Predicting zoonotic spillover

[We’re taking a break from the “how to become a successful parasite ecologist” post series. More on that in a few weeks!]

Poop is pretty gross, and some poop is more disgusting than other poop. I’m sure you’d agree with both of those statements, but why? Imagine, if you will, that you are participating in one of my favorite activities: crawling in a narrow cave passage, with just enough room above you to wear your pack while you’re crawling. You round a corner and discover a very interesting conundrum: the small passage forks momentarily, and one fork contains a large pile of fresh raccoon poop, while the other is sprinkled with bat guano (less fresh). You’ll obviously avoid crawling directly through either one, but which is most important to avoid?

When parasites and pathogens that infect wildlife or domesticated species spillover into humans, it can be pretty terrible – think Ebola, SARS, rabies, etc. And depending on how you define “zoonosis” – we’ll get back to that in an upcoming post – you might say that most emerging infectious diseases of humans are caused by zoonotic parasites and pathogens. So disease ecologists should and do spend a lot of time trying to understand what causes the spillover of wildlife parasites into human populations, and how to predict and even control such spillover events.

The EcoHealth Alliance group is well known for tackling this important and complicated issue, and they recently published some great synthesis science in Nature that works towards understanding and predicting the origins of zoonotic viruses (Olival et al. 2017). Olival et al. (2017) created a database that contained every known virus of mammals and the 754 mammal species infected by those viruses. They also had trait information for each virus and each mammal species. Then they explored their massive mammal-virus data mountain with the intention of  answering ~4 big questions:

Which mammal species host the most known viruses, and what makes some mammal species have more viruses than others? As we’ve seen in other studies, the most important determinant of viral richness in each mammal species was the total disease-related research effort that has focused on that mammal species in the past. (This was also true for the number of zoonotic viruses per host species – see next). In other words, the more we look, the more we find! But Olival et al. (2017) take this one step further, and use model predictions to tell us where we should look to find the most new viruses and the most new zoonotic viruses – see below.

Which mammals host the most known zoonotic viruses, and what makes some mammal species have more zoonotic viruses than others? For the purposes of this paper, zoonotic viruses were defined as viruses detected at least once in humans and at least once in another mammal species. Proportionally speaking, bats, primates, and rodents had more zoonotic viruses than other mammal taxa. And some host traits that correlated with the number of zoonotic viruses per species included phylogenetic distance to humans, ratio of urban to rural human population in the host’s range (a possible measure of human-wildlife contact), and whether the species was hunted (another measure of human-wildlife contact). Even after controlling for all of those covariates, bats hosted higher proportions of zoonotic viruses than other mammal taxa.

If you’re a long time follower of this blog or the disease ecology literature, then you know that this isn’t the first study to find that bats host more than their fair share of zoonotic viruses. For instance, previous work had shown that bat species have more zoonotic viruses than rodent species, on average. (But there are more rodent species than bat species, so rodents host more total zoonotic viruses). Olival et al. (2017) confirm this with a dataset including many more viruses and mammal taxa, so the “bats are special” pattern is quite robust! If you’re wondering why bats host more proportionally more zoonotic viruses than other mammal taxa, you might be interested in these previous posts: here, here, and here.

Where do we expect to find the most undescribed viruses, and in particular zoonotic viruses? It turns out that if you want to find new zoonotic viruses, the best place to look would be bats in Northern South America. Cool! You can check out the neat maps in the paper if you’re interested in other taxa or geographic areas.

Did particular virus traits correlate with whether a virus has been observed to be zoonotic or not? Yes! For instance, viruses that that infected a greater range of non-human host species (i.e., host breadth), replicated in the cytoplasm, or were transmitted by vectors were more likely to be zoonotic. Of course, these viral traits don’t 100% predict whether a newly discovered virus will be zoonotic or not, but these descriptive models help to identify hypotheses that can explain why some viruses easily jump into humans and others don’t.

So… what does all of this tell us about poop in caves? Well, not much, actually. The Olival et al. (2017) study was meant to describe broad patterns and make predictions to guide future survey/surveillance efforts, not to inform specific risk assessments. But to follow up on my admittedly tenuous hook, we DO know that some mammals are far more likely to pass on viruses to humans than others. So if you have to choose between hugging a bat or a rabbit (or crawling through their poop), pick the rabbit!

But of course, it isn’t just viruses that we need to worry about; I was worried that the raccoon poop might contain Baylisascurus eggs. I’ll keep my eye out for their next Nature paper that does this study with all parasites and pathogens!


Where will the next bat virus spillover?

In a previous post, I discussed potential characteristics of bats that might make them “good” at “sharing” novel viruses with humans. There are many hypotheses out there, and probably all of the proposed important bat characteristics play a role for spillover of some viruses in some places at some times. There’s still a lot of research that needs to be done there, so investing in bat research is still a high priority. For today, we’ll just talk about one characteristic: for whatever reason, bats are hosts for a huge diversity of viruses, so there are a lot of viruses for them to potentially transmit to humans.

