Wing shape evolution is not constrained by ancestral genetic covariances in the invasive Drosophila suzukii

This study demonstrates that the rapid evolution of wing shape during the global invasion of *Drosophila suzukii* was driven by selection and drift rather than being constrained by ancestral genetic covariances, as the genetic covariance matrix remained relatively stable while phenotypic divergence exceeded neutral expectations.

The Gist

Imagine a group of tiny, invasive flies called Drosophila suzukii (spotted wing drosophila) that escaped from their home in Asia and spread across Europe and North America. They are like a biological "invader" army, arriving in new territories and trying to survive.

Scientists wanted to know: As these flies moved to new places, did their bodies change because they were forced to adapt, or did they just drift along randomly? More specifically, they looked at the shape of the flies' wings.

To understand this, the researchers used a concept called the G-Matrix. Think of the G-Matrix as the "Genetic Blueprint" or the "DNA Playbook" for the flies.

• The Analogy: Imagine the flies' DNA is a giant, flexible trampoline. The G-Matrix describes how bouncy that trampoline is in different directions.
  • If the trampoline is very bouncy in one direction but stiff in another, the flies can only evolve easily in that one direction. This is a constraint.
  • If the trampoline is bouncy and round in every direction (like a perfect sphere), the flies can evolve in any direction they want. This is no constraint.

The Big Questions

The scientists asked four main questions:

1. Did the flies' wings change shape as they invaded new lands?
2. Did their "Genetic Playbook" (the G-Matrix) stay the same, or did it get twisted and reshaped by the journey?
3. Were the flies stuck evolving in only one direction because of their ancient DNA, or were they free to go anywhere?
4. Was this change driven by natural selection (survival of the fittest) or just genetic drift (random luck)?

What They Found

1. The Wings Did Change, But Not in a Straight Line
The flies in France and the USA developed slightly different wing shapes compared to their ancestors in Japan. However, the French flies changed in one direction, and the American flies changed in a completely different direction. It wasn't a uniform "team effort" to change the same way; each group found its own path.

2. The "Playbook" Stayed Mostly the Same
Despite the long journey and the stress of invading new lands (which usually messes up DNA), the shape of the Genetic Playbook remained remarkably stable.

• The Metaphor: Imagine if you moved a family from a small house to a huge mansion. You might lose some furniture (genetic diversity) or gain some new pieces (mixing with other families), but the shape of the house (the blueprint) stayed roughly the same.
• The scientists found that the "trampoline" of the flies' DNA was still round and bouncy in all directions. It didn't get squashed or stretched into a weird shape. This means the flies were not constrained by their ancestors' DNA. They had the freedom to evolve in any direction they needed.

3. It Wasn't Just Random Luck (Selection vs. Drift)
If the changes were just random luck (drift), the differences between the groups would be small and predictable. But the scientists found the differences were larger than expected by chance.

• The Conclusion: This suggests that natural selection was at work. Something in the new environments (maybe temperature, food, or flight conditions) pushed the flies to change their wing shapes.

4. Was it for Faster Flight?
The researchers wondered: "Did they change their wings to fly faster?" They tested this by simulating a scenario where faster flight was the goal.

• The Result: The actual changes the flies made did not match the changes needed to fly faster. So, while selection was happening, it probably wasn't about speed. It might be about something else entirely, like mating rituals or surviving different temperatures.

The Takeaway

This study is like watching a group of travelers move from a small village to different cities around the world.

• Old Theory: We might have thought their ancient family rules (DNA) would force them to change in only one specific way.
• New Reality: The family rules were flexible (a spherical G-Matrix). The travelers were free to change in whatever way the new city demanded. They didn't get stuck in a rut; they adapted freely, driven by the pressure of their new environments, not by the limitations of their past.

In short: The invasive flies didn't have their wings tied behind their backs by their ancestors. They had the genetic freedom to evolve, and they used that freedom to adapt to their new homes, likely due to specific environmental pressures we are still trying to identify.

Technical Summary

1. Problem Statement and Research Questions

The study addresses a central question in evolutionary biology: To what extent are phenotypic evolution and adaptation constrained by ancestral genetic covariances (the $G$-matrix)?

Specifically, the authors investigate whether the rapid phenotypic divergence observed in invasive populations is channeled along "lines of least evolutionary resistance" (defined by the ancestral $G$-matrix) or if it occurs freely in response to selection and drift. They focus on the worldwide invasion of the spotted wing drosophila, Drosophila suzukii, which has spread from its native range in Asia to Europe and North America within the last decade.

The study aims to answer four specific questions:

1. How has wing shape evolved across multiple invasive populations?
2. Has the structure of the genetic variance-covariance matrix ($G$) remained stable during the invasion?
3. Has phenotypic divergence been constrained by ancestral genetic covariances?
4. Can the observed divergence be attributed to natural selection rather than genetic drift?

2. Methodology

Study System and Sampling:

• Populations: Six natural populations were sampled: two from the native range (Japan: Sapporo and Tokyo) and four from invasive ranges (USA: Dayton and Watsonville; France: Paris and Montpellier).
• Breeding Design: An isofemale line breeding design was used. Wild-caught gravid females were used to establish lines, which were then expanded via full-sib mating for five generations to minimize environmental variance and isolate genetic effects.
• Sample Size: 1,323 individuals across 177 isofemale lines were phenotyped.

Data Acquisition and Phenotyping:

• Geometric Morphometrics: Right wings were mounted and imaged. Fifteen landmarks were digitized on the dorsal side.
• Shape Analysis: Procrustes Generalized Least Squares (GLS) superimposition was used to remove size, rotation, and translation effects. Principal Component Analysis (PCA) reduced the data to 26 non-null components representing wing shape. Centroid size was used as a covariate to account for allometry.
• Divergence Vectors: Phenotypic divergence vectors were calculated between invasive populations and the ancestral mean (average of Japanese populations).

