Forests have been globally reduced by unprecedented rates of deforestation and habitat transformation, which can affect the species’ genetic structure. Understanding patterns of spatial genetic structure is fundamental for proposing conservation actions that contribute to their long-term persistence. This study evaluated the spatial genetic structure and the effects of landscape features on gene flow in two conifers, Abies religiosa and Pinus montezumae, with different successional affinities in the temperate forest of La Malinche National Park (LMNP) in central Mexico, which exhibits a complex topography and high land-use changes. We recovered 7,326 single-nucleotide polymorphisms (SNPs) for A. religiosa and 34,423 for P. montezumae across 12 sites in four mountain slopes. Clustering analyses did not reveal genetic structure in both species, and Moran Eigenvector Maps showed that only 3% of the genetic variation was spatially structured in A. religiosa, whereas in P. montezumae the model did not fit the data (R² = -0.01). A weak isolation by distance pattern was found only for A. religiosa (r = 0.12, P < 0.05), and in P. montezumae, 11% of the genetic variation was structured among slopes. Landscape genetic analyses showed that topography, aspect, and land use did not explain genetic differentiation more than a null model of geographic distance in A. religiosa, in contrast to P. montezumae, in which the aspect-sine surface was the top-ranked model, indicating a reduction of gene flow Eastward. Our study contributes information for the conservation of Mexican temperate forests.

Three collection sites were established on each slope of the La Malinche National Park (12 in total). In each population, five 1,000 m² circular plots were randomly located, separated by at least 50 m from edge to edge. Within each plot, vegetative tissue was collected from at least five individuals for each species, separated from one another by at least 30 m. For dehydration and further processing, all samples were stored in resealable bags with silica gel and labeled with the species name, population, cohort (adults, ≥10 m in height; seedlings, < 0.5 m in height; Ramírez-Marcial et al. 2001), and geographic coordinates.

Genomic DNA was extracted with the Plant Pro Kit (Qiagen). DNA was quantified with the PicoGreen fluorometric method using a NanoDrop™ ND 3300 fluorospectrometer (Thermo Scientific). Libraries were prepared in the Institute of Integrative Biology and Systems (IBIS) at the University of Laval by double digestion (ddRADSeq) using the enzymes PstI and MspI. Sequencing was obtained by the Genome Quebec Centre D'expertise et de Services through four lines of Illumina NovaSeq 6000 (PE150).

Genotyping was carried out using the pipeline in Stacks v.2.62 (Catchen et al. 2013) with de novo alignment and applying the following criteria: exclusion of sites containing an indel (remove-indels), a minor allele count (mac) of 1, a maximum missing data per site (max-missing) of 0.2, inclusion of only biallelic sites displaying Hardy Weinberg Equilibrium, and only loci present in at least 90 % of individuals in each population (-r 0.10). Additional filtering was done with VCFtools v. 0.1.15 (Danecek et al., 2011), considering a minor allele frequency (MAF) of 0.01, a minimum mean depth (min-meanDP) of 10×, and a maximum mean depth (max-meanDP) of 196× for A. religiosa and 306× for P. montezumae (twice the mean depth obtained with the site-mean-depth command; Taylor et al, 2020, Yamasaki et a,l2020). Subsequently, we identified outliers through single-nucleotide polymorphism (SNP) scanning, using an alpha = 0.1, with PCAdapt v.4.3.3 (Duforet-Frebourg et al. 2015; Luu et al. 2017) in the R program v.4.2.2 (R Core Team 2022). Loci identified as adaptive were removed to retain only the neutral ones.