Risks and opportunities of trophic rewilding for arthropod communities

Abstract

Trophic rewilding is a restoration strategy focusing on the restoration of trophic interactions to promote self-regulating, biodiverse ecosystems. It has been proposed as an alternative to traditional conservation management in abandoned or defaunated areas. Arthropods constitute the most species-rich group of eukaryotic organisms, but are rarely considered in rewilding. Here, we first present an overview of direct and indirect pathways by which large herbivores and predators affect arthropod communities. We then review the published evidence of the impacts of rewilding with large herbivores on arthropods, including grey literature. We find that systematic monitoring is rare and that a comparison with a relevant control treatment is usually lacking. Nevertheless, the available data suggest that when the important process of top-down control of large-herbivore populations is missing, arthropod diversity tends to decrease. To ensure that rewilding is supportive of biodiversity conservation, we propose that if natural processes can only partially be restored, substitutes for missing processes are applied. We also propose that boundaries of acceptable outcomes of rewilding actions should be defined a priori, particularly concerning biodiversity conservation, and that action is taken when these boundaries are transgressed. To evaluate the success of rewilding for biodiversity, monitoring of arthropod communities should be a key instrument.

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

1. Introduction

Human influence on the natural world has grown to a planetary scale, leading scientists to propose the beginning of a new era, the Anthropocene [1]. Rates of biodiversity loss have arrived at alarmingly high levels under the combined pressures of habitat conversion, increasing land-use intensity, climatic warming and anthropogenic nitrogen and phosphorus inputs [2,3]. Biodiversity losses have especially been recorded for vertebrates [4], but the available evidence indicates that invertebrates show even stronger declines [2,5].

Traditionally, biodiversity conservation in most European ecosystems has focused on providing a modern substitute for the low-intensity agricultural systems that gave rise to species-rich biotopes, with a focus on livestock grazing and mowing. However, instead of re-instating or mimicking these traditional land-use practices [6], rewilding is increasingly suggested as an alternative [7–10]. Rewilding aims at the restoration of missing or dysfunctional ecological processes, with an emphasis on trophic networks involving keystone species, such as large herbivores, large predators or beavers (Castor sp.), which have strong ecosystem impacts (trophic rewilding) [11–13]. This may be achieved by facilitation of natural recolonization or (re-) introductions of these species.

Arthropods constitute by far the most species-rich group of eukaryotic organisms, representing over 60% of all known species [14], but they are rarely considered in the development of rewilding concepts. Whether rewilding benefits arthropod communities thus remains unclear, which is largely due to a lack of an evidence base, resulting from two main factors. First, rewilding is a predominantly process-oriented strategy [11], where biodiversity targets are deliberately considered of secondary importance and therefore rarely monitored [7], and second, there are very few good examples of trophic rewilding in practice [15,16]. This hampers the development of evidence-based insights into the effectiveness of rewilding for biodiversity [11,17].

Rewilding often carries the underlying assumption that biodiversity will automatically benefit from the restoration of large-scale ecosystem processes [13]. However, there is an accumulation of evidence showing that ecosystem restoration and biodiversity trends do not necessarily go hand in hand [18], because of land-use legacies [19] and the failure of species to re-colonise after restoration [20,21]. For arthropods, there could be additional negative consequences because of the vulnerability of many species to increases in grazing or browsing intensity [22–24]. Hence, it is crucial to include diversity of arthropods in the rewilding agenda, a call first made by Merckx [25].

Here, we address the knowledge gap of the impact of trophic rewilding on arthropods. We first present an overview of the pathways by which trophic rewilding can affect arthropod communities. We then review the available literature, including grey literature, to evaluate the impacts of trophic rewilding with large mammalian herbivores on arthropod communities. Finally, we discuss the balance of risks and opportunities for arthropod communities under rewilding and arrive at recommendations on how to integrate arthropods in an evidence-based strategy that reconciles rewilding with biodiversity conservation objectives.

2. How trophic rewilding affects arthropod communities

The effects of trophic rewilding on arthropods are mediated by the changes in biotic and abiotic environment induced by the (re-)introduced animals. Such changes are primarily caused by large mammalian herbivores, through their food intake and other behaviour, by ecosystem engineers such as beavers (Castor sp.), which remove trees and create ponds, and by large predators, which can change herbivore densities or behaviour. To predict the effects of trophic rewilding on arthropod communities, we extend the framework of Van Klink et al. [22], which focused only on the effects of large herbivores, to include beavers and large predators. We apply it specifically to the rewilding setting (electronic supplementary material, figure S1).

