spatial genetic structure of aquilegia taxa endemic to the...

Spatial genetic structure of Aquilegia taxa endemic to the island of Sardinia

J. L. Garrido1,*, G. Fenu2, E. Mattana2 and G. Bacchetta2 1Estacio n Biolo gica de Don ana, Consejo Superior de Investigaciones Cientıficas (CSIC), Avenida Americo Vespucio, s/n,

41092, Sevilla, Spain and 2Centro Conservazione Biodiversita (CCB), Dipartimento di Scienze della Vita e dell’Ambiente, Universita degli Studi di Cagliari, Viale S. Ignazio da Laconi, 13, 09123, Cagliari, Italy

* For correspondence. E-mail [email protected]

† Background and Aims The Mediterranean Basin is one of the most important regions for the Earth’s plant biodiversity; however, the scarcity of studies on fine scale patterns of genetic variation in this region is striking. Here, an assessment is made of the spatial genetic structure of all known locations of the three Sardinian endemic species of Aquilegia in order to determine the relative roles of gene flow and genetic drift as underlying evolutionary forces canalizing the divergence of Sardinian Aquilegia taxa, and to see if the spatial genetic structure found fits the current taxonomic differentiation of these taxa. † Methods DNA from 89 individuals from all known locations of Aquilegia across Sardinia was analysed by means of amplified fragment length polymorphism (AFLP) markers. Both principal co-ordinates analysis (PCoA) and Bayesian clustering analyses were used to determine the spatial genetic structure irrespective of any taxonomic affiliation. Historical effects of gene flow and genetic drift were assessed by checking for the existence of isolation-by-distance patterns. † Key Results STRUCTURE and PCoA analyses revealed a pattern of genetic variation geographically structured into four spatial genetic groups. No migration – drift equilibrium was detected for Aquilegia in Sardinia, when analysed either as a whole or in individual groups. The scenario approached a Case III pattern sensu Hutchinson and Templeton, which is associated with extreme isolation conditions where genetic drift has historically played a dominant role over gene flow. † Conclusions The pattern of genetic variation of Sardinian taxa of Aquilegia indicates that genetic drift has been historically more influential than gene flow on population structure of Sardinian species of Aquilegia. Limited seed dispersal and divergent selection imposed by habitat conditions have been probably the main causes reinforcing post-Pleistocene geographical isolation of Aquilegia populations. The spatial genetic structure found here is not fully compatible with current taxonomic affiliations of Sardinian Aquilegia taxa. This is probably a consequence of the uncoupling between morphological and genetic patterns of differentiation frequently found in recently radiated taxa.

Key words: Mediterranean Basin, Sardinia Island, endemic species, Aquilegia barbaricina, Aquilegia nugorensis, Aquilegia nuragica, biodiversity hotspot, AFLP markers, spatial genetic structure, genetic drift, divergent selection, habitat specialization.

I N TR OD UCTI ON

The Mediterranean Basin is one of the most important regions for the Earth’s plant biodiversity (Medail and Myers, 2004; Thompson, 2005). The Tyrrhenian Islands (Sardinia and Corsica), together with the rest of the western Mediterranean Islands (Sicily and Balearic Islands), constitute one of the most important biodiversity hotspots within this region (Blondel and Medail, 2009). Large Mediterranean islands, due to high long-term climate stability, combine a high rich- ness of plant lineages and species with the persistence of many endemic taxa (Medail and Diadema, 2009). The out- standing levels of biodiversity and endemism turn this area into an optimal scenario for the study of the effects of spatial isolation on taxa diversification through the interplay between evolutionary forces and geographic isolation. As continental islands (sensu Whittaker, 1998) they are also considered as the best natural laboratories in which to study the effects of geographical isolation on allopatric speciation via selection and/or genetic drift (Bittkau and Comes, 2005).

Sardinia is, after Sicily, the largest island in the Mediterranean Sea. The Central, Eastern and Northern mountain ranges of the island have been identified as one of the 52 putative glacial refugia within the Mediterranean region (Medail and Diadema, 2009). Its flora consists of 2295 species (Conti et al., 2005) and 347 of these are endemics (e.g. Sardinian, Corso-Sardinian, Corso-Sardinian-Balearic endemics), with 45.8 % being exclusive Sardinian endemics (Bacchetta et al., 2005). Among the latter, three species of the genus Aquilegia are distributed across these mountain ranges: A. barbaricina, A. nugorensis and A. nuragica (Arrigoni and Nardi, 1977, 1978). The genus Aquilegia has become a model for the study of adaptive radiations from the knowledge acquired from American lineages (Hodges and Arnold, 1994; Hodges et al., 2004; Whittall and Hodges, 2007); however, the Euroasiatic counterpart of this model genus has been studied only recently. In a recent study by Bastida et al. (2010), phylogenetic analyses of Euroasiatic taxa revealed that the most recently radiated species are endemic to South European mountain ranges (dated approx. 1 million years ago). Among these species are the Sardinian

mailto:[email protected]�

A. barbaricina and A. nugorensis which currently may be involved in very active stages of radiation (Bastida et al., 2010).

Despite the widely recognized ecological and evolutionary relevance, the scarcity of studies on fine scale patterns of genetic variation of species inhabiting Mediterranean islands is striking (but see, Affre et al., 1997; Quilichini et al., 2004; Molins et al., 2009). Following theoretical expectations, endemic species including island endemics (Frankham, 1997) should exhibit lower levels of genetic diversity than widespread species (Hamrick and Godt, 1989; Gitzendanner and Soltis, 2000). Island endemic species worldwide usually exhibit spatially structured patterns of genetic variation [e.g. Nigella arvensis in the Aegean region (Bittkau and Comes, 2005) or Aster asa-grayi in Japan (Maki, 1999)]. High spatial structuring is assigned to geographical isolation coupled to the limited dispersal abilities of the species, circumstances that restrict gene flow among populations and result in subsequent population differentiation by drift and selection (see Edh et al., 2007). Evaluating the relative influences of gene flow and genetic drift on the distribution of genetic variation at different spatial scales provides invaluable insights into the evolutionary interpretation of how patterns of genetic variation have been historically established (Hutchison and Templeton, 1999).

The study of recently radiated taxa, such as the genus Aquilegia (Ro and McPheron, 1997; Bastida et al., 2010), often entails taxonomic and phylogenetic discrepancies, since it usually leads to the uncoupling of morphological and genetic patterns of differentiation (Wang et al., 2005; Whitfield and Lockhart, 2007; Parks et al., 2009; Rymer et al., 2010). Specifically, the rapid speciation which occurred in Aquilegia has led, in both the North American (Hodges and Arnold, 1994; Ro and McPheron, 1997; Whittall et al., 2006; Whittall and Hodges, 2007) and the Euroasiatic taxa (Bastida et al., 2010), to a high number of species with marked morphological differences but low genetic divergence. This circumstance makes the usual approaches of studying intra- and interspecific genetic variation among populations based on a priori defined species and populations difficult. Our approach to evaluate the spatial genetic structure does not consider any taxonomic affiliation.

In this study, we assess the spatial genetic structure of all known Sardinian populations of the genus Aquilegia to determine the relative roles of genetic drift and gene flow as underlying evolutionary forces that canalize the divergence of Sardinian Aquilegia taxa. Specifically, we asked the following questions. (1) How is the spatial pattern of genetic variation of Sardinian Aquilegia populations spatially structured irrespective of their taxonomic affiliation? (2) What is the main evolutionary force underlying this spatial genetic structure? (iii) Does the current taxonomic differentiation of Sardinian Aquilegia taxa fit to the spatial genetic structure found?

MATERI ALS A N D METH OD S

Study species, sites and plant material

Aquilegia species (Ranunculaceae) are perennial rhizomatous herbs widely distributed across Asia, Europe and North America (23, 21 and 22 taxa, respectively). Mature plants consist of an erect, usually pubescent stem growing from a

basal rosette with bi- or triternate compound leaves. Hermaphrodite, pendant flowers are radially symmetrical and exhibit five sepals and five petals, each petal consisting of a flat limb and a backwardly directed, nectar-secreting spur. Basal rosettes grow during April and May and develop one or several inflorescences, eventually flowering from May to August (Cullen and Heywood, 1964; Whittemore, 1997; Nold, 2003).

In Sardinia, three narrow endemic species have been described. Aquilegia barbaricina and A. nugorensis are similar in height (30 – 80 cm), while A. nuragica is usually smaller (20 – 45 cm). As in almost all Euroasiatic taxa of the genus, flowers are pale blue to purple, with the exception of the whitish ones of A. barbaricina. Flower size is larger in A. nugorensis (40 – 56 mm diameter) than in the other two species (25 – 30 mm). Spurs exhibit different degrees of curva- ture among species, being straighter in A. nugorensis (Arrigoni and Nardi, 1977, 1978).

