The COVID-19 pandemic has brought the threat of zoonoses into the public spotlight, creating widespread demand for better management of the ecological sources of disease spillover and emergence. However, even prior to this pandemic, there has been an increasing recognition amongst experts of the ties between healthy ecosystems and human health. This has led to broader support for global conservation initiatives and spurred the United Nations' adoption of sustainable development goals (the 2030 Agenda). The prevention of zoonotic spillovers is a biosecurity imperative with a patent connection to the human-wildlife interface; thus, efforts are underway to identify solutions that both promote biodiversity conservation and facilitate zoonotic disease management 1 . However, given our incomplete understanding of the mechanisms linking biodiversity to infectious disease spillover, a clear vision of how to effect positive solutions for both human health and the environment is needed. Increased attention to, and resources for, zoonotic disease prevention make it an opportune time to study the mechanisms connecting changes in biodiversity with zoonotic disease spillover, and to identify (potentially synergistic) solutions for biodiversity conservation and global health.
There has been a contentious debate about the existence and generality of the relationship between biodiversity and disease: in particular, the extent to which maintaining biodiversity protects against disease via a dilution effect versus the alternative possibility that biodiversity can increase infectious disease transmission via an amplification effect (see for example references 2- 9 ). With a few notable exceptions [10][11][12][13][14][15][16] , this debate has largely focused on correlations between host-species richness and the prevalence of pathogens in host reservoir populations. However, this narrow way of framing the impacts of species richness on host prevalence in most of the empirical literature provides limited insight into the range of mechanisms by which biodiversity affects disease, rendering it difficult to integrate into public health interventions. Here, we expand the focus to the broader mechanistic relationships among a variety of biodiversity components and the zoonotic spillover process. We then follow with a review of general principles with applied relevance. Finally, we highlight opportunities where ongoing conservation initiatives could consider and possibly integrate these mechanisms further in order to reduce disease spillover risks (Figure 1, Table 1, and Table 2).
Biodiversity encompasses all forms of variability among living organisms and the ecological complexes of which they are a part; these different forms of variability have long been studied and summarized into related but alternative definitions of biodiversity by other ecological fields 17 (Box 1). Change in taxonomic diversity, including species richness, is often an observable outcome of changes in other types of biodiversity, which more explicitly guide conservation efforts such as the loss of functional groups, changes in interaction networks, and heterogeneity in habitat composition. Identifying how these underlying axes drive proximate changes in ecosystem processes like disease transmission is critical for responding to human-mediated (that is, anthropogenic) change [10][11][12][13][14][15][16] . Zoonotic spillover is influenced by many ecological processes before a pathogen actually spills over into a human host. Therefore, changes in biodiversity can mechanistically affect spillover through several pathways including effects on the density, distribution, and susceptibility of reservoir hosts, as well as pathogen prevalence, infectiousness, survival, dissemination, and reservoir host-human contact 18,19 (Figure 1). Once in the recipient (human) host, a series of biological and epidemiological factors determine whether onward transmission is possible [18][19][20][21] (Figure 1). To harmonize spillover prevention and biodiversity conservation, a clear mechanistic understanding is needed of how increases and decreases in multiple aspects of biodiversity, from individuals to populations to communities to ecosystems, influence spillover processes (Figure 1). This review focuses on how infectious-disease systems change with shifts in biodiversity, highlighting case studies that suggest causal mechanisms (Figure 1 and Table 1). We group case studies based on the leading International Union for Conservation of Nature-classified threats to biodiversity. Although examples that mechanistically link environmental change to zoonotic spillover via at least one metric of biodiversity change are scarce, our review identifies emerging generalities across disease systems and anthropogenic disturbances. We find the best support for an influence of functional, interaction, ecosystem phenological, and landscape diversity on spillover risk but recognize that there are additional dimensions of biodiversity not explicitly studied that are likely to influence spillover (for example, genetic diversity 22 ). Within our description of the generalities, we identify ongoing sustainability initiatives that could incorporate spillover prevention, emphasizing how reframing the discussion about biodiversity and disease may facilitate win-win outcomes for health and the environment.
As of 2019, agricultural expansion and intensification were the leading causes of biodiversity loss 17 . Agricultural development both clears and fragments previously intact ecosystems, creating edge habitats that increase human encroachment on wildlife, homogenizing landscapes to reduce availability of natural resources for wildlife, and releasing pesticides, fertilizers, and antimicrobial compounds into the environment. Urbanization, characterized by the presence of built environments, similarly clears intact ecosystems while increasing air, water, light, and land pollution 23 . Moreover, urbanization significantly increases human density: 70% of the world's population is expected to live in urban areas by 2050 24 . All of these factors contribute to population declines or even local extinctions of species [25][26][27] and may influence the dynamics of infectious diseases that have an important environmental component in their transmission cycle 28 .
Clearing intact ecosystems for agriculture, urbanization, and other land modifications (including development of forestry) drives the loss of large-and medium-bodied animals (that is, defaunation) while supporting the persistence or growth of populations of small-bodied animals 29- 32 . Recent research has made it clear that loss of functional diversity (defined in Box 1) due to nonrandom patterns of defaunation has significant effects on zoonotic spillover risk 10,11,16,[33][34][35][36][37][38][39] . Increase in disease spillover risk due to changes in functional diversity of animal communities may occur through the expansion or invasion of opportunistic zoonotic hosts that thrive in humanmodified landscapes or through the cascading effect of human-induced extirpation of predators and competitors of zoonotic species, as described below.
