🔭 astrophysics

Rise of the forsaken relics: connecting present-day stellar streams and phase-mixed galaxies to the Epoch of Reionization

By incorporating disrupted progenitors like stellar streams and phase-mixed galaxies into "near-far" analyses using FIRE-2 simulations, this study demonstrates that these overlooked systems contribute roughly half of the proto-Milky Way's stellar mass at reionization, thereby significantly improving the accuracy and reducing the variance of inferred high-redshift luminosity functions down to fainter magnitudes than previously possible.

Original authors: Aritra Kundu, Robyn Sanderson, Adam Lidz, Pratik J. Gandhi, Andrew Wetzel, Robert Feldmann, Nondh Panithanpaisal, Jasjeev Singh, Michael Boylan-Kolchin

Published 2026-03-03

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Aritra Kundu, Robyn Sanderson, Adam Lidz, Pratik J. Gandhi, Andrew Wetzel, Robert Feldmann, Nondh Panithanpaisal, Jasjeev Singh, Michael Boylan-Kolchin

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the universe as a giant, bustling city that has been growing for 13.8 billion years. The "Epoch of Reionization" was the city's chaotic teenage years, roughly 13 billion years ago. During this time, the first stars and galaxies turned on their lights, burning away a thick fog of neutral hydrogen that had blanketed the universe, making it transparent for the first time.

Astronomers want to know: Which specific buildings (galaxies) were responsible for turning on the lights? Were they just the massive skyscrapers, or were there millions of tiny, dim streetlamps (low-mass galaxies) doing the heavy lifting?

The problem is that the "teenage" galaxies are so far away and faint that even our most powerful telescopes (like the James Webb Space Telescope) can only see the biggest skyscrapers. The tiny streetlamps are too dim to spot directly.

The "Near-Far" Detective Story

To solve this mystery, astronomers use a clever trick called the "Near-Far" approach.

The Far: We look at the distant, ancient galaxies we can see.
The Near: We look at our own cosmic neighborhood, the Milky Way and its neighbors (the Local Group), right here and now.

The idea is simple: Our neighborhood is a time capsule. The Milky Way didn't just appear out of nowhere; it grew by swallowing hundreds of smaller galaxies over billions of years. If we can identify the "fossils" of those ancient, swallowed galaxies, we can reconstruct what the universe looked like back in the day.

The Missing Puzzle Pieces

For a long time, scientists only looked for intact fossils. They looked for small, whole galaxies that survived the journey and are still orbiting the Milky Way today.

But this paper argues that looking only at intact galaxies is like trying to solve a murder mystery by only interviewing the people who are still standing up. You're missing the people who were knocked down, dragged away, and scattered.

In the universe, most small galaxies didn't survive intact. They were torn apart by the gravity of bigger galaxies, stretching into long, ghostly ribbons of stars called stellar streams, or they were completely mixed into the background soup of the Milky Way, becoming phase-mixed galaxies.

The Analogy: Imagine a giant smoothie.

Intact Galaxies are the whole strawberries you can still see floating in the drink.
Stellar Streams are the long, stretched-out strings of strawberry pulp.
Phase-Mixed Galaxies are the tiny strawberry seeds and juice that have blended so thoroughly you can't see them individually anymore.

Previous studies only counted the whole strawberries. This paper says, "Wait a minute! The pulp and the seeds contain just as much (or even more) strawberry flavor!"

What the Paper Found

The authors used super-computer simulations (like a video game of the universe) to track every single star particle from the "teenage" era to today. They found three major things:

The "Scattered" Stars are the Majority: About 50% of the stars in our cosmic neighborhood today came from galaxies that were completely torn apart. If you ignore these scattered stars, you are ignoring half the story.
Better Accuracy: When they included the "scattered" stars (streams and mixed galaxies) in their calculations, their reconstruction of the ancient universe became much more accurate. It fixed the "volume" of the data (normalization) and made the results consistent across different simulations. It's like adding the missing half of a puzzle; suddenly, the picture makes perfect sense.
The "Fossil Record" Bias: There's a common method where scientists assume one ancient galaxy became one modern galaxy. The paper shows this is often wrong. One ancient galaxy might have been torn apart, with its stars ending up in three different modern streams. If you don't account for this, you get the wrong math about how many tiny galaxies existed back then.

The Good News: We Can See Them!

You might ask, "If these galaxies are torn apart and mixed up, how can we possibly find them to study them?"