Arthropods are affected by keystone species and ecosystem engineers in direct and indirect ways [22]. Direct positive effects of (re-)introduced animals on arthropods include the use of their living bodies by parasites, feeding on their excrements by coprophagous fauna and feeding on their deceased bodies by necrophagous fauna. Negative direct effects include trampling and accidental or deliberate ingestion. Logically, parasites, coprophages and necrophages can only benefit from the presence of large-bodied mammals if the local management precludes the use of antiparasitic medication (which detrimentally affects dung-feeding fauna [26]), and the removal of carcasses.

Indirect effects of keystone species operate via the vegetation, the soil and, in the case of beavers, the water (electronic supplementary material, figure S1). The effects of large herbivores on arthropods have been studied well, and their most obvious effect is mediated through the vegetation structure and species composition [22]. Recent reviews have shown that increases in densities of large herbivores lead to decreases in arthropod species numbers [22], abundances [23] and population sizes [24]. Under high herbivore densities, the vegetation will consist of short grazing lawns, but under low-to-intermediate densities, a mosaic of short and tall vegetation of various successional stages can develop [27]. Since the species of almost all terrestrial arthropod groups show a variety of responses to vegetation height and successional stages (e.g. [28–31]), the development of a heterogeneous vegetation mosaic should be beneficial to the diversity of arthropods. Although direct evidence for this outcome is limited [22], using data from experimental grazing exclosures in the rewilding site Oostvaardersplassen (OVP; The Netherlands) [32], where natural grazing refuges are absent due to the high grazer densities, we can show that mosaics of grazed and ungrazed vegetation are richer in arthropod species than homogeneously grazed vegetation, a pattern that is consistent across spatial scales (figure 1). An aspect of grazing that is specific for rewilding is winter grazing. Herbivores then strip bark from trees [34], contributing to tree mortality [27]. This might directly benefit saproxylic insect species [35], and change vegetation structural heterogeneity.

Figure 1.

The presence of exclosures as grazing refuges enhances arthropod species richness. We used data from the exclosure study of Van Klink et al. [32], but excluding dung- and carrion-feeding species, to assess arthropod species richness across scales. Arthropods were caught in pitfall traps in grazed areas, in the exclosure centres and exclosure edges. The heterogeneous landscape of grazed and ungrazed patches and the transition zones showed a consistently higher richness than the homogeneously short grazed vegetation. Solid lines are observed accumulation curves, the dashed lines extrapolations and shaded areas are bootstrapped 95% confidence intervals following the study by Chao et al. [33].

Large herbivores can also alter soil conditions by treading, rooting and wallowing (dust bathing). This creates open patches that can be used by, for example, bees as nesting habitat, and thermophilic arthropods for basking [30]. It also creates germination niches for plants that require disturbed soil, on which herbivorous insects may be specialized. This was exemplified by the positive effect of wild boar rooting on the grizzled skipper butterfly (Pyrgus malvae) [36].

Beavers (Castor fiber and C. canadensis) have specific effects on the environment by building dams, which causes large changes in almost all physiochemical parameters of the stream bed [37]. In beaver ponds, densities of freshwater macrofauna are generally higher and characteristic of low velocity streams and soft substrate (e.g. Chironomidae, Odonata and Hemiptera), which often contrasts with low densities of the hard-substrate, high-flow fauna up- or downstream (Ephemeroptera and Plecoptera) of beaver dams [37]. Also, larger wetland complexes created by beavers are characterized by a large number of distinct biotopes, each with its own flora and fauna. The abandonment of beaver ponds can lead to the development of a successional vegetation mosaic, which was shown to provide suitable habitat for the nymphalid butterfly Neonympha mitchellii [38]. Owing to the creation of such varying biotopes, the diversity of the whole stream bed is thus often higher in streams with beaver presence than streams lacking beavers (reviewed by Bush & Wissinger [37]). The positive effect of beaver dams on invertebrate diversity is thus primarily attributable to high habitat heterogeneity.

The effects of large predators on arthropod communities have received little attention, but can be expected to be primarily mediated though their predation on (large) herbivores. Although it is contested that large predators control population sizes of large herbivores [17], there is no doubt that they affect herbivore behaviour and spatial terrain use by creating a landscape of fear [39]. This suggests that the presence of large predators will cause variation in grazing intensity, and thus the creation of grazing refuges, which are essential for many arthropod species. To our knowledge such effects have only been shown by Ripple & Beschta [40] in Zion National Park (USA). At locations where pumas (Puma concolor) were common, riparian vegetation, wildflowers, butterflies, lizards and amphibians were significantly more abundant than where cougars were absent and, hence, deer densities were high.

