Biol Invasions DOI 10.1007/s10530-013-0486-z

ORIGINAL PAPER

Hybridization and introgression between the exotic Siberian elm, Ulmus pumila, and the native Field elm, U. minor, in Italy

Johanne Brunet • Juan E. Zalapa • Francesco Pecori • Alberto Santini

Received: 31 July 2012 / Accepted: 4 May 2013 Ó Springer Science+Business Media Dordrecht (outside the USA) 2013

Abstract In response to the first Dutch elm disease groups had higher levels of heterozygosity relative to (DED) pandemic, Siberian elm, Ulmus pumila, was U. pumila. The programs Structure and NewHybrids planted to replace the native elm, U. minor, in Italy. indicated the presence of first- (F1) and second- The potential for hybridization between these two generation (F2) hybrids and of backcrosses in the species is high and repeated hybridization could result hybrid population. The presence of healthy DED in the genetic swamping of the native species and resistant U. minor individuals combined with the self- facilitate the evolution of invasiveness in the intro- compatibility of U. minor could help explain the duced species. We used genetic markers to examine presence of F2 individuals in Italy. The presence of F2 the extent of hybridization between these two species individuals, where most of the variability present in and to determine the pattern of introgression. We the hybrids will be released, could facilitate rapid quantified and compared the level of genetic diversity evolution and the potential evolution of invasiveness between the hybrids and the two parental species. of U. pumila in Italy. Hybrids between U. pumila and U. minor were common. The pattern of introgression was not as Keywords Dutch elm disease Field elm strongly biased towards U. pumila as was previously Hybridization Introgression Microsatellites observed for hybrids between U. rubra and U. pumila Siberian elm in the United States. The levels of heterozygosity were similar between U. minor and the hybrids and both

& J. Brunet ( ) Introduction USDA-ARS, VCRU, Department of Entomology, University of Wisconsin, 1630 Linden Drive, Madison, WI 53706, USA The repeated introduction of exotic species by humans e-mail: [email protected]; [email protected] has accelerated the range expansion of many organ- isms (Crowl et al. 2008). While various species were J. E. Zalapa USDA-ARS, VCRU, Department of Horticulture, inadvertently introduced, some introductions were University of Wisconsin, 1575 Linden Drive, Madison, intentional. For example, exotic trees have been WI 53706, USA planted as ornamental trees in urban areas or to replace native tree species decimated by a disease F. Pecori A. Santini Institute of Plant Protection, C.N.R., Via Madonna del epidemic (Machon et al. 1997; Cogolludo-Agustin Piano, 10, 50019 Sesto Fiorentino, Italy et al. 2000; Zalapa et al. 2010). The introduced species 123 J. Brunet et al. were often close congeners of the native species, (Ulmus pumila L.) were distributed by authorities and which implied a high potential for hybridization nurseries throughout the whole Italian territory (Go- between the native and introduced species. While the idanich 1936). In addition, seedlings of Siberian elm potential impact of hybridization on the preservation were used as rootstocks (Goidanich and Azzaroli of genetic diversity of the native species was not 1941), or scions were grafted onto Field elms (Ansa- recognized when most of these exotic species were loni 1934). Moreover, two clones of Field elm, introduced, it is now clear that hybridization can pose ‘Christina Buisman’ and ‘Villagrappa 3’, were planted an increased risk of genetic assimilation and eventual as trials because they showed enough resistance to O. loss of the native taxa (Rhymer and Simberloff 1996; ulmi. Following the first DED pandemic, Goidanich Hedge et al. 2006). This is especially true for small (1941), an Italian plant pathologist, affirmed that the populations already at risk from biotic or abiotic impact of DED was successfully controlled, partly as a stresses (Rieseberg et al. 1989; Ellstrand and Elam result of the increased use of U. pumila and the 1993; Daehler and Strong 1997; Collin 2002; Burgess selection of tolerant U. minor clones. While the et al. 2005; Prentis et al. 2007). In addition, it has been widespread distribution of Siberian elms stopped established that hybridization can increase genetic shortly after World War II, many thousands of diversity and novel gene combinations which may, in Siberian elms remained in the landscape, mainly turn, facilitate the process of adaptation and help a along roads, rivers and streams or in abandoned fields. species spread into new habitats (Ellstrand and Elm species are known to cross-hybridize (Mit- Schierenbeck 2000; Vila et al. 2000; Sakai et al. tempergher and La Porta 1991; Goodall-Copestake 2001; Rieseberg et al. 2003; Hedge et al. 2006). et al. 2005) and the ability of U. pumila to produce Many of the European and American species of fertile offspring when crossed with various DED elm, Ulmus spp. (Ulmaceae), were decimated by susceptible elm species has been exploited to develop Dutch elm disease (DED) during the 20th century, DED tolerant varieties (Smalley and Guries 1993; with a first pandemic caused by the ascomycete fungi, Santamour and Bentz 1995; Ware 1995; Santini et al. Ophiostoma ulmi (Buisman) Nannf. and a more 2002, 2007; Mittempergher and Santini 2004). The aggressive species, O. novo-ulmi, responsible for the ability for elm species to cross-hybridize suggests, second pandemic (Brasier 1988, 1991). While Euro- however, a strong potential for the formation of hybrid pean and North American species of elms were very seeds between Siberian elms and the native elm susceptible to DED, with infected trees dying within species. To date, widespread hybridization has been 1–2 years, several Eurasian species exhibited varying documented between the exotic Siberian elm, U. degrees of tolerance to both the first (Smalley and pumila, and Field elm, U. minor, in Spain (Cogolludo- Guries 1993) and the second DED pandemics (Santini Agustin et al. 2000) and between U. pumila and the et al. 2005; Solla et al. 2005). In response to the two native red elm, U. rubra Muhl in the Midwestern DED pandemics, Siberian elm, U. pumila, was planted United States (Zalapa et al. 2009, 2010). to replace the native elms in different countries, A pattern of introgression biased towards U. including Italy (Goidanich 1936) and the United States pumila, with few backcrosses between hybrid indi- (Zalapa et al. 2010). The native Field elm, U. minor viduals and the native species, was detected in both Mill., was widely used as living support for grapevine, Spain and the United States. This pattern of biased fodder for cattle, timber for construction, and firewood introgression has been attributed to the lower abun- when the first DED epidemic reached Italy in the dance of the DED susceptible native elm species over 1930s (Sibilia 1930). Field elm was also important as the landscape (Cogolludo-Agustin et al. 2000; Zalapa shadow tree in pastures and as an ornamental tree in et al. 2009, 2010). In addition, the biased introgression streets and in parks (Goidanich 1936). Given its wide has been considered a threat for the conservation of the usage, the progressive disappearance of Field elm was genetic diversity of the native elm, U. rubra in the considered disastrous, which stimulated private nurs- United States (Zalapa et al. 2009) and U. minor in erymen (Ansaloni 1934) and the scientific community Spain (Cogolludo-Agustin et al. 2000). Furthermore, (Sibilia 1932; Passavalli 1935) to look for a suitable hybridization between U. pumila and U. minor or replacement. Within a few years following the onset of between U. pumila and U. rubra has been shown to the first DED epidemics, thousands of Siberian elms increase the genetic diversity (Cogolludo-Agustin 123 Hybridization and introgression et al. 2000; Zalapa et al. 2009, 2010) and, at least in the hairless twigs; small, blunt, hairless buds; shallowly Midwestern United States, to affect the genetic furrowed, gray or brown bark; and comparatively structure of U. pumila populations (Zalapa et al. small, smooth samaras (Wyman 1951); the crown is 2010). Hybridization could, therefore, have contrib- wide from roundish to vase-shaped, secondary shoots uted to the increased range of habitats where U. pumila are generally pendulous. In contrast, Field elm leaves can establish in the United States compared to its are oblanceolate to nearly circular, coarse and pubes- native range (Zalapa et al. 2010). Moreover, hybrid- cent when young, smooth and glossy above at ization may help explain the fact that Siberian elm has maturity, glabrous beneath with axillary pubescence, become an invasive species in 41 out of 50 states in the base uneven, corky wings (Santini et al. 2008); the United States (USDA, NRCS 2002; Ding et al. 2006). crown shape varies from cylindrical to conical. Based In the current study, we use genetic markers to on these morphological descriptors, eleven trees were examine the extent of hybridization between the native classified in the field as U. pumila,41asU. minor, and Field elm, U. minor, and the exotic Siberian elm, U. 42 as putative hybrids between U. pumila and U. pumila, in Italy. We examine the pattern of introgres- minor while two trees could not be identified sion between these two species to determine whether (Table 1). In addition to these 96 samples, to facilitate introgression is biased towards U. pumila, as was the genetic identification of hybrid individuals, we previously found between U. pumila and U. minor in used as a reference population for U. pumila,49 Spain (Cogolludo-Agustin et al. 2000) and between U. accessions from China whose genetic composition had pumila and U. rubra in the United States (Zalapa et al. been previously described (Zalapa et al. 2008a). 2009). We examine the level of genetic diversity of the hybrids relative to the two parental species and Genotyping using microsatellites quantify the degree of genetic differentiation between these three groups. We discuss the similarities and We extracted DNA from leaf tissue using a E.Z.N.A.Ò differences between the patterns of hybridization kit (Omega Biotek, Norcross, GA, USA) and mea- between U. pumila and U. minor in Italy relative to sured DNA concentrations with a BioPhotometer U. pumila and U. minor in Spain and U. pumila and U. (Eppendorf AG, Hamburg, D). All 96 samples were rubra in the Midwestern United States. Finally, we genotyped using the following ten microsatellite address some of the potential consequences of hybrid- primer-pairs (UR138, UR141, UR153, UR158, ization for the conservation of the native elm species, UR159, UR175, UR188a, ULM-2, ULM3, and U. minor, in Italy and for the potential spread of Ulmi1-98). These 10 primer pairs amplified alleles naturalized populations of U. pumila throughout Italy. in both U. pumila and U. minor and were originally developed in either U. rubra (UR- primers: Zalapa et al. 2008b), U. laevis (ULM- primers: Whiteley et al. Materials and methods 2003)orU. minor (Ulmil-98 primer: Collada et al. 2004). A total volume of 15 ll was used for the PCR Sampling of plant material reactions which included 1.5 ll109 PCR buffer,

