Caryologia. International Journal of Cytology, Cytosystematics and Cytogenetics 73(1): 11-26, 2020 Firenze University Press www.fupress.com/caryologiaCaryologia International Journal of Cytology, Cytosystematics and Cytogenetics ISSN 0008-7114 (print) | ISSN 2165-5391 (online) | DOI: 10.13128/caryologia-510 Citation: M. Meloni, C.A. Dettori, A. Reid, G. Bacchetta, L. Hugot, E. Conti (2020) High genetic diversity and pres- ence of genetic structure characterise the endemics Ruta corsica and Ruta lamarmorae (Rutaceae). Caryologia 73(1): 11-26. doi: 10.13128/caryolo- gia-510 Received: July, 2019 Accepted: February, 2020 Published: May 8, 2020 Copyright: © 2020 M. Meloni, C.A. Dettori, A. Reid, G. Bacchetta, L. Hugot, E. Conti. This is an open access, peer-reviewed article pub- lished by Firenze University Press (http://www.fupress.com/caryologia) and distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distri- bution, and reproduction in any medi- um, provided the original author and source are credited. Data Availability Statement: All rel- evant data are within the paper and its Supporting Information files. Competing Interests: The Author(s) declare(s) no conflict of interest. High genetic diversity and presence of genetic structure characterise the endemics Ruta corsica and Ruta lamarmorae (Rutaceae) Marilena Meloni1, Caterina Angela Dettori2, Andrea Reid3, Gianlui- gi Bacchetta2,4,*, Laetitia Hugot5, Elena Conti1 1 Institute of Systematic Botany, University of Zurich, Zollikerstrase 107, 8008 Zurich, Switzerland 2 Centro Conservazione Biodiversità (CCB), Dipartimento di Scienze della Vita e dell’Ambiente - Università degli Studi di Cagliari. Viale S. Ignazio da Laconi, 13 - IT-09123 Cagliari, Italy 3 Department of Environmental Systems Science, ETH Zurich, Universitätstrasse 6, 8006 Zurich, Switzerland 4 Banca del Germoplasma della Sardegna (BG-SAR), Hortus Botanicus Karalitanus (HBK), Università degli Studi di Cagliari. Viale Sant’Ignazio da Laconi, 9-11, IT-09123, Cagliari, Italy 5 Conservatoire Botanique National de Corse, Office de l’Environnement de la Corse, Ave- nue Jean Nicoli, 20250 Corte, France * Corresponding author. E-mail: m.marilena75@gmail.com, c.angeladettori@gmail. com, areid@student.ethz.ch, bacchet@unica.it, Laetitia.Hugot@oec.fr, Elena.Conti@ systbot.uzh.ch. Abstract. Corsica and Sardinia form one of the areas with highest biodiversity in the Mediterranean and are considered one of the priority regions for conservation in Europe. In order to preserve the high levels of endemism and biological diversity at different hierarchical levels, knowledge of the evolutionary history and current genet- ic structure of Corso-Sardinian endemics is instrumental. Microsatellite markers were newly developed and used to study the genetic structure and taxonomic status of Ruta corsica and Ruta lamarmorae, rare endemics of Corsica and Sardinia, respectively, and previously considered a single species. Our analyses identified high levels of genetic variation within each species (P=0.883, He=0.543 for R. corsica; P=0.972, He=0.627 for R. lamarmorae). Intrinsic traits of the species (hermaphroditism, proterandry and poly- ploidy) and island-dependent factors (i.e. age, origin and history of the islands) might explain the detected high levels of genetic variation. We discovered differentiation between R. corsica and R. lamarmorae, and genetic structure within each species, which are consistent with the observation of low dispersal ability for both species. Our genetic results support the recent taxonomic classification of R. corsica and R. lamarmorae as separate species and suggest that they diverge at only few loci. One R. corsica population (SA) strongly differed from all other studied populations and appeared to be the prod- uct of hybridization between the two species in STRUCTURE analyses. Our results pro- vide important insights for the conservation of the two rare endemics. Further genetic analyses are recommended for R. lamarmorae and for population SA (R. corsica). Keyword. Genetic diversity, endangered species, Ruta, Corsica, Sardinia, microsat- ellite. 