Introduction

The European rabbit (Oryctolagus cuniculus) is native to the Iberian Peninsula (and perhaps also northern Africa), which has colonised almost everywhere in the world since the times of the Roman Empire1,2. The first record of rabbit introduction to America—specifically to Isabela Island, now Santo Domingo—dates back to November 14933. Although it is currently classified as an endangered species in its native lands4,5, it has become one of the most damaging invasive species in the world6. The first reference to rabbits in central Chile was by Ignacio Molina in 17887, but Delibes and Delibes-Mateos8 provided an earlier date: 1765 in Tierra del Fuego Island, in the extreme south of the continent, now shared by Argentina and Chile. Currently, the rabbit has a discontinuous distribution from the Atacama to Los Lagos regions and then to the Magallanes region7,9,10,11,12. However, some uncommon records of rabbits (sightings) have been reported in the Arica y Parinacota and Tarapacá regions (personal communication).

The Magallanes region extends from the Pacific Ocean (48° 36′ to 56° 30′ S, 66° 25′ to 75° 40′ W) eastwards across the Strait of Magellan and halfway to Tierra del Fuego Island (52°–56° S, 63°–73° W)13,14. The introduction of rabbits was attempted several times from 1873–1874 to 1913, but none of these introductions bore fruit, and it was the one of 1936 that generated the rabbit outbreak of 1950–195211,12,15. Once well established, its population grew rapidly due to the rabbits’ high reproductive capacity, along with a negligible predation impact by their potential predators because of low population density derived from hunting for their skins11,12,15,16. Subsequently, the high rabbit population in the Magallanes region generated an economic income due to commercialisation of their meat and skins7,17,18, but also had serious consequences for ranchers for their competition for forage with sheep and thus transforming grasslands into wastelands7,11,12,15. This led to various control efforts, the most effective being that of 1953, when the myxoma virus was introduced to Tierra del Fuego Island11,12,15,19, which—although reducing the rabbit population by 99%11,12,20—failed to eradicate it. The individuals that survived possibly developed viral resistance19, and today rabbits are still found in both continental and insular areas of the Magallanes region11,12.

The rabbit in Chile is considered an invasive species that causes significant damage, generating a loss of 3.25 million US dollars per year9,21 and numerous ecological effects at the population, community, and ecosystem levels, summarized by Correa-Cuadros et al.11,12. The long history of rabbit invasion (since the XVIII century)7,8 means that it has become an integral part of the Magellanic/Fuegian ecosystem, interacting with species at different trophic levels, simultaneously being prey, herbivore, and competitor11,12. Rabbits in continental Magallanes and neighbouring Tierra del Fuego Island consume plant biomass and cause important negative impacts on their abundance22,23. They feed on seedlings of various native shrubs and trees (e.g., Berberis and Nothofagus) and non-native herbs (e.g., Poa and Vicia), thus affecting their regeneration22,23,24. Plants not consumed by rabbits spread and generate changes in the ecosystem24,25. In addition, rabbit burrows increase soil erosion11,12,26. Rabbits also have negative impacts on sheep production by consuming the forage available for livestock, thus reducing the output of meat and wool for export11,12,24,25. They also compete with native mammals for food and habitat (e.g., Lama guanicoe, Oligoryzomys longicaudatus) and favour the population increase of predators and scavengers by being a subsidy of abundant prey27,28,29. Among several other Magellanic/Fuegian predators, the puma (Puma concolor), Lycalopex foxes, Leopardus cats, and birds of prey such as hawks, incorporate rabbits as staple prey to their diet30,31. Also, facultative scavenging bird species such as the Buzzard eagle (Geranoaetus melanoleucus) and caracaras (Caracara plancus and Milvago chimango), and strict scavengers such as the Andean condor (Vultur gryphus) benefit from rabbit presence32. Thus, the rabbit has become a key species (herbivore, competitor, and prey) of the regional food web.

Given the overwhelming evidence of the multiple types of damage generated by invasive species33, such as the European rabbit5,11,12, it cannot be overlooked that the rabbit is now functionally integrated into several ecosystems and interacts with multiple species27,28,29,34,35,36. Therefore, a change in its abundance (e.g., a decrease) can spread through trophic interactions and affect the species with which it interacts both directly and indirectly37,38, triggering impacts such as release of herbivory for plants, decreased competition for herbivores, or switching prey in predators, among others28,29,36,37,39,40,41,42.

