Competition avoidance drives foraging plasticity

A friend of mine recently published a great article that features several interesting topics in animal behavior research right now. The lab I worked in as an undergrad works on personality of threespine stickleback fish. Dr. Kate Laskowski, who just successfully defended her thesis last week, conducted an experiment that tested models predicting that a variable environment can drive between-individual differences in the plasticity of behavior. It’s a bit complicated but the implications are really, really cool. This blog post will be about how avoiding competition can drive different personality types. Dr. Laskowski was kind enough to provide a commentary at the end of the post. If you have access to an academic database, you can download the full article here.

Article details
- Laskowski KL, Bell AM. 2013. Competition avoidance drives individual differences in response to a changing food resource in sticklebacks. Ecology Letters. 16: 746-753.
- School of Integrative Biology, University of Illinois at Urbana-Champaign

Very brief summary
Behavioral plasticity is the ability to change your behavior in response to a change in the environment. Models predict that trying to avoid competition in a heterogeneous environment can promote between-individual variation in plasticity. In other words, if competition avoidance is possible, it'll be beneficial for individuals to consistently differ from one another in how responsive they are to changes in the environment (i.e. some are very responsive, others not responsive, and others in between). This study provides empirical support for these models.

Glossary

- behavioral plasticity – the ability to change one's behavior in response to the environment. There is a cost associated with behavioral plasticity, for example the time spent traveling to a different foraging patch.

- between-individual variation – in a group, the variation between all the individuals. So,

how much one individual will differ from another individual. An increase in between-individual behavioral variation means individuals are behaving more consistently (decrease in within-individual variation) or there is a bigger variation in how the whole group acts.

In the graph on the right, you can see that by decreasing the within-individual variation (smaller bars) would increase the between-individual variation. Decreasing within-individual variation would mean the animals are behaving more consistently. (Note: in reality, this would be much harder to show as a graphic because you'd have lots of individuals, with many of the bars overlapping. This is an extremely simplified version.)  

- heterogeneous – non-uniform, patch. e.g. a salad is a heterogeneous mixture. (Contrast with homogeneous.)

- personality – more accurately referred to as “consistent individual differences in behavior.” These consistencies extend over time and across contexts. Personality has been documented in a wide range of taxa, from insects to primates.

- within-individual variation - how much one individual varies. As you take multiple measurements of a behavior, you will have variation because the animal doesn't act exactly the same every time. 

Article summary

Models on plasticity

In nature, we'd expect animals to be able to react perfectly to their environment (i.e. be perfectly plastic). Animal personality means that animals are not perfectly plastic, however. Animals don't conform to the best behavior for every context, instead differing from one another in aggressiveness, sociability, exploratory behavior, boldness/shyness, and more. This means the most aggressive fish in a group will still be the most aggressive across different contexts, like foraging, exploring, but also in the presence of a predator. The graph on the right, a reaction norm, illustrates this. Each line is a different individual.

Darwin's theory of natural selection clearly shows how variety can be beneficial. A species where everyone is identical means that the possibility exists that one flaw they all possess - susceptibility to a certain disease, for example - could wipe everyone out. Done, extinct, left only to artist's impressions in textbooks. But being different from one another in the case of personality isn't always beneficial; in an extreme example, some female fishing spiders are born so aggressive that they will kill any males before they can mate, meaning their genes never get passed on. (There's a lot of discussion on why limited plasticity should exist at all.)

Interestingly, not only are animals not perfectly plastic; they also differ in how plastic they are. In other words, individuals vary in how much they'll adjust their behavior to a change in the environment. Consider this example: a heavy rain leaves puddles of water in a forest that attract mosquitoes and other insects that lay eggs in water. According to currently-accepted behavioral plasticity models, some bats will shift their flying pattern to include these new sources of food, while other bats will stick to the spots where they've found food before. Think of it like deciding whether or not to eat at McDonald's or some new restaurant that opened somewhere a few streets over.

But now, picture that both McDonald's and the new place (let's call it Joe's) are severely limited in food supply. McDonald's only has 100 happy meals and Joe's only has 100 Joe burgers. You and 199 of your classmates have half an hour to eat lunch. Everyone used to just go to McDonald's without thinking about it, but now what do you choose? There's not enough happy meals, but who knows what a "Joe burger" even is? You're not sure where Joe's actually is, either. 

Some animal behavior models predict that this uncertainty about the environment can drive differences in plasticity. So while McDonald's was always open before (i.e. a stable environment), the appearance of a new restaurant (a change in food availability and distribution) could drive these differences between people's decisions on what you do on your lunch break (plasticity). To add an additional layer of complexity, social dynamics (who's in a group with you) have been predicted to also affect behavioral plasticity, but it's unsure how. It's like quantifying the effect of who's in the car with you when you're deciding where to eat.

Testing the models

As logical as these models (and the McDonald's example) may sound, there have been few studies on real animals to back up what the models say. Laskowski decided to test these models using threespine sticklebacks (right). Threespines are social fish commonly used in animal personality research, ideal for a study on behavioral plasticity and social dynamics. They're also easy to work with in a lab. 

Specifically, Laskowski tested these predictions:

1. When it's possible to avoid competition in a social group, the between-individual variation in plasticity will increase
 - Put simply: If everyone has the same level of responsiveness to the environment (e.g. a new foraging patch appears so everyone goes there instead of staying at the old patch), everyone will be competing for food. If it's possible to avoid this competition (e.g. the old patch is still there, so there's food for those who stay and those who leave will find food at the new patch too), there will be an increase in how responsive individuals are to a change in food availability.

