Genetic Variability of Alnus cordata (Loisel.) Duby Populations and Introgressive Hybridization with A. glutinosa (L.) Gaertn. in Southern Italy: Implication for Conservation and Management of Genetic Resources

: Alnus cordata (Loisel.) Duby (Neapolitan alder) is an endemic tree species with a restricted distribution range, limited to Corsica and southern Italy. The economic value of its wood, its rapid growth, the tolerance to drought stress and the nitrogen ﬁxation capacity make A. cordata an excellent candidate for breeding, as well as for conservation and management of genetic resources. In this context, we evaluated the genetic variability of southern Italy populations and veriﬁed the hybridization capacity with the simpatric species A. glutinosa (L.) Gaertn. Eight pure A. cordata populations, two pure A. glutinosa populations and six mixed A. cordata / A. glutinosa populations located in southern Italy were analyzed using seven microsatellite markers. A low genetic diversity within and among populations was observed, but no inbreeding effects were evident. A variable frequency of F 2 interspeciﬁc hybrids was observed in most of the mixed populations and few backcross individuals were scored. These results suggest a limited capacity of hybrid individuals to cross back with the parent species, reducing the risk of genetic pollution of A. cordata . This work provides meaningful knowledge for the conservation and management of the endemic species A. cordata, which represents a valuable source of biodiversity to be conserved.


Introduction
Hybridization plays an important role in evolution, speciation and even species extinction [1]. An increased rate of hybridization with related and more abundant species could represent a threat to the survival of rare and endemic species. Moreover, extensive hybridization of a widely-distributed species with endemic ones may result in "genetic swamping" or "demographic swamping" and can wipe out the endemic species [2][3][4]. Several studies have documented the decline of rare plant species due to hybridization phenomena [5][6][7][8][9].
Four species of Alnus are present in Italy: A. incana (L.), A. glutinosa, A. alnobetula (Ehrh.) K.Koch subsp. alnobetula (L.) and A. cordata (Loisel.) Duby. A. cordata, known as "Neapolitan alder", is an endemic species that grows in small areas of the southern Apennines, and mountains of north-eastern Corsica [28]. In southern Italy and in other areas of sympatry, A. cordata shares its natural distribution range with A. glutinosa, species widespread throughout Europe. Despite its limited natural range, A. cordata is not considered an endangered species. This species has experienced little human intervention; its most important threats are the competition with other species, the isolation due to reduction and/or absence of gene flow between populations and the isotherm shift in the Mediterranean region due to climate change [28]. Literature on A. cordata is rather scarce, but previous studies highlighted its excellent tolerance to drought stress, a relevant feature to face climatic changes [29,30]. Previous research [31] on A. cordata and A. glutinosa populations of southern Italy and Corsica highlighted different gene pools and hypothesized paleo-introgression events between the two species. A.cordata and A. glutinosa show strong phenotypic similarities [31], but little information is available on hybrids. Their interspecific hybrid, named A. × elliptica Req., was observed in Corsica, with rare individuals identified by leaf morphology [15], but no reports are available for sympatric areas of southern Italy.
Starting from the 1980s, considerable attention has been paid to the management, and genetic improvement of the genus Alnus [10,32] due to the economic value of its wood [33,34], its rapid growth and its nitrogen fixation capacity conferred by the symbiotic association with the soil actinomycete Frankia (Actinomycetales). All these features also made alder an interesting option for bioremediation, "energy plantation systems" and for promoting the growth of other species and the ecosystem development [14,26,35,36]. Moreover, A. cordata ecosystems showed N-rich litter, large organic C and total N stock indicating this species as ideal for afforestation and reforestation [37]. Given its particular traits, A. cordata has gained interest as genetic resources in breeding programs, with the aim of transferring its favorable characteristics to other Alnus species and obtain superior interspecific hybrids [26]. In this context, the management and conservation of endemic genetic resources as A. cordata require the evaluation of genetic variability of the species, the evaluation of the presence of hybrids and the understanding of the factors that contribute to the destructive outcomes versus the constructive ones of hybridization.
The main objectives of this study were (1) to evaluate the genetic variability of natural populations of A. cordata in its major natural distribution range, (2) to identify hybridization phenomena with A. glutinosa in areas of sympatry and (3) to evaluate whether hybridization and introgression processes may threat the genetic integrity of A. cordata.