12 Marilena Meloni et al. INTRODUCTION The Mediterranean Basin is characterised by high species richness (10.8 species/1000 km2, Médail and Quézel 1999) and considered one of the main “hot spots” of biodiversity in the world (Médail and Myers 2004, Thompson 2005). Additionally, this area is par- ticularly rich in endemic taxa (about half of the approxi- mately 25000 plant species native to this region are endemics), which are mainly concentrated in moun- tain chains and islands (Médail and Quézel 1997, 1999, Thompson 2005, Cañadas et al. 2014). Corsica and Sardinia (Figure 1), the two largest islands of the Western Mediterranean Basin, form one of the areas with highest plant diversity in the Mediterra- nean and are particularly rich in endemics (Médail and Quézel 1997, Thompson 2005, Blondel et al. 2010). The Sardinian flora consists of 2498 taxa, with about 11.61% endemic (290 species) to the island (Conti et al. 2005, 2007; Bacchetta et al. 2012; Fenu et al. 2014). Corsica’s flora consists of 2325 taxa, of which ca. 10% are endem- ics (230 species; Jeanmonod and Gamisans 2013). Both islands are considered a major glacial refugium (Médail and Diadema 2009) and together host nine of the 50 most threatened plant species occurring in Mediterra- nean islands (de Montmollin and Strahm 2005). The high level of biodiversity and the number of endemics found in Corsica and Sardinia are often ascribed to their noteworthy ecosystem diversity (Bac- chetta and Pontecorvo 2005) and to past geologic and paleoclimatic processes. Indeed, tectonic movements during the Tertiary (from 66 to 2.58 million years ago, MYA), the Messinian Salinity Crisis (MSC, ca. 5 MYA), the establishment of a Mediterranean climate type in the Pliocene (2-3 MYA), and climate changes associated with glacial and interglacial phases (Pleistocene: 0.01-2 MYA) have shaped the history of Mediterranean plant lineages (Hewitt 2000, Gentili et al. 2015, Médail and Quézel 1997, Thompson 2005). Corsica and Sardinia are continental fragment islands belonging to a single microplate (the Corso- Sardinian (C-S) microplate; Alvarez et al. 1972) and are currently separated by a narrow (11 km) and shallow (less than 50 m deep) water channel through the Boni- facio Strait. The C-S microplate was attached to South- ern France and Northeastern Spain until the late Oli- gocene (28-30 MYA), when it broke off and rafted east- ward, until it collided with the Apulian microplate (i.e., the current Italian peninsula) ca. 18-20 MYA (Rosen- baum and Lister 2004, Speranza et al. 2002). It reached its current position in the middle of the Western Medi- terranean ca. 9 MYA (Rosenbaum and Lister 2004). The separation between Corsica and Sardinia may have begun as early as 15 MYA and was complete by 9 MYA (Alvarez 1972, 1976, Cherchi and Montadert 1982, Orsi- ni et al. 1980). Species that now occur in Corsica and Sardinia could have originated in different ways: 1) they were present in the region before the split of the C-S micro- plate from the Iberian peninsula; 2) they reached the C-S microplate when it was temporarily con- nected with the Apulian microplate during the Mio- cene (ca.10-20 MYA, Rosenbaum et al. 2002); 3) they reached the C-S microplate through the land bridges that formed among Corsica, Sardinia, the Apulian plate and the African continent during the MSC (5 MYA; Hsü et al. 1977, Krijgsman et al. 1999, McKenzie 1999) or 4) during the glacial marine regressions concomi- tant with the glacial cycles of the Pleistocene (Thomp- son 2005); 5) they reached the islands via long distance dispersal at any point in time (LDD). If insular popula- tions differentiated sufficiently from their closest rela- tives due to isolation and/or extinction, they gave ori- gin to island endemics. Although Corsica and Sardinia are one of the pri- ority regions for conservation in Europe (Myers et al. 2000, Mittermeier et al. 2005), knowledge of the evolu- Figure 1. Localities of populations taxonomically assigned to R. cor- sica (green dots) and R. lamarmorae (red dot) sampled for this study. Ruta lamarmorae population was divided in two subpopulations. Detailed information on each population is provided in Table 1. 