Population analysis

The visualisation of the genetic relationships amongst the individuals sampled is presented in the PCO analysis and the dendrogram generated by cluster analysis (Figure 6.5 and Figure 6.6).

Figure 6.5. Principal coordinate analysis for 3 populations of Gomortega keule using microsatellites for 9 loci.

The groups Bosques Arauco (BA, n=75) and Forestal Tierra Chilena (FTC, n=63) are the southernmost populations of the species, while the group North (populations Reserva Nacional Los Queules and Ralbún, n=14) is in the northern area of its natural distribution. The graph was generated using MVSP with the first 2 axis of eigenvalues, which accounted for 32.5% of the variance (Axis 1 19.7%, and Axis2 12.7%).

Figure 6.6. Dendrogram generated by cluster analysis for populations of Gomortega keule.

Populations BA (Bosques Arauco) and FTC (Forestal Tierra Chilena) are the southernmost of the species, while population North (Reserva Nacional Los Queules and Ralbún) is in the northern area of its natural distribution. The dendrogram was generated by UPGMA using the index of Nei and Li (1979) with the software MVSP.

The most frequent alleles across loci had a frequency ranging from 0.48 to 0.84 for the overall populations calculated with Powermarker. Hence, they were all polymorphic (frequency

<0.95) according to Laurentin (2009). The majority of private alleles were from the population North (Table 6.4), with intra-population frequencies from 0.04 to 0.86, while the populations in the South had frequencies always <0.03 (except one allele with 0.06), which meant that private alleles were often rare in the South.

Table 6.4. Number of alleles and private alleles per locus and population.

Number of alleles Number of private alleles Locus

The largest difference between HO and HE (>0.2) was present in the locus Gk-31 for BA and locus Gk-35 for North (Table 6.5). The inbreeding coefficient for each locus and population showed a range of values from -0.38 to 0.86 (Table 6.6), while the highest values (F_IS>0.8) were observed for locus Gk-31 in BA and locus Gk-35 in North. Fixation indices (ρST and FST) for each pair of populations are shown in Table 6.7. Exact test for HWE was significant (P<0.05) for locus Gk-31 in populations BA and FTC, Gk-35 in BA and North, and CS8 in North. Linkage disequilibrium was present in all loci although for different pairs, ranging from a single linked pair to up to 4 linked pairs within a single locus for a given population (Appendix 9.24). Two pairs of loci (Gk-6 locus 1 / Gk-6 locus 2 and Gk-39 / Gk-30) showed significant linkage disequilibrium in the 3 populations. Estimated frequencies of null alleles are presented in Table 6.8.

Table 6.5. Observed (HO) and expected heterozygosity (HE) for each locus and population.

Table 6.6. Inbreeding coefficient (FIS) for each locus and population.

Locus BA FTC North All

FIS Indicates deficit (>0) or excess (<0) of heterozygotes.

Table 6.7. Fixation index for population pairwise comparision.

Population pairwise ρST FST

BA - FTC 0.179 0.177

BA - North 0.852 0.790

FTC - North 0.868 0.530

ρST was caulated using Genepop (procedure by Michalakis and

Table 6.8. Estimation of null allele frequencies for each locus and population.

Null allele frequencies were estimated using the population inbreeding model (PIM) and the individual inbreeding model (IIM) with INEst (Chybicki and Burczyk 2009). Frequencies significantly different from zero are indicated by asterisks (*, P< 0.01 and **, P<0.001).

6.4 Discussion

The yield of DNA from different samples seemed to be highly influenced by the type of tissue (actively growing, old and oxidised, dead), amount of tissue and grinding procedure.

Although without an extensive evaluation, the method by Lander et al. (2007) gave the most consistent results, especially for dried samples.

The fact that samples from dried tissues amplified unreliably for microsatellites 5 and Gk-44, could be attributed to the lower concentration and some degree of degradation of the DNA in dried tissues, since samples from embryos and in vitro material, which in general had good quality (undegraded DNA) and a higher quantity, amplified well. For future studies, this situation may be improved with the optimization of the MgCl2 concentration in the reaction mixture and quickly dehydrating samples during field collection.

The amplification of 2 loci by the primer pair Gk-6 was not found by Lander et al. (2007) in the 20 individuals that they analysed. However, Baldoni et al (2009) reported that 6 of 37 microsatellites they assessed in olive amplified for 2 or more loci. Ying et al. (2009) also mentioned the occurrence of a primer pair which amplified for 3 loci in avocado.

In the population BA, the locus Gk-31 presented an estimated frequency of null allele significantly higher than zero (Table 6.8). For population FTC, the values were much smaller, and within its range, the highest also occured in the locus Gk-31, although with no significance. For population North, the frequency of null allele was only significant for locus Gk-35. The presence of null alleles can be considered of low importance and, although it can have some influence on the accuracy of the statistical analyses, it was not regarded as invalid.

6.4.1 Clonal genetic fidelity

No genetic difference was detected with microsatellites for each of the clones studied, suggesting that in G. keule there is no genetic variation or mutation, detectable by microsatellites, due to culture involving somatic embryogenesis. Low levels of somaclonal variation are unlikely to be detected by microsatellites, making this class of markers suitable for tracing clonal lines and quality control.

Similar findings have been reported using microsatellites to assess genetic variation in plant material derived from embryogenic cultures, where no genetic variation was found in trees including Olea spp. (Lopes et al. 2009), Picea abies (Harvengt et al. 2001) or Quercus suber (Santos et al. 2007). In Betula pendula (Ryynänen and Aronen 2005), Quercus robur (Valladares et al. 2006) and in the herbs Swertia spp. (Chaudhuri et al. 2007) and Chlorophytum arundinaceum (Lattoo et al. 2006), genetic variation was not detected by RAPD. Genetic variation was not present in Swertia spp. using ISSR (Joshi and Dhawan 2007).

In contrast, Marum et al. (2009) observed around 10% genetic variation using microsatellites in plants regenerated from somatic embryos of Pinus pinaster. Similar or lower levels of genetic polymorphism were reported with the use of ISSR in Robinia ambigua (Guo et al.

2006) and Platanus acerifolia (Huang et al. 2009). With the use of RAPD in Cedrus spp.

(Renau-Morata et al. 2005), shoots cultured for 1 year showed low genetic variation, while the use of variable number tandemly repeats (VNTR) in Hypericum perforatum (Urbanová et al.

2006) detected minor changes after cryostorage, but not for cultured material not subjected to cryopreservation.