Declining Genetic Diversity and Increasing Genetic Isolation toward the Range Periphery of pennata, a Eurasian Feather Grass Author(s): Viktoria Wagner, Jan Treiber, Jiři Danihelka, Eszter Ruprecht, Karsten Wesche, and Isabell Hensen Reviewed work(s): Source: International Journal of Sciences, Vol. 173, No. 7 (September 2012), pp. 802-811 Published by: The University of Chicago Press Stable URL: http://www.jstor.org/stable/10.1086/666663 . Accessed: 19/09/2012 15:37

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

.

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to International Journal of Plant Sciences.

http://www.jstor.org Int. J. Plant Sci. 173(7):802–811. 2012. Ó 2012 by The University of Chicago. All rights reserved. 1058-5893/2012/17307-0006$15.00 DOI: 10.1086/666663

DECLINING GENETIC DIVERSITY AND INCREASING GENETIC ISOLATION TOWARD THE RANGE PERIPHERY OF STIPA PENNATA, A EURASIAN FEATHER GRASS

Viktoria Wagner,1,* Jan Treiber,y Jirˇi Danihelka,z Eszter Ruprecht,§ Karsten Wesche,y and Isabell Hensen*

*Institute of Biology/Geobotany and Botanical Garden, Martin Luther University Halle-Wittenberg, Am Kirchtor 1, D-06108 Halle, Germany; ySenckenberg Museum of Natural History, Go¨rlitz, P.O. Box 300 154, D-02806 Go¨rlitz, Germany; zDepartment of Vegetation Ecology, Institute of Botany, Academy of Sciences of the , Lidicka´ 25/27, CZ-60200 Brno, Czech Republic, and Department of Botany and Zoology, Faculty of Science, Masaryk University, Kotla´rˇska´ 2, CZ-61137 Brno, Czech Republic; and §Hungarian Department of Biology and Ecology, Babesx-Bolyai University, Str. Republicii 42, RO-400015 Cluj Napoca, Romania

A common assumption in ecology and evolutionary biology is that genetic diversity declines and differentiation increases toward the edge of a species’ geographic range, where populations tend to be smaller and more isolated. We tested these predictions in a characteristic Eurasian steppe plant, Stipa pennata,by inspecting 230 AFLP bands in 26 populations (345 individuals) along a 3300-km longitudinal gradient from the range core, in Russia, to the range periphery, in central . Overall, our study species showed low genetic diversity within populations (mean proportion of polymorphic bands ¼ 21:2%) and moderately high genetic differentiation among them (mean FST ¼ 0:29). As predicted, genetic diversity declined significantly from the range core to the periphery but was not correlated with population size. Pairwise genetic dif- ferentiation was significantly higher among peripheral populations than central populations but did not show a pronounced relationship with geographic distance. Our results indicate that peripheral populations may experience higher genetic drift and lower gene flow than their central counterparts, possibly because of smaller population sizes, spatial isolation, and a more complex landscape structure. In addition, historic range fluctuations and the mixed breeding system could have enhanced the observed patterns in our study species.

Keywords: abundant-center model, AFLP, fragmentation, geographic distribution range, range center.

Online enhancements: appendix tables.

Introduction at the range edge could lead to bottleneck events and, conse- quently, to decreased genetic diversity and increased genetic Genetic variation forms the basis for evolutionary pro- differentiation (Hewitt 1996; Hampe and Petit 2005). cesses and shapes a species’ adaptive potential and popula- Although the hypothesis of lower genetic diversity and tion viability (Reed and Frankham 2003; Boulding 2008). It higher genetic differentiation at the range edge has been sup- is commonly assumed that genetic variation is not evenly ported by theoretical considerations, it has received only distributed across a species’ geographic distribution range: mixed empirical support in the past (Eckert et al. 2008; Har- at the range edge, genetic diversity is expected to be smaller die and Hutchings 2010). Results that diverge from expecta- and genetic differentiation higher than in the range core tions have been explained by the fact that species do not (Hoffmann and Blows 1994; Eckert et al. 2008; Hardie and show an abundant-center distribution to begin with (Yaki- Hutchings 2010). Such a pattern could arise as a consequence mowski and Eckert 2008). Indeed, it has been pointed out of an abundant-center distribution, in which population size that the abundant-center model does not apply to a variety and abundance decline toward the range periphery because of organisms (Sagarin and Gaines 2002). However, even if of unfavorable environmental conditions and/or interactions species naturally show an abundant-center distribution, an- with new competitors, pathogens, or parasites at the range thropogenic fragmentation in the range core could increase edge (Brown 1984; Lawton 1993; Gaston 2009). The decline genetic drift, inbreeding, and genetic isolation among popula- in population size and stronger spatial isolation should in tions and distort the expected pattern. Meanwhile, biological turn lead to inbreeding, high genetic drift, and hampered traits, such as a species’ longevity and ability for long-distance gene flow (Ellstrand and Elam 1993; Vucetich and Waite dispersal, could counteract diversity loss and dampen genetic 2003). Furthermore, repeating expansions and contractions differentiation. We tested the hypothesis of decreasing genetic diversity 1 Author for correspondence; current address: College of Forestry and increasing genetic differentiation toward the range edge and Conservation, University of Montana, Missoula, Montana 59812, in Stipa pennata, one of the most characteristic of the U.S.A.; e-mail: [email protected]. western Eurasian steppes (Lavrenko 1970; Nosova 1973). Manuscript received December 2011; revised manuscript received April 2012. Dry grassland plants have received considerable attention by