Sardinian Aquilegia taxa are exclusively distributed across the central and eastern mountain ranges of the island, in the regions of Gennargentu massif, Tacchi and Supramontes (Fig. 1). They inhabit different environments; A. barbaricina appears in wet woodlands, meadows and stream margins, mainly occurring on siliceous substrates and secondarily on limestone. Aquilegia nugorensis inhabits wet rocky outcrops and appears mainly on limestone and secondarily on siliceous substrates. Finally, A. nuragica is known from a unique population on a wet limestone cliff extremely difficult to access (Arrigoni and Nardi, 1978). They are all highly threatened species, with scattered distributions and small population sizes. Aquilegia barbaricina and A. nuragica are listed both as ‘critically endangered’ in the IUCN red list (Camarda, 2006a, b; IUCN, 2010), and also as one of the 50 most endangered plants of the Mediterranean islands (de Montmollin and Strahm, 2007). Aquilegia nugorensis is included as ‘endangered’ in the Italian regional red list (Conti et al., 1997).

During June and July of 2009 and 2010, plant material from a total of 89 wild individuals was collected in all 19 locations where any species of Aquilegia is known to occur in the whole Sardinia Island (see Fig. 1 and Table 1 for further geographic and ecological details of sampling sites). Plant material samples were homogeneously distributed across population area and, in the case of proximity, a minimum 2 m distance between individuals was always taken into account to ensure sample independence. Fresh young leaves were immediately desiccated in the field with abundant silica gel at room temperature until DNA extraction. During the sampling, extreme care was taken to minimize the damage to these highly threatened species. DNA extraction and amplified fragment length polymorphism (AFLP) protocol

Total genomic DNA was extracted from approx. 30 mg of dried leaf material by means of the DNeasy Plant Mini Kit (Qiagen) and using the Quiacube extraction robot (Qiagen). In general, amplified fragments were obtained following the protocol of Vos et al. (1995) with some modifications related to the use of fluorescent dye-labelled selective

0 4

8 16

Monte Corrasi

5

Km

8 14

11 15

7

2

Orbisi

Gorropu 14 Canyons

15 19

8

18 11

19 1

18

7

Corrasi Gorropu Pischina

GTF (F) 6 4 Canyons

Sardinia

3

12 9 17

5 1 2 6 CCD 4

3

16 9

17

13 12

16

10 13

10

GTF (GT)

F IG . 1 . Study area and results of the genetic structure analyses. The 19 known locations of Aquilegia in Sardinia are distributed over the Central and Eastern mountain systems. Four locations at Monte Corrasi (North): Arco Corrasi (8), Punta Corrasi (11), Ahottadoglios (14) and Palumbrosa (15). Four Eastern locations: Gorropu Canyon (19), Pischina Urtaddala (18), Flumineddu (7) and Codula Orbisi (1). Five Central locations: Badde Enis (2), Talana (3), Rio Correboi (4), Rio Olai (5) and Monte Spada (6). Three locations at SE-Gennargentu Massif: Baccu Seardu (12), Su Furciddu (17) and Calavrighe Grossu (9). Finally, three locations in the Tacchi region: Monte Arquerı (10), Funtana Dorada (13) and Funtana Sa Ceraxia (16). Filled symbols represent populations already described as A. barbaricina (circles), A. nugorensis (squares) and A. nuragica (diamond). Open symbols (triangles) represent Aquilegia locations not previously described. Spatial genetic structure is inferred by a model-based clustering method implemented in STRUCTURE. At each location, pie charts indicate the mean proportion of membership of individuals at each location for the K ¼ 4 genetic groups. The right-hand bar diagram indicates membership of each individual for the 4 genetic

groups. Spatial genetic groups are indicated to the right of the figure.

primers and the use of a different restriction enzyme as rare cutter. Although EcoRI and MseI (Eco-AFLPs) are the most widely used restriction enzymes for the generation of AFLP markers, here we employ PstI and MseI (Pst-AFLPs) as rare and frequent cutter enzymes, respectively. Recent works suggest that PstI– MseI produces a higher number of bands, with a higher proportion of these bands being informative, a higher efficiency in detecting polymorphisms, a more random distribution of fragments along chromosomes instead of clustered in centromeric regions as in Eco-AFLPs, and clearer and more easily scorable fragments (Castiglioni et al., 1999; Vuylsteke et al., 1999; Young et al., 1999; Yuan et al., 2004; Ojha, 2005).

Digested DNA was ligated to the PstI and MseI adaptors. Fragments were selectively amplified with MseI and PstI primers bearing one selective nucleotide. The final selective amplification was carried out using PstI and MseI primers with two and three selective nucleotides primers, respectively. In total 48 selective primers combinations were assayed in a pilot study carried out on a random sub-sample of 15

individuals. Eight primer combinations were finally selected based on fragment abundance and polymorphism (Table 2). Fragment separation and detection were carried out in an ABI PRISM 3100 DNA automatic sequencer (Applied Biosystems). The presence or absence of fragments in each individual profile was scored with the software GeneMapper v.3.7 (Applied Biosystems).

To assess the reproducibility and reliability of our AFLP fragments, we replicated 25 % (18 individuals) of the samples. Instead of replicating just the AFLP production, we also replicated DNA isolations. Doing so provides more reliable replicates, since repeating only the process of AFLP production does not generate true replicates (Bonin et al., 2007; Holland et al., 2008). AFLP data analyses

Since there seem to be taxonomic inconsistencies in the delimitation of species in the Sardinian Aquilegia, we did not perform analyses separately for each taxon. To avoid

TA B LE 1. Localities, identification numbers and geographic co-ordinates of the sampled locations

Locality (Municipality, Co-ordinates Province) ID (N/E) Altitude Habitat Substrates Attributed taxa Collector Np Ni

Codula Orbisi (Orgosolo, NU)

1 408 08′ /098 29′ 800 – 885 Rocky outcrops Limestones NP AC, GF, EM 53 5

Badde Enis (Orgosolo, NU)

2 408 07′ /098 26′ 840 – 1000 Rocky outcrops Limestones/ granites

NP AC, GF, EM 76 5

Ponte Crobine (Talana, OG) Rio Correboi (Vilagrande Strisaili, OG)

3 408 02′ /098 24′ 870 – 930 Riparian wood of Alnus glutinosa

4 408 03′ /098 20′ 1100 – 1370 Riparian wood of A. glutinosa and Rhamnus persicifolia

Granites NP GB, RB, AC, GF

Metamorphytes A. barbaricina GB, RB, AC, GF

400 10 321 5

Rio Olai (Orgosolo, NU) 5 408 08′ /098 21′ 950 – 985 Riparian wood of A. glutinosa and R. persicifolia

Granites A. barbaricina GB, RB, EM 134 5

Monte Spada (Fonni, NU) 6 408 04′ /098 16′ 1340 – 1350 Riparian wood of A. glutinosa and Ilex aquifolium

Metamorphytes/ granites

A. barbaricina (LC)

GB, EM 110 5

Rio Flumineddu (Urzulei, OG)

7 408 09′ /098 28′ 738 – 780 Mesophylous wood and shady rocky outcrops

Limestones/ granites

NP GB, AC . 10

Arco Corrasi (Oliena, NU) 8 408 14′ /098 25′ 1370 – 1420 Shady rocky outcrops Limestones A. nugorensis AC, GF 156 5 Calavrighe Grussu (Vilagrande Strisaili, OG) Monte Arcuerı (Ussassai, OG)

9 408 00′ /098 21′ 1095 – 1105 Riparian wood of A. glutinosa

10 398 49′ /098 22′ 920 – 925 Mesophylous wood of Taxus baccata and I. aquifolium

Metamorphytes A. nugorensis RB, GF, EM 1209 5 Travertines NP GB, GF 19 5

Punta Corrasi (Oliena, NU)

11 408 14′ /098 25′ 1380 – 1410 Shady rocky outcrops Limestones A. nugorensis GB, EM 97 5

Baccu Seardu (Vilagrande Strisaili, OG)


Granites A. nugorensis GB, RB, EM 250 5

Funtana Dorada (Seui, OG)

13 398 52′ /098 22′ 890 – 915 Riparian wood of Ostrya carpinifolia and T. baccata

Travertines A. nugorensis GB, RB, AC, GF

152 5

Ahottadoglios (Oliena, NU)

14 408 14′ /098 25′ 1345 – 1355 Shady rocky outcrops Limestones A. nugorensis AC, GF 87 5

Palumbrosa (Oliena, NU) 15 408 14′ /098 25′ 1330 – 1332 Shady rocky outcrops Limestones A. nugorensis RB, EM 139 5 Funtana sa Cerasia (Seui, OG)

16 398 53′ /098 23′ 900 – 1060 Riparian wood of O. carpinifolia and I. aquifolium

Travertines A. nugorensis (LC)

GB, RB, AC, GF

553 5

Su Furciddu (Vilagrande Strisaili, OG)


Metamorphytes A. nugorensis RB, GF, EM 1016 5

Pischina Urtaddala (Urzulei, OG)

18 408 10′ /098 29′ 700 – 705 Shady rocky outcrops Limestones A. barbaricina AC, MD, EM

102 5

Gorropu Canyon (Urzulei-Orgosolo, OG-NU)

19 408 11′ /098 30′ 620 – 680 Shady rocky outcrops Limestones A. nuragica (LC)

GBR, AC, FF, GF, EM

61 6

Altitude, habitat and substrate characteristics are also shown. The attributed taxa for already known populations are reported, as well as new populations discovered in this study (NP). Locations 4 and 5 attributed to

A. barbaricina by Arrigoni (2006). Location 6 described as A. barbaricina by Arrigoni and Nardi (1977) and location 18 attributed to the same taxa by Fenu et al. (2010). Locations 8, 11, 14 and 15 attributed to A. nugorensis by Fenu et al. (2010). Locations 9, 12, 13, 16 and 17 described as A. nugorensis by Arrigoni and Nardi (1978). Location 19 described as A. nuragica by Arrigoni and Nardi (1978).

