Identiﬁcation of Alnus glutinosa L. and A. incana (L.) Moench. Hybrids in Natural Forests Using Nuclear DNA Microsatellite and Morphometric Markers

: Two alder species ( Alnus glutinosa and A. incana ) have overlapping distribution, naturally occur in Lithuania, and are considered ecologically and economically important forest tree species. The objective of our study was to estimate the likelihood of spontaneous hybridizations between native alders in natural stands of Lithuania based on leaf morphology and nuclear microsatellite markers. The sampled trees were assigned to the three taxonomic groups of A. glutinosa , A. incana, and potential hybrids based on the leaf and bark morphological traits. The genetic differentiation and potential hybridization between these three groups was tested based on 15 nSSR markers. We identiﬁed studied Alnus spp. individuals as pure species and hybrids. Two microsatellite loci were reported as discriminating well between these species. We concluded that our results showed the highest likelihood of two genetic group structures, a clear genetic differentiation between the morphology-based groups of A. glutinosa and A. incana , and rather variable likelihood values in the putative hybrid group. The results provide important implications for genetic conservation and management of Alnus spp.


Introduction
Spontaneous natural interspecific hybridization provides a source of genetic variation, where hybrid populations often contains higher genetic diversity than their parental species, upon which natural selection may act [1]. Thus, natural interspecific hybridization of plants has been studied for a decades, and it is common to many species [1][2][3][4][5]. Hybridization can increase allelic variability and transmit adaptively important genetic information, which can increase the fitness of an introgressed lineage [6][7][8][9][10]. Introgressive hybridization plays an important role in evolution, increasing genetic diversity by creating many new genotypes, which can lead to the creation of new strains, ecotypes, or even sexual species adapted to certain conditions [1,4,[11][12][13]. However, hybridization can have a negative impact as well; for example, it can cause genetic erosion and interrupt species integrity and lead to species extinction [1,[13][14][15][16].
A. incana Alnus glutinosa (L.) Gaertn. is naturally widespread across all of Europe, from mid-Scandinavia to the Mediterranean countries, including northern Morocco and Algeria [17]. A. incana (L.) Moench. is native to most of Central Europe, to the west towards France and east into Russia, the Caucasus, and is common in Scandinavia and western Siberia. A. incana distribution range overlaps with the black alder and extends further north. In contrast, its southern distribution is more limited in comparison with black alder, and it lower frequency in parental populations [26]. King and Ferris [27] have used chloroplast DNA (PCR-RFLP) and nuclear DNA (ISSR) variation to study alder species A. cordata (Lois.) and A. glutinosa (L.) in Corsica and southern Italy. Some of their findings of geographically correlated cpDNA variation was explained by introgressive hybridization. Zhuk et al. [29] have tested the transferability and amplification of 15 microsatellite markers developed for Betula pendula Roth. on A. glutinosa, A. incana, and their hybrids. The transferability rate was high, and eight out of fifteen nSSR have amplified in Alnus spp. Moreover, SSR marker results confirmed the species separation but were not entirely correlated for the putative hybrids. Later, Šmid et al. [31] used nuclear microsatellites in combination with chloroplast DNA markers to find putative hybrids among A. glutinosa and A. rohlenae at a narrow hybrid zone in southern Serbia. They found that, despite the selection, which acts against the triploid hybrids, they are important for gene exchange and enrichment of haplotype diversity and serve as a bridge for introgression and development of novel genetic lineages [31]. Stanton et al. [32] was investigating the hybridization of red alder (A. rubra Bong.) and white alder (A. rhombifolia Nutt.) in Cascade Mountains. Authors have developed and tested 30 SSR markers, which successfully produced high-quality multilocus genotypes. Furthermore, they were looking for SSR markers that can identify putative species-specific alleles that could best differentiate among the red alder, white alder, and hybrid samples. TO the end, Stanton et al. [32] found four SSR markers, which amplified loci with specific alleles that could serve for species and hybrid identification. Villani et al. [33] used seven nuclear microsatellites to study hybridization and genetic variation among endemic tree species-A. cordata and A. glutinosa in southern Italy. Authors found low genetic diversity, a variable frequency of F2 interspecific hybrids, and few backcross individuals. They have concluded that hybridization among the two Alnus spp. is limited, and the risk of genetic pollution is relatively low [33].
The objectives of our study were (i) to identify spontaneous hybridization events among A. glutinosa and A. incana species based on the morphology and DNA markers, (ii) to evaluate the genetic diversity and morphological traits of A. glutinosa and A. incana and their spontaneous hybrids, (iii) and to assess the growth characteristics of the hybrids in common habitats of both pure alder species.

