CHILES XII: The H I evolution of Luminous Compact Blue Galaxies between 0<z<0.48

D. J. Pisano; E. Momjian; H. Arlow; H. B. Gim; J. Blue Bird; J. Donovan Meyer; L. R. Hunt; M. A. Bershady; N. Luber

arxiv: 2604.13679 · v2 · submitted 2026-04-15 · 🌌 astro-ph.GA

CHILES XII: The H I evolution of Luminous Compact Blue Galaxies between 0<z<0.48

H. Arlow , D. J. Pisano , M. A. Bershady , L. R. Hunt , N. Luber , J. Donovan Meyer , E. Momjian , J. Blue Bird

show 1 more author

H. B. Gim

This is my paper

Pith reviewed 2026-05-10 12:54 UTC · model grok-4.3

classification 🌌 astro-ph.GA

keywords Luminous Compact Blue Galaxiesneutral hydrogenHI gas evolutionCHILES surveyCOSMOS fieldgas depletion timescalesstar-forming galaxies

0 comments

The pith

Luminous Compact Blue Galaxies show little change in average neutral hydrogen mass from redshift 0 to 0.48 while depleting their gas much faster than typical galaxies.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper uses stacking of radio observations from the CHILES survey to track the average amount of neutral hydrogen in Luminous Compact Blue Galaxies across a range of redshifts up to 0.48. It finds no strong evolution in this average HI mass and notes that these galaxies keep gas fractions similar to other star-forming systems. At the same time, their gas is used up nearly ten times quicker, which corresponds to the era when their numbers in the universe start to fall off sharply. This combination suggests these compact blue galaxies are efficient at turning gas into stars but do not run out of fuel suddenly.

Core claim

Using cubelet stacking on CHILES HI data in the COSMOS field, the average HI mass for LCBGs is found to be an upper limit of 4.89×10^9 solar masses at z=0.26, 2.49±0.75×10^9 at z=0.35, and 6.44±2.71×10^9 at z=0.45. There is no strong evidence for evolution in average HI mass. LCBGs retain substantial gas reservoirs with constant gas fractions consistent with the larger star-forming population, but their gas depletion timescales are nearly an order of magnitude shorter than in normal star-forming galaxies, aligning with the drop in their number density.

What carries the argument

Cubelet stacking of HI emission line data from the CHILES survey to measure average HI masses in redshift-binned samples of LCBGs.

If this is right

LCBGs maintain roughly constant average HI masses across 0 < z < 0.48.
Gas fractions in LCBGs stay steady and match those of the broader star-forming galaxy population.
Gas depletion timescales for LCBGs are about ten times shorter than for typical star-forming galaxies.
The short depletion times coincide with the sharp decline in LCBG number density at these redshifts.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

If the short depletion times are driven by intense star formation, LCBGs may represent a brief evolutionary phase where galaxies rapidly build stars before settling into more stable forms.
Surveys covering larger areas than the single COSMOS field could test whether the lack of HI mass evolution holds beyond cosmic variance effects.
Direct HI detections in individual high-redshift LCBGs would help confirm if the stacked averages accurately reflect the population.

Load-bearing premise

The stacking method assumes that LCBG samples are selected without bias across redshift and that averaging non-detections does not create systematic errors from differences in galaxy properties or noise levels.

What would settle it

Measuring individual HI masses for a statistically significant number of LCBGs at z approximately 0.45 that are substantially higher or lower than the stacked value of 6.44×10^9 solar masses would challenge the no-evolution conclusion.

Figures

Figures reproduced from arXiv: 2604.13679 by D. J. Pisano, E. Momjian, H. Arlow, H. B. Gim, J. Blue Bird, J. Donovan Meyer, L. R. Hunt, M. A. Bershady, N. Luber.

**Figure 1.** Figure 1: The RMS noise per channel (in black, right axis) over the full frequency range of the CHILES cube. The histogram (in grey, left axis) indicates the distribution of LCBGs in our sample, binned according to the frequency at which we expect to detect the H i 21 cm line [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗

**Figure 2.** Figure 2: Our full catalog of 552 LCBGs (blue) identified from all galaxies in the G10 region (grey) via their color (𝐵 − 𝑉), B-band magnitude (𝑀𝐵) and surface brightness (𝑆𝐵𝑒 (𝐵)). All galaxies represented here have spectroscopic redshifts < 0.48. We see the LCBGs populating the one corner of the parameter space. In each sub-figure, the grey points in the LCBG quadrant represent those galaxies that do not meet the… view at source ↗

