Philosophical Transactions of the Royal Society B: Biological Sciences
You have accessReview article

Soil biological responses to, and feedbacks on, trophic rewilding

    Abstract

    Trophic rewilding—the (re)introduction of missing large herbivores and/or their predators—is increasingly proposed to restore biodiversity and biotic interactions, but its effects on soils have been largely neglected. The high diversity of soil organisms and the ecological functions they perform mean that the full impact of rewilding on ecosystems cannot be assessed considering only above-ground food webs. Here we outline current understanding on how animal species of rewilding interest affect soil structure, processes and communities, and how in turn soil biota may affect species above ground. We highlight considerable uncertainty in soil responses to and feedbacks on above-ground consumers, with potentially large implications for rewilding interactions with global change. For example, the impact of large herbivores on soil decomposers and plant–soil interactions could lead to reduced carbon sequestration, whereas herbivore interactions with keystone biota such as mycorrhizal fungi, dung beetles and bioturbators could promote native plants and ecosystem heterogeneity. Moreover, (re)inoculation of keystone soil biota could be considered as a strategy to meet some of the objectives of trophic rewilding. Overall, we call for the rewilding research community to engage more with soil ecology experts and consider above-ground–below-ground linkages as integral to assess potential benefits as well as pitfalls.

    This article is part of the theme issue ‘Trophic rewilding: consequences for ecosystems under global change'.

    1. Introduction

    Trophic rewilding aims at (re)introducing large herbivores and/or their predators where they are missing due to humans [1,2]. In the debate on its merits and risks, a crucial component of the ecosystems whose functioning rewilding seeks to restore has been thus far quite neglected—soil. Perhaps tellingly, in two recent reviews on rewilding terms such as ‘soil', ‘below ground', ‘decomposition', ‘mineralization', ‘root' and ‘rhizosphere' never appear—not even in the titles of the references [1,2]. This is at odds with the ‘holistic' objectives of rewilding, given that gains and losses of herbivores and predators above ground may affect soil processes [3,4] and induce shifts in abundance, activity and community composition of soil biota [5,6]. The latter in turn contribute to carbon and nutrient cycling, primary productivity and plant diversity [79]. These feedbacks are especially important in the context of climate change: rewilding could either ameliorate or aggravate carbon losses to the atmosphere, but the discussion so far has focused more on plants than soil [1,10,11], although the latter is the largest terrestrial carbon pool [12], and soil microbes and animals drive its dynamics [1214]. If rewilding is to restore ecosystems without compromising carbon sequestration, responses cannot be assessed solely above ground, as plants and soil communities may respond to vertebrates differently [1517].

    Our aim is to show how soil biota and processes may respond to trophic rewilding, and how in turn these responses would facilitate or hinder the ecosystem shifts and functions that are sought to be restored. Since any rewilding initiative will occur in the context of a rapidly changing climate, we emphasized the role of climate as a control of soil biological responses and consider how these could ameliorate or worsen climate change. We searched the available literature on ISI Web of Science using keywords ‘reintroduc*', ‘introduc*' or ‘rewild*' in combination with ‘soil' and other descriptors of species of rewilding interest (e.g. ‘herbivor*', mammal*', ‘deer', ‘ungulat*’, ‘megafauna', ‘elephant*', ‘rhino*'), yielding 522 results (10 January 2018), the abstracts of which were examined to assess their relevance; all research articles cited in topical reviews [1,2,5,6,10,18] were similarly examined. As the number of suitable studies from actual rewilding projects was limited, we also drew from the wider literature on other species of herbivores and carnivores that can be used as proxies to understand the functional ecology of rewilding species. We included some studies that investigated the effects of non-native species (e.g. free-roaming cattle and horses in the New World) because they can be informative of the impact of a reintroduction in an ecosystem decades or centuries after its disappearance; moreover, some allochthonous species are actually at the centre of some rewilding proposals [19,20]. We emphasize that the evidence from studies on non-native species should be used with caution, as the impact of a novel species in an ecosystem may differ from that of a reintroduced species, for instance, due to coevolution of native organisms with the latter [2123].

    We articulate the following discussion along the following lines. First, we consider how species that have been or could be (re)introduced in rewilding programmes, or that would be directly affected as competitors or prey, affect soil biota and processes. We then consider how soil biological responses, in turn, affect those animals above ground, what implications their interactions have for ecosystem functioning (including climate change mitigation), and whether soil biota themselves should be considered in trophic rewilding, for instance through inoculation of keystone microbes and invertebrates.