Small-bodied mammals are common pathogen reservoirs, with the rodent and bat orders containing the highest number of known zoonotic hosts [40][41][42][43] . Certain taxa of small-bodied animals are likely to predominate in human-modified landscapes due to traits that make them adaptable to living in proximity to humans 44,45 . These traits, including diet and habitat generalism, fast-paced life history, high population density, and proximity with human settlements are positively correlated with zoonotic reservoir status 12,34,41 . On a global scale, the richness and abundance of zoonotic hosts (especially birds, bats, and rodents) positively correlate with the degree of humanmediated land modification 34,46 . Local studies in Kenya, Tanzania, and Madagascar found that this change in functional diversity, such that communities are dominated by animals with traits conducive to adaptation to human environments, increases zoonotic disease risk: rodent communities in croplands had a higher proportion of competent zoonotic reservoir hosts and higher prevalence of zoonotic pathogens than in unmanaged areas 16,35,47 .
Loss of functional diversity in ecological communities may also be driven by the loss of predators and competitors that help regulate populations of reservoir hosts and vectors. Land conversion can drive the replacement of large herbivores with small herbivores, altering the overall effect of herbivores on the plant community and ecosystem as a whole 33,48 . In savanna ecosystems in central Kenya, exclusion of large herbivores through fencing, an experimental simulation of what often occurs with agricultural intensification, resulted in changes in the plant community and competitive release of small herbivores, leading to the increase in abundance of competent rodent hosts (Saccostomus mearnsei) and prevalence of Bartonella and its vectors 33,49 (Figure 1, Table 1, and Figure 2A). Predators of reservoir hosts and vectors might also exert a crucial role in modulating the risk of disease spillover for humans 10,11 . In Senegal, the construction of the Diama dam in 1986 to prevent saltwater intrusion and support agricultural intensification blocked the migration of a native predator (the giant river prawn, Macrobrachium vollenhoveni) that consumes snail vectors and free-living Schistosoma spp., resulting in increased transmission of vector-borne parasites to humans 36 . These findings have been linked to construction of other large dams as well, and the subsequent increases in schistosomiasis transmission throughout Africa 38 . In terrestrial zoonotic disease systems, the presence of leopards may decrease risk of rabies transmission to humans by preying on stray dogs in Mumbai, India 37 . Further, predator loss can trigger significantly more complex trophic cascades. The loss of wolves in the northeastern United States was followed by an increase in coyotes. This resulted in increased predation by coyotes on mesopredators (such as foxes), leading to a dramatic reduction of predators of small-mammals that control the abundance of rodents that carry Lyme disease 11 . This release of competent rodent reservoir hosts from predation has been linked to expansions in Lyme disease in the last two decades 10,11 .
In general, land conversion for agriculture can affect landscape diversity (Box 1), thereby altering species distributions and changing contact patterns between wildlife and humans [50][51][52] . Landscape diversity can be described as compositional diversity (including patch-type diversity, defined as richness of habitat types among patches) and configuration diversity (including number, size, and arrangement of habitat patches). These aspects of landscape diversity have nonlinear and complex responses to anthropogenic change 53 . As many existing biodiversity initiatives center around land conservation and restoration, including landscape diversity in the biodiversity-disease discussion is crucial for identifying synergistic solutions for biodiversity conservation and preventing zoonotic spillover. Within monocultures, all metrics of landscape diversity are reduced. However, in relation to intact ecosystems, moderate agricultural conversion has various effects on patch-type diversity, decreases patch size and thus variation in patch size, and increases the distance among intact habitat patches [54][55][56] . Fragmenting of habitat into small patches can shift the distribution of reservoir species, causing them to aggregate at high densities near humans and increasing their contacts -with humans, previously unencountered mammals, and vectorsthereby increasing potential for transmission 57 . For example, Plasmodium knowlesi malaria is expanding in Malaysia and across Southeast Asia, partially due to forest loss and agricultural land conversion [58][59][60][61][62][63] . These disturbances drive the primary reservoir hosts, long-tailed macaques (Macaca fascicularis) and pig-tailed macaques (Macaca nemestrina), to occupy small forest fragments within or next to agricultural areas where they overlap with anthropophilic mosquito vectors and people [63][64][65] . This shift in distribution not only increases the density of reservoirs, potentially increasing transmission among reservoir hosts, but also increases potential for macaque-vector-human transmission 63 (Table 1). High profile zoonotic pathogens, such as Ebola virus, may similarly spill over in forest fragments 66,67 , highlighting the links between changes in landscape configuration and diversity on zoonotic spillover risk.
Shifts in landscape diversity that skew functional diversity towards favoring reservoir hosts may also increase the risk of zoonotic spillover by antimicrobial-resistant organisms. Runoff from antibiotic-fed livestock forms wastewater lagoons where diverse bacteria mix. There they face strong selective pressures to evolve and share (via horizontal gene transfer) genes conferring resistances to those antibiotics 68,69 . This also occurs in aquaculture waters 70 , wastewater from antibiotic-treated crops 71 , and effluent from wastewater treatment plants 72 . Wildlife that come in contact polluted waters or soils can pick up these antimicrobial-resistant bacteria and transport them to both neighboring and distant croplands or livestock operations, where they can spill over to people [73][74][75][76][77] . Global rates of antimicrobial resistance are on the rise, driven by the misuse of antibiotics in both clinical settings and agriculture, with an estimated 700,000 deaths worldwide caused by antimicrobial-resistant bacterial infections 78 . Although existing research on wild animal reservoirs of antimicrobial-resistant bacteria is severely limited 79 , initial studies show that animal populations proximate or adaptable to human-modified habitats have higher prevalence of antimicrobial-resistant bacteria than animals with little to no contact with humans 80 , perhaps due to higher host competency and/or exposure rates to these potentially infectious agents. Smith et al. 80 found that the prevalence of antimicrobial-resistant bacteria in agricultural areas decreased as the amount of native habitat increased, possibly due to reduced contact between birds and livestock runoff. As a result, landscape composition and configuration may reduce the likelihood of birds becoming inoculated with and transmitting antimicrobial-resistant bacteria. Landscape diversity may decrease antimicrobial-resistance risk both by protecting croplands from livestock wastewater runoff and by providing vegetation that acts as a natural ecosystem filter 81 . The effect of biodiversity changes on antimicrobial-resistance spillover is severely understudied but warrants significant attention 79,80 , given the threat of antimicrobial-resistant bacteria to global public health 82 .