The Magellanic/Fuegian communities face multiple threats due to anthropogenic activities (e.g., hydrocarbon development, grazing pressure, wildfires, and exotic species23,43,44), and it thus seems necessary to understand the role that the rabbit plays in the regional ecosystem. One way to study this is through a multi-trophic food web approach45, which takes a picture of species' positions in different trophic levels across a wide range of direct and indirect interactions and describes trophic relationships and their complexity46,47. This community picture of multiple interactions plus node extinction simulations enables the understanding of how a given disturbance (e.g., management or eradication of invasive species) may affect the presence and abundance of each node of the local food web46,48,49,50,51,52. Thus, a multi-trophic approach is a complementary and holistic tool to make more informed decisions for developing any management of an invasive species, identifying the non-target species that may be affected, and thereby generating early warnings against unwanted impacts.

Therefore, our objective is to elucidate the extent to which the rabbit is currently embedded into the food web of the Magellanic/Fuegian steppe and scrub ecosystem, with which species it interacts, and hypothesise the possible short-term effects that can be triggered if the rabbit is managed. With this theoretical information, this study seeks to highlight the complexity and potential consequences of rabbit management on the community network and to recommend an ecosystem approach rather than a single-target approach to safeguard against unwanted ecological impacts on other species.

Materials and methods

Study site

The Magallanes region is centred at ca. 53° S and 71° W (ranges 48° 40′–56° 00′ S and 66° 30′–75° 40′ W) and comprises 132,297 km2 excluding the Antarctic territory14. The corresponding adjective is Magellanic13. Geographically, there is a west-northerly continental portion, separated from an extensive east-southerly archipelago by the Strait of Magellan. The largest island of that archipelago is Tierra del Fuego Island (Isla Grande de Tierra del Fuego, in Spanish), situated within a polygon ca. 52°–55° S and 65°–72° W. This large island (ca. 48,000 km2) is split east–west between Argentina and Chile (40:60, respectively) at meridian 68° 34′ W. The corresponding adjective is Fuegian13. The surrounding archipelago contains seven medium-sized islands (Hoste, Santa Inés, Navarino, Dawson, Aracena, Clarence, and Staten; ranging from 4100 to 500 km2 in decreasing sequence), and ca. 3000 smaller islands and islets, most of them located to the southwest of Tierra del Fuego Island (Argentina’s Staten Island excepted) and thus are in Chilean territory.

The Magallanes region (Fig. 1) harbours three major climates53: temperate cold rainy (mean annual temperature 6 °C and mean annual rainfall 3000 mm), cold steppe (short and cool summers and slightly cold winters, mean annual rainfall of 500 mm), and high ice (mean monthly temperature not higher than 0 °C and annual snowfall 1000–3000 mm). It is divided into three major biomes54,55,56: the Andean-Patagonian forest biome, the Evergreen forest and peat bog biome, and the Patagonian steppe and scrub biome. The first consists of deciduous forests, with woodland physiognomy and a slight herbaceous layer. The lenga (Nothofagus pumilio) is the characteristic tree of this biome, together with the canelo (Drimys winteri), and the coihue (Nothofagus betuloides). The Evergreen forest and peat bog biome, also has a woodland aspect with Nothofagus trees and peat bogs of Sphagnum magellanicum as representative species. The Patagonian steppe and scrub biome is characterised by low-shrub vegetation, herbs, and grasses. The rabbit is present chiefly in the latter biome57.

Fig. 1
figure 1

The Magellanic/Fuegian ecosystems in southernmost Chile. The steppe and scrub biome was the main study area used to build the food web. The whole region has 48 million hectares: the pure steppe has 2.5 million ha and the scrub has 14 million ha (Universal Transversal de Mercator, Datum WGS84, Zone 18S). *We have a copyright holder to publish it under a CC BY open-access license.

Full size image

The Magallanes region has ca. 58% of its surface within the Chilean System of Protected Wilderness Areas (SNASPE, in Spanish), because it harbours a high number of unique species13,14,54. Among the productive activities of the region are sheep farming (56% of the national total), artisanal and industrial fishing, aquaculture, tourism, and forestry58. According to the national inventory of species of the Ministry of the Environment, the region harbours 176 species of animals and 242 species of plants54,59.

