Genotypic similarities among the parthenogenetic Darevskia rock lizards with different hybrid origins

Abstract

Background: The majority of parthenogenetic vertebrates derive from hybridization between sexually reproducing species, but the exact number of hybridization events ancestral to currently extant clonal lineages is difficult to determine. Usually, we do not know whether the parental species are able to contribute their genes to the parthenogenetic vertebrate lineages after the initial hybridization. In this paper, we address the hypothesis, whether some genotypes of seven phenotypically distinct parthenogenetic rock lizards (genus Darevskia) could have resulted from back-crosses of parthenogens with their presumed parental species. We also tried to identify, as precise as possible, the ancestral populations of all seven parthenogens.

Results: We analysed partial mtDNA sequences and microsatellite genotypes of all seven parthenogens and their presumed ansectral species, sampled across the entire geographic range of parthenogenesis in this group. Our results confirm the previous designation of the parental species, but further specify the maternal populations that are likely ancestral to different parthenogenetic lineages. Contrary to the expectation of independent hybrid origins of the unisexual taxa, we found that genotypes at multiple loci were shared frequently between different parthenogenetic species. The highest proportions of shared genotypes were detected between (i) D. sapphirina and D. bendimahiensis and (ii) D. dahli and D. armeniaca, and less often between other parthenogens. In case (ii), genotypes at the remaining loci were notably distinct.

Conclusions: We suggest that both observations (i-ii) can be explained by two parthenogenetic forms tracing their origin to a single initial hybridization event. In case (ii), however, occasional gene exchange between the unisexual and the parental bisexual species could have taken place after the onset of parthenogenetic reproduction. Indeed, backcrossed polyploid hybrids are relatively frequent in Darevskia, although no direct evidence of recent gene flow has been previously documented. Our results further suggest that parthenogens are losing heterozygosity as a result of allelic conversion, hence their fitness is expected to decline over time as genetic diversity declines. Backcrosses with the parental species could be a rescue mechanism which might prevent this decline, and therefore increase the persistance of unisexual forms.

Keywords: Allele conversion; Backcrosses; Caucasian rock lizards; Darevskia; Microsatellites; Mitochondrial DNA; Parthenogenesis.

Fig. 1

Map of sampling locations (a) and Median Joining network linking D. raddei and its daughter parthenogenetic forms (b). Location numbers (same as in Table 3) are shown on the map. Parental species in boldface. D. armeniaca: 1 - Hrazdan, Armenia; 2a - Didgori, Georgia; 2b - Khospio, Georgia; 3a - Ardahan, Turkey; 3b - Çıldır, Turkey; D. dahli: 4a – Kojori, Georgia; 4b – Didgori, Georgia; D. bendimahiensis: 5a - Muradiye, Turkey; 5b - Çaldıran, Turkey; D. sapphirina: 6a - Patnos, Turkey; 6b - Pınarlı, Turkey; D. rostombekowi: 7 - Dilijan, Armenia; D. unisexualis: 8a - Hrazdan, Armenia; 8b - Hanak, Turkey; 8c- Horasan, Turkey; 8d – River Ağrı, Turkey; D. uzzelli: 9a – Kars, Turkey; 9b - Sarıkamış, Turkey; 9c – Horasan, Turkey; D. mixta: 10a – Akhaldaba, Georgia; 10b – Abastumani, Georgia; 11 – Ambrolauri, Georgia; D. raddei raddei: 12а - Sotk Village, Armenia; D. raddei nairensis: 12b - Digor, Turkey; 12c - Vardzia, Georgia; D. raddei vanensis: 13a - Doğubeyazıt, Turkey; 13b - Muradiye, Turkey. D. portschinskii: 14a - River Khrami, Georgia; 14b - Kojori, Georgia; 15 - Mets Sepasar, Armenia; D. valentini: 15 – Akhalkalaki, Georgia; 16 - Ardahan, Turkey; 17 - Erzurum, Turkey; 18 - Çaldıran, Turkey. Circles of the same color are delimiting sampling areas in (a) and related haplogroups (b). Brown circle delimits haplogroups of D. raddei from Iran downloaded from GenBank, unrelated to any known parthenogenetic form, and hence not included in further analyses. The Caucasus Map was prepared in software QGIS (http://qgis.osgeo.org). Eurasia’s map at the bottom of the figure is modified from World Map Blank.svg (Public domain)

