Limitations
The core challenge in this study lies in the quality and comparability of the historical surveys. While we here focused on overarching trends in species richness, these trends mask a substantial turnover in species composition between surveys, which is likely at least partially the result of methodological differences. These methodological differences seem to have been most substantial for the 1983 effort, yet these can be interpreted when considering the following aspects: 1) monitoring differences, 2) weather conditions, and 3) disturbance intensity.
While the vegetation survey of Lewejohann & Lorenzen (in 1983) was conducted during the same month (July) as Sylvén’s surveys (in 1903 and 1913) and our own in 2021, the covered perimeter (20 areas in total) was much larger. As a result, they presumably had the time to survey all areas only once, therefore likely missing ruderal species that appear later in the season. Additionally, their data missed perennial species such as Trifolium pratense L., Trifolium repens L., andVicia cracca L., which were present in 1903 and 1913 and are currently very abundant along the Rallarvägen trail. Even when these ruderal species are not flowering, they are easily identified, suggesting some selectivity in documenting species during their survey. An unexpected discovery, as Lewejohann & Lorenzen set out to survey vegetation changes following the disturbance caused by the building of the E10 road, and thus should have shown particular interest in ruderal species (Lewejohann & Lorenzen 1983; Lembrechts et al. 2014). We did learn that tourism in the area declined substantially in the early 1980s due to a temporarily reduced availability of tourist accommodation during that time. Consequently, we cannot exclude a recovery period from ruderal dominance in the vegetation. Nevertheless, the train stations would have seen continuous disturbance even in those days, and we thus deem it unlikely that these common ruderals mentioned above would have disappeared entirely. We do believe that all survey locations are still comparable as the landmarks in the surveyed villages have remained the same since 1903: the train stations have always been an anchor point for the villages.
Importantly, 1983 was also an anomalous year in terms of weather conditions (supplementary Table 16). Summer temperatures were on average 1°C lower than in 1913, and summer precipitation was exceptionally high (238.8 mm). Snow cover and snowmelt timing are among the most important drivers of structuring subarctic community composition and distribution (Wipf, 2010). Depending on the thickness, snow cover insulates species from the harsh winter conditions as it decouples the soil surface temperature from the air temperature (Niittynen & Luoto, 2017). If the snow layer disappears too early, the species underneath will be exposed to the spring frost. Snowmelt timing determines the growing season length (Niittynen & Luoto, 2017; Wipf, 2010). Moderate rainfall can speed-up the snow-melting process, expediting the onset of the growing season. Heavy rainfall, on the other hand, causes earlier snowmelt, yet then subsequent cold spells could potentially kill off ruderals, or reduce the surviving species’ ability to exploit the whole growing season. Such negative effects are especially disadvantageous for annual species as they do not get the chance to adapt their physiology or morphology in similar ways to perennial plants (Li et al. 2019). While we do not have information on the spring snow cover in 1983, the anomalous precipitation value could suggest that such a scenario of early snowmelt followed by spring forest events might have happened. Annual species are rare in tundra ecosystems (Weidema, 2000), but their distribution in 1983 is rather noteworthy. Seven of 57 (about 13%) of the observed ruderal species in 1983 had an annual life cycle, which was substantially lower than in the other observational years: 54% in 1903, 49% in 1913, and 16% in 2021. Five out of those seven occurred only at Björkliden. Interestingly, this was also the subregion in 1983 that accommodated species with highest average EIV-Ts (Fig. 3). This, in addition to the observed degree of heterogeneity between the subregions in 1983 (supplementary Fig. 4), is probably caused by the timing of observation, which seems of particular importance during that summer.
Another aspect that could have contributed to the low number of ruderal observations in 1983 was the low disturbance intensity during the E10 highway building between 1976 and 1982 (Bäck & Jonasson, 1998). We found that ruderal species distributions along the Rallarvägen were in the most recent survey still clearly correlated to the railroad, rather than the E10 (Fig. 4). In contrast to the railroad construction, the E10 was not built using the Rallarvägen as a transport road, and hence its direct impact on the Rallarvägen vegetation during that time was substantially smaller. Since then, the highway facilitated easier movement in and around the area by cyclists, cars, and trucks, while parking lots now enable hikers to enter the trail at numerous locations, which were less accessible in the past (Bäck & Jonasson, 1998; Frenkel, 1977; Lembrechts et al. 2014). Nevertheless, the resulting influx of ruderals along the Rallarvägen was possibly not yet detectable in 1983. Indeed, ruderal species numbers increased again since then, yet whether this was simply methodological, disturbance, or climate related we do not know. The proportionally rising number of non-native ruderal species is potentially an indication of a delayed response, yet the above discussion should make clear that caution regarding species patterns in 1983 is warranted.