Figure 6. General Linear Mixed Model fits for the relations between herbaceous and woody cover in fence (red colour) and grazed (cyan) plots across the Fennoscandian tundra. Significant relationships are represented with solid lines and confidence intervals are given in grey. The model consists of herbaceous abundance (measured as hits.100 pins-1) as a response, woody abundance (as hits.100 pins-1) and treatment (fence vs. grazed plots) as predictors and location and treatment block as random factors Model results are: r2 marginal=0.05, r2 conditional=0.86, fence b=-0.37, fence p-value <0.001, grazed b=-0.13 and grazed p-value=0.002.

Discussion

Tyler et al., (2021) developed an IBR that expresses the number of non-plant species that utilize a plant species as a source of food and energy for all Swedish vascular plant species. By combining these so-called biodiversity relevance values with vegetation data from a network of fenced and grazed plots scattered across sites with low and high primary productivity in the Fennoscandian tundra, we revealed that reindeer decreased IBR for total vegetation and a subset of taxonomic groups, despite having an increasing effect on plant diversity as a response to the interaction between grazing and herbaceous plant abundance. The lack of an age effect indicates that the vegetation changes in the fenced plots have already approached an asymptote after two decades and further changes will depend on grazing-intolerant species colonizing the exclosures (Olofsson et al. 2013).
 

Biodiversity relevance

As we predicted, vegetation changes in response to grazing led to shifts in the biodiversity associated with the vegetation. When considering all vegetation, the IBR that is computed as a log scale was 67.4 points higher in the fenced compared to the grazed plots, probably because taller and more complex vegetation in fenced plots, as revealed in other studies, provides a greater variety of niches for different organisms to occupy (Kaarlejärvi et al. 2017). The species associated with fenced plots were occurring at higher abundance and had a high IBR value compared to the grazed plots (Appendix A.2-3). When considering narrower vegetation groups, we also found that IBR was higher both for herbs (10.9 points) and non-fruit-bearing shrubs (68.8 points) in the fenced compared to the grazed plots and this mechanism is driven by plant abundance and plant species richness. The differences among treatments are substantial as the IBR is based on a logarithmic scale (Tyler et al. 2021) and includes insects, fungi and some bacteria, which are responsible for important ecosystem processes (Tilman et al. 1997). Therefore, the ecological food web community in a situation without herbivory is not only larger and more diverse but may also perform a distinct spectrum of ecosystem functions (Naeem et al. 1994; Naeem and Wright 2003), which in the long-term can thread the conservation of the tundra by leading to fundamental changes in how this ecosystem is structured and functions.
A series of empirical studies that targeted the effect of herbivory on different taxa in the Fennoscandian Arctic corroborate our findings. Lightly grazed areas for example were associated with denser shrub vegetation and larger microbial communities compared to heavily grazed areas (Stark et al. 2015). Whereas, the abundance of soil fungi, lichens and mosses was higher in fenced plots compared to grazed plots, plausibly because herbivory reduced woody vegetation (Olofsson et al. 2010; Santalahti et al. 2018). Regarding invertebrate herbivores, previous findings are inconsistent. One study conducted reported higher densities of common insects (leaf beetles and gall-inducing sawflies) inside fenced plots compared to grazed plots (den Herder et al. 2004), while a second study stated higher abundance of Curculionid beetles and lower abundance of Carabid beetles inside fenced plots compared to grazed plots (Suominen et al. 2003), suggesting that herbivory effects are species-specific and plausibly mediated by external factors like primary productivity and temperature. Thus, our IBR values serve as tools to identify overall biodiversity patterns.
Biodiversity changes registered in our grazing experiments are comparable to changes occurring in warming experiments that partially exclude grazing. Plant primary productivity and canopy cover in warming experiments, also known as Open Top Chambers (OTCs), are higher than in the controls. Small and tall shrubs increased in abundance with experimental warming, while mosses and lichens decreased (Arft et al. 1999; Elmendorf et al. 2012). These shifts in composition led to an overall decrease in plant species diversity and evenness (Walker et al. 2006). OTCs have also shown that the composition of the invertebrate community responds to warming treatments in Svalbard and Abisko, with a higher abundance of chewing, cryptostigmatic and predatory mites in warm treatments and a higher number of springtails, aphids, hymenopterous parasitoids and weevils in control plots (Sjursen et al. 2005; Dollery et al. 2006). A warming experiment conducted at the lower mountain birch forest in Kevo led to increased herbivory on birch saplings (Silfver et al. 2020). Yet the occurrence of these changes is not consistent across study sites mainly because of context dependency and the duration of the experiments (Arft et al. 1999). Suggesting that some regions across the circumpolar are more vulnerable to climate warming. Hence, the overall high scores of IBR in our network of fenced plots denote shifts in biodiversity composition and, with less confidence, an increase in the absolute number of species.
Herbivory had little or no effect on the lower-end of biodiversity when maximum species redundancy is assumed and also at the higher-end of biodiversity when no species redundancy is assumed. The true biodiversity value for each of the treatments grouped by vegetation type probably falls somewhere along this range, yet, it is impossible to predict where and it would be unreasonable to assume that the mean would be a good representation since this is a logarithmic scale and we have no insight into the true degree of species redundancy. We can conclude, when combining this result with the evidence that herbivory reduces IBR, that reindeer shape vegetation composition by reducing the abundance of palatable species and not by extirpating plant species.
 