Material
The material for laboratory analysis was collected in several study sites all over the country based on the presence of alder species and the alder hybrid swarms ( Figure 1). In total, om two alders seed orchards (I-black alders, II-black, grey, and hybrid alders), two black alder progeny field trials, and the natural mixed forests of Lithuania, four temporary research plots (500 m 2 ) of juvenal alders and individually selected black, grey, and hybrid alders were sampled for the study. During summer 2015, in each sample plot, 5-10 leaves were collected from the southern side at height of 5-8 m from the selected alder trees.
In total, for laboratory analysis, we used leaves from 162 putative spontaneous hybrids between black (Alnus glutinosa L) and gray (A. incana (L.) Moench.) alders as well as 13 and 14 individuals of A. glutinosa and A. incana. We collected leaves only from trees that looked like hybrids in seed orchard I, progeny field trials, and temporary research plots. Some hybrids and the pure alders leaves were collected in seed orchard II and individual selected plots. The putative hybrid alders were identified onsite based on a visual examination of the leaf morphology traits and stem morphotype [28,48]. After the morphological evaluation of the leaves was performed in the second half of the summer, we used the degree of leaf injury by alder leaf beetle (Agelastica alni L.) as an additional trait for the morphological discrimination between the species groups. The degree of the leaf injury by A. alni was the highest in A. incana (damaged > 50% leaf), lowest in black alder (damaged < 30% leaf), and often reached an intermediate level in the group of putative hybrids. This injury may be related to phenolic compound concentration, which is the highest in black alder leaves [76]. Nevertheless, we used these pest injury scores as a pure species indicator, but not for the determination of the hybridization degree. In total, for laboratory analysis, we used leaves from 162 putative spontaneous hybrids between black (Alnus glutinosa L) and gray (A. incana (L.) Moench.) alders as well as 13 and 14 individuals of A. glutinosa and A. incana. We collected leaves only from trees that looked like hybrids in seed orchard I, progeny field trials, and temporary research plots. Some hybrids and the pure alders leaves were collected in seed orchard II and individual selected plots. The putative hybrid alders were identified onsite based on a visual examination of the leaf morphology traits and stem morphotype [28,48]. After the morphological evaluation of the leaves was performed in the second half of the summer, we used the degree of leaf injury by alder leaf beetle (Agelastica alni L.) as an additional trait for the morphological discrimination between the species groups. The degree of the leaf injury by A. alni was the highest in A. incana (damaged > 50% leaf), lowest in black alder (damaged < 30% leaf), and often reached an intermediate level in the group of putative hybrids. This injury may be related to phenolic compound concentration, which is the highest in black alder leaves [76]. Nevertheless, we used these pest injury scores as a pure species indicator, but not for the determination of the hybridization degree.

Morphology Traits
In total, leaf morphology/morphometric traits of 192 alder trees with 5 to 10 leaves per tree were scanned and assessed: A. glutinosa (13 trees), A. incana (14 trees), and Alnus putative hybrids (162 trees). The mean values of the leaf traits were used in the data analysis. For the morphological investigation, we chose the key leaf traits for identifying the alder species [28,48,50]. The WinFolia 2016 Leaf analyzer program (Regent Instru-

Morphology Traits
In total, leaf morphology/morphometric traits of 192 alder trees with 5 to 10 leaves per tree were scanned and assessed: A. glutinosa (13 trees), A. incana (14 trees), and Alnus putative hybrids (162 trees). The mean values of the leaf traits were used in the data analysis. For the morphological investigation, we chose the key leaf traits for identifying the alder species [28,48,50]. The WinFolia 2016 Leaf analyzer program (Regent Instruments Inc.) was used to score six leaf traits: blade length (A, mm), blade maximum width (B, mm), length to position where maximum blade (D, mm), petiole length (I, mm), blade width measured at 90% blade length (E, mm), and upper angle of leaf (W) (Figure 2). One trait was assessed visually: secondary veins pairs (N, un). The pubescence of the lower half of the blade was determined in the five-point scale by [28] and was scored on a microscope (4× digital zoom) (P, 0-5). Four leaf traits were derived (D/A, E/B, I/A, and B/A). ments Inc.) was used to score six leaf traits: blade length (A, mm), blade maximum width (B, mm), length to position where maximum blade (D, mm), petiole length (I, mm), blade width measured at 90% blade length (E, mm), and upper angle of leaf (W) (Figure 2). One trait was assessed visually: secondary veins pairs (N, un). The pubescence of the lower half of the blade was determined in the five-point scale by [28] and was scored on a microscope (4× digital zoom) (P, 0-5). Four leaf traits were derived (D/A, E/B, I/A, and B/A).

