Andricus lignicola (Hartig, 1840)

Mutun, Serap, Received, Hülya Karagözoğlu & Online, Published, 2015, Investigation of genetic variation among Turkish populations of Andricus lignicola using mitochondrial cytochrome b gene sequence data, Turkish Journal of Zoology 39 (5), pp. 721-733 : 723-729

3.1. Genetic variation in Andricus lignicola populations Amplification results of 117 individuals yielded 18 haplotypes with no indels, nonsense mutations, or stop codons. In the sequences of 433 bp of cyt b gene segment, 375 sites were constant and 58 characters were variable. Of the total variable sites, 21 characters were parsimony uninformative and 37 characters were parsimony informative. Nucleotide frequencies in the haplotypes were 34%, 11%, 9%, and 44% for A, C, G, and T, respectively. There were multiple hits at eight sites (127, 142, 146, 272, 304, 376, 415, and 421). Among other variable sites, there were 30 transitions and 20 transversions (ti/tv = 1.5). T h e protein coding region contained 16 amino acid replacements without any indels or nonsense mutations in the translated protein sequences.

When haplotypes and their frequencies were examined we found that the most abundant haplotype was H1, which was detected in 36 individuals representing 8 populations ( Table 2). Three haplotypes (H4, H5, and H6) were found as private haplotypes. Interestingly, the Çanakkale population had two of the private haplotypes (H5 and H6). When the populations were examined with respect to their haplotype number the Kütahya population had the highest number of haplotypes (n = 4) followed by Çanakkale, Kahramanmaraş, Uşak, and Kayseri (n = 3). The Balıkesir, İstanbul, Konya, and Manisa populations each had two distinct haplotypes. The remaining populations (Afyon, Antalya, Denizli, Düzce, Eskişehir, and Kırıkkale) had a single type of haplotype in each population. In our study, haplotype richness was uncorrelated with the sampling effort (Spearman correlation: r S = 0.161, n = 117, P = 0.56). Rarefaction analysis results showed that the Chao- 1 estimator would not be calculated for 10 populations, in which there were no doubleton haplotypes (S* 1 = –1). However, Kütahya (S* 1 = 4.5 ± 1.12), Uşak (S* 1 = 3.0 ± 0.01), Kayseri (S* 1 = 3.0 ± 0.01), Balıkesir (S* 1 = 2.5 ± 1.12), and İstanbul (S* 1 = 2.0 ± 0.01) displayed a range of Choa-1 estimator values .

The genetic diversity estimates of A. lignicola populations showed that haplotype diversity varied from 0.00 to 0.777 with an average of 0.32519 ( Table 3). The highest haplotype diversity (h = 0.7778) was detected in the Kütahya population, followed by the Uşak (h = 0.6889) and Kayseri, Balıkesir, and İstanbul populations (h = 0.6667). Six of the remaining populations showed no haplotype diversity due to the occurrence of only a single type of population, N h

: number of haplotypes, h: haplotype diversity, π: nucleotide diversity. haplotype in these populations. Nucleotide diversity ranged between 0.00 and 0.042494. The average nucleotide diversity was calculated as π = 0.0087955 (0.8%). The Uşak population displayed the highest nucleotide diversity estimate (π = 0.042494) followed by the Konya population (π = 0.039415). The mismatch distribution, including all samples, indicated a bimodal profile. The Harpending raggedness index was low (r = 0.0044) but not significant. Overall, Tajima’s D neutrality test (Tajima’s D = 0.47873, P> 0.10) and Fu and Li tests (D * = 0.41064, P> 0.10; F * = 0.50013, P> 0.10) were not significant.

A pairwise comparison among A. lignicola haplotypes was conducted to determine the sequence differences. Among all pairwise comparisons, the sequence divergence varied from 0.02% to 10.6% (1 to 46 bp, respectively) ( Table 4). The most divergent haplotypes H5 (n = 1, Çanakkale population) and H18 (n = 2, Uşak population) were separated from each other by 46 nucleotides. The least divergence, with 1 bp difference, was determined between H3 (n = 2 Balıkesir) and H4 (n = 1 Balıkesir), H1 (common haplotype) and H8– (n = 2 İstanbul) H16 (n = 4 Kütahya, n = 3 Manisa), H6 (n = 1 Çanakkale) and H7 (n = 8 Çanakkale), H9 (n = 6 Kahramanmaraş, n = 2 Kayseri, n = 1 Kütahya) and H10 (n = 3 Kahramanmaraş)–H11 (n = 5 Kayseri)– H12 (n = 2 Kayseri), H14 (n = 4 Konya, n = 3 Kütahya, n = 5 Uşak), and H17 (n = 3 Uşak) haplotypes.

The pairwise F st calculations showed significant genetic differentiation among some populations ( Table 5). In particular, two locations showed complete differentiation from some other populations; Antalya was significantly different from Afyon, Denizli, Düzce, Eskişehir, and Kırıkkale (F st = 1); and Kırıkkale was different from Afyon, Antalya, Denizli, Düzce, and Eskişehir (F st = 1). Some of the populations had no genetic differentiation (F st = 0) from each other (Afyon and Denizli, Düzce and Eskişehir populations) because these populations share a single haplotype (H1). Other populations displayed genetic differentiation values on a scale of 0 and 1, with statistically significant support (P <0.001) indicating some degrees of differentiation from each other.