1.8 ll 25 mM MgCl2, 2.4 ll dNTPs (1.25 mM of each Leaf material was collected from 96 elm trees at 61 dATP, dGTP, dTTP, and dCTP), 1.0 ll5lM primer, different locations throughout Italy and some parts of 2 ll10ngll genomic DNA, 1 U Taq DNA polymer- France (Table 1). Twelve specimens were collected at ase (Lucigen, Middleton, Wisconsin, USA), and 6.2 ll the ex situ elm clonal collection of the Institute of Plant H2O. The thermocycling conditions consisted of an Protection (C.N.R.) located in Antella (43°430N initial melting step at 94 °C for 3 min, followed by 30 11°220E; altitude: 170 m), in the province of Florence, cycles each of 94 °C for 15 s, 55/60 °C for 90 s and Italy. All other samples were collected in fields 72 °C for 2 min, and an elongation step of 72 °C for throughout Italy, with one to three accessions sampled 20 min. We used fluorescent labeled primers 50 end per site (Table 1). The following criteria were used to 6-FAM [6-carboxyfluorescein] fluorophore; IDT Cor- assign each tree to a putative species at the time of alville, Iowa, USA. Microsatellites were run on an collection: Siberian elm possesses symmetrical, once- ABI 3730 fluorescent sequencer (POP-6 and a 50 cm serrate, small leaves (3–7 cm long); slender, smooth, array; Applied Biosystems, Foster City, California, 123 J. Brunet et al.