LC, the locus classicus for each species; Np, population size; Ni, the number of sampled individuals (direct field count). Collectors abbreviations: G. Bacchetta (GB), G. Broi (GBR), R. Buttau (RB), A. Congiu (AC), M. Daws (MD), F. Fele (FF), G. Fenu (GF) and E. Mattana (EM).

possible fragment size homoplasy (Vekemans et al., 2002) we evaluated the correlation between locus frequency and fragment size by means of the program AFLP-surv 1.0 (Vekemans, 2002). Only fragments within the 150 – 500 bp size range were finally considered since the non-significant correlation indicated that they were not prone to size homoplasy (Herrera and Bazaga, 2008).

Genetic diversity within locations was estimated by the total number of markers, percentage of polymorphic loci ( pPL) and Nei’s gene diversity (H; Nei, 1987). Three parameters of marker rarity provided indirect estimates of population divergence and isolation: unique markers (markers present only in one population; Mpr), diagnostic markers (unique markers

present in all the individuals of a population; D) and frequency-down-weighted marker values (DW; Scho nswetter and Tribsch, 2005). Since populations which have been isolated for a long time accumulate rare markers by mutations, estimates of marker rarity may indicate divergence and isolation of populations. Specifically, DW marker values may be a standardized measure of divergence and long-term isolation (Scho nswetter and Tribsch, 2005). Several software packages were used to calculate these parameters; H and DW were determined by means of R Statistical Software v.2.11.1 (R Development Core Team, 2010) and AFLPdat (Ehrich, 2006), while pPL, Mpr and D were obtained using FAMD v.1.25 (Schlu ter and Harris, 2006). To avoid the possible

,

TAB LE 2. Primer combinations used, number of markers (loci) obtained in the size range 150 – 500 bp, percentage polymorphic loci at the 99 % level ( pPL; i.e. percentage of polymorphic loci lying within the range 0.01 – 0.99) and scoring error rates, in the amplified fragment length polymorphism (AFLP) analysis of the 89 individual Aquilegia plants considered in this study across

the whole island

Primer combination No. of markers pPL Scoring error rate (%)*

PstI + AA/MseI + CCC 79 72.15 0.91 PstI + AC/MseI + CAC 60 83.33 0.64 PstI + AG/MseI + CAT 82 79.26 0.67 PstI + AT/MseI + CCC 73 86.30 1.06 PstI + AA/MseI + CAT 71 70.42 1.72 PstI + AC/MseI + CTT 92 84.78 0.96 PstI + AG/MseI + CCC 60 73.33 2.22 PstI + AT/MseI + CAT 73 78.08 2.13 Total 590 Average 73.75 88.1 1.29

*Scoring error rate: [number of misscorings on two completely

independent analyses/(number of markers × number of individuals)] × 100, for 18 individual replicates.

effect of unequal sampling size among populations, all these measures were calculated for four randomly selected samples per location (Scho nswetter and Tribsch, 2005; Ronikier et al., 2008), except at locations 2, 4 and 10, with two, two and three samples, respectively.

To explore the possible effect of population size on genetic diversity, we performed a simple regression of the number of Aquilegia individuals per population on the estimate of Nei’s genetic diversity and pPL (arcsine square-root transformed).

To explore the genetic relationships among individuals and assess the spatial genetic structure of the sampled Aquilegia across the island, we used two different approaches. First, we analysed the Euclidean genetic distances among individuals with a principal co-ordinates analysis (PCoA) by means of the software Genalex v.6.4 (Peakall and Smouse, 2006). Secondly, we performed a Bayesian clustering of individual multilocus AFLP profiles by means of the software STRUCTURE v.2.2.3 (Pritchard et al., 2000; Falush et al., 2003, 2007; Hubisz et al., 2009). This program provides the important analytical advantage of not requiring any pre- defined delimitation of genetic populations. To perform this analysis, we used an admixture ancestry model and correlated allele frequencies (Pritchard et al., 2000). We performed 50 runs for each K value ( from 1 to 19), each with a length of 105 MCMC iterations and a burn-in period of 5 × 104. To evaluate the similarity among runs and assess whether clustering solutions for each K value were consistent, we calculated similarity coefficients using the R-script Structure-sum-2009 (Ehrich et al., 2007). To detect the true number of genetic groups, we used the modal DK parameter, proposed by Evanno et al. (2005), which was estimated as 4. For this K value, runs gave consistent results and individuals were clearly assigned to non-empty groups To combine the results of different runs and represent STRUCTURE outputs, we used the programs CLUMPP v.1.1.2 (Jakobsson and Rosenberg, 2007) and DISTRUCT v.1.1 (Rosenberg, 2004), respectively. Genetic covariance among STRUCTURE

groups was also explored through a population graph by means of the software GENETICSTUDIO (Dyer, 2009). The analyses of the resulting network topology (see Garrick et al., 2009; Baker and Dyer, 2011) did not reveal additional details improving the current interpretation of results, and are shown in Supplementary Data Fig. S1.

Additionally, we quantified the genetic differentiation among and within the observed genetic groups from the PCoA and STRUCTURE by performing hierarchical analyses of molecular variance (AMOVAs; Excoffier et al., 1992) with Arlequin v.3.5.1.2 (Excoffier and Lischer, 2010). An additional AMOVA was used to quantify the molecular variation under the current taxonomic affiliation of samples.

To explore the relative historical effects of gene flow and genetic drift on the regional structure defined over the whole distribution range of the genus Aquilegia in Sardinia, we checked for the existence of isolation-by-distance (IBD) patterns (Wright, 1943; Slatkin, 1993) and compared them with the model proposed by Hutchison and Templeton (1999). This model permits the evaluation of the relative historical influences of gene flow and drift on regional population structure by comparing the observed relationship between the genetic and geographic distances separating pairs of populations in a region with those expected under a stepping-stone model of population structure (Hutchison and Templeton, 1999). This approach seems particularly suitable to the system under study where several species may be occurring. To reduce potentially confounding effects of geographical range sub-structure on IBD patterns, the comparisons were performed both across the whole distribution range and for each spatial genetic group (cf. Bittkau and Comes, 2005). IBD patterns were checked by regressing pairwise FST

values and geographic distances, and significance was assessed by Mantel tests using 9999 random permutations in Genalex 6.4 (Peakall and Smouse, 2006). Using this approach, accept- ance of the null hypothesis of regional drift – migration does not rely just on the mere significant positive and monotonic correlation between genetic and geographic distances. Additionally, variance in differentiation should increase with distance (Hutchison and Templeton 1999; Griffin and Barrett, 2004). Thus, we also regressed the residuals from the former regressions on the pairwise geographic distances.

