Genomic-Guided Conservation Actions to Restore the Most Endangered Conifer in the Mediterranean Basin.

1 of 12 Molecular Ecology, 2024; 0:e17605 https://doi.org/10.1111/mec.17605 Molecular Ecology ORIGINAL ARTICLE OPEN ACCESS Genomic- Guided Conservation Actions to Restore the Most Endangered Conifer in the Mediterranean Basin JoséCarlosdelValle1 | MontserratArista1 | CarmenBenítez- Benítez1 | PedroLuisOrtiz1 | FranciscoJ.Jiménez- López2 | AnassTerrab1 | FranciscoBalao1 1Departamento de Biología Vegetal y Ecología, Universidad de Sevilla, Facultad de Biología, Sevilla, Spain | 2Departamento Biología y Geología, Física y Química Inorgánica, Universidad Rey Juan Carlos, Móstoles, Spain Correspondence: Francisco Balao ([email protected]) Received: 19 February 2024 | Revised: 14 October 2024 | Accepted: 17 November 2024 Handling Editor: Christine Edwards Funding: This work was supported by European Climate, Infrastructure and Environment Executive Agency, LIFE18/NAT/IT/000164. Keywords: Abies nebrodensis| conservation genomics| hybridisation| inbreeding| reforestation| small populations| withinpopulation assisted gene flow ABSTRACT Species with extremely small population sizes are critically endangered because of reduced genetic diversity, increased inbreeding and hybridisation threats. Genomic tools significantly advance conservation by revealing genetic insights into endangered species, notably in monitoring frameworks. Sicilian fir (Abies nebrodensis) is the most endangered conifer in Europe with only 30 adult trees in an 84- ha area. Using 20,824 SNPs from RAD- seq, employing genome assembly and a custom 120 SNP- array, we evaluated genetic diversity, mating patterns, and effective population size in adult trees, 118 natural seedlings, and 2064 nursery seedlings from past conservation actions. We assessed introgression from neighbouring nonnative fir plantations (~6%) and established an intrapopulation assisted gene flow (AGF) program selecting the most genetically dissimilar individuals and investigating the outcome through simulations. Genomic analysis unveiled significant genetic diversity among adult Sicilian firs, comparable to nonendangered Mediterranean firs with larger populations. However, the genetic diversity of the forthcoming generation declined due to high selffertilisation, leading to marked inbreeding (FIS = 0.38) and an alarmingly low effective population size (Ne = 6). Nursery seedling monitoring revealed similar selfing rates and introgression (~2%) from nonnative firs. Although intrapopulation AGF could help to mitigate genetic loss, it may not alleviate the species vulnerability to imminent environmental challenges, perpetuating the risk of an extinction vortex. Hence, investigating the impact of Sicilian fir population decline and selfing on inbreeding depression, along with exploring the potential of hybrids for genetic load alleviation and future adaptation, is crucial for effective conservation strategies. 1 | Introduction Over the past few centuries, many species of animals and plants have become extinct, while many others teeter on the brink of extinction due to habitat loss, fragmentation, and the reduction of population size to inviable levels (Palomares etal.2012; Benazzo etal.2017; Humphreys etal.2019). Species with extremely small populations present a pressing challenge in contemporary biodiversity conservation, referred to as the ‘small population paradigm’ (Norris2004). These species face an increased risk of extinction due to a confluence of factors, including reduced genetic diversity, increased inbreeding, vulnerability to stochastic events and susceptibility to environmental pressures (Frankham2015; Ralls etal.2018). Hybridisation This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. © 2024 The Author(s). Molecular Ecology published by John Wiley & Sons Ltd. José Carlos del Valle and Montserrat Arista are cofirst authors. 2 of 12 Molecular Ecology, 2024 presents an additional peril, endangering the species integrity and exacerbating the risks for its already impoverished populations (Balao etal.2015). To persist over the shortterm without facing detrimental effects, it is imperative to maintain an effective population size above 100 individuals (i.e., the minimum viable population sensu Frankham, Bradshaw, and Brook2014). Consequently, genetic management has become an essential tool to identify strategies for preserving genetic integrity, mitigating inbreeding depression, and fostering the adaptive potential needed for the survival and longterm persistence in their respective ecosystems (Hogg etal.2022). Furthermore, genomic tools facilitate the monitoring of current conservation status and the assessment of outcomes from conservation actions (Flanagan etal.2018; Humble etal.2023). Assisted gene flow (AGF), replacing natural gene flow by guidedoutcrossing, can thus be a powerful conservation tool to preserve biodiversity, particularly for populations that are experiencing genetic erosion (Pregler et al. 2023). Restoring gene flow in small, isolated populations can increase genetic diversity and fitness (i.e., genetic rescue), reducing the risk of inbreeding depression and, ultimately, population extinction, and, at the same time, accelerate the adaptive potential of species to environmental modifications such as climate change (Aitken and Whitlock2013; Castilla etal.2019). Although the benefits and risks of this practice have been discussed extensively (Frankham2016; Ralls etal.2020; Grummer etal.2022), the practical application of AGF in tree conservation remains largely unexplored (Aitken, Bemmels, and Sally Aitken2015; Browne etal.2019). Conifers have specific conservation needs due to their slow growth, long reproductive cycles, low mutation rates, and specific breeding, pollination and dispersal systems, requiring a meticulous and sustained approach to population restoration (Givnish 1980; Savolainen and Pyhäjärvi 2007; De La Torre etal.2017). Most conifer species are found in large populations characterised