802 WAGNER ET AL.—GENETIC DIVERSITY AND ISOLATION IN STIPA PENNATA 803 population geneticists in the past, given that many of these is not known. Caryopses are enclosed in the lemma and are plants are red-listed in Europe, including Anthericum liliago dispersed by wind or in animal fur. (Peterson et al. 2008), Astragalus exscapus (Becker 2003), Stipa pennata is native to the temperate zone of Europe and Iris aphylla (Wro´ blewska and Brzosko 2006), Silene chloran- Asia (fig. 1). In its geographic range core, at ;50°–55° latitude tha (Lauterbach et al. 2011), Silene otites (Lauterbach et al. in Russia (fig. 1), it grows in steppes, meadow steppes, and 2012), and Stipa capillata (Hensen et al. 2010). However, forest steppes (Nosova 1973). Steppes occur naturally in this few studies covered large areas across the species’ distribu- region (Adams and Faure 1997), but nomadic tribes have ex- tion range (but see Bylebyl et al. 2008; Wro´ blewska 2008). tended the steppe area since the Eneolithic Age by logging ad- Many dry grassland plants occur at the edge of their distribu- jacent forests, mowing, and livestock grazing (Chibilyov 2002; tion in central Europe but are more frequent in the steppes of Sarychev 2003). In the Middle Ages, when Slavic settlers intro- eastern Europe and Asia. Only one study has used a Eurasian duced agriculture in the region, steppes still covered vast areas dry grassland plant to directly compare genetic variation in (Sarychev 2003). It is only since the eighteenth century that the range core and periphery (Wagner et al. 2011, for S. ploughing resulted in a massive decline of steppe vegetation capillata). The categorical sampling scheme of this study has (Chibilyov 2002; Boonman and Mikhalev 2005). Given the been favored by the majority of other population genetic dramatic habitat loss, S. pennata and many other steppe spe- studies because of its cost-effectiveness and logistic feasibility cies are today red-listed even in their distribution core in Rus- (Eckert et al. 2008). However, it did not allow for broader sia (Golovanov 1988). generalizations beyond the two regions that were compared. At its western geographic periphery, in central Europe, S. To obtain more comprehensive results, we used a continuous pennata is confined to small dry grassland fragments that are sampling scheme across the distribution range of S. pennata, embedded in a matrix of forests, arable fields, and settlements. spanning a 3300-km longitudinal gradient from the geographic Steppe communities used to be naturally distributed during the range core in Russia to the range periphery in central Europe. late Pleistocene in and the Czech Republic (Jankov- We expected genetic diversity to be higher and genetic differen- ska´ and Pokorny´ 2008). Macrofossils of S. pennata s. lat. have tiation lower toward the range core in Russia, where the spe- been found as early as from the Holocene, but populations cies has historically been widely distributed. However, recent may have been subject to oscillations in the following millen- transformation of steppe habitats in Russia has resulted in in- nia because of climatic changes and anthropogenic activities creased fragmentation, which could have altered the predicted (Pott 1996; Poschlod and WallisDeVries 2002). Dry grassland mechanisms and could have led to similar patterns of genetic vegetation has been created by deforestation as recently as in the diversity and differentiation across the sampling gradient. Neolithic, Bronze, and Middle Ages (Ellenberg 1996; Opravil 1999; Bieniek 2002; Kohler-Schneider and Caneppele 2009). It has been maintained by mowing and grazing, but the aban- Material and Methods donment of traditional land use regimes in the last century led to a decline in dry grassland habitat in central Europe Study Species (Poschlod and WallisDeVries 2002). Stipa pennata is cur- Stipa pennata L. is the type species of the genus Stipa;its rently red-listed in several European countries, including the name was initially applied to any central and eastern Euro- Czech Republic (Holub and Procha´zka 2000) and Germany pean feather grass species with a plumose seta. Because of Lin- (Ludwig and Schnittler 1996). naeus’s ambiguous description (Linnaeus 1753), S. pennata was repeatedly rejected as a nomen ambiguum. In the second Sampling Scheme half of the nineteenth century, the name Stipa joannis Cˇ elak. was introduced for our study species. Martinovsky´ and Ska- In 2008, we collected fresh leaves in 26 S. pennata popula- licky´ (1969) proposed to apply the Linnaean name to the tions from five countries (Russia, , Romania, Czech taxon currently known as Stipa eriocaulis Borba´s (see Marti- Republic, and Germany; fig. 1). Our study was confined to novsky´ 1980), but their lectotypification was incorrect. Freitag populations of S. pennata s. str. and did not include popula- (1985) settled these disputes by designating a new lectotype; tions of similar taxa that are sometimes treated as its subspecies the name S. pennata is accepted by recent taxonomic mono- (e.g., Stipa borysthenica Prokudin, Stipa eriocaulis Borba´s, graphs and also here. Despite the nomenclatural ambiguities, Stipa pulcherrima K. Koch, and Stipa crassiculmis P. A. Smirn.; S. pennata can be well distinguished from other feather grasses Tsvelev 1976; Freitag 1985). Leaf samples were collected from by both generative and vegetative traits (Prokudin et al. 1977). different tussocks within 30 3 30-m plots, immediately dried, Stipa pennata is a perennial and tetraploid grass (2n¼44; andstoredinsilicagelforfurtheranalysis.Populationswere Prokudin et al. 1977; Krasnikov 1991; Sheidai et al. 2006). defined as a group of individuals that were separated by at least Tetraploid Stipa species are thought to be of ancient hybrid 1 km from the next group. Population size was estimated by origin and thus allopolyploid (Johnson 1945; Tsvelev 1977). counting the approximate number of flowering tussocks. The species does not form any spreading tillers; individual tussocks are thought to represent a genet. The inflorescence is composed of stalked spikelets bearing a single floret with AFLP Analysis a lemma armed with a conspicuous 22–37 cm long bigenicu- We used AFLP markers to analyze dominant cellular DNA late awn consisting of a glabrous columna and a plumose fingerprint patterns in our samples (Vos et al. 1995). Geno- seta. Flowers can be either cleistogamous or chasmogamous mic DNA extraction and subsequent AFLP analysis followed (anemophilous); the ratio of selfing versus outcrossing flowers the protocol of Hensen et al. (2012). After an initial primer 804 INTERNATIONAL JOURNAL OF PLANT SCIENCES