 - If this is true, this should happen: 
When a new patch becomes available (and the old one remains), some individuals will consistently choose to stay and others will consistently choose to go to the other patch. (This differs slightly from the concept of ideal free distribution in that the individuals themselves will consistently make the same choice about staying or going, as opposed to just choosing where to go based on how many individuals are at each patch.)

- If this is not true, this should happen:
When a new patch becomes available (and the old one remains), individuals will just conform to ideal free distribution. Whether they choose to stay or change patches will be random and solely based on how many individuals are at each patch.

2. The social environment influences the variation in plasticity  
 - Put simply: Those around you can influence whether you consistently stay or switch to a different patch.

- If this is true, this should happen:
When a new patch becomes available (and the old one remains), an individual's tendency to stay or switch patches will depend on who is in the social group.

- If this is not true, this should happen:
The social group will not affect tendency to stay or switch.

Methods
Laskowski used two "regimes" to measure behavioral plasticity. In the "simultaneous patch regime," bloodworms were mechanically dropped into one side of a tank with six sticklebacks. After 5 min, half of the food was dropped into one side of the tank and the other half was dropped into the other side for 5 min. In the "sequential patch regime," after 5 min of food dropping into one side of the tank, the food started dropping into only the other side of the tank for 5 min.

Laskowski used six groups of six sticklebacks that were assigned to one of the two regimes. Each group was tested in two trials per day on five consecutive days to get the repeatability of behavior. The main variable she was interested in was switch delay, or latency to switch to the new patch.

To test the influence of social group, she then shuffled individuals between groups so that only two individuals from the same group were in their new one. Then, the fish were tested in the same regime as their original group. She also tested whether individual behavior in a group was related to individual behavior while alone. This was done by also testing fish from the simultaneous patch regime while alone. 

Data analysis 

Question 1: Competition avoidance
Bayesian statistics was used (as opposed to frequentist) with Markov Chain Monte Carlo simulations. Because there were multiple data for each individual - making the data  non-independent - Group and Individual (nested within Group) were included as random effects in the models. The repeatability of latency to switch patches, this experiment's measure of behavioral plasticity, was estimated.

Question 2: Effect of social group
A separate bivariate mixed model was used to estimate the covariance between switch delay in the original and shuffled groups.

Results

Result 1: The opportunity to avoid competition promotes between-individual variation in plasticity
Individuals in the simultaneous patch regime showed consistent individual differences in switch delay (repeatability = 0.18, 95% CI: 0.05, 0.38). Individuals in the sequential patch regime showed very low between-individual variation (repeatability = 0, 95% CI: 3.0 x 10^-10, 6.6 x 10^-9).

This means that in the simultaneous patch regime, which allows for variation in staying or switching foraging patches, individuals did adopt a consistent strategy to avoid competing with one another. When there was no option to change your strategy to avoid competition (i.e. the sequential patch regime), there was no point in trying to act differently than everyone else. This makes sense, because there's only one source of food.

Importantly, the repeatability in the simultaneous patch regime increased over the course of the five days of the experiment. Specifically, the within-individual consistency grew higher. This means that individuals were becoming more consistent in their strategy. This positive feedback was not observed in the sequential patch regime.  

Result 2: Individuals maintain behavioral plasticity across different social environments in the simultaneous patch regime

In the simultaneous patch regime, there was significant covariance between an individual's switch delay in the original and shuffled groups. This indicates that in an environment where competition can be avoided, the social group does not influence behavioral plasticity.

In the sequential patch regime, there was essentially zero variation in switch delay between the original and shuffled groups, making it impossible to estimate covariance. However, this also supports that there's little carryover across social contexts. This means that for behavioral plasticity in patch choice, this study found no support for social environment having an influence.

Conclusions

Recent models have predicted that individual differences in plasticity are more likely to emerge in a spatiotemporally variable environment and when opportunities to forage are limited by competitors. This study provides strong empirical support for these models. Consistent individual differences were only apparent in the simultaneous patch regime, where competition avoidance was possible. This suggests that ecological factors such as food availability and predictability might influence variation in plasticity. Also, social environment did not have a strong influence on plasticity of switch delay: while individuals differed somewhat in their behavior between the two social groups (i.e. the covariance was not perfect), there was still evidence of consistency in behavior across the two groups.

This study also showed that the presence of highly plastic individuals in a population is likely to drive ideal free distribution, as competition decreases when plastic individuals hurry to the new patch and non-plastic individuals stay at the old patch. 

If you have access to an academic database, you can download the full article here.

The author comments - Kate's thoughts: 
"Running this experiment was a serious labor of love. Just collecting the data took a long time: I had so many groups of fish and they needed to be tested twice a day for five days. At the time, I only had one feeding arena in which to test the fish, so this mean that I could only test one group a week – resulting in a total of 15 weeks of data collection! I started having dreams about dropping bloodworms into fish tanks; it seriously felt like it was never going to end. But the really hard part of the experiment came with the data analysis.  I had to learn how to use R to run the Bayesian analyses, both of which I was completely unfamiliar with. Those were long dark weeks that I don’t remember a lot about other than staring at a computer screen for hours. But finally, I figured it out and was able to apply these fancy new statistical techniques to my data, which really improved the paper. And not to mention now, I’ve gotten quite good at both R and mixed modeling techniques which is an awesome (and very marketable) skill to have. So even though it felt like it took forever, in the long run it's totally worth it."