1 of 1

The Evolutionary Origins of Phenotypic Plasticity

What are the evolutionary stepping stones for phenotypic plasticity?

[1] Cameron K.Ghalambor, Lisa M. Angeloni, and Scott P. Carroll. Behavior as phenotypic plasticity. Evolutionary Behavioral Ecology, pages 90–107, 2010.

[2] Beldade, Patrícia, and Paul M. Brakefield. "The genetics and evo–devo of butterfly wing patterns." Nature Reviews Genetics 3, no. 6 (2002): 442-452.

[3] Charles Ofria, David M. Bryson, and Claus O. Wilke. Avida: A software platform for research in computational evolutionary biology. In Artificial Life Models in Software, pages 3–35. Springer, 2009.

Unconditional Precedes Conditional

Sub-optimal Precedes Optimal

NAND

NOT

100%

89.5%

100%

91.7%

87.5%

Treatment

Baseline

Low Mutation Rate

Short Env. Cycle Length

Long Env. Cycle Length

High Mutation Rate

90.3%

92.1%

100%

77.8%

100%

94.1%

100%

90.9%

100%

90%

Non-plastic Lineages

Plastic Lineages

Non-plastic Lineages

Plastic Lineages

Updates

Results

Alexander Lalejini and Charles Ofria

Long Cycle Length Lineages

Unconditional task performance often leads to partial plasticity and eventually to optimal plasticity. Through visualizations of evolved lineages, we find that stochastic phenotype switching can emerge as an alternative to phenotypic plasticity.

Background

Phenotypic Plasticity is the capacity of organisms to express different traits in response to different environments. [1]

Populations are exposed to temporally or spatially varying environments
Environments are differentiable by reliable signals
Different environments favor different phenotypes
No single phenotype can exhibit high fitness across all environments [1]

When does phenotypic plasticity evolve?

Alternative seasonal phenotypes in Precis coenia

[2]

References

The Avida Digital Evolution Platform

Virtual hardware of an Avida Organism [3]

Avida organisms are self- replicating programs that evolve computational ‘tasks’ that improve their replication rate.

Organisms can sense their environment and alter their phenotype (the tasks they perform) in response.

A toroidal Avida Grid

Experimental Design

ENV-NAND

ENV-NOT

ENV-NAND

Time

Experimental Fluctuations

Environments Experienced Over Time

ENV-CONTROL

Time

Control (Static Environment)

Environments Experienced Over Time

Mutation Rate

baseline

low

high

Cycle Length

baseline

short

long

baseline

Treatment

Baseline

Low Mut.

Short Env. Cycle

Long Env. Cycle

High Mut.

Experimental Treatments

Experimental treatments alternated between two environments where NAND or NOT was rewarded and the other was punished.

Control treatments always rewarded both tasks.

Task Profile

Plastic?

ENV-NAND

ENV-NOT

NAND

NOT

NAND

NOT

–

YES

–

Color Code

Actively Harmful Plasticity

Not expressing right task

Expressing right task

Expressing wrong task

Not expressing wrong task

Baseline Treatment Lineages

Hi, I’m Alex.

And, This research is looking at the evolutionary stepping stones for phenotypic plasticity.

So, phenotypic plasticity is the capacity of an organism to exhibit different phenotypes in response to different environmental conditions.
There’s been theoretical and empirical work on what conditions are necessary for phenotypic plasticity to evolve, and it’s been established that there are four conditions that must be satisfied: 1) 2) 3) and 4)

Because natural evolution is slow and it’s hard to collect really high resolution data on natural systems, we study the evolution using digital organisms.
For this work we used the Avida Digital Evolution Platform where we evolve organisms of self-replicating computer programs to perform computational tasks.
For these computer programs, plasticity is going to involve execution flow control to change what computational tasks they do based on their current environment

The experiment we used to look for the evolutionary stepping stones for plasticity is setup entirely to satisfy these 4 conditions for plasticity.

We have a temporally changing environment: for some period of time NAND is rewarded and NOT is punished, then vice versa
We give organisms a sensory instruction that can be used to differentiate between these two environments
Each of these environments favor different tasks; so no single phenotype will perform optimally in both environments

In addition to a changing environment setup, as a control, we subjected organisms to an environment that always rewarded NAND and NOT and never changed, expecting phenotypic plasticity not to evolve.
We also ran different experimental treatments varying the mutation rate and the environment cycle length

This brings me to the results.

Across the board we saw two fairly intuitive results: 1) organisms generally evolved to perform computational tasks unconditionally, meaning they did the task all the time regardless of the environment, before they evolved to perform them conditionally.
2) Organisms generally evolved sub-optimal forms of phenotypic plasticity before evolving optimal plasticity. In this context, optimal plasticity is only performing the rewarded task in the current environment, so…. Sub optimal plasticity would, for example, be using flow control to suppress performing a punished task in one environment but still performing both tasks in the other.
These things make sense: to do things conditionally, you need to be able to do them unconditionally first. To do things optimally, you must first traverse states of sub-optimality.

So, to further explore how plasticity evolves, we implemented a visualization tool – the results of which are in these two charts. I’ll explain it piece by piece. First I want to point out that if we consider an organisms phenotype to be… there are 16 possible phenotypes. These are mostly enumerated here in this legend.
So, we can assign each one of those phenotypes a color. Then we can take one of our evolved organisms and trace all of its ancestors back to the original starting organism at the beginning of the experiment. We can test each one of those ancestors and categorize its phenotype. That’s what each of these individual bars are. Okay, we can also document what the environment would have been like at each point in time. That’s what each of these are on the vertical axis. So these charts show you the phenotypic state of a lineage at any point in time during the experiment. You’ll notice everything starts the same – not able to do any computations; and evolution happens. Also, these are just very pretty to look at.

I’ve separated plastic lineages from non-plastic lineages.
Plastic on the left, non plastic on the right; we can find examples of the results I just talked about in these charts, but what’s even more interesting are these cyclic patterns; and they’re most prevalent in the nonplastic data. ….