    2. How species of rewilding relevance affect soil organisms and functioning

    (a) Grazers and browsers

    The effects of large herbivores on soil functioning have been reviewed extensively [5,18,24]. In short, grazing and browsing can change quality (e.g. recalcitrance) and quantity of plant litter, a major source of energy to soil food webs, and the resulting shifts in decomposer communities can enhance or slow down elemental cycling rates [2527]. Plants may respond to above-ground herbivory also, by altering resource allocation to roots [28], also a major source of energy to soil food webs. Through urine and dung, herbivores provide labile carbon to soils and relocate nutrients across the landscape [29], which may result in hotspots of microbial activity [30]. Also, their cadavers provide ‘fast' resources to soil [3133], and predators regulate and distribute such resource pulses (see §2c). Lastly, many herbivores affect soils through physical disturbance. Trampling by ungulates often reduces the abundance of soil biota and rates of nutrient cycling [34,35]. Extant megaherbivores such as elephants shape soil heterogeneity by forming trails, digging and creating open patches of vegetation [36], and many smaller species restructure soils through bioturbation (see §2b). Given such a wide array of mechanisms (figure 1), predicting the effects of herbivore introductions on soil functioning is no trivial task. A framework based on plant physiological and community composition responses outlined positive effects of herbivores on carbon and nutrient cycling in fertile conditions versus negative effects in unfertile conditions [5], but empirical support to this generalization is mixed [6,16,37,38].

    Figure 1.

    Figure 1. Conceptual diagram of interactions among the main species (re)introduced by trophic rewilding and keystone groups of soil biota, and resulting effects on plant–soil linkages. Large mammalian herbivores (box 1) affect soils indirectly by altering plant biomass and community composition (3), which may either speed up or slow down litter decomposition and nutrient cycling (4). Herbivores are in turn preyed upon by carnivores (2), which therefore influence plants and decomposers indirectly, and the latter also directly through provision of energy and nutrients via carrions. Herbivores also affect soil biota directly through three mechanisms: provision of energy via excreta (dung, urine) and carrions, soil compaction due to trampling, and loosening and translocation of soil through digging. Here we focus on two keystone groups of soil biota: soil microbes closely associated to plants, prominent among them mycorrhizal fungi (5), and macroinvertebrate ecosystem engineers (6). For simplicity, microbial decomposers and smaller invertebrates are implicit in box 4, as their activities drive litter decomposition and nutrient cycling. The interactive effects of herbivores and soil biota (wide arrows) could lead to synergistic, desirable effects on the ecosystem, but antagonistic effects or trade-offs could arise in some conditions. In particular, the interaction between ecosystem engineers (sensu lato) above and below ground might promote soil structure, or instead lead to soil erosion (bottom left); herbivore effects on vegetation and decomposition could improve soil fertility, but this could lead to higher carbon losses to the atmosphere (bottom right). The interactions here depicted by the black arrows represent a mixture of trophic and non-trophic effects, and the net results depend on the balance resulting from different mechanisms. Some potentially important interactions are not shown for simplicity (e.g. herbivores may directly affect the plant community also through digging and compacting soil; some mycorrhizal fungi may directly affect litter decomposition by acting as saprotrophs; and some mammals of relatively large size, such as bears and boars, may feed directly on soil macrofauna). (Online version in colour.)

    Trampling was highlighted alongside grazing as a mechanism to explain the maintenance of the late Pleistocene high-latitude steppe biome that disappeared after the extinction of megaherbivores [39]. Soil disturbance by large herbivores still exerts a widespread control on ecosystem functioning. In African savannah, increased soil compaction and reduced moisture due to trampling may be as important as plant consumption for the formation of grazing lawns [40], patches of fast-growing vegetation that constitute hotspots of nutrient cycling and plant diversity [23,41]. Such ecosystem engineering by large herbivores interacts with ecosystem engineering by soil fauna, with important consequences (see §3c). The disturbance of ungulate trampling could be especially important in insular ecosystems where non-avian herbivores are an ecological novelty. Avian megaherbivores such as moa differed from mammalian herbivores not only in diet [42] but also in their physical impact on soil, because they exerted much lower foot pressures than even medium-sized ungulates [43]. Therefore, before human arrival, plant and soil communities on many insular ecosystems evolved in the absence of mammalian trampling disturbances [44]. In fact, in a large-scale ungulate exclusion study in New Zealand, forest litter macrofauna tended to respond negatively to the ungulate presence, likely due to trampling disturbance rather than plant-mediated effects such as shifts in litter quality [45]. Soil biota evolved in such ecosystems could be more sensitive to trampling disturbance than soil biota in ecosystems shaped by large mammalian herbivores, but to our knowledge, no study has sought to assess it.

