Introduction
Herbivores have important effects on elemental cycling in terrestrial ecosystems (Bardgett and Wardle 2003, Schmitz et al. 2018). Most attention has been given to vertebrates, but insect herbivores can exert a similar or even stronger control on ecosystem functioning than mammals (Risch et al. 2018, Kristensen et al. 2020), particularly in forest ecosystems, where their impact is likely to intensify substantially with global change (e.g. Logan et al. 2003).
Studies on insect herbivory in high-latitude forests have hitherto emphasised the importance of outbreaks (e.g. Jepsen et al. 2013, Sandén et al. 2020), especially as some species are increasing their ranges into new areas with climate warming (Jepsen et al. 2008, 2011). However, non-outbreak, low-intensity rates of insect herbivory, termed background insect herbivory (BIH), has attracted increasing attention in recent years (Kozlov and Zvereva 2017). Although BIH at high latitudes is generally low - 1-2 % of the leaf area (Barrio et al. 2017) compared to the global average of ~8% (Kozlov et al. 2015) - the rate will likely increase at high latitudes with global warming (Kozlov et al. 2015), despite the observation that insect predation and parasitism also increase with temperature (Virtanen and Neuvonen 1999, Roslin et al. 2017). It has been argued that BIH, despite the small annual contribution, may be more important for long-term ecosystem functioning in a wide range of globally important ecosystems, including tropical (Metcalfe et al. 2014), boreal (Zvereva et al. 2012, Metcalfe et al. 2016), and arctic systems (Barrio et al. 2017).
The fact that most of the global terrestrial soil C is stored at high latitudes (Hugelius et al. 2014), where global warming is most pronounced (ACIA 2004), makes it particularly important to understand perturbations in this region. While low productivity ecosystems dominate at high latitudes (Higgins et al. 2016), subarctic birch forests constitute a relatively productive ecosystem (Sjögersten and Wookey 2009). Therefore, perturbations in birch forests may exert disproportionately large impacts on regional biogeochemical cycling. The end consequences of herbivory, by primarily the geometrid mothsEpirrita autumnata and Operophtera brumata, have been demonstrated in terms of plant traits (Haukioja 2003; Karlsson et al. 2004), plant (Jepsen et al. 2013, Sandén et al. 2020) and soil community composition (Saravesi et al. 2015, Parker et al. 2016, Kristensen et al. 2018), soil nutrient and carbon (C) turnover (Kaukonen et al. 2013, Parker et al. 2016, Kristensen et al. 2018, Sandén et al. 2020), and photosynthetic C-fixation (Heliasz et al. 2011, Bjerke et al. 2014), but quantification of the key mechanism driving these changes - canopy to soil fluxes of C, N and P through insect deposits - is still lacking. Elements channelled through insects are not a novel element input to the ecosystem, but rather an alternative pathway of transferring high quality organic matter from the canopy to the soil in the early growing season rather than the usual transfer of senesced litter by the end of the season. Most studies indicate that N is the main plant growth-limiting nutrient in most forest systems (e.g. LeBauer and Treseder 2008), but an increasing body of literature suggests that P is co-limiting plant growth (Vitousek et al. 2010, Sundqvist et al. 2014). Further, nutrient limitation will become even more widespread in the future due to warming and CO2-fertilisation, which may weaken the terrestrial ecosystem C-sink (Fisher et al. 2012, Wieder et al. 2015). Therefore, it is relevant to assess the amounts of both N and P channelled through labile insect deposits and recalcitrant litter respectively. Insect deposits contain much larger amounts of nutrients compared to senesced litter (e.g. Kristensen et al. 2018) because insects feed on green leaves before the highly conservative Subarctic birches resorb up to ~70 % of their nutrients during senescence (Nordell and Karlsson 1995). Thus, the balance of elemental transfer between the litter and insect pathways regulate the timing and quality of soil substrate inputs, which can in turn lead to both increased and reduced soil C and nutrient turnover (Parker et al. 2016, Kristensen et al. 2018, Sandén et al. 2020).
