which is based on a similar transect setting.
Constitution of populations in the (north)westernmost isolated exclave
In the German exclave, the species’ reproductive success was not obviously reduced, while at the same time our analyses of genetic composition showed a partial genetic impoverishment. This pattern was somewhat surprising, as we did not expect to find reduced genetic diversity
and high reproductive success in this region. The overall low level of genetic diversity observed in Western Germany, seems to be mainly a result of very low values found in two of the four investigated populations (i.e. Eckelsheim, G1, PLP % [7]: 21.7; Weilersberg, G2, PLP % [7]: 24.0). The populations Heidesheim (G3, PLP % [7]: 49.7) and Eierfels (G4, PLP % [7]: 44.7) did not show reduced levels of genetic diversity; the values observed there are comparable to those of the continuous distribution in Central Hungary (PLP % [7]: 41.1; see Table
1). Hensen and Wesche (
2007) observed comparatively high levels of genetic diversity (PLP %: 50.0–74.2; genetic diversity: 0.20–0.29) in Central German populations of
Poa badensis, but a moderate to low genetic differentiation among populations. Compared to other thermophilic steppe species from the same plant family (i.e.
Stipa capillata, Hensen et al.
2010;
Stipa pennata agg., Wagner et al.
2012; Durka et al.
2013), or other steppe species occurring in calcareous rock/sand vegetation (i.e.
Globularia bisnagarica, Honnay et al.
2007;
Alyssum montanum, Španiel et al.
2012), these values represent high levels of genetic diversity at the (north)westernmost border of their distribution. Consequently, Hensen and Wesche (
2007) described the species as only slightly affected by fragmentation and assessed genetic exchange between populations in their study region as still possible. In our more western German study region,
P. badensis is obviously affected by reduced gene flow (cf. Figures
1,
2), which is also clearly demonstrated by the strong genetic within-region population differentiation (Online Resource 3). Although populations of wind-pollinated and -dispersed species in general are assumed to have a greater potential for genetic exchange, the low dispersal capacity of our study species together with the scattered availability of suitable habitats in our study region have led to this increased genetic differentiation. In addition, we found no clear evidence for isolation by distance, possibly indicating genetic drift already occurring in the Western German populations.
Even though the regional genetic diversity is lowest in the German exclave, we basically interpret its comparatively high reproductive success as indicator for good reproductive capability. As we obtained indications of different ploidy levels within our study transect (i.e. diploid populations in Austria and Hungary, but tetraploid populations in Germany, Online Resource 1), a possible explanation for the high performance parameters observed in the (north)westernmost exclave might be the occurrence of one (or more) polyploidization events, likely occurring during warm (postglacial) stages. This may have led to a stronger reproductive potential, probably as an evolutionary strategy to overcome the more unfavorable conditions at the absolute range limit. Moreover, polyploid populations may also show generally higher levels of genetic diversity (cf.
Alyssum montanum, Španiel et al.
2011).
However, the species’ biogeographical history and specific traits (e.g. longevity) may have aided to retain a certain level of genetic diversity, as observed in the populations Heidesheim (G3) and Eierfels (G4). On the contrary, the populations Eckelsheim (G1) and Weilersberg (G2) are likely less viable and show low genetic diversity, which is driven by their histories. Although population size did not correlate with the genetic diversity in general, at least one of these two populations is characterised by small population size (G1, less than 30 individuals). Nevertheless, low genetic diversity of populations does not necessarily imply that they have to be less viable. Habel and Schmitt (
2012,
2018) argued that specialised species, as a consequence of habitat loss and degradation, are commonly characterised by such low levels of genetic diversity accompanied by high genetic distinctiveness. However, they may persist in small-sized remnants of suitable habitats over long periods of time, just slightly affected by population bottlenecks and inbreeding due to their limited, but locally adapted genetic composition.
Regional biogeography and conservation history of the German exclave
Ellenberg and Leuschner (
2010) quote
Poa badensis among other steppe species as a relict, surviving on sandy/rocky grassland sites in Central Europe since the last glacial. However, they also remarked that most of the populations occurring today are likely younger (i.e. originating from Neolithic Age or early mediaeval times; cf. Willis et al.
1998). In Germany, relict localities of thermophilic species are described from dry regions like the Unstruttal (Becker
1999), the (Lower) Nahe region, aeolian deposits of large rivers (e.g. the north and east of the Rhinehessian plateau; Blaufuss and Reichert
1992; Witschel
1991) or south-exposed rock faces of low mountain ranges (e.g. the Swabian Alb; Wilmanns
2005). Our German study region is situated in two of these refugial areas, the Rhinehessian plateau and the adjacent Lower Nahe region. They are characterised by calcareous aeolian sands, deposited along the northern slopes of the plateau in glacial and early postglacial times (Jännicke
1892; Ellenberg and Leuschner
2010), as well as by the basic conglomerate of the so-called Waderner Schichten of the Upper Rotliegend Group (Weiss
1889).
