Rewilding with invertebrates and microbes to restore ecosystems: Present trends and future directions

Abstract Restoration ecology has historically focused on reconstructing communities of highly visible taxa while less visible taxa, such as invertebrates and microbes, are ignored. This is problematic as invertebrates and microbes make up the vast bulk of biodiversity and drive many key ecosystem processes, yet they are rarely actively reintroduced following restoration, potentially limiting ecosystem function and biodiversity in these areas. In this review, we discuss the current (limited) incorporation of invertebrates and microbes in restoration and rewilding projects. We argue that these groups should be actively rewilded during restoration to improve biodiversity, ecosystem function outcomes, and highlight how they can be used to greater effect in the future. For example, invertebrates and microbes are easily manipulated, meaning whole communities can potentially be rewilded through habitat transplants in a practice that we refer to as “whole‐of‐community” rewilding. We provide a framework for whole‐of‐community rewilding and describe empirical case studies as practical applications of this under‐researched restoration tool that land managers can use to improve restoration outcomes. We hope this new perspective on whole‐of‐community restoration will promote applied research into restoration that incorporates all biota, irrespective of size, while also enabling a better understanding of fundamental ecological theory, such as colonization and competition trade‐offs. This may be a necessary consideration as invertebrates that are important in providing ecosystem services are declining globally; targeting invertebrate communities during restoration may be crucial in stemming this decline.


| INTRODUC TI ON
Globally, ecosystems have suffered extensive, largely negative change through human activity. In efforts to ameliorate our impact, we invest billions into ecological restoration each year to repair environments (Palmer et al., 2016). Although there has been considerable discussion concerning the goals of such large monetary investments (including debate around embracing novel communities or aiming for a predisturbance remnant site (Hobbs et al., 2009), see Section 5), there are clear trends in how we have approached restoration so far. For example, although ecological restoration is the process of whole-ecosystem recovery, plant-only restoration dominates current practices (67% of projects) with only 24% of projects restoring both plants and animals simultaneously (McAlpine et al., 2016) (9% of projects were animal-only restoration and this likely occurs when the plant community is already in good condition). This focus on plants suggests that ecosystems are expected to conform with the "Field of Dreams" paradigm that is embedded within restoration ecology (Palmer et al., 1997;Prach et al., 2019), that is, if you build the habitat, other organisms will recolonize passively.
Plants also receive considerably more attention than nonplants in postrestoration monitoring: Plants are surveyed in 54% of projects, whereas less visible groups such as invertebrates and microbes are monitored in only 32% of projects (27% and 5%, respectively) (Kollmann et al., 2016). Studies of passive recolonization suggest that, although many species do recolonize without additional effort (Barber et al., 2017;Wodika et al., 2014), there are many factors that restrict fauna passively recolonizing restoration sites, most notably the suitability of the restored habitat, the proximity to source populations, and dispersal limitations of some fauna (Kitto et al., 2015;Parkyn & Smith, 2011). Dispersal limitations may be especially pertinent in reconstructing communities postdisturbance for smaller organisms such as invertebrates and microbes which are often dispersal-constrained (Brederveld et al., 2011;Chen et al., 2020;Jourdan et al., 2019;Peay et al., 2010).
Invertebrates and microbes are immensely important for restoration processes as they are key drivers of landscape-scale ecosystem functions such as nutrient cycling (Eisenhauer, 2019) and carbon sequestration (Anthony et al., 2020). However, it is often assumed that they colonize independently following restoration of plant species (Strickland et al., 2017). Although some invertebrates and microbes passively recolonize revegetated areas (Barber et al., 2017;Wodika et al., 2014), not all species can disperse to, colonize, or establish successfully. Indeed, invertebrate and microbe communities in revegetated areas do not often become indistinguishable from those in remnant sites. Some macroinvertebrate communities in restoration sites are only ~35% similar to reference sites 20 years postrestoration, whereas the relative abundance of key bacterial Phyla was only half recovered as compared to nearby target sites 16 years postrestoration (Strickland et al., 2017;Wodika & Baer, 2015). Although some of this difference is likely related to the complex interaction between temporal changes in habitat suitability and the movement of metacommunities, a significant proportion may be due to dispersal limitations (Kitto et al., 2015). For example, dispersal constraints have been suggested as a limiting factor in recolonization of restored streams by macroinvertebrates (Brederveld et al., 2011), restored meadows by snails (Knop et al., 2011), and restored arable land by microbes (Chen et al., 2020).
Where passive recolonization fails, more proactive attempts to improve ecosystem function and biodiversity in revegetated areas include actively reintroducing or "rewilding" missing biota. Rewilding is an increasingly popular conservation tool whereby select fauna are reintroduced to reinstate ecosystem function and restore degraded areas (Corlett, 2016). Although a relatively new term, the concept of rewilding can be seen as a subset of restoration (Hayward et al., 2019). As such, rewilding is similarly biased toward highly visible groups (vertebrates in this instance), with comparatively few published examples of rewilding with less obvious groups such as invertebrates and microbes. In the related field of reintroduction biology, invertebrates make up as little as 3% of reintroduction studies, despite their roughly 95% contribution to species diversity (Bajomi et al., 2010). Rewilding projects have therefore tended to ignore the "unseen majority": functionally important yet overlooked groups such as invertebrates and microbes. Examples of invertebrate and microbial rewilding are however common in soil inoculation studies, which often rewild whole communities during soil transplants. There are significant knowledge gaps within these studies as few monitor changes in invertebrates and microbes postsoil inoculation. The effect of rewilding was thus difficult to quantify in these instances (see Section 3). The potential for rewilding dispersal-constrained invertebrates and microbes into areas they fail to recolonize naturally is under-researched outside of soil inoculation studies and is therefore rarely considered during restoration. However, rewilding may increase the likelihood of achieving restoration goals, particularly where the aim is to restore to a state of biodiversity and ecosystem function that is similar to the source area.

