pith. sign in

arxiv: 2509.13408 · v2 · pith:WQ4X2QTLnew · submitted 2025-09-16 · 🌌 astro-ph.GA · astro-ph.SR

Chemical decoding of kinematic substructures in the Galactic halo

A. Mori , P. Di Matteo , S. Salvadori , M. Mondelin , S. Khoperskov , M. Haywood , A. Mastrobuono-Battisti This is my paper

Pith reviewed 2026-05-21 22:44 UTC · model grok-4.3

classification 🌌 astro-ph.GA astro-ph.SR
keywords Galactic halostellar substructureschemical taggingMilky Way accretionGaia-Sausage-Enceladuskinematic selectionAPOGEE abundancesGaussian Mixture Model
0
0 comments X

The pith

Kinematic substructures in the Galactic halo are chemical mixtures rather than single-origin populations.

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

The paper applies kinematic cuts from Gaia data to select stars in known halo substructures and then uses chemical abundances from APOGEE to compare those stars to reference populations. A Gaussian Mixture Model on elements including iron, magnesium, silicon, calcium, manganese, aluminum and carbon shows that the substructures overlap heavily with the Gaia-Sausage-Enceladus merger, the metal-poor disc, Sagittarius and omega Centauri. A sympathetic reader cares because this means the Milky Way's halo did not assemble from a set of cleanly separated mergers; instead the dynamical labels mix stars from multiple sources, changing how we map the galaxy's accretion history.

Core claim

Halo stars chemically compatible with GSE are spread throughout the E-Lz space and considerably contaminate every halo substructure studied in this work. None of these substructures appears to be a unique population of stars with its own origin. In addition to GSE, they all appear to be mixtures of stars chemically compatible either with the metal-poor disc, Sagittarius, ω Cen, or with a combination of them.

What carries the argument

Gaussian Mixture Model applied star-by-star to seven elemental abundances to chemically tag and compare stars across kinematically selected substructures.

If this is right

  • Sequoia shares a chemical origin with GSE.
  • Heracles, Thamnos and the Helmi Stream each contain large fractions of both GSE and heated disc stars.
  • The Helmi Stream also contains stars chemically matching Sagittarius while Thamnos contains stars matching ω Cen.
  • GSE itself is contaminated by Sagittarius stars.
  • Chemically GSE-like stars appear across the full E-Lz plane and affect every substructure examined.

Where Pith is reading between the lines

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

  • Kinematic selections alone are likely to misattribute the origins of halo stars, so future work will need tighter chemistry-dynamics combinations to reconstruct accretion events.
  • Milky Way formation simulations can be tested against the observed degree of mixing in E-Lz space.
  • Larger samples with more precise abundances from upcoming surveys should allow quantitative fractions for each progenitor inside every substructure.
  • The heated metal-poor disc's contribution suggests that in-situ heating during the GSE merger was more important than some models currently assume.

Load-bearing premise

The selected abundances plus the mixture model can separate stellar origins even when measurement errors, internal scatter inside each progenitor, and possible overlaps between progenitors are present.

What would settle it

High-resolution spectra of the same stars that reveal abundance patterns cleanly separated into groups that do not match the proposed GSE, disc, Sagittarius or omega Cen templates would falsify the claimed chemical mixtures.

Figures

Figures reproduced from arXiv: 2509.13408 by A. Mastrobuono-Battisti, A. Mori, M. Haywood, M. Mondelin, P. Di Matteo, S. Khoperskov, S. Salvadori.

