Tuesday, February 14, 2017

Virtual fire ants

Last month I published my first computer modeling paper.  I understand the power and value of simulation studies, but I’m not usually a fan of them.  I tend instead to do projects that involve field experiments, collecting specimens, or measuring things in a lab.  But after years of collecting data on dozens of real-life ant species, I thought I had enough information to create a realistic virtual one.

A computer simulation would allow us to test ideas that would be difficult or impossible to test in real life.  We could, for example, build a virtual ant population and watch it evolve over generations.  And we could do that repeatedly with many populations at the same time.  In just a few days we could have 25 virtual years of data from hundreds of simulated worlds.  Getting the same information in the field could take a whole career.

So we got to work designing virtual ants.  We made a program that created worlds inhabited by ant colonies whose properties we could tailor as we saw fit.  We could then see how tweaking the biology of ant colonies—how they grew, died, and reproduced—affected their populations.

Our program followed ant colonies over several generations as they were founded, grew, acquired territory, reproduced, and died (figure adapted from Helms & Bridge 2017)

Our ant of choice was the red imported fire ant (Solenopsis invicta).  We chose to work with fire ants for three reasons.

1) We know a lot about them.  Programming a realistic animal requires a lot of information, and fire ants are perhaps the most studied ant on the planet.  We have detailed field measurements of how and when they reproduce, how colonies grow and compete for space, and how long they live.

2) They are important.  Fire ants are an invasive species in the US and elsewhere.  Farmers, governments, and conservation organizations around the world want to know how they spread.

Fire ants were accidentally introduced to Alabama in the 1930s and have since spread throughout the southern US (figure from Calcott & Collins 1996, The Florida Entomologist)

3) They are interesting.  Fire ants are a model system for studying alternate reproductive strategies.  They have a complicated life history and reproduce in several ways.  Colonies produce queens that fly off and found new colonies in vacant soil, but they also produce parasitic queens that take over other colonies of the same species.  So by studying them we could learn how parasitic ants evolve, and how the evolution of parasitism affects other aspects of their biology.  (Fire ants have plenty of other odd twists in their biology as well, but for this study we focused on parasitism)

Fire ant colonies practice two reproductive strategies—they produce some queens that found new colonies and other, parasitic, queens that take over existing colonies whose previous queens have died

Once we had our program and our virtual ants, we ran a couple experiments.  In the first, we set up different ant populations that varied in how colonies reproduced.  In some populations colonies only produced queens that founded new colonies, whereas in others half the queens produced by colonies were parasites, and other populations fell somewhere in between.  We then tested whether producing parasitic queens affected the demographics of populations—how large or how spread out the colonies were.

In our second experiment we created mixed populations inhabited by several lineages of ants.  Some lineages produced lots of parasitic queens and others didn’t.  We then watched the populations evolve over time as some lineages survived and reproduced more effectively than others.

We found that populations that produced more parasitic queens had larger colonies and occupied more of the available habitat.  Populations that don’t produce any parasitic queens, on the other hand, spread faster.  This is because parasitic queens can only take over existing colonies—they can only survive in areas already occupied by fire ants.  But queens that found new colonies can survive anywhere that isn’t already claimed by an existing colony.

When it comes to evolution, it turns out that location plays a deciding role in which strategy is favored by natural selection.  Populations at the edge of an expanding range, which are surrounded by suitable empty habitat just waiting to be colonized, evolved to produce almost no parasitic queens.  Conditions are nearly opposite in the interior of the range, which is saturated with existing colonies and contains little free space in which to found new ones.  Not surprisingly, populations in those areas evolved to produce more parasitic queens.

No computer model is supposed to be entirely accurate.  Models are instead meant to be useful for teasing apart how different factors influence a process.  What we need now is a field study to see whether we see the same patterns in nature that we do in our model.  That’s a project for another day…

So while a computer simulation was way outside my comfort zone, it did have one thing in common with all good research—it raised just as many questions as it answered.