STRUCTURE OF P MACULATA POPULATIONS AS REVEALED BY MICROSATELLITE DATA
F- stat Source of variation Nested in %var SS var F value P value
4.3.9 Demographic analysis
Population expansion for the five main populations was tested with two tests: k and g
tests. Alleles are expected to be distributed into a few distinct groups in constant populations, whereas distinct allele groups are not expected in expanding populations Figure 4.6 Test of isolation by distance, based on the microsatellite data. Relationship between the pairwise genetic distances (FST) and geographical distances (km).
(Reich et al., 1999).The kg-tests are based on these assumptions (Luikart and Cornuet, 1998). The k-test is a within-locus test, and it assumes that the pattern of distribution of allele-lengths follows a bimodal shape in constant populations, while this distribution is uni-modal and peaked in expanding populations. The k statistic, which measures the peakedness of the distribution, tends to acquire negative values in expanding
populations (Reich et al., 1999). The proportion of loci that provides negative k values is evaluated with a binomial distribution (Bilgin, 2007). For the P. maculata dataset, k- tests suggested negative values at the majority of loci ranging from eight to ten out of the twelve loci when the populations were pooled into a single population. This can be a sign of an expansion. However, the test returned significant results for only the TP population (P=0.0156) when the analysis was performed for each population separately. No significance was observed for the other populations (P>0.0596). The g-test
compares the observed and expected variance of allele length distribution across loci assuming that the variance of the widths of the distribution will be lower in a recently expanding population. The g value is the ratio of the observed to the expected variation under assumptions of constant population size. The significance of the g value is
determined according to theoretical 5% percentile cut-off values for a given sample size and loci (Reich et al., 1999). The g values calculated in this study do not support a history involving population expansion for the P. maculata data. It is worth noting that both tests, but especially the g-test, have a low power to detect population expansion with a small number of samples and loci (Bilgin, 2007).
In a bottlenecked population, the number of alleles decreases more dramatically than heterozygosity (Cornuet and Luikart, 1996). Consequently, the heterozygosity calculated from allele frequencies (He) becomes higher than the heterozygosity
calculated from allele number (Heq) in a bottlenecked population (Cornuet and Luikart, 1996). In this concept, BOTTLENECK (Piry et al., 1999) was used to test whether a significant number of loci show heterozygosity excess (He>Heq) (Luikart and Cornuet,
1998) in order to investigate the possibility of recent population reduction in each population using three different mutational models: IAM, SMM and the two-phase model (TPM), which is a mixture of IAM and SMM. The Wilcoxon test that was used to evaluate the significance of the heterozygosity excess did not detect recent bottlenecks in any population under the TPM and SMM models (Table 4.7). However, the TR population was found to have experienced recent bottleneck (P=0.0031) under the IAM
model. There is no information available on the mutation model of P. maculata
microsatellites; however, IAM seems to be inappropriate for microsatellite evolution based on empirical data. TPM and SMM, which take size homoplasy into account, are said to provide a more realistic explanation of allele evolution in microsatellites (Bhargava and Fuentes, 2010). Therefore, the bottleneck in TP suggested by IAM may not be realistic. When the distribution of allele frequencies is taken into consideration, nonbottlenecked populations that are under mutation-drift equilibrium are expected to show an L-shaped pattern in which there are high proportions of low frequency alleles. On the other hand, a shift is observed in the mode of allele frequency distribution in bottlenecked populations where intermediate frequencies become more abundant than low frequency alleles (Luikart and Cornuet, 1998). Analysis of mode-shift in the distribution of allele frequencies for the P. maculata dataset with BOTTLENECK suggests that all the populations exhibit a normal L-shaped pattern. All this evidence suggests that none of the populations is likely to have experienced a recent bottleneck. Table 4.7 Summary results of population expansion and decline analysis for the NZ P. maculata populations based on microsatellite data. Expansion and decline were assessed with kg-tests and BOTTLENECK, respectively, using data obtained from the 12
microsatellite loci. Ti Point (TP) Auckland (AKL) Tauranga (TR) Wellington (WL) Nelson (NL)
k-test (# of negative loci) 10 8 9 9 9
k-test (P value) 0.0156 0.1661 0.0596 0.0596 0. 0596 g-values 1.229 1.5642 1.7533 2.0630 3.7239 IAM 0.0756 0.1167 0.0031 0.0881 0.3386 TPM 0.7407 0.7153 0.6890 0.6333 0.9539 SMM 0.9451 0.9933 0.9948 0.0523 0.9994
4.4
DISCUSSION
4.4.1 Genetic diversityIn this study, I investigated the diversity and genetic structure of 146 samples from five main localities representing four regions of the North Island (Ti-Point, Auckland, Tauranga and Wellington) and one region of the South Island (Nelson) using twelve nuclear microsatellite markers. High genetic diversity was observed for all five
populations. All the populations are in HWE. Allelic richness and heterozygosity, which are two important diversity estimators, are high for all five populations, indicating that all the populations contain significant amounts of genetic diversity. According to
population genetics theory, there is positive correlation between effective population size and expected genetic diversity at a neutral locus that is under mutation-drift
equilibrium. This is due to the fact that the effect of genetic drift is less dramatic in large populations compared to small ones (Hartl and Clark, 2007). High diversity in marine invertebrates is attributable to their large population size (Zhan et al., 2009), which may explain the high diversity observed in P. maculata. There are other factors affecting genetic diversity, such as the geographical range of the organism (Frankham, 1996), selection and the mating system (Bazin et al., 2006). Selection affects variation in an increasing or decreasing manner depending on the nature of the selection. However, microsatellites are assumed to be neutral (Oliveira et al., 2006). The possibility that they are hitchhiking with the regions under selection cannot be ruled out. However, the microsatellites used in this study are in linkage equilibrium as revealed by the LD test. All these unlinked loci are unlikely to be under the same type of selection. P. maculata
is an outcrossing simultaneous hermaphrodite (Willan, 1983). This mating type
increases the effective population size, and consequently the genetic diversity (Silva and Russo, 2000). Genetic variation is expected to be high in species with wide
geographical ranges (Frankham, 1996). This effect may also contribute to genetic variation in P. maculata as it has a wide range of geographical distribution in the South- Eastern Pacific (Willan, 1983).