Materials and Methods
A total of 319 individuals were collected from sampling sites in southern Italy, in the regions of Campania, Basilicata and Calabria (Figure 1), according to the distribution range of A. cordata and A. glutinosa [28] and the census made by Regional Forestry Agency. We analyzed 16 natural populations, consisting of eight pure A. cordata, six mixed A. cordata/A. glutinosa and two pure A. glutinosa found in the area where the presence of A. cordata is fragmented and restricted to few sites. Based on the population size, 16 to 26 trees per site were sampled and georeferenced (Table 1, Figure 1).   The preliminary taxonomic classification of each sampled tree as A. glutinosa or A. cordata was performed in the field according to morphology of leaves and bark (Acta Plantarum. Available online http://www.actaplantarum.org, accessed on 20 May 2021). In A. cordata, leaves are ovate or circular-ovate, cordate at base and bark is smooth and greyish brown; in A. glutinosa, leaves are obovate to circular, wedge-shaped at base and bark is fissured and dark brown [31]. In sympatric areas, where A. cordata and A. glutinosa coexisted, several individuals were classified as "uncertain", based on the difficult interpretation of leaf morphology (Table 1, Figure 2). Fresh leaves were collected from individual trees and subsequently stored at −20 • C for subsequent DNA extraction and analyses.

DNA Isolation, SSR Amplification and Genotyping
Frozen leaves were ground to fine powder using liquid nitrogen. Up to 50 mg of ground tissue were used for DNA extraction with the DNeasy 96 Plant Kit (Qiagen, Valencia, CA, USA), according to the manufacturer's protocol. Twenty-one SSR markers, previously developed in A. glutinosa and A. maritima [38,39], were tested and seven of them, which gave polymorphic amplification products, were used for the analysis (Table S2). Two different PCR multiplex reactions were set up based on the amplicon size, using fluorescent dye-labelled primers (6-FAM, VIC, NED, PET; Applied Biosystems, Foster City, CA, USA). Amplifications were performed with the Type-It Microsatellite PCR Kit (Qiagen, Valencia, CA, USA). PCR mix consisted of 4 ng of genomic DNA, 1× Type-it Multiplex PCR Master Mix, 2 µM of each primer and RNase-free water for a total volume of 12.5 L. Amplification conditions were as follows: initial heat activation step at 95 • C for 5 min, followed by 27 cycles of denaturation step at 95 • C for 30 s, annealing step at 57 • C for 1.5 min, and extension step at 72 • C for 30 s. A final extension step at 60 • C for 30 min was executed. PCR fragments were run on an ABI PRIMS ® 3130 XL Genetic Analyzer for separation and sizing. GeneScan 250 LIZ was used as an internal size standard. Genotyping was performed using GeneMapper v4.0 software (Applied Biosystems, Foster City, CA, USA). The alleles were determined by automated binning and checked by visual inspection. The unbiased probability of identity (PIunb) [40] computed for the combination of the seven markers was PI = 0.00. This value indicates the probability that two unrelated trees selected at random from a population would have identical genotypes at multiple loci: the lower is this value, the higher is the capacity of the markers used to capture the variability present in the data set. The absence of null alleles was verified using FREENA software for each locus chosen for the analysis [41].

Genetic Validation and Assignment of Species and Hybrids
NEWHYBRID 1.1 software [42] was used to estimate the posterior probability that genetically sampled individuals fall into each of a set of user-defined hybrid categories. NEWHYBRID assigned each individual to a different hybrid category, expressed in term of percentage of membership of each individual to a specific group. The categories in which individuals could fall are: parental (P0 = A. cordata; P1 = A. glutinosa); F1 and F2 hybrids, and two backcrosses (0Bx = backcross with A. cordata parental group; 1Bx = backcross with A. glutinosa parental group). Principal Coordinate Analysis (PCoA) for pure A. cordata and A. glutinosa populations and for hybrid mixed populations, based on genetic distance (GD), was performed using GeneAlEx 6.503 software [43]. Spatial analysis of NEWHYBRID results was performed using QGIS software [44], basing on a pie chart classification method.