Food web construction

A bibliographic review was carried out through Google Scholar, Scopus, Web of Science, and other search engines for scientific publications or technical reports (but not abstracts), including theses, journal articles, books, and book chapters. Keywords such as south, southern, southernmost, Magallanes, Patagonia, Tierra del Fuego, Chile, diet, species genus (e.g. Oryctolagus, Lama, Alopecurus), species common name (e.g. Rabbit, Guanaco, Mariano), species scientific name (e.g. Oryctolagus cuniculus, Lama guanicoe, Alopecurus magellanicus) were searched in English and Spanish languages, solely or in combination.

To build a Magellanic/Fuegian food web, literature was first searched based on the diet and trophic relationships of the species directly connected to the rabbit (i.e., carnivores and scavengers that consumed rabbits and those plants that the rabbit consumed). Second, we expanded the network by looking for the predators and prey of each rabbit interaction. Third, the predators and prey of each new species that appeared during the expanding network process were searched for. Fourth, the last step was repeated until no new information could be found (whether they were new species or new trophic interactions). Each new node was searched for its consumers and basal resources (Table 1 Supplementary material) and information was added on diet studies from neighbouring Patagonian regions of Chile and Argentina, which shared similar/same ecosystems and species. Still, Argentinean Patagonia food web construction was excluded because Chile and Argentina differ in policies and regulations (e.g., legal hunting of native carnivores, different policies among provinces, lack of unified control measures against invasive herbivores36). Additionally, the two countries have different management practices, and the main focus of this paper is to demonstrate the high integration of rabbits in Chilean ecosystems. The results will be useful for future management actions and national control policies.

Additionally, to verify the bibliographic register, to confirm predator–prey relationships, and to estimate missing trophic links, we consulted with wildlife specialists, managers and field technicians from CONAF-Magallanes (National Forestry Corporation) and SAG-Magallanes (Agricultural and Livestock Service). Simultaneously, the list of species was compared with the National Species Inventory (https://especies.mma.gob.cl/CNMWeb/Web/WebCiudadana/Default.aspx)59, where searched information data to species for which had not found diet.

During the bibliographic search, we observed that diet analyses had different taxonomic resolutions of prey, generating shortcomings in constructing the trophic network. For example, some publications reported predators that consumed “passerines” versus those identifying passerine species (e.g., Zonotrichia capensis). In this case, it is not advisable to consider these prey as two different nodes because a given species also belongs to the passerine category. Thus, we decided that if > 50% of predator diets reported only the clustered node (e.g., passerines), such bundled nodes would be used, removing species-specific nodes. This way, any predator that consumed a given passerine species would be connected to the bundled node and not to separate specific nodes. In the contrary case, some consumers were reported to used one or more species of Nothofagus (N. antarctica, N. obliqua, N. pumilio, or N. nitida), while others were reported to consume only the cluster node (Nothofagus spp.). In this case, we assumed that the latter could consume either species. Finally, another criterion we used to build the food web was that those species with predators but without prey were designated as basal (i.e., within the primary producers level in the food web), and they included species without dietary data as well as insectivorous birds and insects, either because we could not find studies or because they were not specific enough (e.g., they consumed "seeds").

Food web analysis

Through a list of interactions between species, we constructed a binary adjacency matrix using the Network 3D program60,61, where 0 is absence and 1 is the presence of trophic interactions. We analysed the structural attributes of the network, including species richness (S), number of links (L), connectance (C = L/S2), generality SD (standard deviation of generality; generality of a node is the number of species consumed as normalised by L/S), and vulnerability SD (standard deviation of vulnerability; vulnerability of a node is the number of species consumed as normalised by L/S)46,62,63). In turn, we classified the nodes as either native, exotic, or mixed nodes, the latter representing some Orders, Families and Genera that have native and exotic representatives (e.g., the node Agrostis spp. contains native Agrostis glabra and introduced Agrostis capillaris). We performed this classification to better understand the relationship between rabbits as invasive species and their nodes, and how their interactions and the management of rabbits could structure or modify the food web/community. To quantify how embedded the rabbits were into the food web, we calculated each node's connectivity, number of predators, and number of prey, and ranked the rabbit position.