Figure 1
Figure 1. Figure 1: GSE (K) distribution in the [Mg/Fe] vs. [Fe/H] space, colour￾coded by distance from the Sun. possible. In particular, we analysed the accreted debris of the fol￾lowing merger events: GSE, Sequoia, Thamnos, Helmi Stream, Heracles, and Sagittarius. We included ω Cen in the analyses, as a candidate remnant of an accretion event and possibly the nu￾clear star cluster of GSE/Sequoia (Massari et al. 2019; Myeong… view at source ↗
Figure 2
Figure 2. Figure 2: Distribution of the halo substructures in the orbital energy (E) versus angular momentum with respect to the Galactic disc (Lz) space. Beyond the dynamical definitions in [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Upper panel: MDFs of the halo substructures in different colours with the same colour legend as in [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: GSE distribution in the [Mg/Fe], [Si/Fe], [Ca/Fe], [C/Fe], [Al/Fe], and [Mn/Fe] vs. [Fe/H] spaces. The blue circles highlight the bulk of the distribution, while the yellow ones are the outliers of the GSE distribution with respect to itself when a threshold of 20% is considered. The same approach has been used to quantify the chemical compatibility of the different halo substructures with the kine￾matical… view at source ↗
Figure 5
Figure 5. Figure 5: Fractions (%) of chemically compatible stars of the Milky Way halo substructures with respect to GSE (K) in blue and GSE (M) in cyan. The relative uncertainties are obtained by bootstrapping the data a thousand times, accounting for the observational uncertainties. The fractions are computed assuming a threshold equal to the 20th percentile of the probability density, thus the chemical compatibility normal… view at source ↗
Figure 6
Figure 6. Figure 6: Fractions (%) of chemically compatible stars of the Milky Way halo substructures with respect to the metal-poor disc ([Fe/H] ≤ −0.6). The relative uncertainties are obtained by bootstrapping the data a thou￾sand times, accounting for the observational uncertainties. The fractions are computed assuming a threshold equal to the 20th percentile of the probability density, thus the chemical compatibility norma… view at source ↗
Figure 7
Figure 7. Figure 7: Abundance distribution in the [Mg/Fe], [Si/Fe], [Ca/Fe], [C/Fe], [Al/Fe], and [Mn/Fe] vs. [Fe/H] spaces for the outliers of the Sequoia (K) (in green) and GSE (K) (in blue) samples with respect to GSE itself. The black contours show the 68%, 80% and 95% of the GSE probability density distribution for reference, when a threshold of 20% is considered. both samples, meaning that the a significant part of the … view at source ↗
Figure 8
Figure 8. Figure 8: Abundance distribution in the [Mg/Fe], [Si/Fe], [Ca/Fe], [C/Fe], [Al/Fe], and [Mn/Fe] vs. [Fe/H] spaces for the outliers of the Thamnos (in brown) and ω Cen (in turquoise). The black contours show the 68%, 80% and 95% of the GSE distribution for reference, when a threshold of 20% is considered. – Regarding Thamnos, despite being significantly compatible with both GSE and the disc, its outliers feature some… view at source ↗
Figure 9
Figure 9. Figure 9: Abundance distribution in the [Mg/Fe], [Si/Fe], [Ca/Fe], [C/Fe], [Al/Fe], and [Mn/Fe] vs. [Fe/H] spaces for the outliers of the Helmi Stream (in magenta) and Sagittarius Stream (in orange). The black contours show the 68%, 80% and 95% of the GSE distribution for reference, when a threshold of 20% is considered [PITH_FULL_IMAGE:figures/full_fig_p012_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Distribution in the E −Lz space of all the stars that are chemically compatible to GSE (K) (left panel) and the metal-poor disc (right panel), within each halo substructure (the number of the compatible stars are reported in the legend). For reference we show the kinematic definitions of these substructures as rectangles with the same colour legend as the points, along with the GSE (K) and (M) definitions… view at source ↗
Figure 11
Figure 11. Figure 11: For each of the halo substructures listed to the left, we show the fractions (%) of chemically compatible stars with respect to GSE (K) (in blue) and the metal-poor disc (in lilac), along with the lower limit for the outliers (in grey), computed through the GMM analysis assuming a threshold of 20. The patterned grey highlights the similar chemical distribution of the outliers of Sequoia (K) to those of GS… view at source ↗
read the original abstract