Genetic Diversity and Population Structure
For each population, the number of alleles (Na), observed and expected heterozygosity (Ho, He), fixation index (Fis) and pairwise Fst were calculated using GeneAlex 6.503 software [43]. Allelic richness (Ar) was evaluated with HP-Rare 1.0 software [45]. The analysis of molecular variance (AMOVA) of the pure populations of A. cordata and A. glutinosa was calculated with Arlequin 3.1.1 software [46].
Population structure was inferred for all of the sampled populations using a Bayesian approach as implemented in STRUCTURE 2.3.4 software [47,48]. Analyses of population structure used admixture model with correlated allele frequencies. Parameters were set for a burn-in period of 100,000 and a MCMC (Markov chain Monte Carlo) with 200,000 iterations. Potential clusters (K) were tested using 20 iterations. To determine the most likely number of K, the ∆K method by Evanno et al. [49] was applied using STRUCTURE HARVESTER software [50]. A graphical representation of the STRUCTURE results was performed using CLUMPAK software [51].

Genetic Validation and Assignment of Species and Hybrids
The analysis by NEWHYBRID allowed us to ascertain the species and hybrids assignment. Figure 3 presents the NEWHYBRID results as a map of the partitioning of individuals in 216 A. cordata, 71 A. glutinosa and 32 hybrids. The plants sampled in the eight populations of pure A. cordata showed an average degree of assignment of 98.6% to the P0 category (A. cordata) and genetically confirmed that all the individuals belonged to expected pure species. Similarly, the individuals sampled in the two pure A. glutinosa populations showed an average assignment of 98.9% to the category P1 (A. glutinosa) confirming, again, the expected botanic classification. The results of the mixed populations with sympatric species highlighted the presence of hybrids in all the A. cordata/A. glutinosa populations, except for the population "VIT". All the identified hybrids showed a high percentage of membership (mean of 79.5%) to the F 2 hybrid category, with no individuals assigned to F 1 . Moreover, a low percentage of Backcross toward both parental species was found in all the mixed populations and never exceeded a percentage of assignment greater than 3% (Figure 3). The highest number of F 2 hybrids was identified in the population "SAS" (50%), while the population with the lowest value was "STA" (9.1%). An interesting exception was the mixed population "VIT", where, despite the co-presence of both species, there were no hybrids and all individuals belonged either to the species A. cordata (16 individuals) or A. glutinosa (10 individuals). Among the 28 individuals morphologically categorized as "Uncertain", because of the difficult interpretation of leaf morphology (Table 1), three were classified as A. cordata, 19 as A. glutinosa and six as hybrids.

Genetic Diversity and Population Structure
Genetic diversity indices were calculated for the three taxonomic groups obtained after NEWHYBRID assignment, i.e. A. cordata, A. glutinosa and hybrids ( Table 2)  The genetic diversity indices were also calculated by single population (Table 3). Overall, the mixed populations of A. cordata/A. glutinosa showed higher values for the number of alleles (mean Na = 6.904) compared to pure populations of A. cordata (mean Na = 3.107) and A. glutinosa (mean Na = 5.286). Higher values were also observed for heterozygosity in mixed populations compared to pure ones (mean Ho of 0.489, 0.396, 0.468 for mixed, A. cordata and A. glutinosa, respectively). All mixed populations showed significant positive Fis values, ranging from 0.237 to 0.461. Genetic diversity indices calculated for the single loci are reported in Table 4. Fis, Fit and Fst indices were markedly variable and showed high values for the loci "alma1", "alng4" and "AG20" and low values for the loci "alma7" and "AG13". The index Nm, indirect estimator of gene flow, showed an inverse pattern of variation and ranged from 0.236 of locus "AG20" to 3.789 of locus "alma7". Additional data of differentiation between populations and species are provided in the supplementary Table S1, which presents the allelic frequency of single loci calculated per population and per taxonomic group.   The analysis of molecular variance (AMOVA) was calculated separately for individuals of A. cordata and A. glutinosa (Table S3). In A. cordata, the partitioning of molecular variance was 89.86% within individuals, 5.86% among individuals within populations and 4.28% among populations. In A. glutinosa, the partitioning of variance was 77.5% within individuals, 12.98% among individuals within populations and 9.47% among populations.
The PCoA ( Figure 5), performed on the complete set of individuals (A. cordata, A. glutinosa and hybrids), showed a clear separation between the pure species, with hybrid individuals falling in between the two, with a varying degree of kinship to the parental species. The hybrid individuals were genetically interspersed between the pure species and some of them overlapped with the A. cordata or A. glutinosa cluster, as an indication of "genetic admixture". To better illustrate the relationships between the parental pure species and the hybrids, the average Nei's genetic distance was calculated between A. cordata and A. glutinosa and the group of hybrid individuals (Table S4). A greater genetic distance was found between the hybrids and A. glutinosa (0.429) compared to that observed between the hybrids and A. cordata (0.247).
The results of UPGMA clustering ( Figure 6) were in line with the average Nei's genetic distance (Table S4). The UPGMA tree highlighted three main clusters, the cluster of A. cordata populations, highly divergent from the cluster of A. glutinosa populations, and the cluster of mixed populations intermediate between the other two. Within the A. cordata cluster, subgroups can be identified, which correspond to the geographic position of the sampled populations. The populations with high presence of hybrid individuals were genetically closer to the pure populations of A. cordata then to A. glutinosa. The STA population, which was clustered close to A. glutinosa populations, confirmed this pattern, as it was composed mainly of A. glutinosa and a single hybrid individual was found in it.

