Climate change can have strong effects on species interactions and community structure. Temperature-dependent effects on predator-prey interactions are a major mechanism through which these effects occur. To understand the net effects of predator attack rates and dynamic windows of prey vulnerability, we examined the impacts of temperature on the interaction of a caterpillar (Arctia virginalis) and its ant predator (Formica lasioides). We conducted field experiments to examine attack rates on caterpillars relative to temperature, ant abundance, and body size, and laboratory experiments to determine the effects of temperature on caterpillar growth. We modeled temperature-dependent survival based on the integrated effects of temperature-dependent growth and temperature- and size-dependent predation. Attack rates on caterpillars increased with warming and ant recruitment, but decreased with caterpillar size. Caterpillar growth rates increased with temperature, narrowing the window of vulnerability. The model predicted that net caterpillar survival would decrease with temperature, suggesting that A. virginalis populations could be depressed with future climate warming. Theoretical work suggests that the net outcome of predator-prey interactions with increasing temperature depends on the respective responses of interacting species in terms of velocity across space, whereas the present study suggests the importance of effects of temperature on prey window of vulnerability, or "velocity" across time.

1. The boreal biome exhibits distinct alternative ecosystem states with high and low levels of tree cover. Insect outbreaks facilitated by climate warming could potentially drive transitions from high to low tree cover states. We investigated whether two key premises for such outbreak-induced transitions-critical thresholds (tipping points) and positive feedbacks that could maintain alternative states-are present in the northern-boreal mountain birch forest of Fennoscandia. Here, climate warming has promoted range expansions of defoliating geometrid moths, resulting in novel, severe multispecies outbreaks, most recently during 2002-2010.2. We conducted regional-scale field surveys of forest damage and recovery in 280 mountain birch stands in a northeast Norway immediately after the outbreak (2010) and 6 years later (2016). Satellite-derived time series of the Normalized Difference Vegetation Index (NDVI) provided an index of stand defoliation during the outbreak period.3. The proportion of dead stems per stand displayed a bimodal distribution, with stands generally being either lightly or severely damaged. This was due to a critical threshold in the relationship between defoliation and stem mortality, with mortality rates increasing abruptly in stands experiencing a mean drop in NDVI of more than 4% during the outbreak. The two key forest regenerative pathways-basal sprouting and sapling production-both displayed positive feedbacks with surviving stems and trees, so that regeneration success declined with increasing damage to the mature tree layer. These feedbacks imply that stands which have been forced across critical defoliation thresholds and suffered collapses of living tree cover may struggle to recover, especially if the loss of positive regenerative feedbacks is compounded by ungulate browsing on birch recruits. Synthesis.The north Fennoscandian mountain birch forest displays critical thresholds and positive feedbacks that conform to theoretical expectations for a system that could be vulnerable to abrupt and persistent changes of state in the face of novel, climatically facilitated insect outbreaks. These findings deepen the understanding of the persistent losses of tree cover that have occasionally been observed after outbreaks in this system in the past, and add to the list of mechanisms that could help explain the bistability of tree cover across the boreal biome. tarandus L.; Käyhkö & Horstkotte, 2017), which are abundant in northern Fennoscandia, and have negative impacts on the growth and survival of birch sprouts and saplings (Biuw et al., 2014; den K E Y W O R D S Critical threshold, Epirrita autumnata, facilitation, forest regeneration, global change ecology, Operophtera brumata, positive feedback, range expansion | 1143 Journal of Ecology VINDSTAD eT Al.

Context Patch-based population models predominately focus on factors that affect regional processes namely, patch size and connectivity, as the primary drivers explaining patch occupancy. This trend persists despite the recognition that patch quality can strongly influence population demography at the local scale. The quality of patches is often temporally variable and influenced by abiotic conditions. However, few studies have explicitly investigated how climatic variables influence the spatial and temporal dynamics of spatially-structured populations either directly or indirectly through changes in patch quality. Objectives Using a 10-year census of a spatiallystructured population of an outbreaking caterpillar, we determined the relative importance of patch quality (determined demographically), connectivity, precipitation, and their interactive effects on patch abundance, occupancy, colonization, and extinction. Methods We generated a series of statistical models and performed comparisons using Akaike's information criterion. We subsequently used likelihood ratio tests to determine the influence of each parameter on model fit. Results Patch quality and precipitation were the strongest predictors of the observed dynamics. We found that the dynamics of the spatially-structured population of Arctia virginalis were strongly influenced by precipitation: all patches had a higher probability of occupancy, contained higher abundances of caterpillars, and experienced fewer extinctions following wet winters compared to years following droughts. Conclusion These findings suggest that precipitation may act to influence the strength of heterogeneity of patch quality. This work demonstrates that patchbased models that do not include local and climatic factors may produce poor predictions under future climatic regimes.

The success of invasive species is often thought to be due to release from natural enemies. This hypothesis assumes that species are regulated by top-down forces in their native range and are likely to be regulated by bottom-up forces in the invasive range. Neither of these assumptions has been consistently supported with insects, a group which includes many destructive invasive species. Winter moth (Operophtera brumata) is an invasive defoliator in North America that appears to be regulated by larval mortality. To assess whether regulation was caused by top-down or bottom-up forces, we sought to identify the main causes of larval mortality. We used observational and manipulative field and laboratory studies to demonstrate that larval mortality due to predation, parasitism, and disease were minimal. We measured the response of larval dispersal in the field to multiple aspects of foliar quality, including total phenolics, pH 10 oxidized phenolics, trichome density, total nitrogen, total carbon, and carbon-nitrogen ratio. Tree-level declines in density were driven by density-dependent dispersal of early instars. Late instar larvae dispersed at increased rates from previously damaged as compared to undamaged foliage, and in 2015 field larval dispersal rates were related to proportion of oxidative phenolics. We conclude that larval dispersal is the dominant source of density-dependent larval mortality, may be mediated by induced changes in foliar quality, and likely regulates population densities in New England. These findings suggest that winter moth population densities in New England are regulated by bottom-up forces, aligning with the natural enemy release hypothesis.