RESU LT S

Primer combinations averaged a very low scoring error rate (1.29 %, Table 2) which stressed the suitability and repeatabil- ity of our AFLP data set. Overall, 590 loci were consistent and unambiguously scored for 89 individuals from all known 19 locations occurring in Sardinia. A total of 519 loci (88 %) were found to be polymorphic, and primer combinations resolved on average 73 loci. Genetic diversity and marker rarity across populations

In general, genetic diversity measures within locations were low as estimated by pPL and H, which averaged 13.23 and 0.0281, respectively (Table 3). These measures ranged from extremely low values at location 19 (1.02, 0.0016), to the highest values found at location 16 (19.38, 0.0401). Average

01 11 02 12 03 13 04 14 05 15 06 16 07 17 08 18 09 19 10

Axi

s 2

(23·

2 %

)

TA B LE 3. Estimates of genetic diversity and marker rarity for Sardinian populations of Aquilegia

Spatial genetic groups Locality ID pPL H Mpr (D) DW

Corrasi Arco Corrasi 8 11.11 0.0213 (+0.1048) 2 19.63

Punta Corrasi 11 9.25 0.0182 (+0.0987) 1 19.72 Ahottadoglios 14 12.38 0.0247 (+0.1144) 0 21.78 Palumbrosa 15 13.69 0.0281 (+0.1231) 1 21.18 Average (+s.d.) 11.60 (+1.89) 0.0231 (+0.1102) 1 (+0.81) 20.57 (+1.07)

Gorropu Gorropu Canyon 19 1.02 0.0016 (+0.0290) 2 (2) 34.44 CCD Codula Orbisi 1 16.74 0.0320 (+0.1292) 3 19.64

Badde Enis* 2 8.41 0.0286 (+0.1670) 2 10.43 Ponte Crobine 3 16.82 0.0334 (+0.1330) 3 22.37 Rio Correboi* 4 11.96 0.0421 (+0.2011) 3 10.60 Rio Olai 5 17.45 0.0326 (+0.1272) 5 (1) 25.44 Monte Spada 6 17.56 0.0351 (+0.1336) 1 (1) 24.08 Pischina Urtaddala 18 12.03 0.0224 (+0.1055) 3 (2) 32.21 Average (+s.d.) 14.42 (+3.60) 0.0323 (+0.1423) 3 (+1.63) 20.68 (+7.93)

GTF Flumineddu 7 16.81 0.0362 (+0.1422) 1 24.41 Calavrighe Grussu 9 17.74 0.0387 (+0.1439) 2 33.18 Monte Arquerı* 10 2.45 0.0056 (+0.0610) 1 17.15 Baccu Seardu 12 16.88 0.0345 (+0.1313) 3 (1) 33.37 Funtana Dorada 13 13.30 0.0255 (+0.1136) 3 19.82 Funtana Ceraxia 16 19.38 0.0401 (+0.1434) 3 (1) 26.29 Su Furciddu 17 16.43 0.0340 (+0.1353) 2 20.76 Average (+s.d.) 16.75 (+6.43) 0.0349 (+0.1349) 2.33 (+0.79) 26.30 (+10.48)

Spatial genetic groups in which the whole distribution can be clustered (see Results) are also indicated. pPL, percentage of polymorphic loci; H, Nei’s gene diversity; Mpr, markers unique for each population; D, diagnostic markers, occurring in all the

individuals in a population; DW, frequency-down-weighted marker values. Average values of each parameter for the locations at each spatial genetic group are also indicated. Parameters estimated for populations marked with asterisks, excluded from average values, should be considered cautiously due to sub-sampling (see the

Materials and Methods).

1·0

0·5

0·0

–0·5

–1·0

–1·5

–2·0

–2·5

–1·5 –1·0 –0·5 0·0 0·5 1·0 Axis 1 (34·7 %)

F IG . 2 . Principal co-ordinates analysis of Sardinian Aquilegia individuals based on Euclidean distances of AFLP phenotypes. Axis 1 (34.7 %) and axis 2 (23.2 %) segregated populations into four groups: Monte Corrasi (upper- left cloud), Gorropu Canyon (lower cloud), Pischina Urtaddala (right small cloud) and a huge group with Central, Eastern, Gennargentu Massif and Tacchi populations (large upper-right cloud). Each symbol represents a

different location.

values for the estimates of marker rarity, Mpr and DW, were 2.15 and 22.97, respectively. Unique markers ranged from none at location 14 to five at location 5. DW values ranged from 19.63 at location 8 to 34.44 at location 19. Locations 2, 4 and 10 presented very low DW values but, due to a smaller sampling size, these values should be considered with caution. Six locations presented diagnostic markers, i.e.

unique markers characterizing all of their individuals. Specifically, locations 19 and 18 had two of these markers each, and locations 5, 6, 12 and 16 showed one diagnostic marker each.

Regression analyses performed to explore the possible effect of population size on the estimates of genetic diversity (H and pPL) only revealed a marginally positive effect on the levels of H (R2 ¼ 0.27, P ¼ 0.04 for H and R2 ¼ 0.14, P ¼ 0.16 for pPL). Spatial structuring of genetic variation

Four genetic clusters could be clearly recognized in the PCoA analysis (Fig. 2). The main axis strongly segregated locations 8, 11, 14 and 15, all situated at Monte Corrasi, which explained 34.7 % of the total genetic variation. The sec- ondary axis, which explained 23.2 % of the total genetic variation, segregated location 19 (Gorropu Canyon) and location 18 (Pischina Urtaddala). The last cluster grouped the rest of individuals without a clear genetic structure inside.

STRUCTURE analyses provided further structuring insights into the PCoA pattern. The occurrence of four distinct genetic lineages in the Aquilegia of Sardinia was the most likely partition found in the analysis (Fig. 1): Monte Corrasi (locations 8, 11, 14, 15; Corrasi group, hereafter); Gorropu Canyon (location 19; Gorropu group, hereafter), both groups identified in PCoA; central (locations 1 – 6, 18; ‘CCD’ group, hereafter); and SE-Gennargentu/Tacchi + Flumineddu (locations 9, 12, 17/10, 13, 16 + 7; ‘GTF’ group, hereafter). It is also note- worthy that within CCD, individuals from location 18

Corrasi/Gorropu/CCD Among groups

/GTF 3 795.80 31.66

TAB LE 4. Analyses of molecular variance of Sardinian

populations of Aquilegia Fig. 3B). For the GTF group (excluding the Flumineddu population, located apart from the group), correlation was not significant (rM ¼ – 0.200, P ¼ 0.127; Fig. 3D). Finally,

S ource of variation d.f.

Sum of squares

% Total variance FSC FST FCT

correlation for the Gorropu group could not be performed since it was defined by a single population. Correlation between residuals and geographical distance was not signifi- No groups 0.70

Among populations 18 1733.39 70.03 Within populations 70 567.91 29.97

0.59 0.72 0.31

Among populations 15 937.60 40.90 Within populations 70 567.91 27.45

cant in any case (data not shown), suggesting the same scatter- ing degree over all the range of geographical separations.

Despite the positive and significant correlation for CCD and Corrasi, in these groups variance in genetic differentiation did not increase with distance. Therefore, the results reject the null

Taxonomic structuring (A. barbaricina, A. nugorensis, A. nuragica) Among groups 2 527.13 28.42 Among populations 14 942.82 45.49 Within populations 59 469.76 26.09

0.63 0.73 0.28 hypothesis of regional drift – migration equilibrium, both for the whole island and for all spatial genetic groups.

DI SCUSSI ON

Genetic variation was partitioned into the groups of populations resulting from PCoA and STRUCTURE analyses. Partition without considering any grouping is also shown. All P-values were at least ,0.0003.

(Pischina Urtaddala; Pischina, hereafter), segregated by PCoA, showed a particularly high genetic admixturing.

All detected groups were spatially cohesive except for GTF. The Flumineddu population was located fairly far apart from the rest of the populations, which in turn were spatially segregated in three locations at SE-Gennargentu (9, 12 and 17) and three in the Tacchi region (10, 13 ande 16) (see Fig. 1).

Genetic differentiation among groups

Estimates of genetic diversity and marker rarity across genetic groups supported the results obtained by the PCoA and Bayesian grouping analyses and reflected strong genetic differentiation among groups (Table 3).

The AMOVAs showed significant differences in the proportion of variance explained at all levels and for all grouping schemes tested (all P-values ,0.0003, Table 4). Dealing with the four spatial genetic groups (Corrasi, Gorropu, CCD and GTF), genetic variation among groups was high (31.66 %, Table 4). Grouping populations according to their taxonomic affiliation resulted in a lower genetic variation among groups (28.42 %). Irrespective of the grouping scheme, genetic variation among populations within groups was always higher than variation among groups, suggesting that historical rates of gene flow among populations have been limited, probably as a consequence of stronger selection processes among populations than among groups.

Spatial genetic structure

To test the null hypothesis of regional drift – migration equilibrium, correlations of genetic distance (FST) and its variance vs. geographic distance were performed, both for the whole Sardinian geographical range and for each spatial genetic group.

Over the whole of Sardinia, pairwise FST values were not significantly correlated with geographic distances (rM ¼ 0.038, P ¼ 0.312; Fig. 3A). For the CCD group, correlation was significantly positive (rM ¼ 0.686, P ¼ 0.003; Fig. 3C). The Corrasi group exhibited a positive, but just marginally significant correlation (rM ¼ 0.893, P ¼ 0.042;

Genetic diversity

Our analyses across the whole distribution range of the three Sardinian exclusive and endangered taxa of Aquilegia revealed extremely low levels of genetic diversity. This result is in accordance with the general expectation that endemic species, and particularly island endemics (Frankham, 1997), exhibit lower levels of genetic diversity than widespread species (Hamrick and Godt, 1989, 1996).

