Introduction
The Arctic is warming at twice the global mean, resulting in profound changes in not only temperature but also precipitation and growing season length (
Mudryk et al. 2019). Biological processes in the Arctic are closely tuned to environmental cues and as such are showing signals of change in response to a changing climate (
Post et al. 2009). This is important because living organisms hold critical control over biogeochemical, energy, and hydrological fluxes, with huge potential to further exacerbate climate change (
Wookey et al. 2009). Plant communities across the Arctic have shown particularly striking changes in response to warming as they grow taller (
Bjorkman et al. 2018), increase in cover, and undergo shifts in dominance, with mosses often in decline and deciduous shrubs in ascendance (
Elmendorf et al. 2012b). Natural observations of change are supported by experimental evidence showing that there are clear winners and losers in the plant community as the climate continues to change (
Elmendorf et al. 2012a).
The direct effects of warming on Arctic plant community composition and growth have been well studied through a circumpolar network of open-top chamber (OTC) experiments (
Elmendorf et al. 2012a). These generally show that plant growth increases with warming and that deciduous shrubs increase in dominance, but also that responses are mediated by site conditions such as local climate and soil moisture (
Elmendorf et al. 2012a). Wider observation networks are detecting “greening” signals with increases in height and cover at the plot level (
Bjorkman et al. 2018) and increases in the normalized difference vegetation index (NDVI) at satellite levels (
Epstein et al. 2012). Observations of the expansion of deciduous shrub cover are consistent with these trends (
Myers-Smith et al. 2011). One of the key findings is that certain groups in the community, such as mosses, decrease in cover as the community responds to warming, while the response of other groups, such as sedges, is mixed (
Elmendorf et al. 2012a). It is important to understand how all constituents of the plant community will change in the future because they all contribute significantly to ecosystem processes such as primary productivity, reflectance, and phenology, among others (
Myers-Smith et al. 2019).
Climate change in the Arctic is multifaceted and will affect aspects of plant performance in different ways (
Post et al. 2009;
Box et al. 2019). For example, summer growing seasons are extending in the Arctic owing to reductions in snow cover duration (SCD) (
Box et al. 2019). Model projections indicate that SCD over much of the Arctic will decline by about 10%–30% by the end of this century as a consequence of delayed onset of snow cover as well as earlier snowmelt (
Brown et al. 2017). The projected decrease in SCD implies that the potential growing season should lengthen, as found by
Park et al. (2016), who used NDVI to analyze changes in growing season length in boreal and Arctic vegetation. Broadly speaking, plant phenology in the Arctic has been shown to be sensitive to abiotic conditions (
Prevéy et al. 2017;
Assmann et al. 2019). At the beginning of the growing season, earlier snowmelt should result in earlier green-up, as abundant sunshine and the disappearance of snow produces good growing conditions. Many studies have documented the importance of snowmelt timing for controlling the phenology of Arctic plants with earlier snowmelt, which usually results in earlier onset of growth (
Høye et al. 2007;
Bjorkman et al. 2015;
Khorsand Rosa et al. 2015;
Semenchuk et al. 2016;
May et al. 2020). Once the growing season is underway, it is less clear whether higher mean temperatures will affect plant phenology, in part because of interactions with snowmelt timing (
Oberbauer et al. 2013). Geographical patterns in phenology further complicate the response of Arctic plants to climate change. Across the Arctic, phenology of plants from more northern sites have exhibited greater sensitivity to warming temperatures than plants from sites at more southern latitudes (
Prevéy et al. 2017).
Increasing temperatures in autumn (
Box et al. 2019) may offer an opportunity to plant communities to grow for longer, but it is difficult to forecast the effect of mid- and late-season growing conditions on phenology in the autumn. If autumn temperatures increase, it is not clear that Arctic plants will respond by extending their growing season (
Parker et al. 2017). Many species start to turn yellow in August, when temperatures are still warm (
Shaver and Laundre 1997). This may be because, in the Arctic, harsh winter conditions can appear suddenly, which could result in the loss of valuable resources through frost damage to live aboveground biomass that has not fully senesced. Some functional groups, notably some graminoids, may be able to delay senescence in response to warming conditions, while other functional groups may have fixed leaf life spans that are correlated with average growing season lengths (
Oberbauer et al. 2013). Manipulation of the timing of green-up by removing snow or adding it with snow fences has shown that the length of phenological stages such as growth, flowering, or seed setting remained invariant even though the dates of start-up varied greatly (
Khorsand Rosa et al. 2015;
Semenchuk et al. 2016).
Semenchuk et al. (2016) concluded that a range of herbaceous and shrub species in their study are periodic, meaning that the duration of phenological periods is genetically fixed. By extension, therefore, even if the end-of-season environment is suitable for continued growth, tundra plants may senesce early if their green-up was early.
While many studies have focused on variation at the species and community levels of organization, few studies have looked at intraspecific variation in the phenology of tundra plants. Since most Arctic plants have widespread distributions, local adaptations are likely to be important for many species (
Linhart and Grant 1996). Local adaptation is widespread in plant populations, especially those consisting of many individuals covering a wide geographic range (
Leimu and Fischer 2008;
Hereford et al. 2009).
Wagner and Simons (2009) reported differences between the phenology of the Arctic and alpine populations of the annual
Koenigia islandica, with the Arctic population flowering earlier than the alpine population.