by high gene flow, which contributes to significant genetic diversity (Petit and Hampe2006). Conversely, many endangered conifer species occur in small, isolated populations with restricted gene flow, leading to lower levels of genetic diversity (Farjon and Page1999). Sicilian fir (Abies nebrodensis) is the most threatened conifer in the Mediterranean Basin, and likely one of the most endangered species in the world. It is categorised as critically endangered by the IUCN Red List (Thomas2017), listed in AppendixI of the Bern Convention and considered a priority species in Annexes II and IV of the 92/43 EC Habitats Directive (Code 9220). This species constitutes an example of a highly vulnerable species due to its very low population size. The Sicilian fir occupied a broad range in Sicily from the Paleogene to the Holocene (Tinner etal.2016). Its population decreased drastically since the Middle Ages due to anthropogenic pressures (mainly to obtain firewood), fire and soil degradation (Venturella, Mazzola, and Raimondo 1997; Pasta et al. 2020; Frascella et al. 2022). The current population consists of 30 adult trees with a scattered distribution, averaging 621.7 m apart (Figure1a) and 118 seedlings (over 5 yr. old) growing in a sole population over approximately 84 ha within the integral reserve of Madonie Natural Park in Sicily (Venturella, Mazzola, and Raimondo1997). Previous studies suggested that reproductive trees maintained high genetic diversity but typically exhibited FIGURE 1 | Geographical distribution (a) of the 30- adult trees from the unique population of the Sicilian fir (A. nebrodensis). The coloured pie graphs represent the membership probability of each individual according to the Structure results (b) when using K = 2, the most supported numbers of genetic clusters. Structure analysis was performed for the 30 adult trees and the 118 seedlings in the natural population. Adults are sorted by proximity and arrows show seedlings origin from the closest adult tree. 37.0N 37.5N 38.0N 12.0E12.5E13.0E13.5E14.0E14.5E15.0E 15.5E Longitude Latitude 0 1000 2000 3000 4000 Altitude 1M 23M 24M 25M 11M 21M 22M 30M 31M 32M 26M 6M 4M 9M 8M 13M 27M 29M 14M 15M 16M 17M 18M 20M 19M 2M 21M 31M 30 M 22M 32M 29M 20 M 19 M 18 M 17M 16M 2M 11M 10 M 15M 14 M 28 M 9M 13 M 27 M 26 M 8M 12 M 7M 4M 6M 1M 23M 25M 24M 10M 28M (a) (b) 12M 7M 1365294x, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/mec.17605 by Readcube (Labtiva Inc.), Wiley Online Library on [28/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 3 of 12 reduced seed viability (Vicario etal.1995; Vendramin etal.1996; Parducci et al. 2001; Scialabba 2019). This low seed viability strongly indicates the existence of high rates of inbreeding. Moreover, fir species show a pronounced tendency to hybridise (Klaehn and Winieski1962; Balao etal.2020). Extensive plantations of exotic forest species in Sicily, such as silver and Greek fir (A. alba and A. cephalonica, respectively), from 1920s onwards, have thrived in the favourable conditions of the Madonie mountains, leading to their naturalisation (Badalamenti etal.2020; Frascella etal.2022). The potential hybridisation of the Sicilian fir with these introduced firs, some of them located less than 5 km from the native population, may adversely impact seed viability and species integrity (Conte and Cristofolini 2003; Scialabba, Schicchi, and Cordi2005). Many in situ and ex situ conservation efforts for the Sicilian fir have been implemented over the last two decades. One of the most important is the foundation of the forest nursery ‘Piano Noce’ within the Madonie Natural Park. The nursery harbours more than 16,000 seedlings of different cohorts resulting from several breeding programs via natural pollination of the adult trees of A. nebrodensis. In addition, interventions aimed at removal of nonnative fir species have also been carried out in the vicinity of the population (Venturella, Mazzola, and Raimondo1997; Frascella etal.2022; De Rogatis etal.2022). However, there is scarce genetic information available for both the naturally occurring and the nurserypropagated seedlings, hindering the assessment of a viable longterm conservation plan. To address this limitation, we conducted a genomewide analysis of Sicilian fir genetic variation and compared our results to the genetic diversity observed in other Mediterranean fir species. We then developed an SNP- array for genetic monitoring of the Sicilian firs. We evaluated the genetic diversity and relationships among the 30 adult trees in the natural population. We assessed the degree of introgression with neighbouring plantations of nonnative firs. Through simulations, we designed an AGF program that prioritises crossing individuals with greater genetic dissimilarity, with the goal of ensuring the shortterm survival. Finally, we determined the parentage of 118 naturally regenerated seedlings, along with 2064 seedlings cultivated in the forest nursery ‘Piano Noce’, aiming to ascertain outcrossing rates and occurrences of selffertilisation. 2 | Materials and Methods 2.1 | Fir Species Sampling and RAD- Seq Libraries We used restriction siteassociated DNA sequencing (RAD- Seq) to identify highquality and informationrich SNPs for genotyping of Sicilian firs and detect hybridisation with alien silver and Greek fir. We sampled five individuals from the unique population of Sicilian fir, 15 individuals from three silver fir populations from the centralsouthern Italian Peninsula and 15 individuals from three populations of Greek fir. All genetic material (Table1) was obtained from previous studies on the genetic diversification of Mediterranean firs (Balao etal.2020; Balao etal.2024). We selected these natural ‘genetically pure’ populations from outside Sicily (putatively nonadmixed) for comparative purposes. In the case of silver fir, we chose several populations that share a more similar phylogeographical TABLE 1 | Population information and genetic diversity indices of different Abies species analysed using 