System drift in the evolution of plant meristem development

This study combines computational modeling and empirical genomic data to demonstrate that developmental system drift, characterized by the continual rewiring of gene regulatory networks and turnover of conserved non-coding sequences without altering phenotypic outcomes, is a pervasive mechanism in the evolution of complex plant developmental systems.

The Gist

The Big Idea: The "Same Cake, Different Recipe"

Imagine you bake a perfect chocolate cake. It tastes delicious, looks beautiful, and everyone loves it. Now, imagine that over the next 1,000 years, your family keeps baking this exact same cake. But here's the twist: every few generations, someone swaps out an ingredient. Maybe they stop using cocoa powder and start using carob. Maybe they switch from baking soda to baking powder. Maybe they change the oven temperature.

Despite all these changes to the recipe (the instructions), the cake (the final result) tastes exactly the same.

This is the core concept of Developmental System Drift (DSD). It's the idea that living things can keep their physical appearance (phenotype) exactly the same for millions of years, even while the genetic instructions (genotype) underneath are constantly changing, swapping parts, and rewiring themselves.

The Study: Plants as the Test Kitchen

The authors of this paper wanted to know: Does this happen in complex plants, or is it just a trick of simple systems?

They focused on the Shoot Apical Meristem (SAM). Think of the SAM as the "construction site" at the very tip of a plant stem. It's a tiny cluster of stem cells that builds all the leaves, flowers, and stems above ground. It's a critical, complex machine that has to work perfectly for the plant to survive.

How They Did It: The Digital Evolution Lab

Since we can't wait 50,000 years to watch plants evolve in real-time, the scientists built a computer simulation.

1. The Virtual Plants: They created a digital population of plants. Each plant had a "genome" (a list of rules) that told its cells how to talk to each other.
2. The Goal: The plants had to build a specific pattern of cells at their tip (like a specific arrangement of bricks). If they built it right, they got a high "fitness score" and got to reproduce.
3. The Mutation: When they reproduced, their rules got slightly scrambled (mutated), just like in real life.
4. The Experiment: They ran this simulation for 50,000 generations.

The Surprising Discovery: The "Deeply Conserved" Lie

At first, the computer plants evolved quickly. They found a set of rules that built the perfect tip. Some of these rules seemed so important that they never changed for thousands of generations. The scientists thought, "Aha! These are the essential, unchangeable rules of life!"

But then, they kept watching.

Over a longer period (another 50,000 generations), something amazing happened. Even those "essential" rules started to disappear and get replaced by different rules.

• The Old Way: Gene A told Gene B to turn on.
• The New Way: Gene A stopped talking to Gene B. Instead, Gene C started talking to Gene B.

The result? The plant still built the perfect tip. The "cake" was still perfect. But the "recipe" had completely changed.

The Analogy: Imagine a city where the traffic lights are broken. Instead of fixing the lights, the city installs a new system where police officers stand at every intersection directing traffic. The traffic flows perfectly (the phenotype is conserved), but the mechanism (the genotype) has completely shifted from electronic lights to human officers.

Why Does This Happen? (The "Safety Net")

The paper explains that this happens because of redundancy.

In a complex system, there are often multiple ways to do the same job.

• If you have a backup generator, you can turn off the main power without the lights going out.
• In the plant's genetic network, if one gene stops working, another gene can step in and take over the job.

Once a new gene steps in, the old gene becomes "optional." Over time, the old gene might disappear or change its job, and the new gene becomes the "essential" one. This creates a step-by-step rewiring of the entire system.

The Real-World Proof: Looking at Real Plants

To prove this wasn't just a computer glitch, the scientists looked at real data from six different plant species (ranging from close cousins like Arabidopsis to distant relatives like Brachypodium).

They looked at Conserved Non-Coding Sequences (CNSs).

• Analogy: Think of DNA as a book. The "coding" parts are the words that make the story. The "non-coding" parts are the footnotes, margins, and instructions on how to read the words (the switches that turn genes on/off).
• The Finding: They found that these "instruction manuals" (CNSs) were changing rapidly between species. Some instructions were lost, new ones appeared.
• The Result: Even though the instructions were totally different, the plants still had the same gene expression patterns (the story was told the same way).