Table 1 Collected samples Sample ID Province Morphology Genotype Latitude Longitude of putative Ulmus pumila (P), U. minor (M) and their M100 Torino M M 45.13 7.47 Hybrids (H) with the classification based on M101 Cosenza M M 39.49 16.30 morphology and the M102 Massa M M 44.10 10.11 classification based on the M103 Palermo M M 37.54 13.26 genetic profile (genotype) M104 Oristano M H 40.80 8.47 determined from the program Structure M105 Salerno M M 40.24 15.35 M106 Frosinone M M 41.38 13.17 M107 Crotone M M 39.10 17.70 M108 Avellino M M 40.58 15.12 M109 Agrigento M M 37.36 12.56 M110 Lodi M M 45.45 9.17 M111 Teramo M M 42.40 13.42 M112 Benevento M M 41.54 12.30 M113 Piacenza M M 45.20 10.00 M114 Bolzano M M 46.33 11.10 M115 Vercelli M M 45.22 8.27 M116 Bergamo M M 45.43 9.52 M117 Modena M M 44.32 10.54 M118 Pordenone M M 45.52 12.56 M120 Matera M M 40.10 16.37 M121 Imperia M – 44.60 7.53 M122 Padova M M 45.23 11.50 M123 Sassari M M 41.30 9.11 M124 Foggia M M 41.29 15.10 M126 Piacenza M – 44.48 9.50 M128 Venezia M H 45.46 12.50 M130 Seine et Marne (FR) M H 48.57 3.74 M131 Calvados (FR) M M 49.43 2.92 M132 Charente (FR) M – 45.30 1.57 M133 Finistere (FR) M – 48.26 3.40 M134 Loiret (FR) M M 47.50 2.44 M135 Finistere (FR) M – 47.59 4.25 M136 Calvados (FR) M – 49.43 2.92 M137 Orne (FR) M H 48.55 1.23 M138 Somme (FR) M – 49.48 2.50 M139 Firenze M M 43.41 11.25 M225 – – H – – M227 – – M – – M80 Catania M M 37.54 14.54 M81 Enna M M 37.23 14.18 M87 Treviso M M 45.47 12.35 M97 Vibo Valentia M M 43.47 11.15 M98 Genova M H 44.21 9.14 MP01 Aquila H M 42.02 13.56 MP02 Aquila H H 41.46 14.06

123 Hybridization and introgression

Table 1 continued Sample ID Province Morphology Genotype Latitude Longitude

MP03 Aquila H H 41.46 14.06 MP04 Aquila H P 41.46 14.06 MP05 Aquila H H 41.46 14.06 MP06 Aquila H H 41.46 14.06 MP07 Aquila H H 41.46 14.06 MP08 Aquila H H 42.02 13.56 MP09 Aquila H P 42.02 13.56 MP10 Aquila H M 42.02 13.56 MP11 Aquila H H 42.02 13.56 MP12 Aquila H H 42.02 13.56 MP13 Aquila H H 42.02 13.56 MP14 Aquila H H 42.02 13.56 MP15 Aquila H H 42.02 13.56 MP16 Aquila H H 42.02 13.56 MP17 Aquila H P 42.02 13.56 MP18 Aquila H P 42.02 13.56 MP19 Aquila H H 42.02 13.56 MP20 Siena H H 43.25 11.10 MP21 Firenze H M 43.29 11.09 MP22 Firenze H M 43.29 11.08 MP23 Firenze H M 43.36 11.11 MP24 Firenze H M 43.36 11.11 MP25 Firenze H M 43.36 11.11 MP26 Firenze H M 43.36 11.11 MP27 Firenze H M 43.36 11.11 MP28 Firenze H H 43.36 11.10 MP29 Firenze H H 43.36 11.10 MP30 Firenze H H 43.36 11.10 MP31 Firenze H H 43.36 11.10 MP32 Firenze H H 43.36 11.10 MP33 Firenze H – 43.42 11.24 MP34 Firenze H H 43.42 11.24 MP35 Firenze H M 43.47 11.27 MP36 Firenze H H 43.49 11.29 MP37 Firenze H – 43.49 11.29 MP38 Firenze H M 43.51 11.20 MP39 Firenze H H 43.46 11.13 MP40 Firenze H H 43.46 11.13 MP41 Firenze H H 43.46 11.13 MP42 Firenze H H 43.46 11.13 P25 Firenze P H 43.49 11.29 P26 Firenze P H 43.49 11.29 P27 Firenze P H 43.49 11.29 P28 Firenze P H 43.49 11.29 P29 Firenze P P 45.45 13.18

123 J. Brunet et al.

Table 1 continued Sample ID Province Morphology Genotype Latitude Longitude

P30 Gorizia P P 45.50 13.30 P31 Trento P H 45.55 11 .00 P32 Lecco P H 45.45 9.18 P33 Trento P M 45.53 11.10 The provinces are in Italy P34 Trento P H 45.42 10.55 unless otherwise noted P35 Parma P H 44.48 10.20 where FR indicates France

USA) using a Gensize Rox 650 ladder (GENPAK Ltd., proportion of an individual’s genotype that originated Brighton, UK), at the UW Biotechnology Center DNA from each reference population. Therefore, with Sequence Facility. Alleles were scored using Gene- K = 2, q values for the two parental species are Marker Software version 1.5 (SoftGenetics, State expected to be close to 1 and first-generation hybrids

College, Pennsylvania, USA). (F1) are expected to have q values of 0.5. Similarly, individuals with q values of 0.75 from species 1 and Identification of parental species and hybrids 0.25 from species 2 likely represent first-generation

backcrosses (BC1) to species 1, while individuals with We used three different methods to classify individ- q values of 0.25 species 1 and 0.75 species 2 most uals as hybrids or as belonging to one of the two likely are BC1 to species 2. A posterior probability of parental species (U. pumila or U. minor), based on 0.92 or greater was used to assign a genotype to one of their multilocus genotypic information. These meth- the two parental species. We first associated a ods included the Bayesian clustering method available genotypic profile to the morphological phenotype in Structure (v. 3.1) (Pritchard et al. 2000), the when running Structure. To optimize data presenta- Bayesian algorithms provided in NewHybrids v. 1.0 tion, we ran Structure a second time and associated a (Anderson and Thompson 2002) and Principal Coor- genotypic profile with the categorical species classi- dinate Analyses (PCoA). We used the 49 accessions fication provided by the first Structure run (the genetic from China as reference U. pumila population in these classification of an individual as U. pumila, U. minor analyses (Zalapa et al. 2008a). We could not do a or as hybrid). Finally, we ran Structure for K = 1–8 to manual identification of hybrids based on the geno- confirm the use of K = 2 as the optimal value. typic profiles of U. minor and U. pumila, as was We used PCoA to examine the clustering of previously done to separate U. rubra from U. pumila individuals. These analyses were based on the genetic in the United States, because we did not find sufficient distances calculated by GeneAlEx 6.0 (Peakall and species-specific alleles between U. minor and U. Smouse 2006) for PCoA analyses (Appendix 1of pumila (Zalapa et al. 2009). This could reflect a more GeneAlEx). We first ran PCoA where genotypic distant relationship between U. pumila and U. rubra profiles were associated to the morphological pheno- relative to U. pumila and U. minor as suggested by typic classification of an individual as U. pumila, U. phylogenetic analyses (Wiegrefe 1992). minor or hybrid; we then ran a second PCoA where In the Bayesian clustering method available in the genotypic profiles were associated to the genetic program Structure (v. 3.1) (Pritchard et al., 2000), we classification obtained from the Structure results. used a priori value of K = 2 (2 genetic clusters) to Finally, the program NewHybrids v. 1.1beta (Anderson account for the two parental species. We ran Structure and Thompson 2002) permitted an independent classi- using 50,000 Markov chain Monte Carlo iterations fication of an individual as U. minor, U. pumila or with 50,000 burn-in iterations and 10 replicates per hybrid, based on its genotypic profile, and it helped run. Analyses were performed with no a priori further classify the hybrids into specific categories. We taxonomic information and used the genetic admixture considered the following hybrid classes: first- (F1)and option and the correlated allele frequencies model. second- (F2) generation hybrids and first- (BC1)and The program Structure calculates an admixture coef- second-generation (BC2) backcrosses. The NewHy- ficient (q) for each individual, where q represents the brids algorithm was run without prior information for 123 Hybridization and introgression