Fig. 1 Map showing the distribution of Stipa pennata (gray area). Study populations are shown as triangles. screening, we selected the following three combinations of tions are compared only in terms of their band (dis)similar- AFLP primer pairs that produced clear and polymorphic peaks ities (Michalski et al. 2010). This method sticks most closely for further analysis: *FAM 59-EcoRIþAAG-39/59-MseIþCCA-39, to the observed data but might overlook important genetic *HEX 59-EcoRIþAGC-39/59-MseIþCCA-39,and*HEX59- information. Furthermore, it does not generate standard pop- EcoRIþAGC-39/59-MseIþCAA-39. ulation genetic measures, rendering comparisons to other We scored peaks in Fragment Profiler (ver. 1.2; Amersham studies impossible. A second is the fragment-frequency ap- Biosciences), confining our analysis to peaks ranging between proach: allele frequency is estimated by equating it to the 60 and 400 bp in length and reaching 50 relative fluores- observed fragment frequency (Wagner et al. 2011). This ap- cence units. After exporting data as a 1/0 matrix, we com- proach generates common population genetic parameters that pared band presence and absence with original peak runs by can be compared across studies. However, it assumes fixed ho- eye. Ambiguous and nonrepeatable bands were excluded. The mozygosity. The third is the allele-frequency approach: the final data set comprised 230 polymorphic bands (86% of all frequency of the marker allele is estimated using methods de- scored fragments). The error rate (sensu Bonin et al. 2004) veloped for diploid species. Although formally invalid, this was 3.8%, as assessed in 24 repeated samples (7% of the total analysis is still more realistic than the previous one in that it sample size, repeated starting from the extraction). relaxes the assumption of fixed homozygosity. Furthermore, this approach can be justified on the basis of the disomic (or classic Mendelian) inheritance in allotetraploids that is simi- Statistical Analysis lar to diploids (Gallais 2003). Given these advantages and Most methods for estimating allele frequencies from domi- disadvantages, we employed all three methods. For the third nant markers assume that data originate from a diploid or- approach, we used a Bayesian estimation method with non- ganism. For tetraploid species, three common approaches uniform prior distribution (Zhivotovsky 1999). have been used. One is the band-based approach: allele- Genetic diversity. For the band-based approach, we cal- frequency estimation is abandoned altogether, and popula- culated the number of private bands (PrivB) and the propor- WAGNER ET AL.—GENETIC DIVERSITY AND ISOLATION IN STIPA PENNATA 805 tion of polymorphic bands (PPB), defined as the proportion (GER-2; table 1). However, population size was not signifi- of polymorphic bands among the total number of bands in cantly correlated with longitude (a proxy for the core-edge the data set, using the program FAMD (Schlu¨ ter and Harris gradient; Spearman’s r ¼ 0:4, P ¼ 0:07). Results for several 2006). Using the same program, we calculated average Jac- genetic diversity measures (PPB, JD, HE,FF, and HE,AF) were card dissimilarity (JD) among individuals within a population highly correlated (Pearson’s r > 0:9, P < 0:01). Thus, we re- (JD ¼ 1 J, with J indicating Jaccard similarity). The num- port only the results for PPB and the remaining, less corre- ber of rare bands (RB; occurring in <25% of the data set) lated variables (PrivB and RB; see table A2, available in the was assessed in GenAlEx (Peakall and Smouse 2006). Nei’s online edition of the International Journal of Plant Sciences, gene diversity (HE) was calculated in AFLP-SURV (Vekemans for all genetic diversity results). 2002), for both the fragment-frequency (HE,FF) and the allele- Mean genetic diversity across all populations was low frequency (HE,AF) approach. Population size was log trans- (PPB ¼ 21:2%; table 1). The proportion of polymorphic formed to achieve normality. Linear regression analyses were bands decreased significantly from the range core toward the performed in R (R Development Core Team 2010). periphery (fig. 2). The highest proportion of polymorphic Genetic differentiation. To test the hypothesis that differ- bands was found in the Russian population RUS-9 (32.6%), entiation is higher at the range edge than in the range core, we and the lowest was found in the German population GER-3 calculated pairwise Jaccard dissimilarity (for the band-based (9.1%). Neither private bands nor rare bands showed a signifi- approach) and FST distances (for the fragment-frequency and cant linear relationship with the core-edge gradient. Popula- allele-frequency approaches) among populations. Jaccard dis- tion size and population genetic diversity were not correlated similarity was first computed among individuals in a population (PPB: R2 < 0:001, P ¼ 1:0). by means of the vegan package in R (Oksanen et al. 2010) and then averaged for each compared pair of populations. Genetic Differentiation We used AFLP-SURV to obtain pairwise FST values among populations. Jaccard dissimilarity and FST distances were in- In the neighbor-net network, central populations in Russia spected by constructing neighbor-net networks in SPLITSTREE formed a cluster separate from the peripheral populations in (Huson and Bryant 2006). In this graph, parallel branches the Czech Republic and Germany (fig. 3). Populations from that form boxes indicate conflicting signals. For subsequent Romania were genetically intermediate between the two groups, analyses, data were divided into a group of central (Russia) whereas the Ukrainian population was placed among the and peripheral (Czech Republic, Germany) populations. To central European ones. Branches among central populations keep the two regions comparable in their geographic extent, were shorter than those among peripheral populations, indi- we included the Romanian populations in the range periph- cating lower genetic differentiation. The AMOVA revealed ery and omitted population UKR-1 (Ukraine). We performed that in the total data set 29% of the genetic variation was an analysis of molecular variance (AMOVA) in GenAlEx found among populations and 66% was found within popula- (Peakall and Smouse 2006) to inspect the partition of genetic tions, while only 5% was partitioned among core and periph- variation at different hierarchical levels. Significance of vari- eral regions (FPT ¼ 0:34; table 2). When analyzed separately, ance components was based on 999 generated permutations. peripheral populations showed a higher among-population ge- Using the same program, we computed FRT (an analogue of netic variance (40%, FPT ¼ 0:40) than central populations FCT), FPR (FSC), and FPT (FST). To directly compare genetic (16%, FPT ¼ 0:16). differentiation in the two regions, we assessed mean Jaccard Pairwise genetic dissimilarity and distance measures among dissimilarity and pairwise FST distance values among central populations (JD, FST,FF,andFST,AF)werecorrelated(r > 0:7, and peripheral populations with Welch’s t-test, which is ro- P < 0:01); therefore, we report only the results of one pairwise bust to unequal variances within groups. P values were calcu- distance measure (FST,AF). Mean pairwise genetic distance lated as the proportion of 1000 randomized runs in which was on average significantly higher among peripheral popula- simulated t values exceeded or were equal to observed ones. tions than central populations (0.316 vs. 0.108, t ¼ 14:3, In addition, we compared the variance in Jaccard dissimilar- P < 0:001; see table A3, available in the online edition of the ity and FST distance among peripheral and central popula- International Journal of Plant Sciences, for an FST,AF distance tions with Fisher’s F-test and by calculating P values as the matrix among all populations). Genetic distance among pe- proportion of 1000 randomized runs in which the region was ripheral populations was also more variable than that among found to yield a higher or equal variance than the other re- central populations (0.015 vs. 0.003, F ¼ 5, P < 0:001). Ge- gion. We performed a Mantel test in the vegan package to netic distance did not show a linear relationship with geo- test for genetic isolation by distance patterns. Prior to this graphic distance, as shown by a Mantel test for the range 2 analysis, FST values were linearized as FST=ðÞ1 FST , accord- core (linearized FST,AF: R ¼ 0:19, P ¼ 0:093) and the range 2 ing to Rousset (1997). periphery (linearized FST,AF: R ¼ 0:05, P ¼ 0:388).