This overview suggests that, setting aside the effects of beavers, trophic rewilding will primarily affect arthropod communities via the impacts of large herbivores on the vegetation and soil, which may or may not be regulated by large predators. As is the case in agricultural or semi-natural systems [22], it can generally be expected that high densities of large herbivores in rewilding reserves will be detrimental to arthropod diversity. In the next section, we review the evidence for this prediction across various reserves where free-ranging large herbivores have been re-introduced or released from human intervention.

3. Published effects of rewilding with large herbivores on arthropod communities

To obtain a first qualitative overview of the effects of trophic rewilding with large herbivores on arthropod communities, we compiled a list of studies investigating rewilding effects on any terrestrial arthropod group that is not directly associated with large mammals (thus excluding coprophagous and necrophagous fauna). For finding relevant grey literature, we mostly relied on our professional networks. Because ‘rewilding’ is not always defined unambiguously, we set restrictions on site inclusion. Since the primary goal of trophic rewilding is the reinstatement of natural processes, a minimum criterion is that the introduced mammals must be able to display all their natural behaviours within the reserves. Therefore, we include only year-round grazing systems with free-ranging wild herbivores or robust livestock. This still leaves considerable variation in the implementation of rewilding in these sites, ranging from uncontrolled herds of wild or feral animals, to domestic grazers kept at predefined densities by culling or translocation (here termed ‘near-natural grazing’). In these sites with near-natural grazing, the large herbivores have the legal status of livestock and are therefore usually treated with parasiticides, and dead or dying individuals are usually removed.

We focused on contrasting herbivore impacts at high versus low densities. We therefore estimated herbivore density as live biomass per km² and determined whether this exceeded a rough estimation of the herbivore carrying capacity at which heterogeneous vegetation mosaics can develop, after WallisDeVries [41]: 3000 kg km⁻² at low soil fertility, 6000 kg km⁻² at intermediate fertility and 15 000 kg km⁻² at high soil fertility. Hence, ‘low’ here refers to densities below this carrying capacity, ‘moderate’ refers to at carrying capacity and ‘high refers to exceeding 1.5 times the carrying capacity.

We found studies on arthropods from 10 reserves with free-ranging large herbivores. The reserves are spread over five countries, all in Western and Central Europe (electronic supplementary material, table S1), where large predators have been rare or absent for centuries. The ecosystems cover the entire climatic range that can be found in this region, from coastal lowlands to alpine systems and continental steppes, but have in common that they were all used for livestock grazing until the introduction of free-ranging herbivores. Five cases concerned high herbivore densities and five low to moderate densities in relation to the estimated carrying capacity. Not surprisingly, low or moderate densities were only found in reserves with population control (near-natural grazing). The outcomes of the studies are summarized in table 1; more details on the evidence, as well as background information on the locations and management of each reserve, can be found in electronic supplementary material, appendix 1.

Table 1.

Summarized effects of rewilding and near-natural grazing (NNG) on arthropod diversity in ten reserves. Herbivore density was classified as high when the biomass exceeded 1.5 times the estimated carrying capacity, as moderate between 1 and 1.5 times the carrying capacity and low when it was below carrying capacity. Reserves are ordered from high to low herbivore density.

reserve	rewilding type and reserve size	live ungulate biomass (km2)	approximate carrying capacitya	herbivore density	impact period (< or >10 years)	comparison made	taxa studied	qualitative result	outcome
Höltigbaum (DE)	NNG 375 ha	13 800	6000	high	short	change over time	spiders, ground beetles, grasshoppers, moths	+/−	increase in grasshopper and moth species richness, decline in carabid richness, no change in spiders [42]
OVP (NL) 2016 densities	fenced rewilding 5600 ha	28 400	15 000	high	long	presence or absence of artificial grazing refuges	spiders, true bugs, plant and leafhoppers, beetles, soil macrofaunab	−	species richness is higher in the presence of ungrazed areas (figure 1)
New Forest (UK)	NNG 38 000 ha	5400	3000	high	long	change over time	moths, butterflies, singing cicada	−	loss of at least 124 species of butterflies and moths (59 associated with herb and shrub layer in forest clearings and grasslands), and singing cicada [43,44]
Kraansvlak (NL)	fenced rewilding 205 ha	5300	3000	high	short	high or low spatial terrain use	grasshoppers, butterflies	+/−	positive for grazing-tolerant butterflies and grasshopper species, negative for grazing-sensitive species [45]
AWD (NL)	accidental rewilding 3400 ha	5200	3000	high	long	nearby reserve with low densities	butterflies	−	negative for grazing-sensitive butterflies; no change in grazing-tolerant species ([46], figure 2)
Swiss NP (CH)	unfenced rewilding 17 000 ha	1400	1000	moderate	long	change over time	butterflies	+/−	loss of two grazing-sensitive butterfly species [47]
Hortobágy (HU)	NNG 3000 ha	9700	9000	moderate	long	nearby intensive sheep grazing	spiders	+/−	no effects on species richness, only small differences in community composition [48]
Ohrdrufer Platte (DE)	NNG 2500 ha	15 600	15 000	moderate	short	nearby cut meadow	plant and leafhoppers	+	increase in species richness, including arboreal species [49]
Knepp wildland (UK)	NNG 1400 ha	13 000	15 000	low	long	change over time	butterflies	+	increase in species richness (electronic supplementary material, appendix 2)
Rodach floodplain (DE)	NNG 67 ha	8200	15 000	low	short	nearby cut meadow	plant and leafhoppers	+	increase in species richness, including arboreal species [49]