Rabbit extinction simulation

We carried out a simulation of rabbit extinction to evaluate the ecological mechanisms involved (e.g., release from predation, reduction of trophic niche) that could trigger short-term responses in the ecosystem, using the Network Extinction package52 and graph64 in R65. For this, we removed from the network the node belonging to the rabbit, and quantified the percentage of prey lost to predators (reduction of trophic niche) and the percentage of predators that lost prey (release from predation/herbivory) in the food web.

Results

The Magellanic/Fuegian steppe and scrub food web resulted in a network with 206 nodes and 535 trophic interactions (Links, L) and a connectance of 0.013 (Fig. 2). In the network trophic levels, 82% were primary producers (plants and basal species—see Methodology), 10% intermediate species (carnivores, herbivores, and omnivores), and 8% corresponded to top predators (carnivores and scavengers). The food web was made up of 132 native nodes (64%), 29 exotic (13%), and 45 (22%) mixed—native and exotic representatives—(Table 1 Supplementary material, Fig. 2). The link density (L/S) was equal to three, representing the number of interactions per node. The network had a vulnerability SD = 1.4 and a generality SD = 2.9, which implies that it was more homogeneous in terms of vulnerability than in generality, having nodes with very wide dietary breadth (e.g., guanaco, Lama guanicoe, with 47 consumed items/nodes, or sheep, Ovis aries, with 40) versus others with much narrower dietary breadth (e.g., quique, Galictis cuja, with five prey items).

Fig. 2
figure 2

Food web of the European rabbit Oryctolagus cuniculus in the Magellanic/Fuegian steppe and scrub. Each circle represents a node (species), and each line represents a link (trophic interaction). The size of each node represents the number of trophic interactions associated with it: larger sphere size indicates more trophic interactions. Species inside the yellow box represent plants consumed exclusively by rabbits, where the green colour indicates a native species, blue a mixed-origin species, and red an exotic species. *We have a copyright holder to publish it under a CC BY open-access license.

Full size image

The European rabbit was the most connected species in the network (connectance = 10.8), which means the rabbit node had the most trophic interactions (56 links, 10% direct interactions within the food web). At the same time, it was the second in generality (16.9) after the guanaco Lama guanicoe (18.1). The other species in sequence of connectivity after European rabbits were: Lama guanicoe (native herbivore; connectance = 10.2; 53 links), Ovis aries (domesticated herbivore; connectance = 9.6; 50 links), Oligoryzomys longicaudatus (native herbivore; connectance = 8.5; 44 links), and Hippocamelus bisulcus (native herbivore; connectance = 7.9; 41 links) (Fig. 2). These five species accounted for half of the links in the entire network (45%, 244 out of 535).

The rabbit was preyed on by 12 species, of which 11 were native and one was an exotic invasive (Neogale vison) (Fig. 3A). Among predators/scavengers with the broadest dietary breadth were Vultur gryphus, Lycalopex griseus, and Neogale vison. Rabbit predators with the narrowest dietary breadth were Galicitis cuja, Leopardus pajeros, L. geoffroyi, and Lyncodon patagonicus.

Fig. 3
figure 3

The relative importance of the European rabbit Oryctolagus cuniculus in the Magellanic/Fuegian steppe and scrub of southernmost Chile. (A) Dietary breadth of rabbit predators in decreasing order. (B) Resources used by rabbits in decreasing order. *We have a copyright holder to publish it under a CC BY open-access license.

Full size image

The Magellanic/Fuegian food web focused on predator–prey trophic relationships. Nevertheless, it also displays the rabbit's potential competitors for resources (Fig. 3B). Rabbits consumed 14 species exclusively, most of which were native (78%). The remaining plants they consumed (30 species) were shared with other herbivores, 40% native and 60% exotic, including Lama guanicoe, Ovis aries, and Bos taurus, which shared 19–22 plants with rabbits, followed by Hippocamelus bisulcus (sharing 12 species), Oligoryzomys longicaudatus (11 species), Sus scrofa (4 species), Lepus europaeus (3 species), Zaedyus pichy (1 species), and Chaetophractus villosus (1 species). The percentages of native, exotic, and mixed-origin nodes of the species consumed by rabbits were 59%, 14%, and 27%, respectively.