Molecular Data Analysis
We have used 15 microsatellite loci (nSSR) for Alnus species differentiation and putative hybrid identification analysis. We use the Bayesian clustering approach as the model-based clustering algorithm in STRUCTURE ver. 2.3.4 software [85] to assess the genetic structure within the sampled groups of alders with the number of genetic clusters (K) ranging from 1 to 6 and using 100,000 Markov chain Monte Carlo iterations with a burn-in period of 100,000 and 20 replicates per run. We used the admixture model with correlated allele frequencies among the species groups. To determine the most likely number of clusters (K), the Delta K method by Evanno et al. [86] was applied using STRUCTURE HARVESTER software v0.6.94 [87].
The STRUCTURE analysis as described above was performed on all samples from the three species groups: the group of A. glutinosa, the group of A. incana, and the group of potential hybrids-189 individuals in total. To detect the putative interspecific hybrids among the studied individuals, the genetically pure individuals of the two respective parental species (A. glutinosa and A. incana) were used as the reference samples. This method has been used in several studies [5,54,56] to identify intra-and inter-specific hybridization in trees. The sampled individuals were sorted according to leaf and stem traits into the most probably pure individuals of each species and into one group of putative hybrids. Then, we used the program STRUCTURE to assign individuals to pure species or identify possible hybrids when pools were mixed. When two pure parental species were sampled as references, it was expected that the optimal value of K would consist of two genetic clusters (K = 2). This could be confirmed by testing values of K from one up to the number of populations in the respective groups using the STRUCTURE HARVESTER software [87], and we selected the optimum K following the method of Evanno et al. [86]. The program STRUCTURE generates an admixture coefficient (q) that represents the proportion of an individual's genotype that originates from each of the K genetic clusters. STRUCTURE can be run with the option ANCESTDIST, which computes the 95% posterior probability for each q value, equivalent to a 95% confidence interval. Following Blair and Hufbauer [88], individuals were classified as hybrids if their q value was <0.90. If an individual's proportion did not include one, introgression likely occurred. The online software CLUMPAK was used to identify clustering modes and packaging population structure inferences across the K values, as well as for graphical representation of the STRUCTURE results [89].
Finally, we used the Bayesian algorithms provided in NewHybrids v.1.1 beta [90], which estimate the posterior probability, and performed the independent classification of individuals as Alnus glutinosa and Alnus incana, or a hybrid based on their DNA genotypes. We considered the following categories: AI-A. incana, AG-A. glutinosa, F1 and F2 firstand second-generation hybrids, and two backcrosses: first to A. incana (0_Bx) and second to A. glutinosa (1_Bx). The NewHybrids algorithm was run with Jeffreys-like priors with 500,000 iterations following a 500,000-iteration burning. At the end, we combined the information obtained from Structure and NewHybrids to determine the specific hybrid class to which an individual tree was most likely to belong.
Finally, when the putative hybrids and pure species individuals were identified, the genetic diversity parameters were calculated for the three Alnus spp. groups: number of different alleles (Na), number of effective alleles (Ne), observed (Ho)/expected (He)/unbiased (uHe) heterozygosity, and fixation index (F) based on 15 microsatellite loci using the GenAlEx 6.5 software [91]. Pairwise Nei's genetic distance [92] was estimated among three groups identified by STRUCTURE analysis [91,93]. The assessment and visualization the number of private alleles were performed in the Poppr R package [94]. Allelic richness (Ar) was estimated with the FSTAT 2.9.3. software [95], the lowest number of samples (23) was used for rarefaction. The software estimates allelic richness per locus, sample, and samples overall. Allelic richness is a measure of the number of alleles independent of sample size, thus allowing for comparison between different sample sizes among populations. Missing data were assessed among the loci and three groups of Alnus spp. and visualized by the R package poppr [94]. Putative species-specific alleles were identified by a manual examination of the GenAlEx output and GeneMapper profiles (data not shown). Discriminant analysis of principal components (DAPC) was used to proof the clustering of individuals based on results from STRUCTURE analysis (R package adegent 2.0.0 [96,97]). To test the associations among the species groups based on traditional Nei's [98] genetic distances, we ran UPGMA cluster analysis with the R package poppr with 10,000 bootstrap replicates.

