Localized defense induction in trees: a mosaic of leaf traits promoting variation in plant traits, predation, and communities of canopy arthropods?

Forest canopies are a spatially complex three-dimensional environment. This complexity is one of the reasons forest canopies harbor a substantial part of global biodiversity, including hyper-diverse communities of herbivorous arthropods (Basset et al., 2012). These communities vary substantially both between trees and within the large canopies of individual trees. Exploring the factors that promote variation between individual trees or their modules is crucial for understanding how the diversity of canopy arthropods is maintained. A substantial proportion of variability in herbivore communities can be attributed to the variation in the chemical traits of their hosts (Richards et al., 2015; Volf et al., 2018). A multitude of factors shape the variation in tree chemistry, including macro- and micro-evolutionary processes, seasonality, and abiotic conditions (Lämke and Unsicker, 2018; Volf et al., 2019). As reviewed by Lämke and Unsicker (2018), substantial changes in tree chemistry also appear due to induction of defenses following biotic stress. These changes can be asynchronous among plants or their parts and may support spatial variation in chemical composition of host plants (Rubert-Nason et al., 2015). Here we discuss localized defense induction as a source of chemical variation in canopy trees. We postulate how localized induced defenses can cascade to several trophic levels and can possibly drive variation in arthropod communities within a growth season. Many plants respond to herbivore and pathogen attacks by employing induced defenses (Kant et al., 2015; Turlings and Erb, 2018). These defenses involve the production of deterrent or toxic direct defenses targeted directly at herbivores (Kant et al., 2015). In addition, they may involve indirect defenses, such as herbivore-induced plant volatiles (HIPVs), which attract predators and parasitoids (Turlings and Erb, 2018). Such induced responses can significantly contribute to different levels of variation in plant chemistry. First, they support temporal variation in plant chemistry because induced defenses are deployed only when plants are attacked (Kant et al., 2015). Second, they support variation within plant populations consisting of herbivore-induced and non-induced individuals (Backmann et al., 2019). Moreover, the strength of the response may differ between the feeding site and distal plant parts, depending on how they are connected by the phloem (Viswanathan and Thaler, 2004; Rubert-Nason et al., 2015). Thus, induced responses are a causal agent of spatial variation in chemical traits among plants or their modules. Localization of induced responses can be especially important in large plants, such as trees, because it helps them to direct their defenses only to the branches where needed. Compared to herbs, trees are large, have a complex architecture, and are long-lived, which makes them apparent to insect herbivores (Lämke and Unsicker, 2018). The herbivore pressure, however, is not uniform across the canopy. Localizing induced responses only to the attacked modules can make the involved defenses more efficient. Indeed, although herbivory normally induces some systemic response, there is also a substantial level of localization (Lämke and Unsicker, 2018). Preliminary results from a garden experiment with oak, lime, and hornbeam trees indicate that increased volatile emissions are localized to the induced branches. Untreated branches on the same trees were largely unaffected by the induction treatment; their volatile emissions were not significantly different from those of untreated trees (M. Volf et al., unpublished results). Similarly, Clavijo McCormick et al. (2014) reported that although some upregulation of volatiles occurs in both induced leaves and non-induced adjacent leaves in 1-year-old poplars, the differences between the two are still significant, suggesting a relevant level of localization. Such localized induction events in trees have a potential to create a "canopy mosaic" of variously induced branches. They can contribute to variation in leaf quality across the canopy and possibly drive turnover in arboreal herbivore communities (Rubert-Nason et al., 2015; Lämke and Unsicker, 2018). This effect can be further pronounced by the diversity of defense mechanisms induced. One induction event may trigger the production of various metabolites, including both direct and indirect defenses (Kant et al., 2015; Lämke and Unsicker, 2018). As arthropod herbivores vary in their response to individual plant defenses (e.g., Volf et al., 2018), the diversity of mechanisms involved in defense induction could further promote herbivore turnover between variously induced branches (Fig. 1). Whereas some herbivores prefer weakly protected leaves, specialized herbivores may prefer high levels of secondary metabolites while being sensitive to other factors, such as enhanced predation (Sipura