A pleiotropic hitchhiking model recapitulates alignments between fly wing divergence and variation

This paper proposes and validates a pleiotropic hitchhiking model demonstrating that univariate selection on wing size, acting through the structure of mutational variance, can explain both the alignment of Drosophila wing divergence with mutational lines of least resistance and the observed rate paradox without requiring hidden deleterious pleiotropic costs.

The Gist

Imagine you are watching a flock of birds fly. Over millions of years, you notice something strange: the birds' wings have changed shape in very specific directions, and they have changed at a very specific speed.

Scientists have long been puzzled by two things about this:

1. The Direction: The wings change in the exact same directions that random mutations (genetic "typos") naturally create. It's as if the birds are only flying down paths that are already paved with genetic possibilities.
2. The Speed: Despite having plenty of genetic "fuel" (variation) to change quickly, the wings evolve incredibly slowly. It's like having a Ferrari engine but driving in a 5 mph school zone.

This is the "Rate Paradox." Why are they so slow?

The Old Explanations (The "Hidden Hand" Theories)

Previously, scientists thought there must be a "hidden hand" holding them back.

• Theory A: Maybe changing the wing shape hurts the bird in other ways we can't see (like making its heart beat slower). This is called "deleterious pleiotropy."
• Theory B: Maybe the environment is constantly demanding specific, complex combinations of wing shapes, forcing the birds to stay in a narrow lane.

However, recent studies found no evidence that wing shape hurts the bird in other ways. So, the "hidden hand" theories are looking shaky.

The New Idea: The "Pleiotropic Hitchhiking" Model

The author of this paper, Haoran Cai, proposes a much simpler explanation. He suggests that nature isn't trying to change the wing shape at all. Nature is only trying to change the size of the wing.

Here is the analogy:

The "Train and the Passengers" Analogy

Imagine a train (the wing size) moving along a track. The train is the only thing the conductor (natural selection) cares about. The conductor wants the train to go faster or slower to match the wind.

Now, imagine the train is pulling 24 passenger cars (the wing shapes and vein patterns). These cars are not on their own tracks; they are physically bolted to the train.

• The Mechanism: Because the cars are bolted to the train, they have to move whenever the train moves. They "hitchhike" on the train's motion.
• The Constraint: The train (wing size) is under strict rules. It can't go too fast or too slow, or it crashes (the bird can't fly). This strict rule on the train creates a "drag" on the passenger cars.
• The Result: Even though the passenger cars (wing shapes) aren't being steered, they end up moving in a very specific, predictable pattern because they are tied to the train. They also move much slower than they would if they were free-floating, because the train's strict speed limits hold them back.

What the Paper Found

The author ran computer simulations based on real data from fruit flies to test this "Train and Passengers" idea.

1. It Works: When he simulated a world where only wing size was being selected for, the wing shapes naturally evolved in the exact same directions as the real-world data. The "hitchhiking" explained the alignment perfectly.
2. It Explains the Slowness: Because the "train" (size) is tightly controlled, the "passengers" (shape) are dragged along slowly. They can't drift away freely. This solves the "Rate Paradox" without needing to invent invisible fitness costs.
3. The "Squishy" Effect: The model predicts that because the size is being squeezed by selection, the genetic variation for size should be very low, while the variation for shape remains higher. Real-world data from fruit flies confirms this: the genes for size are "squeezed" tight, while the genes for shape are more relaxed.

Why This Matters

This paper suggests that we don't need to assume nature is constantly fine-tuning complex wing shapes. Instead, nature is just focused on one simple thing: How big is the wing?

The complex, beautiful geometry of the veins and the specific curves of the wing are just passengers along for the ride. They evolve slowly and in specific directions not because they are being carefully designed, but because they are stuck to the size of the wing, which is being carefully designed.

In short: The wing shape isn't the driver; it's just the passenger holding on tight to the driver's seat.

Technical Summary

1. Problem Statement

The paper addresses a central paradox in evolutionary biology regarding the alignment of mutational variance ($M$), standing genetic variance ($G$), and macroevolutionary divergence ($R$) in Drosophila wing shape.

• The Alignment: Empirical data shows a strong positive correlation between the structure of new mutations ($M$), existing genetic variation ($G$), and the direction of long-term species divergence ($R$). This suggests evolution follows "lines of least resistance" defined by mutational input.
• The Rate Paradox: While the direction of evolution aligns with available variation, the rate of evolution is orders of magnitude slower than quantitative genetic theory predicts based on the abundance of standing genetic variation.
• Existing Hypotheses & Limitations:
  • Deleterious Pleiotropy: Suggests hidden fitness costs on unmeasured traits constrain evolution. However, recent empirical tests found little evidence of such costs.
  • Correlational Selection: Suggests selection acts directly on trait covariances to maintain optimal shapes.
  • Moving Corridor: Suggests the fitness optimum moves along mutational lines of least resistance.
  • Gap: There is a need for a parsimonious model that explains both the alignment ($M-G-R$) and the slow evolutionary rate without invoking hidden fitness costs or complex multivariate selection regimes.

2. Methodology

The author proposes and tests a Pleiotropic Hitchhiking Model using a combination of theoretical derivation and individual-based simulations.