**Figure 3.** Figure 3: The measured RMS noise of the stacked simulated spectrum as a function of the number of galaxies included in the stack, computed from line-free channels. The red line shows a least-squares fit to the data in log–log space. For ideal stacking, the noise is expected to scale as 𝑁−1/2 , corresponding to a slope of approximately −0.5, as shown by the dashed line. Our simulated noise scales down with a slope o… view at source ↗

**Figure 4.** Figure 4: H i column density contours overlaid on optical DECaLS images. Left: The directly detected LCBG at 𝑧 = 0.045 with an estimated H i mass of (3.53 ± 0.06) × 109 𝑀⊙. The spectrum of this detection had 𝑊20 = (317 ± 26) km/s. Right: The directly detected LCBG at 𝑧 = 0.072 with an estimated H i mass of (7.55 ± 0.23) × 109 𝑀⊙. The spectrum of this detection had 𝑊20 = (211 ± 26) km/s. S/N = 5.25 0.22 < z < 0.48 N … view at source ↗

**Figure 5.** Figure 5: The integrated, stacked spectrum from a stacked cubelet, smoothed to 50 km/s for the LCBG sample between 0.22 < 𝑧 < 0.48. The spectrum is taken from the aperture indicated by the cyan box. The reference spectrum is taken from the aperture indicated by the white box. The filled purple region indicates the off-line RMS noise of the stacked spectrum. The moment 0 map on the right of each panel is produced fro… view at source ↗

**Figure 6.** Figure 6: The integrated, stacked spectrum from a stacked cubelet, smoothed to 50 km/s for the LCBG sample between 0.22 < 𝑧 < 0.3 (top), 0.3 < 𝑧 < 0.4 (center) and 0.4 < 𝑧 < 0.48 (bottom). The spectrum is taken from the aperture indicated by the cyan box. The reference spectrum is taken from the aperture indicated by the white box. The filled purple region indicates the off-line RMS noise of the stacked spectrum. Th… view at source ↗

**Figure 7.** Figure 7: Spectra taken at noise peaks present in some of our moment 0 maps indicated with white crosses in [PITH_FULL_IMAGE:figures/full_fig_p007_7.png] view at source ↗

**Figure 8.** Figure 8: We show the measured RMS noise of the stacked spectra as a function of the number of galaxies included in the stack, computed from line-free channels. The red line shows a least-squares fit in log–log space. The RMS follows the expected 𝑁−1/2 scaling (the dashed line), with a fitted slopes of −0.54 ± 0.01, −0.45 ± 0.01 and −0.48 ± 0.01 for the 0.22 < 𝑧 < 0.3, 0.3 < 𝑧 < 0.4 and 0.4 < 𝑧 < 0.48 stacks respect… view at source ↗

**Figure 9.** Figure 9: Evolution of LCBG properties compared to literature samples. Top left: average H i masses as a function of redshift. Top right: star formation rates (SFRs) calculated from the COSMOS2015 catalogue (Laigle et al. 2016). Note that error bars are smaller than the symbol size where not visible. Bottom left: H i gas fractions, 𝑓HI = 𝑀HI/𝑀∗. Bottom right: gas depletion timescales, 𝜏HI = 𝑀HI/SFR. Blue markers sho… view at source ↗

read the original abstract

We study the evolution of Luminous Compact Blue Galaxies (LCBGs) by making use of H I emission line data provided by the full 856 h COSMOS H I Large Extragalactic Survey (CHILES), which spans a redshift range of $0\leq z\leq 0.48$ within the COSMOS field. We report the results on a cubelet stacking analysis, which we use to estimate the average H I mass evolution of LCBGs in the field up to $z=0.48$. For the stacks that do not show a detection, we report an upper limit estimate of the average H I mass. We also report on two directly detected LCBGs. We find the average H I mass in LCBGs at redshifts $z=0.26$, $z=0.35$ and $z=0.45$ respectively to be $\langle M_{\rm HI}\rangle<4.89\times10^9$ M$_\odot$, $\langle M_{\rm HI}\rangle=(2.49\pm0.75)\times10^9$ M$_\odot$ and $\langle M_{\rm HI}\rangle=(6.44\pm2.71)\times10^9$ M$_\odot$. We see no strong evidence for evolution in the average H I mass over this redshift range, consistent with other recent studies of the evolution of the H I in galaxies at $z<0.5$. On average, LCBGs appear to retain substantial gas reservoirs, with gas fractions staying constant and remaining broadly consistent with those of the larger star-forming population. LCBG gas depletion timescales are nearly an order of magnitude shorter than in normal star-forming galaxies across the studied redshift range, aligning with the period during which their number density drops sharply.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper adds three new stacked HI mass points for LCBGs at z=0.26-0.45 but the no-evolution story hinges on unverified assumptions about selection and single-field stacking.

