Despite the potential for interactive effects of multiple stressors at each trophic level, plantmediated processes mostly predicted herbivore abundance and composition responses to multiple anthropogenic perturbations in the fragmented islands. For instance, the strongest driver, fertilization, reduced plant richness and increased dominance by a specific set of species, which in turn increased abundance of a few dominant herbivore taxa thus reducing among-islands spatial turnover. These among-island perturbation effects generated an important spatial homogenization of the plant community compared to the continuous mainland, which explain the observed decline in herbivore richness along the transition from continuous mainland to the fragmented island habitat. Therefore this study suggests that resource dependencies might play a fundamental role in the regulation of insect herbivore communities in response to the interactive effects of landscape fragmentation and perturbations (here fertilization and defoliation). Top-predators can be especially sensitive to perturbations, mainly because of their large habitat range requirement, high metabolic demand (McKinney 1997; Purvis et al. 2000), and stronger trophic dependencies (Gravel et al. 2011; Harvey & MacDougall 2014). This study did find evidence for a reduction in top-predator abundance from the continuous mainland to the fragmented islands habitat. However, this decline was not associated with any density or diversity-based response from the herbivore community that would be indicative of a disruptive effect on top-down control. Despite clear experimental evidence, in closed experimental systems, of the impact of top-predators on grassland insect herbivores (for a review see Schmitz 2010), empirical findings generally emphasize bottom-up control from plants (Haddad et al. 2009; Borer et al. 2012; Rzanny et al. 2013). My results confirm the well-documented sensitivity of top-predators to landscape alteration (Purvis et al. 2000; Martinson & Fagan 2014), but suggest that for this grassland insect community, the decline of top-predators does not lead to predator-mediated effects on herbivore communities. However, recent studies have shown that the regulatory effect of top-predators on insects can be context-dependent as a function of seasonality (Gratton & Denno 2003), wind speed (Barton 2014), and especially temperature (Hoekman 2010; Shurin et al. 2012; Barton & Ives 2014). In the current context of global change, caution is therefore required when predicting the relative importance of predator versus plant mediated processes as it is likely to change as a function of the focus stressors. In conclusion, this study illustrates that habitat fragmentation interacts with other important stressors associated with global change to strengthen plant-mediated control of insect primary consumers. Past work had already made the connection between habitat fragmentation and the potential disruption of trophic control (Nelson et al. 2013; Martinson & Fagan 2014). However, mechanistic linkages with multiple stressors are needed to accurately describe potential effects of global change, which are intrinsically characterized by the simultaneous occurrence of multiple stressors. I showed that among island dynamics associated with nutrient enrichment, defoliation and habitat size interacted to generate lower among-island spatial variability in plant composition compared to continuous habitat. These plant-mediated effects re-enforced trophic dependencies, leading to dominance by fewer but more abundant and less spatially variable herbivorous taxa. Given that generalist feeders are generally favored by landscape simplification (Rand & Tscharntke 2007; Bommarco et al. 2010; Filippi-Codaccioni et al. 2010), my results suggest that regional outbreaks of generalist insect herbivores are likely to occur more often with increasing plant spatial homogenization associated with landscape alteration and nutrient pollution (Bobbink et al. 2010; Barbosa et al. 2012; Hautier et al. 2014), and that specialist consumers might be at risk of population collapse because of increased pressure to find suitable resource (Bommarco et al. 2010; Clavel et al. 2010; FilippiCodaccioni et al. 2010). However, hyper specialist insect consumers might also be favored with regional homogenization because they can potentially outcompete other consumers for a single resource across the whole landscape (Pillai et al. 2011). These predictions remain to be tested more thoroughly given their important implications for conservation, and pest management efforts.