The paper is optimistic. They estimate that upcoming telescopes, specifically the Vera C. Rubin Observatory (which will scan the whole sky) and the Roman Space Telescope, will be able to see these faint, ghostly streams and mixed-up galaxies.

Analogy: It's like having a new camera that is so sensitive it can see the faint glow of a firefly in a dark forest, even if the firefly is flying in a long, winding path.

Why Does This Matter?

If we ignore the torn-apart galaxies, we might think that only the big, bright galaxies were responsible for clearing the cosmic fog (reionization). But this paper suggests that the tiny, faint, and scattered galaxies were actually the heavy lifters. They provided a massive amount of the ultraviolet light needed to clear the universe's fog.

In a nutshell:
To understand how the universe grew up, we can't just look at the survivors. We have to look at the scars, the broken pieces, and the scattered dust. By including these "Forsaken Relics," we get a much clearer, more accurate picture of our cosmic origins.

1. Problem Statement

The Epoch of Reionization (EoR, $z \gtrsim 6$ ) is driven by ionizing photons from early galaxies. Direct observations (e.g., by JWST) are limited to relatively bright galaxies ( $M_{UV} \gtrsim -15$ ), leaving the faint, low-mass end of the Ultraviolet Luminosity Function (UVLF) and Stellar Mass Function (SMF) largely unconstrained.

The "near-far" technique attempts to infer these high-redshift functions by reconstructing the star formation histories (SFHs) of present-day Local Group (LG) and Milky Way (MW) galaxies. However, previous applications of this technique suffered from two major limitations:

Incomplete Progenitor Sampling: They considered only intact low-mass galaxies (satellites and field galaxies) surviving to $z=0$ , neglecting the fact that $\sim 50\%$ of the proto-MW/LG stellar mass at $z=6-9$ originated from galaxies that were subsequently disrupted by mergers and tidal forces.
Fossil Record Bias: The standard assumption that a present-day galaxy descends from a single high-redshift progenitor ignores hierarchical assembly, potentially biasing the inferred slopes of the SMF/UVLF at the low-mass/faint end.

This paper investigates whether including disrupted galaxies (stellar streams and phase-mixed objects) in the near-far analysis improves the accuracy and precision of reconstructing the high-redshift SMF/UVLF and the cosmic UV luminosity density.

2. Methodology

The authors utilized a suite of high-resolution cosmological zoom-in hydrodynamic simulations from the FIRE-2 (Feedback in Realistic Environments) project.

Simulations:
- Core Suite: 13 MW/LG-like halos (7 isolated MW-like from the Latte suite, 3 MW/M31 pairs from the ELVIS suite). These track the assembly of present-day substructures.
- Benchmark: The FIREboxHR simulation (a high-resolution cosmological volume) serves as the ground truth for the universal SMF/UVLF at $z \sim 6-14$ .
Substructure Classification (at $z=0$ ):
Using a modified version of the Panithanpaisal et al. (2021) algorithm, present-day substructures were classified into four categories:
1. Intact Satellites: Low-mass galaxies within the virial radius.
2. Stellar Streams: Disrupted galaxies with coherent tidal tails.
3. Phase-mixed Galaxies: Disrupted galaxies fully merged into the host's equilibrium phase-space distribution.
4. Field Galaxies: Intact galaxies outside the virial radius.
Progenitor Tracking:
Star particles in present-day substructures were tracked backward to $z \ge 6$ $z \geq 6$ . The study analyzed two reconstruction scenarios:
- All-Progenitors Scenario: Counting every distinct high-redshift progenitor galaxy separately.
- Fossil-Record Scenario: Combining all stellar mass from multiple progenitors of a single present-day substructure into a single "effective" progenitor (mimicking observational limitations where merger trees are unknown).
Statistical Analysis:
- Reconstructed SMFs and UVLFs were compared against the proto-MW/LG and the universal FIREboxHR functions.
- Bayesian Modeling: A Schechter function was fitted using PyMC to estimate the low-mass/faint-end slope ( $\alpha$ ) without binning biases.
- Halo-to-Halo Scatter: Quantified via bootstrapping to assess the robustness of slope estimates across different accretion histories.
Detectability Estimates:
Surface brightness maps were generated for LSST (internal view of MW) and Roman (external view of nearby galaxies) to estimate the fraction of streams and disrupted systems detectable in future surveys.