Statistical Analysis of Leaf Morphometric Traits
We used an analysis of variance (ANOVA XLSTAT 2020.3.1, Addinsoft [99]) to estimate differentiation in leaf morphology traits among the species groups as assigned by the DNA analyses. The traits with the highest F-values were used for further PCA analysis aimed to cluster the individuals based on the morphology traits (PC-ORD5 soft [100]). We used variance/covariance cross-products matrix and, for scores for variant, used distance-based biplot in PCA calculation. Significance of PCA scores of morphological traits was calculated using Pearson (R) and Kendell (tau) correlations. Based on this analysis, we identified the key leaf morphology traits for the discrimination among and identification of hybrids of alder species.

Species Genetic Differentiation
Based on the on-site morphologic identification according to leaves and stem traits, we grouped all alder individuals into three groups: A. glutinosa (13 trees), A. incana (14 trees), and putative A. × hybrid (162 trees). In total, 189 trees from the three target groups were analyzed using 15 microsatellite loci. All the microsatellite loci were polymorphic and amplified 162 alleles in total. The mean number of alleles varied from 2.67 at loci A2 to 12.33 at loci A26 (Table S2).
The STRUCTURE Bayesian clustering revealed two genetic clusters that best explain the molecular variation (delta K = 975.007; Figure S1 and Table S3). A clear separation of Alnus glutinosa (AG) and A. incana (AI) as pure reference samples was observed with a few exceptions and among the remaining group of alder hybrids (AH), they were clearly visible ( Figure 3). The STRUCTURE program identified 24 individuals as A. incana, 132-A. glutinosa, and 33-hybrids, of which five individuals are closer to A. glutinosa, six to A. incana, and 22 are F1 hybrids (Table S4). package adegent 2.0.0 [96,97]). To test the associations among the species groups based on traditional Nei's [98] genetic distances, we ran UPGMA cluster analysis with the R package poppr with 10,000 bootstrap replicates.

Statistical Analysis of Leaf Morphometric Traits
We used an analysis of variance (ANOVA XLSTAT 2020.3.1, Addinsoft [99]) to estimate differentiation in leaf morphology traits among the species groups as assigned by the DNA analyses. The traits with the highest F-values were used for further PCA analysis aimed to cluster the individuals based on the morphology traits (PC-ORD5 soft [100]). We used variance/covariance cross-products matrix and, for scores for variant, used distance-based biplot in PCA calculation. Significance of PCA scores of morphological traits was calculated using Pearson (R) and Kendell (tau) correlations. Based on this analysis, we identified the key leaf morphology traits for the discrimination among and identification of hybrids of alder species.