Discussion
Our study focused on A. cordata, an endemic species in southern Italy, we investigated its genetic variability and the introgression with A. glutinosa, a wide distributed species, which coexist in the same area.
Based on the genetic analysis, each individual was assigned to either pure A. cordata, pure A. glutinosa or different categories of hybrids (F 1 , F 2 , Backcrosses). We found the presence of hybrids in the mixed populations of A. cordata/A. glutinosa with a percentage from 9.1% (STA) to 50% (SAS). The majority of the hybrids identified in the mixed populations belonged to the F 2 category, with low frequency of backcrosses toward the parental species. Interestingly, F 1 hybrids were not detected in any of the mixed populations. We hypothesized that the F 2 individuals observed in our study could be the result of an "old" hybridization event; once rare F 1 hybrids are produced, they could have a good capacity to cross-breed with each other, producing F 2 individuals. The mixed population of VIT, represents a particular case, where the presence of both species was observed, but no hybrids were identified. This particular result indicates that hybridization between these two species might be not common, as are the favorable conditions for its occurrence to arise. The variation of hybridization rate among the mixed populations suggests that natural crossing between the two species occurs at low and variable frequency, likely controlled by favorable local factors, which are unknown at present.
Other studies documented that crossing between two Alnus species might be difficult, even in the presence of a continuous mixed distribution where no natural barriers prevent gene flow between the two species [26,52,53]. A possible reason for this is the asynchronous flowering of the different species. The reproductive biology of A. glutinosa and A. cordata allows unidirectional crosses, with pollen of A. glutinosa fertilizing A. cordata flowers [31]. Experimental evidence supports the importance of climatic factors on hybridization success. Parfenov [53] studied the hybridization of A. incana with A. glutinosa in Belarus and noticed that the barrier to natural crossing laid in a six days shift in the flowering time between the two species. Such an impairment could be overcome in years with anomalous climate (e.g., with cold prolonged spring), when flowering time of the species had the chance to overlap. In another study of natural hybridization between A. glutinosa and A. incana, Banaev and Bazant [52] observed that hybrids occur very sporadically, even in a zone of continuous distribution of the species. They highlighted the increase in hybridization frequency in areas more affected by climate change. A similar hypothesis could explain the occurrence of hybrids observed in our case, as the study area is at the southern boundary of A. glutinosa geographic range and yearly climate fluctuations could rarely offset the flowering times of the two species and facilitate the hybridization. This hypothesis needs to be tested by further studies, but it raises the concern that climate change could affect interspecific hybridization in sympatric areas.
The structure and composition of the pure and mixed populations is further elucidated by STRUCTURE analysis, which highlighted two genetic clusters (K = 2), one represented by the A. cordata populations, and the other one by the A. glutinosa populations. A strong genetic homogeneity between individuals was evident within the pure populations of the two species, whereas in the mixed populations several hybrid individuals showed different levels of kinship to the main gene pools, sharing genetic background from the parental species.
The other main objective of this study was to evaluate the genetic diversity within and among the three main groups: A. cordata, A. glutinosa species and hybrids. All genetic diversity indices showed lower genetic variability level in A. cordata as compared to A. glutinosa, with hybrids showing the highest values in most indices. These results demonstrate that A. cordata is potentially more vulnerable in terms of genetic erosion. However, the observed level of interspecific hybridization and introgression did not reveal permeability between the two species, which could represent a threat for the genetic integrity of A. cordata. The analysis of genetic diversity of single populations highlighted that most of the pure A. cordata populations, except for the population ORS, showed a fixation index (F) close to zero, indicating that random mating occurred in these populations and no post-zygotic selection mechanisms favored inbreeding. Among the two A. glutinosa pure populations, the population CAM showed a significant positive F value (0.187). The geographic isolation of this population could explain its positive fixation index as a result of genetic drift. Despite their higher genetic diversity, all the mixed populations showed significant positive F values, corresponding to a deficiency of heterozygotes compared to the expected. This could be the result of the assortative mating between the two Alnus species with a small cohort of compatible individuals. Another point of great interest is the molecular signature of interspecific introgression as revealed by the differentiation pattern of the loci analyzed. The index Fst was highly variable across loci, with low value in the loci Alma 7 (0.062) and AG13 (0.088), and high values in the loci AG20 (0.515) and Alma1(0.320). This variation of Fst suggests a differential contribution of these loci and the associated genomic regions to the genetic structure and divergence between populations and species. The loci with high Fst are discriminant between populations, but also informative about the admixture rate between A. cordata and A. glutinosa. Actually, the allelic frequency of the loci with high Fst were also highly divergent between populations and between species (Supplementary, Table S1). Reproductive isolation and introgression between species are largely controlled by "speciation" genes and the associated genome regions can be more or less porous to gene flow between species [54]. The analysis of Fst and allelic frequency in hybrid zones is a powerful approach to investigate the interspecific hybridization and the differential permeability of genomic regions to introgression [55]. In sympatric populations of Juglans regia L., J. sigillata Dode and J. cathayensis Dode, the analysis of genetic structure and Fst pattern by SSR markers have identified historical introgression phenomena and Fst outliers functionally linked to adaptive genes [56,57]. Genomic studies of sympatric populations of Populus balsamifera L., P. angustifolia E. James, and P. trichocarpa Torr. et A. Gray ex Hook. have discovered genomic clines of non-neutral introgression, with adaptive significance, especially at the geographic boundaries of species range [58]. Similarly, the variable Fst level observed in our study, at the southern margin of the A. glutinosa range, suggests that introgression between A. cordata and A. glutinosa might differentially affect loci and genome regions, with possible adaptive implications. This hypothesis needs to be elucidated in further studies of Alnus hybrid zones by genome-wide analyses.