Why Should We Care?

This changes how we understand evolution:

1. Evolution is Messy: We often think evolution is a straight line where "better" genes replace "worse" ones. This paper shows evolution is more like a game of "Telephone" where the message stays the same, but the person passing it along changes every time.
2. Hidden Diversity: Two plants might look identical and act identical, but their internal genetic wiring could be completely different. This means there is a huge amount of "hidden" genetic diversity that we can't see just by looking at the plant.
3. Resilience: This "drift" makes life robust. Because there are so many ways to build the same thing, life can survive massive genetic changes without falling apart.

Summary

The paper tells us that nature is incredibly flexible. Plants can keep their shape and function perfectly stable for millions of years, even while their internal genetic machinery is constantly being dismantled and rebuilt with new parts. It's like a house that looks exactly the same from the outside, but every few decades, the plumbing, wiring, and foundation are completely replaced with new materials, yet the house never stops being a home.

Technical Summary

1. Problem Statement

The central question in evolutionary developmental biology (evo-devo) addressed by this study is the evolvability of complex phenotypes. Specifically, the authors investigate Developmental System Drift (DSD): a phenomenon where a phenotypic trait remains conserved over evolutionary time while the underlying genetic and regulatory mechanisms diverge.

• The Debate: While DSD is well-documented in simpler genotype-phenotype maps (GPMs) like RNA folding, its prevalence in complex, multicellular developmental systems is debated. Some theories suggest complex phenotypes occupy isolated "islands" in genotype space, limiting neutral evolution. Others suggest extensive neutral paths exist.
• The Gap: DSD in plants, particularly in the Shoot Apical Meristem (SAM) (the stem cell niche generating above-ground tissues), has received little attention. The study aims to determine if the regulatory networks governing SAM development can undergo extensive rewiring (neutral evolution) while maintaining the functional tissue pattern.

2. Methodology

The authors employed a dual approach combining computational evolutionary modeling and bioinformatic analysis of empirical plant data.

A. Computational Model (In Silico Evo-Devo)

• System: A population of 1,000 individuals, each representing a simulated SAM tissue (a 2D grid of 130 cells).
• Genotype: Modeled as a "beads-on-a-string" genome containing genes (encoding Transcription Factors, TFs) and Transcription Factor Binding Sites (TFBSs).
• Phenotype: The spatiotemporal distribution of protein concentrations within the tissue, governed by a Gene Regulatory Network (GRN).
  • TFs have specific properties: diffusion, dimerization, cell-cell communication, and cell autonomy.
  • Development is modeled using stochastic differential equations (SDEs) to account for molecular noise.
• Selection: Fitness is determined by the expression pattern of two specific "fitness genes" in specific regions: the Central Zone (CZ) and the Organizing Center (OC).
• Evolutionary Process:
  1. Adaptation Phase (0–50,000 generations): Populations evolve to achieve a target expression pattern (fitness score ≥ 75).
  2. Stabilizing Phase (50,000–100,000 generations): The top 8 populations were cloned (5 clones each) and evolved further under stabilizing selection to observe neutral drift without fitness gains.
• Analysis Metrics:
  • Conservation: Tracking how long specific regulatory interactions persist in ancestral lineages.
  • Divergence: Calculating the distance between GRNs using Boolean adjacency matrices (qualitative differences in interactions).
  • Robustness: Measuring mutational robustness (fitness of mutated offspring) and developmental robustness (variance in clonal expression).