Species Genetic Differentiation
Based on the on-site morphologic identification according to leaves and stem traits, we grouped all alder individuals into three groups: A. glutinosa (13 trees), A. incana (14 trees), and putative A. × hybrid (162 trees). In total, 189 trees from the three target groups were analyzed using 15 microsatellite loci. All the microsatellite loci were polymorphic and amplified 162 alleles in total. The mean number of alleles varied from 2.67 at loci A2 to 12.33 at loci A26 (Table S2).
The STRUCTURE Bayesian clustering revealed two genetic clusters that best explain the molecular variation (delta K = 975.007; Figure S1 and Table S3). A clear separation of Alnus glutinosa (AG) and A. incana (AI) as pure reference samples was observed with a few exceptions and among the remaining group of alder hybrids (AH), they were clearly visible (Figure 3). The STRUCTURE program identified 24 individuals as A. incana, 132-A. glutinosa, and 33-hybrids, of which five individuals are closer to A. glutinosa, six to A. incana, and 22 are F1 hybrids (Table S4). Genetic differentiation of the alder species groups based on the Bayesian admixture clustering with K = 2 (highest delta K value of 975.007 was for K = 2; Figure S1 and Table S3). In the plot, individuals are represented by a thin vertical bar divided into K = 2 colored segments that represent the individual's estimated membership fractions. The black vertical lines separate the three species groups.
The analysis by NewHybrids program allowed us to verify the species and hybrids assignment. NewHybrids have identified 20 individuals of A. incana, 132 of A. glutinosa and 37 of hybrids, of which 15 are pure F1 hybrids (q > 0.9), 9 are F1 hybrids with 0.9 < q > 0.7, 3 are hybrids are closer to A. incana, 5 are closer to A. glutinosa, and the other 5 hybrids as backcrosses to A. incana or to A. glutinosa (Table S4). Genetic differentiation of the alder species groups based on the Bayesian admixture clustering with K = 2 (highest delta K value of 975.007 was for K = 2; Figure S1 and Table S3). In the plot, individuals are represented by a thin vertical bar divided into K = 2 colored segments that represent the individual's estimated membership fractions. The black vertical lines separate the three species groups.
The analysis by NewHybrids program allowed us to verify the species and hybrids assignment. NewHybrids have identified 20 individuals of A. incana, 132 of A. glutinosa and 37 of hybrids, of which 15 are pure F1 hybrids (q > 0.9), 9 are F1 hybrids with 0.9 < q > 0.7, 3 are hybrids are closer to A. incana, 5 are closer to A. glutinosa, and the other 5 hybrids as backcrosses to A. incana or to A. glutinosa (Table S4).
After species identification based on the Bayesian clustering, we used the discriminant analysis of principal components (DAPC) to further confirm the A. incana, A. glutinosa, and A. × hybrid genetic differentiation on the ordination axes based on SSR scores (Figure 4). The results showed a clear speciation of the tree species groups as identified by the NewHybrids analyses (Figure 4).
The UPGMA clustering based on Nei's [98] genetic distances showed closer genetic ties between the hybrid group and A. incana ( Figure 5). Similar results were obtained based on pairwise genetic distance according to Nei [92] (Table S5).
After species identification based on the Bayesian clustering, we used the discriminant analysis of principal components (DAPC) to further confirm the A. incana, A. glutinosa, and A. × hybrid genetic differentiation on the ordination axes based on SSR scores (Figure 4). The results showed a clear speciation of the tree species groups as identified by the NewHybrids analyses ( Figure 4). The UPGMA clustering based on Nei's [98] genetic distances showed closer genetic ties between the hybrid group and A. incana ( Figure 5). Similar results were obtained based on pairwise genetic distance according to Nei [92] (Table S5).  [98]. The significance of branch nodes was tested with 10,000 bootstraps among loci (indicated by percentage of bootstraps separating a given branch).
Finally, after successful hybrid identification based on genetic markers, we sorted all the individuals into three groups of pure A. glutinosa (132 individuals), A. incana (24 individuals), and A. × hybrid (33 individuals) and identified two microsatellite loci discriminating A. incana from A. glutinosa and their hybrids. Putative species-specific alleles were identified by manual examination of the GenAlEx output of loci Ag01 and Ag30. Locus Ag01 was monomorphic and amplified only one allele (129/129) in A. incana. In A. glutinosa and their hybrids, this locus amplified 11 and 8 alleles, respectively, and locus ( Figure 4). The results showed a clear speciation of the tree species groups as identified by the NewHybrids analyses ( Figure 4). The UPGMA clustering based on Nei's [98] genetic distances showed closer genetic ties between the hybrid group and A. incana ( Figure 5). Similar results were obtained based on pairwise genetic distance according to Nei [92] (Table S5).  [98]. The significance of branch nodes was tested with 10,000 bootstraps among loci (indicated by percentage of bootstraps separating a given branch).
Finally, after successful hybrid identification based on genetic markers, we sorted all the individuals into three groups of pure A. glutinosa (132 individuals), A. incana (24 individuals), and A. × hybrid (33 individuals) and identified two microsatellite loci discriminating A. incana from A. glutinosa and their hybrids. Putative species-specific alleles were identified by manual examination of the GenAlEx output of loci Ag01 and Ag30. Locus Ag01 was monomorphic and amplified only one allele (129/129) in A. incana. In A. glutinosa and their hybrids, this locus amplified 11 and 8 alleles, respectively, and locus  [98]. The significance of branch nodes was tested with 10,000 bootstraps among loci (indicated by percentage of bootstraps separating a given branch).
Finally, after successful hybrid identification based on genetic markers, we sorted all the individuals into three groups of pure A. glutinosa (132 individuals), A. incana (24 individuals), and A. × hybrid (33 individuals) and identified two microsatellite loci discriminating A. incana from A. glutinosa and their hybrids. Putative species-specific alleles were identified by manual examination of the GenAlEx output of loci Ag01 and Ag30. Locus Ag01 was monomorphic and amplified only one allele (129/129) in A. incana. In A. glutinosa and their hybrids, this locus amplified 11 and 8 alleles, respectively, and locus Ag30 amplified only two alleles for A. incana (96 and 98) and for A. glutinosa, and their hybrids seven and six, respectively.

