Fluctuating reproductive isolation and stable ancestry structure in a fine-scaled mosaic of hybridizing Mimulus monkeyflowers

Hybridization among taxa impacts a variety of evolutionary processes from adaptation to extinction. We seek to understand both patterns of hybridization across taxa and the evolutionary and ecological forces driving those patterns. To this end, we use whole-genome low-coverage sequencing of 458 wild-grown and 1565 offspring individuals to characterize the structure, stability, and mating dynamics of admixed populations of Mimulus guttatus and Mimulus nasutus across a decade of sampling. In three streams, admixed genomes are common and a M. nasutus organellar haplotype is fixed in M. guttatus, but new hybridization events are rare. Admixture is strongly unidirectional, but each stream has a unique distribution of ancestry proportions. In one stream, three distinct cohorts of admixed ancestry are spatially structured at ~20-50m resolution and stable across years. Mating system provides almost complete isolation of M. nasutus from both M. guttatus and admixed cohorts, and is a partial barrier between admixed and M. guttatus cohorts. Isolation due to phenology is near-complete between M. guttatus and M. nasutus. Phenological isolation is a strong barrier in some years between admixed and M. guttatus cohorts, but a much weaker barrier in other years, providing a potential bridge for gene flow. These fluctuations are associated with differences in water availability across years, supporting a role for climate in mediating the strength of reproductive isolation. Together, mating system and phenology accurately predict fluctuations in assortative mating across years, which we estimate directly using paired maternal and offspring genotypes. Climate-driven fluctuations in reproductive isolation may promote the longer-term stability of a complex mosaic of hybrid ancestry, preventing either complete isolation or complete collapse of species barriers.

Methods
For full collection and processing details, see the associated manuscript and the analysis code available at https://github.com/mfarnitano/CAC_popgen.

Along three streams in the Columbia River Gorge area, Washington, USA, 0.5x0.5m square plots were laid down where Mimulus species were growing, during flowering seasons (April through June) of 2012, 2019, 2021, and 2022 (note that not all streams were sampled every year, for details see the associated manuscript).

At each plot for the years 2012, 2019, and 2022, phenological data was recorded: the total number of open flowers within each plot was counted every few days for the duration of the flowering season. 

In addition, leaf or bud tissue was collected for sequencing from wild samples of Mimulus guttatus, Mimulus nasutus, and admixed individuals, as well as the allopolyploid species Mimulus sookensis, growing in these streams during the years 2019, 2021, and 2022. For a subset of these wild samples in 2019 and 2022, one or two flowers were marked on the date they opened and the flowering date was recorded; later, the fruits resulting from these marked flowers were collected. Seeds from these fruits were germinated in greenhouse conditions and tissue was collected for sequencing. Note that 2012 samples were collected and presented as part of a previous study (see manuscript for details). 

DNA was extracted from both wild (maternal) and greenhouse (offspring) tissue samples using a CTAB extraction protocol. Samples were sequenced using a low-coverage Tagmentation library prep on an illumina sequencer, in three sequencing batches. 

Illumina sequence reads were aligned to the Mimulus guttatus IM62_v3 reference genome and processed according to the details in the associated manuscript. Ancestry proportions (hybrid indices) were obtained using a local ancestry hidden markov model (Ancestry_HMM). A PCA was conducted using genotype likelihoods from ANGSD and the PCAngsd program. Selfing and sibship estimation was conducting using BORICE.