13Genetic diversity and structure of Ruta corsica and Ruta lamarmorae tion and genetic characteristics of their endemic flora, instrumental for the long-term conservation of these species, is still poor. Some molecular phylogenetic anal- yses have been performed to infer when and how C-S endemics reached the two islands (Yesson et al. 2009, Mansion et al. 2008, Salvo et al. 2008, 2010) and few studies focused on the more recent history of these spe- cies (Bacchetta et al. 2008; Coppi et al. 2008; Mameli et al. 2008; Bacchetta et al. 2011; Garrido et al. 2012). Nev- ertheless, several questions remain unanswered, includ- ing: How did C-S endemics evolve after island coloni- zation? What is their current genetic structure? Are Corsican and Sardinian populations of the same species genetically differentiated? Ruta corsica DC. and R. lamarmorae Bacch., Brullo & Giusso are endemics of Corsica and Sardinia, respec- tively. The two species belong to the small genus Ruta, which also includes four species widely distributed in the Mediterranean (R. angustifolia Pers., R. chalepensis L., R. montana L., and R. graveolens L.) and three spe- cies endemic to the Canary Islands (R. oreojasme Webb & Berth, R. pinnata L.f. and R. microcarpa Svent.). Ruta corsica and R. lamarmorae exhibit some features not found in the other species of the same genus (i.e. pulvi- nate, subspinescent habit; green-glaucescent leaves; white to pale yellow petals) and have been interpreted by tax- onomists as relictual paleo-endemics (Bacchetta et et al. 2006, Cardona and Contandriopoulos 1979), in other words as ancient lineages that were more widespread in the past and are now restricted to a local region (Nekola 1999, Mishler et al. 2014), in this case to the C-S micro- plate (Arrigoni 1983, Thompson 2005). They were treat- ed as one species (i.e., R. corsica) until 2006, when they were split in two different taxa based on morphological (i.e., leaf shape and size of flowers, stamens and ova- ries; Bacchetta et al. 2006) and karyological differences (R. corsica is diploid, R. lamarmorae is tetraploid; Con- tandriopoulos 1957, Honsell 1957). Phylogenetic analyses of chloroplast DNA sequences from only two individu- als each from Corsica and Sardinia supported the sepa- ration of R. corsica and R. lamarmorae, with individuals from the two islands grouped in mutually exclusive sis- ter clades (Salvo et al. 2008). Molecular dating analyses and inference of ances- tral areas of distribution for Ruta species suggested that the genus originated during the Eocene in Eurasia and subsequently expanded westward and southward, colo- nising several landmasses of the forming Mediterranean Basin (Salvo et al. 2010). The ancestor of the two endem- ics likely colonised the C-S block from the Apulian plate (i.e., the emerging Italian peninsula) during the early Miocene. The divergence between the C-S endemics and the remaining Ruta species apparently occurred dur- ing the middle Miocene (ca. 14 MYA). Finally, R. corsica diverged from R. lamarmorae most likely in the Pliocene (ca. 3.7 MYA), when Corsica and Sardinia had already attained their current position in the middle of the Western Mediterranean sea and were separated by the Bonifacio strait. The two islands were occasionally con- nected by land corridors during the MSC of the Miocene and during the glacial maxima of Pleistocene climatic oscillations (Salvo et al. 2010). Table 1. Description of populations of R. corsica and sub-populations of R. lamarmorae surveyed in this study. Species Population abbreviation Location Population size Sample number Coordinates Altitude R. lamarmorae BC Broncu