Results Discussion

Genetic Diversity Preserving genetic variation is one of the most important Estimatedpopulationsizerangedbetween100and goals in conservation, given its influence on species’ perfor- 100,000 individuals, with the largest population located in mance and evolutionary processes (Boulding 2008). Genetic Russia (RUS-10) and the smallest population in Germany variation is predicted to be unevenly distributed across a spe- 806 INTERNATIONAL JOURNAL OF PLANT SCIENCES

Table 1 Overview of Stipa pennata Study Populations Population code Country Region NnPrivB PPB RB Range core: RUS-1 Russia Bashkortostan 5000 14 0 22.6 7 RUS-2 Russia Bashkortostan 1500 14 2 28.7 8 RUS-3 Russia Bashkortostan 300 13 1 27.4 11 RUS-4 Russia Bashkortostan 1000 14 0 21.7 8 RUS-5 Russia Lipetskaya Oblast’ 20,000 12 3 19.1 3 RUS-6 Russia Lipetskaya Oblast’ 50,000 15 1 23.5 6 RUS-7 Russia Lipetskaya Oblast’ 5000 11 1 20.0 5 RUS-8 Russia Kurskaya Oblast’ 3000 14 7 28.3 8 RUS-9 Russia Kurskaya Oblast’ 20,000 13 2 32.6 8 RUS-10 Russia Kurskaya Oblast’ 100,000 13 1 22.2 4 RUS-11 Russia Kurskaya Oblast’ 50,000 14 1 16.1 0 Intermediate range position: UKR-1 Ukraine Dnipropetrovs’ka Oblast’ 20,000 14 0 23.0 8 ROM-1 Romania Turda, Cheile Turzii 200 12 0 22.2 8 ROM-2 Romania Cluj, Faˆnatxele Clujului 400 14 0 19.6 7 Range periphery: CZ-1 Czech Republic District Brˇeclav 25 14 3 23.5 8 CZ-2 Czech Republic District Brˇeclav 400 13 11 21.3 5 CZ-3 Czech Republic District Brˇeclav 600 14 0 13.9 0 CZ-4 Czech Republic District Brˇeclav 500 13 3 28.7 12 CZ-5 Czech Republic District Brˇeclav 90 14 3 22.2 10 GER-1 Germany Saxony-Anhalt 3000 9 3 14.8 3 GER-2 Germany Saxony-Anhalt 100 9 2 20.0 6 GER-3 Germany Thuringia 3000 12 1 9.1 6 GER-4 Germany Thuringia 500 11 1 20.4 13 GER-5 Germany Hesse 500 12 0 12.2 5 GER-6 Germany Hesse 3000 11 3 18.3 7 GER-7 Germany Hesse 3000 10 1 19.1 7 Mean 11,197 13 2 21.2 7 Note. N ¼ population size, n ¼ sample size, PrivB ¼ number of private bands, PPB ¼ proportion of polymorphic bands, RB ¼ number of rare bands. See table A1 for more detailed information on the geographic localities of populations and table A2 for further genetic diversity measures, both available in the online edition of the International Journal of Plant Sciences. cies’ geographic range, with genetic diversity decreasing and (43%), S. lagascae (31%), and S. parviflora (43% [H. R. genetic differentiation increasing from the geographic range Hamasha, A. N. Schmidt-Lebuhn, W. Durka, M. Schleuning, core toward the range periphery (Eckert et al. 2008). We con- and I. Hensen, unpublished manuscript]). A causal relation- firmed this prediction in populations of Stipa pennata, a char- ship between the mixed breeding system and low to moder- acteristic Eurasian steppe grass, along a 3300-km longitudinal ate genetic diversity in Stipa species has also been suggested gradient from the range core to the range periphery. by Hensen et al. (2010) and Wagner et al. (2011). As expected, genetic diversity declined from the range core to the range periphery. Similar trends have been observed in Genetic Diversity several other plant species—for example, Corylus avelana Overall, our study species showed much lower within- (Persson et al. 2004), Gypsophila fastigiata (Lo¨nn and Pren- population genetic diversity (mean PPB ¼ 21:2%) than other tice 2002), Juncus atratus (Michalski and Durka 2007), Vera- perennial grasses, as evaluated by AFLP markers (e.g., Dacty- trum album (Treier and Mu¨ ller-Scha¨rer 2011), and three lis glomerata: 66%–74% [Peng et al. 2008]; Festuca cam- Viola subsect. rostratae species (Eckstein et al. 2006). Despite pestris: 89% [Fu et al. 2005]; see the supplement in Wagner its intuitive predictions, this concept is, however, far from be- et al. 2011 for further examples). The low overall genetic di- ing a biogeographic rule. In a review by Eckert et al. (2008), versity in our study species may be attributed to its mixed only 64% of all examined studies followed this prediction. breeding system (Hamrick and Godt 1996; Nybom and Bart- Similarly, Hardie and Hutchings (2010) found only 60% of ish 2000). Indeed, Stipa species are known for their faculta- all studies of plants to be in agreement with this model. The tive cleistogamous pollination (Ronnenberg et al. 2011); they high number of divergent cases points to the possibility that tend to self-pollinate, especially under drought conditions historic processes, including land use patterns, and biological (Brown 1952; Ponomarev 1961). AFLP studies of other Stipa traits may override spatial constraints across the geographic species have also revealed low to moderate levels of genetic range or that many species simply do not have an abundant- diversity: S. pulcherrima (5% [Nossol 2007]), S. capillata center distribution (Eckert et al. 2008; Yakimowski and Eckert (21% [Wagner et al. 2011]), S. arabica (36%), S. capensis 2008). WAGNER ET AL.—GENETIC DIVERSITY AND ISOLATION IN STIPA PENNATA 807