Bjorkman et al. (2017) reported that southern populations of the Arctic plants
Oxyria digyna and
Papaver radicatum were slower to leaf out and to initiate senescence than northern (local) populations. Likewise,
Parker et al. (2017) showed that senescence of
Eriophorum vaginatum grown in a common garden occurs later for populations from the southern portions of a latitudinal gradient in the Alaskan Arctic. Although growth rates were the same, the southern populations were able to accumulate more biomass because of the longer growing season (
Parker et al. 2017). Thus, it is important to base models of phenology on not only a generalized phenotype but also the variation within species across their range where local dynamics may vary, although the assemblage remains the same.
Many Arctic plant species are distributed along the latitudinal gradient from the Low to the High Arctic, which provides ample scope for locally adapted populations or ecotypes. Strong adaptation to local climates may render Arctic plants vulnerable to rapid climate change in their locales if they are not able to respond quickly enough (
McGraw et al. 2015). The degree of phenotypic plasticity of ecotypes of Arctic plants may determine their potential to take advantage of, or survive, warmer conditions.
Eriophorum vaginatum is a foundational species of moist acidic tundra, meaning that it strongly dictates the system’s physical structure as well as its process rates (
Chapin and Shaver 1985).
Eriophorum vaginatum demonstrates clear ecotypic differentiation in phenotypes (
Shaver et al. 1986;
Fetcher and Shaver 1990) and gene expression (
Mohl et al. 2020) across its South–North distribution in Alaska, which reflects a wide range of growing season conditions.
McGraw et al. (2015) showed that the optimal environment for tussock survival and tiller population growth in
E. vaginatum had shifted northwards, meaning that this important species may suffer from “adaptational lag” and thus may not keep pace with current rates of climate change. To address this lag in performance, local populations may need to be supplemented by gene flow from the south (
McGraw et al. 2015). Performance of the northern ecotypes of
E. vaginatum is less flexible than that of the southern ecotypes in both net ecosystem exchange (
Curasi et al. 2019) and leaf growth (
Fetcher and Shaver 1990). But as previously stated, changes in growing season length offer plants new opportunities to grow for longer and remain competitive in their environment.
Here we investigate the role of genetic background and environmental conditions as they affect the phenology of
E. vaginatum growing in a reciprocal transplant experiment in northern Alaska, USA. We used this system to ask the following questions:
1.
Can the phenology of E. vaginatum ecotypes match growing conditions when transplanted into warmer ecosystems with longer growing seasons?
2.
Do southern populations retain their growth patterns when transplanted north?
3.
Do local ecotypes increase growth and growing season length when experimentally warmed in situ?
4.
Does E. vaginatum exhibit a fixed periodicity in its phenology, that is, if it starts growing early will it senesce early?
Results
Across all populations, tillers of
E. vaginatum initiated growth earlier at Coldfoot than at the other two sites (
Table 1;
Fig. 2) but there was no significant difference between populations across all gardens (
P = 0.195,
Table 1). However, the CF population responded to differences in site growing conditions more than the other populations, resulting in a significant interaction between populations and site (
Table 1). The CF population started to senesce later than the northern ecotypes as represented by the TL and SG populations at all the sites (
Fig. 2). But the onset of growth at Toolik Lake and Sagwon was significantly delayed after snowmelt in 2016, during which there were low temperatures in early June (
Table 1;
Supplementary Fig. S12). Thus, early June temperatures appeared to exert some control on the initiation of growth.
The southern ecotype had significantly longer leaves than did the northern ecotypes (
P < 0.001,
Table 1), although this difference was less pronounced at Toolik Lake than at the other two sites (
Table 1;
Supplementary Fig. S32). Warming with OTCs had no effect on leaf length (
Table 1), but on average, over the 2 years, it did result in a significantly (
P < 0.05) longer growing season, defined by the number of days between S1 and A1 (
Table 1;
Fig. 2). Warming did not affect spring phenology or autumn phenology in a statistically detectable way, but the combined effects may have increased the overall season length slightly.
The effect of transplanting was analyzed by comparing tussocks that were transplanted into their “home” site with nontransplanted “control” tussocks. Across all sites, transplanting did not affect the onset of growth (metric S1;
Supplementary Figs. S4 and
S5 and
Table S32), but on average made the onset of senescence marginally earlier (metric A1,
P = 0.066;
Supplementary Figs. S4 and
S5 and
Table S32); however, this pattern was not consistent and depended on population. Growing season length was not affected by transplanting (metric S1A1;
Supplementary Figs. S4 and
S5 and
Table S32), but transplanting did significantly reduce maximum green length compared with nontransplanted controls (
P = 0.01;
Supplementary Figs. S4 and
S5 and
Table S32).
There was no significant relationship between the date of growth onset and the date of senescence onset across populations and no interaction between onset of growth and population (
Fig. 3;
Supplementary Table S42). CF populations consistently senesced later than the others, but this was unrelated to the onset of growth. Over the whole growing season, the actual growing season of leaves (metric S1A1) of all populations was positively affected by potential growing length, but the CF tussocks responded particularly strongly. This resulted in a statistically significant effect of potential growing season, population origin (CF was highest on average), and an interaction between the two (
Fig. 4;
Supplementary Table S42). The onset of growth in spring was positively related to the day of snowmelt across all populations, but the CF population was particularly responsive, with initiation of growth closely tracking the loss of snow at any given site (
Fig. 5;
Supplementary Table S42). In autumn, none of the populations in either year were responsive to differences in late-season environmental conditions (in this case, temperature in the first half of August). Instead, the populations maintained a significant difference in senescence timing regardless of the garden they were present in, with CF senescing particularly late (
Fig. 6;
Supplementary Table S42). There were significant differences between years, with most tussocks senescing later in 2017 than in 2016.