| OBJEC TIVE S
In this review, we discuss the current incorporation of invertebrates and microbes into rewilding and restoration projects and how their use can be improved in the future. First, we explore how invertebrates and microbes have been used in ecosystem restoration to date and provide a summary table that highlights significant knowledge gaps in our approach to invertebrate and microbial rewilding so far. Next, as rewilding has significant ecosystem ramifications (both intentional and unintentional), we discuss scenarios in which invertebrate and microbial rewilding is justified during restoration.
Finally, we discuss how invertebrate and microbial rewilding can move forward in the future by utilizing their unique characteristics.
This includes specific examples of empirical invertebrate and microbial rewilding projects that land managers can use during restoration to improve the recovery of ecosystem functions and biodiversity.
Our goal is to challenge the current plant-focused view of restoration and provide the foundations for a more holistic approach that

| AC TIVE RE S TOR ATI ON OF INVERTEB R ATE S
The return of invertebrates to revegetated areas is crucial for restoration goals as they are critical components of functioning ecosystems. Invertebrates may fail to actively recolonize due to inadequate habitat within the restoration site or characteristics that limit dispersal, such as a lack of wings (Haase & Pilotto, 2019 (2017)).

TA B L E 1 (Continued)
The paucity of invertebrate rewilding projects demonstrates that there are significant knowledge gaps regarding if, how, and when invertebrates should be used to restore ecosystem function. However, the diversity of ecosystem functions provisioned by invertebrates may be matched by an equally diverse range of situations which call for active rewilding efforts.

| MICROB IAL RE S TOR ATION: MOVING B E YOND INTER AC TI ON S WITH PL ANTS
Like invertebrates, it is generally assumed that microbial diversity and function in revegetated areas will naturally attain the level  (Table 1).
Further, for ecological restoration, it might make more sense to consider whole communities: the ultimate success for restoration would be to reinstate biodiversity and ecosystem function in its entirety.
To do this, microbial rewilding will need to venture from the plantfocused singular AMF inoculation studies, while invertebrate rewilding should broaden from earthworms to communities that include a greater diversity of functional groups, such as those found in litter ( Figure 1).