Conclusions
We evaluated the genetic diversity of the endemic A. cordata populations in southern Italy and we reported the first molecular evidence of its natural hybridization with the widespread species A. glutinosa. The hybridization between the two species naturally occurs in sympatric areas, but the phenomenon varies in different ecological sites. The very low frequency of backcross individuals suggests a limited capacity of hybrid individuals to cross back with the parent species, reducing the risk of genetic pollution of A. cordata. Therefore, the hybrid individuals seem not to have the potential to wipe out the parental species and the genetic integrity of the endemic A. cordata would not be endangered. The results support the hypothesis that the hybridization occurs under favorable, but rare circumstances. The factors that actually favor this phenomenon are not investigated in this study, but reasonable hypotheses are related to particular climatic conditions that allowed the species to cross and produce F 1 hybrids in an initial event. In this scenario, climate change could affect the flowering time of the two species, and possibly facilitate interspecific hybridization. Further studies could clarify the conditions under which hybridization between the two species has the potential to occur. In conclusion, this work might provide meaningful knowledge for the conservation and management of the endemic species A. cordata, which represents a valuable resource of biodiversity in the Mediterranean ecosystems.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/f12060655/s1. Table S1: Allele frequency at seven SSR loci, observed in pure and mixed populations of A. cordata, A. glutinosa and their interspecific hybrids. Table S2: List of the seven polymorphic SSR markers used in this study. Table S3: Hierarchical AMOVA calculated considering the two species, A cordata e A. glutinosa. Table S4: Pairwise Matrix of average Nei's Genetic distances between designated parental type individuals of A. cordata, A. glutinosa and hybrids.