When removing the rabbit node from the food web through a rabbit extinction simulation (Fig. 4A), 11 plant nodes lost from 10 to 20% of their consumers, 12 plants lost from 20 to 30%, one plant lost 30–40%, and six plants lost 40–50% of their consumers. Also, 14 plant nodes lost 100% of their consumers—meaning predation release—(Gunnera tinctoria, Chiliotrichum spp., Drimys sp., Adenocaulon chilense, Tetroncium magellanicum, Symphyotrichum glabrifolium, Pratia repens, Urtica magellanica, Codonorchis lessonii, Gavilea lutea, Asplenium dareoides, Geranium spp., Viola sp., and Dactylis spp.), among which 78% were native species.

Fig. 4
figure 4

Rabbit extinction simulation in the Magellanic/Fuegian steppe and scrub food web. (A) Number of prey species that lose a certain percentage of predators by removing the rabbit from the network. (B) Number of predator species that lose a certain percentage of their prey by eliminating the rabbit. *We have a copyright holder to publish it under a CC BY open-access license.

Full size image

Most rabbit predators were not greatly affected by rabbit removal because they did not lose more than 20% of their prey (Fig. 4B). Only one species, Galictis cuja, lost 20% of its prey (trophic decrease). Therefore, the main ecological mechanism triggered by the removal of the rabbit node was the release of pressure on plants from its herbivory.

Discussion

Historically, biological invasions due to human actions have resulted in new environments and community structures worldwide66,67. Further, invasive species that have naturalised in their new environment have become adapted to the trophic network, interacting with many species at different trophic levels28. Although the ecological effects of these species are generally negative, it is relevant to increase knowledge about their role in food webs, about their impacts/effects on an ecosystem, and about the possible consequences for the local community29,68. European rabbits can modify and impact the structure of food webs, and can even destabilise them27,36,69, as has been evidenced throughout the world5,28,41,70,71,72. The Magellanic/Fuegian steppe and scrub food web showed the rabbits' key role as a consumer of native and exotic plants, a competitor for other herbivores, and a prey for various native and exotic predators and scavengers. This study shows that the rabbit has integrated deeply into the Magellanic/Fuegian steppe and scrub food web and highlights the importance of studying the possible short-term impacts of their control.

Primary producers had the most nodes compared to other trophic levels in the network, where most were native species. The importance of plant ecological roles is evident through the consumption by the species most connected to the food web, which are all herbivores (e.g., rabbits—Oryctolagus cuniculus-, guanaco—Lama guanicoe, sheep—Ovis aries, long-tailed mouse—Oligoryzomys longicaudatus, huemul—Hippocamelus bisulcus), highlighting the relevance of conserving plants for the community against threats such as invasive species37. The rabbit consumed 36% of the plant species in the Magellanic/Fuegian food web, among which more than half were native, which highlights its potential impact on the conservation of the magellanic flora.

As medium-sized abundant herbivores, rabbits have the potential to alter many ecological processes29,36, strongly affecting primary producers through massive consumption, and by creating strong interaction links through feeding or competition with other herbivores28,29,40,73. Therefore, rabbit management could relieve herbivory pressure on plants—as observed in the rabbit extinction simulation (see “Results”)—likely increasing their biomass and controlling the dispersal of exotic species that alter the composition and structure of the ecosystem74,75. This lagomorph consumed 36% of the plants in the network, where 14 species were consumed only by rabbits (e.g., Gunnera tinctoria, Urtica magellanica, Tetroncium magellanicum, and Codonorchis lessoni). In this case, rabbit control could directly benefit the growth and abundance of these plants, thus becoming a food resource available for other native nodes that could replace the ecological functions of the rabbit and conserve important plants in the steppe and scrub ecosystems, such as T. magellanicum, which is a monotypic species of evolutionary significance76, or C. lessonii, which is endemic and the only representative of the subfamily Codonorchideae in Chile77. Further, several studies have shown a positive impact on plant biomass after removing invasive species, including species reappearance37,40,78. Still, in the Magellanic/Fuegian steppe and scrub ecosystems, the impact generated by the rabbit may extend beyond native to exotic plants, where the possible effects of a reduction in rabbit density are unknown and need to be evaluated37,38. A potential impact could be an uncontrolled increase in exotic plants that are consumed only by rabbits (e.g., Dactylis spp.), thus becoming detrimental to the ecosystem—perhaps this genus should be monitored under experimental pilots without the presence of the rabbit to evaluate the possible mechanism that may be unchained. Notice that rabbits consumed exotic species such as Poa spp. and Taraxacum spp., which are highly effective colonisers although not dispersed by rabbits because they are anemochorous79. Therefore, rabbit control would not necessarily affect their dispersal.

