In low-productive tundra systems, both small and large herbivores exert strong control over plant biomass and community composition. They may also modulate ecosystem process rates, including availability and cycling of nutrients and soil decomposition rates. However, evidence from different tundra systems is highly idiosyncratic, with reports of both deceleration and acceleration of ecosystem process rates in response to herbivore presence. Acceleration of process rates contradicts dynamics predicted by prevailing theories of herbivore-plant-soil interactions, which assume primacy of selective foraging and defection/urination in driving ecosystem process rates. In many tundra systems, such assumptions should therefore be re-evaluated. In my thesis, I used empirical (Chapters I and II) and conceptual (Chapters II and IV) approaches to address this research gap. I studied the effects of prominent activities small rodents and large ungulates - non-selective dwarf-shrub decimation and trampling, respectively - which up until now have barely been addressed in tundra. The dearth of research on spatially explicit interlinkages between small rodents, plants and soil may partially result from a lack of cost-efficient methods to estimate in situ rodent abundances. Therefore, I developed a novel method to study small rodent populations (Chapter III) that could augment current methods and allow addressing small rodent-plant-soil dynamics with small spatial grain and large extent.

I provided both observational and experimental evidence that small rodents decimate poorly palatable dwarf-shrubs during their population peaks. An exceptionally strong vole and lemming peak resulted in severe plant decimation across a productivity gradient, as temporarily relaxed top-down control of predators allowed for strong herbivore-plant interactions in the productive tundra-forest ecotone (Chapter I). I showed, for the first time, that such decimation of dominant dwarf-shrubs can promote higher ecosystem process rates, indicated by increased community-level plant N content, increased soil inorganic N content and increased litter decomposition rates (Chapter II).

In Chapter III, I assessed the ability of a near-infrared reflectance spectroscopy (NIRS) -based method to identify individual rodent fecal pellets to species. The model predictions were highly accurate especially with feces exposed to ambient weather. Moreover, I showed that a model based on feces from two regions predicted accurately samples from both regions, indicating feasibility of inter-regional or circumpolar calibrations.

Ungulate trampling may contribute to tundra state-shifts towards accelerated process rates, yet little empirical knowledge and no conceptual synthesis exists of trampling effects on tundra soils. In Chapter IV, I review original papers on trampling effects on tundra soil structure, biota, microclimate and biogeochemistry, and present a conceptual model on ungulate trampling effects on tundra soils. Trampling may drive changes in process rates by e.g. compacting soil, altering soil fauna and microbiota, and by modulating plant-soil feedbacks. In tundra, trampling may accelerate process rates especially by reducing insulating bryophyte cover, which increases soil temperatures and promotes temperature-limited microbial activity and decomposition.

I conclude that Arctic herbivores are likely to alter ecosystem process rates through both trophic and non-trophic activities, the latter being overlooked in prominent theories. Especially non-trophic or non-selective activities may be key to accelerated process rates. A realistic view on herbivore effect on tundra ecosystem function now and under a warming climate needs to build on a better understanding of small rodent and ungulate effects on soil, as well as an explicit consideration of all salient herbivore activities and their multiple roles.