Land conversion can also reduce the phenological diversity of natural ecosystems and food sources (that is, diversity of temporal or cyclical biological cycles, see 'Ecosystem phenological diversity', defined in Box 1), which can cause nomadic and migrating species to forgo migration in favor of occupying the same habitat year-round. In some cases, formation of resident populations may shift reservoir-host dynamics and alter zoonotic spillover risk, particularly when loss of seasonal, high-quality natural resources is paired with provisioning of non-seasonal, subpar food 83 . For example, the reservoir hosts of Hendra virus, the Pteropus spp. fruit bats, form large nomadic groups that track seasonally abundant nectar sources. Loss of optimal winter resources, at least in part due to habitat loss, drives these animals into small resident groups feeding on permanent, suboptimal food within and around cities 21,84,85 (Figure 1, Figure 2B, Table 1). Not only does this bring these bats into closer proximity to humans but also food stress associated with these suboptimal resources may promote viral shedding, increasing the likelihood of the virus spilling into amplifying hosts (that is, hosts in which a pathogen can rapidly replicate to high concentrations, for example horses in this case) and humans 86 . Similarly, agricultural conversion has limited the availability of high-quality winter resources for elk, which serve as reservoir hosts of Brucella abortus (Figure 2C). Large elk populations are now supported by lower-quality supplemental feeding, which reduces migration and promotes high-density aggregations, thereby increasing the spread of Brucella among these animals and potentially spillover to livestock [87][88][89][90] . Climate change may further exacerbate loss of phenological diversity and interrelated shifts in animal movement, however, this has not been explicitly linked to zoonotic spillover 91 .
Finally, the rural to urban transition that has occurred over time has released local economies from dependence on local agriculture, and opened up trade of goods, services, and ideas with more distant places 92 . Through trade with rural areas, urbanization interacts with other biodiversity threats to drive changes in zoonotic spillover; for example, via introduction of pathogens through the wildlife trade and introduction of invasive species 93 . Drastic reduction of non-human-adapted animals in completely converted land (that is, cities) may reduce the frequency of spillover of novel zoonotic pathogens 94 . At the same time, interactions between urbanization and other anthropogenic disturbances creates circumstances for pathogen introduction, especially if pathogens can be sustained via human-human transmission. For example, urban centers serve as hubs for long-distance shipping, with urban wildlife markets often containing higher densities and diversity of wildlife. Thus, urban wildlife markets create unique assemblages of species, subsequently increasing the likelihood of novel cross-species transmission 95 . Then, in the rare case where the biology of the pathogen allows frequent humanhuman transmission (for example, high infectivity to humans, asymptomatic transmission, aerosol transmission 19 ), the large and dense human population found in cities can facilitate rapid pathogen spread, resulting in large epidemics 94 or even pandemics. Spread of novel zoonotic pathogens may be mitigated by increased health and subsequent reduced susceptibility in affluent urban areas 96 . However, the opposite may be true in urban areas that are unplanned or designed to oppress groups of people (that is, without centralized infrastructure and equitable distribution of resources). In these areas, human health might be compromised by increased pollution, lack of affordable healthcare, and limited access to healthy food and clean water 93,97 .
Species may respond to climate change through phenotypic plasticity 98 , rapid adaptive evolution 99 , and altitudinal and latitudinal range shifts to the edge of their geographic range [100][101][102] . Alternatively, species may undergo local population extinctions, range shifts, or even global extinction [103][104][105][106][107] . Further, the velocity of rising temperatures varies across different regions of the world, affecting species and populations differently 108 . Together these responses can drive biodiversity change in complex, nonlinear, and interdependent ways. Here, we focus on case studies of range shifts in response to rapid anthropogenic climate change, as it is the most immediately observable impact of climate change on wildlife hosts that harbor zoonotic pathogens 109,110 . Plastic, adaptive, and local declines or extirpation responses are currently well researched [111][112][113] , with the amphibian decline being perhaps the most emblematic case 114 , but they are rarely connected to pathogen spillover.
The abundance of different species with certain traits or ecosystem functions (for example, diet, habitat, activity patterns, etc.), and thus functional diversity, may decline with range shifts, especially at high latitudes, although taxonomic diversity (Box 1) of some systems may increase with range shifts [115][116][117] . This is largely attributed to generalists outnumbering specialists in systems impacted by global change, as generalists are able to thrive in a variety of ecological conditions, including human-modified landscapes, whereas specialists need specific resources and/or habitats to survive. At the same time, correlative analyses suggest that zoonotic reservoirs are more likely to be generalist species 34,39,118 , as they are more likely to live in closer proximity to people and contact a wider range of other host species. Further, climate-induced loss of forest habitat may lead to an increase in abundance of extreme generalists with zoonotic reservoir potential, as in the case of the highly adaptable deer mice harboring Sin Nombre virus 119 .
