What is a vector?

Precise definitions are important in science, because I said so (and other better reasons). In parasite ecology, the tricky definitions that students often mix up are things like parasite versus parasitoid and microparasite versus macroparasite. In fact, at least 20 visitors per week happen upon this blog because they want look up one of those terms.

But it isn’t just students and the general public who struggle with tricky definitions in parasite ecology. Within the field, we can’t even agree on what to call our discipline! (Disease ecology? Parasitology? Epidemiology?) And it has recently come to my attention that a term that I thought had bullet-proof definition is somewhat controversial among parasite ecologists.

In an awesome special issue of the Philosophical Transactions of the Royal Society B that just came out this month, there was a thought-provoking article entitled, “What is a vector?” The idea for the article came from a working group of the British Ecological Society’s ‘Parasites & Pathogens’ Special Interest Group, where participants unexpectedly found that they did not all use the same definition of “vector.” The article contained a whole list of possible definitions that the authors found in the literature, including this subset, which I have re-numbered for my own purposes:

Definition 1A: “Any organism (vertebrate or invertebrate) that functions as a carrier of an infectious agent between organisms of a different species.”

Definition 1B: “Any organism (vertebrate or invertebrate) or inanimate object (i.e., fomite) that functions as a carrier of an infectious agent between organisms.”

Definition 2: “Any organism that can transmit infectious diseases between humans or from animals to humans.”

Definition 3A: “Hosts that transmit a pathogen while feeding non-lethally upon the internal fluids of another host.”

Definition 3B: “Blood-feeding arthropods such as mosquitoes, ticks, sandflies, tsetse flies and biting midges that transmit a pathogen while feeding non-lethally upon the internal fluids of another host.”

I couldn’t believe that parasite ecologists differed so much with their working definitions, so I put them into a poll, and then I asked you guys to tell me which definitions you use. (Thanks for your participation!) To my great surprise, your answers were all over the place! No one really used Definition 2 (the “anthropocentric” definition), but all of the other definitions received some support. As of 21 March, the most popular definition was 1A, and 1A+B were more popular than 3A+B.

Wilson et al. (2017) do a great job of discussing the pros and cons of each definition, and they also take a stab at a possible mathematical definition (the “sequential” definition), so I’d recommend giving their paper a read for a lot more coverage than you’re going to get here. I’m just going to cover two points that were surprising to me.

First surprise: I did not expect that so many people would prefer 1B over 1A, because I don’t like including fomites in the definition of a vector. My primary reason for preferring IA is that we already have a term for inanimate objects that transmit infectious agents (i.e., fomites). Wilson et al. (2017) provide a good discussion on the utility of thinking about some fomites (e.g., drug needles) in pseudo-biological terms that would normally apply to vectors. But I think that I’d still prefer to call things like needles “fomites,” even if it’s helpful to think of their parallels with living vectors.

Second surprise: I did not expect that so many people would prefer 1A+B to 3A+B. As Wilson et al. (2017) discuss, the 1A+B definitions are broad; for instance, the wording suggests that we include intermediate hosts (i.e., snails infected by trematodes) as vectors! It also suggests that any host capable of interspecific transmission could be a vector. 

In the end,Wilson et al. (2017) suggested that parasite ecologists think carefully about their definitions of the term “vector,” and then they scored a closing home run with, “all vector definitions are wrong, but some are (we hope) useful.” WOMP WOMP.

Give the paper a read, and share your thoughts in the comments!

Reference:

Wilson, A. J., E. R. Morgan, M. Booth, R. Norman, S. E. Perkins, H. C. Hauffe, N. Mideo, J. Antonovics, H. McCallum, and A. Fenton. 2017. What is a vector? Phil. Trans. R. Soc. B 372:20160085.

Food provisioning and wildlife disease dynamics

Humans change environments in many different ways, including accidentally or purposefully provisioning wildlife with novel food resources. For example, bird feeders, salt licks, ecotourism feeding stations, and dumpsters all provide concentrated food resources for wildlife. Does this food provisioning influence disease dynamics?