Although the largest S. pennata populations were found in the range core in Russia, there was no correlation between current population size and the core-edge gradient. Genetic di- versity was not correlated with population size, which can be attributed to different or fluctuating population sizes in the past or to the longevity of the species (Schiemann et al. 2000; Honnay et al. 2007). Meanwhile, genetic diversity significantly declined toward the range periphery. This discrepancy may be explained by the fact that Russian populations were larger and better connected in the past, before steppe was converted into agricultural land. By contrast, S. pennata populations in cen- tral Europe have been restricted for much longer times to rare sites, such as rock outcrops and south-facing slopes on calcium- rich soils. The fragmentation process of steppe habitats in Russia might have been too recent to reduce genetic diversity in a perennial grass. However, decreasing population sizes and poor connectivity in the range core of S. pennata has the potential to deteriorate genetic diversity in the future. Our results differ from the observations of a recent study in the grass species S. capillata (Wagner et al. 2011). In this study, genetic diversity did not decline toward the range periphery in central Europe compared with the range core in . This could be explained by the fact that European S. capillata populations were generally larger (>1000 individuals) than S. pennata populations. However, our study did not reveal a significant relationship between population size and genetic diversity in S. pennata. It is possible that our study species ex- perienced stronger range fluctuations at its western periphery in the past, where it has more specific habitat requirements than S. capillata. This may be particularly relevant if migration has been advanced by few long-distance dispersal events and asso- ciated bottleneck effects (Ibrahim et al. 1996).

Genetic Differentiation Our study species showed overall moderate among-population genetic differentiation (AMOVA FPT ¼ 0:34). This result is similar to values reported by AFLP studies for grasses with a mixed mating system—for example, Arrhenatherum elatius (FST ¼ 0:24 [Michalski et al. 2010]); S. arabica, S. capensis, S. lagascae, and S. parviflora (FST ¼ 0:57, 0.36, 0.60, and 0.39, respectively [H. R. Hamasha, A. N. Schmidt-Lebuhn, W. Durka, M. Schleuning, and I. Hensen, unpublished manu- script]); and S. capillata (FST ¼ 0:16, 0.4 [Wagner et al. 2011]). In the neighbor-net network, populations of Russia and central Europe were separated, indicating low genetic ex- change on a large spatial scale. Within both regions, genetic differentiation was not correlated with geographic distance, as shown by the Mantel test. This pattern could be explained by strong genetic drift that counteracts gene flow or by rare long- distance dispersal events. Our results corroborated our second initial hypothesis: geographically peripheral populations of S. pennata were ge- netically more differentiated than central populations. Higher

Fig. 2 Relationship between genetic diversity parameters and range edge to the range core, with central populations located at high longitude in Stipa pennata populations for number of private bands longitudes. The P value indicates whether the slope was significantly (A), number of rare bands (B), and proportion of polymorphic bands different from 0. The regression line in A and B is not shown, as the (C). The longitudinal gradient corresponds to a gradient from the slope was not significantly different from 0. 808 INTERNATIONAL JOURNAL OF PLANT SCIENCES

Fig. 3 Neighbor-net network of Stipa pennata populations based on pairwise FST values, as computed by the allele-frequency approach. genetic differentiation at the range edge was supported by drift among peripheral populations than central populations 70% of studies in the review by Eckert et al. (2008). A similar (Hutchison and Templeton 1999). Furthermore, the mixed pattern was detected by several other studies of, for example, breeding system of our study species may have accelerated ge- Corylus avellana (Persson et al. 2004), Geum triflorum netic differentiation at the range edge (Loveless and Hamrick (Hamilton and Eckert 2007), and Iris aphylla (Wro´ blewska 1984). 2008). In addition to the long-lasting spatial isolation, land- In conclusion, our study of S. pennata populations sup- scape barriers—such as forests, roads, agricultural fields, and ported the hypotheses that (1) genetic diversity declines and urban areas—could have hampered pollen and seed flow among (2) genetic differentiation increases toward the geographic peripheral populations of S. pennata in the past. More variable range periphery. Smaller historic population sizes, increased FST values at the range edge point to a stronger role of genetic spatial isolation, a more complex landscape structure, and

Table 2 Analysis of Molecular Variance for All Populations, Central Populations, and Peripheral Populations of Stipa pennata df SS MS Estimated variation % F All populations:

Among regions 1 143.6 143.6 .57 5*** FRT ¼ .05 Among populations 24 1120.4 46.7 3.01 29*** FPR ¼ .31 Within populations 319 2176.1 6.8 6.82 66*** ...