The Alaskan Arctic is currently exhibiting climate-induced shifts in host species, with an increase in the abundance of zoonotic hosts more likely to contact humans. Before contemporary climate change, the ranges of Arctic and red foxes (Figure 3A,B), both of which serve as reservoir hosts for rabies, were separated 120 . However, with climate change, the home range of the generalist red fox has expanded northward, encroaching on the territory of the Arctic fox, which is more of a habitat specialist 121 . Arctic fox numbers were already in decline due to other effects of climate change, such as the loss of sea ice and tundra habitat as well as loss of lemming prey, but red foxes are expediting this decline through intraguild predation and competition for resources [122][123][124] . As Arctic fox populations are replaced by red fox populations, the red fox will become the primary reservoir for rabies spillover. As immigrant red foxes increasingly interact with resident Arctic foxes, this shift in the reservoir community will likely increase epizootic peaks of rabies, increasing both the transmission rate and the overall density of susceptible individuals 125 . Further, because the larger-bodied red fox displays more aggressive behavior than the Arctic fox 120 , and because it is more adaptable to human-dominated landscapes, contact rates between wild rabies reservoirs and dogs or humans might increase, thus increasing rabies spillover risk (Figure 1, Table 1, Figure 3A,B).
Climate change may reduce other dimensions of biodiversity beyond functional diversity. For instance, climate change may reduce landscape diversity by reducing patch diversity and subsequently increase the likelihood of cross-species transmission through increased habitat overlap and taxonomic diversity in confined areas 126 . For instance, the melting of sea ice alters, disrupts, or even prevents migration patterns of animals such as wild caribou 127 , increasing the chance of intermingling among caribou and with other wild or domestic ungulates. Thus, people who rely on caribou and/or other livestock might be at higher risk of brucellosis spillover under a warming climate in temperate regions 128 . Similarly, in water-stressed parts of Africa, extreme droughts can force many animals -that may previously have used different water bodies and had little to no contact with one another, such as humans, wildlife, and livestock -to congregate at common water sources 129,130 (Figure 3C), increasing traffic and reducing water quality due to elevated fecal loads. In Chobe National Park, Botswana, these patterns and processes are associated with increased loads of Escherichia coli, the leading cause of diarrheal outbreaks 130 . Following drought events in and around Chobe National Park, heavy seasonal rainfall and flooding mobilize pathogen-containing feces, subsequently leading to human diarrheal outbreaks in neighboring communities 131 (Table 1). Further, these water sources have the potential to serve as melting pots of antimicrobial resistant bacteria and sources of novel pathogen emergence 132 .
Invasive species (that is, organisms that establish and spread outside their native range) present a significant threat to ecosystems and human well-being by negatively impacting native biodiversity and ecosystem services 133 . Through processes such as predation, competition, and environmental modification, invasive species can drastically decrease the biodiversity of an ecosystem; an estimated 30 species of invasive predators alone are responsible for at least 58% of all bird, mammal, and reptile extinctions globally 134 . Invasive species can indirectly impact infectious disease by altering the structure and composition of the native community in ways that either increase or decrease pathogen transmission.
Altering a native community in a way that increases zoonotic spillover risk has been empirically demonstrated for the Everglade virus, a mosquito-borne zoonotic virus. The introduction of the Burmese python (Python bivittatus, Figure 2D) to the Florida Everglades has led to large-scale declines in functional and taxonomic mammalian diversity due to predation and subsequent precipitous loss of large and small-bodied mammals 135,136 . With the loss of deer, racoons, and opossums as food sources for blood-sucking arthropods, mosquito vectors of Everglades virus turned increasingly to the primary reservoir host of the virus, the hispid cotton rat (Sigmodon hispidus). This has resulted in higher rates of Everglade virus infection in mosquitoes, potentially increasing the risk of virus exposure to humans 136,137 . The Burmese python-Everglade virus case study is a clear example of the dilution effect: higher taxonomic diversity of hosts reduces disease risk because the vector takes 'wasted bites' (from a pathogentransmission perspective) on non-competent hosts. The loss of taxonomic diversity therefore increases disease spillover risk, with the dilution effect most commonly occurring for vectorborne, zoonotic pathogens, as is the case here 9 .
In contrast, introduction of invasive species can sometimes reduce transmission of infectious disease from vectors to people through predation on various vector life stages: for example, larvivorous fish preying on malaria vectors 138 and crayfish consuming schistosome intermediate hosts 139 . However, despite crayfish lowering the risk of schistosomiasis by voraciously consuming snail intermediate hosts and free-living parasites, invasive crayfish compromised other dimensions of human health by consuming rice and degrading canal banks with their burrows 140 . Consequently, in scenarios where invasive species reduce disease risk there can still be a tension between biodiversity impacts of invasive species and their specific ecological roles in infectious-disease dynamics.
Invasive species may also affect infectious disease dynamics by acting as vectors or reservoir hosts 40,47,[141][142][143] , sharing pathogens with native species [144][145][146] , or providing resources for reservoirs and/or vectors 143,147 . In these cases, biodiversity conservation via invasive species control may simultaneously reduce zoonotic spillover risk 143 . The same processes that drive species introductions, including global trade and travel, may also drive disease emergence, suggesting that win-win solutions for protecting ecosystems from species invasion and humans from pathogen spillover might be possible, albeit potentially challenging from a technical or political perspective 148 .
One in five vertebrate species is impacted by trade 149 , with some wildlife facing population declines and/or species extinction due, mainly or in part, to the impacts of wildlife trade -some legal but primarily illegal (for example, tigers, rhinoceroses, elephants, sharks, and pangolins) 150,151 (Figure 3D). The illegal wildlife trade is estimated to be the world's second largest underground businesses (hypothesized to be a 5-20 billion-dollar industry) after narcotics 152 . The legal wildlife trade, estimated to be an even larger business (300 billion-dollar industry), also poses a threat to biodiversity as the majority of legal wildlife trade (78%) is composed of wild-caught animals, as opposed to those reared in captivity 153 . The local increase or decrease of biodiversity, as well as novel contacts made during translocation and trade between species that do not co-occur naturally in the wild, may drive spillover and disease emergence, as explained below.