In the hierarchical assembly framework, the accretion history of the Milky Way is crucial to understand its evolution. However, in massive mergers, integrals of motion are not strictly conserved, redistributing accreted stars across dynamical spaces, such as energy-angular momentum ($E-L_z$). Additionally, the in situ disc becomes kinematically heated, acquiring halo-like orbits. Consequently, even for minor mergers, which should preserve dynamical coherence, we expect their kinematic-defined samples to be contaminated by both the massive merger(s) and the disc stars. This study aims at quantifying this contamination in known accreted halo substructures. As they are defined by kinematics, we aim at cleaning their samples analysing only chemical properties. We applied the kinematic selection criteria for the halo substructures to the Gaia EDR3 and APOGEE DR17 data. Then we adopted a Gaussian Mixture Model approach to chemically compare different substructures on a star-by-star basis, taking into account several abundances (Fe, Mg, Si, Ca, Mn, Al, and C). We argue that the chemical properties of Sequoia point towards a shared origin with GSE. Heracles, Thamnos and the Helmi Stream all likely comprise GSE and heated disc stars in a significant amount. Besides these two populations, we identified stars with chemical and orbital properties compatible with Sagittarius in the Helmi Stream and with $\omega$ Cen in Thamnos. Finally, GSE itself is contaminated by Sagittarius. Halo stars chemically compatible with GSE are spread throughout the $E-L_z$ space and considerably contaminate every halo substructure studied in this work. None of these substructures appears to be a unique population of stars with its own origin. In addition to GSE, they all appear to be mixtures of stars chemically compatible either with the metal-poor disc, Sagittarius, $\omega$ Cen, or with a combination of them.

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.

Referee Report

3 major / 3 minor

Summary. The manuscript applies kinematic selections from Gaia EDR3 to APOGEE DR17 stars to isolate known halo substructures (GSE, Sequoia, Heracles, Thamnos, Helmi Stream) and then uses a Gaussian Mixture Model on seven abundances (Fe, Mg, Si, Ca, Mn, Al, C) to perform star-by-star chemical classification. The central claim is that GSE-compatible stars are distributed across E-Lz space and contaminate every substructure, rendering none of them chemically unique; instead, the substructures are reported as mixtures of GSE, metal-poor disc, Sagittarius, and ω Cen stars.

Significance. If the GMM-based separations hold after proper validation, the result would demonstrate that dynamical mixing from massive mergers and disc heating renders purely kinematic halo substructures chemically composite. This has direct implications for interpreting the Milky Way's accretion history and for the reliability of substructure catalogs derived from integrals of motion alone. The analysis is grounded in public catalogs, which in principle supports reproducibility.

major comments (3)
  1. [GMM methods] Section describing the GMM implementation (likely §3): no details are provided on how measurement uncertainties in the seven abundances are propagated into the mixture model, how the number of components is chosen (e.g., via BIC, cross-validation, or silhouette score), or on initialization stability. Because the reported contamination fractions and mixture interpretations (Heracles/Thamnos as GSE+disc; Helmi with added Sgr) rest directly on the per-star posterior assignments, these omissions make the quantitative claims sensitive to untested modeling choices.
  2. [Results for Helmi and Thamnos] Results sections on individual substructures (e.g., Helmi Stream and Thamnos): the identification of Sagittarius- or ω Cen-compatible stars is presented without a confusion-matrix test, synthetic-population validation, or quantification of overlap probabilities between the fitted Gaussians. Given known intrinsic scatter within progenitors and APOGEE abundance errors, the evidence that these specific contaminants are required (rather than being absorbed into broader GSE or disc components) is not yet load-bearing.
  3. [Data and sample selection] Sample selection and cleaning (likely §2): the manuscript does not specify the quality cuts applied to APOGEE DR17 abundances or how stars with large uncertainties or missing values in the chosen elements are handled before GMM fitting. This directly affects the reliability of the chemical comparison across substructures.
minor comments (3)
  1. [Abstract] The abstract states that 'GSE itself is contaminated by Sagittarius' but does not quantify the fraction or show the corresponding E-Lz distribution; adding a brief numerical summary would improve clarity.
  2. [Methods] Notation for the abundance vector and the GMM covariance matrices should be defined explicitly in the methods section to avoid ambiguity when readers attempt to reproduce the fits.
  3. [Figures] Figure captions for the E-Lz and abundance projections could explicitly state the number of stars assigned to each GMM component per substructure.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their constructive and detailed comments, which highlight important areas for improving the methodological transparency and validation of our analysis. We address each major comment point by point below and will revise the manuscript accordingly where needed.