Vegetation richness and diversity
We predicted and found that across the latitudinal gradient (61°32'49" N-70°25'30" N), excluding reindeer did not affect overall plant richness and diversity. Albeit the lack of overall differences in plant diversity, there were clear differences in the composition of the plant community with more shrubs and lichens in the fenced plots. This became evident in this study when assessing the associations of richness and diversity to herbaceous cover between fenced and grazed plots. Plant richness and diversity were positively associated with herbaceous cover in grazed plots; whereas in fenced plots, plant richness presented a less steep increase and plant diversity peaked at intermediate herbaceous cover. Grazing is therefore a key mechanism that not only holds back climate-driven shrub dominance (Olofsson et al. 2001, 2009) but also promotes a wider plant diversity by favouring the establishment of other plant species that are commonly associated with grazed steppes. Alternatively, it can also be explained by the composition of the herbaceous group. For example, if the herbaceous group compared to other vegetation groups has more species that share a set of grazing tolerant traits (Barton and Koricheva 2010), this would ultimately facilitate a higher richness and diversity. The absence of a herbivory effect in overall plant diversity patterns is explained by a low net density of herbivores and variation in primary productivity across our research sites.

 

Variation in plant co-existence

We predicted and found that herbaceous cover was negatively associated with woody cover, particularly inside the fenced plots. Woody plants overall limit the establishment, development and reproduction of herbaceous plants by shading them (Eskelinen 2008). In contrast, herbivory modulates the strength of the competition between these two groups by directly limiting the expansion of shrubs by consuming their leaves and by trampling and indirectly, by alleviating light limitation for ground herbaceous cover (Olofsson et al. 2001; Lindén et al. 2021). This result highlights the key function large herbivores play in maintaining the open structure and the vegetation composition of the Arctic tundra in the face of climate change.
 

Outlook

Biodiversity indicators are convenient tools that can rapidly assess the overall biological state of less complex ecosystems like the tundra. This approach still needs to be tested in intricate ecosystems like in the tropics that host a vast number of interacting species. Indicators overall provide insights that empirical data cannot provide since collecting data across large spatial and temporal dimensions is logistically challenging and requires the participation of staff with extensive taxonomic knowledge to conduct field inventories. Therefore, a high investment of resources is required for conducting these methodological procedures.
The downside of using IBRs is that the resolution of the results is lost because a single index value does not transmit the species identification or the exact number of organisms. More importantly, it assumes that the biodiversity values are constant within species and not context-dependent (Tyler et al. 2021). IBR estimates are likely to correlate with overall biodiversity by summing the number of unique plant species interacting with other non-plant species, but IBR estimates inherently hold a high degree of species redundancy since different plant species, in particular, if closely related, may be interacting with the same non-plant species (and vice versa). Even if not directly translatable to species richness, the number of unique species-to-species interactions may be highly relevant as a measure of the complexity of the food web or ecosystem which, in turn, may indicate its stability and ability to cope with climate change or other environmental stressors.
 
Conclusion
We found in this study evidence that reindeer herbivory in the Fennoscandian promotes the conservation of the tundra’s biodiversity and ecosystem functions by favouring small-size (i.e., herbaceous) over the climate-driven tall-size (i.e., woody) plants and by doing so, they increase plant diversity and decrease the overall number of taxa interacting with vegetation based on an IBR analysis. Interacting diversity was quantified by binding the abundance of plants in our network of plots (i.e., fence and grazed) with a biodiversity index developed for vascular plants. This approach allowed us to quantify in a tangible way the breadth and magnitude of trophic interactions led by herbivores. We reckon that most advancement in this line of research, including complex ecosystems, will be done by combining biodiversity indices with surveys performed in a range of habitats that include keystone species belonging to different trophic levels, enabling a holistic understanding of the important role that herbivores have on the structure, composition and functioning of ecosystems.