Epidemiological and genetic analyses have linked wildlife hunting, trade, and consumption to spillover and spread of many high-profile zoonotic pathogens: rabies virus, Crimean-Congo hemorrhagic fever virus, the plague-causing bacteria Yersinia pestis, monkeypox virus, coronaviruses, HIV, Marburg, and Ebola viruses 150,151,[153][154][155][156] . However, in order to stop or mitigate the spillover process, we need to better understand the mechanisms linking the wildlife trade to the eco-epidemiological process of spillover (Figure 1).
The wildlife trade highlights how anthropogenic pressures can increase spillover risk via a direct increase in both taxonomic diversity and the number of interactions across taxa on very small spatial scales (see 'Interaction diversity' defined in Box 1). Throughout the supply chain, the wildlife trade brings together high densities of species that typically would not contact each other in natural habitats. These unique assemblages and interactions can promote cross-species transmission, increasing the likelihood that a pathogen may be transmitted to amplifying hosts and/or humans 154,[157][158][159][160][161][162][163] . Trade may also impact the spillover process by promoting pathogen shedding from animals because of stress during transportation and unsanitary conditions at markets 154,[157][158][159][160][161][162][163] . For example, the ancestor to SARS-CoV-1 is suspected to have been transmitted from horseshoe bats (most likely Rhinolophus sinicus) to palm civets, two species that do not interact in wild settings. However, palm civets served as amplifying or intermediate hosts within wildlife markets, bringing the virus in closer proximity to humans [164][165][166] . Seroprevalence and virological testing surveys of civets on farms versus those brought to markets in Guangdong, China suggest that palm civets were exposed to the virus at the end of the supply chain [165][166][167] . In a study in Vietnam, the prevalence of coronaviruses in field rats caught or reared for human consumption and sold in markets was more than double than that of field rats in the wild 162 . Further, coronavirus prevalence was ten times higher in field rats sold or served in restaurants compared with field rats in the wild 162 . Thus, the wildlife trade creates opportunities for increased transmission among multiple wild animal species and puts humans in closer proximity to stressed and infected wildlife, fueling the potential for spillover of pathogens (Figure 1, Table 1).
The wildlife trade for human consumption can take on various forms, including commercial harvesting of wild animals on land and at sea. Together, these interact to amplify the effects of overharvesting, leading to a decrease of many types of biodiversity, such as taxonomic, genetic, functional, interaction, and landscape diversity (Box 1). For example, the wild meat trade in Ghana, which has driven population declines of some mammalian species in the last few decades, correlates with local declines in fish supply, probably due to commercial overfishing off the coast 168,169 . Conceivably, during periods when the demand for wild meat is high, hunters and people involved with butchering and preparation are at a higher risk of disease spillover from bites, scratches, and contacts with bodily fluids of animals that serve as pathogen reservoirs. In the Congo basin and other regions where pathogens have recently emerged, wild meat serves as an important protein source in impoverished households. This makes the banning of wild meat a controversial topic 170 even though genetic and epidemiological evidence suggest that wild meat consumption has contributed to the rise of emerging diseases and recent outbreaks via spillover from wildlife to humans of pathogens like Ebola (Table 1), HIV, Marburg, and monkeypox viruses 154,171,172 . In Cameroon, simian foamy viruses regularly spill over and infect wild meat hunters, but no human-human transmission has yet been established 154 . Conversely, although HIV has adapted to undergo human-human transmission, phylogenetic analyses suggest that approximately ten spillover events occurred over the past century before it eventually evolved to cause a pandemic, suggesting that frequent spillover during bushmeat hunting was critical for its emergence 151 .
Overexploitation of wild meat and other anthropogenic pressures have also been correlated with a decrease in the proportion of large-bodied mammals and an increase in the proportion of small-bodied mammals brought to market 173,174 . As a result, preliminary research suggests that overharvesting of wildlife may influence the types of wild animals that hunters and consumers are contacting, potentially presenting new zoonotic spillover risks. However, mechanistic links between change in composition of wildlife markets and zoonotic disease risk have not yet been established.
Although mechanistic research linking changes in biodiversity to zoonotic spillover risk is limited due to expense and logistical challenges, by considering more mechanism-based changes in biodiversity, we collect enough empirical examples to propose four general concepts that have potential to inform biodiversity conservation. These generalities may motivate further integration of biodiversity and zoonotic pathogen spillover research, potentially opening more avenues of funding and facilitating the incorporation of multi-disciplinary methods for collecting and analyzing data. To illustrate this application of our synthesis, we identify ongoing biodiversity and sustainability initiatives that could use these generalities to incorporate spillover prevention. Using the framework we propose may, for example, help to avoid unintended harms from biodiversity conservation or broaden the benefits of biodiversity conservation. Echoing Halsey 8 , we distinguish between generality, that which is mostly considered true, and universality, that which is considered true in all possible contexts. These four generalities (described below) may be more or less applicable for different ecosystems and disease threats.