N is also the growth-limiting nutrient for the moths in our study system (Metcalfe et al. 2019). The foliar content of N is therefore expected to be an important driver of moth success, hence BIH level. In order to conserve nutrients from herbivores, host plants may increase the level of foliar chemical defence compounds (Haukioja 2005, Fürstenberg-Hägg et al. 2013). For example, the leaf content of bioactive specialised compounds, such as condensed tannins (CT), has been found to increase in leaves subjected to herbivory (Fürstenberg-Hägg et al. 2013). Yet, the relationship between herbivory and plant defence compounds is not simple in natural systems. In fact, geometrid moth species in subarctic birch forests are rather tolerant to the birch chemical defence compounds (Haukioja 2003, 2005). Therefore, it is only relevant for the birch to pursue this defence strategy when the growth-limiting nutrient level in the leaves is so low that the insects have to eat large amounts of foliage to compensate for low nutrient concentrations (Haukioja 2003). Thus, a negative relationship between the foliar content of the moth growth-limiting nutrients and foliar defence compounds should be expected at ecosystem scale.
Apart from the links to foliar chemistry, BIH also varies with climate, with an expected increase with warmer temperatures in mid-high latitude systems (Kozlov et al. 2015, Barrio et al. 2017, Galmán et al. 2018). Nonetheless, moth outbreaks most often appear close to the treeline in birch forests across subarctic Scandinavia rather than in valley bottoms, probably due to the higher likelihood of winter temperatures below their egg survival limit (< -35-37 °C) along the valley bottom due to thermal inversion of air masses during winter (Ruohomäki et al. 1997, Hagen et al. 2007), and/or higher parasitism during the summer (Virtanen and Neuvonen 1999). Yet, such decrease in top-down controls on herbivory at higher local elevation may be confounded by decreasing bottom-up controls. For example, the food quality (leaf N content) is also expected to increase with elevation (Körner 1989, Read et al. 2014). Nonetheless, no systematic increase in BIH with the local elevation above the valley bottom has been found previously (Virtanen and Neuvonen 1999), and such an increase would oppose the overall elevational decrease in herbivory at the global scale (Galmán et al. 2018).
In this study, we quantified BIH on the mountain birch (Betula pubescens ssp. czerepanovii (Orlova) Hämet-Ahti) along 9 elevation gradients in Subarctic birch forests spanning a considerable portion of the environmental variation in the Fennoscandian mountain birch forests. We estimated the canopy-to-soil fluxes of C, N and P per unit ground area in litter and insect deposits and compared them to other relevant sources of soil input of the same elements. We also evaluated the dependence of BIH and the resulting elemental fluxes on leaf chemistry, climate, elevation and relative position over the valley bottom to identify potential abiotic and biotic drivers. Our setup with multiple gradients spanning a larger regional elevational gradient allowed us to test the universality of relationships between elevation and foliar loss to herbivory. We hypothesised that:
(H1) Elemental fluxes : The BIH levels in subarctic birch forests are minor compared to lower latitude systems, and consequently the annual contribution of nutrients to the soil through this channel is much smaller than other internal (i.e. recycling from litter) and external sources (i.e. fixation, deposition, weathering).
(H2) Biotic controls : The BIH level increases with concentration of leaf N, which is the growth-limiting nutrient in the system. Consequently, we expected no strong relationship with leaf condensed tannin concentration, as this only play a role as a defence compound at low N-levels, where we already expected low foliar loss to herbivory. Moreover, we hypothesised that the importance of insect herbivores for channelling nutrients from the canopy to the soil would increase with stronger foliar nutrient resorption, as this would decrease the transfer of nutrients through senesced litter on annual basis.
(H3) Abiotic controls : We expected a local scale increase in BIH with the relative position above the valley bottom along each transect towards the treeline, due to decreasing likelihood of extreme winter cold and summer parasitism, which are both strongest in the valley bottom. Yet, we also expected an overall BIH level increase with site temperature, as insect herbivory levels are generally higher in warmer mid-latitude systems, but this effect may be substantially weakened by the expected local elevational increase in BIH.