^aApproximate carrying capacity under which vegetation mosaics can develop following the study by WallisDeVries [41].

^bDung- and carrion-dependent fauna excluded.

From table 1, it becomes clear that experimental work on arthropods in rewilding reserves is virtually non-existent, and that systematic monitoring is rare. From the limited information available, it is hard to draw conclusions about the effects of rewilding on arthropod communities, due to variability in study design, ecosystem type and implementation of rewilding. Most importantly, a comparison with a control site is only rarely made, and even when this is done, the controls vary from intensive agriculture [48,49] to nature reserves ([46]; figure 2). Nevertheless, the results in table 1 show that an absence of top-down control of herbivore populations was detrimental to arthropod diversity in most cases (especially OVP, New Forest and AWD). This agrees with recent reviews of the effects of both wild and domestic herbivores on arthropod communities [22–24]. The effects of low-density near-natural grazing were often positive (see Ohrdufer Platte, Knepp Wildland and Rodach floodplain), suggesting that when top-down control of large-herbivore populations is missing, a substitute (in this case culling) can benefit biodiversity.

Figure 2.

Mean additive trend slopes (±s.e.) of butterfly abundance on monitoring transects in periods of low (1992–2005) and high (2005–2016) densities of fallow deer (Dama dama) in the coastal dunes of AWD (Amsterdam Water-supply Dunes) for 16 grazing-sensitive and five grazing-tolerant butterfly species. In the adjacent control area NPZK (National Park Zuid-Kennemerland), deer densities remained low. Mean trend slopes were only significantly different between periods for grazing-sensitive species in AWD (paired t-test; t = −3.45, d.f. = 15, p = 0.0036 (**)). This was also the only group with a trend deviating significantly from zero.

4. Risks and opportunities of rewilding for arthropods

Although trophic rewilding holds the promise of restoring the natural processes that have maintained biodiversity since time immemorial [9,13], it is clear that in the current human-dominated landscape, a full reversion to pre-Anthropocene conditions will be impossible. For arthropods, the simple approach of passive rewilding, i.e. abandonment without further intervention, will only lead to a brief period of bloom, before habitat heterogeneity will diminish with shrub encroachment, leading to a closed young woodland and the disappearance of open-land species [25,50]. Trophic rewilding intends to retain habitat heterogeneity through the influence of natural processes performed by keystone species. Since rewilding explicitly promotes the reinstatement of natural dynamics, it is crucial to consider the limitations set by the surrounding landscape, the constraints on species re-introductions and land-use legacies.

The landscape and land use around rewilding reserves will determine the degree of natural dynamics considered acceptable, as well as the possibilities of colonization by new species. The introduction of large herbivores and predators can lead to population growth and spillover into the human-dominated landscape, causing human-wildlife conflicts. Additionally, the extent to which floods and wildfire are acceptable should be considered. Connectivity and/or stepping stones between reserves are also of key importance for colonization by poorly dispersing species (e.g. [13,20,21]), but it is essential to recognize that arthropods often have very different habitat requirements than larger vertebrates, so that existing corridors may not be functional.

Within the constraints set by the surrounding landscape, the implementation of trophic rewilding will be of crucial importance for arthropod diversity. In the absence of large predators or other types of top-down control, populations of large herbivores will increase, with detrimental effects on arthropod diversity (see above). In Europe, introduced large herbivores are typically considered as domestic stock from a legal perspective. Therefore, they are subject to animal welfare regulations, which has major consequences for parasites and coprophagous and necrophagous species.