Loss of spatially and phenologically diverse habitat alters the spatiotemporal distributions of reservoirs, leading to increased overlap with other vertebrate hosts, vectors, and humans. This generality suggests an opportunity: preserving and restoring large, contiguous, and heterogeneous habitats could minimize harmful contact between humans and wildlife and between host species that do not commonly interact (for example, a reservoir and an amplifying host). Such an approach might additionally reduce the density of reservoir hosts and subsequent intraspecific contact and transmission. The Bonn Challenge 175 , Thirty-by-Thirty Resolution to Save Nature 176,177 , Payments for Ecosystem Services [178][179][180] , and Project Finance for Permanence projects [181][182][183] all include conservation and/or restoration of natural ecosystems but do not incorporate spillover prevention in project design and implementation (Table 2). Intact and diverse contiguous landscapes may also promote landscape immunity, defined as ecological conditions that maintain and strengthen immunity in resident fauna so as to reduce pathogen susceptibility and shedding, particularly for potential reservoir species including bats and rodents 184,185 . Further, targeted habitat conservation and restoration could encourage previous migration patterns by re-creating or maintaining phenological diversity of high-quality food sources, such as nectar resources for bats 21,143 . However, in some cases, resource provisioning -through invasive species, crops, and even waste disposal practices -may reduce migration even when endemic, phenologically diverse habitats are available 186,187 . More research differentiating the impact of habitat restoration versus limiting human provisions is needed. Importantly, some biodiversity conservation initiatives such as Payment for Ecosystem services in Costa Rica 179 include agroforestry, which could hypothetically increase human exposure risk to zoonotic disease 188 . In these cases, the effect of landscape diversity and specific agroforestry practices on spillover should be considered so as not to put biodiversity conservation and public health at odds. Overall, studying the mechanistic effect of landscape diversity and ecosystem phenological diversity on each spillover process (Figure 1) should lead to new insights that can guide evidence-based policy for both conserving natural ecosystems and reducing spillover risk.
Loss of large consumers and predators (changes in functional diversity) can result in increased abundance of animals with fast growth rates and relatively small ranges, such as rodent reservoirs and arthropod vectors. Regulation of poaching (for example, via the Convention on International Trade in Endangered Species 189 initiative) and habitat conservation, preservation, and restoration of contiguous, intact ecosystems could support populations of large predators and herbivores 174,190,191 . In turn, predators and large consumers may be important in ecotones between intact and anthropogenic landscapes, where they can regulate populations of small-bodied reservoirs that thrive in human-modified areas. The initiatives aimed to restore and conserve habitat in Table 2 could be adapted to support populations of wildlife that help regulate rodent populations. For example, the Thirty-by-Thirty Resolution to Save Nature 176,177 proposes conservation of wildlife habitat and corridors for safe passage of wildlife between intact habitats. This plan could be improved by configuring habitats and corridors to best support populations of keystone predators and large consumers in areas of zoonotic disease risk. More research is needed to understand the impacts that large herbivores and predators have on zoonotic disease regulation, especially within and around ecotones. If more evidence supports a beneficial effect of conserving predators and large herbivores for reducing spillover risk without increasing human-wildlife conflict, conservation of predators and large consumers may offer another promising solution.
Human modification further affects functional diversity by changing habitats and shifting communities toward dominance by species that are resilient to anthropogenic disturbance or thrive in human-dominated landscapes. Change in functional diversity towards such 'synanthropic' species has been observed across taxonomic groups of vertebrates including rodents, birds, and carnivores. Similar effects have been observed for disease vectors: generalists thrive in urban areas and have high capacity to transmit pathogens to humans 38,192,193 . Integrative approaches, such as direct management of invasive rodents and vectors or indirect management through preserving intact habitat and mitigating impacts of climate change to reduce range shifts of reservoirs and vectors, are likely necessary 143,194 . Initiatives that guide policy and coordinate action to protect biodiversity from multiple anthropogenic threats, such as the Convention on Biological Diversity 195 , may be particularly well suited to prevent spillover from these human-adapted reservoirs and vectors. For example, the Convention on Biological Diversity sets global priorities and coordinates global action on invasive species and climate change, providing a platform to jointly manage invasive reservoir hosts and vectors while advocating for climate resilient ecosystems on a global scale.
Commercial wildlife trade, introduction of invasive species, and transportation of livestock and companion animals are activities that increase interaction diversity, introduce more opportunities for cross-species transmission, and facilitate the emergence of new pathogens with zoonotic spillover potential. The Convention on International Trade in Endangered Species 189 aims to control the illegal wildlife trade but does not include objectives that prevent spillover. Adopting global regulations on pathogen screening and ethical and sanitary animal husbandry standards in the international wildlife trade could be a natural next step in advancing management of zoonotic spillover. Overall, regulations and initiatives that reduce diversity of novel interspecies interactions should be adjusted to incorporate spillover prevention.
Other international initiatives are currently working towards sustainable solutions for promoting both public health and conservation, such as the UN Sustainable Development Goals 196 201 , and the collaboration among Food and Agriculture Organization (FAO), World Organisation for Animal Health (OIE), and WHO (FAO-OIE-WHO Collaboration) 202 . The initiatives included in Table 2 have not yet incorporated spillover prevention.
We emphasize that the initiatives described here must only be implemented based upon local context, centered around the needs, demands, and culture of the local people. A number of global restoration and conservation efforts have been criticized as colonialist and thus detrimental to vulnerable and marginalized groups of people. For example, the Bonn Challenge has been criticized for foresting historically savannah ecosystems, thereby impacting ecosystem function and rangeland livelihoods 203 . The Payment for Ecosystem services in Costa Rica has been rebuked as not adequately compensating people for the service they provide 204 . Further, Thirty by Thirty has been challenged for disproportionately, negatively impacting Indigenous communities via exclusion from land ownership and management, despite the outsized, positive effect that some Indigenous practices have had on biodiversity 205 . These initiatives may be improved by creating context-dependent management plans that are designed around and implemented by local communities and Indigenous groups. One way to achieve this is through conservation of land via Indigenous Protected Areas: although defined differently depending on the country, Indigenous Protected Areas are generally large areas of intact ecosystems managed or co-managed by Indigenous groups. More than 46% of national reserves within Australia are Indigenous Protected Areas 206 , and a small but increasing proportion of protected land in Canada is comprised of Indigenous Protected Areas (for example, Thaidene Nëné Indigenous Protected Area, the homeland of the Łutsël K'é Dene First Nation) 207 . The United States and countries with similar Thirty by Thirty goals can and should create similar protected areas. Another successful model is Health in Harmony's programs in Borneo, Madagascar, and Brazil, which start with 'radical listening' within rainforest communities to co-develop community-based conservation and health programs that reduce deforestation and provide affordable healthcare access 208 .