20,824 SNPs from RAD- seq. Species Locality Country Altitude Latitude Longitude N Private alleles HSπ FIS A. alba Piano dábete (S Apennines) Italy 991 38.26388 16.02539 5592 0.179 ± 0.0015 0.223 ± 0.0020 0.093 ± 0.0115 A. alba Laurenzana (S Apennines) Italy 1192 40.41872 15.97601 5660 0.169 ± 0.0015 0.212 ± 0.0020 0.074 ± 0.0116 A. alba Selva di Castiglione (C Apennines) Italy 1005 41.73423 14.31645 5483 0.163 ± 0.0016 0.204 ± 0.0021 0.065 ± 0.0121 A. cephalonica Kalavryta Greece 633 38.07316 22.12514 5589 0.172 ± 0.0016 0.216 ± 0.0021 0.057 ± 0.0118 A. cephalonica Oros Ainos Kefalonia Greece 891 38.15680 20.66164 5715 0.161 ± 0.0015 0.201 ± 0.0020 0.072 ± 0.0118 A. cephalonica Oros Pateras Greece 747 38.14394 23.31128 5728 0.171 ± 0.0016 0.218 ± 0.0021 0.060 ± 0.0117 A. nebrodensis Parco delle Madonie (Sicily) Italy 1659 37.87568 14.04049 51029 0.161 ± 0.0015 0.200 ± 0.0020 0.073 ± 0.0116 1365294x, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/mec.17605 by Readcube (Labtiva Inc.), Wiley Online Library on [28/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 4 of 12 Molecular Ecology, 2024 history, allowing for a more precise comparison of genetic diversity (Piotti etal.2017; Balao etal.2024). We extracted DNA and prepared RAD libraries following Balao etal.(2020). Genomic DNA was digested by using highfidelity SbfI (New England Biolabs) and the resulting fragments were double barcoded. The library was sequenced in a separate lane of an Illumina flowcell HiSeq 2500 at the VBCF NGS Unit (www. vbcf. ac. at/ ngs) as 100 bp singleend (SE) reads. 2.2 | SNP Identification and Population Genomics Quality filtering and demultiplexing of the RAD- seq library were performed with deML (Renaud etal. 2015) and STACKS v. 2.53 (Rochette, Rivera- Colón, and Catchen2019). Following this, we aligned each raw read fastq file to the silver fir genome (Mosca et al. 2019), using Bowtie2 v. 2.4.1 (Langmead and Salzberg2012) with - - sensitive and - - nounal settings. Singlebest alignments were then sorted and converted from SAM to BAM format using Samtools v. 1.10 (Li etal.2009). To assemble RAD loci, we used the ref_map.pl. pipeline implemented through Stacks v. 2.53. Then, we utilised the populations script to export SNP data into various formats for subsequent analyses. We investigated the nucleotide diversity (π), heterozygosity (Hs) and inbreeding (FIS) of the Sicilian fir and compared to silver and Greek fir populations with a nonparametric Kruskal–Wallis oneway analysis of variance. The effect sizes of these tests were calculated using ε2, using the rcompanion package in R v. 4.3.2 (R Core Team2023). 2.3 | OpenArray Development for Genotyping and Hybrid Detection We customised a panel of 120 SNPs by PCR- based OpenArray Technology (Thermo Fisher Scientific, USA) for genotyping the Sicilian fir samples. This panel was designed as a costeffective tool for paternity analysis and hybrid identification among the Sicilian fir seedlings. This strategy has been effective in tree breeding programs for other conifers (Fuentes- Utrilla etal.2017; Silva etal.2020). The relatively small number of SNPs of this panel has been demonstrated to be sufficient for accurate parentage determination and for assessing genetic introgression in both animal and plant populations (Flanagan and Jones2019). The entire process leading to the selection of the SNPs is described in detail in FigureS1. First, we selected 100 highly informative SNPs to assess the genetic structure of the natural population as well as the origin of the seedling from the forest nursery ‘Piano Noce’. These unlinked highquality SNPs with a maximum information for genotyping and paternity exclusion were selected using a combination of Stacks, Plink2 v. 2.00a3.7 (Chang etal.2015), PopLDdecay v. 3.40 (Zhang etal.2019) based on the following criteria: ‘Biallelic’, ‘No other SNPs within 4 kb’, ‘r2 < 0.01’, ‘Minor allele frequency [MAF] > 0.05’, ‘Minimum 1 individual heterozygote’, ‘Minimum 1 individual homozygote for alternative alleles’, ‘100 bp flanking sequence available on both sides of the variable position’. To minimise the probability of SNPs being affected by natural selection, we kept the SNPs in intergenic regions (avoiding 1 kb downstream and 2 kb upstream regions of genes) using SnpEff and SnpSift v. 4.3.1 t (Cingolani et al. 2012). Secondly, we selected 20 highquality SNPs (unlinked markers with high MAF) to detect hybridization between the Sicilian fir and the other firs inhabiting the Madonie Natural Park. These markers followed the same selection criteria previously described and showed the highest speciesspecific diagnostic power, showing the highest loadings in the first two axes of a Discriminant Analysis of Principal Components (DAPC) performed with the adegenet package v. 2.1.10 (Jombart and Ahmed, 2011) in R v. 4.0.3 (R Core Team 2020). A Principal Component Analysis (PCA) using the R package dartR v. 2.7.2 (Mijangos etal. 2022) was used to corroborate the discriminant power of the 20 selected SNPs. 2.4 | SNP- Array Validation and Genotyping of Plants From the Natural Population To investigate the genetic diversity and relatedness of Sicilian fir individuals, we first collected leaf material for genetic analysis from the 30 adult trees and the 118 seedlings inhabiting the population. DNA was isolated using the NucleoMag Plant kit (Macherey- Nagel, Germany) according to the manufacturer's protocol. Before starting the genotyping of samples, we initially validated the specially developed 120- SNPs arrays for this species. To achieve this, we genotyped 24 samples (12 in duplicate), evaluating replicability and genotype accuracy. Additionally, we calculated the genotyping error rate and allele dropout. Subsequently, we carried out a PCA to verify that replicates clustered together. We used the 120 SNP panel to genotype all the individuals in the natural population. As tree spatial isolation and selfing can limit gene dispersion and create finescale spatial genetic structure within the population (Vekemans and Hardy2004), we explored the population genetic structuring of the natural population using Structure v. 2.3.4 (Pritchard, Stephens, and Donnelly2000). We estimated the number of genetic clusters (K) by assigning individuals in undefined mixture clusters under a Bayesian framework. We conducted 10 independent runs of 1,000,000 iterations each, with a burnin period of 100,000 for each value of K from 1 to 5. Then, the best K that fit the data was calculated using the Δk method (Evanno, Regnaut, and Goudet2005) in Structure Harvester (Earl and VonHoldt2011). We inferred the pedigree relationships in the natural population of Sicilian fir using Colony v. 2.0.6.6 (Jones and Wang2010) with the fulllikelihood approach. This method incorporates locusspecific error rates, enhancing the estimation of parent– offspring probabilities, and shows high efficiency in offspring assignment even with a low number of SNPs (Hauser etal.2011; Harrison etal. 2013). Our analysis incorporated prior information on parental (all reproductive adults) and maternal genotypes (the closest adult to the seedling). We assumed monoecy, diploidy, both male and female polygamy, and potential inbreeding. For the full likelihood calculation, we employed ‘very high likelihood precision’ setting, ‘very long’ run length, sibship scaling, updated allele frequencies and weak sibship priors. The genotyping error rate (estimated from replicates) was set to 0.01 for all loci. We performed three runs with different random number seeds to check the reliability of the results. Additionally, the seedlings with a putative hybrid origin (i.e., with an uncertain genetic origin) were examined with a PCA using the 20 SNPs from 1365294x, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/mec.17605 by Readcube (Labtiva Inc.), Wiley Online Library on [28/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 5 of 12 the 120 SNP panel designed for the discrimination of fir species. We also investigated introgression patterns for these seedlings using a Bayesian MCMC approach with the R package gghybrid v. 1.0.0.0 (Bailey, 2024), which uses the maximumlikelihood approach (Buerkle, 2005) to estimate the genomewide hybrid index. We estimated hybrid indices using samples from silver fir and Greek fir as independent parental populations, with indices ranging from 0 (pure Sicilian fir) to 1 (pure alien fir), where 0.5 indicates a maximally hybrid individual. For the adult trees, we used the adegenet package to estimate the genetic diversity at the individual level by calculating the following coefficients: HsExp, standardised heterozygosity based on the mean expected heterozygosity; HL, homozygosity by loci; and INBR, inbreeding coefficient. Finally, the effective population size (Ne) and inbreeding level (FIS) were estimated using the software Colony2. Colony2 calculated a maximum likelihood Ne estimate assuming nonrandom mating using the sibship relatedness method, which is more flexible and accurate than other existing methods such as linkage disequilibrium, heterozygote excess, and the temporal method (Wang2009). To assess whether genetic information could be used in management strategies to mitigate the alleged low population diversity/ inbreeding, we studied the population genetic structure and relatedness of the 30- adult trees to design a genomeinformed AGF program. To do so, we used the adegenet package to perform a DAPC using a K- mean clustering (K = 1–30, the total number of adult trees). Additionally, we used the related package v.1.0 (Pew etal.2015) to calculate the Ritland estimator (RIT) of pairwise coancestry (Ritland1996), which allowed us to infer the genetic relatedness of all adult trees. Finally, considering the absence of other Sicilian fir populations for outcrossing, we selected the 12 pairs of adults with the lowest RIT values (i.e., those individuals genetically more dissimilar) to simulate an artificial population in the AGF conservation program. We used the function hybridise from the adegenet package to generate 10 progenies from each of the 12 proposed crosses. Then, we compared the genetic diversity, inbreeding, and effective population size of adults, natural seedlings, the simulated AGF population (n = 120) and the AGF nextgeneration population. This latter comprised the 120 seedlings from the AGF population, the 118 natural seedlings, and the 19 seedlings from the nursery that resulted from outcrossing (see Results section). Effective size (Ne) was estimated for each population using the software NeEstimator v. 2.1 (Do etal.2014), utilising the Linkage Disequilibrium (LD) approach and setting the lowest allele frequency value to 0.05. 2.5 | Pedigree Analysis of Seedlings From the Forest Nursery We also investigated the genetic diversity and pedigree of 2064 seedlings at the forest nursery using the 120- SNP panel. These seedlings were derived from eight maternal individuals, which were established for forest restoration using seeds collected between 2008 and 2019, ensuring the analysed seedlings were at least 4 years old to increase transplant survival. Plant material sampling was carried out proportionally to the number of seedlings available from each maternal individual (ranging from 100 to 5000; Table S1). To assess parentage and detect putative hybrids with other alien firs, we genotyped the samples with the 120 SNP panel. Then we conducted paternity tests with the 30 adult trees and determined the rate of outcrossing, inbreeding and selffertilisation using Colony2, as described above. We also estimated the hybrid index for the seedlings with a putative hybrid origin (i.e., with an uncertain genetic origin) using silver fir and Greek fir as parental populations and gghybrid R package, as previously described. 