Circadian period manifests genetic diversity among and within natural plant populations
As hypothesized, we found that populations of B. stricta along an 800-m elevational gradient were genetically differentiated in circadian period of leaf movement, with population averages varying between 21.9 h and 24.9 h. Thus, the broad-ranging regional population sample in this study expanded the among-population range of variation by approximately 2 h from that previously described by Salmela et al. (2016) with a smaller elevational gradient. The only other study to estimate among-population genetic differentiation in circadian period found that in Mimulus laciniatus in the California Sierra Nevada, population averages varied by 1.6 h within ca. 100 km and along an elevational range of 1000−2600 m (Leinonen et al. 2020). The current 3-h range among population means accounts for approximately 46 % of the genetic variation documented among 150 A. thaliana accessions sampled across the Northern hemisphere (Michael et al. 2003), and for 68 % of the variation among 191 A. thaliana accessions sampled in Sweden (Rees et al. 2021). These contrasts are notable because we restricted our sampling to a considerably narrower geographic range.
Although a high degree of self-fertilization in species like B. stricta is often expected to erode genetic diversity on a fine spatial scale (e.g., Wright et al. 2013), a significant effect of maternal family nested within population showed that genetic variation in circadian period was present locally, i.e., on a scale of only a few hundred meters. Further, we observed that the magnitude of within-population genetic diversity was variable, with eight populations expressing a genetic range in period length that exceeded four hours. Variance components indicated that among-population genetic differences accounted for a slightly larger proportion of total variation in circadian period than did among-family differences within populations, but overall, the amounts of genetic diversity at these two levels were similar. This finding agrees with Song et al. (2006) who found that up to 40 % of molecular marker diversity segregated within and up to 47 % segregated among populations in B. stricta , but it differs from the study by Salmela et al. (2016) in which within-population genetic diversity in circadian period exceeded that found among populations. The difference most likely arises from a greater number of populations sampled in the current study. The 3.5-h within-population range in genotypic means found by Salmela et al. (2016) is in agreement with the current results. In M. laciniatus in the Sierra Nevada, among-population genetic differences explained ca. 12 % of total variation in circadian period while genetic variation within populations contributed only ca. 2 % (Leinonen et al. 2020). This interspecific difference may stem from the contrasting life histories of the two species: M. laciniatus is a small annual plant with a very rapid life and a limited distribution in the Sierra Nevada, whereas B. stricta is a short-lived perennial found across North America. In the widely distributed M. guttatus in western North America, within-population genetic variation in circadian period was large compared to the magnitude of among-population differentiation (Greenham et al. 2017).
In considering variation in biological rhythms, it is worth reviewing the observed range of natural genetic variation in circadian rhythms from the perspective of the classical circadian resonance theory, which postulates a fitness advantage to endogenous circadian periods whose length matches that of the exogenous environmental cycle (Dodd et al. 2005, Ouyang et al. 1998, Pittendrigh and Minis 1972, Woelfle et al. 2004). We detected a wide range of genetic variation in circadian period in natural populations that have experienced only 24-h diel cycles in their native habitats. This raises the possibility of weak daylength-imposed natural selection on the trait, which in recent animal studies has been proposed to explain the considerable variation detected in period length of locomotor activity within and among three spider species and in mouse lemurs (Hozer et al. 2020, Mah et al. 2019). Indeed, these parallel observations across plant and animal species provide evidence that 24-h exogenous cycles do not select just for endogenous cycling patterns of a matching length. Considering that variation in circadian period is not manifested under 24-h environmental cycles, it is possible that the trait is influenced by selection indirectly via its association with another circadian trait like phase that is relevant in a naturally oscillating environment. Indeed, circadian period and phase are correlated in some (Rees et al. 2021, Rubin et al. 2017) but not all datasets (Michael et al. 2003) inA. thaliana . Genetic variation in the clock could be affected by selection on genetically related phenological traits like flowering time: Salmela et al. (2018) found that maternal families from a single population of B. stricta that had longer circadian periods tended to flower earlier and at a larger size after vernalization, and experimental progeny from controlled crosses of A. thaliana andBrassica rapa accessions indicate that clock traits covary with shoot growth patterns (Rubin et al. 2018) and photosynthetic traits (Edwards et al. 2011, Yarkhunova et al. 2019). To identify environmental factors that may select for natural genetic variation in circadian period in B. stricta , it is necessary to investigate among-population differences in relation to the steep spatial environmental gradients that characterize the study area in the Rocky Mountains.