and Tahvanainen, 2000; Volf et al., 2018). The spatial variation in defensive traits and predator or parasitoid pressure between damaged and undamaged leaves thus could not only lower the dominance of abundant insects, but also create niches suitable for less dominant taxa due to the increase in overall chemical diversity. So far, the relative importance of induced defenses in structuring communities of canopy arthropods is largely unknown (Lämke and Unsicker, 2018). For example, many temperate broadleaf tree species rely primarily on phenolic-based direct defenses, including various tannins and lignin. Athough such metabolites are known to affect herbivore community structure (Segar et al., 2017), they do not always have pronounced effects on herbivore mortality (Barbehenn et al., 2009). The main defensive value of phenolics may result from their interplay with indirect defenses because they can retard larval growth and increase the exposure of herbivores to natural enemies attracted by HIPVs. Indeed, HIPVs could help predators and parasitoids to efficiently navigate toward their prey through the dense jungle of foliage (Vet et al., 1990), theoretically making them an important form of anti-herbivore defense in the canopy. However, recent models show that HIPV efficiency in attracting predators and parasitoids is highly dependent on their chemical properties (Douma et al., 2019). The efficacy of HIPVs as well as the predators and parasitoids themselves are also highly affected by the environmental conditions (Posa et al., 2007). The shape of the canopy, its density, or the tree site are likely of great importance. For example, the upregulation of volatile production increased invertebrate and bird predation on artificial clay caterpillars in our garden plot including oaks, limes, and hornbeams. However, the overall predation rates on the induced trees inside the relatively densely vegetated plot were still much lower than on trees at the margins. Although the trees on the plot margin were not induced, they were probably preferred by predators that avoided the dense canopies inside the plot. This preliminary evidence implies that although predation rates can be elevated by HIPVs inside the canopy, their efficiency as indirect defenses is affected by vegetation structure (M. Volf et al., unpublished results). It also remains to be established on which time scale induced defenses affect communities of canopy arthropods. The induction of direct defenses in herbaceous plants can be relatively rapid, leading to upregulation of defensive metabolites over several days. While relatively rapid upregulation of direct defenses has been also recorded in trees, direct defense induction can transcend between seasons and herbivore generations (Tuomi et al., 1988; Eyles et al., 2010). The induction of indirect defenses such as HIPVs is more rapid; in our garden experiment, significant changes in tree volatile profiles occurred within 24 h after the first induction treatment (M. Volf et al., unpublished results). The time scale on which induced defenses promote arthropod species turnover may also differ among herbivores. Both mobile and sedentary herbivores can be removed from the community, either because of increased host toxicity or increased predation. However, mobile herbivores, such as beetles or true bugs, can also rapidly change their location in response to changes in host chemistry, whereas less mobile or sedentary herbivores can change their distribution only in the following generation. We are now exploring some of these mechanisms and roles of induced defenses in canopy trees using the Leipzig Canopy Crane Facility (Fig. 1B). We combine branch-localized induction to promote chemical variation among individual oak branches and explore how this affects herbivore performance in laboratory experiments and predator pressure in the field. Furthermore, we aim to explore the effect of induced changes on herbivore communities and compare this to the variation driven by other important environmental gradients. These involve factors such as vertical stratification of the canopy, seasonality, or variation between individual trees. Our preliminary results show that such gradients are especially important drivers of insect community structure in the studied systems. We expect that such a combination of experimental approaches will allow us to identify whether induced defenses can truly contribute to high diversity and species turnover in diverse communities of canopy arthropods. We thank P. Diggle for the invitation to contribute this essay and for her valuable feedback. We also thank two anonymous reviewers for their excellent comments that helped to improve our manuscript. M.V. acknowledges funding by Alexander von Humboldt Foundation and the Federal Ministry for Education and Research Ref.3.3-CZE-1192673-HFST-P and Grant Agency of the Czech Republic 20-10543Y. C.W. and N.M.v.D. thank the German Research Foundation for funding the German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig (FZT 118).