3 | Results 3.1 | Genomic Diversity of the Sicilian Fir and Relatives The RAD- seq of the 35 individuals from populations of Sicilian, silver and Greek fir produced a final STACKS catalogue a total of 365,582 RAD loci (composed of 34,325,726 genomic sites) with an average coverage per sample (SD) of 51.4 (38.5) reads per locus. After filtering for polymorphic RAD loci with MAF > 0.05, we obtained a final data set with 20,824 SNPs. The raw data was deposited in the NCBI Sequence Read Archive (BioProject ID PRJNA563575). Estimates of genetic diversity in the populations were notably consistent across the three fir species (Table1). Although the Sicilian fir exhibited the lowest values for π and Hs, these differences were statistically similar when compared to the populations of the other two firs (Kruskal–Wallis tests with p > 0.05, n = 7 and ε2 = 0.381). The Sicilian fir population showed similar FIS (Kruskal–Wallis test, χ2 = 2.285, p > 0.05, n = 7, and ε2 = 0.524) to those of the silver fir (0.07 [0.014]) and Greek fir (0.06 [0.007]). Furthermore, it is noteworthy that the Sicilian fir displayed the highest number of private alleles, with 1029, surpassing the populations of the other species which ranged from 483 to 728 (Table1). The pairwise F'ST values varied between 0.07 (silver fir pop1- silver fir pop2) and 0.26 (Sicilian fir—Greek fir pop 2). Finally, the Sicilian fir showed the lowest F'ST pairwise with the silver fir pop1 (0.14; TableS2). 3.2 | Development of SNP- Array for Conservation of Sicilian Fir We selected 120 highquality SNPs from the 3536 SNPs called with MAF > 0.05 in the five individuals of Sicilian fir. After initial quality filtering, 459 SNPs remained. In addition, functional annotations showed that 351 out of 459 SNPs were in intergenic regions. After excluding promotors and gene regions (46 downstream and 59 upstream), 268 candidate SNPs for the panel remained and 120 SNPs were finally selected. Validation of the 120 SNP- array revealed > 99% replicability and the PCA successfully separated different samples but clustered duplicates together. Call rates (i.e., the proportion of samples that were assigned a genotype call compared to the total number of samples) of the analysed samples ranged from 64.6% to 77.1%. High genotyping failures (> 50%) was detected at 15 out of 100 SNPs designed for paternity assignment, which were discarded from downstream analyses. Furthermore, the 20 SNPs selected to discriminate putative hybrids maximised the separation of the three fir species (the first two components of the PCA accounted for 42.5% and 20.8% of the total variation respectively; FigureS2). 1365294x, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/mec.17605 by Readcube (Labtiva Inc.), Wiley Online Library on [28/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 6 of 12 Molecular Ecology, 2024 3.3 | Gene Diversity and Population Structure in the Natural Population The structure analysis showed that the most likely number of genetic groupings was K = 2 (Figure 1b), exhibiting mild geographical structure. The first cluster comprised four adult trees and 41 out of the 118 seedlings, which were related to one (22 M) of the four adult trees in the genetic cluster. The second cluster included the remaining adult trees and seedlings from the population. The effective population size was very low (Ne = 6; 3–21; lower and upper confidence interval at 95%) and the level of inbreeding was moderate (FIS = 0.373). The results of the paternity assignments identified 109 of the 118 seedlings (92.4%) as purebred Sicilian firs. Of those, 103 seedlings from the natural population originated by selffertilisation and only six of them resulted from outcrossing (only five out of the 30 adult trees in three parental combinations were involved in outcrossing). Additionally, our results disclosed a potential hybrid origin of nine seedlings which were preliminary assigned by proximity to the mother trees 16 M, 18 M, 20 M, 21 M and 22 M. The genetic origin of three of the putative hybrids could not be assigned to any of the 30 adult trees occurring in the population., For the remaining six seedlings only one parent was identified, originating from four different maternal plants. Of the nine seedlings with a putative hybrid origin, the PCA showed that only one of them (20.1P) was notably closer to the cloud point of the silver fir (Figure2a). However, estimates of the hybrid index suggest other potential hybrid seedlings (Figure2b,c). For these two seedlings (20.1P and 22.18P), hybrid values were higher in comparisons involving Greek fir than those involving silver fir, suggesting that these two putative hybrid seedlings were likely hybrids between Sicilian fir and Greek fir, with hybrid indices ranging from 0.41 to 0.78 (Figure2c). Additionally, five more seedlings showed low hybrid indices ranging between 0.11 and 0.22, likely due to backcrossing of F1 hybrids with other Sicilian firs (Figure2c). A notable finding is that one Sicilian fir adult (31 M) was likely a Greek fir (hybrid index = 1), while two other adults (21 M and 26 M) exhibited a hybrid index of 0.5, suggesting an F1 hybrid origin between these two species (Figure2c). These results were corroborated by their positions in the PCA analysis (Figure2a). For adult trees, analyses showed mild genetic diversity at the individual level (TableS3). The mean value of the standardised observed heterozygosity (HsExp) was 0.74 (0.19), and ranged from 0.32 to 1.13. The individual homozygosity by loci (HL) ranged from 0.50 to 0.86, with an average value of 0.67 (0.09). Finally, the inbreeding coefficient (INBR) showed values that ranged from −0.20 to 0.79, with an average value of 0.21 (0.23). DAPC analysis found three genetic clusters with some genetic admixture that were comprised of 12, 11 and seven individuals, respectively (Figure S3a). Coancestry analysis using RIT pairwise estimations revealed overt differences of the genetic relatedness of the 30- adult trees, ranging from −0.384 to 1.058 (a matrix of pairwise coancestry among all adult trees is depicted in FigureS3b). Finally, the Mantel test showed absence of any spatial correlation between geographical distance and genetic diversity from adult trees (r = −0.06; n = 30, p > 0.05). FIGURE 2 | (a) Principal