Spina ~1000 30 40° 07’ 34’’ N 9° 31’ 26’’ E 1650m SS Su Susciu ~1000 30 40° 01’ 02’’ N 9° 19’’ 33’’ E 1620m R. corsica MU Muvrella 20 8 42° 24’ 0” N 8° 54 0”’ E 1050m MC Monte Cinto 150 16 42° 23’ 0’’ N 8° 55’ 0’’ E 1750m SA Saltare 20 8 42° 21’ 41” N 8° 52’ 53”E 1180m GH Ghisoni 40 23 42° 6’ 37” N 9° 9’ 16”E 1650m AL Albertacce 120 20 42° 17’ 39” N 8° 52’ 44”E 1300-1500m BA Bastelica 750 21 42° 0’ 26” N 9° 6’ 4”E 900m 14 Marilena Meloni et al. Given the inferred biogeographic history of R. cor- sica and R. lamarmorae and their current distribution in Corsica and Sardinia, respectively, these species rep- resent an ideal case study to gain new knowledge on the genetic characteristics of the Corso-Sardinian endemic flora. The aims of the present study are thus to: (1) assess the current amount and distribution of genetic diversity for the two species, testing whether the taxonomic sta- tus of R. corsica and R. lamarmorae as separate species is warranted; and (2) use the results of genetic analyses to recommend proper conservation strategies for these spe- cies, with a particular focus on R. lamarmorae, recently listed as endangered according to the IUCN criteria and categories (Dettori et al. 2014a). MATERIALS AND METHODS Study species Ruta lamarmorae was described as a species separate from R. corsica in 2006 (Bacchetta et al.). It is a small, erect, perennial shrub, 15-50 cm tall, with woody, sub- spinescent branches. It is characterised by bipinnate, obovate-rounded leaves, 1.5-8 cm long. The whitish, pale yellow flowers (12-13 mm in diameter) are hermaphro- ditic and proterandrous. The capsules, 6-7 mm long, are obtuse at the apex. As with most members of its genus, R. lamarmorae is tetraploid, with 2n=36 (Honsell 1957). It blooms in June-July, fruiting in August-October. It occurs mainly on siliceous substrates, at an altitude of 1500-1750 m a.s.l. Ruta lamarmorae is found in a single, fragmented population in the Gennargentu massif (central-eastern Sardinia). It is categorized as endangered (EN) according to the IUCN criteria (2013; Dettori et al. 2014a). The main threats to this species are habitat fragmentation, overgraz- ing and fires (Bacchetta et al. 2006; Dettori et al. 2014a). Ruta corsica, first described in 1824 by De Candolle, shares the same habit with R. lamarmorae and has over- all similar leaves, flowers and fruits. It differs from R. lamarmorae in leaf shape (obovate to cuneate-oblong), smaller flower size (8-10 mm in diameter) and fruit size and shape (7-8 mm long, with apiculate apex; Bacchetta et al. 2006). Ruta corsica is the only diploid species of the genus, with 2n=18, while the other karyotyped spe- cies are tetraploid (Contandriopoulos 1957). It blooms and fruits slightly later than R. lamarmorae (Septem- ber-November vs. July-August, respectively). The spe- cies occurs at an altitude of 1000-1900 m asl, mainly on siliceous substrates, and is widespread on the main Cor- sican massifs with about 15 populations. It is listed as Least Concern (LC) under the French red list of threat- ened species (UICN, France 2013). To date, no information is available on the mating system, the pollination biology and the mode of disper- sal of the diaspores of neither of the two species. Sample collection and DNA extraction Plant material was collected during summer 2010 (see Table 1). Since the only existing population of R. lamarmorae has a scattered distribution on the Gennar- gentu massif, we sampled plants from the two opposite sides of the massif, in other words from the Northwest- ern (BC, 30 individuals) and Southwestern slopes (SS, 30 individuals; Figure 1). Because sub-populations BC and SS are large (thousands of individuals; Gianluigi Bac- chetta, pers. obs.), sampling was carried out in order to minimize collection of related individuals and cover the entire occupied area. Samples of R. corsica were collected from a total of 96 individuals from six populations cho- sen in order to cover the entire distribution