Genetic structure and diversity of the endangered fir tree of Lebanon (Abies cilicica Carr.): implications for conservation

Abstract

The threatened conifer Abies cilicica currently persists in Lebanon in geographically isolated forest patches. The impact of demographic and evolutionary processes on population genetic diversity and structure were assessed using 10 nuclear microsatellite loci. All remnant 15 local populations revealed a low genetic variation but a high recent effective population size. FST -based measures of population genetic differentiation revealed a low spatial genetic structure, but Bayesian analysis of population structure identified a significant Northeast-Southwest population structure. Populations showed significant but weak isolation-by-distance, indicating non-equilibrium conditions between dispersal and genetic drift. Bayesian assignment tests detected an asymmetric Northeast-Southwest migration involving some long-distance dispersal events. We suggest that the persistence and Northeast-Southwest geographic structure of Abies cilicica in Lebanon is the result of at least two demographic processes during its recent evolutionary history: (1) recent migration to currently marginal populations and (2) local persistence through altitudinal shifts along a mountainous topography. These results might help us better understand the mechanisms involved in the species response to expected climate change.

Figure 1. Range of distribution of Abies cilicica in the Eastern Mediterranean.

Remnant populations of Abies cilicica in Turkey, Syria and Lebanon are plotted in red onto the map of the Eastern Mediterranean. Remnant populations of Abies cilicica in Lebanon are plotted in green onto the map of Lebanon where they are currently found on the western slopes of the northern part of Mount Lebanon. Geographical position of Abies cilicica in Turkey and Syria were plotted from a euforgen map while the geographical position of Abies cilicica in Lebanon corresponds to our sampling data. Map was prepared using qgis version 1.8.0 and the gmt package within r .

Figure 2. Genetic variation of Abies cilicica and other Mediterranean Abies species.

Pie charts are used to depict the gene diversity corrected for sample size , and the rarefied allelic richness , , averaged over 10 nuclear microsatellite loci, for A. cilicica in Lebanon (CILI), A. bornmuelleriana in Turkey (BORN), A. cephalonica in Greece (CEPH), A. alba in the French Southern Alps (ALBA1), A. alba in the French Pyrenees (ALBA2), and A. marocana in Morocco (MARO). Diameter of pie charts is relative to the correspondent rarefied allelic richness of the species. Proportion of red or blue color within pie charts is relative to the correspondent gene diversity of the species. Blue and red colors are employed to separate the studied fir populations into a group with significant lower genetic variation (red pie charts), and a group with significant higher genetic variation (blue pie charts) based on 10000 permutations of populations among groups performed in fstat version 2.9.3.2 .

Figure 3. Asymmetric Northeast-Southwest migration of Abies cilicica in Lebanon.

The Northeastern and Southwestern ridges correspond to the two genetically-distinct demes assigned by baps . Populations of the Northeastern ridge are plotted in green color while those of the Southwestern ridge are plotted in magenta color. A likelihood-based Bayesian assignment test performed in geneclass2 allowed the detection, in the Southwestern ridge, of 3 F₀ migrants originating from the Northeastern ridge. It has also allowed the detection, in the Northeastern ridge, of 1 F₀ migrant originating from the Southwestern ridge, suggesting an asymmetric Northeast-Southwest migration.

Figure 4. Isolation-by-distance and non-equilibrium conditions between dispersal and genetic drift.

The logarithm of genetic similarity is plotted against the logarithm of geographic distance among population pairs . Estimations of slope, intercept and R² of the relationships were calculated using Reduced Major Axis (RMA) regression . The RMA regression explained only 5% of the variance in the whole distribution area of Abies cilicica in Lebanon. Analysis was performed in ibdws .

Figure 5. Observed and simulated values of the modified Garza-Williamson index for Abies cilicica populations.

This figure shows the different observed values of the M-ratios calculated for the microsatellite loci and the values of the simulated critical M threshold (M _c) below which a bottleneck is evident. The observed M-ratio values are displayed in black solid line. The M _c values calculated for θ = 10, θ = 1, θ = 0.5, and θ = 0.1 are respectively displayed in red, blue, green and orange solid lines. All the observed M-ratio values were significantly higher than the simulated M _c values for different pre-bottleneck θ values. Values of the M-ratio and the M _c parameters were obtained using m-p-val and critical_m .

Figure 6. Allele frequency distribution over different allele frequency classes for each population of Abies cilicica.

This figure shows the distribution of allele frequencies across the 15 local populations. The values were obtained using the mode-shift test . The distribution of allele frequencies across the 15 local populations is plotted using different colors. All populations showed a higher abundance of rare alleles (frequency<0.1) than alleles at intermediate frequencies (0.1–0.2).