Total 344 3440.1 10.40 100*** FPT ¼ .34 Central populations:

Among populations 10 252.9 25.3 1.3 16*** FPT ¼ .16 Within populations 141 985.0 7.0 7.0 84*** ... Total 151 1237.9 8.3 100*** Peripheral populations:

Among populations 13 824.9 63.5 4.5 40*** FPT ¼ .40 Within populations 164 1082.3 6.6 6.6 60*** ... Total 177 1907.2 11.1 100*** Note. SS ¼ sum of squares, MS ¼ mean of squares. P < 0:001. WAGNER ET AL.—GENETIC DIVERSITY AND ISOLATION IN STIPA PENNATA 809 past range pulsation dynamics may all have led to the ob- Larisa Skolzneva; Nikolay Skolznev; Andrey Vlasov; the served pattern at the distribution periphery. These effects staff of the Zapovednik Galichya Gora, the Tsentral’no- may well have been accelerated by the mixed breeding sys- Chernozemnyy Zapovednik, and the Bashkiriya National tem of the species. The genetic distinctiveness of peripheral Park (Russia); Ivan Moysienko; Vasiliy Kucherevskiy; the S. pennata populations makes them a valuable target for con- staff of the Krivorozhskiy Botanical Garden (Ukraine); and servation. In the range core, populations show no evidence of Toby Spribille (United States). We are also thankful to Chris- deteriorating genetic variation, in spite of recent spatial frag- toph Rosche, Elke Seeber, and Birgit Mu¨ ller (Germany) for mentation. Here, the longevity of the study species may have their assistance during laboratory work. This work was delayed any loss of genetic diversity. Nonetheless, genetic supported by a travel grant to V. Wagner from the German monitoring should be implemented in the range core to deter- Academic Exchange Service (DAAD); grants to J. Danihelka mine whether fragmentation might have detrimental effects from the Ministry of Education of the Czech Republic on genetic variation in the long term. (MSM0021622416 and LC06073) and the Institute of Bot- any, Czech Academy of Sciences (long-term research plan Acknowledgments AV0Z60050516); and a grant to E. Ruprecht from CNCSIS- UEFISCSU (project PN II-RU TE 296 no. 71/29.07.2010). Support during fieldwork was kindly provided by Vasiliy The editor and two anonymous reviewers provided valuable Martynenko; Liliya Sultangareeva; Rafael Sultangareev; comments on the manuscript.

Literature Cited

Adams JM, H Faure, eds. 1997 Review and atlas of palaeovegeta- small population size: implications for plant conservation. Annu tion: preliminary land ecosystem maps of the world since the Last Rev Ecol Syst 24:217–242. Glacial Maximum. Oak Ridge National Laboratory, Oak Ridge, Freitag H 1985 The genus Stipa (Gramineae) in southwest and south TN. http://www.esd.ornl.gov/projects/qen/adams1.html. Asia. Notes R Bot Gard Edinb 42:355–489. Becker T 2003 Auswirkungen langzeitiger Fragmentierung auf Po- Fu YB, D Thompson, W Willms, M Mackay 2005 Long-term grazing pulationen am Beispiel der reliktischen Steppenrasenart Astragalus effects on genetic variability in mountain rough fescue. Rangel Ecol exscapus L. (Fabaceae). Diss Bot 380:1–210. Manag 58:637–642. Bieniek A 2002 Archaeobotanical analysis of some early Neolithic Gallais A 2003 Quantitative genetics and breeding methods in settlements in the Kujawy region, central , with potential autopolyploid plants. INRA Quae, Paris. plant gathering activities emphasised. Veg Hist Archaeobot 11: Gaston KJ 2009 Geographic range limits of species. Proc R Soc B 33–40. 276:1391–1393. Bonin A, E Bellemain, P Bronken Eidesen, F Pompanon, C Golovanov VD, ed 1988 Krasnaya kniga RSFSR: Rasteniya [Red Brochmann, P Taberlet 2004 How to track and assess genotyping data book of the Russian Soviet Federative Socialist Republic: errors in population genetic studies. Mol Ecol 13:3261–3273. plants]. Rosagropromizdat, Moscow. Boonman JG, SS Mikhalev 2005 The Russian steppe. Pages 381–416 Hamilton JA, CG Eckert 2007 Population genetic consequences of in JM Suttie, SG Reynolds, C Batello, eds. Grasslands of the geographic disjunction: a prairie plant isolated on Great Lakes world. Food and Agriculture Organization of the United Nations, alvars. Mol Ecol 16:1649–1660. Rome. Hampe A, RJ Petit 2005 Conserving biodiversity under climate Boulding EG 2008 Genetic diversity, adaptive potential, and pop- change: the rear edge matters. Ecol Lett 8:461–467. ulation viability in changing environments. Pages 199–219 in SP Hamrick JL, MJW Godt 1996 Effects of life history traits on Carroll, CW Fox, eds. Conservation biology: evolution in action. genetic diversity in plant species. Philos Trans R Soc B 351:1291– Oxford University Press, Oxford. 1298. Brown JH 1984 On the relationship between abundance and distribu- Hardie DC, JA Hutchings 2010 Evolutionary ecology at the extremes tion of species. Am Nat 124:255–279. of species’ ranges. Environ Rev 18:1–20. Brown WV 1952 The relation of soil moisture to cleistogamy in Stipa Hensen I, A Cierjacks, H Hirsch, M Kessler, K Romoleroux, D leucotricha. Bot Gaz 113:438–444. Renison, K Wesche 2012 Historic and recent fragmentation Bylebyl K, P Poschlod, C Reisch 2008 Genetic variation of Eryngium coupled with altitude affect the genetic population structure of campestre L. (Apiaceae) in Central Europe. Mol Ecol 17:3379– one of the world’s highest tropical tree line species. Glob Ecol 3388. Biogeogr 21:455–464. Chibilyov A 2002 Steppe and forest-steppe. Pages 248–266 in M Hensen I, C Kilian, V Wagner, W Durka, J Pusch, K Wesche 2010 Shahgedanova, ed. The physical geography of northern Eurasia. Low genetic variability and strong differentiation among isolated Oxford University Press, Oxford. populations of the rare steppe grass Stipa capillata L. in Central Eckert CG, KE Samis, SC Lougheed 2008 Genetic variation across Europe. Plant Biol 12:526–536. species’ geographical ranges: the central-marginal hypothesis and Hewitt GM 1996 Some genetic consequences of ice ages, and their beyond. Mol Ecol 17:1170–1188. role in divergence and speciation. Biol J Linn Soc 58:247–276. Eckstein RL, RA O’Neill, J Danihelka, A Otte, W Ko¨ hler 2006 Hoffmann AA, MW Blows 1994 Species borders: ecological and Genetic structure among and within peripheral and central pop- evolutionary perspectives. Trends Ecol Evol 9:223–227. ulations of three endangered floodplain violets. Mol Ecol 15:2367– Holub J, F Procha´zka 2000 Red list of vascular plants of the Czech 2379. Republic—2000. Preslia 72:187–230. Ellenberg H 1996 Vegetation Mitteleuropas mit den Alpen: in Honnay O, D Adriaens, E Coart, H Jacquemyn, I Roldan-Ruiz 2007 o¨ kologischer, dynamischer und historischer Sicht. Ed. 5. Ulmer, Genetic diversity within and between remnant populations of the Stuttgart. endangered calcareous grassland plant Globularia bisnagarica L. Ellstrand NC, DR Elam 1993 Population genetic consequences of Conserv Genet 8:293–303. 810 INTERNATIONAL JOURNAL OF PLANT SCIENCES