range of the species (see Figure 1 and Table 1 for details on each pop- ulation). Leaf-tissue samples were dried and preserved in silica gel. Total genomic DNA was isolated from dried leaves using the QIAGEN® DNeasy plant mini kit, fol- lowing the manufacturer’s guidelines, with minor modi- fications. In particular, an increased volume of buffer AP1 (from 400 μl to 600 μl), buffer AP2 (from 130 μl to 200 μl) and RNase A (from 4 μl to 6 μl), as well as a longer incubation time with buffer AP1 (30 minutes) for cell lysis were applied. Microsatellite development and genotyping DNA isolated from one specimen of R. lamarmo- rae sampled from population BC was used by Genetic Marker Services (Brighton, UK, www.geneticmarkerser- vices.com) to develop an enriched library, design and test microsatellite primer pairs. Enrichment involved incubating adaptor-ligated restricted DNA with filter- bonded synthetic repeat motifs, (AG)17, (AC)17, (AAC)10, (CCG)10, (CTG)10 and (AAT)10. Twenty-one positive library colonies were selected for sequencing, from which 15 microsatellites were designed and tested. The primer sets were designed using PRIMER 3.0 (Rozen and Skaletsky, 2000). Each primer pair was tested for amplification ability and polymorphism on eight indi- viduals of R. lamarmorae (four from BC and four from SS). Cross-species amplification was tested on four indi- viduals of R. corsica and four individuals of R. chalepen- sis (another species of the same genus with wider dis- tribution also occurring in Sardinia; Salvo et al. 2010). Polymerase chain reactions (PCRs) for primer screening 15Genetic diversity and structure of Ruta corsica and Ruta lamarmorae were performed in 25 μL and contained approximately 50 ng of DNA, 5 pmol of each unlabelled primer, 1.5 mM of MgCl2, 0.2 mM of each dNTP, 1X PCR buffer and 0.5 U of SupraTherm DNA Polymerase (GeneCraft, Cologne, Germany). In-vitro amplifications consisted of: 60 s denaturation at 95°C, followed by 25 cycles of 95°C for 60 s, annealing at 55°C for 60 s and 72°C for 60 s, ending with a final extension at 72°C for 5 min. This first screening of microsatellites was performed on high- resolution agarose gels. The primer pairs able to amplify polymorphic prod- ucts were then used to genotype all sampled individuals via amplification with fluorescently labelled primers and separation of PCR products by capillary electrophore- sis. Amplifications were performed following the two- step method described by Schuelke (2000), using ca. 20 ng of genomic DNA, 2.5 μl of 10X reaction buffer, 0.5 μl of each dNTP (10 mM), 1 μl of MgCl2 (50 mM), 0.2μl of the forward primer with M13(−21) tail at the 5’ end (10 μM), 0.5 μl of the reverse primer (10 μM), 0.5 μl of the fluorescently labelled M13(−21) primer (FAM, NED, VIC, PET; 10 μM) and 0.1 μl of Taq DNA polymerase (5 U/μl; Bioline GmbH, Luckenwalde, Germany) in a final volume of 25μl. Amplification reactions started with 94°C for 3 min, followed by 30 cycles of 94°C for 30 s,s 55°C for 45 s (see Table 2), and 72°C for 1 min. The fluo- rescently labelled M13(−21) primer was incorporated in the following eight cycles of 94°C for 30 s, 53°C for 45 s, and 72°C for 1 min, followed by a final extension step of 72°C for 5 min. Up to four PCR products of different primer sets were pooled for each individual and separat- ed by capillary electrophoresis on an AB3130xl Genetic Analyzer. Alleles were sized against the internal size standard GeneScan™ LIZ500™ (Applied Biosystems) and scored using GeneMapper® software Version 4.0 (Applied Biosystems). The observed allele size of all genotyped individuals was decreased by 18bp in order to account for the M13(−21) universal sequence tag. Population genetic analyses Even though R. lamarmorae is known to be tetra- ploid (Honsell 1957), a maximum of two alleles per locus and per individual were detected in all populations, thus showing disomic inheritance. Because genetic analyses can be performed with standard population genetic tools