Photoperiod is an environmental cue that many organisms use to regulate seasonal changes in behavior, migration, and reproduction (Kardailsky et al., 1999; Yano et al., 2000; Bradshaw and Holzapfel, 2001; O’Malley and Banks, 2008; Zhang et al., 2008). At temperate and subarctic latitudes, photoperiod, light quality, and temperature are major environmental signals that plants sense in order to synchronize their flowering time with the changing seasons (Bradshaw et al., 2004; Putterill et al., 2004; Andres and Coupland, 2012; Song et al., 2013). Variation in flowering time of plants contributes to local adaptation to different growth conditions, and hence clinal variation in flowering time is believed to be a sign of adaptive evolution (Endler, 1977).

Arabidopsis thaliana is a facultative long-day plant and flowering time in response to photoperiod has been studied extensively (Lee and Amasino, 1995; Onouchi and Coupland, 1998; Simpson and Dean, 2002; El-Assal et al., 2003; Shindo et al., 2005; Balasubramanian et al., 2006; Schwartz et al., 2009; Giakountis et al., 2010). Since it is crucial for populations originating from higher northern latitudes to reproduce at a proper time of the year, one would expect that they are more sensitive to changes in photoperiod than populations from lower latitudes. However, contradictory findings have been reported. Banta et al. (2007) observed that the fitness (individual reproductive success) of accessions from Northern and Southern Europe did not differ under different photoperiods. Moreover, Samis et al. (2008) observed that photoperiodic responses in Eurasian A. thaliana accessions correlated with longitude, and not with latitude. However, latitudinal clines in flowering time (Stinchcombe et al., 2004; Hopkins et al., 2008), hypocotyl length in response to red and far-red light (Stenøien et al., 2002), altitudinal clines in life history traits (Montesinos-Navarro et al., 2010), and flowering time in response to vernalization have been demonstrated (Lewandowska-Sabat et al., 2012).

On a continental-wide scale, essential climatic factors vary with latitude and therefore latitudinal coordinates may be a significant proxy for climatic factors such as temperature and photoperiod. However, on smaller geographic scales, transects are present at the same latitude, e.g., due to climatic gradients related to the distance from the ocean or altitude. In these cases, latitude is not a suitable proxy for ecologically important environmental cues. Furthermore, an earlier study in our group found that altitude, and not latitude, is a better proxy for environmental cues at the northernmost range of distribution of a species, like A. thaliana at higher northern latitudes (Lewandowska-Sabat et al., 2012). Due to the influence of the North Atlantic Current, the temperature fluctuates during the winter season in coastal and low-altitude regions of Norway and the precipitation comes not only in the form of snow but more often in the form of rain, sleet and hail. Moderate winter temperatures, and increased winter and yearly precipitation characterize these locations. On the contrary, in regions at higher altitudes, stable low temperatures during the winter season and moderate winter and yearly precipitation result in a long lasting snow cover. In the same way as environmental cues, varying across broad geographic regions, will impose natural selection on flowering time (Engelmann and Purugganan, 2006; Hall and Willis, 2006; Verhoeven et al., 2008); we expect populations originating from restricted geographical regions, that experience extensive variation in environmental cues but also share a common demographic history, to be under selection and thus be ideal for detecting signs of local adaptation.

The photoperiod that plants experience during the growing season at higher northern latitudes depends not only on latitude of origin but also on the duration of the snow cover, determined by temperature and precipitation during autumn and winter. In the present study, we used lineages derived from high and low altitude local populations of A. thaliana from Norway, which represent the northernmost distribution (59–68°N) of the species. We determined photoperiodic flowering time responses by exposing plants to five different photoperiodic treatments under controlled growth conditions and scored flowering time responses, i.e., days to bolting, days to flowering, rosette leaf numbers at bolting, and flower stem length. The aim of this study was to test for relationships between variation in these flowering time responses and altitude at the site of origin of the populations, and characterize the relationship between climatic factors and variation in photoperiodic sensitivity.

We have demonstrated that photoperiodic responses in Norwegian populations of A. thaliana originating from its northernmost distribution range were associated with altitude and several climatic factors (Table 3 and Supplementary Figure S3). It is therefore highly likely that climatic factors that are changing with altitude, e.g., winter mean temperatures and precipitation, can impose selection pressures on photoperiodic responses in A. thaliana. All of the populations flowered in SD, which is due to the vernalization treatment they were subjected to prior the exposure to five different photoperiodic treatments. The vernalization treatment of the populations were applied in order to fully saturate their vernalization requirements, as the large majority of the populations did not flower without a prolonged cold treatment requiring 6–9 weeks to fully saturate their vernalization requirement (Lewandowska-Sabat et al., 2012). Despite the fact that the vernalization requirement was fully saturated for all populations, flowering was markedly promoted by increasing the photoperiod from 8 to 16 h (i.e., earlier flowering at 16 compared to 8 h; Figure 2), which indicates that the critical photoperiod for flowering in these populations is between 8 and 16 h (Roberts and Summerfield, 1987; Pouteau et al., 2008; Giakountis et al., 2010).

