Ginkgo biloba's footprint of dynamic Pleistocene history dates back only 390,000 years ago

Abstract

Background: At the end of the Pliocene and the beginning of Pleistocene glaciation and deglaciation cycles Ginkgo biloba went extinct all over the world, and only few populations remained in China in relict areas serving as sanctuary for Tertiary relict trees. Yet the status of these regions as refuge areas with naturally existing populations has been proven not earlier than one decade ago. Herein we elaborated the hypothesis that during the Pleistocene cooling periods G. biloba expanded its distribution range in China repeatedly. Whole plastid genomes were sequenced, assembled and annotated, and sequence data was analyzed in a phylogenetic framework of the entire gymnosperms to establish a robust spatio-temporal framework for gymnosperms and in particular for G. biloba Pleistocene evolutionary history.

Results: Using a phylogenetic approach, we identified that Ginkgoatae stem group age is about 325 million years, whereas crown group radiation of extant Ginkgo started not earlier than 390,000 years ago. During repeated warming phases, Gingko populations were separated and isolated by contraction of distribution range and retreated into mountainous regions serving as refuge for warm-temperate deciduous forests. Diversification and phylogenetic splits correlate with the onset of cooling phases when Ginkgo expanded its distribution range and gene pools merged.

Conclusions: Analysis of whole plastid genome sequence data representing the entire spatio-temporal genetic variation of wild extant Ginkgo populations revealed the deepest temporal footprint dating back to approximately 390,000 years ago. Present-day directional West-East admixture of genetic diversity is shown to be the result of pronounced effects of the last cooling period. Our evolutionary framework will serve as a conceptual roadmap for forthcoming genomic sequence data, which can then provide deep insights into the demographic history of Ginkgo.

Keywords: Evolutionary history; Ginkgo biloba; Phylogenomics; Pleistocene.

Fig. 1

Phylogenetic network reconstruction of 71 Ginkgo accessions. a TCS network based on complete plastomes. 17 haplotypes were clustered into eight haplotype group (A-H), different colors represent populations of origin. Six sample names denote the samples used in the gymnosperm-wide dataset. Size of haplotype pie charts is proportional to the number of individual sequences assigned to the respective haplotype. b Geographic distribution of the eight haplotype groups. Map was created with packages ‘maps’ [62], ‘mapdata’ [63] and ‘mapplots’ [64] in R version 3.2.3 [65] using the ‘worldHires’ map. Detailed geographic coordinates are provided with Additional file 1

Fig. 2

Maximum likelihood phylogenetic reconstruction of Ginkgo in a gymnosperm-wide context Bootstrap values from 1000 bootstrap replicates are indicated (** for bootstrap support of 100% and * for support of 95-99%). Bootstrap support was 90% for the split between Ginkgo and the group of Pinaceae, Gnetophytes and Cupressophytes. Amborella trichopoda (angiosperms) was set as outgroup. Images taken by M.A. Koch

Fig. 3

Divergence time estimation of Ginkgo in a gymnosperm-wide context. Combined BEAST results from four independent MCMC runs of 5*10⁸ generations each. Calibration was based on 11 fossils (see Additional files 7 and 8). Divergence time estimates are shown with their 95% HPD intervals

Fig. 4

Ginkgo secondary calibration extracted from the gymnosperm-wide divergence time estimation. a Age distribution from the four different combinations of tree model and topology. Ginkgo crown age and the two oldest splits within the species (TM12/TM22 and TM12/ES10) were used for calibration. Displayed are combinations of all four runs for each respective parameter set, excluding burn-in. b Magnification of the Ginkgo intrageneric nodes from Fig. 3. Upper and lower 95% HPD limits are given. Red stars denote nodes used for secondary calibration in the Ginkgo only analysis (Fig. 5)

Fig. 5

Maximum likelihood phylogenetic reconstruction of 71 Ginkgo accessions. Bootstrap values from 1000 bootstrap replicates are indicated (** for bootstrap support of 100% and * for support of 95-99%). The clade containing SNJ1 and SNJ6 (i.e. haplotype groups E-H) was set as outgroup following the analysis in the gymnosperm-wide context

Fig. 6

Divergence time estimation of 71 Ginkgo accessions. Combined BEAST results from four independent MCMC runs of 1*10⁸ generations each. Calibration was based on secondary calibration derived from the gymnosperm-wide analysis. Divergence time estimates are shown with their 95% HPD intervals for those nodes that represent splits between haplotypes in maximum likelihood reconstruction

Fig. 7

Schematic timeline of Ginkgo evolution. a Graphical representation of our hypothesis of range contraction and isolation during warming phases and range expansion and admixture during cooling phases. b Simplification of divergence time estimations and branching patterns as shown in Fig. 6. Respective warming and cooling phases (S1-S4 and L1-L5 [55]) are indicated