| WHEN IS RE WILD ING INVERTEB R ATE S AND MI CROB E S NECE SSARY ?
Whether or not a practitioner chooses to rewild invertebrates or microbes is highly dependent on the first critical step in restoration: setting goals and targets (Prach et al., 2019) (Figure 2). For example, practitioners that accept a novel ecosystem may let a postdisturbance community form from whichever biota are best adapted to the novel abiotic conditions, regardless of their status as native to the area or their functional role, thereby avoiding active intervention (Hobbs et al., 2009). Other approaches aim to restore an area to a "natural" predefined target state in terms of species composition or ecosystem function. This is a common goal in ecological restoration and is the first of six key concepts underpinning best practice in ecological restoration as defined by the international Society for Ecological Restoration (Mcdonald et al., 2016). These target states are often based on the species, trait, and/or functional diversity of one or more nearby remnant sites, or if no remnant sites exist, literature that describes the community predisturbance (Prach et al., 2019). This paradigm is inherently interventionist as it can take significant effort and resources to push a degraded ecosystem toward its predisturbance state. As such, practitioners may be more inclined to rewild fauna from remnant sites when there is a desired remnant target state ( Figure 2).
Restoration success or failure can often depend on the ability of dispersal-limited species to reach and recolonize restoration sites and how this factor interacts with temporal changes in habitat conditions (Baur, 2014 Zealand would often reach their desired invertebrate community reference state in between 10 and 50 years, whereas poorly connected streams may never reach this state, regardless of improving environmental conditions. The latter scenario might be common in highly disturbed systems and may have stimulated emerging studies that examine the feasibility of rewilding whole communities of invertebrates into streams undergoing restoration (Dumeier et al., 2020; Haase & Pilotto, 2019).

| FUTURE P OSS IB ILITIE S FOR INVERTEB R ATE AND MICROB IAL RE WILDING
Stepwise restoration of communities by adding individual species is becoming increasingly unattainable, unrealistic, and ineffective in our rapidly changing and dynamic world. This has no doubt influenced the trajectory of restoration and rewilding projects, which have increasingly focused on reinstating ecosystem function and self-organizing communities, rather than compiling set groups of species (Harris, 2014;Perino et al., 2019). This changing paradigm suits the unique characteristics of invertebrates and microbes, F I G U R E 2 Conceptual framework of trajectories and restoration options for degraded communities modified from Bradshaw (1996) and Hobbs and Norton (1996) (a). Each step of restoration is associated with key questions practitioners need to answer to justify active interventions or to evaluate restoration goals (b). Following these stages, the degraded community (S1) is replanted with vegetation (S2). Fauna from the reference remnant community (S5) then passively recolonize the new restoration habitat. Where biodiversity and function are exceedingly slow or unlikely to reach remnant levels, active intervention via rewilding (S3) may push the restoration community closer to the reference community. Over time, biodiversity and function in the restoration community may sit within the natural variation (wavy lines) of the target reference community (S4) further encouraging their use in future rewilding projects. For example, the astounding diversity of invertebrates and microbes, the lack of knowledge of their functional roles, and their high spatial turn-over rates means that in any given community we are often unsure of which species are functionally critical (Setälä et al., 2005).
Thus, targeted rewilding of single species may not lead to desired changes in ecosystem function efficiency. However, invertebrates and microbes are miniscule and thus easily manipulated, meaning we can readily move whole communities, and the functions they provision, from one place to another (given the habitat is appropriate and enough species establish). This is already how a majority of invertebrate and microbial rewilding projects operate. For instance, soil inoculation is a common form of invertebrate and microbial rewilding which consists of moving soil from target sites (with invertebrate and microbe communities in situ) into restoration sites (Wubs et al., 2016). We term this practice "whole-of-community" rewilding, and although it is evident within soil inoculation studies, it is highly under-researched outside of soil transplants and thus rarely considered during restoration (

BOX 1 Rewilding litter invertebrates and microbes to improve nutrient cycling
Litter-dwelling detritivore invertebrates and microbes are critical; yet overlooked, components of ecosystems (Bender et al., 2016;Eisenhauer, 2019). They support the breakdown of leaf litter, which turns organically bound nutrients into nutrients available for uptake by plants.  (Srivastava & Vellend, 2005). Although debate surrounds the generality of patterns between biodiversity and ecosystem function (e.g., how the effect varies over spatial scales (Thompson et al., 2018)), it may be of particular use to restorationists as postdisturbance communities are biologically depauperate and their diversity can be easily and directly manipulated through practices such as rewilding (Srivastava & Vellend, 2005).

