Plant cells harbor two membrane-bound organelles containing their own genetic material-plastids and mitochondria. Although the two organelles coexist and coevolve within the same plant cells, they differ in genome copy number, intracellular organization, and mode of segregation. How these attributes affect the time to fixation or, conversely, loss of neutral alleles is currently unresolved. Here, we show that mitochondria and plastids share the same mutation rate, yet plastid alleles remain in a heteroplasmic state significantly longer compared with mitochondrial alleles. By analyzing genetic variants across populations of the marine flowering plant Zostera marina and simulating organelle allele dynamics, we examine the determinants of allele segregation and allele fixation. Our results suggest that the bottlenecks on the cell population, e.g. during branching or seeding, and stratification of the meristematic tissue are important determinants of mitochondrial allele dynamics. Furthermore, we suggest that the prolonged plastid allele dynamics are due to a yet unknown active plastid partition mechanism. The dissimilarity between plastid and mitochondrial novel allele fixation at different levels of organization may manifest in differences in adaptation processes. Our study uncovers fundamental principles of organelle population genetics that are essential for further investigations of long-term evolution and molecular dating of divergence events.

Background: Spinocerebellar ataxia types 2 (SCA2) and 3 (SCA3/MJD) are diseases due to dominant unstable expansions of CAG repeats (CAGexp). Age of onset of symptoms (AO) correlates with the CAGexp length. Repeat instability leads to increases in the expanded repeats, to important AO anticipations and to the eventual extinction of lineages. Because of that, compensatory forces are expected to act on the maintenance of expanded alleles, but they are poorly understood. Objectives: we described the CAGexp dynamics, adapting a classical equation and aiming to estimate for how many generations will the descendants of a de novo expansion last. Methods: A mathematical model was adapted to encompass anticipation, fitness, and allelic segregation; and empirical data fed the model. The arbitrated ancestral mutations included in the model had the lowest CAGexp and the highest AO described in the literature. One thousand generations were simulated until the alleles were eliminated, fixed, or 650 generations had passed. Results: All SCA2 lineages were eliminated in a median of 10 generations. In SCA3/MJD lineages, 593 were eliminated in a median of 29 generations. The other ones were eliminated due to anticipation after the 650th generation or remained indefinitely with CAG repeats transitioning between expanded and unexpanded ranges. Discussion: the model predicted outcomes compatible with empirical data - the very old ancestral SCA3/MJD haplotype, and the de novo SCA2 expansions -, which previously seemed to be contradictory. This model accommodates these data into understandable dynamics and might be useful for other CAGexp disorders.