read point-by-point responses

  1. Referee: [GMM methods] Section describing the GMM implementation (likely §3): no details are provided on how measurement uncertainties in the seven abundances are propagated into the mixture model, how the number of components is chosen (e.g., via BIC, cross-validation, or silhouette score), or on initialization stability. Because the reported contamination fractions and mixture interpretations (Heracles/Thamnos as GSE+disc; Helmi with added Sgr) rest directly on the per-star posterior assignments, these omissions make the quantitative claims sensitive to untested modeling choices.

    Authors: We agree that additional details on the GMM procedure are required for full reproducibility. In the revised manuscript we will expand the methods section to describe: propagation of APOGEE abundance uncertainties via Monte Carlo resampling of the input data before fitting; selection of the number of components using the Bayesian Information Criterion with explicit reporting of BIC values for 2–8 components; and assessment of initialization stability through 50 random initializations, retaining only assignments that remain consistent across runs. These additions will directly support the robustness of the reported contamination fractions. revision: yes

  2. Referee: [Results for Helmi and Thamnos] Results sections on individual substructures (e.g., Helmi Stream and Thamnos): the identification of Sagittarius- or ω Cen-compatible stars is presented without a confusion-matrix test, synthetic-population validation, or quantification of overlap probabilities between the fitted Gaussians. Given known intrinsic scatter within progenitors and APOGEE abundance errors, the evidence that these specific contaminants are required (rather than being absorbed into broader GSE or disc components) is not yet load-bearing.

    Authors: We acknowledge that the current presentation would benefit from explicit validation of the additional components. In the revision we will add a supplementary section that generates synthetic populations drawn from literature abundance distributions for GSE, metal-poor disc, Sagittarius, and ω Cen, injects realistic APOGEE uncertainties, and quantifies the overlap probabilities and recovery rates of the fitted Gaussians. This will demonstrate whether the Sagittarius and ω Cen components are statistically required or can be absorbed into broader mixtures. revision: yes

  3. Referee: [Data and sample selection] Sample selection and cleaning (likely §2): the manuscript does not specify the quality cuts applied to APOGEE DR17 abundances or how stars with large uncertainties or missing values in the chosen elements are handled before GMM fitting. This directly affects the reliability of the chemical comparison across substructures.

    Authors: We thank the referee for noting this omission. The revised Section 2 will explicitly list the applied quality criteria, including ASPCAPFLAG and STARFLAG cuts, a minimum S/N threshold of 100, and the exclusion of stars with abundance uncertainties exceeding 0.2 dex in any of the seven elements or with missing values in more than one element. This will clarify the sample definition and allow direct assessment of its impact on the chemical tagging results. revision: yes

Circularity Check

0 steps flagged

No circularity: data-driven GMM classification from public catalogs

full rationale

The paper applies standard kinematic cuts to Gaia EDR3 and APOGEE DR17, then runs a Gaussian Mixture Model on the seven abundances (Fe, Mg, Si, Ca, Mn, Al, C) to assign chemical membership on a star-by-star basis. All reported contamination fractions and mixture interpretations follow directly from these empirical assignments and orbital comparisons; no equation or parameter is defined in terms of the final claim, no self-citation supplies a uniqueness theorem, and no fitted input is relabeled as an independent prediction. The derivation therefore remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The work rests on the domain assumption that chemical abundances serve as reliable origin tracers and on the statistical modeling choice of GMM component number; no new physical entities are postulated.

free parameters (1)
  • Number of GMM components
    Chosen to describe the observed chemical distributions across the compared substructures.
axioms (1)
  • domain assumption Stars accreted from the same progenitor share sufficiently distinct multi-element abundance patterns to be separable by GMM despite observational errors.
    Invoked when the GMM is used to chemically compare substructures on a star-by-star basis.

