Caught in the web: galaxy mergers along cosmic filaments

Using deep-learning analysis of 43,922 galaxies from the 4MOST CHANCES survey, this study reveals that galaxy mergers are significantly more likely to occur near cosmic web filaments than in other regions, suggesting that these filaments facilitate pre-processing interactions that transform galaxies before they enter cluster cores.

The Gist

Imagine the universe not as a random scattering of stars, but as a giant, invisible spiderweb stretching across the cosmos. This is the "cosmic web."

In this cosmic web, galaxies are like houses. Some houses are lonely, but most are part of neighborhoods called galaxy groups or massive cities called galaxy clusters. The "roads" connecting these neighborhoods are the filaments—long, thin strands of dark matter and gas that guide galaxies toward the big clusters.

For a long time, astronomers thought these filaments were just passive highways. They believed galaxies simply drove along them until they crashed into the chaotic, high-speed traffic of a galaxy cluster, where they would get stripped of their gas and stop making new stars.

But this new paper suggests the filaments are actually active construction zones.

Here is the story of what the researchers found, broken down simply:

1. The Setup: A Massive Traffic Study

The team, led by Carolina Dulcien and Yara Jaffé, used a powerful new telescope survey called CHANCES to look at 33 nearby galaxy clusters. They didn't just look at the clusters themselves; they looked at the vast area around them, out to five times the size of the cluster's main body.

They analyzed nearly 44,000 galaxies. To figure out which ones were crashing into each other (merging), they used a super-smart AI named Zoobot. Think of Zoobot as a digital art critic trained on millions of human opinions. It looked at photos of galaxies and said, "That one looks like two galaxies hugging and merging," or "That one looks like a peaceful, single galaxy."

They found about 700 galaxies that were in the middle of a merger.

2. The Discovery: The "Web" is Where the Action Happens

The researchers asked a simple question: Where are these merging galaxies located?

• The Old Idea: Maybe mergers happen mostly inside the crowded cluster cities, where galaxies are packed tight.
• The New Finding: No! The merging galaxies were found much closer to the filaments (the cosmic roads) than the peaceful, non-merging galaxies.

This was especially true outside the main cluster. Once galaxies get deep inside the cluster, they are moving so fast that they zoom past each other without colliding (like cars on a highway going 100 mph). But in the filaments, the traffic is slower. The "roads" act like a gentle funnel, slowing galaxies down just enough so they can bump into each other and stick together.

3. The Analogy: The "Pre-Processing" Kitchen

Think of a galaxy cluster as a fancy restaurant kitchen where the chefs (galaxies) are under extreme pressure. The environment is so harsh that the ingredients (gas) get stripped away, and the chefs stop cooking (star formation stops).

The researchers found that before the galaxies even enter this high-pressure kitchen, they are getting "pre-processed" in the filaments, which act like a prep kitchen or a dining hall just outside the restaurant.

• In the prep kitchen (the filament): Galaxies are moving slowly. They have time to bump into each other, merge, and mix their ingredients (gas and stars). This changes their shape and personality before they ever enter the main cluster.
• In the main kitchen (the cluster): By the time they arrive, they have already been transformed. They are no longer the same galaxies they were when they started their journey.

4. Why This Matters

This changes how we understand the life cycle of a galaxy.

• Filaments aren't just roads; they are factories. They aren't just moving galaxies; they are actively changing them.
• The "Pre-Processing" is real. A huge chunk of the changes we see in galaxies inside clusters actually happened before they got there, while they were still traveling along the cosmic web.

The Bottom Line

The universe is a busy place. Galaxies don't just drift aimlessly; they travel along cosmic highways. And as they travel these highways, they are bumping into neighbors, merging, and changing their shapes. By the time they reach the massive clusters, they have already been "caught in the web" and transformed.

The filaments are the stage where the drama of galaxy evolution begins, long before the galaxies reach the main event.

Technical Summary

Here is a detailed technical summary of the paper "Caught in the web: galaxy mergers along cosmic filaments" by Dulcien et al. (2026).

1. Problem Statement

The hierarchical structure formation paradigm posits that galaxies accrete into massive clusters via large-scale cosmic filaments. While it is well-established that dense cluster cores drive galaxy evolution through mechanisms like ram-pressure stripping, a significant fraction of galaxies are already "quenched" (star formation halted) upon entering clusters. This suggests a process of "pre-processing" occurs in lower-density environments (groups, filaments) prior to cluster infall.

Despite theoretical predictions that galaxy-galaxy mergers are more efficient in intermediate-density environments due to lower relative velocities, observational evidence specifically linking galaxy mergers to cosmic filaments feeding clusters has been limited. The specific role of filaments in facilitating these mergers as a pre-processing mechanism remains unexplored systematically.