First, let’s consider how food provisioning might influence disease dynamics for parasites with any given transmission mode. The big rates that we care about here are the transmission rate (a function of contact rate and transmission success), birth and death rates, and immigration and emigration rates. If resource provisioning increases host population density via increased aggregation of individuals, increased birth rates or decreased death rates, and/or increased immigration or decreased emigration rates, then transmission rates for pathogens with direct contact density-dependent transmission should increase. In the same scenarios, pathogens with frequency-dependent transmission may not be affected by provisioning, or transmission rates may even decline if high birth rates dilute the prevalence of infection in the population. For pathogens that are transmitted via environmental stages, environmental stages may build up at resource provisioning sites when high densities of animals hang out there for long periods, increasing transmission. For pathogens that are transmitted via intermediate hosts, transmission may be reduced if hosts switch from foraging on intermediate hosts to foraging on human-provided resources.

Factors besides transmission mode might also be important. For instance, regardless of transmission mode, if resource provisioning increases host resistance – for instance, by increasing body condition – then transmission rates should decrease. Conversely, if resource provisioning decreases host resistance – for instance, if the provisioned food is nutritionally poor or there is high competition at sites with provisioning and body condition is reduced – then transmission rates should increase. Resource provisioning might also alter host tolerance to infection, so that hosts aren’t as sick but continue shedding infectious agents longer than they would if they didn’t have supplemental resources.

Ok, I think that covers most of the possibilities. Back to our question: does food provisioning influence disease dynamics? Yes, sometimes. Most of the mechanisms listed above were supported by at least one study in a recent review by Becker et al. (2015). Based on the long, messy list above, you can probably guess that sometimes food provisioning increases transmission, sometimes it decreases transmission, and sometimes nothing notable happens. The big take-home message is that there isn’t just one universal outcome when we provision wildlife with supplemental resources, and it’s important that we conduct more and better studies aimed at elucidating the epidemiological mechanisms underlying the observed relationships. Cool stuff!

Have you read this post about house finch conjunctivitis yet?

curseofmagicalseeddispenser

Reference:

Becker, D.J., D.G. Streicker, and S. Altizer. 2015. Linking anthropogenic resources to wildlife–pathogen dynamics: a review and meta-analysis. Ecology Letters 18: 483–495.

Refugia, connectivity, and transmission

When populations become small, their probability of extinction typically goes up, because demographic and environmental stochasticity are more likely to set the population on an irreversible decline. However, when one population goes locally extinct, a species is not necessarily lost; the area might be re-colonized by migrants from a different population later, if other populations exist. Metapopulation theory tells us that a balance between population extinctions and re-colonizations in heterogeneous patches that are linked by dispersal can allow a species to persist regionally, even when it goes extinct locally.

What happens when we add infectious diseases into our host metapopulation model? Connectivity might be detrimental to regional persistence when infectious diseases are introduced, because pathogens in one population can invade the other populations via host dispersal, whereas pathogens are limited to a single population when populations aren’t linked via dispersal. Or…not?

Heard et al. (2015) recently published a quite complicated and fancy metapopulation model that suggests that connectivity actually increases the probability of metapopulation persistence in an Australian frog species endangered by the fungal pathogen (Bd), which causes the disease chytridiomycosis. From survey work, they knew that the prevalence of Bd in growling glass frogs was lower in warm and/or salty wetlands. They could also show that the probability of a local extinction in any given frog population increased with the prevalence of Bd in the frog population. By linking local extinction risk to Bd prevalence and microclimate, they could create metapopulation models using known dispersal distances for the growling glass frog, and they could run simulations regarding metapopulation persistence under scenarios where they eliminated frog dispersal among populations or not. They had two important findings. First, if you ignore the fact that Bd prevalence varies with microclimate, the probability of metapopulation persistance is predicted to be much lower than it actually is. The warm and/or salty wetlands act as important low Bd prevalence frog population refugia that can seed the other populations in a metapopulation when they go locally extinct, such that microclimate variability increases persistence. Second, “re-seeding” can only happen if dispersal occurs among populations, so connectivity increases metapopulation persistence in this system.

