Background losses of woody plant foliage to insects show variable relationships with plant functional traits across the globe

Abstract

Summary Despite a long history of research, we lack a comprehensive understanding of the reasons behind pronounced variation in foliar losses to insects among plant species. Our aim was to test the hypothesis that plant functional traits and/or ecological strategies are good predictors of the background losses of woody plant foliage to insects (i.e. losses occurring when herbivore populations are at their ‘normal’ densities) at the global scale. We conducted a quantitative synthesis of published and original data on foliar losses of 793 species of woody plants belonging to 128 families from ca. 500 localities world‐wide, representing tropical, temperate and polar climate zones. The background foliar losses to insects varied among plant life forms, increasing from dwarf shrubs to large trees, and were higher in inherently fast‐growing species than in slow‐growing species. These patterns supported predictions of both the apparency and growth rate (resource availability) hypotheses for data combined across localities, but only outside the tropics. Foliar losses to insects slightly but significantly increased with specific leaf area (SLA) in all climate zones. No differences were noted in background herbivory between woody plant species with deciduous and evergreen foliage, with different shade tolerance and with different successional status. Synthesis. Factors affecting the distribution of herbivory among species of woody plants differed among the climate zones, and the predictive power of at least some of the theories/hypotheses addressing plant–herbivore interactions at large spatial scales varies among climates and/or biomes. The average background losses of woody plant foliage to insects across localities can be predicted from the inherent growth rate of the plant species, its life form and SLA, although these plant characteristics jointly explain only a minor part of the total variation observed in the primary data.