Many insects possess intracellular bacterial symbionts that they depend on for essential nutrients, such as amino acids or vitamins, which are missing from their diet. These insects are often additionally infected by secondary symbionts that provide context-dependent benefits, such as defense against parasitoid wasps or heat stress. Pea aphids are a model for both obligate and facultative symbioses; they possess an ancient symbiont, Buchnera aphidicola, and can harbor secondary symbionts, including Serratia symbiotica. This dissertation begins with a review on animal-microbe symbioses, including background on symbiont population genetics, that helps give shape to the subsequent research chapters (Chapter 1). Then, I investigate the capacity of culturable S. symbiotica strains, which are closely related to non-culturable, beneficial strains, for evolving into intracellular symbionts of aphids (Chapter 2). After identifying pathogenicity as one of the limiting factors preventing long-term symbiosis between culturable S. symbiotica and aphids, I investigate whether knockout of pathogenicity factors or general mutagenesis can be used to produce avirulent, symbiotic strains (Chapter 3). In my last research chapter, I use Buchnera haplotypes to explore the evolutionary pressures faced by intracellular, maternally transmitted symbionts of aphids (Chapter 4). I conclude by summarizing the main findings of my dissertation research and by offering some remaining questions and future directions (Chapter 5). Chapter 1 sets the stage by reviewing how animal-microbe symbioses adapt, or ‘innovate’, over long periods of time and changing environmental conditions. Recent publications have shown that symbioses adapt by diverse strategies, including horizontal gene transfer to hosts or symbionts, the evolution of novel protein functions, and sometimes – the supplementation or replacement of symbionts altogether. In this review, we provide a framework for understanding why different symbiotic relationships consistently adopt specific approaches for innovation. We explain that these routes depend on their symbionts’ population structures which can be categorized as ‘open’, ‘mixed’, or ‘closed’. We first review how these structures influence symbiont evolution, ultimately producing distinct genomic ‘syndromes’ for symbionts. We then dive into the specific types of innovations that repeatedly arise in each category of symbiosis. Chapter 2 investigates the capacity of cultured S. symbiotica gut pathogens, strains CWBI2.3ᵀ and HB1, to evolve towards intracellular symbiosis in aphids. By injecting these strains into the aphid body cavity and imaging dissected ovaries, we show that both strains can enter aphid embryos through a specialized transmission route previously described for Buchnera or mutualistic and non-culturable S. symbiotica strains. As embryos develop, CWBI-2.3ᵀ is sorted into aphid cells (bacteriocytes) that are distinct from those that house Buchnera, in a fashion similar to mutualistic strains. However, we show that CWBI-2.3ᵀ and HB1 do not evolve to an intracellular, vertically transmitted lifestyle because both strains act as pathogens in aphid hemolymph (blood), killing their hosts or otherwise preventing transmission to the following generation. Chapter 3 explores whether virulence of S. symbiotica CWBI-2.3ᵀ to aphids can be attenuated by knockout of specific virulence factors, or by non-targeted mutagenesis. Using comparative genomics, we found that S. symbiotica gut pathogens possess a ysa-like type III secretion system (T3SS) which was horizontally acquired from Yersinia and has degenerated in S. symbiotica mutualists. T3SSs are important to many symbiotic relationships, but we found that the ysa T3SS does not contribute to pathogenicity or long-term symbiosis (vertical transmission) between S. symbiotica and aphids. We additionally tested whether the accumulation of deleterious mutations through ultraviolet (UV) mutagenesis can reduce S. symbiotica virulence. We treated 48 lines of S. symbiotica with 16 cycles of UV exposure but found only slight and nonsignificant reductions in overall virulence. Together, we demonstrate that mutagenized strains are still unable to form long-term symbioses with aphids. Chapter 4 considers the evolutionary pressures faced by intracellular Buchnera symbionts of aphids. This project was a co-first author collaboration with Dr. Bo Zhang, a former post doc in the Moran lab. Bo used microinjection to generate two aphid matrilines, each possessing two Buchnera haplotypes (LSR1 and 5A, or LSR1 and 5AY). Bo maintained these matrilines under two thermal conditions (cool or hot) and quantified Buchnera haplotype frequencies at each generation, over five generations total, using next generation sequencing of a variable genomic region. This frequency data was used to estimate two key parameters known to shape symbiont evolution: within-host selection and the population bottleneck size. During initial injections, a third Buchnera haplotype (designated 5Aᴱ) arose from the 5A background and quickly outcompeted the two other strains (5A and LSR1) under both thermal conditions. We described this new haplotype and performed a population-level competition experiment of aphids clonally infected with their native Buchnera haplotype or with the new Buchnera haplotype. We demonstrated that, despite being strongly selected for within hosts, this novel Buchnera haplotype does not appear to be deleterious to hosts under our experimental conditions. Chapter 5 summarizes the main findings of my PhD and suggests future directions for researchers interested in the evolution of aphid-bacterial symbioses.