One lingering question is whether this system is a good example of the role of connectivity in all host metapopulations plagued by infectious diseases. Heard et al. (2015) argue that Bd is basically everywhere already – and there to stay – because it can be maintained in both environmental reservoirs and reservoir hosts. Therefore, they suggest that for growling glass frogs, dispersal of hosts among populations doesn’t really play a role in disease dynamics. In systems where the pathogen is not yet widespread (i.e., regions currently being invaded), where reservoirs are less likely to provide long-term maintenance of the pathogen, or where dispersal of reservoir hosts is particularly important to pathogen spread, host dispersal could start to have detrimental impacts on long-term metapopulation persistence. This is cool stuff that deserves more attention!

Reference:

Heard, G.W., C.D. Thomas, J.A. Hodgson, M.P. Scroggie, D.S. Ramsey, and N. Clemann. 2015. Refugia and connectivity sustain amphibian metapopulations afflicted by disease. Ecology Letters, 18: 853–863.

frogald.png

Dilution Effect Debate Continues!

Admit it, the title of this post made you cringe! Or it set off a little fuse in your brain and you’re just seconds away from your head exploding. Take a zen moment, and then continue reading for some interesting science.

If you don’t know what the dilution effect hypothesis is or why disease ecologists are debating the hypothesis, you might want to check out some of my previous posts before reading this one. But briefly: scientists have found a negative relationship between host biodiversity and the risk of infection to particular host species in some disease systems in some areas. That negative relationship is called the “dilution effect,” because host biodiversity is “diluting” parasite transmission. The debates have arisen because disease ecologists can’t agree on is how often the dilution effect occurs in natural systems: always, sometimes, or never? I’ve described the core arguments on both sides of the debate in this post.

Meta-analyses are one way to figure out how commonly the dilution effect occurs in natural systems. By collecting all of the available empirical and/or observation  studies that consider biodiversity-disease relationships and lumping them together  into one analysis, we can figure out whether the dilution effect always, sometimes, or never occurs in natural systems. Before my last post on the dilution effect debates was published, two such meta-analyses argued that the dilution effect only sometimes occurs (it’s “idiosyncratic”) in systems where the focal hosts are humans and non-human primates. Additionally, in the meta-analysis regarding how biodiversity influences human risk of infection with zoonotic pathogens, Salkeld et al. (2013) found evidence of a publication bias for studies that find a dilution effect, suggesting that studies finding neutral or amplification effects are less likely to be published.

Fast forward to Civetello et al. (2015), who did a larger meta-analysis that included more studies in more host-parasite systems. As I posted about a few weeks ago, Civetello et al. (2015) found broad support for the dilution effect, which I suppose we can say means that the dilution effect “often” or “usually” occurs, which puts us somewhere between “always” and “sometimes.”

Now back to the “debate” part of this post: Salkeld et al. (in press) responded to the Civetello et al. (2015) paper with some concerns regarding the larger meta-analysis. In particular, they worried that including laboratory studies might muddle the analysis, because the way that we manipulate systems in the lab doesn’t necessarily correspond to what really happens in nature. Also, they pointed out that if there is a publication bias – as they previously found – then it might not be particularly meaningful that the dilution effect is commonly reported in the ecological literature. However, Civetello et al. (2015) didn’t evaluate whether there was a publication bias in their analysis. (Note that McCallum et al. 2015 also pointed out some of these possible concerns.)

Civetello et al. (in press) responded to these worries by doing another meta-analysis. They used only a subset of the studies from their previous paper, so that they were including only field studies of human pathogens, like in Salkeld et al. (2013). Civetello et al. (in press) still found an overall dilution effect, and they suggested that adding in the studies published since Salkeld et al. (2013) provided more statistical power to see the dilution effect than Salkeld et al. (2013) had. Civetello et al. (in press) also looked for publication bias in their subsetted dataset and didn’t find any evidence for bias, but they point out that the analysis to look for a publication bias had to violate some assumptions of independence, so it might not be particularly meaningful.

