COS 174-1 - CANCELLED - Baseline expectations for the high variability in outcomes of hybridization.

Background/Question/Methods

Hybridization can happen whenever two related species interact, but the rate and consequences of hybridization are variable. While there is a general expectation that some evolutionary processes (specifically selection or environmental dependence) have the potential to act in a similar way across hybrid zones, genetic drift is known to make outcomes of hybridization extremely variable. Recently there has been growing interest in comparing replicate hybrid zones to understand the ecological and evolutionary processes that shape the frequency and outcomes of hybridization. For conservation, understanding variability in hybridization is critical for risk assessment and mitigation. However, we lack baseline expectations for variation in outcomes of hybridization.

We used replicates of simulated hybrid zones (including treatments where we varied selection, migration, and the genetics of incompatibilities between species), to estimate a distribution of outcomes and quantified variability among replicates. We quantified hybridization outcomes for individuals in the resulting populations, based on their admixture proportions, diploid ancestry complement, and number of ancestry junctions across simulated genomes.

Results/Conclusions

The distributions of admixture score, ancestry complement, and the number of junctions for replicates resulting from the same evolutionary processes were extremely variable. We found decreased predictability with more generations since hybridization began, and with rare gene flow. This suggests that, particularly for finite population sizes, the outcomes of hybridization are expected to be highly variable.

We used simulated replicate hybrid zones, with the same exact rate of migration and selection, and still found that it is difficult to predict the outcomes of hybridization in held-out samples. Empirical studies will not be perfect replicates; our simulations represent a best-case-scenario for prediction. Additionally, empirical data will often be consistent with multiple alternative evolutionary and ecological contexts or histories (low identifiability). Our understanding of hybrid zone dynamics will improve by combining analyses of variablities of outcomes from simulated data to provide a null expectation for the identifiability of processes from empirical data.

To manage hybrids for conservation purposes, our work suggests that extrapolating from a few locations with hybrids is ill-advised and that effective strategies may require prohibitive sampling effort, as expectations for outcomes of hybridization may not extend across multiple locations. Targeted management could benefit from information at focal sites.