developed for diploid organisms in tetraploid species characterised by disomic inheritance (Stift et al. 2008; Meloni et al. 2013), our analyses were conducted assum- ing a diploid status for R. lamarmorae. We assessed genetic diversity by quantifying the number of alleles (NA), proportion of polymorphic loci (P), number of private alleles (AP), observed (Ho) and expected (He) heterozygosity for each population across loci. Populations were tested for deviations from Hardy- Weinberg (HW) equilibrium using Fisher’s exact test with the Markov chain algorithm (Guo and Thompson 1992). The fixation index, FIS, was estimated in order to assess departure from Hardy-Weinberg expectations due to non-random mating. Fisher’s exact test was performed within each population to check for linkage disequi- librium (LD) between all different pairs of loci. These analyses were performed with the web-based Genepop (Raymond and Rousset 1995; Rousset 2008) and GenAl- Ex v. 6.5 (Peakall and Smouse 2012). The program BOT- TLENECK (Piry et al. 1999) was used to detect recent genetic bottlenecks in the studied populations. Based on the observed number of alleles and the sample size of a population, the program computes the gene diversity expected under the assumption of mutation-drift equi- librium and compares it to Hardy–Weinberg gene diver- sity (He; Nei 1978) to establish whether there is a sig- nificant deficit of gene diversity resulting from a recent bottleneck. As recommended by the authors of the pro- gram, a Wilcoxon sign-rank test (Luikart et al. 1998) was performed using 1000 bottleneck simulation repli- cates under the stepwise mutation model (SMM). FSTAT 2.9.3 (Goudet 1995) was used to estimate genetic differentiation among populations by FST. We also measured RST, an analogue of FST specific for micro- satellite data, employing a stepwise mutation model (SMM, Slatkin 1995); RST was measured using the soft- ware SPAGeDi 1.4 (Hardy and Vekemans 2002). The same program was used to perform an allele-size per- mutation test (10000 permutations) to evaluate the con- tribution of stepwise mutations to the observed genetic differentiation, hence the relative suitability of FST vs. RST (Hardy et al. 2003). In addition, since FST can consider- ably underestimate differentiation when loci are highly variable (as commonly found with microsatellite mark- ers; Hedrick 2005; Jost 2008; Meirmans and Hedrick 2011), we also calculated Jost’s Dest (Jost 2008) using 1,000 bootstrap replicates in SMOGD (Crawford 2010). An Analysis of Molecular Variance (AMOVA) was performed to assess the hierarchical partitioning of genetic variation among populations and species. We followed the procedure of Excoffier et al. (1992), Huff et al. (1993), Peakall et al. (1995), and Michalakis and Excoffier (1996) by estimating FST and using 999 ran- dom permutations of the data in GenAlEx v.6.5 (Peakall and Smouse 2012). With the same program, a Principal Coordinate Analysis (PCoA) based on a genetic distance matrix was performed to visualise genetic relatedness among individuals. To test for isolation by distance, a 16 Marilena Meloni et al. Mantel test (Mantel 1967) was applied to the matrix of pairwise Nei’s genetic distances (Nei 1972, 1978) and the matrix of geographical distances with 999 random per- mutations in GenAlEx v. 6.5 (Peakall and Smouse 2012). Population structure was inferred using the Bayes- ian clustering method implemented in STRUCTURE (v. 2.3; Pritchard et al. 2000; Hubisz et al. 2009). The pro- gram uses a Markov Chain Monte Carlo (MCMC) pro- cedure to estimate P(X|K), the posterior probability that the data fit the hypothesis of K clusters, and assigns individual genotypes to clusters by estimating the mem- bership coefficient Q for each individual based on allele frequencies at unlinked loci (independent of locality information). We tested all possible values of K from 1 to 9; for each K we ran an admixture model (each indi- vidual draws some fraction of the