Document Type


Date of Award


Degree Name

Doctor of Philosophy (PhD)


Biological Sciences

First Advisor

Heather Fiumera


Mitochondrial haplotypes contribute to functional diversity in natural populations. Uniparental inheritance makes it difficult to characterize the genetic architecture of mitochondrially driven phenotypes. In this work I explored the natural diversity of mitochondrial genomes in the yeast Saccharomyces cerevisiae. Few complete mitochondrial genomes were available for S. cerevisiae owing to challenges in high-throughput sequencing. I developed sequencing strategies using new technologies to generate complete high quality yeast mitochondrial genomes. Comparisons of 100 complete yeast mitochondrial genomes demonstrated extensive variation between populations in coding sequences and variable introns.

I demonstrated that these mitochondrial variants directly caused growth differences in strains with isogenic nuclear genomes but divergent mitochondrial genomes. Synthetic mitochondrial-nuclear combinations generally performed worse than coadapted combinations, consistent with mitonuclear coevolution. Using mitochondrial recombination that naturally occurs in S. cerevisiae I showed that the genetic architecture of these growth differences is complex and likely due to multiple loci. Mitochondrial recombinants showed poor growth in ecologically relevant conditions consistent with negative epistasis between mitochondrial loci (mito-mito epistasis). I generated recombinants between mitotypes from various populations and found significant estimates of mito-mito epistasis affecting growth that were more frequently negative in sign. Crosses between mitotypes from divergent populations also negative impacted fitness through the generation of progeny that cannot respire. Negative mito-mito epistasis may act in concert with mitochondrial-nuclear epistasis as a post-zygotic barrier contributing to speciation.

I attempted to map mitochondrial variants that confer temperature dependent growth advantages through two methods. Using a bulk segregant approach I created large pools of mitochondrial recombinants and sequenced these pools before and after selection for high temperature growth to determine shifts in allele frequencies in response to selection. The sequencing pools showed evidence of low levels of recombination which were inconsistent with evidence of recombination events in sequencing reads. Simulations of recombination better support a low rate of recombination in the data. I also endeavored to map functional mitochondrial variants by identifying regions of the mitochondrial genome remaining after large deletions that still contained causative alleles. I did not identify a causative region but found evidence that several regions likely do not contain the causative site. The analyses presented here did not identify a causative site but did narrow the range of possibilities and showed that the rate of mitochondrial DNA recombination in S. cerevisiae is lower than prior estimates.

Included in

Biology Commons