component analysis (PCA) displaying nine seedlings of uncertain origin from the natural Sicilian fir population and other alien fir species. (b) Bayesian estimates of hybrid index and the 95% credible interval for Sicilian fir adults, silver fir samples and seedlings with an uncertain origin (i.e., seedlings with a putative hybrid origin) (c) Bayesian estimates of hybrid index and the 95% credible interval for Sicilian fir adults, Greek fir samples and seedlings with an uncertain origin (i.e., seedlings with a putative hybrid origin). Coloured dots represent silver fir (orange), Greek fir (light yellow), Sicilian fir adults (green), and seedlings with an uncertain origin (red). Adult trees and seedlings showing introgression are highlighted. Mother trees are labelled with an ‘M’ and seedlings with a ‘P’. For the seedlings, the maternal origin is indicated to the left of the dot, followed by the seedling number. 1365294x, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/mec.17605 by Readcube (Labtiva Inc.), Wiley Online Library on [28/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 7 of 12 3.4 | Gene Diversity and Population Structure in Seedlings From the Forest Nursery We were able to genetically characterise 1776 of the 2064 selected seedlings from the nursery. Paternity tests confidently assigned 1525 seedlings (85.8%) to one or more parents of the Sicilian fir. We identified 897 purebred seedlings from this species with a very high autogamy rate (97.9%), only 19 of them being the result of outcrossing (involving 10 out of the 30 adulttrees in nine parental combinations). Finally, the effective population size of the seedlings in the nursery was very low (Ne = 12; 95% confidence interval 6–26) and its level of inbreeding was moderately high (FIS = 0.354). Additionally, we identified 879 seedlings as putative hybrids, including 251 seedlings that could not be genetically characterised with confidence, 345 seedlings where only one Sicilian fir parent was identified, and 283 seedlings with unknown parents. Estimation of the hybrid index confirmed that only 32 individuals (1.8% of the analysed seedlings) were putatively admixed hybrids with Sicilian and Greek fir, showing hybrid indices that ranged from 0.29 to 0.83 (FigureS4). 3.5 | Simulating the Effects of the Assisted Gene Flow RIT values of 12 pairs of adult trees selected to simulate an artificial AGF population ranged from −0.384 to −0.284. The simulated AGF population from the selected crosses showed higher Ho (Kruskal–Wallis's test, χ2 = 91.882, df = 3, n = 400, p < 0.001, ε2 = 0.230) and lower FIS (χ2 = 179.82, df = 3, n = 400, p < 0.001, ε2 = 0.451) than the natural population (i.e., adults and seedlings) and the AGF nextgeneration population (i.e., seedlings from the AGF population, natural seedlings and outcrossed seedlings from the nursery). This latter population maintained similar values as current adults in the population whereas natural seedlings showed significatively the lowest Ho and the highest FIS (Figure3). The nucleotide diversity was similar for all the groupings (χ2 = 1.984, df = 3, n = 400, p > 0.05, ε2 = 0.004). Natural seedlings exhibited the lowest Ne (95% CI 2.6–2.9) while adults showed the highest one (18–25.2). AGF seedlings and the AGF nextgeneration population displayed intermediate Ne values (Figure3). 4 | Discussion 4.1 | Differential Genetic Diversity Across Age Cohorts of the Sicilian Fir The genomewide levels of genetic diversity among adult individuals in the Sicilian fir population were comparable to those of the analysed silver and Greek fir populations. The significant genetic variation observed in adult Sicilian fir samples, highlighted in prior studies (Vicario etal.1995; Ducci, Proietti, and Favre1999; Conte etal.2004), correlated with the recent demographic dynamics of the Sicilian fir, as evidenced by the increased private alleles, slower heterozygosity response, and stable inbreeding coefficients (Llorens, Ayre, and Whelan2018). Palaeobotanical and humanistic studies revealed the occurrence of firs along Sicily from Paleogene to Holocene (Tinner etal.2016), along with a significant demographic decline since the Middle Ages (Pasta et al. 2020). Anthropogenic pressures nearly decimated the Sicilian fir forests, driving the species to the verge of extinction, with only a few surviving individuals comprising the current population (Venturella, Mazzola, and Raimondo 1997). The extended lifespan of firs (> 150 years) and adaptive selection against homozygous individuals could counteract the effects of demographic decline on genetic diversity (Yao etal.2007; Fraser etal.2014; Su etal.2018). Despite the high genetic diversity observed in adult trees, genetic monitoring of seedlings in the natural population using a SNP- array revealed a concerning scenario for species conservation. The seedlings exhibited a substantial decrease in heterozygosity and increased inbreeding coefficients because of exceptionally high selfing rates (> 94%). Only six outcrossed seedlings were found, which were traced back to just five parental genotypes. This contrasts sharply with the typically high levels of outcrossing observed in Mediterranean firs (outcrossing rate tm = 0.72–0.94; Restoux etal.2008) and other gymnosperms such as spruce and subalpine fir (tm = 0.93 & 0.89, respectively; Shea 1987). The high selfing rate in Sicilian fir is likely influenced by several factors such as sparse tree dispersion and limited pollen dispersal, as occurs in other endangered Mediterranean firs (Arista and Talavera1994; Sánchez- Robles etal.2014). The high selfing rate and subsequent decline in genetic diversity across young cohorts of the Sicilian fir likely account for the previously reported reduced seed and seedling viability (Conte etal.2004; Scialabba2019; Mirabile etal.2023). Indeed, the Sicilian fir population exhibited some inbreeding as indicated by significant positive inbreeding coefficients. This diminished genetic diversity may also increase the species' vulnerability to emerging environmental challenges, particularly those associated with climate change (Frankham2015; Ralls etal.2020). 