genome from each of the K populations) with correlated allele frequencies 20 times with a length of burnin period of 100,000 followed by 100,000 MCMC repetitions. To identify the best K, we measured ΔK (the rate of change in the log probabil- ity of data between successive K values), as suggested by Evanno et al. (2005) and implemented in STRUCTURE HARVESTER (Earl and vonHoldt 2012). This method provides the most accurate estimate of the number of clusters K (Evanno et al. 2005), but does not allow for discrimination between K=1 and K=2. Therefore we also calculated the average posterior probability of the data for each value of K, Ln P(X|K), as proposed by Pritchard et al. (2000). After determining the most effective num- ber of genetic groups (K) for our data, we ran STRUC- TURE with the admixture model and default parameter settings; the inferred genetic composition of individu- als was then determined using 100,000 iterations after a burnin period length of 100,000. RESULTS Microsatellite development and cross-species amplification Of the 15 microsatellites newly developed for R. lamarmorae, 13 were suitable for genetic analyses on the target species (Table S1). Tests on cross-species amplifi- cation showed that 13 markers could be used for R. cor- sica while 11 could be amplified in the more distantly related R. chalepensis (Table S1). Genetic diversity Eleven microsatellites showing a clear amplification pattern after capillary electrophoresis on both R. corsi- ca and R. lamarmorae were used for genotyping (Table 2). Because all studied individuals of R. corsica and R. lamarmorae were heterozygotes for the same two alleles in locus RL9, this marker was excluded from population genetic analyses. In R. lamarmorae, all amplified loci were polymorphic; in R. corsica, locus RL16 was fixed in populations SA, GH, AL, BA; population SA showed fixed alleles also at loci RL15, RL17 and RL18. The num- ber of alleles identified across all loci ranged from 22 to 61 in R. corsica populations, and from 55 to 68 in the two R. lamarmorae sub-populations (Table 3). Private alleles were found in all populations except SA (Table 1). All populations, except for GH, were at HW equi- librium (p>0.05). FIS values were positive in the two sub-populations of R. lamarmorae, as well as in four out of six populations of R. corsica (MU, MC, GH and BA; Table 3), meaning that the departure of genotype fre- quencies from Hardy-Weinberg expectations was always associated with a deficit of heterozygotes. Conversely, an excess of heterozygotes was found in populations SA and AL (R. corsica; Table 3). Significant linkage-disequilibrium (LD) at the 5% level was detected at one pair of loci for populations GH (RL6-RL18), AL (RL13-RL15) and MU (RL12-RL17), three loci for population BC (RL4-RL6, RL13-RL15, RL11-RL17), four loci for SA (RL4-RL12, RL4-RL13, RL11-RL13, RL12-RL13) and BA (RL6-RL11, RL12-RL15, RL13-RL15, RL15-RL18), and eight loci for SS (RL4-RL5, RL5-RL11, RL5-RL13, RL6-RL13, RL11-RL13, RL15- RL17, RL12-RL18, RL15-RL18). It was impossible to per- form most of the tests for populations MU (25 out of 45 pairs of loci) and SA (30 out of 45 pairs of loci). Gene diversity (He) was high in all populations, with a mean value of 0.627 for R. lamarmorae and 0.543 for R. corsica; the only population showing a relatively low value was SA (R. corsica), with He= 0.323 (Table 3). The only population showing significant signs of a recent bottleneck was BA (R. corsica, p = 0.032), while all other populations were at mutation-drift equilibrium. Genetic structure Genetic differentiation among populations meas- ured with FST was always statistically significant (P < 0.05); it was 0.086 between the two sub-populations of R. lamarmorae and ranged between 0.012 (MU-MC) and 0.240 (AL-SA) in R. corsica (Table 4). Genetic differen- tiation among populations across the two species ranged between 0.050 (MU-SS) and 0.212 (SA-BC). Population SA was the most differentiated, with 0.173