It’s unlikely that things are totally resolved here, but I think everyone is on the same page regarding the future directions for diversity-disease relationships: we’ve spent time looking for general trends and determining how common the dilution effect is in natural systems, and now it’s time to switch our focus to the mechanisms underlying the dilution effect.

References:

Civitello DJ, et al. (2015) Biodiversity inhibits parasites: Broad evidence for the dilution effect. Proc Natl Acad Sci USA 112:8667–8671.

McCallum H. (2015) Lose biodiversity, gain disease. Proc Natl Acad Sci USA 112: 8523–8524.

Salkeld DJ, Padgett KA, Jones JH (2013) A meta-analysis suggesting that the relationship between biodiversity and risk of zoonotic pathogen transmission is idiosyncratic. Ecol Lett 16(5):679–686.

Salkeld DJ, Padgett KA, Jones JH, Antolin MF (2015) Public health perspective on patterns of biodiversity and zoonotic disease. Proc Natl Acad Sci USA, 10.1073/pnas.1517640112.

Why infectious disease research needs community ecology

If you haven’t seen it yet, there’s a recent Science paper by Johnson et al. (2015) entitled, “Why infectious disease research needs community ecology.” If you’re a disease ecologist, it probably won’t come as a surprise to you that infectious disease research needs community ecology. If you’re not a disease ecologist, check out this paper for a quick, informative read!

I was happy to see that Johnson et al. (2015) emphasized the importance of looking at symbiont communities as well as host communities when considering the spread of parasites and pathogens. The importance of symbiont communities is still not as widely recognized as I think it should be, and this paper does a great job of giving concrete examples of systems where coinfection by parasites/pathogens or the presence of non-pathogenic symbionts influence the spread of a single parasite within a host population or community.

If you’re looking for more examples where understanding the spread of parasites and pathogens required a detailed understanding of community ecology, check out some of these previous blog posts:

Considering symbiont communities is important:

Symbionts protect hosts from parasitoids

Diverse symbiont communities protect hosts from several natural enemies

Symbionts protect hosts from parasites

Considering host communities is important:

Host community diversity may reduce focal host infection risk

Multiple host and vector species for prairie dog plague

Spillover of pathogens from wildlife to livestock to humans

Fecal transplants and other kinds of microbial community restoration

Competition among host species influences transmission

Considering heterogeneity among individuals and species is important:

House finches and Mg

Tasmanian devils and facial tumor disease

The superhosts!

Mouse raves and contact heterogeneity

Coinfection and superspreaders

Reference:

Johnson, P.T.J., J.C. de Roode, and A. Fenton. 2015. Why infectious disease research needs community ecology. Science 349(6252): 1259504.

Are superspreaders also superreceivers?

For simplicity, we often assume that all hosts have an equal probability of becoming infected by and transmitting parasites and pathogens. But of course, we know that it isn’t how real systems work. For instance, in real systems, hosts vary in their propensity to become infected by pathogens, and that variation is one probable cause of the parasite ecology “law” that macroparasites are aggregately distributed among hosts. We call hosts that are highly susceptible to a given pathogen “superreceivers,” and hosts that are highly likely to transmit a pathogen are “superspreaders.”

Here’s a question for you to ponder: are superspreaders usually superreceivers and/or are superreceivers usually superspreaders? For instance, sex workers are at high risk for contracting HIV (=superreceivers) because they frequently change sex partners, and they’re also highly likely to spread HIV (=superspreader), if they have it, in comparison to the average person. In that case, the superreceivers are also superspreaders. When that happens, we might predict really explosive epidemics whenever “patient zero” is a superreceiver+superspreader, because R0 will be very, very high.

But consider the Tasmanian devil example that I posted about recently. Tasmanian devils that bite lots of individuals are highly likely to contract Tasmanian devil facial tumor disease; they’re superreceivers. But being bitten by an infected individual doesn’t seem to transmit the infectious cancer to the receiving host, so devils that bite frequently don’t transmit any more frequently than devils that don’t bite frequently. Therefore, the superreceivers in that system aren’t superspreaders.

