Increased productivity in hybrid lines suggests recovery from fitness breakdown
Previous long-term evolution experiments in T. californicus have found varying degrees of fitness recovery from F2 breakdown (Hwanget al. 2011; Pritchard et al. 2012; Hwang et al.2016), presumably because hybrids with matching mitonuclear complexes would have higher generalized fitness than their siblings with unmatched mitonuclear complexes. Thus, matched (coadapted) nuclear alleles would outcompete unmatched (incompatible) alleles over multiple generations (Fig. 1). However, those experimental lines started with a mix of pure parental individuals, so that unmixed nuclear genotypes and the two mitochondria could persist well into the experiment. This design made it challenging to distinguish between fitness recovery caused by the exclusion of incompatible nuclear alleles in a hybrid background, from fitness recovery caused by competition and drift of pure parental genotypes. In contrast, we initiated experimental lines with F2 hybrids and in fixed mitochondrial backgrounds, so that differences in fitness and allele frequency can only be explained by competition between parental nuclear alleles.
Our results show that control parental populations went extinct within two months of experimental evolution, probably due to inbreeding depression described in this species (Palmer & Edmands 2000). In contrast, almost all hybrid lines survived the nine months of experimental evolution, with most showing an increase in population size (from 1.1 to 9-fold; Fig. 2A). From the many traits known to be involved in hybrid breakdown in T. californicus (Burton 1990a; b; Edmands 1999; Edmands & Burton 1999; Ellison & Burton 2006; 2008a; 2010), we have assessed recovery in two: female fecundity and nauplii survivorship. In respect to fecundity, we observed a strong F2 breakdown (P-values< 0.017), and no recovery in evolved hybrid lines, irrespective of the mitochondrial background (Fig. 2B). This suggests that this is a complex trait that could not evolve in these experimental conditions of population size, number of generations and selective regime.
In contrast, with respect to survivorship, we observed less severe breakdown in F2 generation (P-values> 0.125). Notably, survivorship breakdown was stronger in the F2 hybrids with the SC mitochondria, suggesting that selection imposed by mitonuclear incompatibilities is stronger in SC relative to SD mitochondrial background. In both backgrounds, several hybrid lines recovered up to or beyond parental fitness levels (P-values< 0.043; Fig. 2C), suggesting that recombination and selection over nine months were sufficient to cause evolution at this trait. Rescue in survivorship up to or beyond parental fitness levels has also been reported in F9 recombinant hybrid lines of T. californicus (Pereira et al. 2014), suggesting that the genetic architecture of this specific trait might be relatively simple. Nevertheless, we note that the line showing the largest increase in productivity (9-fold increase in SD6), did not show recovery in survivorship, therefore selection imposed by our experimental design is likely multifarious and responding with phenotypes beyond those measured here. Future studies employing directed selection on a specific trait (Healy & Burton 2020) over many generations of experimental evolution may provide insights into this question.