pith-pipeline@v0.9.0 · 5902 in / 1517 out tokens · 55464 ms · 2026-05-21T22:44:45.488409+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Probing the Milky Way Halo with RR Lyrae Stars from Gaia Data Release 3

    astro-ph.GA 2026-02 unverdicted novelty 4.0

    RR Lyrae stars yield metallicity measurements for Milky Way halo substructures including Gaia Sausage/Enceladus at [Fe/H] = -1.57 dex, with some features like ED-1 showing disk contamination and others like Shiva and ...

  2. The Nephele ecosystem: stars, globular clusters, and stellar streams associated with the progenitor galaxy of $\omega$ Centauri

    astro-ph.GA 2025-10 unverdicted novelty 4.0

    Using 8D chemical clustering on APOGEE DR17 and kinematic matching to e-TidalGCs simulations, the authors report 470 ω Cen-like stars including 6 kinematically consistent with its stream and additional links to stream...

Reference graph

Works this paper leans on

58 extracted references · 58 canonical work pages · cited by 2 Pith papers

  1. [1]

    2022, ApJS, 259, 35

    Abdurro’uf, Accetta, K., Aerts, C., et al. 2022, ApJS, 259, 35

    work page 2022

  2. [2]

    Amarante, J. A. S., Debattista, V . P., Beraldo e Silva, L., Laporte, C. F. P., & Deg, N. 2022, ApJ, 937, 12

    work page 2022

  3. [3]

    & Vasiliev, E

    Baumgardt, H. & Vasiliev, E. 2021, MNRAS, 505, 5957

    work page 2021

  4. [4]

    W., Koposov, S

    Belokurov, V ., Erkal, D., Evans, N. W., Koposov, S. E., & Deason, A. J. 2018, MNRAS, 478, 611

    work page 2018

  5. [5]

    L., Fattahi, A., et al

    Belokurov, V ., Sanders, J. L., Fattahi, A., et al. 2020, MNRAS, 494, 3880

    work page 2020

  6. [6]

    & Gerhard, O

    Bland-Hawthorn, J. & Gerhard, O. 2016, ARA&A, 54, 529

    work page 2016

  7. [7]

    Brown, A. G. A., Velázquez, H. M., & Aguilar, L. A. 2005, MNRAS, 359, 1287

    work page 2005

  8. [8]

    K., et al

    Buder, S., Lind, K., Ness, M. K., et al. 2022, MNRAS, 510, 2407

    work page 2022

  9. [9]

    M., Cautun, M., Deason, A

    Callingham, T. M., Cautun, M., Deason, A. J., et al. 2022, MNRAS, 513, 4107

    work page 2022

  10. [10]

    Deason, A. J. & Belokurov, V . 2024, New A Rev., 99, 101706 Di Matteo, P., Haywood, M., Lehnert, M. D., et al. 2019, A&A, 632, A4

    work page 2024

  11. [11]

    K., Feltzing, S., Sahlholdt, C

    Feuillet, D. K., Feltzing, S., Sahlholdt, C. L., & Casagrande, L. 2020, MNRAS, 497, 109

    work page 2020

  12. [12]

    K., Sahlholdt, C

    Feuillet, D. K., Sahlholdt, C. L., Feltzing, S., & Casagrande, L. 2021, MNRAS, 508, 1489

    work page 2021

  13. [13]

    Forbes, D. A. 2020, MNRAS, 493, 847

    work page 2020

  14. [14]

    & Bland-Hawthorn, J

    Freeman, K. & Bland-Hawthorn, J. 2002, ARA&A, 40, 487 Gaia Collaboration, Brown, A. G. A., Vallenari, A., et al. 2021, A&A, 650, C3

    work page 2002

  15. [15]

    J., Brook, C

    Gallart, C., Bernard, E. J., Brook, C. B., et al. 2019, Nature Astronomy, 3, 932 Gómez, F. A., Helmi, A., Brown, A. G. A., & Li, Y .-S. 2010, MNRAS, 408, 935

    work page 2019

  16. [16]

