Antlion larvae are some of the more vicious predators among the insects. They build pit traps in sand. These are context-dependent contructs: the steepness of the pit and its diameter are influenced by several factors, including sand properties and hunger level of the larva. Regardless of how large the pit is, it’s a very sensitive structure: as soon as an ant steps on the edge, the sand will collapse into the hole dragging the ant with it, and straight into the waiting open mouth of the larva.
The work I’m doing with antlion larvae isn’t actually my idea. The study design was done by colleague, who also gathered the data a park trail near his house. The data sedimentated in his office for a couple of years and now he handed it over to me. I’m only responsible for data manipulation, statistics, and interpretation (he’s a coral freak who did this on a whim, not an entomologist).
What he did was every day, on set points along this long path (using distance markers and numbered light fixtures to make sure he was at the same place every day), he recorded the distribution of antlion pits across the path using a quadrat. 27 different lines were done at different points along the trail, and all this every day for a bit over 3 months (with only a couple of days missing).
This is a mammoth dataset recorded on paper. Besides the pits, puddles, larval tracks made by moving antlion larvae, pedestrian footprints, and tyre tracks were also recorded. My first job is to convert all this analog data into a digital format. I’ve played around with some sort of image recognition script that will do it automatically if I scan the pages, but I have no experience with this stuff (if any reader does, e-mail me and we can strike up a collaboration!), so it’s manual data entry for 30000+ points. Maybe I can get one of those monkeys that are busy trying to randomly put together Shakespeare’s works to do it for me.
Once the data is in, I will initially be working it as spatial data. This will allow me to plot the points in each plot and see how they’re distributed: is there preference for one side of the path, do the pits cluster in the middle, do these attributes change on different points along the path (if so, why?).
Once that is done, I can then implement the time axis and visualise the change through time, allowing me to see if the pit distributions along the path change as the rainy season progresses in response to increasing rainfall, for example. This will be the tricky part, as I have very little experience with such analyses, so I’ll have to learn some more R. It can be done on a very local scale, examining the effect of single events (e.g. the decimation and gradual recovery of antlion abundances after a particularly rainy day), or it can be done generally.
One additional goal is to find a way to automate the tracking of individual antlions. I can do this manually by looking at the data, but writing a script that can do it would provide a new depth to the analyses – it would allow us to quantify the responses of all individuals rather than the group as a whole. This can probably get coopted from some other program, I don’t know. If you want to give it a shot, e-mail me, this is an easy publication, and there could be an opportunity to develop a full-fledged program useful for behavioural biologists/ecologists.
There are two points to this research. The first is urban planning. We have this comprehensive dataset showing the behavioural responses of antlions to anthropogenic disturbance (walkers and bikers on the path, construction) and environmental disturbances (rainfall). The general feeling from the data is that antlions are pretty damn resilient – even after a huge disturbance, the same individuals start popping back up in a couple of days. This will thus provide data for use in urban ecology.
The second point, and where I am most interested, is biological. Lab observations of antlions give us a lot of knowledge… but they don’t quite jive with what is observed with this data from the wild. For example, pit sizes don’t seem to be associated with hunger level here. I have a weird, consistent pattern of antlions being more dense where there is most destructive disturbance – this is contrary to what one would expect, and I’m tentatively hypothesising that maternal effects have a role to play in where the antlion larvae settle. There is also an exceptional trail recorded that’s longer than predictions based on how much energy we think antlion larvae have. Either this was a superpowered antlion larva, or we’re just underestimating them.
In any case, I’m still midway through raw data entry right now, so all of this will have to wait a while until I can start analysing.