4.2 | Remnants of Recent Hybridization in Sicilian Fir Population and Implications in the Conservation It is remarkable that, despite a high level of selfing, at least seven seedlings in the Sicilian fir population were likely derived from hybridization with an alien fir species. Previous conservation efforts identified more than 5000 alien fir trees in reforested areas located within a few hundred metres of the Sicilian fir population (Ducci 2014). Although many of these nonnative firs were removed, some still persist, presenting a real risk of genetic exchange, particularly the Greek fir, which has been cultivated extensively across Sicily since 1920 (Badalamenti etal.2020). Hybridization is a widespread phenomenon present in the evolutionary history of Mediterranean firs (Krajmerová etal.2016; Balao etal.2020). While hybridisation can potentially harm endangered species (Balao etal.2015), understanding the consequences of hybridization requires detailed study (Arnold2015; Gompert and Buerkle2016). Notably, two adult Sicilian firs exhibited indications of coancestry with Greek fir (and another one was assigned to this alien species), suggesting historical introgression of anthropogenic origin due to plantations that exist for at least a century, as occurred in other tree species (Vanden Broeck etal.2005; Meirmans etal.2014; Scotti- Saintagne etal.2023). Considering that such introgressed trees could potentially enhance the adaptability towards prevailing ecological stress, as observed in other interspecific fir hybrids (Kormutak etal.2013; Stejskal, Horák, and Typta2016), it may be prudent to prioritise their conservation as part of the 1365294x, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/mec.17605 by Readcube (Labtiva Inc.), Wiley Online Library on [28/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 8 of 12 Molecular Ecology, 2024 Sicilian fir recovery program (Jackiw, Mandil, and Hager2015; VonHoldt etal.2022; Brauer etal.2023). However, intentional hybridization (or the preservation of natural hybrids) presents a conservation dilemma. Although this approach is generally criticised because of the risk of outbreeding depression and genetic swamping, it seems appropriate in instances where a species, such as the Sicilian fir, has lost substantial genetic diversity due to genetic drift and exhibits clear signs of inbreeding depression, such as reduced seed viability (Allendorf etal.2001). 4.3 | Monitoring Ex Situ Regenerated Seedlings With the SNP- Array Confirms Low Hybridization Prevalence and High Selfing Similarly, SNP- array analyses revealed a hybridization prevalence of less than 5% in nurserygrown seedlings, mirroring the natural population. Despite its low level, the presence of hybrids in the nursery could potentially compromise the effectiveness of reforestation efforts by promoting genetic swamping (Maunder etal. 2004). However, it seems worthwhile to delve deeper into investigating the fitness and adaptability of these hybrids in their natural habitat (Hirashiki, Kareiva, and Marvier2021). In this context, approximately 98% of the purebred Sicilian fir seedlings in the forest nursery originated from selfpollination, mirroring the selfing rates observed in the wild. While the effective population size (Ne) in the nursery population, established from cones collected from adult trees, slightly surpassed that of the wild population, this discrepancy likely stemmed from the inclusion of hybrid seedlings. However, the strikingly low ratio of effective population size to census size (Ne/N = 0.0058) highlights an alarming genetic bottleneck. This compromises the genetic robustness of the forest restoration stock, akin to endangered conifers such as the Boise araucaria or the Yellow Box (Kettle etal.2008; Broadhurst2013). 4.4 | Implementing Within- Population AGF as the Last Chance to Avoid Extinction In a similar vein, the Sicilian fir population exhibited a notably low effective population size (Ne = 3–21), falling significantly below the recommended threshold (> 100), which poses considerable challenges for its shortterm conservation efforts (Frankham, Bradshaw, and Brook2014). The high selfing rate, coupled with the low mutation rate in firs (De La Torre etal.2017), indicates that relying solely on spontaneous mutations will not sufficiently enhance the population genetic diversity in the short- or mediumterm, emphasising the urgent need for proactive genetic management strategies (Ralls etal.2018). Consequently, genetic rescue, through upcoming outcrossing events with FIGURE 3 | Genetic population parameters based in 120 SNP- array for the adult Sicilian firs (red) and seedlings (orange) from the natural population, the simulated AGF population (green), and the AGF- next- generation population (blue). Distribution (thick black line, median; white box, interquartile range; thin black line, 1.5 × interquartile range) of homozygosity (Ho), inbreeding (FIS), nucleotide diversity (π), and effective size population (Ne; dot, estimate; thin black line, 1.5 × 95% CI). 