First suggested by Stebbins (1942), it has been almost uni- versally assumed that endemic species show reduced genetic variation compared with widespread congeners, but this relationship is not apparent nowadays. Gitzendanner and Soltis (2000) provided only slight evidence (or contradictory evidence) for differences in genetic diversity between rare and widespread plant species, and concluded that, although there is an overall trend associating rare species with low levels of genetic variation, endemic species do not define a homoge- neous group of reduced genetic diversity (but see Nybom et al., 2004). In the Mediterranean Basin, empirical results also show this controversy, with both lower (e.g. Affre and Thompson, 1997a; Affre et al., 1997; Toumi and Lumaret, 2001) and higher (e.g. Affre and Thompson, 1997b; Mateu-Andres, 1999) levels of genetic diversity observed in endemics than in their related widespread congeners.

Several causes have been proposed to explain the reduced genetic variation of endemic species (Thompson, 2005). Endemic species are usually specialized to local ecological conditions, and natural selection may reduce variation by eliminat- ing non-adapted alleles (Babbel and Selander, 1974; Brown and Schoen, 1992). Additionally, small population sizes, founder effects and genetic bottlenecks may also induce high levels of inbreeding and subsequent reductions in genetic variability (Barrett and Kohn, 1991; Ellstrand and Elam, 1993).

Here, we found a weak effect of population size on diversity estimates (marginal on H and not significant on pPL), which does not seem strong enough to be the main cause of the observed reduced genetic diversity. Across Sardinia, Aquilegia populations are found in different and restricted habitat types, ranging from riverside meadows to rocky outcrops (Mattana et al., 2012). Reduced diversity values of Sardinian Aquilegia might be the consequence of local adaptation (divergent selection) to particular environmental conditions, with selection eroding genetic variability. In this sense, the available

4 6 8 10 12 14 16 18 20 0 5 10 15 20 Geographic distance (km) Geographic distance (km)

F ST

F ST

F ST

F ST

1·0

A Whole of Sardinia 0·8

B Mount Corrasi

0·9

0·8

0·7

0·7 0·6

0·6

0·5

0·5

0·4

0 10 20 30 40 50 Geographic distance (km)

0·4 0·1 0·2 0·3 0·4 0·5 0·6

Geographic distance (km)

0·8 C CCD

0·8 D GTF

0·7 0·7

0·6 0·6

0·5 0·5

0·4 0·4

F IG . 3 . Pairwise FST values among populations plotted against the straight-line distance separating them with its regression line for (A) the whole set of Sardinian populations and for spatial genetic groups where this comparison was possible. (B) Corrasi, (C) CCD and (D) GTF. The slope was significantly different from 0 at CCD and marginally significant at Corrasi. Locations 18 (Pischina) and 7 (Flumineddu) were excluded from this representation due to their genotypic

(Pischina) or geographic (Flumineddu) segregation from the rest of the individuals of their respective groups (CCD and GTF, respectively).

phylogeny of the Euroasiatic taxa of the genus establishes that habitat specialization and geographic isolation were the main drivers of species divergence in the process of adaptive radiation of the Euroasiatic taxa of the genus (Bastida et al., 2010). Furthermore, Alcantara et al. (2010) have found divergent selection imposed by environmental habitat parameters on phenotypic divergence between Aquilegia species and on ecotypic differentiation within species.

Geographical structure of the pattern of genetic variation

STRUCTURE results from individual multilocus profiles, also endorsed by PCoA analyses, revealed that the pattern of genetic variation of Sardinian endemic taxa of Aquilegia is structured geographically into four spatial genetic groups: Gorropu, Corrasi, CCD and GTF. However, AMOVAs revealed that, although differences within populations still explained a high proportion, the highest amount of genetic variation was explained by differences among populations. In general, our results reveal a pattern of genetic variation spatially structured under a classical perspective of a priori-

defined populations and also under the ‘blind’ individual- based approach by STRUCTURE.

High levels of spatial genetic structure have been described for other island endemics, such as Brassica insularis (Hurtrez-Bousses, 1996), Cyclamen balearicum (Affre et al., 1997), Anchusa crispa (Quilichini et al., 2004) and Senecio rodriguezii (Molins et al., 2009), in the Western Mediterranean sea; Nigella arvensis complex (Bittkau and Comes, 2005) and Brassica cretica (Edh et al., 2007) in the Aegean sea; Silene struthioloides and S. hawaiiensis in the Hawaiin archipelago (Westerbergh and Saura, 1994); and Aster asa-grayi in Japan (Maki, 1999). In our case, the level of spatial genetic structuring is likely to be the consequence of geographical isolation, probably as the result of the interplay between the complex Sardinian topography and geology and the variation of climate characteristics within the island, and limited dispersal abilities of the species. These circumstances promote restricted gene flow among populations and subsequent differentiation by drift and selection (Edh et al., 2007). Major paleogeographic and paleoclimatic processes may be discarded here due to the recent dating of this group of Aquilegia taxa. European Aquilegia taxa began to diversify

around 2.54 million years ago. Of these, the clade including Sardinian taxa is the most recent (Bastida et al., 2010).

Gene flow and drift within spatial genetic groups

In fact, we have detected no migration – drift equilibrium for the Aquilegia in Sardinia (neither when analysed as a whole nor when analysed in individual groups), approaching a Case III pattern sensu Hutchinson and Templeton (1999) (Fig. 3A). This scenario is associated with extreme isolation conditions where drift historically has played a dominant role in the absence of significant gene flow.

Some degree of association between genetic and geographic distances can be found even in the absence of regional equilibrium especially at shorter geographic distances and, in general, over geographical ranges where gene flow may be acting to some extent (Hutchison and Templeton, 1999). This seems to be the case for the Corrasi and CCD groups. The Corrasi group is made up of four locations situated over a reduced geographic range (,1 km2). This proximity makes it problematic to draw an inference of the roles of gene flow and genetic drift as per Hutchinson and Templeton’s Cases. Despite this, Corrasi showed a significant positive relationship between pairwise FST values and geographic distance which indicates a higher relevance of gene flow over genetic drift. Pollen flow might be the main agent connecting these locations. Bumble-bees usually forage at distances of 400 – 600 m, reach- ing altitudes of 800 – 1400 m (Walther-Hellwig and Frankl, 2000; Osborne et al., 2008), which is in the range of distances separating these populations (581 m).

The genetic group CCD, which encompasses central populations of the Island, approximates a positively sloped Case II, where patterns of isolation by distance are beginning to form, with gene flow reducing differentiation among neighbouring populations. This scenario suggests two hypothetical evolutionary trajectories: (1) an emerging Case IV pattern where gene flow will exert a predominant influence, eventually defining a range of higher separation distances where genetic drift will predominate; or (2) an emerging Case III pattern where environmental and/or geographic conditions enhance isolation levels, and genetic drift will start to predominate over gene flow.

The group GTF approximated a typical Case III pattern, with a wide divergence in the scatter and a non-significant correlation between genetic and geographic distances. This suggests a high degree of isolation between the three locations of SE-Gennargentu (9, 12 and 17), highly differentiated with nearly no admixture, and the three locations at Tacchi region (10, 13 and 16), where genotypes from other genetic groups ( particularly from Corrasi) are present.

Gorropu and Pischina are both constituted by a single population and cannot be evaluated in the present context. Nevertheless, Gorropu is highly isolated spatially. It presents no admixture and only shares some genetic information with Pischina. Thus it is probably subject to high levels of genetic drift.

The heavily admixtured location of Pischina harbours genetic profiles from Gorropu, CCD and GTF ( probably from Flumineddu location 7). Gene flow is presumably playing a relevant role in its genetic composition.

Thus, as in other geographic areas and island-related species (e.g. Bittkau and Comes, 2005), our evidence indicates that

genetic drift has historically been more influential than gene flow on the population structure of the Sardinian species of Aquilegia. However, in some areas with less isolated populations, such as the group CCD, the influence of gene flow has also played a relevant role. Differentiation of spatial genetic groups

The Aquilegia population at Gorropu Canyon is probably the most geographically isolated in Sardinia. It hangs 25 – 50 m away from the soil level, over one of the walls of a deep rocky canyon. The nearest Aquilegia location, Pischina Urtaddala, is 1.67 km upstream, in a temporal water course flowing to Gorropu Canyon. It exhibits the highest DW marker value and a complete fixation of unique markers (i.e. exclusive markers fixed for each one of the individuals; Table 3), which indicates a long-term isolation of individuals. Further, it shows no admixture from any other genetic group and only shares some genetic information with Pischina. The latter suggest a genetic link from Gorropu to Pischina where seed dispersal could be responsible for this connection. The distance separating these locations (1.67 km) is at the upper limit of the distance that pollinators can cover in their foraging activity, and further topographic bar- riers may magnify this straight line separation distance (Herrera and Bazaga, 2008). Seed dispersal by ungulates is the most prob- able cause. Wild European mouflon (Ovis musimon) and domes- tic sheep, goats and cows are abundant in these mountain systems (Ciuti et al., 2009). Although long-distance dispersal of dry-fruited shrubs by ungulates has been traditionally assumed to entail the complete destruction of seeds, recent evidence indicates that herbivore endozoochory may be of potential relevance for colonization of distant sites, because some seeds may survive ungulate digestion (Manzano et al., 2005; Manzano and Malo, 2006).