Now let’s talk about a really cool system that I somehow haven’t blogged about yet. House finches are hosts for an emerging bacterial pathogen (Mycoplasma gallisepticum – Mg) that jumped from poultry into house finches in the 1990s. This pathogen causes conjunctivitis in the house finches – a symptom you don’t often think about in wildlife! In a really neat recent paper, Adelman et al. (2015) showed that birds that spent more time on bird feeders were more likely to become infected by (superreceivers) and transmit (superspreaders) Mg. This is a really cool example of a pathogen that appears to be transmitted by “fomites”: inanimate objects that the pathogen can survive on when off the host.

We probably don’t have enough examples in the literature to determine whether superspreaders are usually superreceivers or to look for generalities in systems where this occurs. But we’re accumulating more examples all the time! Stay tuned.

…if academics were at higher risk of developing conjunctivitis when they sought out free food, I’d have some very squinty-eyed colleagues.

Finch

Reference:

Adelman, J.S., S.C. Moyers, D.R. Farine, and D.M. Hawley. 2015. Feeder use predicts both acquisition and transmission of a contagious pathogen in a North American songbird. Proc Biol Sci. 282(1815): 20151429.

Avoiding cadavers is good for you

There is a caveat to the title of this post. Avoiding cadavers is good for you if you’re very close to the cadaver. But that’s getting ahead of the game, so let’s back up:

There is growing evidence that uninfected animals can often sense and avoid infected individuals or infectious agents in the environment.  Or, conversely, uninfected animals may be attracted to infected individuals (see my old post about how disease is sexy).  Gypsy moths fall under the first category. Specifically, gypsy moths can become infected by a lethal (to them) bacilovirus when they ingest leaves contaminated by infectious gypsy moth cadavers, and previous work with these gypsy moths showed that they will preferentially eat uncontaminated leaves over contaminated leaves.

Obviously, these studies on the avoidance behavior of individuals were conducted because the authors thought that avoidance should affect an individual’s probability of becoming infected. These behaviors may affect population-level infection dynamics, too. So, does avoidance behavior affect pathogen transmission?

Eakin et al. (2015) used a really cool combination of laboratory experiments, field data, and modeling to answer this question. First, they parameterized two submodels using data from a laboratory experiment where they allowed individual caterpillars to feed on leaves that had one infectious cadaver on the surface. In the first submodel, they used model selection to figure out which mechanistic model best predicted a caterpillar’s probability of becoming infected. They found that the best model incorporated both how close to the infectious cadaver the uninfected caterpillar fed and how much the uninfected caterpillar ate. Just one bite of an infectious cadaver (which are ~78 bites total) increased the risk of infection by 0.4-4.7%. In the second submodel, they used model selection to figure out which stochastic simulation model best explained a caterpillar’s feeding decisions. They found that the best model included avoiding infectious cadavers. But here’s an interesting thing: the caterpillars don’t really detect and avoid cadavers until they are 0.7 mm away. Because eating a bite or two of infectious material doesn’t increase infection risk dramatically (just 0.4-4.7%), Eakin et al. (2015) suggest that caterpillars shouldn’t really go out of their ways to avoid cadavers; they should keep munching away until there is a cadaver right in front of them. Neat!

Finally, using the two parameterized submodels, Eakin et al. (2014) showed that the model predictions fit field data quite well, and the model that includes cadaver avoidance slightly outperformed the model without cadaver avoidance. At the highest densities of infectious cadavers, avoidance can reduce infection rates by 7% in a single transmission bout in the field. Cool!

I glossed over some cool math – like stochastic simulation models – so you should go check out the paper. Also, here is a relevant cartoon that, if nothing else, demonstrates my peculiar unique brand of humor:

caterpillarsonitsy3

(I don’t think Eakin et al. (2014) are selling caterpillar art on the Interwebs. But if they suddenly start to sell caterpillar art, I want 10% of the profit. Just saying. PS – this is a thing.)

Reference:

Eakin, L., M. Wang, and G. Dwyer. 2015. The Effects of the Avoidance of Infectious Hosts on Infection. The American Naturalist 185:1.