| HOW C AN WE RE WILD WHOLE COMMUNITIE S?
Successful whole-of-community rewilding, and indeed any form of reintroduction, depends on the suitability of habitat to which the transplantees are moved. For whole-of-community rewilding, the transplants of whole habitat would ideally come from nearby remnant areas of similar original state as they are most likely to contain species appropriate to the environment of the revegetated area (Dumeier et al., 2020;Jourdan et al., 2019;Wubs et al., 2016).
This both increases the likelihood of successful establishment and ensures that communities with appropriate functional and life history traits are used during restoration. Using nearby remnant target sites is the more common method for setting restoration goals (Mcdonald et al., 2016) and is how most documented cases of wholeof-community rewilding choose their source of rewilded populations (86%) ( Table 1) et al., 2020). Litter transplants will therefore be more effective at the height of detritivore activity which is generally during cool and wet conditions.

BOX 1 (Continued) F I G U R E 3
Leaf litter samples taken from remnant patches and moved into revegetation patches will carry a multitude of invertebrate and microbe species and individuals. Inset: detritivorous mites and springtails taken from a leaf litter sample practices (Corlett, 2016). This is inherently advantageous as the purpose of restoration is the complete return of biota and function, not just some specific species. Unlike traditional rewilding projects, whole-of-community rewilding also does not discriminate based on "likeability" of a species, meaning that overlooked, yet functionally important species, can be incorporated into restoration more frequently (Jourdan et al., 2019). These benefits may be why the whole-of-community reintroduction paradigm is ingrained in soil restoration and starting to gain traction in stream restoration (Dumeier et al., 2020;Haase & Pilotto, 2019). This is not to say that single species rewilding has no place in future projects. For instance, single species reintroductions of butterflies and bumblebees in the United Kingdom were the most feasible way to reconstruct historic pollinator communities and improve pollination across the landscape (Steele et al., 2019).

| RIS K S AND CON S IDER ATI ON S FOR WHOLE-OF-COMMUNIT Y RE WILDING
There are, however, risks associated with whole-of-community rewilding that need to be acknowledged and addressed. Land managers will also need to be conscious of the potential to spread invasive invertebrates and microbes during the rewilding event as non-natives may be embedded within remnant sites. Even though remnants sites would ideally be "pristine," thorough sampling of the source population prerewilding is needed to assess the risks of spreading invasive species into areas in which they may not be pres- They found that introduced AMF were scarce as compared to indigenous AMF, indicating the former were ineffective at establishing and proliferating within the in situ soil community (Emam, 2016;Lance et al., 2019). This was expressed as increased soil function in revegetated areas inoculated with indigenous soil whole communities, resulting in greater soil Phosphorous concentration (Lance et al., 2019) and increased plant biomass (Emam, 2016). Conversely, the ecological consequences of commercial AMF outcompeting native species are unknown but could pose a threat for native soil biodiversity and ecosystem function (Hart et al., 2018).
Although there are limited examples of invertebrate rewilding, we can learn much from the success and failures of these projects.
Reducing competition between in situ communities and rewilded communities by removing the former can have significant effects on the establishment of rewilded invertebrates and microbes. For example, a topsoil inoculation study looked at the difference in restoration success between areas where the topsoil and its resident soil community had been removed as compared to areas where the resident soil community was unaltered (Wubs et al., 2016). They found that when rewilded, whole soil communities were more likely to establish in areas where topsoil had been stripped and the competitive effects of resident soil communities removed, which manifested as a more successful restoration effort. Whether this is a general pattern, or dependent on the habitat in question, is not known. Benetková et al. (2020) speculated that it may be more appropriate to strip the resident community in forests as opposed to grasslands as soil formation is much faster under forests. They posit that other soil restoration projects van der Bij et al., 2018) were more successful than theirs due to their different rewilding methodology (Benetková et al. (2020) transplanted soil on top of the resident community as opposed to removing the resident community prior to the transplants). Further, literature on invertebrate translocations revealed that predation from species in the established community was a significant barrier to the establishment of reintroduced invertebrates (Bellis et al., 2019). This should also be a consideration for invertebrate and microbial rewilding.

