30 More than herbivory: levels of silica-based defences in grasses vary with plant species, genotype and location Eeva M. Soininen, Kari Anne Bråthen, Juan German Herranz Jusdado, Stefan Reidinger and Susan E. Hartley E. M. Soininen (eeva.soininen@uit.no), K. A. Bråthen and J. G. H. Jusdado, Dept of Arctic and Marine Biology, he Univ. of Tromsø, NO-9037 Tromsø, Norway. – S. Reidinger and S. E. Hartley, School of Life Sciences, Univ. of Sussex, Falmer, East Sussex, BN1 9QG, UK. Current address: Dept of Biology, Univ. of York, Heslington, York, YO10 5DD, UK. Silica defences in grasses have recently been suggested to be a potential driver of vole population dynamics. However, the ability of grasses to induce silica in response to herbivory has not been tested in northern ecosystems where small rodents are important herbivores. We conducted a large-scale field experiment in subarctic tundra using three river catchments differing in herbivore densities, and examined the effects of small rodent and/or reindeer exclusion on leaf silica levels in five grass species ( Avenella flexuosa, Anthoxanthum nipponicum, Calamagrostis phragmitoides, Deschampsia cespitosa and Phleum alpinum). We also conducted a greenhouse experiment using three of these species ( A. flexuosa, A. nipponicum and D. cespitosa) and Festuca ovina to determine whether intraspecific genotypic variation affects baseline silica concentrations and the capacity to induce silica in response to simulated grazing. Baseline leaf silica concentrations and silica induction varied with plant species in both experiments, with catchment in the field experiment and with genotype in the greenhouse experiment. hese findings show that the allocation to silica defences in grasses is highly variable, and suggest that the combined effects of grazing pressure, plant species and intraspecific genotypic differences are likely to determine the circumstances under which silica induction may be an optimal defence strategy. A better understanding of the interplay between grazing and other factors influencing silica induction is necessary to interpret the role of silica in plant–herbivore interactions. Oikos 122: 30–41, 2013 doi: 10.1111/j.1600-0706.2012.20689.x © 2012 he Authors. Oikos © 2012 Nordic Society Oikos Subject Editor: Stan Boutin. Accepted 22 May 2012 Small rodents, which are renowned for their cyclicity, are the main trophic link between plants and predators in many ecosystems. Particularly in boreal and arctic regions rodent population dynamics have major impacts on the func- tioning of these systems (Ims and Fuglei 2005). Despite a century of research, no consensus has been reached on the causes of these cycles, although most researchers now agree that trophic interactions are the likely underlying mechanism (Berryman 2002, Turchin 2003). he role of predator–prey interactions (Hanski et al. 2001, Gilg et al. 2003) has long been acknowledged (but see Graham and Lambin 2002), whereas several plant-driven mechanisms have recently gained more attention (Kent et al. 2005, Huitu et al. 2008, Massey et al. 2008). One potential driver of cyclic rodent population dynamics are silica (SiO 2 ) based defences in grass leaves (Massey and Hartley 2006, Massey et al. 2008), which can alter both the preference and performance of mammalian herbivores (Massey and Hartley 2006, Massey et al. 2007a, 2009). For instance, grass leaves rich in silica are poorly digested by field voles Microtus agrestis, leading to a decline in the growth rate of both adults and juveniles (Massey and Hartley 2006, Massey et al. 2008). For a plant defence mechanism to cause herbivore populations to cycle, there needs to be a feedback mechanism between herbivore den- sity and levels of defence, i.e. the defence should be induced by herbivory and decline in its absence. Silica defences in grasses seem to fulfill this assumption: field studies have shown that leaf silica concentrations are related to previ- ous grazing densities (McNaughton et al. 1985, Cid et al. 1989), and recent manipulative greenhouse studies have demonstrated an induction of silica defences following sustained periods of grazing (Massey et al. 2007b, Garbuzov et al. 2011). Furthermore, there is evidence that grass silica concentrations and vole population densities show cyclic, synchronized dynamics in an upland grassland ecosystem in northern England (Massey et al. 2008). However, it remains unclear whether silica defences alone can cause small rodent population densities to decline, or whether other fac- tors, such as diseases, predation, or other changes in food quality, are required to produce such population crashes. Silica based defences are the most important defence mechanism identified in grasses (Vicari and Bazely 1993), which are generally considered to mainly tolerate grazing rather than defending themselves against it (McNaughton 1992, del-Val and Crawley 2004). Silica is taken up from