Spatial patterns along the Rallarvägen
As expected, ruderal species were concentrated around points of introduction with continuous disturbance, with a progressive decline in ruderal species richness with increasing distance to these introductory points (Fig. 4 and 5). Disturbance was the strongest driver – especially for the total ruderal community – since the number of ruderals decreased significantly with distance to the railroad at average (9.8 °C) and below average (9.4 °C) MSST. This supports other studies that illustrate the key role of disturbance for new plant species introductions (Alexander et al. 2010; Lembrechts et al. 2016b). However, our findings also showed a climatic response on ruderal species distributions as the predicted number of both the total ruderals and non-native ruderals increased with distance to the railroad in transects where summer soil temperatures (MSST) reached above average values (10.3 °C; Fig. 4). This could indicate a potential role for climate change: if climate warms, the current climatic limitations might decrease, allowing for non-natives to expand further away from disturbed sites (Lembrechts et al. 2018).
Interestingly, average ruderal species abundances were negatively correlated with the total number of ruderal species per transect (Fig. 5), revealing a gradient from sites that accommodated relatively few and predominantly common ruderal species to sites that accommodated relatively many and often rare ruderals. Transects with train stations, such as Björkliden, Abisko Östra and Riksgränsen, experience continuous disturbance and appear to be the sites where rare ruderals reside and are thus (re)introduced (Brandes, 2002). From here they must spread out through local human-mediated dispersal (e.g., hiking) or through self-dispersal. In less disturbed transects along the Rallarvägen, only fewer and more common ruderal species still reside. Those sites were sometimes not easily accessible, lay further away from train stations, and were places where significant disturbance (outside of occasional hiking) only happened once (i.e., during railroad building in the early 20th century) (Rendeková et al. 2019). Exact drivers for this relationship are difficult to determine from observational data only, but we can assume a link with the degree of disturbance intensity. Distance to the E10 highway, railroad, and train station, and hiking-intensity were all interconnected and sum-up to a certain degree of disturbance intensity that facilitates species dispersal (Kowarik, 2003; Pauchard et al. 2009).
We propose that the correlation between human impact and ruderal species distributions demonstrates Horizontal Directional Ecological Filtering (HDEF; Fig. 5). The term Directional Ecological Filtering (DEF) was first coined by Alexander et al. (2010) for elevational (climatic) gradients, but here we demonstrate that the concept is applicable to a horizontal anthropogenic disturbance intensity gradient as well. The original DEF hypothesis states that non-native species migrations happen from anthropogenically disturbed sites in the lowlands to higher elevations in the mountains, and that their richness declines with elevation but their elevational range increases with their maximum elevation (Alexander et al. 2010). Non-native species thus progressively drop out with increasing elevation. Unlike in the DEF, where climate harshness is considered the most likely filter, native and non-native ruderal species originating at the train stations were here thus progressively filtered out with increasing distance to these introductory points as the degree of disturbance intensity declined, showing that such directional filters can still be strong when climatic gradients are minimal. This led to the conclusion that gradients of disturbance intensity coinciding with the elevation gradient should not be ruled out as a critical driver of the DEF either (Pauchard et al. 2009).