3.2. Phylogenetic relationships among Andricus lignicola haplotypes

Using PAUP* 4.0b for estimation of phylogenetic relationships among 18 A. lignicola haplotypes, MP and ML produced similar tree topologies with different bootstrap values; thus, only a single phylogenetic tree was shown in Figure 1. MP is a consensus tree of 123 shortest trees, which was produced with CI = 0.691 and 139-step length. For ML analysis, jModeltest was used to determine the mutational model that best approximated the sequence evolution of the data set and calculate the transition and transversion ratios; it identified HKY + I

Ak

Figure 1. The consensus tree of both MP/ML analyses for the cyt b gene region of A.

lignicola. Bootstrap values are shown on the branches for both MP and ML, respectively.

Outgroup haplotypes: Ac ( Andricus caliciformis ) and Ak ( Andricus kollari ).

( Hasegawa et al., 1985) as the best fit model to the data set (I = 0.3778). Thus, this substitution model was utilized in ML and BI analysis. ML and MP trees produced two well-supported clades: clade A is composed of a basally located haplotype (H15) from the Konya population and a small polytomous group of three haplotypes (H14, H17, and H18). However, clade B is larger and composed of two further subclades. Evolutionary relationships were not clearly resolved in small subclades; however, in clade B in the large polytomous part, in addition to the presence of a commonly shared haplotype (H1), all other haplotypes are geographically restricted to the western populations. In spite of a monophyletic grouping of two Balıkesir haplotypes (H3 and H4), the relationships of all other haplotypes showed polytomy that could be due to insufficient time elapsed since divergence between lineages or incomplete lineage sorting ( Avise, 2000). Some of the haplotypes representing westerly populations such as H14, H17, H18, and H15 (from Konya, Kütahya, and Uşak) seem to be well-separated from other haplotypes .

The tree resulting from the Bayesian analysis (BI) is given in Figure 2 with posterior probabilities on the branches. In the Bayesian tree there is polytomy at the basal part, which comprises H16 (from Kütahya and Manisa) , H8 (from İstanbul), H1 (shared common haplotype from Afyon, Denizli, Düzce, Eskişehir, İstanbul, Kahramanmaraş, Kütahya, and Manisa), and a large clade that covers the rest of the haplotypes. In the large clade there is also polytomy composed of H13 ( KIR) and H7 ( CAN), a monophyletic group including H5 and H6 from Çanakkale population, another monophyletic group including H3 and H4 from Balıkesir, and the third lineagemaking polytomy that includes the remaining haplotypes. Within this lineage two haplogroups are observed to be monophyletic. Of these, the first subclade is composed of a small polyphyletic group, which includes H9 (from Kahramanmaraş, Kayseri, and Kütahya), H10 (from Kahramanmaraş) , and H11 and H12 (both from Kayseri). Another polytomic group has a basally located haplotype H2 from Antalya and the small polytomic group composed of H14 (Konya, Kütahya, and Uşak) and H15, H16, and H17 (Uşak) .

The haplotype network analysis shown in Figure 3 produced three main haplogroups with an additional

ESK, IST, KAH, KUT, MAN single haplotype (H2 from Antalya) that could not be connected under the 95% connection limit. H2, with the uppermost substitution value, could not be grouped with other haplotypes. In the obtained TCS network haplotypes of the largest cluster seemed to be derived from H1, which was the most common and, geographically, the most widely distributed haplotype. In the large cluster, H6 and H8 did not connect to any haplotypes other than H1. However, H1 was connected to three hypothetical haplotypes that provided a connection to several other haplotypes, such as H5 and H7. In addition to H5, H6, and H7 haplotypes, H4, H3, and H13 are clustered in the large group of haplotypes of the network ( Figure 2 ). The third haplotype cluster, which formed a small grouping with the uppermost substitution value, belonged to H9. Haplotype 9 seemed to be connected to H12, H10, and H11. The last haplogroup contains H14 with the uppermost value, which was connected to two hypothetical haplotypes with three haplotypes (H15, H17, and H18).

The same haplotype groupings were also produced by ABGD analysis and generated four groups, including recursive evaluation of group splitting with the initial partition with prior maximal distance P = 1.67e- 03. The ABGD analysis united H1, H3–H8, H13, and H16 as group 1; H2 as group 2; H9 and H1–H12 as group 3; and H14–H15 and H17–H18 as group 4. The genetic distances were d = 0.0062, 0.021, and 0.0036 for the first, third, and fourth groups, respectively. The genetic distance could not be calculated for the second group due to the presence of a single haplotype in this group. The calculated pairwise genetic distances between each of the four detected groups were 0.037 (between groups 1 and 2), 0.0036 (between groups 1 and 3), and 0.0054 (between groups 1 and 4). Likewise, group 4 showed a distance of 0.049 from group 2 and 0.051 from group 3 .

AMOVA for revealing population structuring was conducted through several trials in groupings of the populations. They were tested for significant clustering and partitioning of the genetic variation at two/three levels; however, only two of the trial schemes are included in Table 6. In the first trial scheme all populations were considered a distinct group, and it was determined that variation was partitioned among groups with 71.43%, and the remaining variation (28.57%) was at the “within population” level. These results were statistically significant (P <0.001) ( Table 6a). On the other hand, when all the populations were divided into four groups based on the groupings obtained after ABGD and haplotype network analyses, an even greater level of genetic partitioning value with statistically significant support was detected; 74.12% of variance components were recovered among groups, 16.03% among populations within groups, and 9.86% within population ( Table 6b).

Figures

Figure 2. Bayesian analysis tree. Posterior probability values are given on the branches. Outgroup haplotypes: Ac (Andricus caliciformis) and Ak (Andricus kollari).