In our study, we included four of the five mentioned populations from Rhineland-Palatinate by performing all analyses, which overall characterised the (north)westernmost exclave as genetically impoverished. However, there are two populations, Eierfels (G4) and Heidesheim (G3), showing comparatively high levels of genetic diversity. As their history is very different, we will go into more detail. The population Eierfels (G4) is situated on a flat ridge of basic conglomerate in the Lower Nahe region which is protected as natural monument and harbours a unique xerothermic vegetation (Korneck
1998; Oesau and Frahm
2007). The population of
P. badensis is rather small, but likely represents an old occurrence probably existing there since the early postglacial (cf. Blaufuss and Reichert
1992 and citation therein). In Heidesheim (G3) our study species grows secondary on a cemetery wall. This rather large and likewise old population persists there over a long period of time and represents one of the two still existing populations on the northern edge of the Rhinehessian plateau (i.e. beside the relict population in the Mainzer Sand). It seems possible that the species had arrived there during the construction of the wall with limestone mortar or, shortly after completion, via natural dispersal (e.g. by wind) from the close surrounding area, where the species previously occurred more frequently (cf. Blaufuss and Reichert
1992; Hodvina and Cezanne
2008).
Occurring in the close vicinity of Heidesheim (i.e. only 1.5 km apart), the population Weilersberg (G2) is characterised by a different and considerably younger history. This area was originally used as a sandpit and later designated as a nature reserve. To our knowledge,
P. badensis had been introduced to this site by taking (seed) material from the closely located population Heidesheim (G3). However, in our AFLP data we found the population Weilersberg grouping together with Eckelsheim and not with the assumed source of the seed transfer (Fig.
1). Therefore, we may assume that possibly a second gene pool has been used during the introduction, or that based on a heterogeneous source, lineages were sorted out differently given population bottlenecks. Considering this specific history, the extremely low genetic diversity observed in this large population is not surprising and likely reflects this artificial (recent) founder event. However, the reproductive success in general was not significantly reduced by this bottleneck. Geographically most isolated from the other occurrences of
P. badensis in Western Germany, the population Eckelsheim (G1) is again located on a rocky outcrop belonging to the Waderner Schichten as part of the Rhinehessian Plateau (Fig.
1). This isolated location is also reflected in our cpDNA results, where Eckelsheim represents one distinct haplotype within the German haplotype group (Fig.
2). The estimated genetic diversity of this population was by far the absolute lowest from all analysed populations, which we attribute to the obvious lack of gene flow and the very low population size (Table
1). Nevertheless, in combination with the not reduced reproductive success of this population, an adaptation to low genetic diversity as described by Habel and Schmitt (
2012;
2018) might be assumed.
Our AFLP data revealed admixture between the populations Heidesheim (G3) and Eierfels (G4; together representing the group GII, Fig.
1) and the German group GI. Due to small population size (Eckelsheim, G1) and recent translocation (Weilersberg, G2) the other two German populations missed such an admixture pattern. To a lesser extent we also found admixture between the German group GI and the Pannonian region, indicating an earlier genetic connection of the (north)westernmost exclave with the continuous Pannonian occurrences. Despite that, we observed that 123 (30.4%) of 404 regionally unique AFLP fragments (cf. Petit et al.
1998) were only present in the (north)westernmost exclave. Within the continuous Pannonian distribution area, 281 (69.6%) of these regionally unique AFLP fragments were found. In this context, having nearly one-third of unique AFLP fragments, the peripheral populations in Western Germany represent a considerable and unique part of the total species’ genetic variation (cf. Plenk et al.
2017). Moreover, we obtained evidence for the relict status and high conservation relevance of the Mainzer Sand population (G5) although it was represented by just four individual plants. The four individuals analysed were allocated to both detected haplotypes of the German exclave (cf. Figure
2 and Online Resource 4), demonstrating that this probably oldest relict population (Korneck and Pretscher
1984) might still hold the full spectrum of genetic diversity original to this region.
Therefore, we interpret the observed genetic composition, demographic traits and reproductive success within the German exclave as evidence of a long-term (successful) in situ survival, probably since early postglacial times, which we especially hypothesise for the populations Eckelsheim (G1), Eierfels (G4), and Mainzer Sand (G5). We moreover conclude, that the populations of the Waderner Schichten (e.g. Eierfels, Eckelsheim) may originate from the probably slightly earlier populated occurrences along the northern edge of the Rhinehessian plateau (e.g. Mainzer Sand and other previously existing ones) and spread there along the river systems of Nahe and subsidiary streams to the west and south (cf. Blaufuss and Reichert
1992; Hodvina and Cezanne
2008).
Nevertheless, we also have to consider that the species may have reached suitable sites in Western Germany more recently via a long-distance dispersal (LDD) event. In this case, we would expect low genetic diversity and less genetic differentiation of these founder populations from a potential source in more central/eastern areas (cf. Wróblewska
2008). Although we observed some admixture between the German and Pannonian region, our data clearly showed a genetic distinctiveness of the (north)westernmost exclave in both marker systems (cf. Figures
1,
2). We, therefore, concluded that the populations of the German exclave do not represent descendants of a recent LDD-event. However, analysing additional populations of
P. badensis from the few existing occurrences “linking” the German exclave and the Pannonian region may strengthen this conclusion.
Due to their rarity, peripheral populations are often under protection, although they are generally considered as genetically impoverished and less competitive (Lesica and Allendorf
1995). With our study on
P. badensis, we could clearly demonstrate that these assumptions are not applicable in general. However, due to the low abundance and the lack of connectivity, populations might be nevertheless under pressure if they are not (yet) adapted to low levels of genetic diversity (Habel and Schmitt
2012;
2018). Therefore, we must also consider a possible time lag (Lindborg and Eriksson
2004) between the decline of population sizes and the species’ reproductive success and genetic diversity (cf. Kropf
2012).