Genetic Diversity
The genetic diversity indices were calculated after the Bayesian assignment of 189 individuals into three groups of pure A. glutinosa (132 individuals), A. incana (24 individuals), and A. × hybrid (33 individuals) (Table 1). Interestingly, the hybrid group possessed marked higher values for most of the genetic diversity indexes, including the observed and expected heterozygosity ( Table 1). The mean number of alleles (Na) varied from 5.07 in A. incana to 10.07 in A. glutinosa, with an overall average of Na = 7.76. The mean number  (Table 1 and Figure S2). The inbreeding coefficient (F IS ) was highest in A. incana followed by A. glutinosa (F IS = 0.057) and A. × hybrid (F IS = −0.066) with an overall average 0.063 (Table 1). In total, 39 private alleles were present and differently distributed over three alder species groups, with the highest number of private alleles observed in the group of A. glutinosa (Np = 34) ( Table 1 and Figures S2 and S3). When comparing the two pure species of A. glutinosa and A. incana, for all the genetic diversity indices, A. incana had the lowest values.

Leaf Morphology Variation
After assigning the individuals into the species groups based on the DNA markers, morphological analysis of leaves according to twelve morphological traits was performed. The ANOVA revealed significant species effects on all the leaf morphology traits except for A and I/A (blade length and ratio of petiole length and blade length) ( Table 2).
The PCA analysis was performed on the morphology traits with the highest values of R 2 and F from the ANOVA ( Table 2, rows 1-6). The PCA results showed high correlation and determination coefficients between the above-mentioned leaf traits and the PC1 from the PCA ( Table 3).
The PCA plots of the individual tree values against two major PCs accounted for 95.55%, 4.10%, and 0.28% of the total variance in the first, second, and third PC axes, respectively. Based on leaf morphology traits, pure alder species obtained extreme PC1 scores and clustered onto the opposite ends of the PC1 axis, whereas the putative spontaneous hybrids were mainly located in between the pure species (Figure 6a,b).
For black alder, the values for all investigated leaf morphology traits were high, except for the pubescence (P), secondary veins pairs (N), and blade width and length ratio (B/A). Though the hybrid alder were found to occupy intermediate positions between A. glutinosa and A. incana. The A. incana had more pubescens and secondary veins pairs (Table 2, Figure S4). Variety of alder leaf shapes within the species can be seen in Figure 7. The PCA plots of the individual tree values against two major PCs accounted for 95.55%, 4.10%, and 0.28% of the total variance in the first, second, and third PC axes, respectively. Based on leaf morphology traits, pure alder species obtained extreme PC1 scores and clustered onto the opposite ends of the PC1 axis, whereas the putative spontaneous hybrids were mainly located in between the pure species (Figure 6a,b).  Figure S4). Variety of alder leaf shapes within the species can be seen in Figure  7.   Figure S4). Variety of alder leaf shapes within the species can be seen in Figure  7.

Discussion
Our study focused on two autochthonous Alnus genus (Betulaceae) species-A. glutinosa and A. incana, with overlapping ranges in Lithuania. A. incana is inhabiting drier sites than A. glutinosa and is less common in the southern part and more frequent in the north-eastern part of the country. Meanwhile, A. glutinosa is most common on wet sites