Rewilding always builds on the legacy of past land use, which will determine the effects of the rewilding actions on biodiversity. Former agricultural land use may have increased soil productivity, reduced natural (a)biotic heterogeneity, led to local or regional extinctions of native flora and fauna, and may have facilitated the spread of exotic species [19,51]. Trophic rewilding in such altered or impoverished ecosystems may have unforeseen effects (e.g. alternative stable states), preventing the development of the system to its potential taxonomic or functional composition. This may be pre-empted by active intervention (e.g. topsoil removal, water table restoration or assisted colonization), which, although financially daunting, may ultimately lead to more natural and biodiverse landscapes.

Trophic rewilding offers a possible alternative to traditional conservation practice, or to the passive rewilding following land abandonment [8,10], but our review of the available evidence shows that its effects on biodiversity conservation are not unequivocally positive. We propose two strategies to ensure that rewilding supports biodiversity conservation:

First, we propose the application of substitutes for natural processes when the full suite of natural biotic and abiotic interactions cannot be restored within the constraints of the reserve and its socio-ecological context. This is particularly the case in the densely populated regions of Central and Western Europe, where reserves are often small and embedded in an urban or agricultural landscape. Substitutes for missing processes can, for example, be (periodic) downsizing of herbivore populations, predator-mimicking hunting regimes or controlled fire or flooding regimes.

Second, we propose that the range of acceptable outcomes be defined before the start of any rewilding project, particularly with respect to undesirable outcomes. The outcome of rewilding is by definition uncertain and variable in time and space. However, this does not imply that all outcomes of rewilding are acceptable or desirable. When boundaries are set on the range of possible outcomes, action can be undertaken when these boundaries are transgressed. Examples of undesirable outcomes could be the extinction of an introduced population, loss of other species (e.g. open-land species or red-listed species), severe loss of habitat heterogeneity and, more generally, catastrophic wildfire, flooding or even the loss of public support.

5. Perspective: integrating arthropods in rewilding research

Our review reveals several urgent knowledge gaps that should be addressed in current and future rewilding initiatives. (i) Since trophic rewilding is primarily an alternative to either long-term abandonment (passive rewilding) or (mimics of) traditional low-intensity land use, research should focus on comparing these options for biodiversity conservation, leaving aside the self-evident comparison with intensive agricultural land use. Such research should be conducted over several decades, because the short-term benefits of land abandonment for arthropods are well known [50]. (ii) The trophic cascades from large predators to arthropods via large herbivores and the vegetation are still poorly known. The scant evidence suggests that arthropod diversity benefits from top-down control by large predators, but this needs to be studied across spatial and temporal scales. (iii) The constraints set by former land use require more attention in realizing the full biodiversity potential of rewilding.

In support of Merckx [25], we advocate that rewilding should support biodiversity and therefore be designed as an active process of learning-by-doing, i.e. adaptive management with short feedback cycles between biodiversity monitoring and management. This requires an a priori definition of undesirable potential outcomes, and planning of mitigation measures. In this way, biodiversity can be protected from the uncertainties in the outcomes of rewilding in combination with land-use legacies and climate change.

In a framework of biodiversity conservation, it is essential to monitor the effects of rewilding on arthropods, since these make up the majority of macroscopic life. In the face of their overwhelming diversity, several feasible approaches may be suggested to develop monitoring schemes: (i) involvement of citizen scientists to monitor conspicuous arthropod groups. For butterflies, this has been a successful approach across many countries [52], as exemplified here by the case study from AWD. This approach could be broadened to include other taxa. (ii) Use of (semi-)automated acoustic monitoring of loud chirping insects, such as orthopterans and cicadas [53]. (iii) The use of LiDAR to monitor changes in vegetation structure [54] and (iv) establishing collaborations with taxonomic specialists for (infrequent) monitoring of larger and more difficult arthropod taxa. These indicators can be used to evaluate the outcomes of rewilding, and identify the need for active intervention, which should be highly contingent on local climate, biotope and land-use history.

Data accessibility

Our biomass calculations, all AWD data, can be found in electronic supplementary material, appendix 2. The OVP data are available at [55] and the Knepp Wildland data at https://knepp.co.uk/s/Main-Butterfly-report-2017.pdf (and are copied in electronic supplementary material, appendix 2 for future reference).

Authors' contributions

R.v.K. and M.F.W.d.V. conducted the literature search. R.v.K. analysed the OVP data and M.F.W.d.V. analysed the AWD data. R.v.K. and M.F.W.d.V. wrote the manuscript.

Competing interests

We declare no competing interests.

Funding

R.v.K. was supported by DFG grant no. DFG FZT 118 to iDiv.