Genetic variation in circadian period follows an elevational gradient in the Rocky Mountains
We detected limited spatial autocorrelation in our data, indicating that in rare cases spatially adjacent populations exhibited similar population averages or ranges of within-population genetic variation. On the whole, we found evidence for a moderate negative association between elevation and population averages of circadian period in our sample. This trend may be an outcome of populations sampled close to and above 3000 m exhibiting averages below 24 h while for instance at ca. 2700 m the full range of among-population genetic variation was evident. Similarly, the range of within-population genetic variation decreased towards higher elevations, which agrees with the results of Salmela et al. (2016) and which may result from more intense selection in marginal habitats with long winters and limited growing seasons and resources.
We estimated home site conditions of the populations using the WorldClim data (Fick and Hijmans 2017) and found evidence for pronounced temperature and precipitation gradients along the 800-m elevational range, with estimates of average annual temperature varying from ca. +4 °C at 2500 m to ca. −1 °C at 3400 m. This pattern denotes that growing seasons begin at different timepoints and photoperiods depending on the elevation: by June, the two highest-elevation sites may have yet to reach a monthly mean of +5 °C. On the other hand, average precipitation was estimated to increase towards higher elevations, while soil pH was lower at higher elevations. Although we found that elevation explained a larger proportion of genetic diversity in circadian period (22 %) than did latitude in A. thaliana (Michael et al. 2003, Rees et al. 2021), variation was still largely residual, suggesting that complex spatial gradients are behind the observed levels of diversity. Contrastingly, in M. laciniatus in the California Sierra Nevada, circadian period was not linearly related to elevation although the WorldClim data indicated a strong association between elevation and temperature conditions (Leinonen et al. 2020). Importantly, the WorldClim data provide model-based estimates for a given location with a 1 × 1 km grid, which may mask substantial micro-environmental heterogeneity in complex mountainous landscapes. When a species is found in highly divergent environments, it is possible that different environmental factors drive local adaptation even across adjacent populations. This has been noted for instance in M. guttatus in California, where coastal and inland populations seem to be influenced by divergent selective agents (Popovic and Lowry 2020). Such differentiation might give rise to large-scale spatial clines in quantitative traits with consequential levels of unexplained variation.
A population-level survey on reproductive phenology in this species has yet to be conducted, but in the study by Anderson and Gezon (2014), which sampled one maternal family of B. stricta per location in Colorado at 2800−3700 m, higher-elevation genotypes tended to flower earlier in common garden. Consequently, selection on photoperiodic responses could be a contributing factor to the diversity of circadian period among populations. Beside the circadian clock, other traits will need to be investigated in the same genetic origins to understand adaptation in this species more comprehensively and to uncover the mechanisms that help maintain significant genetic diversity within populations. A limitation of the WorldClim data is that it does not provide information on the extent of potential among-year variation in temperature conditions in the region. However, weather station data from the Rocky Mountains reveal long-term annual variability for instance in spring temperatures and timing of snowmelt, which in turn are correlated with timing of flowering in nearby plant populations (Anderson et al. 2012). Genetic variation in the circadian clock might be sustained across short spatial distances within populations if selection on circadian period – or on a closely correlated trait − varied in strength or direction in response to such environmental fluctuations. Causes of very spatially localized genetic diversity in nature are often poorly understood (e.g., Delph and Kelly 2014), but in A. thaliana on the Iberian Peninsula, spatially and temporally replicated field experiments provide evidence that fluctuating natural selection could preserve genetic variation for instance in phenology within populations (Exposito-Alonso et al. 2018).