Fig. 2

Maximum Likelihood 50% consensus rule tree of haplotypes associated with different populations D. raddei and five its parthenogenetic daughter species, based on the HKY + G model. The 683 bp fragment of the mitochondrial Cytochrome b gene is used. The tree with the highest log likelihood (− 799.7117) is shown. The tree is based on 42 novel sequences and 62 sequences downloaded from GenBank. Bootstrap values exceeding 50% are shown on the nodes and branches. Individual clades are shown in color boxes. Clade 1 (blue box) - D. r. nairensis and D. r. raddei from the northernmost part of the species range and their daughter parthenogenetic species; clade 2 (yellow box) – D. r. vanensis and its daughter parthenogenetic species. Clade 3 (red box) – D. r. raddei from southern Armenia and Azerbaijan and their daughter parthenogenetic species; D. mixta and its daughter parthenogens (green box) are used as an outgroup for D. raddei and its daughter parthenogens. For the relative position of D. raddei and D. mixta in the phylogenetic tree of Darevskia, see [45, 66]

Fig. 3

a Allelic richness (the value of the index of Petit et al. [67]) (AR); b the difference between the observed and expected heterozygosity (OH – EH); c and the overall proportion of homozygous alleles (HoZ). Parthenogenetic taxa (grey) and their presumed ancestral species (black)

Fig. 4

ΔK and BSRK (the number of K selected with broken stick method) based on the clustering of bisexual species only (D. mixta, D. raddei, D. portschinskii + D. valentini) with admixture model, 1 = <K = < 15, 10 runs for each analysis

Fig. 5

STRUCTURE clustering outcome with K = 3 (upper panel) and K = 7 (lower panel). Admixture model applied, no LOCPRIOR used, POPFLAF = 1 for sexually breeding species (D. mixta, D. raddei, D. portschinskii, D. valentini). With K = 7, two geographic populations of D. mixta are from the Lesser and the Greater Caucasus respectively; the red cluster marks populations of D. raddei from Armenia and northern Turkey, blue cluster - D. raddei vanensis, brown cluster - D. valentini from Georgia and northeastern Turkey, yellow cluster - D. valentini from the Lake Van area

Fig. 6

Clustering of all parthenogenetic individuals based on a) Pairwise Bruvo distance between microsatellite genotypes represented as a heatmap with NJ cladogram. b) Same as in A, but all individuals with no or one heterozygous locus removed in an attempt to remove a possible effect of allele conversion. c) An NJ tree was built solely on the pairwise counts of loci with complete shared diploid genotypes (distance between genotypes measured as 10 - no. of shared loci). The colors in the “heatmaps” indicate the distance between the individuals: red color corresponds to small and yellow color to a larger distance

Fig. 7

The proportion of shared alleles among the parthenogenetic forms and their presumed ancestors. The colors in the “heatmap” indicate the proportion of shared alleles. Areas outside the black rectangle show the proportion of shared alleles between different parthenogens and different bisexual species, and the area within the rectangle – the alleles of the parental species shared with the parthenogens

Fig. 8

Connectivity of clonal lineages between (a) D. armeniaca and D.dahli, (b) D. bendimahiensis and D. sapphirina, (c) D. unisexualis and D. rostombekowi, (d) D. uzzelli and D. armeniaca, illustrated as Minimum Spanning Networks (MSNs). Each node represents a unique multilocus genotype, with size proportional to the number of individuals. The edges were constructed using Bruvo distances between microsatellite genotypes. The numbers next to the nodes show the maximum number of loci shared with the other species in a pair

Fig. 9

Probability of independent coincidence of a multilocus genotype between (a) random coincidence in two unisexual populations (heterozygous or homozygous genotypes); (b) random coincidence in a parthenogenetic and a bisexual population: all-heterozygous genotype; (c) random coincidence in a parthenogenetic and a bisexual population: all-homozygous genotype; (d) probability of coincidence in two parthenogenetic populations due to due to allelic conversion (only possible for homozygous genotypes) (see text). X-axis: the number of loci in a multilocus genotype; Y-axis: the respective probability values. All allele frequencies in all populations are set to ½

Fig. 10

Origin and genotypic overlap in parthenogenetic Darevskia. a Matrilineal and patrilineal ancestry of seven parthenogenetic forms of Darevskia, based on mitochondrial DNA sequences and STRUCTURE analysis at K = 7. Solid lines show matrilineal (left) and patrilineal (right) ancestors of the parthenogens, according to [45] and our data (this paper), specifying geographic populations of the presumed ancestors. Some parthenogens, however, were associated with more than two populations, and dashed lines show the links with these “third” populations (see the text below). b The scheme of the most frequent shared genotypes linking the parthenogenetic species of Darevskia to each other. Multiple lines demonstrate the numbers of loci with the shared genotypes of different parthenogenetic species