Fig. 1 Structure results for A the classification of a China Europe reference population of Ulmus pumila from China, U. pumila, U. minor and their hybrids from Italy. A Structure results for K = 2, with genotypes grouped based on their morphology; B structure results for K = 2 with genotypes grouped based on genetic profiles. Each Reference U. pumila (n=49) Putative U. pumila (n=11), U. minor (n=36), hybrids (n=40) individual is represented by a thin vertical line; the black lines separate the groups. B China Europe The colored segments of each line illustrate the individual’s estimated membership fractions to each of the two genetic clusters (K = 2)

Reference U. pumila (n=49) BayesianU. pumila (n=6), U. minor (n=42), hybrids (n=39)

600,000 iterations following a 100,000-iteration burn- variance (AMOVA) (Excoffier et al. 1992)toestimate in. We combined the information obtained from the degree of genetic variation within and among Structure and NewHybrids to determine the specific groups. We first considered 4 groups including both U. hybrid class an individual tree was most likely to pumila from China and U. pumila from Italy and then belong. examined the variation among 3 groups considering only U. pumila from Italy. Lastly, we calculated

Parental species and hybrid genetic diversity pairwise FST values between the four groups, U. pumila from China, U. pumila from Italy, U. minor and the After classifying individuals as U. minor, U. pumila or hybrids using the AMOVA method of GenAlEx with hybrids, based on their genotypic profiles, we used 9999 bootstrap iterations. This approach brings the FST GeneAlEx 6 (Peakall and Smouse 2006) to separately estimates in line with the Weir and Cockerham describe the genetic diversity of U. pumila from China, estimates (Peakall and Smouse 2006). U. pumila from Italy, U. minor and the hybrids between the two species. The genetic data from the 10 micro- satellite loci were used to estimate the average observed Results

(A) and effective (Ae) number of alleles per locus, the number of loci with allele frequencies greater than 0.05, Identification of parental species and hybrids the Shannon’s information index (I), the number of private alleles, where a private allele is defined as the We obtained the genotypic profiles for 87 out of the 96 number of alleles unique to a group, and the levels of elm trees sampled. Based on their morphology, 11 of observed (Ho) and expected (He) heterozygosity for these trees were classified at the time of collection as each of the 4 groups. After describing the genetic U. pumila,36asU. minor and 40 as hybrids (Table 1). diversity within each group, we compared groups by We confirmed using Structure that 2 genetic clusters calculating the degree of genetic differentiation among best explained the genetic diversity (Appendix). Using groups. We performed an analysis of molecular K = 2, the Structure results identified 6 trees as U. 123 J. Brunet et al.

Fig. 2 Principal A coordinates analyses (PCoA 1 and 2) based on 10 microsatellite markers of a reference population of Ulmus pumila from China, and of U. pumila, U. minor, and their hybrids from Italy. A Grouping designations based on morphology; B Grouping designations based on genetic profiles

B

U. pumila(P1) = 56

U.minor (P2) = 43

F1= 0

F2= 35

BCP1 = 2

BCP2= 0

Fig. 3 NewHybrids analyses based on 10 microsatellite mark- colored segments that represent the individual’s estimated ers for a reference population of Ulmus pumila from China, and membership fractions to each of the six categories, U. pumila, U. for U. pumila, U. minor, and their hybrids from Italy. Each minor, first- (F1) or second- (F2) generation hybrids, backcross individual is represented by a thin horizontal line divided into to U. pumila (BCP1)ortoU. minor (BCP2) 123 Hybridization and introgression pumila, 42 trees as U. minor and 39 trees as hybrids originally classified as hybrids based on their mor- (Table 1; Fig. 1). In many instances, however, the phology actually represented U. pumila or U. minor classification based on the genotype using Structure based on their genotypic profile (Table 1; Fig. 1A). (Pritchard et al. 2000) did not correspond to the The morphological traits typically used to identify elm classification based on the morphology (Table 1; species in the field did not reliably distinguish the Fig. 1). For example, only two of the individuals genetic parental species and hybrids. originally classified as U. pumila based on their The PCoA results supported the greater reliability of morphology were identified as U. pumila based on the genotypic profiles relative to morphology to group their genotypic profile, with the majority being individual trees into categories representing the two identified as hybrids and one as U. minor (Table 1; parental species and the hybrids (Fig. 2). Moreover, the Fig. 1A). In addition, six of the individuals originally PCoA results illustrated a bilateral pattern of introgres- classified as U. minor based on their morphology were sion, with backcrosses to both parental species identified as hybrids based on their genotypic profile (Fig. 2B), as was further supported by the Structure (Table 1; Fig. 1A). Finally, a number of individuals results. When using the admixture coefficients (q)gen- erated in Structure to further classify the 39 hybrid individuals, we detected 21 first-generation hybrids Table 2 Genetic diversity characteristics based on 10 micro- satellite loci for Ulmus pumila, U. minor, and their hybrids (F1), and 18 backcrosses (BC). The 18 backcrosses were collected in Italy and for a U. pumila reference population from further categorized as 6 backcrosses to U. minor and 12 China backcrosses to U. pumila (Fig. 1B). Population U. pumila U. pumila U.minor Hybrids The NewHybrids results identified 7 individuals as China Italy U. pumila,43asU. minor and 37 as hybrids (Fig. 3).