read the letter

The main thing to know is that this work delivers concrete average HI mass values from CHILES stacking for Luminous Compact Blue Galaxies across three redshift bins, plus two direct detections, and concludes there is no strong evolution while gas fractions stay roughly constant and depletion times run shorter than in typical star-forming galaxies. That is the actual new content: specific numbers like <4.89e9 solar masses at z=0.26, 2.49 plus or minus 0.75 times 10^9 at z=0.35, and 6.44 plus or minus 2.71 times 10^9 at z=0.45, framed against the drop in LCBG number density. The paper does a clean job of applying cubelet stacking to the full 856-hour COSMOS dataset and reporting both detections and upper limits in a way that lets readers see the quantitative results directly. It also places the findings next to other recent HI evolution studies at z less than 0.5, which is useful for context. The methods are standard for this kind of survey work and the claims stay tied to the measurements rather than overreaching. The soft spots sit mainly in the assumptions needed to compare the stacks across redshift. LCBG selection by luminosity, compactness, and color can shift what population you are actually averaging as redshift changes, and the paper does not appear to show a detailed selection function or k-correction test to confirm the samples are equivalent. Stacking the non-detections further assumes that noise properties and undetected galaxy traits do not introduce systematic offsets, and everything comes from one COSMOS field so cosmic variance is a real possibility. The depletion timescale comparison also needs the star-formation rates to be measured consistently; if those come from the same data or have their own uncertainties, the factor-of-ten difference could move. These are not fatal but they are the places where the central claim could weaken on closer look. This is a targeted observational paper for people who track gas content in specific galaxy types at intermediate redshifts. Galaxy evolution modelers or HI survey teams would get the most out of the numbers. It is worth sending to peer review because the measurements are new and the dataset is solid, even if the interpretation will need the usual checks on selection and variance.

Referee Report

2 major / 2 minor

Summary. The paper analyzes HI emission from the CHILES survey in the COSMOS field to measure the average HI mass evolution of Luminous Compact Blue Galaxies (LCBGs) over 0 ≤ z ≤ 0.48 using cubelet stacking. It reports stacked average HI masses of <4.89×10^9 M⊙ at z=0.26 (upper limit), (2.49±0.75)×10^9 M⊙ at z=0.35, and (6.44±2.71)×10^9 M⊙ at z=0.45, along with two direct detections. The central claims are that there is no strong evidence for evolution in average HI mass, that gas fractions remain constant and consistent with the broader star-forming population, and that LCBG gas depletion timescales are nearly an order of magnitude shorter than in normal star-forming galaxies.

Significance. If the stacking results hold after addressing selection and variance issues, the work supplies direct observational constraints on HI content in LCBGs during the epoch when their number density declines sharply. It aligns with other low-z HI studies and highlights the rapid gas consumption in these galaxies while they retain substantial reservoirs, providing a useful benchmark for models of galaxy evolution and star-formation quenching.

major comments (2)

[§4] §4 (Stacking Analysis): The conclusion of no strong evolution in ⟨M_HI⟩ across the three redshift bins rests on the direct comparability of the reported stacked values. However, the text does not demonstrate that the LCBG selection criteria (luminosity, compactness, color) produce equivalent populations at z=0.26, 0.35, and 0.45 after accounting for k-corrections and possible evolutionary changes in galaxy properties; without this validation or a modeled selection function, the upper limit at z=0.26 and the factor-of-~2.5 variation between the z=0.35 and z=0.45 detections cannot be interpreted as evidence against evolution.
[§4.3] §4.3 and error budget discussion: The stacking of non-detections assumes that the average is unbiased with respect to varying noise properties, undetected galaxy characteristics, and cosmic variance within the single COSMOS field. No Monte Carlo simulations of the stacking procedure or multi-field comparison are presented to quantify possible systematic offsets, which directly affects the reliability of the z=0.26 upper limit and the z=0.45 detection used to support the constant gas-fraction and short depletion-time claims.

minor comments (2)

[Abstract] Abstract: The redshift range is stated as 0<z<0.48 in the title but 0≤z≤0.48 in the text; clarify the exact range and whether z=0 objects are included in any stack.
Notation: Use consistent symbols for average HI mass (⟨M_HI⟩) throughout and define the gas fraction and depletion timescale explicitly in the text or a table when first introduced.