Even if bats are highly likely to share novel viruses with human populations, that sharing could never happen if bats and humans didn’t directly or indirectly interact. Therefore, there are also lots of hypotheses out there about which anthropogenic activities lead to high rates of interaction between humans and bats. For instance, areas with many people, areas with many domesticated animals that interact with bats and people, and areas where people are particularly likely to encounter bats (e.g., by eating them as bushmeat) might be especially likely to experience spillover of bat viruses into human population.

So, wouldn’t it be awesome if we had a map that showed where these drivers of bat virus spillover were particularly prominent, so that we could predict areas where spillover is most likely to occur? Why yes, yes it would be awesome. And just such a map was recently created by Brierley et al. (2016).

Here’s the just of it: using spatial regression techniques, Brierley et al. (2016) came up with a list of drivers that were good at predicting the total number of viruses shared between humans and bats in 1 decimal degree-sized grid blocks all over the world. They found that bat host diversity and annual rainfall were important drivers, and they suggested that these were links between virus diversity and the potential for virus spillover. They also found that things like human population sizes, the number of domesticated pigs, and the use of bats as bushmeat were important drivers, suggesting that anthropogenic activities are also important to spillover.

Interestingly, the areas where risk is high due to the high diversity of bat viruses (South America) are not the same as the areas where the risk is high due to high human-bat interaction rates (Sub-Saharan Africa). This suggests that when we think about preventing spillover of bat viruses into human populations, we probably need different plans for regions with different drivers. That’s not necessarily a new idea, but now we have a great map to show us which areas need which kinds of prevention!

This cartoon is not intended for people eating bats because they have few sources of protein in their lives. Obviously, eating bats isn’t a decision for them, it’s a necessity. But those of us in positions of relative power can work towards alleviating the socioeconomic situations that push people towards the consumption of bushmeat. And if you do have a choice, don’t eat bats!!



Brierley, L., M.J. Vonhof, K.J. Olival, P. Daszak, and K.E. Jones. 2016. Quantifying global drivers of zoonotic bat viruses: a process-based perspective. The American Naturalist.

Are the majority of human EIDs really zoonotic?

 In 2001, Taylor et al. reviewed more than 1415 pathogens that infect humans. They found that 61% of those pathogens were zoonotic, meaning that they are transmitted between animals and humans. Furthermore, 75% of the pathogens classified as causative agents of emerging infectious diseases of humans were considered zoonotic. Clearly, if we want to understand pathogen transmission in human populations, we need to understand how these pathogens are spilling over from wildlife populations. But here’s a sensational question for you: are the statistics that I just summarized accurate?

Before I speculate further, I need to introduce a category of pathogens that may be commonly overlooked: sapronoses. Unlike ‘typical’ pathogens (whatever that means), sapronoses do not need a host to survive. While sapronoses might replicate in a host or even be transmitted among hosts, they can also reproduce and flourish in the environment outside of the host indefinitely. In contrast, pathogens with free-living stages eventually need a host to complete their life cycle.

Brain eating amoebas are one example of a sapronosis. These amoebas typically live in the environment, but they can accidentally enter the human body through the nose. For instance, this might occur when a human is swimming in water containing the amoebas. From there, the protist feeds on nervous tissue, and the human host almost always dies due to infection. Because the amoebas don’t really need a host for reproduction or transmission among environments, there is no selection pressure for the amoebas to keep their hosts healthy for longer periods by evolving reduced virulence. And because the amoebas aren’t transmitted directly among hosts, treating or quarantining infected people won’t reduce the probability that other humans become infected. Instead, limiting human contact with contaminated environments or treating contaminated environments to eradicate the sapronotic agent are the only ways to reduce transmission to other hosts.  Some other examples of sapronoses are anthrax, cholera, and tetanus.

Ok, back to my sensational question:

Kuris et al. (2014) reviewed a subsample of the human pathogens that Taylor et al. (2001) reviewed previously, and Kuris et al. (2014) found that one third of the subsampled pathogens were sapronoses. Cool! When they broke down the percentages by taxa, almost 100% of the fungi that they examined were sapronotic/saprophytic, as well as ~29% of the bacteria and ~13% of the protists. When Taylor et al. (2001) classified the pathogens, they found 113 zoonotic fungi. But Kuris et al. (2014) argue that their subsample suggests that almost all of the fungi should be saprophytic, not zoonotic. It may be that Taylor et al. (2001) classified saprophytic pathogens as zoonotic pathogens, leading to an overestimate of the proportion of human pathogens that are zoonotic.

I think it’s still safe to say that most human pathogens have an environmental and/or animal reservoir. Additionally, even though the proportion of human pathogens that are zoonotic might be less than “the majority” (i.e., <50%), a large proportion of the human pathogens would still be classified as zoonotic, even after reclassifying the potentially sapronotic pathogens. But Kuris et al. (2014) bring up a subtle point that deserves more attention: just because animals can be infected by a human pathogen doesn’t mean that there is transmission of the pathogen between animals and humans. Neat stuff!


Kuris, A.M., K.D. Lafferty, and S.H. Sokolow. 2014. Sapronosis: a distinctive type of infectious agent. Trends in Parasitology 30(8): 386-393.

Taylor, L.H, S.M. Latham, and M.E. Woolhouse. 2001. Risk factors for human disease emergence. Philos. Trans. R. Soc. Lond. B: Biol. Sci. 356: 983–989.