• Model Definition:
  • The system consists of $n$ wing traits ($z$).
  • Selection Target: Only one primary trait, wing size ($z_1$), is under direct stabilizing selection on a moving optimum.
  • Neutral Traits: Wing shape coordinates ($z_2, \dots, z_n$) are effectively neutral regarding fitness ($\beta_i \approx 0$).
  • Mechanism: Shape traits evolve solely as a correlated byproduct ("hitchhiking") due to mutational covariance with the selected size trait. This creates a "mutational drag" effect.
• Simulation Protocol:
  • Platform: Individual-based simulations with $N=500$ hermaphroditic individuals, non-overlapping generations.
  • Genetics: 50 unlinked diploid loci; alleles are fully pleiotropic (affecting all traits).
  • Parameters:
    • Mutational covariance matrix ($M$) derived from empirical data (Houle et al., 2017).
    • Selection strength ($V_s$) varied (Strong: $V_s=0.05$; Weak: $V_s=0.5$).
    • Optimum movement: 20 replicate populations evolved under divergent directional selection (optimum moves at constant velocity $v_i$).
  • Duration: 20,000 generations after a 10,000-generation burn-in.
• Analytical Tools:
  • Common Subspace Analysis (Jiang & Zhang, 2020): Projects $G$ and $R$ matrices onto the eigenvectors of $M$ to quantify alignment via log-log regression of variances.
  • Eigenvalue Diagnostics: Comparison of eigenvalue spectra (eccentricity) and effective dimensionality ($n_{eff}$) between $M$ and equilibrium $G$.
  • Metrics: Ratios of $G/M$ per trait, drift rates, and directional evolvability ratios.

3. Key Contributions

1. Novel Mechanism: Introduces "Pleiotropic Hitchhiking" as a parsimonious explanation for the $M-G-R$ alignment. It posits that complex multivariate selection is not strictly necessary; univariate selection on size can drive shape evolution via vertical pleiotropy.
2. Resolution of the Rate Paradox: Demonstrates that selection on a single primary trait can drastically reduce the effective drift rate of correlated neutral traits, explaining why evolution is slower than predicted by standing variation alone.
3. Distinction from Competing Hypotheses: Provides specific, testable predictions that distinguish hitchhiking from correlational selection, particularly regarding the "sphericalization" (flattening) of the genetic variance matrix ($G$) relative to the mutational matrix ($M$).

4. Key Results

• Recapitulation of Alignment: The simulation results successfully reproduced the empirical alignment. Both the simulated $G$ and $R$ matrices showed strong positive correlations with $M$ along principal axes, matching empirical patterns without invoking correlational selection on shape.
• Resolution of the Rate Paradox:
  • Under strong stabilizing selection on size ($V_s=0.05$), the drift rate of shape traits was reduced by approximately two orders of magnitude compared to neutral or weakly selected scenarios.
  • Mechanism: Strong selection on the primary trait erodes the genetic variance of the neutral trait ($G_{BB}$) and drives the genetic correlation toward the mutational correlation, effectively "leashing" the neutral trait to the constrained selected trait.
• Selective Depletion of Genetic Variance:
  • The model predicts that the $G/M$ ratio (standing variance per unit mutational input) should be lowest for the selected trait.
  • Empirical Validation: Analysis of empirical D. melanogaster data confirmed this: the $G/M$ ratio for centroid size was the lowest among all 25 traits (approx. 17–21% of the median shape trait), supporting the hypothesis of selective depletion.
• Eigenvalue Structure (Hitchhiking vs. Correlational Selection):
  • Hitchhiking Prediction: Selection on a single axis reduces the eccentricity of $G$ relative to $M$ (making $G$ more "spherical" or higher dimensional).
  • Correlational Selection Prediction: Selection preserves the eccentricity of $M$.
  • Result: Simulations showed that hitchhiking significantly increased the effective dimensionality ($n_{eff}$) and reduced the leading eigenvalue proportion of $G$ compared to $M$. Empirical data from Dugand et al. (2021) matched the hitchhiking signature (higher $n_{eff}$, lower evolvability ratio along the size axis) rather than the correlational selection signature.

5. Significance

• Theoretical Impact: The paper challenges the necessity of invoking "hidden" deleterious pleiotropy or complex multivariate selection to explain evolutionary constraints. It suggests that vertical pleiotropy (size-shape integration) combined with univariate selection is sufficient to generate observed macroevolutionary patterns.
• Evolutionary Stasis: It offers a mechanism for evolutionary stasis in complex traits: even if a trait is not directly selected, it can be evolutionarily constrained if it is genetically correlated with a trait under strong stabilizing selection.
• Future Directions: The study highlights that while hitchhiking explains the alignment and a significant portion of the rate reduction, it may not fully account for the extreme magnitude of the rate paradox ($10^{-4}$ of neutral rates). This suggests a potential interplay between hitchhiking and other factors (e.g., environmental variance, horizontal pleiotropy, or selection on multiple targets).
• Methodological Rigor: By using individual-based simulations parameterized with empirical $M$ matrices and comparing specific eigenvalue diagnostics, the study provides a robust framework for distinguishing between competing evolutionary hypotheses.

In conclusion, Haoran Cai demonstrates that pleiotropic hitchhiking—where wing shape evolves as a neutral passenger to selection on wing size—is a powerful, parsimonious mechanism that aligns microevolutionary variance with macroevolutionary divergence while simultaneously resolving the paradox of slow evolutionary rates.