Short-term fluctuating and long-term divergent selection on sympatric Monkeyflowers: insights from decade-spanning reciprocal transplants

Abstract

Sympatric species are often locally adapted to distinct microhabitats. However, temporal variation may cause local maladaptation and species boundary breakdown, especially during extreme climatic events leading to episodic selection. Repeated reciprocal transplants can reveal the interplay between short and long-term patterns of natural selection. To examine evolutionary trajectories of sympatric Monkeyflowers adapted to different niches, Mimulus guttatus and M. laciniatus, we performed three replicated transplants and combined them with previous experiments to leverage a dataset of five transplants spanning 10 years. We performed phenotypic selection analyses on parents and hybrids in parental habitats in Yosemite NP, CA during years of drastically differing snowpack. If there is ecological isolation, then we predicted divergent phenotypic selection between habitats in line with species' differences and local adaptation. We found interannual fluctuations in phenotypic selection, often in unpredicted directions. However, a combined-year analysis detected longer-term divergent selection on flowering time, a key temporally isolating and adaptative trait, suggesting that selection may reinforce species boundaries despite short-term fluctuations. Finally, we found temporal variation in local adaptation with M. laciniatus locally adapted in low snowpack years, while an extremely high snowpack year contributed to average local maladaptation of M. guttatus.

Keywords: adaptation; divergence; environmental variation; evolution; natural selection; phenology; speciation; temporal variation.

Figure 1.

Temporal and spatial scale of the experimental design. Annual changes in a) snow water content and b) accumulated precipitation during experimental years of low snowpack (2013, 2021, 2022) and comparatively higher snowpack (2019 and 2023) with the c) spatial layout of experimental sites: meadow 1 and 2 (filled circles) and granite 1 and 2 (open circles). Nearby parental populations are shown: M. guttatus YVO (filled square) and M. laciniatus WLF (open triangle). Representative meadow and granite habitats are shown in the insets. Water data are averaged from the California Department of Water Resources data stations GIN, WHW, and TUM, located nearby experimental sites along Tioga Road in Yosemite National Park (cdec.water.ca.gov).

Figure 2.

Seasonal changes in (a,b) survival (%), (c,d) soil moisture (%), (e,f) light intensity (μmol m-2 s-1), and (g,h) soil surface temperature (°F) in each block per site in meadow and granite habitats in low snowpack years (2013, 2021, 2022) and high snowpack years (2019, 2023). Data from 2013 and 2019 are published in Ferris and Willis 2018 and Tataru et al. 2023 and re-analyzed here.

Figure 3.

Visual representation of selection gradients (β) of phenotypic traits measured for hybrids during experimental years in granite (open circles) and meadow (filled circles) habitats. Shown are (a-e) predicted values in each habitat based on native species trait differences and estimated values from (f-j) negative binomial models and (k-o) zero-truncated Poisson models. Significance level is indicated with asterisks and negative graph areas are shaded grey for clarity. The absence of a value during an experimental year indicates that the trait was not present in the best-fit model. All β values are based on seed number, except for values from 2013 which are based on fruit number, as seed number was not recorded. Values from 2013 and 2019 are from Ferris and Willis 2018 and Tataru et al. 2023.

Figure 4.

Patterns of cumulative selection with all years combined in meadow (left; filled circles) and granite (right; open circles) habitat in a zero-truncated Poisson analysis on (a,b) flowering time (days from planting), (c,d) width of first flower (mm), (e,f) leaf lobing index, and (g,h) plant height at flowering (mm to apical meristem). Y-axes show adjusted relative fitness (fruit number), statistically corrected for other variables included in the models, and the x-axes show unstandardized trait values for biological relevance. Data from 2013 and 2019 are from Ferris and Willis 2018 and Tataru et al. 2023 and re-analyzed here.

Figure 5.

Survival to flowering (a-f), fecundity (mean seed number per reproductive individual; g-l), and total fitness (mean seed number per planted seed; m-r) of Mimulus guttatus (filled square, solid line) and M. laciniatus (open triangle, dashed line) in granite and meadow habitats, from each experimental year and averaged across years (f, l, r). Calculations of fecundity and total fitness for 2013 are based on fruit number because seed number was not recorded; 2013 was omitted from averaged fecundity and total fitness. Values from 2013 and 2019 are from Ferris and Willis 2018 and Tataru et al. 2023.