Soil-disturbing animals are common globally and play important roles in creating and maintaining healthy functional soils and landscapes. Yet many of these animals are threatened or locally extinct, either by habitat loss, predation by non-native animals or poaching/poisoning. Some reintroduction and rewilding programs have as their core aims both the increase in animal populations and reinstatement of processes that have been lost due to their extirpation. Here we review the effects of soil-disturbing vertebrates on ecosystem processes, using a meta-analytical approach to advance the argument that they can be used to rehabilitate degraded ecosystems, by altering mainly composition and function, but with fewer positive effects on structure. We describe four case studies showing that loss or reintroduction of soil-disturbing vertebrates leads to ecosystem state changes, and highlight the role of spatial scale, co-varying management changes, and species co-occurrence in modulating their effects. We discuss the advantages and disadvantages of using soil-disturbing vertebrates over mechanised engineering approaches such as pitting and furrowing, including more self-sustainable and heterogenous disturbances, creation of new habitats, and added recreational values. Finally, we identify key knowledge gaps in our understanding of the use of soil-disturbing vertebrates for rehabilitating degraded ecosystems.

We used the global database of Mallen-Cooper et al. (2019) and updated it with papers published up until September 2022 that contained information on soil disturbance by vertebrates published between November 2016 and September 2022. We maintained the same search string used by Mallee-Cooper et al. (2019) and used the ISI Web of Knowledge and Google Scholar: biopedturbation, biopedoturbation, faunalpedoturbation, faunalturbation, “foraging pits” and zoogeomorphology. Studies were restricted to terrestrial environments and where a publication reported data on the effects of vertebrates on plants, soils, or ecological processes, and compared this with a paired undisturbed control. Our searching resulted in a combined database of 2437 records for 70 vertebrate species across all continents except Antarctica (Fig. 1a). In total we had data on 35 attributes (electronic supplementary material, table S2), including four that were unclassified (lizard size, cover of micro-depressions, rock cover and proportion of exotic plants, representing 102 records). From these studies we extracted the raw means and variances of the different attributes. We used ImageJ to extract data from figures.

No specific packages, only Microsoft Excel to calculate means and confidence intervals.