Discussion
Our study focused on two autochthonous Alnus genus (Betulaceae) species-A. glutinosa and A. incana, with overlapping ranges in Lithuania. A. incana is inhabiting drier sites than A. glutinosa and is less common in the southern part and more frequent in the north-eastern part of the country. Meanwhile, A. glutinosa is most common on wet sites in the lowlands of central Lithuania and in the southwestern part of the country. However, warming climates often alter the site moisture regime, thus providing better chances for co-occurrence and hybridization among tree species that may lead to sympatric speciation events following the global environmental change [101]. Our study provides strong evidence for such sympatric speciation events among alder species complex in overlapping natural habitats. Assuming climatic predictions for northern Europe of frequent incidence of high precipitation over a short time in otherwise dry ecosystems, we may expect a diverse moisture regime within a single site. Here, a desiccation and moisture stress may occur on the same site, where alders possessing integrated gene pools of A. glutinosa × incana could be an example of an adaptive species introgression for better tolerance of such complex stresses [102].
For this study, we collected leaves mainly from those alders that were morphologically different from the usual morphometric traits of A. glutinosa and A. incana (bark color, number of leaf vein pairs, and tooths of leaf edging) [28,36,48]. The leaves of most of the trees that were assigned to the A. × hybrid group had 7-9 vein pairs, the leaf edge was slightly serrated and double-toothed, and the form of leaf was not traits of pure A. glutinosa or A. incana. However, only about 30 trees of the morphology-based 162 putative hybrids proved to be such based on the DNA markers. Thus, the accuracy of the onsite morphology identification of alder hybrids is low.
In the current study, we have identified a set of 15 nuclear microsatellite markers for an efficient genetic discrimination within the Alnus species complex. Our findings agree with a number of studies where microsatellites or other DNA markers were used to study the level of genetic diversity and hybridization in Alnus species [26,27,29,31,33]. First, based on Bayesian clustering analysis (STRUCTURE soft.), everyone was assigned to either pure A. incana, pure A. glutinosa, or a hybrid; thus, we have identified 24 individuals as A. incana, 132 as A. glutinosa, and 33 as putative hybrids. The analysis with NewHybrids software assigned everyone to either pure species of A. incana or A. glutinosa and different orders of hybrids (F1, F2 and Backcrosses) as follows: 20, 132, and 37 individuals as A. incana, A. glutinosa, and putative hybrids, respectively. For the hybrid populations, 24 individuals were F1 hybrids and 13 have F2, backcrosses, or mixed traits. These multiple backcrossing events usually complicate the morphology-based identification of the hybrids. However, most of the hybrids identified in our study belonged to the F1 generation. In contrast, Villani et al. [33] in southern Italy have not found any F1 hybrids, and most of the hybrids were F2 hybrids among A. cordata and A. glutinosa with a low frequency of backcrosses toward the parental species. According to some authors [49,103], hybrids between A. incana and A. glutinosa most commonly occur when the mother tree is A. incana. It was proven by artificial crosses and was found that successful hybridization takes a place only when the mother tree was A. incana. In contrast, hybridization was unsuccessful when the mother tree was A. glutinosa. Moreover, in pilot plantation, where we have collected alders as a putative hybrid, genetic data analysis recognized only one hybrid. All other trees were identified as black alders. Furthermore, from other leaf collection sites, we identified 82 putative hybrids by morphometric traits; however, genetic analysis determined 32 hybrids. These results shows that their morphological and genetic boundaries do not fit and are much wider than the phenotypic or vice versa. The variation of hybridization rate among the sampled trees suggests that natural crossing between the two species occurs at lower frequency than we were expecting and probably depends on diverse local factors such as synchronization, which favors hybridization.
Overall, in our study, evidence for hybridization between A. glutinosa and A. incana was found from morphological and genetic backgrounds; however, it was inconsistent. Based on Bayesian clustering results, putative hybrids constituted 17.5% (33 out of 189 individuals), which was comparable with the results of Villani et al. [33], which identified significant introgression (from 9.1% to 50.0%) in some natural Alnus spp. populations in southern Italy.
However, the initial number of pure reference samples per parental species was not high in our study (e.g., A. glutinosa-13 trees and A. incana-14 trees), which may influence the genetic assignment into hybrid groups. Furthermore, morphological leaf traits typically used to identify alders in the field did not reliably distinguish taxonomy within the Alnus complex [28]. However, our results show that a combination of molecular data and morphological traits has high potential and could help the classification of the genus Alnus, especially in hybrid individuals. Therefore, further autochthonous Alnus species conservation, breeding, and management measures should take a more detailed genetic examination to enable better species discrimination, which is the basis for in situ and ex situ conservation and for successful breeding programs.
After clear pure species and hybrids identification, the level of genetic diversity was assessed on the three groups: A. glutinosa, A. incana, and their hybrids were compared based on 15 nSSR markers. Our results agreed with previous studies of Alnus spp. and showed a moderate level of genetic diversity among three groups. In the group of pure A. incana trees, the inbreeding coefficient (F IS = 0.197) was highest, and all genetic variation indices were lower (e.g., mean values of mean number of alleles (Na), mean number of effective alleles (Ne), allelic richness (Ar), and expected heterozigosity (He)) in comparison to A. glutinosa and their hybrids. Overall, despite unequal sample size, Alnus spp. hybrids contained higher values of Na, Ne, Ar, and He. In comparison, results presented by Šmid et al. [31], which used nuclear and chloroplast DNA markers to study hybridization zone and find putative hybrids among A. glutinosa and A. rohlena, showed clear separation of the two Alnus species based on both marker types. Genetic diversity was moderate (Na = 4.78 and He = 0.72) but the inbreeding coefficient was high (F IS = 0.33-0.42). The results presented by Villani et al. [33] showed low genetic diversity among and within A. cordata, endemic tree species, populations in comparison with A. glutinosa or A. cordata × glutinosa hybrids. For example, A. cordata × glutinosa hybrids showed higher values for the number of alleles (Na = 6.904) and observed heterozygosity (Ho = 0.489), compared to pure populations of A. cordata and A. glutinosa (Na = 3.107, Ho = 0.396, Na = 5.286, and Ho = 0.468, respectively). Furthermore, all populations with admixture of putative hybrids indicated significant positive inbreeding coefficient (F IS = 0.237-0.461). However, when comparing the results, we should consider the peculiarities of each species and the variation in sample size, geography, and number of loci used.
A. glutinosa leaves vary in their shape, and they are emarginate, rounded, or acute at tip and cuneate, obtuse, or acute at base, and their margins are entire, serrate with a slightly wavy edge, or double serrate ( Figure S5). Many interesting morphological forms were found between alders in our sampling sites ( Figure S6). These results may be related to natural variability within the species, or they might be the consequence of backcrossbreeding or heritability. Morphological investigations in our study were carried out to help to distinguish the species of A. glutinosa from A. incana and their hybrids. We can highlight that F1 A. hybrids in their morphometric traits are closer to A. incana and have about nine pairs of veins (N), pubescence (P) less than A. incana, and a blade maximum width (B) greater than A. incana. However, morphometric data gives just first insights into interspecific hybridization but are not sufficient to assess the level of hybridization among the alder species. Although our morphological assessment in many cases showed alders as pure A. glutinosa or A. incana, the genetic analysis showed that it is a putative hybrid (Table S4, alders id J9, 17BtPL25, BL1, BM1, B3).
In the temporary research plots, A. glutinosa accounted for 88.4%, A. incana for 10.3%, and alder hybrids for 1.3% of all alders (results not shown). Similarly, hybridization events have been reported in other studies on A. glutinosa and A. incana [25,28,36,39,48,50,51,73,84]. Our results are in line with other studies showing that, by a complex of morphometric traits, hybrids of alders hold an intermediate position (e.g., Villani et al. [33] found from 9.1% to 50% of hybrids between A. cordata and A. glutinosa in different populations in Italy).

