Relationships between topography, vegetation, soil and
microclimate
Our study sites covered a range of abiotic and biotic environmental
conditions (Figure 2c), indicating that the stratified random sampling
approach overall captured environmental gradients in the area. Yet,
relationships between topographic, vegetation and soil variables and
microclimate only partly reflected our expectations (Figure 5). As
expected, soil temperature decreased with higher elevation and
underneath higher vegetation cover, corroborating trends previously
found in Fennoscandian tundra (Kemppinen et al. 2021). In contrast,
solar radiation predicted trends opposite to expected relationships,
particularly lower rather than higher soil temperature but higher rather
than lower soil moisture, possibly due to its close association with
slope values. Local topography as quantified by TPI did not explain
variation in any vegetation or soil parameter, perhaps because variation
in TPI was more constrained than in other variables (Figure 2c), thus
not well representing differences in snow accumulation or wind exposure
across the landscape. Also, neither TWI nor soil texture explained
variation in growing-season soil moisture among plots. Here, our
gradient most likely did not cover the entire spectrum of conditions
present in the area, as very wet locations were not covered (Figure 2c),
probably as a result of the dry summer in 2019 and of standing water
areas being masked out during the stratification process. The example
highlights the difficult issue of obtaining reliable soil moisture
predictions in tundra ecosystems (Kemppinen et al. 2018). In addition,
implicit uncertainty in gridded topographic variables might have added
further noise to our dataset.