Herbivory mediated canopy-to-soil fluxes and their controls
Our results showed that BIH constitute a minor fraction (~1.6 %) of the annual soil input of organic matter from the birch canopies, in line with our hypothesis (H1). These rates are similar to the estimates of insect herbivory in dwarf shrub tundra (Barrio et al. 2017), but lower than ecosystems at lower latitudes (Kozlov et al. 2015, Galmán et al. 2018). Despite the relatively short recurrence interval of insect population peaks of about 10 years (Jepsen et al. 2008), major outbreaks only return to the same site every ~50-100 year (Tenow and Bylund 2000). Thus, the low contributions to fluxes at low insect densities may cumulatively recycle similar or larger amounts of elements than the major outbreaks, where most of the leaf area is lost, but it is much less disruptive to the ecosystem. Further, as background densities may increase with future climate warming (Kozlov et al. 2015, Galmán et al. 2018), BIH may increase in overall importance for nutrient cycling. The variation in insect herbivory level across our sites was considerable, with leaf area loss at some sites of up to 7%. When the nutrient resorption during leaf senescence (~55% N, ~30% P) was accounted for, the average herbivore-mediated fraction of the annual soil input was slightly higher than for C (~3.5% N and ~2.5% P), yet still a relatively small contribution. Nonetheless, at the sites with the highest herbivory levels, this corresponded to ~14% and ~9% of annual N and P fluxes, respectively. These estimates of herbivory mediated fluxes may be in the low end, as our estimated mean N resorption efficiency of 55% was lower than what has previously been found in these forests (~70 %, Nordell and Karlsson 1995). This suggests an even larger gap between the substrate quality of litter and insect deposits. The higher substrate quality (availability to soil decomposer microbes) of insect deposits compared to litter (Kristensen et al. 2018) may trigger increased soil N and C-turnover early in the season. Yet, these belowground responses to BIH are most likely within the ranges of what the soil biota can take up and keep within the ecosystem, and so are unlikely to cause nutrient losses, in contrast to outbreaks which can cause substantial nutrient losses from the soil (Lovett et al. 2002, Kristensen et al. 2020).
The positive relationship between herbivory and leaf N-concentration in the green leaves was in line with our expectations (H2), suggesting that leaves of higher nutritive quality suffer greater herbivore damage, with no apparent effect of foliar condensed tannin concentration. Moreover, the strong negative relationship between leaf N content and condensed tannins (linear regression w. CT:C log-transformed, p<0.001) corroborates the theory of Haukioja (2003) predicting that it is only beneficial for the plants to increase the content of defence compounds when the nutritive quality of the leaves is low. Only under such conditions will the insects eat large enough amounts of the leaves to encounter a negative effect from the tannins. We note that our measurements of leaf condensed tannin content were quite low compared to some other literature observations from similar trees and ecosystems (Paaso et al. 2017; Stark et al. 2007). This may be partly due to large variation between sampling years, genotypes, and responses to other regulators than herbivory, such as soil nutrient content and photochemical conditions (Madritch and Lindroth 2015, Rubert-Nason et al. 2015).
The significant increase in insect herbivory with relative position above the valley bottom, i.e. towards the treeline, was in line with our expectations (H3) derived from the patterns found for outbreaks (Hagen et al. 2007), and previous indications from studies of non-outbreak conditions in subarctic Finland (Virtanen and Neuvonen 1999). Yet, this is, to our knowledge, the first time a significant effect has been shown for background herbivore densities. Nonethless, the mechanisms remain unclear. The proposed drivers of the patterns – higher parasitism in valley bottoms during the summer and eggs being killed by extremely cold air masses creeping into valley bottoms during the winter (e.g. Virtanen and Neuvonen 1999) – are confounded by the consistent increase in leaf N content with local elevation. Yet, if foliar chemistry was the paramount driver of BIH across scales, the N concentration should show a neutral-negative trend with elevation at the regional scale. In contrast, the leaf N content also increase with elevation at the regional scale, although the relationship is weaker than at the local scale as should be expected (Read et al. 2014). Thus, abiotic drivers, e.g. climate, are likely moderating the biotic relationships at the regional scale.
The local increase in herbivory with elevation found here contrasts with the more general decrease with elevation along individual transects across the World’s woody species (Galmán et al. 2018). This suggests that the proposed responsible mechanisms in subarctic mountain birch landscapes may only apply to high-latitude systems. Thus, our study serves as a great example of how global generalisations may lead to wrong assumptions in certain ecosystems highly important for the global climate. Interestingly, we found no significant relationship between BIH and the most commonly studied climate variables, temperature and humidity, nor with a commonly used climate proxy, absolute elevation (Table 2). Instead, solar radiation was a marginally significant positive predictor of BIH, which may suggest that this is in fact a better integrator of relevant abiotic conditions, when working in mountainous landscapes. This should be particularly relevant to consider at high latitudes, where sun-angles are low, yielding a particularly marked difference in solar energy input between north- and south-facing slopes in steep-sided valleys (~50 % higher annual solar radiation on south-facing slopes in our dataset, e.g. N8 compared to N7, Figure 1). Hence, our data point to the potential value of including solar radiation as an abiotic predictor of ecological characteristics in other landscapes with complex topography.