Premise of the study: Interspecific hybridization can cause genetic structure across species ranges if the mating system and degree of sympatry/parapatry with close relatives varies geographically. The coastal dune endemic Camissoniopsis cheiranthifolia (Onagraceae) exhibits genetic subdivisions across its range, some of which are associated with shifts in mating system from outcrossing to selfing, while others are not. For instance, strong differentiation between large-flowered, self-incompatible (LF-SI) and large-flowered, self-compatible (LF-SC) populations occurs without much reduction in outcrossing or obvious barriers to gene flow. We hypothesized that LF-SI diverged from LF-SC via hybridization with the predominantly inland SI sister species C. bistorta.

Methods: We analyzed spatial proximity using 1460 herbarium records, and genetic variation at 12 microsatellites assayed for 805 and 404 individuals from 32 C. cheiranthifolia and 18 C. bistorta populations, respectively. We also assayed nine chloroplast microsatellites for 124 and 111 individuals from 27 and 19 populations, respectively. 

Key results: Closer parapatry was associated with unexpectedly high genetic continuity between LF-SI C. cheiranthifolia and C. bistorta. LF-SI genotypes clustered with C. bistorta exclusive of other C. cheiranthifolia genotypes. Similarly, pairwise FST among SI C. cheiranthifolia and C. bistorta, adjusted for geographic proximity, was not higher between heterospecific than conspecific populations. 

Conclusions: The lack of genetic differentiation between LF-SI C. cheiranthifolia and C. bistorta populations, even those located away from the zone of parapatry, suggests that LF-SI C. cheiranthifolia instead of hybridizing with C. bistorta is rather an ecotype of C. bistorta that has adapted to coastal dune habitat independent of other lineages in C. cheiranthifolia proper.

The nuclear microsatellite (nSSR) dataset represents 129 genotypes at 12 loci from 50 populations across the range of Camissoniopsis cheiranthifolia (32 populations) and C. bistorta (18 populations) collected in 2009 and 2010. Fragment sizes were binned using the MsatAllele package (version 1.04, Alberto 2009) for the R statistical computing environmt.For details about PCR conditions and other laboratory methods please see López-Villalobos et al., 2014.

We provide individual and population codes, latitude, longitude, location (coastal vs. inland), species (C.ch = C. cheiranthifolia, C.bi = C. bistorta) and the categorical variable used to identify gentic clusters and mating system variation (please, refer to the paper). Loci names are as presented in the supplementary information of this study and in López-Villalobos et al., 2014 and López-Villalobos and Eckert 2018.

The chloroplast microsatellite data represents number of variants (base pair repeats)  at nine polymorphic chloroplast DNA microsatellites (cpSSR; following Weising and Gardner, 1999; Chung and Staub, 2003) assyed for 124 individuals from 27 populations of  C. cheiranthifolia and 111 C. bistorta individuals from  19 populations.

The third dataset are pairwise F_ST [Weir and Cockerham,1984] and other genetic differentiation statistics from nSSR data estimated using the R package DiveRsity.This data was obtained using the methods described in the isolation-by-distance analyses from this study. Metadata for each colum is as follows: population comparison (codes as in nSSR dataset), population 1, population 2, latitude and longitude of populations 1 and 2, grpspp (species grouping, wCc = within C. cheiranthifolia, wCbi = within C. bistorta, bSp = between species), group according to whether populations are coastal (Co) or inland (In), gouping according to their genetic cluster as in Figure 5 (P=pink, B=Blue), geographic distance estimated as the great circle surface distance calculated using the R package geosphere, version 1.3-11 and several genetic differentiation statistics (Fst used in this paper) estimated using the R package DiveRsity.