3. Key Contributions

Quantification of Disrupted Mass: Demonstrated that progenitors of disrupted galaxies (streams and phase-mixed objects) contribute up to $\sim 50\%$ of the total stellar mass budget of the proto-MW/LG at $z=6-9$ , significantly exceeding the contribution from intact satellites.
Improved Reconstruction: Showed that including disrupted galaxies in the near-far technique recovers $\sim 75\%$ of the normalization of the proto-MW/LG SMF/UVLF (Sample III: Intact + Streams + Phase-mixed), compared to only $\sim 25\%$ when using only intact galaxies (Sample I).
Reduction of Scatter: Proved that including disrupted systems reduces the halo-to-halo scatter in the inferred low-mass slope by $\sim 20-40\%$ , providing more robust constraints on the universal SMF/UVLF shape.
Bias Identification: Highlighted that the "fossil record" assumption (single progenitor) introduces a mass-dependent bias, flattening the inferred slope at the low-mass end, particularly when massive disrupted systems are included.
Observational Feasibility: Estimated that upcoming surveys (LSST and Roman) can detect a significant fraction of these disrupted systems, making the near-far technique with disrupted galaxies a viable observational strategy.

4. Key Results

Stellar Mass Budget:
- In isolated MW-like simulations, phase-mixed galaxies are the dominant contributor to the EoR stellar mass, followed by streams.
- In paired LG-like simulations, field galaxies contribute the most, followed by phase-mixed objects, streams, and satellites.
- Neglecting disrupted galaxies leads to a severe underestimation of the ionizing photon budget.
SMF/UVLF Reconstruction:
- Normalization: Adding streams (Sample II) improves normalization by a factor of $\sim 2$ ; adding phase-mixed galaxies (Sample III) improves it by a factor of $\sim 3$ relative to intact-only samples.
- Slope ( $\alpha$ ):
  - Sample III (Intact + Streams + Phase-mixed) reproduces the proto-MW/LG slope most accurately (offset $\approx 0.04$ ) with the smallest scatter ( $\tilde{\sigma}_\alpha \approx 0.15$ ).
  - Sample I (Intact only) has a large scatter ( $\tilde{\sigma}_\alpha \approx 0.24$ ) and a steeper slope bias.
- Fossil Record vs. All-Progenitors: The fossil-record approach yields flatter slopes than the all-progenitors approach because it merges multiple low-mass progenitors into single massive systems. This bias is sensitive to the selection criteria (mass cutoffs) of both present-day and high-redshift galaxies.
Cosmic UV Luminosity Density:
- Intact-only reconstructions under-predict the contribution of bright galaxies ( $M_{UV} \lesssim -15$ ).
- Including disrupted galaxies reveals that UV-bright progenitors ( $M_{UV} \lesssim -15$ ) contribute $\sim 65-70\%$ of the total cosmic UV density at $z=7$ , a figure significantly higher than inferred from intact satellites alone.
Slope Discrepancy (SMF vs. UVLF):
- The faint-end slope of the UVLF ( $\alpha \approx -1.4$ ) is shallower than the low-mass slope of the SMF ( $\alpha \approx -1.7$ ).
- This is attributed to a superlinear dependence of UV luminosity on stellar mass ( $L_{UV} \propto M_\star^{1.22}$ ), as low-mass galaxies are less efficient at forming the massive O/B stars that dominate UV output.
Detectability:
- LSST (MW): $\sim 29\%$ of streams are detectable with visible tidal tails; an additional $\sim 29\%$ are detectable as intact low-mass galaxies (total $\sim 58\%$ ).
- Roman (External): $\sim 12\%$ detectable as streams; $\sim 39\%$ detectable as intact galaxies.
- This confirms that a significant population of "forsaken relics" will be observable, enabling the application of the near-far technique to disrupted systems.

5. Significance

This paper fundamentally shifts the paradigm of the "near-far" technique by demonstrating that disrupted galaxies are not noise but essential data.

Cosmological Impact: It resolves the discrepancy between the low normalization of fossil-record reconstructions and the universal SMF/UVLF, suggesting that the faint-end slope is steeper and the normalization higher than previously thought when only intact galaxies are considered.
Reionization Physics: It implies that the ionizing photon budget during reionization was significantly bolstered by galaxies that no longer exist as distinct entities, altering our understanding of which galaxy populations drove reionization.
Future Observations: It provides a roadmap for LSST and Roman to utilize stellar streams and phase-mixed objects as direct tracers of the high-redshift universe, moving beyond the limitations of direct high- $z$ imaging.
Methodological Rigor: It warns against the uncritical use of the "single progenitor" fossil record assumption, advocating for the integration of merger tree data (from simulations) with observational SFHs to correct for hierarchical assembly biases.