    (b) Digging and burrowing vertebrates

    Some species that are being reintroduced, or that would be affected by reintroduced species due to predation or competition, alter soil structure (e.g. mix soil horizon) by digging for food. Wild boars, a target for rewilding in Europe [1,46], create disturbed soil patches with faster nutrient cycling, resulting in higher carbon emissions relative to undisturbed soil [47]. Whether this scales up to landscape-level effects is unclear. Short-term exclusion studies on invasive boars and feral pigs showed weak effects on soil properties [48,49], but studies of longer duration showed strong effects on nutrient cycling and uptake by plants [50] and on soil microarthropod populations [51]. In fact, in European ecosystems in which they occur naturally, boars significantly affect soil properties [52] and fauna [53,54]. Among carnivores, European badgers in forests [55] and American badgers in steppes [56] were shown to form ‘cold spots' of nutrients by translocating deep soil to the surface, although this may increase the concentration of some micronutrients [55]. Effects of such disturbances tend to be transient, but repeated use of the same patches may have long-term consequences. In coniferous forests, oribatid mites declined in abundance near badger mounds, and although their numbers recovered within a few years, effects on community structure were longer lasting [57]. In alpine meadows, grizzly bears foraging on glacier lily bulbs tend to select the same, resource-rich patches every year, and it is precisely their digging activity that increases nitrogen availability and thus maintains those hotspots [58].

    Species that form and occupy long-lived systems of burrows have perhaps the greatest effects on soil structure and functioning [5961]. Burrowing and waste deposition by fossorial mammals often form soil patches with faster rates of litter decomposition and nutrient turnover [6264], and maintenance of such hotspots over multiple generations leads to long-term soil heterogeneity [65]. This, in turn, drives patchiness in above-ground communities, for instance by hindering plant growth on mounds and burrow openings but promoting it in adjacent areas [63]. Formation and maintenance of fertile soil patches are well-documented in Australia, where the decline of native fossorial mammals has led to a deterioration in ecosystem functioning that has not been compensated by invasive species such as rabbits [6669]. Reintroduction of native species could restore hotspots of soil moisture and nutrient cycling in Australian drylands, promoting plant and animal diversity [67,68]. On the other hand, mammalian bioturbation can also degrade soil structure [70] and reduce carbon sink capacity [71,72]. How mammalian bioturbation affects soil biota is not well-studied, but it has the potential to induce important functional shifts. In North American grassland, high density of prairie dogs may promote plant-parasitic nematodes, likely subjecting plants to more intense root herbivory [73]. Litter decomposition by termites declined when fossorial mammals were reintroduced in an Australian desert [74], suggesting that the loss of the latter had been partially compensated, in terms of soil functioning, by invertebrate ecosystem engineers.

    (c) Top predators

    Top predators have the potential to affect soil functioning by regulating abundance and habitat use of their prey. Among species of rewilding interest, only wolves are well-studied in their effects on soil properties, although their effects on soil biota have not been assessed. Within a few years of reintroduction in Yellowstone, wolves reduced numbers of ungulates (particularly elk) and facilitated the recruitment of trees, such as aspen, that had declined in their absence [75]. A wolf-induced trophic cascade on understorey plant diversity was found also in Wisconsin [76]. These plant responses may affect soil food webs through the deposition of a litter of different quality to the surrounding landscape. Less intensive grazing in sites with a higher perceived risk of wolf predation was observed in Yellowstone and in Poland [7779], and these ‘landscapes of fear' likely contribute to soil heterogeneity. In fact, soil nitrogen mineralization in Yellowstone declined after wolf reintroduction particularly in the most productive sites [77]. Similarly, dingo predation cancelled negative effects of kangaroos on carbon and nitrogen retention in an Australian desert [80]. Arctic foxes, which are being reintroduced in Scandinavia after being hunted to near-extinction [81], were shown to curtail the seabird-mediated nutrient subsidy from marine to terrestrial food webs to the point of inducing dramatic vegetation shifts on previously predator-free islands [82].