    Grand, R. J. J., Deason, A. J., White, S. D. M., et al. 2019, MNRAS, 487, L72

    work page 2019

  17. [17]

    R., Lian, J., et al

    Hasselquist, S., Hayes, C. R., Lian, J., et al. 2021, ApJ, 923, 172

    work page 2021

  18. [18]

    2015, MNRAS, 453, 758

    Hawkins, K., Jofré, P., Masseron, T., & Gilmore, G. 2015, MNRAS, 453, 758

    work page 2015

  19. [19]

    D., et al

    Haywood, M., Di Matteo, P., Lehnert, M. D., et al. 2018, ApJ, 863, 113

    work page 2018

  20. [20]

    H., et al

    Helmi, A., Babusiaux, C., Koppelman, H. H., et al. 2018, Nature, 563, 85

    work page 2018

  21. [21]

    & de Zeeuw, P

    Helmi, A. & de Zeeuw, P. T. 2000, MNRAS, 319, 657

    work page 2000

  22. [22]

    & White, S

    Helmi, A. & White, S. D. M. 1999, MNRAS, 307, 495

    work page 1999

  23. [23]

    Helmi, A., White, S. D. M., de Zeeuw, P. T., & Zhao, H. 1999, Nature, 402, 53

    work page 1999

  24. [24]

    P., Mackereth, J

    Horta, D., Schiavon, R. P., Mackereth, J. T., et al. 2021, MNRAS, 500, 1385

    work page 2021

  25. [25]

    P., Mackereth, J

    Horta, D., Schiavon, R. P., Mackereth, J. T., et al. 2023, MNRAS, 520, 5671

    work page 2023

  26. [26]

    A., Gilmore, G., & Irwin, M

    Ibata, R. A., Gilmore, G., & Irwin, M. J. 1994, Nature, 370, 194

    work page 1994

  27. [27]

    2017, A&A, 604, A106

    Jean-Baptiste, I., Di Matteo, P., Haywood, M., et al. 2017, A&A, 604, A106

    work page 2017

  28. [28]

    2023d, arXiv e-prints, arXiv:2310.05287

    Khoperskov, S., Minchev, I., Steinmetz, M., et al. 2023d, arXiv e-prints, arXiv:2310.05287

    work page arXiv

  29. [29]

    Knebe, A., Gill, S. P. D., Kawata, D., & Gibson, B. K. 2005, MNRAS, 357, L35

    work page 2005

  30. [30]

    H., Bos, R

    Koppelman, H. H., Bos, R. O. Y ., & Helmi, A. 2020, A&A, 642, L18

    work page 2020

  31. [31]

    Kruijssen, J. M. D., Pfeffer, J. L., Chevance, M., et al. 2020, MNRAS, 498, 2472

    work page 2020

  32. [32]

    Kruijssen, J. M. D., Pfeffer, J. L., Reina-Campos, M., Crain, R. A., & Bastian, N. 2019, MNRAS, 486, 3180

    work page 2019

  33. [33]

    Laporte, C. F. P., Johnston, K. V ., & Tzanidakis, A. 2019, MNRAS, 483, 1427

    work page 2019

  34. [34]

    R., Majewski, S

    Law, D. R., Majewski, S. R., & Johnston, K. V . 2009, ApJ, 703, L67

    work page 2009

  35. [35]

    Leung, H. W. & Bovy, J. 2019, MNRAS, 489, 2079

    work page 2019

  36. [36]

    O., Pérez-Villegas, A., et al

    Limberg, G., Souza, S. O., Pérez-Villegas, A., et al. 2022, ApJ, 935, 109

    work page 2022

  37. [37]

    Mackereth, J. T. & Bovy, J. 2018, PASP, 130, 114501

    work page 2018

  38. [38]

    T., Schiavon, R

    Mackereth, J. T., Schiavon, R. P., Pfeffer, J., et al. 2019, MNRAS, 482, 3426

    work page 2019

  39. [39]