The majority of plants that rabbits consumed were shared with a variety of herbivores, such as guanaco (Lama guanicoe, 22 shared nodes), sheep (Ovis aries, 21 shared nodes), and cattle (Bos taurus, 19 shared nodes), which could suggest competition for food80,81,82, thus affecting both native species and livestock. For instance, Empetrum rubrum, Senecio spp., Acaena spp., and Rumex spp. are not only part of the diet of rabbits, but also of native species with conservation importance, such as the huemul83 and the guanaco84. These plants are also economically valuable for livestock (sheep and cattle), an important productive system in the Magallanes region. Additionally, species of the Nothofagus genus have high ecosystem value by hosting diverse species and playing a key role in livestock grazing85. The Magallanes region has seen a decrease in the number of cattle in the last 17 years: 11% reduction in bovines (140 thousand to 120), 35% (10 thousand to 6) in horses, and 37% (2.2 million to 1.4) in sheep86. A decrease in the carrying capacity of the fields for sheep production has been attributed to the rabbit herbivory in Isla Grande de Tierra del Fuego during the 1950s87. Furthermore, because of carnivore-livestock conflicts, especially with the puma88, there has been a conversion from medium-sized (sheep) to large-sized livestock (cattle). In this sense, any management aimed at controlling rabbits would contribute to the recovery of soils and basal resources, benefiting other native prey species41, and increasing the availability of pastures for sustainable livestock management through regenerative grazing89,90,91.

High rabbit abundance has resulted in them becoming the primary prey for some predators, even enabling the rise of a rabbit-eating guild36. The European rabbit is now considered the main prey of many native predators in invaded regions and has been generating a dietary shift among them, channelling most of the biomass input to their diets, generating strong interaction links, and affecting the food web configuration and stability11,12,27,28,29,34,35,36,41,42,92. Here, the rabbit is now common prey for carnivores, scavengers, and omnivores—mainly native to the Magallanes region, where most have a prey range of 10–20 species, being considered generalists. Rabbit population must be managed in Chilean Patagonia. Still, this requires consideration of possible flow-on effects to top- and meso-predator diets and potential secondary impacts on prey due to predators switching their feeding behaviour37,38,40. The predators' short-term response to a shortage in their rabbit prey may be to adjust their attack rates on native prey, altering the food web and making it necessary to monitor key and at-risk species (e.g., Galictis cuja—see Results)29,41. Speculatively, perhaps regional predators would direct higher pressure on another abundant invader, the European hare (Lepus europaeus), which would not be unwelcome news93. A plausible scenario for generalist predators, recognizing their plastic feeding behaviours, is to adapt to the availability of resources and shift their diets, weakening other links as a consequence of establishing a strong link with rabbits28,80.

Our study shows that the network approach can provide a powerful tool for elucidating potential ecosystem-wide effects of European rabbits and hypothesising the consequences of management interventions. Rabbit control is more complex than just controlling a single species. It needs to carry out fauna monitoring and flora experiments to evaluate the effects on the network. Still, two limitations of our approach should be highlighted: (a) Missing links in the predator–prey or herbivore-plant interactions which are not considered because of a lack of information on them. These could affect estimates of niche breadth and of predation or herbivory pressure. (b) Variable diet records, with widely different sampling intensities in diet studies. Better sampled diets are more trustworthy than others based on a small sample. For future work we propose: (a) To collect information on the frequency of feeding interactions and—if possible—the amount of food consumed. (b) To estimate the total amount of biomass coming from the prey or plants and going to the consumers, to better capture the functional status of a species in a food web, (c) To carry out bioenergetic mathematical models to quantify the strength of the interactions considering the flow of nutrients or energy across the food web.