The genetic admixture shown by individuals at the Pischina Urtaddala location from CCD, Gorropu and GTF groups is particularly interesting. During the wettest time of the year, waters from Orbisi canyon reach Pischina Urtaddala (a small pond) and then flow into the Flumineddu river, just before Gorropu Canyon. This scheme reflects to some extent the pivotal geographic location interconnecting the Gorropu group (discussed above) with GTF and CCD. Genotypes from GTF probably reached Pischina from Flumineddu (location 7) and genotypes from CCD probably from Codula Orbisi (location 1) situated upstream, to which it is connected by river flow during the wettest season. The particular admixture scheme and the pivotal position in the spatial confluence of different spatial genetic groups make this location an important one for hybridization; a circumstance that will be discussed in the next section when we incorporate taxonomic information.

Individuals from Corrasi, representing four locations, show no evidence of admixture. However, Corrasi genotypes do occur in other groups of the island, especially within the central location 4 (Rio Correboi, CCD) and within the southern location 16 (Funtana Sa Ceraxia, GTF). This observation is consistent with more recent gene flow between these localities than is seen with CCD and GTF. The spread of Corrasi genotypes to other areas may be due to the aforementioned seed dispersal events by different ungulates.

GTF genotypes are spread across all central locations situated

from 3 to 18 (Fig. 1). That is, these genotypes occur in all CCD locations (1 – 6), in Pischina (18) and in Flumineddu (7; whose high proportion of GTF genotypes led it to be clustered within the GTF genetic group). This high occurrence of GTF genotypes in all locations, except Gorropu and Corrasi (a circumstance particularly evident at Flumineddu), may also be explained by seed dispersal events by ungulates. An alternative but tentative explanation could be an ancestral generalized occurrence of this group of genotypes across the majority of the current distribution range of Aquilegia in Sardinia, except Corrasi. Nowadays it is relegated to SE-Gennargentu, Tacchi, Flumineddu and Pischina (which in turn may be subject to dispersal events from Flumineddu, as discussed earlier).

Potential implications for taxonomy

The results obtained here from the analyses of AFLP multilocus individual profiles are not fully compatible with the current taxonomic affiliations of Sardinian Aquilegia taxa. Our analyses revealed the existence of four genetic lineages in the Aquilegia genus in Sardinia, while current taxonomic treatments (Arrigoni and Nardi, 1977, 1978) consider three species of Aquilegia in Sardinia: A. barbaricina, A. nugorensis and A. nuragica (Fig. 1). Aquilegia barbaricina is described from locations 4 – 6 and 18, with location 6 (Monte Spada) being the locus classicus (Arrigoni and Nardi, 1977). Aquilegia nugorensis is described from locations 8, 9 and 11 – 17, with location 16 being the locus classicus (Arrigoni and Nardi, 1978). Aquilegia nuragica is uniquely described from population 19 (Arrigoni and Nardi, 1978). Detailed taxonomic insights are beyond the scope of this article but, given the genetic evidence presented here, the spatial genetic structure does not match the taxonomically defined entitites.

This genetic/taxonomic uncoupling is reflected by the AMOVAs which indicated that the proportion of variance explained when adopting the taxonomic grouping was lower than that obtained when considering the spatial genetic groups from the STRUCTURE results (28.4 % vs. 34.7 %).

The group GTF includes all populations currently ascribed to A. nugorensis (9, 12 – 13 and 16 – 17; Fig. 1) with the exception of populations at Monte Corrasi, that constitute a well differentiated spatial genetic group. This group shows a high correspondence to what is currently described as A. nugorensis, including the two populations not previously described.

The CCD group includes all the populations of what is currently ascribed to A. barbaricina (4 – 6 and 18), and three populations not formerly known (1 – 3). Our analyses agree with this view and also consider that the three populations not previously described should also be potentially ascribed to this taxon, given also the high morphological similarity of their individuals to those from locations 4 – 6. Location 18, Pischina Urtaddala, exhibits a high genotypic contribution of this CCD cluster, attributed to A. barbaricina, but also presents important contributions from Gorropu, the group associated with A. nuragica, and from the group corresponding to A. nugorensis (see below), which may indicate that this location could be subject to recent interspecific introgressive hybridization involving three different lineages. Future research should be devoted to the study of genetic diversity at this specif- ic location.

Locations of the Corrasi group, described as A. nugorensis (Arrigoni, 2006), constitute a well differentiated spatial genetic group, distinct from the rest of the A. nugorensis populations. The strong segregation signal exhibited in the STRUCTURE analyses suggests that the designation of the populations of Monte Corrasi as A. nugorensis deserves further study, as previously suggested by Fenu et al. (2010).

Gorropu constitutes the most well-differentiated group in the Island. Its unique population (19) was described as A. nuragica (Arrigoni and Nardi, 1978). The fact that it is described as a single-population species, together with evidence commented on in preceding sections, suggest extremely high values of isolation and low genetic diversity. These observations help to explain its problematic conservation status (‘critically endangered’; IUCN, 2010). Concluding remarks

The only phylogenetic study available to date of the Euroasiatic taxa of Aquilegia (Bastida et al., 2010) establishes that habitat specialization and geographic isolation are the main drivers of species divergence in the process of adaptive radiation of these taxa. Furthermore, Alcantara et al. (2010) have found divergent selection imposed by environmental habitat parameters on phenotypic divergence between several south European species of Aquilegia and on ecotypic differentiation within them. Evidence presented here also points towards this. Across Sardinia, Aquilegia occurs in different and restricted habitat types, ranging from riverside meadows to rocky outcrops (Mattana et al., 2012), and also from limestone to granites and metamorphic substrates (Table 1). This intra- and interspecific diversity of habitats may indicate that habitat specialization by divergent selection may have been a very relevant force contributing to population differentiation and species divergence, although this needs to be specifically studied and demonstrated. In addition, overall reduced levels of genetic diversity, as commented on at the beginning of Discussion, are also in agreement with this possibility, since divergent selection would be eroding genetic variability.

Together with divergent selection imposed by habitat environments, the limited dispersal ability of these species would also contribute to reinforce post-Pleistocene geographical isolation of Sardinian Aquilegia populations. This ecological and evolutionary scenario determines the historical dominant influence of genetic drift over gene flow on population structure of Sardinian species of Aquilegia.

SUPPLEMENTA R Y DATA

Supplementary data are available online at www.aob.oxford- journals.org and consist of Figure S1: population graph analysis conducted on the four spatial genetic groups pre- defined by STRUCTURE.

AC KNO WLED GEMENTS

We thank Rafael G. Albaladejo and Carlos M. Herrera for helpful comments on earlier drafts. Two anonymous reviewers provided valuable suggestions to improve the paper. Additionally, we also benefited from the valuable help

http://www.aob.oxford-/�

http://www.aob.oxford-/�

provided by C. Perez and E. Lo pez with lab work and also from the helpful advice of P. Bazaga. A. Congiu and R. Buttau helped in the field work, and two rock-climbers, F. Fele and G. Broi, provided us with the plant material from the otherwise inaccessible population at Gorropu Canyons. J.L.G. also thanks A. Sariego for providing crucial keys during the work. We also thank the Remote Sensing and Geographic Information Systems Laboratory of Estacio n Biolo gica de Don ana (LAST-EBD). All bioinformatic analyses regarding STRUCTURE were carried out on the freely available Bioportal (www.bioportal.uio.no). The work was partially supported by grant PIE-200730I012 (Consejo Superior de Investigaciones Cientıficas) and by the RNM-154 research group (Junta de Andalucıa).

LITE RATURE CITED

Affre L, Thompson JD. 1997a. Variation in the population genetic structure of two Cyclamen species on the island of Corsica. Heredity 78: 205 – 214.

Affre L, Thompson JD. 1997b. Population genetic structure and levels of inbreeding depression in the Mediterranean island endemic Cyclamen creticum (Primulaceae). Biological Journal of the Linnean Society 60: 527 – 549.

Affre L, Thompson JD, Debussche M. 1997. Genetic structure of continental and island populations of the Mediterranean endemic Cyclamen balearicum (Primulaceae). American Journal of Botany 84: 437 – 451.

Alca ntara JM, Bastida JM, Rey PJ. 2010. Linking divergent selection on vegetative traits to environmental variation and phenotypic diversification in the Iberian columbines (Aquilegia). Journal of Evolutionary Biology 23: 1218 – 1233.

Arrigoni PV. 2006. Flora dell’Isola di Sardegna. Sassari: Delfino Carlo Editore and C. snc.