    R., Skrutskie, M

    Majewski, S. R., Skrutskie, M. F., Weinberg, M. D., & Ostheimer, J. C. 2003, ApJ, 599, 1082

    work page 2003

  40. [40]

    H., & Helmi, A

    Massari, D., Koppelman, H. H., & Helmi, A. 2019, A&A, 630, L4

    work page 2019

  41. [41]

    McMillan, P. J. 2017, MNRAS, 465, 76

    work page 2017

  42. [42]

    McMillan, P. J. & Binney, J. J. 2008, MNRAS, 390, 429

    work page 2008

  43. [43]

    A., Lane, J

    Monty, S., Venn, K. A., Lane, J. M. M., Lokhorst, D., & Yong, D. 2020, MNRAS, 497, 1236

    work page 2020

  44. [44]

    2024, A&A, 690, A136

    Mori, A., Di Matteo, P., Salvadori, S., et al. 2024, A&A, 690, A136

    work page 2024

  45. [45]

    C., Evans, N

    Myeong, G. C., Evans, N. W., Belokurov, V ., Sanders, J. L., & Koposov, S. E. 2018, ApJ, 856, L26

    work page 2018

  46. [46]

    C., Vasiliev, E., Iorio, G., Evans, N

    Myeong, G. C., Vasiliev, E., Iorio, G., Evans, N. W., & Belokurov, V . 2019, MNRAS, 488, 1235

    work page 2019

  47. [47]

    P., Conroy, C., Bonaca, A., et al

    Naidu, R. P., Conroy, C., Bonaca, A., et al. 2020, ApJ, 901, 48

    work page 2020

  48. [48]

    P., Ji, A

    Naidu, R. P., Ji, A. P., Conroy, C., et al. 2022, ApJ, 926, L36

    work page 2022

  49. [49]

    2025, ApJ, 982, L14

    Nandakumar, G., Ryde, N., Schultheis, M., et al. 2025, ApJ, 982, L14

    work page 2025

  50. [50]

    Nissen, P. E. & Schuster, W. J. 2010, A&A, 511, L10

    work page 2010

  51. [51]

    2025, A&A, 693, A155

    Pagnini, G., Di Matteo, P., Haywood, M., et al. 2025, A&A, 693, A155

    work page 2025

  52. [52]

    S., et al

    Ruiz-Lara, T., Matsuno, T., Lövdal, S. S., et al. 2022, A&A, 665, A58

    work page 2022

  53. [53]

    2025, ApJ, 979, 174

    Ryde, N., Nandakumar, G., Schultheis, M., et al. 2025, ApJ, 979, 174

    work page 2025

  54. [54]

    P., Phillips, S

    Schiavon, R. P., Phillips, S. G., Myers, N., et al. 2024, MNRAS, 528, 1393 Skúladóttir, Á., Ernandes, H., Feuillet, D. K., et al. 2025, ApJ, 986, L21

    work page 2024

  55. [55]

    F., Battaglia, G., Grand, R

    Thomas, G. F., Battaglia, G., Grand, R. J. J., & Aguiar Álvarez, A. 2025, arXiv e-prints, arXiv:2504.10398

    work page arXiv 2025

  56. [56]

    & Baumgardt, H

    Vasiliev, E. & Baumgardt, H. 2021, MNRAS, 505, 5978

    work page 2021

  57. [57]

    A., Irwin, M., Shetrone, M

    Venn, K. A., Irwin, M., Shetrone, M. D., et al. 2004, AJ, 128, 1177

    work page 2004

  58. [58]

    2019, MNRAS, 487, L47 Article number, page 15 of 23 A&A proofs:manuscript no

    Vincenzo, F., Spitoni, E., Calura, F., et al. 2019, MNRAS, 487, L47 Article number, page 15 of 23 A&A proofs:manuscript no. aa Appendix A: Chemical comparison of the halo substructures to GSE (K) and the metal-poor disc In this appendix, we show the detailed chemical comparison of each of the halo substructures with respect to GSE (K) and the metal-poor d...

    work page 2019