*2.4. Genetic Diversity of the Mitochondrial and Plastid Genome in Common Buckwheat*

The assembled sequence of the buckwheat mitogenome allowed us to estimate the diversity among cultivars. We found that it is unexpectedly low (Table S2). Only three-four substitutions differ Dasha from most other Russian cultivars. The most divergent are the cultivars Koto and Shinanonatsusoba, which have six substitutions. All variants are located in the non-coding regions. Interestingly, three out of six substitutions in Koto are shared with Russian cultivars and two more are shared with Shinanonatsusoba, the Japanese cultivar. This may indicate that Koto and Shinanonatsusoba have a recent common maternal ancestor. Shinanonatsusoba is a Japanese cultivar that originated in 1984 (Nishimaki Et al., 1984). Koto is a cultivar produced by the Canadian breeding company Kade Research in 2002. According to Ikeda [37], Canadian buckwheat breeders (not excepting Kade Research) specialized in the breeding of buckwheat for Japanese market, even the names of many Canadian cultivars (including Koto) are derived from Japanese. Thus, it is highly likely that Japanese cultivars, in particular Shinanonatsusoba, might contribute to the Koto genome. In contrast to the low divergence between cultivars, *F. esculentum* ssp. *ancestrale* differs from Dasha in 220 positions. The prevalent type of the variant is single nucleotide variation (SNV), followed by deletions and insertions (Table S2). All variants are located in the non-coding regions or in the synonymous positions of the codon and do not affect the amino acid sequence.

In an earlier study [5], we characterized the plastid genome of *F. esculentum* ssp. *ancestrale*. However, we found that it contains a number of inaccuracies, mostly indels associated with the homopolymer and low-complexity regions, which are error-prone for Sanger sequencing. Since indels in the reference genome may result in mapping and variant calling errors, we decided not to use this sequence as a reference. Taking advantage of the long reads, we assembled and annotated the plastid genome of the Dasha cultivar and used it as a backbone for the analysis of the plastid genome diversity. The diversity among cultivars is also extremely low; as well as for the mitochondrial genome, this concerns not only Russian cultivars but also Koto and Shinanonatsusoba. For Dasha and Dizajn, we found no differences at all. Dasha and other cultivars differ in one position: T to G substitution in the intron of the trnG-UCC gene. Shinanonatsusoba carries an additional change, the substitution in the trnG-trnM spacer. As well as for the mitochondrial genome, the number of SNPs with *F. esculentum* ssp. *ancestrale* is much higher: 159 (Table S4) (148 if SNPs located in the IR counted once). The SNP density is the highest in the small single copy region and the lowest in the inverted repeat (IR), as expected. In total, 81 SNPs are located in the spacers, 20 in the introns, and 58 in the coding regions. What is surprising is that 32 substitutions are non-synonymous (Table 2). While several types of mutations (C-to-T substitutions) can potentially be silenced by RNA editing (see, for example, [38]), only three substitutions out of 32 are C-to-T. Most substitutions affect highly variable genes, such as *matK*, *ycf1*, and *rpoC2*; however, there are also ones that change the amino acid sequence of highly conserved genes encoding photosystem components (*psbA*, *psbH*, *psbC*). While missenses in photosynthesis-related genes usually adversely affect photosynthesis, there are several examples of a positive effect compared to the wild type [39]. We hypothesize that this mutation(s) increases the photosynthesis efficiency and their fixation in buckwheat cultivars is the result of artificial selection. The increase of the photosynthesis efficiency has indeed been an important trend in buckwheat breeding since the middle of the 20th

century. The survey of photosynthesis efficiency in different cultivars and landraces showed that in modern cultivars (developed in 1990–2010, including three sampled in our study: Demetra, Dizajn, and Devyatka), it is higher than in old cultivars and local landraces up to approximately 20% [40]. At the same time, modern cultivars have a low adaptive potential, and they are less resistant to unfavorable environmental conditions (in particular, drought) [41].


**Table 2.** Non-synonymous substitutions that differentiate Dasha and *F. esculentum* ssp. *ancestrale* plastid genes (substitutions located in the IR are counted once).

Taken together, the analysis of the extranuclear genome diversity shows that buckwheat cultivars sampled in this study share a very recent common maternal ancestor. This would not be surprising for Russian cultivars, but our study also included Japanese and Canadian (presumably of Japanese origin) cultivars. This might indicate the loss of genetic diversity as a side effect of the intensification of buckwheat breeding in the last several decades. A similar situation is observed for sunflower: While a high diversity exists in wild populations, most cultivated varieties arise from a limited pool of germplasm [42,43]. This calls for better characterization of the buckwheat germplasm in order to understand the patterns and limits of its diversity (including that of extranuclear genomes). We expect that the availability of reference sequences for plastid and mitochondrial genomes will facilitate this research program.