1365294x, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/mec.17605 by Readcube (Labtiva Inc.), Wiley Online Library on [28/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 9 of 12 genetically diverse individuals, stands as the final opportunity to avert extinction by mitigating inbreeding and reducing levels of homozygosity (Ingvarsson2001; Fady etal.2020). Given the single remaining wild Sicilian fir population, we propose a withinpopulation AGF approach, selecting crosses among trees with more distant coancestry in the population. Note that this represents an extreme case of AGF, as this method typically involves translocating unrelated individuals or gametes from an external source population within species ranges (Aitken and Whitlock2013). The withinpopulation AGF method appeared to be potentially effective in enhancing Sicilian fir genetic diversity and Ne and reducing inbreeding. This genomicguided conservation method is scheduled for implementation in practical Sicilian fir conservation practices. Additionally, it could be practically implemented in other endangered species facing similar conservation challenges (i.e., with small populations sizes and high selfing rates) to increase their genetic diversity and effective population sizes, simultaneously mitigating potential negative effects associated with traditional AGF, such as outbreeding and the introduction of maladaptive alleles (Aitken and Whitlock2013; Grummer etal.2022). However, the gain in genetic diversity through AGF appears insufficient to recover the past levels of diversity in the Sicilian fir, and the longterm effectiveness of this conservation action to avoid the extinction vortex remains uncertain. Hence, there is a necessity to investigate inbreeding depression in Sicilian fir (Pečnerová etal. 2023), along with exploring the potential of hybrids to alleviate genetic load and facilitate future adaptation (VonHoldt etal. 2018). A combined strategy (i.e., withinpopulation AGF and controlled hybridisation) could be optimal for enhancing species survival, effectively mitigating the drawbacks associated with each individual approach. 5 | Conclusions Our study revealed that adult Sicilian firs harbour high genetic diversity despite their small population size, comparable even to that observed in other Mediterranean firs spread across large populations. Despite the high genetic richness found in adult trees, the genetic variability of the population will dramatically decrease in the upcoming generations, even if humanmediated actions are taken to increase, or at least maintain, the genetic diversity of the population. Intrapopulation AGF can contribute to preventing the loss of genetic variation in the population. However, this does not seem to be sufficient to break free from the extinction spiral because of the predictable species' vulnerability in upcoming generations to emerging environmental challenges, such as climate change. Interspecific crosses with other intersectional fir species, namely the silver and Greek fir, could represent a potential alternative strategy for the future conservation of the Sicilian fir that merits exploration. Our results emphasise the relevance of genomicguided conservation actions, which can assist in identifying suitable individuals for reforestation and provide a solid foundation for conservation management. We expect that this approach could serve as a model to be replicated in the conservation strategies for other endangered firs, or even for other conifers, around the Mediterranean region. Author Contributions M.A., A.T., P.L.O. and F.B. conceived the ideas and designed the methodology; J.C.d.V., M.A., A.T., P.L.O. and F.B. collected the samples; J.C.V., C.B.B. and F.J.J.L. generated the genomic data; J.C.d.V., C.B.B., F.J.J.L. and F.B. analysed the data; J.C.d.V. and F.B. led the writing of the manuscript. All authors contributed critically to the drafts and gave final approval for publication. Acknowledgements The authors acknowledge the Madonie Natural Park and the forest nursery of Piano Noce (Polizzi Generosa) for hosting the project area and maintaining A. nebrodensis seedlings. We also thank University of Seville Biology and Herbarium General Research Services (CITIUS) for providing the facilities. We particularly thank the anonymous referees and the editor for their helpful comments. This research was funded by the European Union through the LIFE4FIR project, grant number LIFE18/NAT/IT/000164. Conflicts of Interest The authors declare no conflicts of interest. Data Availability Statement All raw data generated for this project is stored in the NCBI Short Read Archive (SRA), under accession number PRJNA563575. The dataset for Open Array Loci is available at idUS (https:// doi. org/ 10. 12795/ 11441/ 133288). We have hosted the scripts used in this project in a public repository (https:// github. com/ fbalao/ nebrodi) and archived them at Zenodo (DOI: 10.5281/zenodo.11622037). Benefit- Sharing Section Our study is an integral component of a multidisciplinary EULIFE project dedicated to the conservation of Abies nebrodensis. It brings together scientists from various institutions across the country where the research is being carried out. However, there is a noticeable absence of direct involvement from local scientists in the study area, a responsibility that has been allocated to our group. The findings of our research, aimed at enhancing the conservation strategies for the Sicilian fir, have been effectively shared with local policymakers, including DRSRT and Madonie Natural Park, as well as the broader scientific community. References Aitken, S. N., J. B. Bemmels, and C. N. Sally Aitken. 2015. “Time to Get Moving: Assisted Gene Flow of Forest Trees.” Evolutionary Applications 9: 271–290. Aitken, S. N., and M. C. Whitlock. 2013. “Assisted Gene Flow to Facilitate Local Adaptation to Climate Change.” Annual Review of Ecology, Evolution, and Systematics 44: 367–388. Allendorf, F. W., R. F. Leary, P. Spruell, and J. K. Wenburg. 2001. “The Problems With Hybrids: Setting Conservation Guidelines.” Trends in Ecology & Evolution 16, no. 11: 613–622. Arista, M., and S. Talavera. 1994. “Pollen Dispersal Capacity and Pollen Viability of Abies pinsapo Boiss.” Silvae Genetica 43: 155–158. Arnold, M. L. 2015. Divergence With Genetic Exchange. Page Divergence with Genetic Exchange: Oxford University Press, Oxford, UK. Badalamenti, E., S. Sferlazza, D. S. La Mela Veca, F. Maetzke, G. Sala, and T. La Mantia. 2020. “Which Are Southern Italy's Fastest Growing Tree Species? Lessons From the Past for Future Perspectives, With a Special Focus on Sicily.” Annals of Silvicultural Research 45, no. 1: 31–43. 1365294x, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/mec.17605 by Readcube (Labtiva Inc.), Wiley Online Library on [28/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License