Frequency-dependent fitness effects are ubiquitous

This study demonstrates that frequency-dependent fitness effects are ubiquitous in *E. coli* populations, challenging the assumption of constant selection by revealing that competitive advantages typically decline as mutant frequency increases and that these dynamics significantly alter evolutionary trajectories.

The Gist

Imagine you are watching a race. In the classic way we think about evolution, we assume that if a runner has a "super-shoe" (a beneficial mutation), they will always run faster than everyone else, no matter how many other runners are wearing normal shoes. If they are the only one with super-shoes, they win easily. If they are surrounded by a crowd of normal runners, they still win easily. Their speed advantage is constant.

This paper says: "Actually, that's not how it works."

The researchers took a famous, long-running experiment with bacteria (the E. coli Long-Term Evolution Experiment, or LTEE) and looked at the "super-shoes" these bacteria had developed. They found that in almost every case, a bacterium's speed advantage depends entirely on how many of its friends are in the race with it.

Here is the breakdown of their discovery using simple analogies:

1. The "Crowded Room" Effect (Frequency Dependence)

The main discovery is that fitness is not a fixed number; it's a social dynamic.

• The Old View: If you have a better tool, you are always better, regardless of who you are working with.
• The New View: Imagine you have a really cool new smartphone.
  • Scenario A (You are the only one): Everyone is jealous. You are the "king" of the room. Your phone is incredibly valuable because it's unique.
  • Scenario B (Everyone has one): Suddenly, your phone isn't special anymore. The advantage you had disappears because the environment is now saturated with phones.

The researchers found that for about 80% of the bacterial mutations they tested, this "crowded room" effect happened. When a mutant bacterium was rare, it had a huge advantage. But as it became common, its advantage shrank or even disappeared.

2. The "Traffic Jam" Analogy (Why does this happen?)

Why does being common make you slower? The paper suggests it's like traffic on a highway.

• The Resource: The bacteria are eating glucose (sugar). Think of the sugar as the fuel for a car.
• The Mutant: A mutant bacterium is a sports car that burns fuel very efficiently and goes fast.
• The Ancestor: The original bacteria are sedans that are a bit slower.

When the sports car is rare (Low Frequency):
It zooms past the sedans. There is plenty of fuel for everyone, and the sports car eats its share quickly, leaving the sedans behind. The sports car wins big.

When the sports car is common (High Frequency):
Suddenly, you have a whole fleet of sports cars. They are all so efficient and fast that they eat all the fuel in the tank way too quickly. The fuel runs out, and the race stops early. Because the race ended early, the sports car didn't get to run long enough to build up a massive lead over the sedans. The "advantage" was canceled out by the fact that they ran out of gas too soon.

This is called negative frequency-dependence: The more common you are, the less of an advantage you have, because you deplete your own resources.

3. The "Rock-Paper-Scissors" Surprise (Non-Transitivity)

In school, we learn that if A beats B, and B beats C, then A must beat C. This is called transitivity.

The researchers found that bacteria break this rule.

• Imagine Strain A beats Strain B.
• Strain B beats Strain C.
• But... Strain C might actually beat Strain A!

This happens because the "winner" depends on the specific mix of the group. It's like the game Rock-Paper-Scissors. Rock beats Scissors, Scissors beats Paper, but Paper beats Rock. There is no single "best" bacterium; the winner changes depending on who is in the room. This makes predicting evolution much harder than we thought.

4. The "Hidden Battle" (It changes during the day)

The researchers didn't just look at the start and end of the race; they watched the whole 24-hour cycle. They found that the "winner" changes throughout the day!

• Early in the day: The mutant might be winning.
• Middle of the day: The ancestor might take the lead.
• End of the day: The mutant might win again.

The final result is just a messy average of these ups and downs. It's like a tug-of-war where the teams switch who is pulling harder every few minutes.

Why Does This Matter?

For a long time, scientists thought that in simple lab environments (like a test tube with just sugar and water), evolution was a simple, straight line. They thought mutations just made things "better" in a constant way.

This paper proves that even in a simple test tube, life is complex.

• Diversity stays alive: Because being "too good" makes you run out of resources, no single mutant can completely wipe out the others. This explains why we see so many different types of bacteria coexisting in nature.
• Evolution is slower: Because the advantage shrinks as you become common, it takes much longer for a "super-mutant" to take over a population.
• Prediction is hard: You can't just measure how fast a bug is in isolation and predict how it will do in the wild. You have to know who else is there.

In short: Evolution isn't just about having the best tool; it's about how that tool interacts with the crowd. Being rare is often the best strategy, and being common can actually be a disadvantage.

Technical Summary

1. Problem Statement

In classical population genetics and microbial evolution, the fitness effects of beneficial mutations are typically assumed to be constant, independent of the mutant's frequency within the population. This assumption underpins models of selective sweeps, epistasis, and the maintenance of genetic diversity. However, this framework ignores ecological realities where organisms interact with competitors and modify their shared environment. While frequency-dependent selection is known to maintain diversity in complex natural populations, it is often assumed to be negligible in simple, well-mixed laboratory environments (like the E. coli Long-Term Evolution Experiment, or LTEE) where ecological complexity is minimized.

The authors challenge this assumption, asking: Are frequency-dependent fitness effects rare exceptions or the norm, even among well-characterized beneficial mutations in simple laboratory settings?

2. Methodology

The study utilized a systematic, high-resolution approach to measure fitness across varying frequencies.