N 49 6 42 39 A 4.90 2.70 5.70 7.20

Afreq. C 5 % 2.90 2.70 3.60 4.20 Table 4 Pairwise genetic differentiation (FST) based on 10 Ulmus pumila U. minor A 2.46 2.09 2.66 3.10 microsatellite loci for , , and their e hybrids from Italy and a U. pumila reference population from I 0.79 0.60 1.17 1.28 China No. private 0.70 0.00 0.60 0.60 alleles U. pumila U. pumila U. minor Hybrids China Italy Ho 0.34 0.35 0.54 0.56 He 0.36 0.32 0.59 0.59 U. pumila China N number of individuals, A average number of alleles per locus, U. pumila Italy 0.05 Afreq. C 5%mean number of alleles with frequency greater U. minor 0.37 0.32 than 0.05 per locus, Ae average effective number of alleles per Hybrids 0.11 0.08 0.14 locus, I Shannon index of diversity, Ho average observed heterozygosity per locus, He average expected heterozygosity Values in bold are statistically significant (P \ 0.01) using the per locus AMOVA method of GenAlEx with 9,999 permutations

Table 3 Analysis of molecular variance (AMOVA) based on population from China and for B. three groups: U. pumila, U. 10 microsatellite loci for A. four groups: Ulmus pumila, U. minor and their hybrids from Italy minor, and their hybrids from Italy and U. pumila reference Source of variance d.f. MS Variance components Total variance (%) F-Stat Value

A. Four groups

Among populations 3 100.01 3.03 34 FST 0.22 Within populations 132 5.99 5.99 66 B. Three groups

Among populations 2 63.84 2.33 25 FST 0.16 Within populations 84 6.89 6.89 75

Significance level of P \ 0.01 for FST using the AMOVA method of GenAlEx with 9,999 permutations is in bold 123 J. Brunet et al.

The programs Structure and New Hybrids classified was previously found between U. minor and U. pumila individuals quite similarly as belonging to one of the in Spain (Cogolludo-Agustin et al. 2000) and between two parental species or as representing a hybrid but the U. rubra and U. pumila in the Midwestern United two programs differed into how the hybrids were further States (Zalapa et al. 2009, 2010). Although hybrids classified. Structure more commonly classified individ- were common in Italy, the pattern of introgression was uals as F1 and backcrosses (Figs. 1B, 2B). The majority not as strongly biased towards U. pumila as was of hybrid individuals were identified as F2 by the previously observed in Spain for hybrids between U. program NewHybrids; 9 individuals had a posterior pumila and U. minor (Cogolludo-Agustin et al. 2000) probability of 0.94 or greater of belonging to the F2 and in the Midwestern United States for hybrids category and 13 had a posterior probability of 0.90 or between U. pumila and U. rubra (Zalapa et al. 2009). In greater; probabilities were lower for the other hybrid fact, approximately 33.3 % of the observed back- individuals (Fig. 3). crosses in Italy were towards U. minor and 66.6 % towards U. pumila, in contrast to 12.5 % and 87.5 % Genetic diversity and genetic differentiation respectively for backcrosses to U. rubra or U. pumila in for parental species and hybrids the United States (Zalapa et al. 2009). No distinctions were made between different classes of hybrids in the The hybrid individuals had the largest number of Spanish study (Cogolludo-Agustin et al. 2000). The alleles, followed by U. minor and then U. pumila biased pattern of introgression towards U. pumila (Table 2). The lower number of alleles in the U. suggests that U. minor will likely become assimilated pumila individuals sampled from Italy relative to over time, but the process will be slower than between China reflected the low sample size for this group. The U. pumila and U. rubra in the Midwestern United level of heterozygosity was greater for the hybrids and States where introgression is more strongly biased U. minor relative to U. pumila from China. The towards U. pumila (Zalapa et al. 2009). While the number of private alleles was similar for these three threat of genetic assimilation of U. minor may be less groups. Overall, genetic diversity was greatest for the imminent in Italy, planting U. minor trees over the hybrids, followed by U. minor and lastly U. pumila. landscape would increase the probability of backcross When all four groups were considered, the AMOVA with U. minor and help prevent its genetic assimilation. indicated that 66 % of the genetic variation was found The stronger pattern of biased introgression towards within each group with 34 % of the variation detected U. pumila observed in previous studies was attributed among groups (Table 3A). When only U. pumila from to the low abundance of healthy U. minor over the Italy was considered, 75 % of the variation was landscape in Spain (Cogolludo-Agustin et al. 2000) identified within groups and 25 % among groups and of healthy U. rubra in the United States (Zalapa (Table 3B). Such a pattern would be expected if the et al. 2009). The high susceptibility of U. rubra to U. pumila individuals in Italy were involved in hybrid DED, with many populations consisting of only young formation. As expected, we observed a greater level of or diseased trees with little pollen production (Lester genetic differentiation between the two parental species and Smalley 1972a, b), could explain the low abun- than between each of the parental species and the dance of U. rubra over the landscape in the United hybrids (Table 4). Similar levels of genetic differenti- States. The species U. minor is, however, less suscep- ation were observed between the hybrids and each of the tible to DED relative to U. rubra (Heybroek 1968). In two parental species, supporting bilateral introgression fact, some DED resistant U. minor clones were planted (Table 4). A low level of genetic differentiation was in Italy as part of the efforts to reduce the impact of observed between U. pumila from China and U. pumila DED on elms, following the first DED pandemic from Italy (Table 4). (Goidanich 1941). In addition, some resistance to the more aggressive O. novo-ulmi species, responsible for the second DED epidemics, has also been observed in Discussion Field elm (Santini et al. 2005; Solla et al. 2005). Therefore, resistance to DED could help explain the We detected many hybrid individuals between U. maintenance of healthy Field elms over the landscape minor and U. pumila in Italy, a pattern similar to what in Italy. The differences in the patterns of introgression 123 Hybridization and introgression

observed between Italy and Spain remain, however, theoretically distinguish between F1 and F2 hybrid difficult to explain. The pattern of biased introgression individuals, the lack of species-specific alleles and the in Spain was, however, based on patterns of allele relatively low number of loci can render the inference distributions and genetic distances (Cogolludo-Agu- of the specific genotype frequency classes to which an stin et al. 2000) and not on direct observations of individual belongs and the classification of individuals genetic backcrosses as was done in Zalapa et al. (2009, into specific categories difficult (Anderson and 2010) and in the current study. The grouping of trees Thompson 2002). However, NewHybrids identified 9 into U. pumila, hybrids and U. minor in Cogolludo- individuals with a greater than 0.94 posterior proba-

