Fluctuating selection in a monkeyflower hybrid zone

Abstract

While hybridization was viewed as a hindrance to adaptation and speciation by early evolutionary biologists, recent studies have demonstrated the importance of hybridization in facilitating evolutionary processes. However, it is still not well-known what role spatial and temporal variation in natural selection play in the maintenance of naturally occurring hybrid zones. To identify whether hybridization is adaptive between two closely related monkeyflower species, Mimulus guttatus and Mimulus laciniatus, we performed repeated reciprocal transplants between natural hybrid and pure species' populations. We planted parental genotypes along with multiple experimental hybrid generations in a dry (2021) and extremely wet (2023) year in the Sierra Nevada, CA. By taking fine-scale environmental measurements, we found that the environment of the hybrid zone is more similar to M. laciniatus's seasonally dry rocky outcrop habitat than M. guttatus's moist meadows. In our transplants hybridization does not appear to be maintained by a consistent fitness advantage of hybrids over parental species in hybrid zones, but rather a lack of strong selection against hybrids. We also found higher fitness of the drought-adapted species, M. laciniatus, than M. guttatus in both species' habitats, as well as phenotypic selection for M. laciniatus-like traits in the hybrid habitat in the dry year of our experiment. These findings suggest that in this system, hybridization might function to introduce drought-adapted traits and genes from M. laciniatus into M. guttatus, specifically in years with limited soil moisture. However, we also find evidence of genetic incompatibilities in second generation hybrids in the wetter year, which may balance a selective advantage of M. laciniatus introgression. Therefore, we find that hybridization in this system is both potentially adaptive and costly, and that the interaction of positive and negative selection likely determines patterns of gene flow between these Mimulus species.

Keywords: adaptation; hybridization; natural selection.

Figure 1.

Pictures of (A) M. guttatus parental habitat at the end of the season (B) hybrid zone habitat (C) experimental plot set-up in M. laciniatus parental habitat, and (D) crossing design of six genotypes, with colors indicating parental genetic material (created with Biorender.com).

Figure 2.

Predicted fecundity in each site given hypothesized fitness advantage of hybrids in hybrid sites, and local adaptation of each species in its native habitat. Plant genotypes are grouped by site, and genotype is indicated by color. Predictions show higher fitness of individuals with M. guttatus genes in the M. guttatus habitat (A), higher hybrid fitness in the hybrid habitat (B), and higher fitness of individuals with M. laciniatus genes in the M. laciniatus habitat (C).

Figure 3.

Soil moisture (A) and plant survival (B) decrease over time in M. guttatus, hybrid, and M. laciniatus sites. Solid lines connect weekly site means in 2021 and dotted lines connected weekly site means in 2023. Principal component analyses of soil moisture decrease over time in the dry year (C and D) and wet year (E and F). PC1 indicates differences in total soil moisture between sites (C and E), while PC2 and PC3 indicate differences in shape of soil moisture curves between sites (D and F).

Figure 4.

Average lifetime fecundity (total seed number/number of plants planted) per genotype in each site, with sites plotted separately. Upper row (A–C) is 2021 sites and lower row (D–F) is 2023 sites. Genotypes are, from left to right, M. guttatus habitat M. guttatus (GG), hybrid habitat M. guttatus (HG), back-crossed M. guttatus (BCG), first-generation hybrid (F₁), second generation hybrid (F₂), backcrossed M. laciniatus (BCL), hybrid habitat M. laciniatus (HL) and M. laciniatus habitat M. laciniatus (LL). Error bars represent 95% confidence intervals from bootstrap analysis. Values for average fecundity and 95% CI can be found in Supplementary Table 1.

Figure 5.

Phenotypic selection analysis in 2021 of all genotypes combined in the M. guttatus and Hybrid sites. There was no significant selection for leaf roundedness for genotypes combined in either species habitat in 2021. Pictures and arrows above the graph indicate which species’ traits the direction of selection is moving towards, M. laciniatus (left) and M. guttatus (right). Strength of selection refers to the selection gradient (β) from the zero-truncated poisson best fit model for all genotypes combined. Values from this graph can be found in Table 3, and model parameters can be found in Supplementary Table S6.

Figure 6.

Directional selection on traits with all genotypes combined in M. guttatus, Hybrid, and M. laciniatus sites, visualized using partial residuals from multiple regression of best fit zero-truncated poisson models (Breheny & Burchett, 2017). The fitted curves show best fitting linear regression in 2021 (A–F) and both years (G). Plant height (A and C) and flower width (B and D) were in the best fit models for both sites, while stigma-anther separation (E) and leaf size (F) were only in best fit models for the hybrid habitat. Graph (G) also includes selection differentials for flowering time measured in 2023 in M. guttatus, hybrid and M. laciniatus habitats.