Arrigoni PV, Nardi E. 1977. Le piante endemiche della Sardegna: 1. Bollettino della Societa Sarda di Science Naturali 16: 265 – 268.

Arrigoni PV, Nardi E. 1978. Le piante endemiche della Sardegna: 19 – 20. Bollettino della Societa Sarda di Science Naturali 17: 215 – 225.

Babbel GR, Selander RK. 1974. Genetic-variability in edaphically restricted and widespread plant species. Evolution 28: 619 – 630.

Bacchetta G, Iiriti G, Pontecorvo C. 2005. Contributo alla conoscenza della flora vascolare endemica della Sardegna. Informatore Botanico Italiano 37: 306 – 307.

Baker SA, Dyer RJ. 2011. Invasion genetics of Microstegium vimineum (Poaceae) within the James River Basin of Virginia, USA. Conservation Genetics 12: 793 – 803.

Barrett SCH, Kohn JR. 1991. Genetic and evolutionary consequences of small population size in plants: implications for conservation. In: Falk DA, Holsinger KE. eds. Genetics and conservation of rare plants. New York: Oxford University Press, 3 – 30.

Bastida JM, Alca ntara JM, Rey PJ, Vargas P, Herrera CM. 2010. Extended phylogeny of Aquilegia: the biogeographical and ecological patterns of two simultaneous but contrasting radiations. Plant Systematics and Evolution 284: 171 – 185.

Bittkau C, Comes HP. 2005. Evolutionary processes in a continental island system: molecular phylogeography of the Aegean Nigella arvensis alli- ance (Ranunculaceae) inferred from chloroplast DNA. Molecular Ecology 14: 4065 – 4083.

Blondel J, Medail F. 2009. Biodiversity and conservation. In: Woodward JC. ed. The physical geography of the Mediterranean. Oxford: Oxford University Press, 615 – 650.

Bonin A, Ehrich D, Manel S. 2007. Statistical analysis of amplified fragment length polymorphism data: a toolbox for molecular ecologists and evolu- tionists. Molecular Ecology 16: 3737 – 3758.

Brown AHD, Schoen DJ. 1992. Plant population genetic structure and biological conservation. In: Sandlund OT, Hindar K, Brown ADH. eds. Conservation of the biodiversity and sustainable development. Oslo: Scandinavian University Press, 88 – 104.

Camarda I. 2006a. Aquilegia barbaricina. In: IUCN red list of threatened species. Version 2010.1. Available from: www.iucnredlist.org. IUCN. [Accessed 30 May 2010].

Camarda I. 2006b. Aquilegia nuragica. In: IUCN red list of threatened species. Version 2010.1. Available from: www.iucnredlist.org. IUCN. [Accessed 30 May 2010].

Castiglioni P, Ajmone-Marsan P, van Wijk R, Motto M. 1999. AFLP markers in a molecular linkage map of maize: codominant scoring and linkage group distribution. Theoretical and Applied Genetics 99: 425 – 431.

Ciuti S, Pipia A, Grignolio S, Ghiandai F, Apollonio M. 2009. Space use, habitat selection and activity patterns of female Sardinian mouflon (Ovis orientalis musimon) during the lambing season. European Journal of Wildlife Research 55: 589 – 595.

Conti F, Manzi A, Pedrotti F. 1997. Liste rosse regionali delle piante d’Italia. Camerino: WWF Italia, Societa Botanica Italiana.

Conti F, Abbate G, Alessandrini A, Blasi C, Bonacquisti S, Scassellati E. 2005. La flora vascolare italiana: ricchezza e originalita a livello nazio- nale e regionale. In: Scoppola A, Blasi C. eds. Stato delle conoscenze sulla flora vascolare d’Italia. Roma: Palombi Editori, 18 – 22.

Cullen J, Heywood VH. 1964. Aquilegia L. In: Tutin TG, Heywood VH, Burgess NA et al. eds. Flora Europea. Cambridge: Cambridge: University Press, 238 – 240.

Dyer RJ. 2009. GeneticStudio: a suite of programs for spatial analysis of genetic-marker data. Molecular Ecology Resources 9: 110 – 113.

Edh K, Widen B, Ceplitis A. 2007. Nuclear and chloroplast microsatellites reveal extreme population differentiation and limited gene flow in the Aegean endemic Brassica cretica (Brassicaceae). Molecular Ecology 16: 4972 – 4983.

Ehrich D. 2006. AFLPdat: a collection of R functions for convenient handling of AFLP data. Molecular Ecology Notes 6: 603 – 604.

Ehrich D, Gaudeul M, Assefa A, et al. 2007. Genetic consequences of Pleistocene range shifts: contrast between the Arctic, the Alps and the East African mountains. Molecular Ecology 16: 2542 – 2559.

Ellstrand NC, Elam DR. 1993. Population genetic consequences of small population size: implications for plant conservation. Annual Review of Ecology and Systematics 24: 217 – 242.

Evanno G, Regnaut S, Goudet J. 2005. Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Molecular Ecology 14: 2611 – 2620.

Excoffier L, Lischer HEL. 2010. Arlequin suite ver 3.5: a new series of programs to perform population genetics analyses under Linux and Windows. Molecular Ecology Resources 10: 564 – 567.

Excoffier L, Smouse PE, Quattro JM. 1992. Analysis of molecular variance inferred from metric distances among DNA haplotypes – application to human mitochondrial-DNA restriction data. Genetics 131: 479 – 491.

Falush D, Stephens M, Pritchard JK. 2003. Inference of population structure using multilocus genotype data: linked loci and correlated allele frequencies. Genetics 164: 1567 – 1587.

Falush D, Stephens M, Pritchard JK. 2007. Inference of population structure using multilocus genotype data: dominant markers and null alleles. Molecular Ecology Notes 7: 574 – 578.

Fenu G, Mattana E, Congiu A, Bacchetta G. 2010. The endemic vascular flora of Supramontes (Sardinia), a priority plant conservation area. Candollea 65: 347 – 358.

Frankham R. 1997. Do island populations have less genetic variation than mainland populations? Heredity 78: 311 – 327.

Garrick RC, Nason JD, Meadows CA, Dyer RJ. 2009. Not just vicariance: phylogeography of a Sonoran Desert euphorb indicates a major role of range expansion along the Baja peninsula. Molecular Ecology 18: 1916 – 1931.

Gitzendanner MA, Soltis PS. 2000. Patterns of genetic variation in rare and widespread plant congeners. American Journal of Botany 87: 783 – 792.

Griffin SR, Barrett SCH. 2004. Post-glacial history of Trillium grandiflorum (Melanthiaceae) in eastern North America: inferences from phylogeogra-

phy. American Journal of Botany 91: 465 – 473. Hamrick JL, Godt MJW. 1989. Allozyme diversity in plant species. In:

Brown ADH, Clegg MT, Kahler AL, Weir BS. eds. Plant population genetics, breeding and genetic resources. Sunderland, MA: Sinauer, 43 – 63.

Hamrick JL, Godt MJW. 1996. Effects of life history traits on genetic diversity in plant species. Philosophical Transactions of the Royal Society B: Biological Sciences 351: 1291 – 1298.

Herrera CM, Bazaga P. 2008. Adding a third dimension to the edge of a species’ range: altitude and genetic structuring in mountainous land- scapes. Heredity 100: 275 – 285.

http://www.bioportal.uio.no/�

http://www.iucnredlist.org/�

http://www.iucnredlist.org/�

Hodges SA, Arnold ML. 1994. Columbines – a geographically widespread

species flock. Proceedings of the National Academy of Sciences, USA 91: 5129 – 5132.

Hodges SA, Fulton M, Yang JY, Whittall JB. 2004. Verne Grant and evolutionary studies of Aquilegia. New Phytologist 161: 113 – 120.

Holland BR, Clarke AC, Meudt HM. 2008. Optimizing automated AFLP scoring parameters to improve phylogenetic resolution. Systematic Biology 57: 347 – 366.

Hubisz MJ, Falush D, Stephens M, Pritchard JK. 2009. Inferring weak population structure with the assistance of sample group information. Molecular Ecology Resources 9: 1322 – 1332.

Hurtrez-Bousses S. 1996. Genetic differentiation among natural populations of the rare Corsican endemic Brassica insularis Moris: implications for conservation guidelines. Biological Conservation 76: 25 – 30.

Hutchison DW, Templeton AR. 1999. Correlation of pairwise genetic and geographic distance measures: inferring the relative influences of gene flow and drift on the distribution of genetic variability. Evolution 53: 1898 – 1914.

IUCN. 2010. The 2010 IUCN red list of threatened species. Species information. http://www.redlist.org.

Jakobsson M, Rosenberg NA. 2007. CLUMPP: a cluster matching and per- mutation program for dealing with label switching and multimodality in analysis of population structure. Bioinformatics 23: 1801 – 1806.