Agustin et al. (2000) was based on morphological bility of belonging to the F2 category. Combining the characteristics and was not adjusted once the genetic information obtained from NewHybrids and Structure, results were obtained. In fact, the authors claim that we concluded that the hybrid population consisted of a

20.5 % of the hybrids had only alleles from U. pumila mixture of F1,F2 and BC individuals. The availability which suggests that these individuals were genetically of more loci in the future would improve the U. pumila although they remained classified as hybrids classification of hybrids into distinct categories for in future analyses, based on their morphology. The these populations.

First Canonical Discriminant Function analyses sug- Second-generation (F2) hybrids between U. pumila gested introgression towards U. minor (Fig. 3 in and U. minor were likely present in Italy although they Cogolludo-Agustin et al. 2000). The fact that the were not detected between U. pumila and U. rubra in UPGMA results and allele frequency showed the the Midwestern United States (Zalapa et al. 2010). The hybrids closer to U. pumila than U. minor could simply low number of allozyme loci used in an earlier reflect the fact that some morphological hybrids were hybridization study between U. minor and U. pumila misclassified and really represented U. pumila based in Spain did not permit the detection of F2 hybrids in on their genetic profile. Under such circumstances, that study (Cogolludo-Agustin et al. 2000). Differ- there would be no evidence for a strong pattern of ences in the mating system of these elm species may biased introgression towards U. pumila in Spain and help explain differences in the likelihood of producing the pattern would be more similar to what we observed F2 hybrids between U. pumila and U. rubra relative to in Italy. If U. pumila was only introduced once in Spain between U. pumila and U. minor. While U. minor can in the sixteenth century as an ornamental tree during set some seeds when selfed (3.4 % seed set from the rule of King Philip the Second (Kamen, 1997), Mittempergher and La Porta 1991) and has been constant biased backcrosses towards U. pumila over classified as self-compatible, U. pumila and likely U. close to 400 years would make such individuals rubra have been reported as self-incompatible (Zalapa difficult to identify without a large number of loci et al. 2009). We expect a low probability of F2 hybrid distributed over the genome. We must therefore production from crosses between U. pumila and U. conclude, based on the data currently available, that rubra because both parental species may carry self- there is no evidence for a strong pattern of biased incompatibility (SI) alleles that are transmitted to the introgression towards U. pumila in Spain. F1 hybrids and can prevent mating between F1 The programs Structure (Pritchard et al. 2000) and individuals carrying similar SI alleles (Zalapa et al. NewHybrids (Anderson and Thompson 2002) both 2009). The lack of self-incompatibility alleles in U. identified a similar number of hybrid individuals, minor should increase the probability that F1 individ- although the classification of these individuals into uals mate with one another and form F2, relative to a specific hybrid classes differed significantly between situation where both parental species carry self- the two programs. Structure identified hybrid individ- incompatibility alleles. An increase in the frequency uals as first-generation hybrids (F1) and backcrosses of matings between F1 individuals would, in turn, tend (BC) and did not distinguish between first- (F1) and to reduce the overall frequency of matings between F1 second-generation (F2) hybrids. With K = 2, the and the parental species, including U. pumila. This expected proportion of the genome originating from reduced frequency of backcrossing with U. pumila each of the two parental species (the q values from will further reduce the probability of genetic assim-

Structure) is expected to be 0.5 and similar in the F1 ilation of U. minor through hybridization in Italy and F2 hybrid individuals. While NewHybrids can (Rhymer and Simberloff 1996; Hedge et al. 2006). 123 J. Brunet et al.

Table 5 Results of the structure analysis and the calculations to infer the number of genetic clusters (K) for a reference population of Ulmus pumila from China, and of U. pumila¸ U. minor and their hybrids from Italy, based on 10 microsatellite loci K Reps Mean LnP(K) Stdev LnP(K) Ln’(K) |Ln’’(K)| Delta K

110 -3,588.440000 0.490351 – – – 210 -3,024.940000 0.745654 563.500000 526.010000 705.434353 310 -2,987.450000 2.229723 37.490000 10.210000 4.579043 410 -2,960.170000 1.411894 27.280000 37.070000 26.255514 510 -2,969.960000 7.316526 -9.790000 9.880000 1.350368 610 -2,969.870000 16.828286 0.090000 10.690000 0.635240 710 -2,959.090000 18.650436 10.780000 5.480000 0.293827 810 -2,953.790000 16.701726 5.300000 47.443333 2.840624

