Discussion
Our results based on EDM modelling provide several new insights on joint effects of insect herbivory and climate as drivers of annual willow growth. First, our results support the prediction that negative effects of insect herbivory on shoot growth intensify along with effects of warmer conditions of growing season. However, contrary to our expectation, the effects of warming summer conditions on annual shoot growth were generally, on net, negative, although effects tended to become either less negative or weakly positive under warmer conditions, especially in experimental treatments that included vertebrate herbivores. Second, supporting our expectations, dynamic effects of insect herbivory and climate on shoot growth depend on vertebrate browsing and ramet age. In particular, the negative impact of insect herbivory is weaker on young ramets, while vertebrate browsing tended to enhance insect herbivory effects. Jointly, these results from a long-term experiment and EDM markedly advance our understanding of context-dependent effects of climate and herbivory on shrub growth at forest-tundra ecotones.
The finding that the negative effect of insect herbivory on shoot growth becomes more intense under warmer climates provides support for earlier studies suggesting that at more northerly latitudes, higher summer temperatures extend the summer season and increase the thermal budget for growth and reproduction insect herbivores, and it could result in increasing herbivory (Bale et al . 2002; see also Lehmann et al . 2020). It is also compatible with findings from an experimental system, in which O’Connor (2009) found that increasing temperature increased per capita interaction strength between macroalgae and their grazers, and reversed positive effects of temperature on plant growth. Barrio et al . (2016), however, found that a cold-adapted caterpillar performed worse under warmer conditions in the tundra. Therefore, it is still uncertain how insect herbivory levels will change under climatic warming as well as to what extent insect herbivory can modify climate-growth relationships.
Our finding of negative effects of warming on shoot growth, at least under cooler conditions, although somewhat counter-intuitive, could potentially be a result of the net balance between increasing respiration costs vs. faster overall growth and photosynthetic rates. Although photosynthesis rates generally increase faster than tissue growth rates – leading to an overall positive increase in growth with temperature – these changes depend strongly on the timing of the warming and the physiology of the plant (Grace et al . 2002). For example, in cropping systems, increased night time temperatures can have strong negative impacts on growth and yield due to a disproportionally large increase in respiration rates (Mohammed & Tarpley 2008). Similar imbalances would explain why growth under the herbivory treatments was generally bolstered by warming (where growth was presumably focused on edible tissues such as leaves and buds, which also contribute to photosynthesis), whereas warming effects were more strongly negative in the rejuvenation treatments (where growth presumably also included large amounts of structural tissue). Additionally, it is possible that in the coldest soils, root systems have strongly limited nutrient uptake activity (Havraneck 1972; Day et al . 1989) and cold conditions further decrease sink strength (Hoch & Körner 2003).
Importantly, our EDM analyses of long-term time series data on shoot growth reveal that growth dynamics are not a simple function of summer thermal conditions, but rather are highly context dependent, and vary as a function of herbivory and plant age. Our results are generally compatible with results of the meta-analysis of Arft et al . (1999) who found that deciduous shrubs exhibited weak growth responses to short-term experimental warming. They proposed that deciduous shrubs such as Salix may have a tight control of meristem activity, which may limit their response or ability to respond quickly to warming. Based on our results, we find it likely that vertebrate browsers change meristem activity of buds, and thus either stimulate or inhibit shoot growth and thus influence shoot growth-climate relationships. We find further evidence that young and old ramets differ in their inherent growth dynamics (see also Wijk 1986). In other studies, Vuorinenet al . (2020) found that pine growth reduced at high browsing pressure and warm climate conditions and Marshall et al . (2014) reported that woody vegetation responses to shifts in browsing pressure can depend on plant age. Also, the continuum ontogenetic response model (Massad 2013) predicts that plant herbivore responses can vary depending on plant age. Therefore, there is a growing body of empirical and theoretical evidence pointing out that climate and insect herbivory effects on tundra shrub growth dynamics will necessitate jointly considering both browsing level and age.
Our results have two main implications to current understanding of plant-herbivore interactions under climatic warming. First, these findings seem to challenge a common perspective that temperature is a dominant factor controlling shrub growth at their range edges (Myers-Smith et al . 2015). As our study system is at the edge of the common occurrence of the studied Salix species, we find it possible that effects of herbivory are at strongest at range edges and limit growth of established shrubs. We suspect that this phenomenon is wide-spread but remains rarely documented due to lack of long-term data and dynamic modelling that can link growth and growth regulating drivers. See, however Trotter et al . (2002) and comparable results of Wilmking et al . (2018) reporting Eurois occultaoutbreak effects on Alnus viridis in Greenland and Vuorinenet al . (2020) reporting negative warming effects on pine growth at high deer browsing levels. Second, as has been suggested with regard to vertebrate browsers (Christie et al . 2015) we find it possible that potentially intensifying effects of insect herbivores and their outbreaks (Seidl et al . 2017; Ren et al . 2020), the expansion of deciduous shrub genets is markedly slowed down, while newly established genets and their ramets in warmer and sufficiently enemy free circumstances play the greatest role in the observed tundra shrub expansion.