    Predators may also alter soil properties by providing pulses of carbon and nutrients with their kills. By bringing salmon on land, bears form soil patches with much higher nutrient content and faster turnover than surrounding areas [83,84]. While these biochemical hotspots are highly localized and short-lived [83], the legacies of larger kills can be long-lasting and contribute to the spatial heterogeneity of vegetation, for instance facilitating subordinate plant species [85]. In Isle Royale, moose carcasses left by wolves formed hotspots of soil nutrients and microbial biomass, which resulted in improved nitrogen uptake by plants even three years later [86]. In conclusion, despite the limited available evidence, it can be presumed that reintroduced predators may have substantial effects on soil functioning. Yet, considerable uncertainty remains on the net result of these interactions. Wolves in North America can have either net positive or negative effects on carbon sequestration depending on the system in which they are reintroduced [87].

    3. Predicting, and steering, soil biological feedbacks on rewilding

    (a) Restoration of keystone soil biota

    Promoting or reintroducing native herbivores to replace invasive ones could restore some functional attributes of soil communities. Here we focus on two keystone groups of soil biota: soil microbes closely associated with plants and coprophagous soil macroinvertebrates closely associated with herbivores (figure 1).

    Mycorrhizal fungi have mutualistic associations with many plant species, and therefore they may be sensitive to plant physiological and species shifts induced by rewilding [8890]; in turn, their role in controlling nutrient cycling and plant communities feeds back on herbivores [91]. There is also an even more direct connection: many vertebrates ingest mycorrhizal fungi and other soil microbes and disperse them through their faeces [89,92]. In Australia, reintroduction of native mammals promotes mycorrhizal colonization of rainforest soil [93], and alters fungal community composition in arid soils, with some fungal taxa being found exclusively in their presence [94]. The latter finding suggests that shifts in soil microbial community under invasive herbivores could be reversed by restoring the native herbivore assemblages. However, it is unclear whether fungal communities closely associated to certain species may be supported also by other native species, e.g. two species of insectivorous burrowing mammals in Australian drylands (short-beaked echidna and greater bilby) had dissimilar effects on soil microbial community composition [95]. Moreover, the success of (re)introducing native mammals in restoring native soil communities could only be ascertained by having a pristine baseline, such as a soil community unaffected by non-native species. Fully restoring herbivore–microbial interactions is not possible when the native species are extinct. In New Zealand, moa consumed and dispersed ectomycorrhizal fungi, including species on which native Nothofagus trees depend [96], whereas introduced herbivores (e.g. deer) disperse viable spores of invasive fungi, promoting introduced conifers over native tree species [97]. This suggests that, in addition to removal of non-native herbivores, inoculation of specific soil microbes may aid restoration of New Zealand forests, and of other ecosystems which have suffered prolonged losses of large herbivores. There is experimental evidence that targeted soil microbial inoculations can promote native plant diversity [98,99], but to our knowledge, this has not been considered in a rewilding perspective.

    The redistribution and recycling of herbivore dung is an important ecological function performed by widely distributed groups of soil macrofauna, including earthworms and dung beetles. These act as keystone species in grazed and browsed ecosystems through their contribution to carbon and nutrient cycling, secondary seed dispersal, and maintenance of hotspots of microbial diversity and plant growth [100102]. Dung beetles decline when large mammals become locally extinct [22], and in fact, some species disappeared with the late Quaternary megafauna extinctions [103]. Declines and losses of coprophagous invertebrates should be taken into account when predicting the impact of rewilding, for instance, whether a habitat has a functionally efficient community of dung recyclers. The observation of niche segregation among dung beetle species based on dung type [104] suggests that the disappearance of certain species may leave a functional gap in the community. On the other hand, some beetles that had evolved on islands formerly inhabited by avian megaherbivores, in the absence of large mammalian herbivores, readily feed on ungulate dung [105,106]. Nonetheless, it is doubtful that these generalists can fully substitute some missing species. Large-bodied dung beetles that are more closely linked to large herbivores have a disproportionate role in faeces removal and seed dispersal [22,100,107], and the ecological composition of their communities determines their functional impact [101,102,108]. Therefore, as with mycorrhizal fungi, it is worth considering whether (re)introduction of dung beetles should be a part of the rewilding discussion.