Besides the potential genetic swamping of the et al. (2009) study; fewer hybrids were identified as F1 native species, a second concern with continued in the current study (54 %). The different stages of hybridization between an introduced and a native introgression identified between these two studies, species is the increase in genetic diversity and creation with a lower frequency of F1 individuals and the of new genotypes which can stimulate the evolution of presence of F2 individuals in the current study could invasiveness (Ellstrand and Schierenbeck 2000; Sakai explain some of the difference in the level of et al. 2001; Hedge et al. 2006). In this study, the heterozygosity of the hybrid populations observed hybrids had more alleles than either of the parental between these two studies. In addition, the level of species, were quite heterozygous (H0 = 0.56) and heterozygosity observed in U. minor in Italy equally differentiated from each of the two parental (H0 = 0.54) was greater than the levels reported in species (FST = 0.11 with U. pumila and 0.14 or 0.08 Spain (H0 = 0.22) (Cogolludo-Agustin et al. 2000). with U. minor). The level of heterozygosity in the This may reflect the fact that allozymes were used in hybrids (0.56) was higher than in U. pumila (0.34) but the 2000 study while microsatellites were used in this similar to U. minor (0.59). In this study, Structure study and microsatellites are known to be more identified 46 % of the hybrids as backcrosses (BC) variable than allozymes (Collevatti et al. 2001). When while NewHybrids identified some hybrids as second similar sets of microsatellite loci were used, the level generation hybrids (F2). While the level of heterozy- of observed heterozygosity in U. minor was within the gosity is important and reflects the stage of introgres- range previously detected for the native U. rubra in the sion, most of the variability present in the hybrids will Midwestern United States (Zalapa et al. 2009). be expressed in the F2 individuals. The maintenance of While the introduction of U. pumila in Italy was genetic variability in the hybrid population combined strongly encouraged by the local authorities during the with the variability expressed in an F2 population 1930s as a barrier against the DED epidemic (Passav- would increase the expression of novel gene combi- alli 1935) this species may have become an even more nations which may in turn set the stage for rapid dangerous threat to the native Field elm than DED. evolution (Keim et al. 1989), facilitate the process of Although the disastrous effects of the DED epidemic adaptation and help U. pumila spread into new habitats on Field elm populations are evident, the impact of and become invasive in Italy. hybridization with U. pumila is more difficult to The level of heterozygosity observed in this study evaluate, partly due to the fact that the two species are was lower than the heterozygosity detected for the difficult to discriminate based on morphology alone. hybrids between U. pumila and U. rubra in the The current study indicates that hybridization and

Midwestern United States (H0 = 0.90) (Zalapa et al. introgression between the native Field elm and the 2009). The level of heterozygosity in the hybrid exotic Siberian elm are causing irreversible changes in population is expected to be greatest with F1 hybrids. the genetic structure of the indigenous species. A While Structure identified 69 % of the hybrid indi- potential advantage of introgression toward U. minor viduals as first-generation hybrids (F1) in the Zalapa would be the transmission of DED resistance genes 123 Hybridization and introgression from U. pumila. This would increase the survival of U. Brasier CM (1991) Ophiostoma novo-ulmi sp. nov., causative minor over the landscape in Italy. Introgression toward agent of current Dutch elm disease pandemics. Myco- pathologia 115:151–161 U. pumila could, however, facilitate the acquisition of Burgess KS, Morgan M, Deverno L, Husband BC (2005) useful genes from the native U. minor that would Asymmetrical introgression between two Morus species enhance the ability of U. pumila to invade U. minor (M. alba, M. rubra) that differ in abundance. Mol Ecol habitats and could increase its ability to spread. In 14:3471–3483 Cogolludo-Agustin MA, Agundez D, Gil L (2000) Identification order to limit these risks, the use of Siberian elm of native and hybrid elms in Spain using isozyme gene should be restricted, while the use and spread of U. markers. Heredity 85:157–166 minor genotypes that prove more tolerant to DED Collada C, Fuentes-Utrilla P, Gil L, Cervera MT (2004) Char- should be promoted. acterization of microsatellite loci in Ulmus minor Miller and cross-amplification in U. glabra Hudson and U. laevis Pall. Mol Ecol Notes 4:731–732 Acknowledgments The authors wish to thank Ignazio Collevatti RS, Grattapaglia D, Duvall Hay J (2001) High reso- Graziosi for providing some samples. Eric Collin, Luisa lution microsatellite based analysis of the mating system Ghelardini and Francesca Bagnoli commented on the allows the detection of significant biparental inbreeding in manuscript. We gratefully acknowledge the National Science Caryocar brasiliense, an endangered tropical tree species. Foundation Minority Post-doctoral Fellowship to J.E. Zalapa Heredity 86:60–67 (NSF award #0409651) and support from the USDA-ARS to J. Collin E (2002) Strategies and guidelines for the conservation of Brunet. the genetic resources of Ulmus spp. In: Turok J, Eriksson G, Russell K, Borelli S (eds) Noble Hardwoods Network, 50–67, Report of the fourth meeting, September 1999, Appendix Gmunden, Austria, and the fifth meeting, May 2001, Blessington, Ireland. International Plant Genetic Resources Institute, Rome See Table 5 Crowl TA, Crist TO, Parmenter RR, Belovsky G, Lugo AE (2008) The spread of invasive species and infectious dis- ease as drivers of ecosystem change. Front Ecol Environ 6:238–246 Daehler C, Strong D (1997) Hybridization between introduced smooth cordgrass (Spartina alternifl ora; Poaceae) and native California cordgrass (S. foliosa) in San Francisco Bay, California, USA. Am J Bot 84:607–611 Ding J, Reardon R, Wu Y, Zheng H, Fu W (2006) Biological control of invasive plants through collaboration between China and the United States of America: a perspective. Biol Invasions 8:1439–1450 Ellstrand NC, Elam D (1993) Population genetic consequences of small population size: implications for plant conserva- tion. Ann Rev Ecol Syst 24:217–242 Ellstrand NC, Schierenbeck K (2000) Hybridization as a stim- ulus for the evolution of invasiveness in plants? PNAS 97:7043–7050 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 Goidanich G (1936) La moria dell’olmo (Graphium ulmi). Ramo editoriale degli agricoltori, Roma 134 pp Goidanich G (1941) Il problema della grafiosi dell’olmo nella References fase risolutiva. Annali della R. Accademia di agricoltura di Bologna 1:3–23 Anderson EC, Thompson EA (2002) A model-based method for Goidanich G, Azzaroli F (1941) Relazione sulle esperienze di identifying species hybrids using multilocus genetic data. selezione di olmi resistenti alla grafiosi e di inoculazioni Genetics 160:1217–1229 artificiali di Graphium ulmi eseguite nel 1939–1940. Boll Ansaloni A (1934) La moria degli olmi e la diffusione in Italia R Staz Pat Veg 21:287–306 dell’olmo siberiano. Bologna Ed. La Selva 119 pp Goodall-Copestake WP, Hollingsworth ML, Jenkins GI, Collin Brasier CM (1988) Ophiostoma ulmi, cause of Dutch elm dis- E (2005) Molecular markers and ex situ conservation of the ease. Adv Plant Pathol 6:207–221 European elms (Ulmus spp.). Biol Conserv 122:537–546