• Strains: The researchers used a panel of beneficial mutations derived from the early generations of the LTEE (specifically the ancestor REL606). These included five single mutants (G, R, P, S, T) representing common adaptation targets (e.g., pykF deletion, rbs operon deletion) and their corresponding double mutants.
• Experimental Design:
  • Fluorescent Tagging: Strains were tagged with Red (RFP) or Yellow (YFP) fluorescent proteins via a neutral Tn7 transposon integration to allow for precise tracking without affecting fitness.
  • Competition Assays: Competitions were conducted in DM25 (glucose-limited minimal media) under standard LTEE conditions (daily 1:100 dilution).
  • Frequency Variation: Competitions were initiated at four distinct starting frequencies for the focal strain: ~1%, 20%, 80%, and 99%.
  • Measurement: Fitness was measured using flow cytometry over two growth cycles.
• Fitness Metrics:
  • Fitness Effect ($s$): Calculated as the slope of logit-transformed frequencies over a cycle: $s(f_0) = \text{logit}(f_1) - \text{logit}(f_0)$.
  • Malthusian Fitness ($w$): Calculated as the ratio of growth rates (Malthusian parameters).
• High-Resolution Dynamics: To understand the mechanism, the authors performed within-growth cycle time-course experiments, sampling cocultures hourly for 10 hours (covering exponential to stationary phase) to track instantaneous growth rates and fitness changes.
• Statistical Analysis:
  • Slope Estimation: Orthogonal distance regression was used to determine the slope of frequency-dependence ($\beta_s$).
  • Variance Partitioning: ANCOVA models were used to partition variance in frequency-dependent slopes into invasion fitness, biological idiosyncrasy, and measurement error.
  • Transitivity Tests: The authors tested the additivity of fitness effects (transitivity) by comparing observed fitness in triads against predicted values based on pairwise comparisons.

3. Key Contributions

1. Ubiquity of Frequency-Dependence: The study provides the first systematic evidence that frequency-dependent fitness effects are the norm (occurring in ~80% of strain pairs) rather than the exception, even in the highly controlled LTEE environment.
2. Predictability of Slopes: The authors demonstrate that the strength of frequency-dependence is partially predictable. Invasion fitness (fitness when rare) explains approximately 50% of the biological variation in the frequency-dependent slopes.
3. Violation of Transitivity: The study reveals that fitness relationships are often non-transitive. The outcome of a competition between two strains cannot always be predicted by their individual fitnesses against a common reference strain, indicating context-dependent ecological interactions.
4. Mechanistic Insight (Niche Construction): Through high-resolution time-courses, the authors identified that frequency-dependence arises from niche construction (resource depletion dynamics). Specifically, the duration of the exponential growth phase is frequency-dependent because dominant strains deplete resources faster, altering the time available for fitness accumulation.

4. Key Results

• Prevalence: Approximately 75% of competitions exhibited significant negative frequency-dependence (fitness advantage declines as mutant frequency increases). In some cases (e.g., SG vs. G), the fitness effect crossed zero, suggesting potential for stable coexistence.
• Negative Correlation: There is a strong negative correlation (Pearson/Spearman $\rho \approx -0.5$ to $-0.6$) between invasion fitness ($s_{inv}$) and the slope of frequency-dependence. Strains with higher invasion fitness tend to exhibit steeper negative frequency-dependence.
• Epistasis and Sign Reversal: Double mutants often showed lower mean fitness than single mutants in specific conditions, a signature of sign epistasis. Furthermore, frequency-dependence could reverse the sign of fitness effects in specific contexts.
• Non-Transitivity: Significant violations of transitivity ($\nu \approx \pm 20\%$) were observed. This implies that competitive hierarchies are not static; the "winner" depends on the specific composition of the competing genotypes.
• Within-Cycle Dynamics:
  • Fitness effects are not constant throughout the growth cycle.
  • Phase 1: Mutants often grow slower initially but perform relatively better when the ancestor is in the majority.
  • Phase 2: Mutants gain advantages in mid-exponential phase, often stronger when the mutant is in the majority.
  • Phase 3: In late exponential/stationary transition, mutants generally perform better when the ancestor is in the majority.
  • Resource Depletion: The study quantified that resource depletion (shortening the exponential phase when mutants dominate) accounts for 10–50% of the frequency-dependent fitness effect. However, the majority of the effect is driven by other, strain-specific ecological interactions.
• Evolutionary Impact: Simulations indicate that negative frequency-dependence significantly slows the fixation of beneficial mutations, extending the time required for a mutant to sweep through a population by hundreds of generations.

5. Significance

• Paradigm Shift: The findings challenge the foundational assumption of constant fitness in microbial evolution models. Even in "simple" environments, subtle ecological interactions between closely related genotypes create complex, dynamic fitness landscapes.
• Maintenance of Diversity: The ubiquity of negative frequency-dependence provides a robust mechanism for the maintenance of genetic diversity and the stable coexistence of ecotypes observed in the LTEE, countering the homogenizing effects of directional selection.
• Experimental Design: The results suggest that standard fitness assays (measuring fitness at a single frequency, usually low) may misrepresent the true evolutionary potential of mutations. Future experiments must account for frequency-dependence to accurately predict evolutionary trajectories and epistatic interactions.
• Generalizability: The mechanisms identified (niche construction via resource competition) are likely universal across microbial systems, offering insights into how genetic diversity is generated and maintained in natural communities.

In conclusion, the paper demonstrates that ecological interactions are intrinsic to the evolution of beneficial mutations, fundamentally altering evolutionary trajectories even in the absence of complex environmental structures.