| UNE XPEC TED CON S EQUEN CE S OF THE B I OTI C BARRIER
The biotic barrier can manifest unforeseen results during restoration. Remediation research has shown that the establishment of beneficial microbial inoculants in soil communities is not strictly correlated with measurable macro-ecological outcomes (plant growth in this instance). At least two independent studies have reported beneficial plant growth outcomes even when the inoculant was lost from the soil (Kang et al., 2013;Liu et al., 2015). The beneficial effects are attributed to changes to the native community structure and function triggered by the addition, and subsequent demise, of the inoculum. This research highlights significant knowledge gaps in managing soil function and indicates that monitoring of soil communities postrewilding will be necessary to disentangle outcomes for ecosystem functions and the microbial community. Addressing these knowledge gaps can include more specific instances of rewilding soil microbial communities that extend beyond current soil inoculation methodologies (Box 2).

| HOW C AN WHOLE-OF-COMMUNIT Y RE WILD ING AID CON S ERVATI ON?
Although the emphasis of whole-of-community rewilding often falls on reinstating function (  (Choudoir et al., 2018). Consequently, dispersal distances vary greatly between species, with some fungi exhibiting effective dispersal ranges of only ~1 km (Peay et al., 2010).
Like invertebrates, the return of soil microbes and the functions they provide to revegetated "habitat islands" on degraded farmlands is often assumed to occur passively (Box 1). However, restoration projects may benefit from actively rewilding soil microbes. This could both overcome dispersal constraints and tailor the reconstructed microbe community to a desired trajectory. Local paddock trees are often the last remaining remnant trees on degraded farms.
They are potential reservoirs of soil carbon cycling taxa as they contain mycorrhizal fungi and bacterial species adapted to competitive dynamics within local conditions (Wood et al., 2018). This provides a competitive advantage over communities already established in revegetated areas and increases the likelihood of rewilded communities overcoming the biotic barrier.
Rewilding soil microbial communities would entail moving

| CON CLUS I ON: TOWARD G RE ATER US E OF INVERTEB R ATE S AND MI CROB E S
If ecological restoration is to move forward as a more complete science, it is critical that we further investigate the role that rewilded invertebrates and microbes play during restoration. Recent advances in the field of restoration ecology have explored both the potential of whole-of-community rewilding as a restorative tool (Wubs et al., 2016) and how benthic stream invertebrates can be translocated as whole communities (Dumeier et al., 2020). We hope that the ideas and practical case studies proposed in this article will spur further empirical testing of whole-of-community rewilding which extend beyond soil inoculation studies. Monitoring throughout the life of invertebrate and microbial rewilding projects will be vital to determining their efficacy and the conditions under which it will enhance recovery rates. Ecosystem functions mediated by rewilded vertebrates can vary across abiotic gradients (Decker et al., 2019).
Whether the restorative potential of rewilded invertebrates and microbes varies spatially, and temporally, should therefore be explored.
Restoration projects currently overlook two groups that make up the bulk of biodiversity (Kollmann et al., 2016). This brings into question whether most restoration projects are failing to attain their end goal: the reinstatement of biodiversity and ecosystem function in its entirety.
Our argument for rewilding invertebrates and microbes addresses this shortfall, but we stress that their use should be considered for the novel advantages alone. Whole-of-community rewilding is a unique and potentially very powerful tool that land managers are largely unaware of.
A greater incorporation of invertebrates and microbes in rewilding projects may also simultaneously answer the resounding call for more thorough monitoring of this underappreciated group (Eisenhauer, 2019).
This would also help to fill substantial gaps in baseline knowledge of what species are present and what their functional role is before they are lost, thus assisting future recovery efforts of globally declining invertebrate populations (Klink et al., 2020). We hope that the ideas raised in this discussion engender a greater appreciation for the restoration and rewilding potential that invertebrates and microbes deserve.
This can help mold restoration ecology into a more holistic science that values the role of all biota, irrespective of size.

ACK N OWLED G EM ENTS
We would like to thank the anonymous reviewer for their constructive and helpful comments on an earlier version of the manuscript.

CO N FLI C T S O F I NTE R E S T
None declared.

F I G U R E 4
Collecting soil from the whole rhizosphere region of large established trees is impractical. Collecting rhizosphere communities by sampling 1 m out from the base of trees using a soil corer is a viable methodologic approach and is a more targeted way of rewilding microbial communities than current soil inoculation studies. Previous research has demonstrated that rhizosphere signatures can be detected using this approach for microbe communities from rainforest plant species despite the complex overlapping root networks (Wood et al., 2020)