We additionally emphasize that biodiversity conservation is not a panacea for zoonotic spillover prevention, and many systems are too complex or understudied to define clear links between biodiversity change and spillover risk. For example, highly diverse multi-host, multivector systems such as West Nile Virus, Ross River virus 209,210 , leishmaniasis 211 , and Chagas disease 212 , require more studies to document ecological drivers of reservoir and vector abundances and capacities to transmit disease. Further, reservoir host species that contribute most to transmission may be variable along geographic and land-use gradients [213][214][215][216][217][218] . Even when conservation-related levers for spillover prevention exist, their impacts should be compared to those of other approaches (including economic and biomedical) and implemented from a community-based, environmental-justice perspective. Thus, sustainable solutions for alleviating zoonotic disease burden while conserving biodiversity should be evaluated based on specific knowledge of the socio-ecological context 1 .
We identified mechanistic evidence in the literature that anthropogenically driven biodiversity change may increase zoonotic spillover risk. Several common themes emerged. First, the loss of intact habitat increases overlap between reservoirs and other vertebrate hosts, vectors, and humans. Second, loss of large-bodied consumers and predators (defaunation) can result in increased abundance of rodent reservoirs. Third, human-modified landscapes change the functional diversity of species assemblages, increasing the proportion of species that are able to adapt and thrive in anthropogenic environments, and thereby increasing human exposure to zoonotic pathogens. Fourth, other forms of anthropogenic disturbance generated by agriculture and trade of domestic animals and wildlife lead to the introduction of invasive species and increase interaction diversity, facilitating opportunities for cross-species transmission and thus the potential for emergence of novel pathogens with zoonotic spillover potential. Hence, anthropogenic drivers of biodiversity change interact in complex ways, including synergies and both direct and indirect effects. The combined impacts of many different anthropogenic disturbances may exacerbate the effects of biodiversity change on spillover risk.
Certain disease systems are either understudied or too complex to elucidate the effects of biodiversity changes on spillover risk. In addition, some components of the spillover process (Figure 1) are better studied than others in this context. Based on our review, the effects of biodiversity changes on wildlife-host susceptibility, pathogen shedding, and pathogen prevalence in the reservoir are three important steps of spillover that are understudied and warrant further investigation. These aspects are difficult to observe 219 , but another possible reason that they have been understudied could be a lack of appreciation for how wildlife health -and not just presence or absence of disease agents -may affect zoonotic spillover risk. When exposed to stress from anthropogenic activities, wildlife hosts may experience suppressed immunity, rendering them more susceptible to opportunistic infections, more pathogen shedding, and altered behavior that further increases their exposure to pathogens 185,220 . Thus, increased pathogen surveillance and health assessments of wildlife may be useful for understanding mechanisms by which environmental stressors affect wild animal health and lead to changes in the process of disease spillover to people and domestic animals. Finally, there is an urgent need for spatially and temporally replicated field studies incorporating biodiversity change, pathogen dynamics, and wildlife host immunology 184,185 , in addition to human health outcomes.
The world is undergoing rapid anthropogenic change with detrimental effects on biodiversity and the health of organisms, including humans. Efforts are underway to combat the impact of anthropogenic disturbances on biodiversity. However, since biodiversity change may affect zoonotic disease spillover through multiple mechanisms, it is imperative that biodiversity conservation efforts also incorporate actions to prevent spillover. Spillover is an issue not only for public health, but also for conservation of threatened wildlife. Here, we argue that reframing discussions of biodiversity and disease around a more inclusive definition of biodiversity and considering the context of each of the complex social-ecological systems in which the spillover process occurs (Figure 1, Box 1) are essential to highlight mechanistic links between biodiversity and zoonotic spillover. This approach sheds light on how to develop sustainable interventions that prevent zoonotic spillover while protecting biodiversity-to the benefit of both humans and the environment.
There are a number of dimensions that comprise 'biodiversity', each with multiple axes affecting zoonotic spillover risk. Below are a handful of examples described by Naeem et al. 22 , with suggestions for how to measure and track each aspect using the universally developed Group on Earth Observations Biodiversity Observation Network's essential biodiversity variables (EBVs) 221 .
• Genetic diversity includes aspects of genomic variability, including nucleotide, allelic, chromosomal, and genotypic variation. Genetic diversity has yet to be studied in the context of biodiversity change and zoonotic disease risk; however, multiple reviews 14,15 have described how observable patterns in taxonomic diversity are likely, at least in part, the result of genotypic variation governing phenotypic variation in host physiology and behavior (that is, host resistance, tolerance, and competence) and thus can influence zoonotic disease risk. EBVs: Intraspecific genetic diversity, Genetic differentiation.
• Taxonomic diversity refers to the number and relative abundance of taxa (for example, species, genera, and higher levels of taxonomic organization). Disease-diversity relationships are typically described within the context of species richness. One examples relevant to spillover is an increase in diversity of host species, so that vectors take 'wasted bites' on non-competent hosts. In many cases, change in taxonomic diversity per se does not influence zoonotic disease spillover; however, change in the other dimensions of biodiversity are evident through changes in taxonomic diversity. EBVs: Species distributions, Species abundances, Community abundance, Taxonomic/phylogenetic diversity.
