Testing Light Unaffiliated Mass Clumps in MACS 0416 on galaxy and galaxy cluster scales using JWST

TL;DR

The paper tackles whether dark mass clumps without stellar counterparts (LUMCs) are necessary to explain strong-lensing in MACS 0416 at both cluster and galaxy scales. It combines Lenstool’s parametric modeling with Grale’s non-parametric reconstructions, comparing pre-JWST and JWST datasets, including 303 spectroscopically confirmed images. The cluster-scale analysis shows a DM distribution that can be traced by light, removing the need for cluster-scale LUMCs, while galaxy-scale analysis yields divergent results: Grale preserves a M2-like substructure; Lenstool disfavors but does not rule out M2, with only minor local RMS gains upon inclusion. Across methods, the surface mass density near M2 agrees to about 5–6% within a few arcseconds, highlighting residual degeneracies between compact substructures and diffuse mass distributions, and supporting a loose DM–light association. All models and products are made publicly available, contributing to the understanding of DM tracing light in clusters and galaxies.

Abstract

Light unaffiliated mass clumps (LUMCs), i.e. dark matter (DM) components without any stellar counterparts, have been reported in strong lensing mass reconstructions of MACS 0416, both on galaxy and galaxy cluster scales. On galaxy cluster scale, the most recent parametric study based on 303 multiple images features a LUMC, in the south of the cluster. On galaxy scale, the most recent GRALE non-parametric study based on 237 multiple images features two LUMCs, M1 and M2. Given the implications of these findings in the context of structure formation and evolution, we test these features parametrically, using LENSTOOL. First, we show that a mass model where each large scale DM component introduced in the modelling is associated with a stellar counterpart can reproduce the 303 multiple images, removing the need for any cluster scale LUMC in \lens. We then update the GRALE model using the 303 multiple images, finding that one of the two galaxy scale LUMC, M1, is no longer significant, while M2 remains. We test M2 by explicitely including it in our parametric model. We find that the inclusion of this LUMC does not improve the global RMS, but mildly improves locally the RMS for one multiple image located close to M2. Besides, the preferred mass for M2 corresponds to the lowest mass allowed by the adopted prior. M2 is therefore not strongly favoured by a parametric approach but it is not ruled out. We present a comparison of parametric and non-parametric models in the M2 area. Both approaches show very similar surface mass density at this location, with a 5-6 % difference between the mass maps. The difference is that GRALE favors a distinct mass substructure when LENSTOOL favors a more diffuse mass distribution. We have been able to propose a parametric mass model without including any LUMCs, providing further evidences for DM being associated with light in clusters.

Figures (6)

• Figure 1: Core of MACS 0416 from JWST data. North is up, East is left. We show the location of the cluster scale DM clumps inferred in different studies by ellipses whose semi-axis are equal to the 3$\sigma$ error bars on the position of these clumps: in yellow for Bergamini_2021, in cyan for Bergamini_2023 and in red for Gregor_2025. Green boxes represent the prior on the position of each cluster scale DM clumps proposed by Limousin_2022 and investigated in this work. We show in magenta the location of the two galaxy scale light unaffiliated mass clumps inferred by Perera_2025a. In white, the 303 secure multiple images reported by Gregor_2025.
• Figure 2: Core of MACS 0416 from JWST data. We show the total convergence maps for all models: B21 in yellow; B23 in cyan; R25 in red; our updated Grale model (PCANUCS) in magenta and our updated Lenstool model (L25) in green. We also show the convergence map from Diego_2024 in white (D24). We show the location of M2 in blue.
• Figure 3: Top: Density contours in the $10^" \times 10^"$ area centered on the peak of the M2 substructure recovered at RA = 64.03071, Dec = -24.07974 in PCANUCS. Spacing between contours is $\Delta \log_{10}\kappa = 0.05$. The three primary lens models discussed in this article using the 303 images reported by R25 are shown in solid purple (PCANUCS), dashed green (L25), and dotted red (R25). The solid yellow circle denotes M2 with radius of 2.5" corresponding to its characteristic radius. Bottom: The Median Percent Difference (MPD) between models as a function of the length of the grid comparison window centered about M2. Comparisons between PCANUCS and R25 (solid red), L25 and R25 (dash-dotted green), and PCANUCS and L25 (dashed blue) are shown. To compare these with the earlier works, the comparison using 237 images between P25 and B23 (dotted black) is also shown. For the comparisons between Grale and Lenstool models, the solid horizontal dashes on the right indicate the MPD over the whole cluster for PCANUCS and R25 (red), PCANUCS and L25 (blue), L25 and R25 (green), and P25 and B23 (black). We note that the apparent misalignment of the contours near the galaxy positions is a likely side effect of interpolation using models of varying resolution. These shifts are restricted to the pixel level, and have a negligible effect on our results.
• Figure 4: Corner plots obtained for the parameters of the mass model, for the values of rate & Nb indicated on the legend. Top: NE clump; middle: Main clump; bottom: South clump. The position of each clump is not shown for clarity since it is constrained to be coincident with the associated light component.
• Figure 5: Corner plots obtained for the parameters of the mass model when considering 182 multiple images (green) and 303 multiple images (black/grey). A rate of 0.01 is used for all realisations.