Ecological Constraints on Plague Reservoirs

The primary objective of this project was to identify where plague, a wildlife rodent disease that has caused three pandemics, could survive in the wild and to understand why it could do so in those places. We undertook this project because we still poorly understand why plague occurs where it does: for example, we find plague in the American and Asian mountain ranges, but not in the European Alpes. In fact, at the start of this project, we estimated that the plague maps produced by the WHO overestimate its geographic extent by 50%-90%, based on highly detailed maps from the former USSR. The project planned to employ ecological niche models and a newly compiled dataset of wildlife and human plague cases from 1950 onwards to improve on the current global and regional maps. Additionally, the project planned to determine where plague might have persisted in wildlife in the past. For that we would train a model with contemporary plague and climate data, and use the model to make predictions based on a climate reconstruction of the last 6,000 years. These predictions of historic plague reservoirs would then be compared against a newly compiled historical database of plague outbreaks, and be of great help to paleo-geneticists and historians who try to piece together the history of plague. Furthermore, the project aimed to analyze factors that impact the presence of plague in humans at a more granular level. While such detailed data isn't globally available, it is accessible for several plague reservoirs. This data enabled the project to examine a broader range of factors, including rodent and flea densities. During the project's timeframe, we developed a new ecological niche model for the western United States (published). This model, based on Bayesian additive regression trees, predicted both where plague could persist in rodents and where it could spill over to humans. We discovered that high elevation and species richness correlate positively with plague presence in the United States, while soil properties like calcium and sodium content correlate negatively. Other influential factors, such as annual temperature and iron and sand content of the soil, showed different patterns in the wildlife plague and human spillover models, emphasizing that the presence of plague and its spillover to humans are two different processes! Because our dataset for the western United States spanned seven decades, we could also see how climate change has made the US landscape more suited for plague reservoirs. After that first model, we successfully compiled a global dataset of wildlife and human plague outbreaks from open sources, with 9,000 observations. We fine-tuned the Bayesian models to account for sampling biases in these observations and produced our inaugural global plague map. We also assembled a dataset of historical plague outbreaks (published) that recorded 14,000(!) historical instances where plague reportedly affected humans, and we recalibrated the climate reconstruction data from the IPLS-CM6 global climate model, thus putting in place all the piece we need to predict historic plague reservoirs. We also conducted a detailed analysis of plague spillover using a generalized additive model, focusing on Yunnan, south-west China, where the project gained access to a 30-year dataset of wildlife and domestic rodent and flea surveillance. A significant finding was that plague spillover primarily occurred during periods where the wildlife rodents themselves went through their own plague epidemics, in contrast to periods where the disease was present at lower levels. This finding emphasizes the need to understand what factors influence epidemics in wildlife species. While the project has concluded, several components are nearing completion and will be continued in future research. The regional-scale statistical analysis of plague spillover will be finalized shortly, followed by the global and historical plague maps, together with the datasets.

Two grant products, "Plague risk in the western United States over seven decades of environmental change" and "Mapping the plague through natural language processing", are available to the public. Three others, including a global map of plague reservoirs, a statistical analysis of plague spillovers, and historical maps of plague reservoirs, are in progress. In 2022 we published a detailed prediction of current plague reservoirs in the western United States and the impact of climate change on their locations over 70 years. The paper received significant attention and gained citations from diverse topics. The paper reaffirms the IPCC's latest synthesis report on the increasing risk of infectious diseases due to climate change, but our current paper suggest that the regions at risk should probably be extended. Upon release of the global plague map, we anticipate it reaching a broad audience interested in wildlife disease systems. The map will assess regional variations in plague reservoir suitability and potental expansion areas. Our findings from the U.S. study indicate that human cases of plague are concentrated in specific endemic areas. This underscores a significant gap in our understanding of the conditions that lead to spillovers. While many diseases transition from wildlife to humans, plague and its associated vectors and hosts are monitored far more intensively than most, and our upcoming research on plague spillovers can therefore provide a unique insight into the conditions that foster these spillovers. These findings will serve as a significant case study for the global spillover of various wildlife diseases. More studies are needed to fully grasp the general principles behind spillovers, but our results suggest that increased attention should be given to wildlife disease epidemics. If confirmed, this understanding offers actionable strategies to reduce the risk of spillovers to humans. Our research paper, titled "Mapping the plague...", yielded two publicly available datasets. Given its technical nature, the paper did not immediately garner extensive public interest, but the datasets are picked up by the research community. These dataset are essential for the evaluation of our project's culmination - the projected historical maps of plague reservoirs. As we predict the location of historic plague reservoirs with the help of climate models that themselves carry a high degree of uncertainty, we can only assess the level of confidence we can have in these maps by using historical data sources. A key result from this study, which goes beyond the study of plague, will be a better understanding of how disease reservoirs move or change over time, a topic that is almost completely unexplored within the sciences.

Yersinia pestis caused large plague pandemics in Medieval Europe, Central Asia and the Middle East. While its human burden is far lower today, the bacterium persists widely in the semi-arid deserts, steppes, montane meadows and the tropics of the Americas, Africa and Asia, in ground squirrels, jirds, voles, gerbils, rats & marmots, as well as their fleas. The ecological constraints that define where plague can persist in the wild are unknown. For non-ecologists, it is tempting to assume that regions with the right species (or even genus) of rodent can be potential plague reservoirs, but that intuition breaks down rapidly when comparing the rodent distribution maps with plague reservoir maps. Our primary objective is to understand the ecological constraints that define where plague reservoirs can exist. We will approach this problem by building ecological niche models (ENMs) that avoid the limitations of other attempts (e.g. focused on a regional scope only, using limited datasets, and generic rather than plague-specific input variables). With a vast amount of data now available from the former USSR plague control program and from historical records, the main challenge for us lies in using the right methodology to build and project ENMs as far across the globe as is reliably possible, and to test and select the right plague-relevant input variables for the models (e.g. climate instability, soil properties). Our use cases are both historic and current. We will answer how suitable medieval Europe, Asia and North Africa were as past plague reservoirs, something that since the ancient DNA breakthroughs in plague research is now heavily speculated upon, as well as learn more about the potential lifespan of those reservoirs. Finally, a better understanding what defines a plague reservoir helps with plague surveillance now, and in a future world that is in flux with climate change.