Quantitative Genetic Estimation:

• $G$-Matrix Estimation: Six population-specific $G$-matrices were estimated using a Bayesian multivariate animal model (via the MCMCglmm R package). The model partitioned phenotypic variance into additive genetic variance and residual error, incorporating pedigree information.
• Stability Tests:
  • Size and Shape: Compared matrix volumes (sum of eigenvalues) and shapes (eccentricity/ratio of eigenvalues) using 95% Highest Posterior Density (HPD) intervals.
  • Covariance Tensors: Used a fourth-order genetic covariance tensor analysis (Hine et al., 2009) to identify independent axes of divergence among the six $G$-matrices.
• Constraint and Selection Tests:
  • Proportionality ($D$ vs. $G$): Calculated the angle between the divergence matrix ($D$) and the ancestral $G$-matrix to test for alignment (constrained evolution).
  • Multivariate $Q_{ST}$-$F_{ST}$: Compared the coefficient of proportionality ($\rho_{ST}$) between $D$ and $G$ against neutral expectations derived from microsatellite markers ($2F_{ST}/(1-F_{ST})$). A significant deviation ($\rho_{ST} > F_{ST}$) indicates divergent selection.
  • Breeder's Equation Simulation: Solved the multivariate breeder's equation ($\Delta z = G\beta$) using a hypothetical selection gradient based on flight speed (from a previous study) to see if the predicted response matched observed divergence.
  • Variance Depletion: Tested if genetic variance was depleted along divergence vectors, which would indicate strong selection.

3. Key Results

Phenotypic Divergence:

• Significant differentiation in wing size and shape was found among all populations.
• Directionality: French populations diverged in a similar direction from the ancestor, whereas American populations diverged in different directions. The average divergence vectors for France and the USA were not similar to each other (angles close to random vectors), suggesting independent evolutionary trajectories.

Stability of the $G$-Matrix:

• Volume (Size): The total genetic variance (volume of $G$) was generally stable. The only significant difference was that the US-WAT population had a larger volume than US-DAY and the French populations. This likely reflects genetic admixture in US-WAT.
• Shape (Eccentricity): All $G$-matrices were roughly spherical (low eccentricity, eigenvalues close to 1). This indicates that genetic variation is distributed relatively equally across all directions of the phenotypic space, with no single "line of least resistance."
• Tensor Analysis: Covariance tensor analysis confirmed global stability; the observed divergence among $G$-matrices did not significantly differ from a null distribution generated by random shuffling.

Constraints and Selection:

• Lack of Constraint: Despite the $D$ and $G$ matrices being proportional (aligned), the spherical nature of $G$ implies a lack of constraint. Populations were free to evolve in various directions because genetic variation existed in all directions.
• Evidence of Selection: The multivariate $Q_{ST}$-$F_{ST}$ test revealed that phenotypic divergence ($\rho_{ST}$) was significantly greater than expected under neutral drift alone ($2F_{ST}/(1-F_{ST})$). This strongly supports a role for divergent selection.
• Flight Speed Hypothesis: Simulating selection for increased flight speed did not reproduce the observed divergence patterns. The predicted response to selection did not match the actual divergence vectors, suggesting flight speed is not the primary driver of the observed shape changes.
• Variance Depletion: No significant reduction in genetic variance was found along the divergence vectors, further suggesting that selection has not yet depleted standing genetic variation in these specific directions.

4. Key Contributions

1. Empirical Evidence of $G$-Matrix Stability: The study provides robust evidence that the $G$-matrix for wing shape remained stable in structure (spherical) and largely in volume over a short evolutionary timescale (<10 years) despite severe demographic bottlenecks and admixture events.
2. Decoupling of Drift and Selection: By combining $Q_{ST}$-$F_{ST}$ comparisons with $G$-matrix stability analysis, the authors demonstrate that while drift occurred (evidenced by bottlenecks), the observed phenotypic divergence is driven by divergent selection, not just drift.
3. Absence of Genetic Constraints: The study challenges the assumption that ancestral genetic covariances constrain rapid adaptation. The spherical $G$-matrix of D. suzukii allowed for unconstrained evolution in multiple directions, facilitating rapid adaptation to new environments.
4. Methodological Integration: The paper effectively integrates geometric morphometrics, Bayesian quantitative genetics, and tensor analysis to dissect the evolutionary dynamics of a complex multivariate trait in a natural invasion scenario.

5. Significance

This research is significant for invasion biology and evolutionary theory for several reasons:

• Rapid Adaptation: It demonstrates that invasive species can undergo significant adaptive evolution in very short timeframes (less than a decade) without being hindered by ancestral genetic constraints.
• Role of Demography: It clarifies that while demographic events (bottlenecks, admixture) can alter the volume of genetic variance, they do not necessarily reshape the orientation or structure of the $G$-matrix in the short term.
• Predictability of Evolution: The findings suggest that in species with "spherical" $G$-matrices (high evolvability in all directions), evolutionary trajectories may be less predictable based solely on ancestral genetics and more dependent on specific, local selective pressures.
• Pest Management: Understanding the genetic architecture and evolvability of D. suzukii is crucial for predicting its future spread and potential for adapting to control measures or new climates.

In conclusion, the study concludes that the early stages of D. suzukii wing shape evolution were driven by selection and drift acting on a flexible, unconstrained genetic architecture, rather than being channeled by ancestral genetic covariances.