Spatial genetic structure in fourPinusspecies in the Sierra Madre Occidental, Durango, Mexico

Abstract

In this study, we examined the spatial genetic structure (SGS) in extensively managed, but naturally regenerated forest stands of Pinus cembroides Zucc., Pinus discolor Bailey et Hawksworth, Pinus durangensis Martínez, and Pinus teocote Schiede ex Schltdl. & Cham. at local (within the stands) and large (among the stands) scales using amplified fragment length polymorphisms (AFLP), with respect to conservation and sustainable management of genetic resources of these species. Because these four pine species grow in different landscape structures, we expected to find differences in their SGS, although all of them are widely spread, wind pollinated, and often occur at high population densities. At the local scale, there was no evidence of significant SGS in the four species under study (except in 1 out of 18 seed stands), suggesting that the genetic variants of these species are almost always randomly distributed in space, probably due to high wind pollination and seed dispersal. At a larger scale, the significant SGS found may be the result of isolation by distance among populations. We recommend (i) establishing a tight network of seed stands, with a maximum distance of 3–11 km between seed stands, to prevent greater loss of local genetic structure, and (ii) using these seeds to establish reforestations within a maximal radius of 3–5 km from seed provenances.

Spatial Genetic Structure within and among Seed Stands of Pinus engelmannii Carr. and Pinus leiophylla Schiede ex Schltdl. & Cham, in Durango, Mexico

“…leiophylla) at the local level ( Table 2 and Table S1). This matches several previous reports describing that no SGS could be observed in North American stands of conifer species such as Picea chihuahuana, Pseudotsuga menziesii, Pinus strobiformis [13], P. clausa [11], P. strobus [12], P. lumholtzii [6], P. cembroides, P. discolor, P. durangensis, and P. teocote [14]. According to Parker et al [11] and Vekemans and Hardy [1], the main reason for the lack of significant SGS observed in these two pine species at the local level was probably the wind-mediated dispersal mating system, which leads to a strong omnidirectional gene flow.…”

“…However, there are other causes of random spatial distribution of genetic variants: (i) seed shadows of neighboring adult trees sometimes overlap; (ii) germination of seeds from numerous neighboring adult trees within relatively small stand patches; (iii) demographic mortality, which may also reduce the genetic relatedness at the seedling and sapling stages [6,11,[49][50][51]; (iv) the disguising of SGS by high densities of adult trees; and (v) alteration of the tree stands by human intervention [14,50,52,53].…”

“…The stands were situated in natural populations mixed with other pine species (P. teocote, P. arizonica, and P. cooperi) as well as oak (Quercus sideroxyla), Juniper (Juniperus flacciada, J. deppeana), and Arbutus spp. For analysis of the SGS, samples of needles were randomly collected from 23-35 trees (from 130 trees) in each seed stand in 2013 (a total of 238 needle samples from PE and 298 from PL) (see more in Hernández-Velasco et al [14]; Table 1). …”

“…Tanimoto distance (TD) [34], Genetic Distance of Huff (GD) [35], and pairwise kinship coefficients (F ij ) [9] were all considered because different mathematical approaches can produce contrasting conclusions (see details below). The distances were computed using Spatial Genetic Software v1.0 (SILVOLAB Guyane INRA Station de Recherches Forestières, Campus agronomique, Kourou cedex, French Guiana) [5], SPAGeDi v1.4 (Laboratoire de Gènètique et Ecologie vegetales, Universitè Libre de Bruxelles, Bruxelles, Belgium) [1], and GenAlEx v6.501 (Evolution, Ecology and Genetics, Research School of Biology, The Australian National University, Canberra, Australia and Department of Ecology, Evolution and Natural Resources, School of Environmental and Biological Sciences, Rutgers University, New Brunswick, NJ, USA) [14,36].…”

“…The corrected critical p value (significance level α* = 0.00036) was determined by dividing the original critical p value (0.025) by the number of comparisons or hypotheses (m = 69). Consequently, the Bonferroni critical value of P was 1 − α* = 0.9997 [14].…”