B. Bioinformatic Analysis (Empirical Validation)

• Datasets:
  1. Conserved Non-coding Sequences (CNSs): From the "Conservatory Project" (Hendelman et al., 2021), covering 284 species.
  2. Gene Expression: Organ-specific RNA-seq data from Schuster et al. (2026).
• Species: Six angiosperms: Arabidopsis thaliana, A. lyrata, Capsella rubella, Eutrema salsugineum, Medicago truncatula, and Brachypodium distachyon.
• Target Genes: Focus on SEPALLATA (SEP) MADS-box genes (critical for floral meristem identity) and other meristem-related orthogroups.
• Method:
  • Mapped CNSs to orthologous genes.
  • Quantified CNS turnover (gain/loss of sequences) across species.
  • Correlated CNS similarity with gene expression pattern similarity (using Pearson distance).
  • Hypothesis: If DSD occurs, genes with similar expression patterns should show high CNS turnover (disjoint CNS sets).

3. Key Contributions

1. Demonstration of DSD in Complex GRNs: The study provides strong evidence that even in complex, multicellular developmental systems, the underlying regulatory architecture is highly evolvable and prone to drift.
2. Mechanism of Rewiring: Identified functional redundancy as the primary driver allowing conserved interactions to be lost and replaced. New interactions can emerge to compensate for lost ones, maintaining the phenotype while the genotype changes.
3. Empirical Evidence in Plants: Validated the computational predictions using real-world plant genomic data, showing that deeply conserved non-coding sequences (CNSs) are frequently lost or gained without altering the gene expression phenotype.
4. Reframing Robustness: Showed that robustness (mutational and developmental) does not necessarily increase during stabilizing selection; instead, it fluctuates, allowing for neutral exploration of genotype space.

4. Key Results

Computational Simulation Results

• Conservation and Loss: During the adaptation phase, a subset of regulatory interactions became highly conserved (>5,000 generations). However, during the subsequent stabilizing phase, these "essential" interactions were frequently lost and replaced by new interactions.
• Neutral Drift: The full GRNs diverged rapidly from the common ancestor. Even when restricting analysis to only the "conserved" interactions, significant divergence was observed between independent lineages, indicating that the regulatory logic itself changes over time.
• Phenotypic Stability: Despite massive rewiring of the GRN, the target expression pattern (fitness phenotype) remained stable.
• Redundancy Mechanism: The study traced specific rewiring events (e.g., Gene A regulating Gene B is replaced by Gene C regulating Gene B). This occurred because a new interaction emerged that provided functional redundancy, lowering the "importance" of the original interaction to near zero, allowing it to be deleted without fitness cost.
• Robustness: Mutational and developmental robustness fluctuated but did not show a consistent trend of increasing, suggesting that drift, not adaptation of robustness, drives the divergence.

Bioinformatic Analysis Results

• High CNS Turnover: In the six angiosperm species, there was extensive turnover of CNSs. Many CNSs were conserved within closely related species (e.g., Brassicaceae) but lost in more distant ones, or vice versa.
• Decoupling of Genotype and Phenotype:
  • Genes with similar CNS sets generally showed similar expression patterns.
  • Crucially, genes with disjoint (completely different) CNS sets often exhibited highly similar expression patterns.
  • This lack of correlation between CNS composition and expression pattern provides empirical evidence for DSD in plants: the regulatory "wiring" changes, but the output remains the same.

5. Significance and Implications

• Evolutionary Flexibility: The findings challenge the view that complex developmental systems are rigid. Instead, they possess a "many-to-one" mapping where numerous distinct genotypes can produce the same phenotype.
• Speciation and Adaptation: DSD allows populations to explore different regions of genotype space. This accumulation of cryptic genetic variation can facilitate rapid adaptation to new environments or contribute to reproductive isolation (speciation) even when morphology is conserved.
• Plant Evolution: The study explains how plant meristems can maintain structural homology (e.g., the SAM) across millions of years of evolution despite significant changes in the underlying gene regulatory networks.
• Methodological Advance: The combination of in silico evolution with large-scale comparative genomics provides a robust framework for studying evolutionary dynamics in complex systems where direct experimental manipulation is difficult.

In conclusion, the paper establishes that Developmental System Drift is pervasive in plant evolution, driven by the ability of gene regulatory networks to rewire themselves through redundancy while maintaining functional stability. This suggests that the conservation of a phenotype does not imply the conservation of the genetic mechanisms that produce it.