The occurrence of trophic cascades has become widely accepted in ecology. However, studies of trophic cascades have generally been limited to small spatiotemporal scales with small organisms conducive to manipulation. Evidence for trophic cascades involving large mammals over large spatiotemporal scales has been building over the last decade. We conducted a meta-analysis that included 9 studies with 15 individual datasets to determine whether predator extirpation could lead to recruitment gaps in browse-sensitive tree species. We included studies that quantitatively measured tree population age structure in a location from which large predators have been functionally reduced by human activity. Age class size should decrease exponentially with time since establishment, whereas we hypothesized smaller than expected age class sizes in decades without predators. We tested this hypothesis by using age class sizes before predator extirpation and an exponential model to estimate expected post-extirpation age class sizes. Decadal effect sizes were calculated by comparing observed and expected age class sizes using the log response ratio R = ln (observed/expected). We computed average decadal effect sizes before and after predator extirpation for 12 sites exposed to cervid browsing and 3 refugia sites.
Results/Conclusions
A paired permutation t-test revealed a 30-fold difference in effect sizes between decades before (-0.11 ± 0.056 SE) and after (-3.24 ± 0.53 SE) predator extirpation in non-refugia sites (n = 12 sites; t = -9.83; p = 0.0003; 10,000 permutations), indicating strong recruitment limitation following predator loss. Based on 95% confidence intervals only 1 out of 12 non-refugia sites deviated from expected recruitment patterns when predators were common. When predators were absent, all non-refugia sites exhibited significantly less recruitment than expected. However, observed recruitment in browsing refugia sites did not differ from expected before (-0.26 ± 0.21 SE) or after (-0.19 ± 0.79 SE) predator extirpation (n = 3 sites; t = -0.14; p = 0.60; 20 permutations). Predators must have exerted strong top-down forces in these ecosystems allowing browse-senstive species to thrive. The refugia sites demonstrate that overbrowsing rather than climate or fire regime change was likely the dominant factor restricting recruitment. Our mean log response ratio of –3.24 for recruitment frequencies when predators were absent is among the highest effect sizes recorded for any trophic cascade, terrestrial or aquatic.