Maki M. 1999. Genetic diversity in the threatened insular endemic plant Aster asa-grayi (Asteraceae). Plant Systematics and Evolution 217: 1 – 9.

Manzano P, Malo JE. 2006. Extreme long-distance seed dispersal via sheep. Frontiers in Ecology and Environment 4: 244 – 248.

Manzano P, Malo JE, Peco B. 2005. Sheep gut passage and survival of Mediterranean shrub seeds. Seed Sciences Research 15: 21 – 28.

Mateu-Andres I. 1999. Allozymic variation and divergence in three species of Antirrhinum L-(Scrophulariaceae-Antirrhineae). Botanical Journal of the Linnean Society 131: 187 – 199.

Mattana E, Daws MI, Fenu G, Bacchetta G. 2012. Adaptation to habitat in Aquilegia species endemic to Sardinia (Italy): seed dispersal, germination and persistence in the soil. Plant Biosystems 145: in press. http://dx.doi. org/10.1080/11263504.2011.557097.

Medail F, Diadema K. 2009. Glacial refugia influence plant diversity patterns in the Mediterranean Basin. Journal of Biogeography 36: 1333 – 1345.

Medail F, Myers N. 2004. Mediterranean Basin. In: Mittermeier RA, Robles Gil P, Hoffmann M et al. eds. Hotspots revisited: Earth’s biologically richest and most endangered terrestrial ecoregions. Monterrey: CEMEX. Conservation International, Washington and Agrupacio n Sierra Madre.

Molins A, Mayol M, Rossello JA. 2009. Phylogeographical structure in the coastal species Senecio rodriguezii (Asteraceae), a narrowly distributed endemic Mediterranean plant. Journal of Biogeography 36: 1372 – 1383.

de Montmollin B, Strahm W. 2007. La lista Top 50 de especies vegetales amenazadas de las islas del Mediterra neo: 50 especies silvestres al borde de la extincio n, y las acciones para salvarlas. Gland: UICN.

Nei M. 1987. Molecular evolutionary genetics. New York: Columbia University Press.

Nold R. 2003. Columbines. Aquilegia, Paraquilegia and Semiaquilegia. Cambridge, UK: Timber Press.

Nybom H, Esselink GD, Werlemark G, Vosman B. 2004. Microsatellite DNA marker inheritance indicates preferential pairing between two highly homologous genomes in polyploid and hemisexual dog-roses, Rosa L. Sect. Caninae DC. Heredity 92: 139 – 150.

Ojha BR. 2005. Comparison of Eco-Mse and Pst-Mse primer combinations in generating AFLP map of tomato. Journal of the Institute of Agriculture and Animal Sciences 26: 27 – 35.

Osborne JL, Martin AP, Carreck NL, et al. 2008. Bumblebee flight distances in relation to the forage landscape. Journal of Animal Ecology 77: 406 – 415.

Parks M, Cronn R, Liston A. 2009. Increasing phylogenetic resolution at low taxonomic levels using massively parallel sequencing of chloroplast genomes. BMC Biology 7: 84. http://dx.doi.org/ 10.1186/1741-7007-7-84.

Peakall R, Smouse PE. 2006. GENALEX 6: Genetic analysis in Excel. Population genetic software for teaching and research. Molecular Ecology Notes 6: 288 – 295.

Pritchard JK, Stephens M, Donnelly P. 2000. Inference of population structure using multilocus genotype data. Genetics 155: 945 – 959.

Quilichini A, Debussche M, Thompson JD. 2004. Geographic differentiation of morphological traits and isozymes in the Mediterranean island

endemic Anchusa crispa: implications for the conservation of a protected species. Biological Conservation 118: 651 – 660.

R Development Core Team. 2010. R: a language and environment for statistical computing. Vienna: R Foundation for Statistical Computing. http://www.r-project.org/.

Ro KE, McPheron BA. 1997. Molecular phylogeny of the Aquilegia group (Ranunculaceae) based on internal transcribed spacers and 5.8S nuclear ribosomal DNA. Biochemical Systematics and Ecology 25: 445 – 461.

Ronikier M, Cieslak E, Korbecka G. 2008. High genetic differentiation in the alpine plant Campanula alpina Jacq. (Campanulaceae): evidence for glacial survival in several Carpathian regions and long-term isolation between the Carpathians and the Alps. Molecular Ecology 17: 1763 – 1775.

Rosenberg NA. 2004. DISTRUCT: a program for the graphical display of population structure. Molecular Ecology Notes 4: 137 – 138.

Rymer PD, Manning JC, Goldblatt P, Powell MP, Savolainen V. 2010. Evidence of recent and continuous speciation in a biodiversity hotspot: a population genetic approach in southern African gladioli (Gladiolus; Iridaceae). Molecular Ecology 19: 4765 – 4782.

Schlu ter PM, Harris SA. 2006. Analysis of multilocus fingerprinting data sets containing missing data. Molecular Ecology Notes 6: 569 – 572.

Scho nswetter P, Tribsch A. 2005. Vicariance and dispersal in the alpine perennial Bupleurum stellatum L. (Apiaceae). Taxon 54: 725 – 732.

Slatkin M. 1993. Isolation by distance in equilibrium and nonequilibrium populations. Evolution 47: 264 – 279.

Stebbins GL. 1942. Polyploid complexes in relation to ecology and the history of floras. American Naturalist 76: 36 – 45.

Thompson JD. 2005. Plant evolution in the Mediterranean. New York: Oxford University Press.

Toumi L, Lumaret R. 2001. Allozyme characterisation of four Mediterranean evergreen oak species. Biochemical Systematics and Ecology 29: 799–817.

Vekemans X. 2002. AFLP-SURV version 1.0. Distributed by the author. Laboratoire de Genetique et Ecologie Vegetale, Universite Libre de

Bruxelles, Belgium. Vekemans X, Beauwens T, Lemaire M, Roldan-Ruiz I. 2002. Data from

amplified fragment length polymorphism (AFLP) markers show indica- tion of size homoplasy and of a relationship between degree of homoplasy and fragment size. Molecular Ecology 11: 139 – 151.

Vos P, Hogers R, Bleeker M, et al. 1995. AFLP – a new technique for DNA fingerprinting. Nucleic Acids Research 23: 4407 – 4414.

Vuylsteke M, Mank R, Antonise R, et al. 1999. Two high-density AFLP (R) linkage maps of Zea mays L.: analysis of distribution of AFLP markers. Theoretical and Applied Genetics 99: 921 – 935.

Walther-Hellwig K, Frankl R. 2000. Foraging habitats and foraging distances of bumblebees, Bombus spp. (Hym., apidae), in an agricultural landscape. Journal of Applied Entomology 124: 299 – 306.

Wang AL, Yang MH, Liu JQ. 2005. Molecular phylogeny, recent radiation and evolution of gross morphology of the rhubarb genus Rheum (Polygonaceae) inferred from chloroplast DNA trnL-F sequences. Annals of Botany 96: 489 – 498.

Westerbergh A, Saura A. 1994. Genetic differentiation in endemic Silene (Caryophyllaceae) on the Hawaiian islands. American Journal of Botany 81: 1487 – 1493.

Whitfield JB, Lockhart PJ. 2007. Deciphering ancient rapid radiations. Trends in Ecology and Evolution 22: 258 – 265.

Whittaker RJ. 1998. Island biogeography: ecology, evolution, and conservation. Oxford: Oxford University Press.

Whittall JB, Hodges SA. 2007. Pollinator shifts drive increasingly long nectar spurs in columbine flowers. Nature 447: 706 – 712.

Whittall JB, Medina-Marino A, Zimmer EA, Hodges SA. 2006. Generating single-copy nuclear gene data for a recent adaptive radiation. Molecular Phylogenetics and Evolution 39: 124 – 134.

Whittemore AT. 1997. Aquilegia: the flora of North America North of Mexica. In: Morin NR. ed. Magnoliophyta: Magnoliidae and Hamamelidae. New York: Oxford University Press, 249 – 258.

Wright S. 1943. Isolation by distance. Genetics 28: 114 – 138. Young WP, Schupp JM, Keim P. 1999. DNA methylation and AFLP marker

distribution in the soybean genome. Theoretical and Applied Genetics 99: 785 – 792.

Yuan L, Dussle CM, Muminovic J, Melchinger AE, Lubberstedt T. 2004. Targeted BSA mapping of Scmv1 and Scmv2 conferring resistance to SCMV using PstI/MseI compared with EcoRI/MseI AFLP markers. Plant Breeding 123: 434 – 437.

http://www.redlist.org/�

http://dx.doi.org/10.1080/11263504.2011.557097�

http://dx.doi.org/10.1080/11263504.2011.557097�

http://dx.doi.org/�

http://www.r-project.org/�

spatial genetic structure of aquilegia taxa endemic to the...

Documents