    (b) Soil response feeds back on carbon cycling

    A key question of trophic rewilding is whether it can promote or hinder carbon sequestration [1]. Here we focus on high-latitude ecosystems, as they are disproportionally important to global carbon cycling and are a target for some ambitious rewilding projects [109]. Modelling suggests that the shift from subarctic tundra to shrubland linked to the Pleistocene megaherbivore extinctions resulted in net carbon losses [110], but it is hard to predict whether reversing such habitat changes through rewilding would also reverse the effects on carbon cycling. Restoration of cold steppe biomes through the introduction of large herbivores in subarctic tundra can enhance soil fertility [39], yet it has been suggested that it would also reduce carbon losses under climate change, by mechanisms such as carbon storage in root biomass and reduced decomposition due to lower soil temperature [109,111]. However, quantitative evidence to date shows that, in most biomes, large herbivores increase carbon losses [112]. In high-latitude tundra, vertebrate herbivores tend to reduce the abundance of soil animal decomposers [6,34] and to promote mineralization and soil respiration [6,37,113]. This often results in reduced soil C storage [114116], although positive effects have also been found [117]. Notably, plant responses to herbivory are sensitive to warming, resulting in an altered quantity of plant detritus entering the soil food web [118]. Of particular concern is the possibility that expanding vascular plants in tundra might prime the decomposition of soil organic matter, but a recent study found no such mechanism [119]. Moreover, top-down control of shrubs in high-latitude ecosystems may enhance vegetation resilience to warming [120]. However, retention of fresh carbon is higher under subarctic shrubs than sedges [119], and similar differences between shrubs and non-woody plants are found in other ecosystems [121]. Overall, these findings underscore the need for more research on how herbivores affect carbon cycling in high-latitude soils. We also call for caution in extrapolating effects of currently studied species, such as reindeer, to those of more diverse herbivore assemblages that may result from rewilding, as divergent diet preferences may result in contrasting effects on soil carbon sequestration even at similar densities [122]. In short, herbivore identity matters not only to plants but below ground as well.

    A particularly ambitious and controversial type of rewilding is to introduce allochthonous species where native keystone species are extinct. The argument is that proxy species, i.e. non-natives taxonomically different yet ecologically similar to the lost keystone species, may restore biological interactions needed to maintain local biodiversity. Such proposals have generated an intense ecological and ethical debate, but the repercussions on soil functioning have seldom been considered. For instance, given the geologically recent and ostensibly anthropogenic extinction of wild equids in America [123], the spread of feral horses could be considered unintentional rewilding [124], but how this affects soil carbon cycling is unclear. As for intentional rewilding with allochthonous species, introduction of extant camelids in part of their extinct relatives' range could maintain local plant and pollinator diversity due to their ability to counteract shrub encroachment [19]; on the other hand, shrub encroachment tends to promote carbon sequestration [121], therefore by counteracting it camelids may have undesirable effects in a climate change mitigation perspective. Given the large but idiosyncratic effects of shrub encroachment on ecosystem functioning [121], trade-offs between above-ground and below-ground targets, e.g. plant diversity versus carbon storage, may be inevitable. Finally, extant giant tortoises have been proposed to replace extinct tortoises on some islands [20,125], on the assumption that their selective herbivory and seed dispersal would restore top-down control of the plant communities, making it a low-risk, high-reward scenario [126]. Again, a below-ground perspective might suggest different priorities. For example, it could be argued that the eradication of invasive species that cause erosion through overgrazing, such as goats do in the Galapagos [127], would suffice to limit carbon and nutrient losses.