• Functional diversity refers to the variation in the degree of expression of multiple functional traits: that is, the different types of processes in a community that are important to its structure and dynamic stability. Examples relevant to spillover include loss of predators and competitors and increase in abundance of generalist, synanthropic animals. EBV: Trait diversity.
• Interaction diversity refers to the number and relative abundance of interactions among species in a community 222 . These biotic interactions include contact, competition, facilitation, and predation. Examples relevant to spillover include a loss of interactions regulating reservoir host species or an increased number of novel cross-species interactions via crowding. EBV: Interaction diversity.
• Ecosystem phenological diversity is the diversity in the phenological dates of life within an ecosystem (for example, flowering time). Phenological diversity is a subset of temporal diversity, which is broadly thought of as change in biodiversity over time. An example relevant to spillover is the reduction in the seasonal availability of resources, which in turn affects sedentary movement and eating habits. EBV: Phenology.
• Landscape diversity * is composed of compositional and configuration diversity. Landscape compositional diversity includes diversity of habitat patches, and configuration diversity includes the number, size, and arrangement of habitat patches. An example relevant to spillover is an increase in the number of reservoir habitat patches while decreasing their size, thereby providing increased opportunity for reservoir host-human or host-vector contact. * Note that landscape ecologists commonly refer to 'landscape diversity' as 'heterogeneity'. EBVs: Live cover fraction, Ecosystem distribution. Table 1. Case studies of mechanisms connecting anthropogenic disturbance with biodiversity change and the subsequent effects on infectious disease spillover. Figure 1 illustrates the overall framework for linking anthropogenic disturbance to biodiversity change to disease spillover via the spillover layers being affected in each case study. Several initiatives are listed along with the four generalities discussed in the main text section "Incorporating concepts of ecological diversity to mitigate spillover risk" that may be considered applicable. Generality numbers in the tables refer to: 1) Large, intact habitat reduces overlap among host species and promotes wildlife health; 2) Loss of predators and competitors reduces regulation of reservoir host and vector populations; 3) Reservoir hosts are better adapted to human-modified systems; and 4) Human activity may increase opportunities for novel interspecies contacts. To understand mechanisms connecting anthropogenic disturbance with spillover via biodiversity change, it is imperative to investigate how anthropogenic disturbance impacts biodiversity, and how those effects drive the perforation of the layers (intermediate processes) leading to spillover (shown using four case studies from Table 1 as examples). Zoonotic spillover arises from the alignment of multiple processes (depicted as layers). Apart from human susceptibility to infection, we found that each layer can be affected by biodiversity change, especially when considering biodiversity along multiple axes (Box 1). Connecting biodiversity change to explicit processes helps us to better understand how, when, and why biodiversity change impacts zoonotic disease risk. Numbers next to each layer correspond to the eight case studies highlighted in Table 1. All references for these case studies are included in Table 1. (A) The competent rodent host species (Saccostomus mearnsei) of Bartonella in Kenya (image courtesy of Hillary Young). Reduced functional diversity, due to loss of large consumers and driven by agricultural expansion and intensification, increases rodent richness and abundance and thus Bartonella spillover risk. (B) The natural habitat of the flying fox (a fruit bat of the genus Pteropus) is threatened by land conversion and urbanization (reducing ecosystem phenological diversity), which in turn aggregates flying foxes at higher densities in urban areas and brings humans into closer proximity with these natural reservoirs of Hendra virus (photo by Elizabeth Shanahan). (C) Supplemental feeding of elk (Cervus canadensis) during winter months in Yellowstone National Park (image courtesy of United States Geological Survey). Agricultural conversion of land in North America has limited the availability of natural winter-feeding grounds for elk (reduced ecosystem phenological diversity). Large populations are dependent on supplemental feeding, reducing migration and promoting high density aggregations, thus increasing the risk of brucellosis spillover to livestock and humans. (D) A Burmese python (Python bivittatus) in the Everglades in Florida, USA (photo by Anne Devan-Song). This invasive species has reduced biodiversity in the Everglades (taxonomic, functional, and interaction diversity), thereby increasing the rate at which vectors feed on competent hosts of Everglade virus and thus spillover risk in this region. (A) An Arctic fox (Vulpes lagopus) in Alaska (image courtesy of Alaska Department of Fish and Game). Climate change may increase functional diversity in polar and temperate regions as native fauna, such as the Arctic fox, is being replaced by northwardly range-shifting species, such as the red fox (Vulpes vulpes) (B) (photo by Peter Hudson). This could potentially increase rabies spillover to humans in Alaska as the red fox is generally a more human-landscape adaptable reservoir species. (C) Several species aggregating around a small water hole in southern Africa (photo by Nick Fox). In the tropics and sub-tropics, climate change is reducing water availability, which may increase taxonomic and interaction diversity. This in turn could increase spillover risk of E. coli as more hosts start to share common water resources. (D) Elephants in Tarangire National Park, Tanzania, protected from poaching (photo by Peter Hudson). The wildlife trade is reducing wild elephant populations and other large-bodied animals, thereby decreasing biodiversity (taxonomic, genetic, functional, interaction, and landscape diversity) and leading to a higher demand for meat from small-bodied mammals such as bats, potentially increasing spillover risk of Ebola and other disease borne by small mammals.
Glidden et al. review mechanisms by which biodiversity change-driven by anthropogenic disturbance on the environment-influences disease spillover risk by considering a suite of biodiversity metrics. Finally, this review summarizes general principles that could be used to integrate spillover prevention into biodiversity conservation initiatives.