The overarching goal of this study was to compare and contrast patterns of nucleotide diversity and tests of neutrality in four conifer species of the alpine mountain region of central Europe. These results give an overview of nucleotide diversity in four coniferous species and provide a useful SNP resource that can be applied in landscape genomic studies. In spite of the result that there were no strong and consistent differences in the proportion of outliers detected between candidate and control genes, there were several putative outlier genes that may be related to environmental adaptations. A unique aspect of this study is the comparison of diversity and departure from neutrality among four tree species living in montane ecosystems.
Pinus cembra showed lower diversity than the other tree species
Pinus cembra showed the lowest diversity among the four species, with values falling in the range of species belonging to the subgenus Strobus, such as P. chiapensis (Syring et al. 2007) and P. albicaulis (A. J. Eckert, A. D. Bower, K. D. Jermstad, J. L. Wegrzyn, B. J. Knaus, J. V. Syring and D. B. Neale, unpublished manuscript) and somewhat less than that found in other coniferous species (González-Martínez et al. 2006; Savolainen and Pyhäjärvi 2007). To the contrary, P. mugo showed the highest nucleotide diversity among the four species, with values similar to other species belonging to the subgenus Pinus (Grivet et al. 2009, 2011; Ma et al. 2006; Shiraishi and Shiraishi 2011; Wachowiak et al. 2009; Eveno et al. 2008).
The contrasting patterns of nucleotide diversity of the two pines, growing in similar altitudinal ranges (1200–2300 m for P. cembra and 1000–2000 m for P. mugo), call for an interpretation. This result may be linked to the different demographic histories of the two pines, because P. cembra is characterized by two distinct postglacial refugia in the Carpathians and in the Alps (Höhn et al. 2009), whereas in the Pliocene, the large range of P. mugo was separated into several refugia that are poorly known (Sandoz 1983Heuertz et al. 2010). Moreover, P. cembra, like P. albicaulis Engelm., has bird-dispersed seeds (Tomback 2005), which may lead to higher levels of inbreeding (Rogers et al. 1999). Low genetic diversity within P. cembra populations in the northern Alps may be due to genetic drift by restricted gene flow (Gugerli et al. 2009). The nonpine species of this study, A. alba and L. decidua, had fairly high estimates of nucleotide diversity compared to two other pines, P. sylvestris (Pyhäjärvi et al. 2007; Palmé et al. 2008) and P. luchuensis (Shiraishi and Shiraishi 2011).
Pinus cembra shows proportionally more outlier loci
The compound DHEW test detected the presence of outlier loci in only P. cembra and P. mugo, although the single-locus tests revealed the possible presence of outlier loci in A. alba and L. decidua as well. These results suggest that selection may have acted more in the two pines than in the other two species, although this interpretation could be confounded by the fact that proportionally more highly conserved genes were tested in A. alba and L. decidua or that P. cembra had the lowest average diversity.
The presence of several genes, especially in P. mugo, showing higher nucleotide diversity at synonymous sites compared to the other site categories is an indication of purifying selection (e.g., Palmé et al. 2009), in accordance with the expectation that in coding regions, most mutations are probably disadvantageous. Moreover, divergence at synonymous sites was three times the value at nonsynonymous sites and up to eight genes per species displayed Ka/Ks ratios greater than one, which may indicate the presence of positive selection (Palmé et al. 2008). In the candidate genes, the negative average estimate of Tajima’s D found in A. alba and in the two pines may indicate the presence of recent demographic events, such as population size expansion or purifying selection or selective sweeps. Moreover, the positive value of Tajima’s D in L. decidua may indicate a decrease in population size and/or balancing selection.
Several loci deviating from neutrality were found in both control and candidate gene sets in P. cembra. Among candidate outlier loci, locus 2_6731 is the most interesting as its homologue encodes for the E3 ubiquitin complex protein, an F-box protein that is involved in GA signaling. Ubiquitination controls most of the hormonal responses in plants and is one of the dominant plant regulatory mechanisms (reviewed in Dreher and Callis 2007; Santner and Estelle 2009). Plant DNA viruses (Geminiviruses) may interfere with several responses regulated by the ubiquitin E3 ligases, making the plant more susceptible to virus infection (Ascencio-Ibáñez et al. 2008; Lozano-Durán et al. 2011). Moreover, GA modulates plant growth and development throughout the whole lifecycle of the plant (Sun 2010). Additionally, two outliers encoded for proteins related with membrane transporters (loci 0_2775, CL1659Contig1). In particular, the Arabidopsis homologue of locus 0_2775 is involved in cellular uptake of inorganic phosphate in the root xylem (Wang et al. 2004). In the same species, one outlier (locus CL1661Contig1) encodes for acetyl-CoA carboxylase, the enzyme that catalyzes the first committed step in fatty acid synthesis (Konishi et al. 1996; Li-Beisson et al. 2010). Acyl lipids constitute the membrane between cell and organelles. These genes may be important for tree fitness, because organelle proteins change in abundance during stress, as an immediate response to abiotic stress (Taylor et al. 2009).
Several outliers were also found in both control and candidate gene sets in P. mugo. Among the candidate gene outliers, locus 0_13913 encodes for a member of EXO70 family protein, which is involved in exocytosis. One member of this gene family (AtEXO70A1) was found to be crucial for polar growth and plant development (Synek et al. 2006). Locus 2_8627 has a catalytic activity for the carbon–sulfur lyase that is involved in glucosinolate biosynthesis (Mikkelsen et al. 2004). Glucosinolates are amino acid-derived natural plant products in Arabidopsis, implicated in plant defense (Halkier and Gershenzon 2006). Locus 2_8852 encodes for a galactokinase that is involved in the synthesis of d-galacturonic acid (d-GalA) polysaccharides (Yang et al. 2009).
To compare our results with other coniferous species and between species in this study, it is important to consider our results in the context of possible biases owing to (i) sequence conservation that determined the number of loci successfully resequenced, (ii) number of trees sampled, (iii) species range covered by the sampling and (iv) the effect of demography that differs from the SNM assumptions. The much lower success in the resequencing in A. alba and L. decidua than in the two pines was a direct effect of sequence conservation with P. taeda from which primers were designed. The unbalanced number of sequences per tree may have affected the estimates of nucleotide diversity, especially in L. decidua, which had the lowest average number of sequences per gene (n = 3 ± 1). In the same species, the low number of reads may have biased outlier detection; nevertheless, the sample number was used in the estimation of the neutrality tests.
Furthermore, the small number of trees sampled, the partial coverage of species ranges, and a nonuniform sample distribution according to species demographic history could all affect estimates of diversity (Städler et al. 2009). Abies alba was sampled mainly in the central-west of Europe; this might bias the estimation of nucleotide diversity, because there is a clear separation into two maternal lineages in A. alba (Liepelt et al. 2009). In P. cembra, more trees were sampled in the Alps than in the Carpathian Mountains. These two areas belonged to two different lineages, with the Carpathian populations being more polymorphic than the populations in the Alps (Belokon et al. 2005; Höhn et al. 2009). For P. mugo, the sampling covered the species range, including the area in which the different varieties (P. mugo s.s. and P. uncinata) overlap. Moreover, for the identification of the outlier loci, the estimation of the P-value in the neutrality tests did not take into account species demography. It should be noted, however, that the compound test is fairly robust to demographic deviations from the SNM (Zeng et al. 2006, 2007). The bias in sequence conservation may also have affected the identification of the outliers from SNM in the studied species, because the percentage of outliers per gene set per species ranged from 4.17% to 1.58% in the two pines, while no outliers were identified in A. alba and L. decidua.