Conclusions
We used DNA makers to untangle the complex morphology of pure A. incana and A. glutinosa species and their hybrid swarms in natural stands. The result showed that most were F1 hybrids, and other had backcrosses to parental species or mixed traits. It is these multiple backcrossing events that complicated morphology-based recognition of the hybrids. Two microsatellite loci (Ag01 and Ag30) discriminate well between these species. DNA markers supported clear genetic differentiation between the groups based on the morphology of A. glutinosa and A. incana and quite variable values of the probable envelope in the putative hybrid group.
In conclusion, our study provides strong evidence for spontaneous hybridization between sympatric species of A. incana and A. glutinosa in natural forests of northern Europe. The hybrid alders seem to be genetically closer to A. incana and are of a markedly greater genetic diversity than the corresponding parental species. There are concerns for genetic drift effects in the populations of A. incana.

Supplementary Materials:
The following are available online at https://www.mdpi.com/article/ 10.3390/f12111504/s1, Table S1: List of nuclear microsatellite markers (nSSR's) used in our study; Table S2: The genetic diversity parameters for each of the loci (GenAlex6.5 software [91,93]). Allelic richness (FSTAT 2.9.3 [95]); Figure S1: The results of Bayesian clustering (soft. STRUCTURE2.3.4 [85]) on the most likely number of genetic clusters within the studied three Alnus spp. groups, indicated by the highest delta K value at K = 2 (STRUCTURE HARVESTER soft. [87]);  [85]) on the most likely number of genetic clusters within the studied populations, indicated by the highest delta K value (STRUC-TURE HARVESTER soft. [87]); Table S4: Identification of putative hybrids by STRUCTURE and NewHybrids programs; Table S5: Pairwise genetic distance according to Nei [92] (GenAlEx6.5 [91,93]; Figure S2: Distribution of genetic diversity among three sampled Alnus spp. groups (Na-Mean no. of Different Alleles; Ne-Mean no. of Effective Alleles; Ar-Mean allelic richness (based on min. sample size of 23 diploid individuals.), Npriv-No. of Private Alleles; He-Expected Heterozygosity) (soft. GenAlEx 6.5 [91,93,94]). Groups abbreviations in Table 1; Figure S3: Private alleles distribution among the studied three Alnus spp. groups (189 individuals) (R package poppr [94]); Figure S4: The PCA ordination plots of alder trees given separately for species-specific leaf morphology traits. The symbol size indicates the relative size of the morphology traits in the entity. The minimum value (zero) is shown on an overlay as the smallest size for that symbol. Abbreviations shown in Table 2, species ID-in Figure 6.