Rewilded mammal assemblages reveal the missing ecological functions of granivores

Rewilding is a strategy for ecological restoration that uses reintroductions of animals to re-establish the ecological functions of keystone species. Globally, rewilding efforts have focused primarily on reinstating the ecological functions of charismatic megafauna. In Australia, rewilding efforts have focused on restoring the ecological functions of herbivorous and omnivorous rodents and marsupials weighing between 30 and 5,000 g inside of predator-proof exclosures. In many arid ecosystems, mammals are considered the dominant seed predators. In Australian deserts, ants are considered to be the primary removers and predators of seeds and mammals unimportant removers and predators of seeds. However, most research on granivory in Australian deserts has occurred in areas where ­native mammals were functionally extinct. Here, we compare rates of seed removal by mammals and ants on shrub seeds and abundance of shrub seedlings in two rewilded desert ecosystems (Arid Recovery Reserve and Scotia Wildlife Sanctuary) with adjacent areas possessing depauperate mammal faunas. We used foraging trays containing seeds of common native shrubs (Acacia ligulata and Dodonaea viscosa) to examine rates of seed removal by ants and mammals. We quantified the abundance of A. ligulata and D. viscosa seedlings inside and outside of rewilded areas along belt transects. By excluding ants and mammals from foraging trays, we show that ants removed more seeds than mammals where mammal assemblages were depauperate, but mammals removed far more seeds than ants in rewilded areas. Shrub seedlings were more abundant in areas with depauperate mammal faunas than in rewilded areas. Our study provides evidence that rewilding of desert mammal assemblages has restored the hitherto unappreciated ecological function of omnivorous rodents and bettongs as seed predators. We hypothesize that the loss of omnivorous mammals may be a factor that has facilitated shrub encroachment in arid Australia. We contend that rewilding programs aimed at restoring ecological processes should not ignore consumers with relatively lower per capita consumptive effects. This is because consumers with low per capita consumptive effects often occur at high population densities or perform critical ecological functions and thus may have significant population level impacts that can be harnessed for ecological restoration.