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for their thoughtful and constructive review of our manuscript on the HI evolution of Luminous Compact Blue Galaxies using CHILES data. We address each major comment below in detail, indicating revisions where we agree changes are warranted to strengthen the presentation of our results.

read point-by-point responses

Referee: [§4] The conclusion of no strong evolution in ⟨M_HI⟩ across the three redshift bins rests on the direct comparability of the reported stacked values. However, the text does not demonstrate that the LCBG selection criteria (luminosity, compactness, color) produce equivalent populations at z=0.26, 0.35, and 0.45 after accounting for k-corrections and possible evolutionary changes in galaxy properties; without this validation or a modeled selection function, the upper limit at z=0.26 and the factor-of-~2.5 variation between the z=0.35 and z=0.45 detections cannot be interpreted as evidence against evolution.

Authors: We agree that careful consideration of sample equivalence is needed to interpret the stacked HI masses. The LCBG criteria (luminosity, compactness, and blue color) are applied uniformly across bins using the same rest-frame definitions from the literature, with k-corrections incorporated into the photometry as described in Section 2. We have confirmed that average stellar masses and other ancillary properties remain comparable between bins. A complete modeled selection function incorporating all possible evolutionary effects is beyond the current scope, but we will expand the discussion in the revised Section 4 to explicitly address potential selection biases and their implications for the no-evolution conclusion, emphasizing that the measurements are consistent within the reported uncertainties. revision: partial
Referee: [§4.3] The stacking of non-detections assumes that the average is unbiased with respect to varying noise properties, undetected galaxy characteristics, and cosmic variance within the single COSMOS field. No Monte Carlo simulations of the stacking procedure or multi-field comparison are presented to quantify possible systematic offsets, which directly affects the reliability of the z=0.26 upper limit and the z=0.45 detection used to support the constant gas-fraction and short depletion-time claims.

Authors: We appreciate this feedback on the robustness of our error analysis. The stacking follows standard HI literature methods, and our error budget already incorporates contributions from noise, sample variance, and cosmic variance. To further quantify possible systematic offsets from noise properties and undetected galaxy characteristics, we will include Monte Carlo simulations of the stacking procedure in the revised manuscript. However, because the CHILES data are restricted to the single COSMOS field, a multi-field comparison is not feasible with existing observations; we will add explicit discussion of this limitation and its implications for the results. revision: partial

standing simulated objections not resolved

Inability to perform a multi-field comparison to assess cosmic variance, as the survey is limited to the single COSMOS field.

Circularity Check

0 steps flagged

No circularity: direct observational stacking of CHILES HI data yields reported masses without self-referential fits or derivations

full rationale

This is a purely observational paper reporting stacked HI masses and upper limits from CHILES cubelets for LCBGs at different redshifts. The central results (<M_HI> values at z=0.26, 0.35, 0.45) are direct measurements from the data cubes, with non-detections handled by averaging and upper limits. No equations or derivations reduce these to parameters fitted from the same dataset. Claims of no strong evolution and consistency with other studies rely on external comparisons, not internal self-citation chains or ansatzes. The analysis is self-contained against the survey data without load-bearing self-references that collapse the result to its inputs.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The analysis rests on standard radio astronomy practices for converting HI line flux to mass and on cosmological distance calculations; no new free parameters, ad-hoc axioms, or invented entities are introduced in the reported results.

axioms (1)

standard math Standard flat Lambda-CDM cosmology for converting observed redshift to comoving distance and look-back time
Implicit in mapping the three redshift bins to cosmic epochs.

pith-pipeline@v0.9.0 · 5688 in / 1361 out tokens · 46642 ms · 2026-05-10T12:54:19.799890+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

1 extracted references · 1 canonical work pages

[1]

N., Bagla J

Aghanim N., et al., 2020, A&A, 641, A6 Bera A., Kanekar N., Chengalur J. N., Bagla J. S., 2023, ApJ Letters, 950, L18 Bianchetti A., et al., 2025, ApJ, 982, 82 Blue Bird J., et al., 2020, MNRAS, 492, 153 Blue Bird J., et al., 2026, ApJ, 998, 248 Blyth S.-L., et al., 2016, Proceedings of Science Cardamone C., et al., 2009, MNRAS, 399, 1191 Chen Q., Meyer M...

work page arXiv 2020

[1] [1]

N., Bagla J

Aghanim N., et al., 2020, A&A, 641, A6 Bera A., Kanekar N., Chengalur J. N., Bagla J. S., 2023, ApJ Letters, 950, L18 Bianchetti A., et al., 2025, ApJ, 982, 82 Blue Bird J., et al., 2020, MNRAS, 492, 153 Blue Bird J., et al., 2026, ApJ, 998, 248 Blyth S.-L., et al., 2016, Proceedings of Science Cardamone C., et al., 2009, MNRAS, 399, 1191 Chen Q., Meyer M...

work page arXiv 2020