Evolutionary history ofCastanea sativaMill. in the Caucasus driven by Middle and Late Pleistocene paleoenvironmental changes

Abstract

Due to global climate cooling and aridification since the Paleogene, the members of the Neogene flora were extirpated from the Northern Hemisphere or were confined to a few refugial areas. For some species, the final reduction/extinction came in the Pleistocene, but some others have survived climatic transformations up to the present. This has occurred in Castanea sativa, a species of high commercial value in Europe and a significant component of the Caucasian forest biodiversity. In contrast to the European range, neither the historical biogeography nor the population genetic structure of the species in the isolated Caucasian range has been clarified. Here, based on a survey of 21 natural populations from the Caucasus and a single one from Europe, we provide likely biogeographic reconstruction and genetic diversity details. By applying Bayesian inference, species distribution modelling, and fossil pollen data, we estimated (1) the time of the Caucasian - European divergence during the Middle Pleistocene (436.5 ka), (2) the time of divergence among Caucasian lineages, and (3) outlined the glacial refugia for species. The climate changes related to the Early Middle Pleistocene Transition and the alpine orogenic uplift in the region are proposed as the major drivers of the intraspecific divergence and European-Caucasian disjunction, while the impact of the last glacial cycle was of marginal importance.

Biodiversity protection against anthropogenic climate change: Conservation prioritization of Castanea sativa in the South Caucasus based on genetic and ecological metrics

“…Consequently, C. sativa in the western range (West Georgia) and eastern range (East Georgia and Azerbaijan) in the South Caucasus grows under different precipitation conditions. Studies have shown that the neutral genetic diversity of sweet chestnut in the Caucasus and the Iberian Peninsula are organized along the precipitation gradient (Beridze et al, 2023; Míguez‐Soto et al, 2019). Beridze et al (2023) demonstrated a reduced gene flow between two major clusters covering the western and eastern parts of the species range in the Caucasus, corresponding to the two climatic groups defined in this study.…”

“…However, we know that accurate estimates of N e may be especially difficult in trees because some aspects of their population biology violate methodological assumptions of N e estimation (Santos‐del‐Blanco et al, 2022). The values presented here may be down‐biased because of spatially restricted sampling and the intensive gene flow (Beridze et al, 2023). Also, trees are assumed to be able to efficiently respond to selection despite low N e (Hoban et al, 2021) because of a unique set of life history traits and genetic architecture allowing them for rapid adaptation (Aitken et al, 2008; Petit & Hampe, 2006).…”

“…As the final factor possibly governing the pattern of diversity and differentiation, the impact of historical demographic events (divergence and admixture) was considered. Here, we used the STRUCTURE results previously performed for studied populations by Beridze et al (2023). Specifically, Q ‐membership values at K = 4, which defined the four evolutionary lineages in the contemporary gene pool of sweet chestnut, were extracted to test the IBC and trace the ancestry factor in the genetic diversity pattern.…”

“…This software estimates N e as the mutation‐scaled parameter, which is Ө = 4 N e μ , for diploid nuclear organisms, where μ is the mutation rate. Ө was converted into N e based on the mutation rate of 5.47 × 10 −5 estimated using the set of SSRs by Beridze et al (2023). In Migrate‐n, we used the Brownian motion model, with starting conditions based on F ST and uniform prior distribution, to estimate theta Ө with minimum, maximum, and delta values of 0, 100, and 10, respectively.…”

“…The software was run with 10 5 seeps, 10 4 burn-in, and 10 replications for each K (number of clusters). The tested K varied from 2 to 5 and was chosen based on the results of STRUCTURE's generic clustering performed by Beridze et al (2023). Optimal K and best run were chosen according to DIC value and correlation plot (Appendix S1, Figure S4).…”