Food systems are often related to human health via the impact of unhealthy diets on the emergence of chronic diseases such as type 2 diabetes and cardiovascular syndromes. However, they can also affect human health more indirectly, through land use changes induced by the increasing demand for food commodities such as meat or other animal products, a phenomenon known as “the livestock revolution”17. In many regions of the world the growth in livestock production has often led to the expansion of agriculture, forest destruction, and the encroachment of cropland and concentrated livestock farms into disturbed wildlife habitat14. This study connected the dots between the emergence of SARSr-CoV epidemics and land use changes resulting from both intensive farming and agricultural expansion. Our approach uses the horseshoe bats as a model family because of their key role as hosts of Sarbecovirus coronaviruses, which have caused SARS, COVID-19, and SADS2,9,25,27,28,29,30. Other strains of related viruses have been found in other bat genera, but these relationships are less clear28,29,30. The widespread sampling of other bats may find species-specific relationships, though horseshoe bats appear to be the reservoirs where most SARSr-CoVs have their evolutionary ancestors so we assume they are the most appropriate model. The risk to humans from other coronaviruses, therefore, will be different, because their host distributions are different, and two CoV genera (gamma- and delta-coronaviruses) are mostly bird viruses.
The bat location data and species distribution data also suffer from different, but related issues. The bat location data are presence only data. True absence data are difficult to obtain, therefore we randomly sampled within different locations to generate pseudo-absence data. Choosing where to sample from also present difficulties, so we chose horseshoe bat distribution data for species that existed within China, East Asia, South, and South East Asia. This presents further issues as the distribution of one species, the greater horseshoe bat (Rhinolophus ferrumequinum), extends from Western Europe, Northern Africa, Central Asia and Eastern Asia. We therefore weighted our sampling based on the number of overlapping species distributions to account for this. However, these species distributions are large polygons and the realized niches used within them by the species likely differ, so better niche models using presence and, ideally, presence/absence data are required to develop better species presence predictions33. However, our results for random locations in China and outside China and reported bat observations were comparable.
More generally, though using the relatively specific bat and virus relationships, we took a high-level approach to understand the more distal or ultimate (rather than proximal) causes of infectious disease emergence in China, linking environmental change and human drivers like agricultural intensification. Different infectious diseases have different transmission mechanisms and life cycles, and not all will respond to such changes in the same way. For example, directly transmitted, acute infections with short incubation and infectious periods, like SARSr-CoVs, will likely be dependent on hosts having greater densities, like in China, for them to emerge. The epidemic potential is then also increased through local and global movement and trade, either of people, wildlife, or livestock20,34,35,36. Along with the biological properties of the virus and hosts, the true risk of both the initial cross-species transmission and epidemic potential is either increased or limited by more proximal mechanisms, such as biosecurity, health and safety measures (e.g. personal protective equipment, meat hygiene) that can reduce risk, even if the ultimate factors are present and increasing through the processes of habitat fragmentation and human encroachment20,2 .
Spillover of infectious disease such as SARS, COVID-19 and SADS from wildlife to humans likely requires the coexistence of horseshoe bats and humans in the same environment and is favored by the presence of intermediate animal species, particularly livestock because it is in closer contact with humans. The fragmentation and disturbance of forest ecosystems likely favors habitat generalist bat species. In particular, chickens, ducks, and pigs have been associated with the spread of several zoonotic viral infections, such as influenza viruses. This study demonstrates that in China these important factors responsible for reducing the distance between wildlife and humans co-occur both in horseshoe bat distributions and in the surroundings of actual documented bat occurrence. These results are consistent with the notion that population growth and increasing meat consumption associated with urbanization and economic growth have expanded the footprint of agriculture, leading to human encroachment in wildlife habitat and increased livestock density in areas adjacent to fragmented forest patches. China has dramatically increased animal consumption37, likely as the result of increasing affluence. In China, meat supply is largely reliant on domestic production using imported feed (e.g. soy from the Americas)37, which explains the high livestock density in many rural areas, including those at the forest margins. Likewise, economic growth and the shift to diets richer in animal products explains the increasing demand for wild animal meat delicacies, increasing human-wildlife interactions through multiple pathways and the disturbance of forest habitat in more remote locations – frequently abroad – through trade-related teleconnections38.
The multivariate hotspot analysis highlights how China is the largest hotspot for the concurrence of high forest fragmentation, livestock density, and human presence in our analysis (Fig. 3). The sensitivity analyses identifying the possible transition to new hotspots in response to an increase in one of these attributes (Fig. 4b) highlights areas that could become suitable for spillover and the type of land use change that could induce hotspot activation. Therefore, this analysis highlights region-specific targeted interventions that are urgently needed to increase resilience to SARSr-CoV spillovers. For instance, the green dots in Fig. 4 could be turned into hotspots as a result of forest fragmentation. In these regions resilience can be built through forest conservation or restoration efforts. Indeed, land use change evaluations should consider the risk of activating new hotspots suitable for wildlife-to-human spillover of pathogens such as SARSr-CoV, an aspect that has seldom been included in the impact analysis of land use change. Likewise, other regions such as the China-Indochina transition zone or central Thailand are prone to hotspot transitioning as a result of increased livestock density of urbanization, respectively. In these cases, mitigation of SARSr-CoV emergence can been enhanced by reducing livestock or human density, respectively, thereby inverting ongoing dietary and urbanization trends. Thus, environmental health, is tightly connected to both animal and human health, as recently stressed by planetary and ‘one health’ discourses, which advocate for more holistic views of global health, encompassing environment, animals, and people as well as their interactions39.