2. Methodology

Data Sample:

• Source: The study utilizes the 4MOST CHANCES survey (Low-z sub-survey), targeting galaxies in and around 33 low-redshift clusters ($z < 0.07$).
• Sample Size: A total of 43,922 galaxies were analyzed, selected from the DESI Legacy Surveys Data Release 10 (LS-DR10) with photometric redshifts.
• Selection Criteria: Galaxies were limited to $m_r < 18.5$ mag to ensure high completeness ($\sim80\%$) and purity of cluster members. The analysis covers regions from the cluster center out to $5 \times R_{200}$(where$R_{200}$ is the virial radius).

Morphological Classification (Merger Identification):

• Tool: The authors used Zoobot, a deep-learning framework trained on Galaxy Zoo volunteer classifications.
• Input: Cross-matched with Galaxy Zoo DESI (GZDESI) catalogues derived from DESI Legacy Imaging Surveys DR8.
• Criteria: Mergers were identified based on predicted vote fractions for specific categories:
  • Major Merger (M-M) or Major Disturbance (M-D).
  • Threshold: $F_{MM} \geq 0.5$ OR $F_{MD} \geq 0.5$.
• Resulting Sample: This conservative selection yielded 698 merger candidates ($\sim2\%$ of the sample), prioritizing high purity (78%) over completeness (45%). The remaining 43,224 galaxies served as the non-merging control sample.

Environmental Characterization:

• Filament Detection: Cosmic web filaments were identified in 2D using DisPerSE (Discrete Persistence Structures Extractor) applied to the projected galaxy distribution.
  • Parameters: Persistence threshold at $3\sigma$, smoothness level of 20.
• Distance Metrics:
  • $D_{fil}$: Distance from each galaxy to the nearest filament.
  • $r_{cl}$: Projected distance from the cluster center.
• Regions Analyzed:
  1. Entire region ($r < 5 \times R_{200}$).
  2. Inner cluster ($r < R_{200}$).
  3. Cluster outskirts ($R_{200} < r < 5 \times R_{200}$).

Statistical Analysis:

• The Anderson-Darling (AD) k-sample test was used to compare the distribution of filament distances ($D_{fil}$) between merging and non-merging galaxies.
• A significance threshold of $\alpha = 0.07$ was adopted to account for observational biases toward filamentary environments.

3. Key Results

• Spatial Association: Galaxy mergers are significantly closer to cosmic filaments than the general non-merging population.
• Radial Dependence: The trend is most pronounced in the cluster outskirts ($R_{200} < r < 5 \times R_{200}$).
  • Median Distance: Merging galaxies in the outskirts have a median filament distance of 0.406 Mpc, compared to 0.459 Mpc for non-merging galaxies.
  • Statistical Significance: The AD test yields $p$-values well below 0.07 outside the virial region, indicating a statistically significant difference in the distribution shapes.
• Case Study (Abell 85): In the massive relaxed cluster Abell 85, mergers clearly trace the filamentary structure. The median distance difference between mergers and non-mergers is 0.29 Mpc, with a $p$-value of 0.001.
• Cluster-by-Cluster Analysis: Approximately 39% of the individual clusters in the sample show statistically suggestive evidence ($p < 0.07$) of merger-filament association in the outer regions.
• Mass Distribution: The stellar mass distributions of mergers and non-mergers are similar, and matching them does not alter the primary results.

4. Key Contributions

1. First Direct Observational Evidence: This study provides the first direct observational confirmation that galaxy merger activity is enhanced specifically along the filaments feeding galaxy clusters.
2. Quantification of Pre-processing: It quantifies the spatial relationship between mergers and the cosmic web, demonstrating that filaments are not merely passive conduits for galaxy infall but active sites of morphological transformation.
3. Methodological Framework: The paper establishes a robust pipeline combining deep-learning morphology classification (Zoobot) with topological filament detection (DisPerSE) on large photometric samples to study environmental effects.

5. Significance and Implications

• Mechanism of Pre-processing: The findings support a scenario where filaments moderate the relative velocities of infalling galaxies and enhance gas availability, creating conditions conducive to gravitational binding and mergers. These mergers likely trigger starbursts or feed AGNs, altering galaxy properties before they reach the quenching environment of the cluster core.
• Cosmic Web Dynamics: The results validate simulations predicting that merger rates are higher in intermediate-mass haloes and filamentary structures compared to dense cluster cores, where high relative velocities inhibit mergers.
• Future Directions: The authors note limitations regarding photometric redshift uncertainties and conservative merger selection. Future work will utilize the full CHANCES spectroscopic sample, citizen science projects, and hydrodynamical simulations to further constrain the physical drivers of filament-driven evolution.

Conclusion: The paper concludes that cosmic filaments play a critical, active role in galaxy evolution by promoting mergers as a pre-processing mechanism, fundamentally shaping the galaxy population before it enters the virialized cluster environment.