Huson DH, D Bryant 2006 Application of phylogenetic networks in Oksanen J, FG Blanchet, R Kindt, P Legendre, RB O’Hara, GL evolutionary studies. Mol Biol Evol 23:254–267. Simpson, P Solymos, MHH Stevens, H Wagner 2010 Vegan: Hutchison DW, AR Templeton 1999 Correlation of pairwise genetic community ecology package. Version 1.17-4. and geographic distance measures: inferring the relative influences Opravil E 1999 Archeologicke´ na´lezy kavylu na Moraveˇ [Archaeo- of gene flow and drift on the distribution of genetic variability. logical finds of Stipa in Moravia]. Praveˇk, NS, 9:153–157. Evolution 53:1898–1914. Peakall R, PE Smouse 2006 GENALEX 6: genetic analysis in Excel. Ibrahim KM, RA Nichols, GM Hewitt 1996 Spatial patterns of Population genetic software for teaching and research. Mol Ecol genetic variation generated by different forms of dispersal during Notes 6:288–295. range expansion. Heredity 77:282–291. Peng Y, X Zhang, Y Deng, X Ma 2008 Evaluation of genetic Jankovska´ V, P Pokorny´ 2008 Forest vegetation of the last full-glacial diversity in wild orchardgrass (Dactylis glomerata L.) based on period in the Western Carpathians (Slovakia and Czech Republic). AFLP markers. Hereditas 145:174–181. Preslia 80:307–324. Persson H, B Wide´n, S Andersson, L Svensson 2004 Allozyme Johnson BL 1945 Cyto-taxonomic studies in Oryzopsis. Bot Gaz diversity and genetic structure of marginal and central populations 107:1–32. of Corylus avellana L. (Betulaceae) in Europe. Plant Syst Evol 244: Kohler-Schneider M, A Caneppele 2009 Late Neolithic agriculture in 157–179. eastern Austria: archaeobotanical results from sites of the Baden and Peterson A, IV Bartish, J Peterson 2008 Effects of population size on Jevisˇovice cultures (3600–2800 B.C.) Veg Hist Archaeobot 18:61–74. genetic diversity, fitness and pollinator community composition in Krasnikov AA 1991 Chisla khromosom nekotorykh vidov sosudis- fragmented populations of Anthericum liliago L. Plant Ecol 198: tykh rasteniy iz Novosibirskoy oblasti [Chromosome numbers in 101–110. some species from the Novosibirsk region]. Bot Zh Ponomarev AN 1961 Kleystogamiya u kovyley [Cleistogamy in 76:476–479. feather grasses]. Bot Zh 46:1229–1236. Lauterbach D, M Ristow, B Gemeinholzer 2011 Genetic population Poschlod P, MF WallisDeVries 2002 The historical and socioeco- structure, fitness variation and the importance of population history nomic perspective of calcareous grasslands—lessons from the distant in remnant populations of the endangered plant Silene chlorantha and recent past. Biol Conserv 104:361–376. (Willd.) Ehrh. (Caryophyllaceae). Plant Biol 13:667–777. Pott R 1996 Die Entwicklungsgeschichte und Verbreitung xero- ——— 2012 Population genetics and fitness in fragmented popula- thermer Vegetationseinheiten in Mitteleuropa unter dem Einfluß tions of the dioecious and endangered Silene otites (Caryophylla- des Menschen. Tuexenia 16:337–369. ceae). Plant Syst Evol 298:155–164. Prokudin YN, AG Vovk, OA Petrova, ED Ermolenko, YV Verni- Lavrenko EM 1970 Provintsial’noe razdelenie prichernomorsko- chenko 1977 Zlaki Ukrainy [Grasses of the Ukraine]. Naukova kazakhstanskoy podoblasti stepnoy oblasti Evrazii [The division Dumka, Kiev. of the Black Sea–Kazakhstan subregion of the Eurasian steppe R Development Core Team 2010 R: a language and environment for region]. Bot Zh 55:609–625. statistical computing. Version 2.12.1. R Foundation for Statistical Lawton JH 1993 Range, population abundance and conservation. Computing, Vienna. Trends Ecol Evol 8:409–413. Reed DH, R Frankham 2003 Correlation between fitness and genetic Linnaeus C 1753 Species plantarum. Vol 1. Salvius, Holmiae. diversity. Conserv Biol 17:230–237. Lo¨ nn M, HC Prentice 2002 Gene diversity and demographic Ronnenberg K, I Hensen, K Wesche 2011 Contrasting effects of turnover in central and peripheral populations of the perennial precipitation and fertilization on seed viability and production of herb Gypsophila fastigiata. Oikos 99:489–498. Stipa krylovii in Mongolia. Basic Appl Ecol 12:141–151. Loveless MD, JL Hamrick 1984 Ecological determinants of genetic Rousset F 1997 Genetic differentiation and estimation of gene flow structure in plant populations. Annu Rev Ecol Syst 15:65–95. from F-statistics under isolation by distance. Genetics 145:1219– Ludwig G, M Schnittler 1996 Rote Liste gefa¨hrdeter Pflanzen 1228. Deutschlands. Bundesamt fu¨ r Naturschutz, Bonn. Sagarin RD, SD Gaines 2002 The ‘‘abundant centre’’ distribution: to Martinovsky´ JO 1980 Stipa L. Pages 247–252 in TG Tutin, VH what extent is it a biogeographical rule? Ecol Lett 5:137–147. Heywood, NS Burges, DM Moore, DH Valentine, SM Walters, DA Sarychev V 2005 Steppe communities in central Russia (Lipetsk Webb, eds. Flora Europaea. Vol 5, Alismataceae to Orchidaceae province): on the verge of destruction or a new life? Pages 1–18 in A (Monocotyledones). Oxford University Press, Oxford. Struchkov, J Kuleshova, eds. Facets of grassland restoration. Selected Martinovsky´ JO, V Skalicky´ 1969 Zur Nomenklatur einiger Stipa- papers from the International Field Seminar held at the Galichya Sippen der Pennatae-Gruppe XVI. Beitrag zur Kenntnis der euro- Gora Nature Reserve (Russia), 16–22 June 2003. Biodiversity Con- pa¨ischen Federgrassippen. Preslia 41:327–341. servation Center, Moscow. Michalski SG, W Durka 2007 High selfing and high inbreeding Schiemann K, T Tyler, B Wide´n 2000 Allozyme diversity in relation depression in peripheral populations of Juncus atratus. Mol Ecol to geographic distribution and population size in Lathyrus vernus 16:4715–4727. (L.) Bernh. (Fabaceae). Plant Syst Evol 225:119–132. Michalski SG, W Durka, A Jentsch, J Kreyling, S Pompe, O Schweiger, Schlu¨ ter PM, SA Harris 2006 Analysis of multilocus fingerprinting E Willner, C Beierkuhnlein 2010 Evidence for genetic differentia- data sets containing missing data. Mol Ecol Notes 6:569–572. tion and divergent selection in an autotetraploid forage grass Sheidai M, S Attaei, M Khosravi-Reineh 2006 Cytology of some (Arrhenatherum elatius). Theor Appl Genet 120:1151–1162. Iranian Stipa () species and populations. Acta Bot Croat 65: Nosova LM 1973 Floro-geograficheskiy analiz severnoy stepi evro- 1–11. peyskoy chasti SSSR [Floristic-geographic analysis of the northern Treier UA, H Mu¨ ller-Scha¨rer 2011 Differential effects of historical steppes in the European part of the USSR]. Nauka, Moscow. migration, glaciations and human impact on the genetic structure Nossol K 2007 Genetische Struktur und Reproduktion von Stipa and diversity of the mountain pasture weed Veratrum album.J pulcherrima K. Koch s. l. in Mitteleuropa. Diploma thesis. Biogeogr 38:1776–1791. University of Halle-Wittenberg. Tsvelev NN 1976 Zlaki SSSR [Grasses of the USSR]. Nauka, Nybom H, IV Bartish 2000 Effects of life history traits and sampling Leningrad. strategies on genetic diversity estimates obtained with RAPD markers ——— 1977 O proiskhozhdeniyu i evolyutsii kovyley [On the origin in plants. Perspect Plant Ecol, Evol, Syst 3:93–114. and evolution in feather grasses]. Pages 139–150 in ZV Karamy- WAGNER ET AL.—GENETIC DIVERSITY AND ISOLATION IN STIPA PENNATA 811