    (c) Co-engineering by large herbivores and soil macrofauna

    Ecosystem engineering by soil macroinvertebrates such as earthworms, dung beetles, termites and ants leads to patchiness in soil properties and plant communities, which can influence even the largest consumers above ground. For instance, in African savannahs, nutrient hotspots around large termite mounds (particularly of Macrotermes spp.) facilitate the development of fertile vegetation patches [128], which are intensively grazed by megaherbivores [129,130]. A small but growing number of studies points to co-engineering by large herbivores and soil macrofauna as essential to the formation and turnover of the vegetation mosaic of grazed ecosystems. In particular, burrowing and soil loosening by macrofauna enable the formation of tall vegetation patches in grazed ecosystems, whereas grazing and trampling by large herbivores form lawn patches [40,131]. The mosaics observed in African savannahs and other ecosystems can, therefore, be explained not only in terms of plant responses to consumption but also as a ‘battle' between two contrasting forms of ecosystem engineering, compaction by herbivores and bioturbation by soil macrofauna [131]. Therefore, soil macrofauna may contribute to the ecological success of large herbivore reintroductions. Reintroduction of soil macrofauna could even be considered to meet some objectives of rewilding, as it has recently been shown that earthworm inoculation can speed up the restoration of plant communities [132]. However, even apparently similar ecosystem engineers may have different impacts on soil and therefore on plants and herbivores, e.g. co-occurring earthworm species in West Africa have opposite effects on soil porosity [133]. Therefore, any plan of soil invertebrate reintroductions as part of rewilding would require an understanding of the functional traits of the target species.

    Soil ecosystem engineers could be especially important in shaping ecosystem responses to climate change. In arid tropical ecosystems, termite mounds constitute hotspots of carbon and nutrient cycling in water-limited conditions [134], with implications for ecosystem resilience to drought [135]. In temperate ecosystems, dung beetles may enhance water retention and thus reduce plant stress under drought in temperate climate [136], whereas deep-burrowing earthworms may enhance water infiltration and buffer plant growth against extreme rainfall [137]. Conversely, invasive earthworms in North America appear to worsen the negative impact of climate warming on forest ecosystems [138]. Therefore, whether (re)introduced herbivores would facilitate or hinder soil macrofauna can have major ecological implications. At present, it is unclear how taxonomic identity and environmental condition control the responses of soil macrofauna to large herbivores. For instance, free-roaming equids in North American drylands have negative effects on ants [139], whereas megaherbivores such as in African savannah appear to benefit from habitat modification by termites without in turn affecting their abundance [130]. Such knowledge gaps may compromise the evaluation of potential strategies for managing rewilding projects. Replacement of moose by smaller deer appears to enhance the unwanted effects of non-native earthworms in North American forests [140,141], but it is not known whether reintroducing moose or predators of deer would counteract the ecosystem disservices of those invasive ecosystem engineers.

    4. Concluding remarks

    Overall, we have shown that species used or proposed for (re)introduction to restore trophic interactions above ground, particularly mammalian herbivores, may affect soil functioning and biodiversity through a variety of mechanisms. In turn, the responses of soil biota will feed back on plants and food webs above ground, and may promote species and biotic interactions of rewilding interest. We stress that there is considerable uncertainty in how soil communities respond to (re)introduction of herbivores and their predators in different contexts (e.g. climate, vegetation, soil type). This makes it difficult to predict the full impact of rewilding on biodiversity and ecosystem functioning over the long term, particularly in a rapidly changing climate. We, therefore, urge caution on proceeding with rewilding without considering the potential trade-offs between desired effects above ground and unintended consequences below ground. For instance, will restoration of above-ground food webs in some ecosystems come at a cost of reduced carbon sequestration and increased soil erosion? On the other hand, there are also potential synergies between above-ground consumers and soil biota that could contribute to successful rewilding. For instance, if mycorrhizal fungi and macroinvertebrate ecosystem engineers responded positively to reintroductions of large herbivores, their contribution to soil structure and vegetation heterogeneity could, in turn, promote biodiversity and large herbivore survival. Indeed, the largely unexplored potential for (re)inoculation of some keystone soil biota poses thus far untapped opportunities to complement rewilding as a tool for successful species and ecosystem restoration. We strongly encourage the scientific community leading the trophic rewilding discourse to engage with experts in soil functioning and biodiversity and in above-ground–below-ground interactions to obtain a more comprehensive picture of its feasibility and impact, and to better detect pitfalls as well as synergies.

    Data accessibility

    This article has no additional data.

    Authors' contributions

    W.S.A. drafted manuscript and figure. Both authors contributed ideas and references, critically evaluated and revised manuscript and figure, and approved the final version.

    Competing interests

    We have no competing interests.

    Funding

    We received no funding for this study.

    Footnotes

    One contribution of 16 to a theme issue ‘Trophic rewilding: consequences for ecosystems under global change’.

    Published by the Royal Society. All rights reserved.

    References