Although photoperiod is highly associated with latitude of origin, many other environmental factors, not associated with latitude also affect flowering time, survival, and seed production. Consequently, differences in temperature, precipitation, and duration of snow cover during the winter season may create different selective pressure on photoperiodic control of flowering. Longer photoperiods enhanced flowering in populations from lower altitudes (Table 3), and these populations demonstrated lower vernalization requirements in our previous study (Lewandowska-Sabat et al., 2012). Thus, it seems likely that natural selection at low altitude locations are creating populations which are more photoperiodic sensitive than at higher altitudes since, at lower altitudes, photoperiod is a more reliable indicator of the changing seasons than temperature. Populations from low altitudes might perceive longer periods with increasing photoperiod as signal for the onset of spring. This gives insurance against damages to floral organs caused by sudden frost in early spring (Tonsor and Scheiner, 2007).

Furthermore, the highest absolute values of the structure correlation coefficients were found for responses of all populations to 21 h photoperiod (Table 3). This indicates the highest association of responses to this photoperiod with altitude, which may indicate that these populations are adapted to photoperiods of approximately 21 h.

In low altitude regions, where winters are relatively mild and short, and with unpredictable temperature fluctuations, it is likely that plants respond rapidly to increasing photoperiod in spring, the crucial signal for these populations to flower at the right time. On the contrary, in high altitude locations, where winters are harsh with long lasting snow cover, rapidly increasing temperatures in the spring are more reliable signals than changing photoperiod, for these populations to flower.

In high altitude populations, the enhanced rosette leaf number responses to photoperiod were observed (Table 3). The greater rosette leaf number responses indicate reduction in the leaf number, i.e., fewer rosette leaves were produced in response to longer photoperiods compared to 8 h. This may suggest that at higher altitudes plants with fewer rosette leaves are favored by natural selection because they likely overwinter under snow cover as rosettes and are adapted to water loss during winter months. These is consistent with previous findings of a genetic correlation between water use efficiency and ecological traits (e.g., rosette leaf number) in A. thaliana (McKay et al., 2003; Stinchcombe et al., 2004). Furthermore, trends of smaller plant size (fewer leaves) at high altitudes were observed in A. thaliana (Luo et al., 2015), and it has been suggested that this may be a mechanism that allow plants to benefit from higher temperatures closer to the soil (Körner, 1999).

We have observed that in low altitude populations, flower stem length responses to longer photoperiods were decreased (Table 3), i.e., these populations developed shorter flower stem in response to longer photoperiods compared to high altitude populations. Flower stem elongation in A. thaliana is regulated by gibberellin, an endogenous regulator of plant growth and development (Richards et al., 2001). It has been shown that treatment with inhibitors of gibberellin biosynthesis suppresses stem elongation under long day conditions in spinach (Zeevaart et al., 1993). Also, in A. thaliana and pea (Pisum sativum), lower day temperature was shown to decrease gibberellin levels through enhanced gibberellin inactivation, as compared to higher temperature (Stavang et al., 2005, 2009). Thus, it is likely that endogenous gibberellin regulation play an important role in inhibition of stem elongation in low altitude populations, preventing them from too rapid growth in unfavorable environmental conditions, such as in cold period of early spring.

In order to reflect the natural growth conditions in Norway, we used a much lower ambient temperature (16°C) than the standard temperature (21°C) used in most of the A. thaliana studies (Hoffmann, 2002). Therefore, interacting effects of photoperiod and lower ambient temperature (Halliday et al., 2003), as being used in our study, have the potential to reveal phenotypic responses involving genetic regulations that so far has gone undetected in A. thaliana. On the other hand, these conditions may still be artificial, especially in the autumn when SDs are often accompanied by very cold weather. Furthermore, in order to test whether the genetic differences in photoperiod sensitivity observed in our study are under natural selection, a comparison of results from common garden experiments at low- versus high altitude locations could provide further evidences.

We have shown that photoperiodic sensitivity of A. thaliana populations originating from its northernmost species boundary is associated with altitude. Populations from lower altitudes were more responsive toward changes in photoperiod than populations from higher altitudes. Therefore, it is highly likely that climatic factors such as length of the winter and duration of snow cover may impose natural selection that creates clinal variation in photoperiodic sensitivity in A. thaliana. These results indicate that higher photoperiodic sensitivity in lower altitude populations might be essential for sensing the onset of spring in regions with relatively mild and unpredictable winter climates as opposed to higher altitude climates with more stable winters. However, this needs to be verified by further studies. Nevertheless, this study contributes to the current knowledge on diversity and dynamics of photoperiodic responses in natural populations of A. thaliana.

AL-S participated in the design of the study, carried out the experiments, collected and analyzed the data, and drafted the manuscript. SF, JO, and OR participated in the design of the study, discussion and interpretation of the results, and manuscript drafting. All authors read and approved the final version of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.