sheva, ed. Problemy ekologii, geobotaniki, botanicheskoy geografii i but no reduced genetic diversity in peripheral vs. central popula- floristiki. Nauka, Leningrad. tions of a steppe grass. Am J Bot 98:1173–1179. Vekemans X 2002 AFLP-surv 1.0: a program for genetic diversity Wro´ blewska A 2008 From the center to the margins of geographical analysis with AFLP (and RAPD) population data. Distributed by the range: molecular history of steppe plant Iris aphylla L. in Europe. author. Laboratoire de Ge´ne´tique et Ecologie Ve´ge´tale, Universite´ Plant Syst Evol 272:49–65. Libre de Bruxelles, Brussels. Wro´ blewska A, E Brzosko 2006 The genetic structure of the steppe Vos P, R Hogers, M Bleeker, M Reijans, T van de Lee, M Hornes, A plant Iris aphylla L. at the northern limit of its geographical range. Frijters, et al 1995 AFLP: a new technique for DNA fingerprinting. Bot J Linn Soc 152:245–255. Nucleic Acids Res 23:4407–4414. Yakimowski SB, CG Eckert 2008 Populations do not become less Vucetich JA, TA Waite 2003 Spatial patterns of demography genetically diverse or more differentiated towards the northern limit and genetic processes across the species’ range: null hypothesis of the geographical range in clonal Vaccinium stamineum (Erica- for landscape conservation genetics. Conserv Genet 4:639– ceae). New Phytol 180:534–544. 645. Zhivotovsky LA 1999 Estimating population structure in diploids Wagner V, W Durka, I Hensen 2011 Increased genetic differentiation with multilocus dominant DNA markers. Mol Ecol 8:907–913.