123 J. Brunet et al.

Hedge SG, Nason JD, Clegg JM, Ellstrand NC (2006) The Santamour FS Jr, Bentz SE (1995) Updated checklist of elm evolution of California’s wild radish has resulted in the (Ulmus) cultivars for use in North America. J Arboric extinction of its progenitors. Evolution 60:1187–1197 21:122–131 Heybroek HM (1968) Taxonomy, crossability and breeding of Santini A, Fagnani A, Ferrini F, Mittempergher L (2002) ‘San elms. Typescript presented at international symposium on Zanobi’ and ‘Plinio’ elmtrees. HortScience 37:1139–1141 Dutch elm disease, Ames, Iowa, 26–28 Feb, pp 25 Santini A, Fagnani A, Ferrini F, Ghelardini L, Mittempergher L Kamen H (1997) Philip of Spain. Ed. espanola, Siglo XXI de (2005) Variation among Italian and French elm clones in Espana editores, Madrid their response to Ophiostoma novo-ulmi inoculation. Forest Keim P, Paige KN, Whitham TG, Lark KG (1989) Genetic Pathol 35:183–193 analysis of an interspecific hybrid swarm of Populus: Santini A, Fagnani A, Ferrini F, Ghelardini L, Mittempergher L Occurrence of unidirectional introgression. Genetics 123: (2007) ‘Fiorente’ and ‘Arno’ Elm trees. HortScience 557–565 42:712–714 Lester DT, Smalley EB (1972a) Response of Ulmus pumila and Santini A, La Porta N, Ghelardini L, Mittempergher L (2008) U. pumila X U. rubra hybrids to inoculation with Cerato- Breeding against Dutch elm disease adapted to the Medi- cystis ulmi. Phytopathology 62:848–852 terranean climate. Euphytica 163:45–56 Lester DT, Smalley EB (1972b) Variation in ornamental traits Sibilia C (1930) La moria degli olmi in Italia. Boll R Staz Pat and disease tolerance among crosses of Ulmus pumila, U. Veg 10:281–283 rubra and putative natural hybrids. Silvae Genetica 21: Sibilia C (1932) La resistenza dell’Ulmus pumila a Cerato- 193–197 stomella ulmi. Boll R Staz Pat Veg 11:360–364 Machon N, Lefranc M, Bilger I, Mazer SJ, Sarr A (1997) Al- Smalley EB, Guries RP (1993) Breeding elms for tolerance to lozyme variation in Ulmus species from France: analysis of Dutch elm disease. Ann Rev Phytopathol 31:325–352 differentiation. Heredity 78:12–20 Solla A, Bohnens J, Collin E, Diamandis S, Franke A, Gil L, Mittempergher L, La Porta N (1991) Hybridization studies in the Buron M, Santini A, Mittempergher L, Pinon J, Vanden Eurasian Species of Elm (Ulmus spp.). Silvae Genetica Broeck A (2005) Screening European elms for resistance to 40:237–243 Ophiostoma novo-ulmi. For Sci 51:134–141 Mittempergher L, Santini A (2004) The history of elm breeding. USDA, NRCS. 2002. PLANTS Database, version 3.5. Online, Investigacio´n Agraria, Sistemas y Recursos Forestales website http://plants.usda.gov. National Plant Data Center, 13:161–177 Baton Rouge, Louisiana, US Passavalli L (1935) L’olmo siberiano (Ulmus pumila L.). Sua Vila M, Weber E, D’Antonio CM (2000) Conservation impli- importanza nella difesa da Ceratostomella (Graphium) cations of invasion by plant hybridization. Biol Invasions ulmi Buis. L’Alpe 22:409–418 2:207–217 Peakall R, Smouse PE (2006) Genalex 6: genetic analysis in Ware GH (1995) Little-known elms from China: landscape tree Excel. Population genetic software for teaching and possibilities. J Arboric 21:284–288 research. Mol Ecol Notes 6:288–295 Whiteley RE, Black-Samuelsson S, Clapham D (2003) Devel- Prentis PJ, White EM, Radford IJ, Lowe AJ, Clarke AR (2007) opment of microsatellite markers for the European white Can hybridization cause local extinction: a case for elm (Ulmus laevis Pall.) and cross-species amplification demographic swamping of the Australian native Senecio within the genus Ulmus. Mol Ecol Notes 3:598–600 pinnatifolius by the invasive Senecio madagascariensis?. Wiegrefe, SJ (1992) Molecular genetic variation in the Ulma- New Phytol 176:902–912 ceae: phylogenetic implications. Ph.D. thesis, University of Pritchard JK, Stephens M, Donnelly P (2000) Inference of Wisconsin-Madison population structure using multilocus genotype data. Wyman D (1951) Elms grown in America. Arnoldia 11:79–93 Genetics 155:945–959 Zalapa JE, Brunet J, Guries RP (2008a) Genetic diversity and Rhymer JM, Simberloff D (1996) Extinction by hybridization relationships among Dutch elm disease tolerant Ulmus and introgression. Ann Rev Ecol Syst 27:83–109 pumila L. accessions from China. Genome 51:492–500 Rieseberg LH, Zona S, Aberbom L, Martin TD (1989) Zalapa JE, Brunet J, Guries RP (2008b) Isolation and charac- Hybridization in the island endemic, Catalina mahogany. terization of microsatellite markers for red elm (Ulmus Conserv Biol 3:52–58 rubra Muhl.) and cross-species amplification with Siberian Rieseberg LH, Raymond O, Rosenthal DM, Lai Z, Livingstone elm (Ulmus pumila L.). Molec Ecol Res 8:109–112 K, Nakazato T, Durphy JL, Schwarzbach AE, Donovan Zalapa JE, Brunet J, Guries RP (2009) Patterns of hybridization LA, Lexer C (2003) Major ecological transitions in wild and introgression between invasive Ulmus pumila (Ulma- sunflowers facilitated by hybridization. Science ceae) and native U. rubra. Am J Bot 96:1116–1128 301:1211–1216 Zalapa JE, Brunet J, Guries RP (2010) The extent of hybrid- Sakai AK, Allendorf FW, Holt JS, Lodge DM, Molofsky J, With ization and its impact on the genetic diversity and popu- KA, Baughman S, Cabin RJ, Cohen JE, Ellstrand NC, lation structure of an invasive tree, Ulmus pumila McCauley DE, O’Neil P, Parker IM, Thompson JN, Weller (Ulmaceae). Evol Appl 3:157–168 SG (2001) The population biology of invasive species. Annu Rev Ecol Syst 32:305–332

123