Sequoia: Retrograde Stellar Halo Debris

• Sequoia is a retrograde stellar-halo substructure likely originating from an accreted dwarf galaxy, characterized by high orbital energy and distinct retrograde motion.
• Studies using various dynamical selection methods reveal that Sequoia’s membership overlaps with other retrograde components, complicating its clear separation from structures like Gaia-Enceladus and Thamnos.
• Chemical analyses show Sequoia stars are predominantly metal-poor with low-α and unique r-process signatures, indicating a less efficient star formation history and early enrichment compared to other halo debris.

Sequoia is a retrograde stellar-halo substructure of the Milky Way, generally interpreted as debris from an accreted dwarf galaxy or merger event. In the literature it is consistently associated with strongly retrograde orbital angular momentum and comparatively high orbital energy, but its exact dynamical boundaries, chemical homogeneity, and degree of separation from Gaia-Enceladus and Thamnos remain debated. Much of the modern discussion therefore treats Sequoia not only as a named merger remnant, but also as a test case for how Galactic-archaeology substructures are defined in overlapping action–energy space (Matsuno et al., 2021, Koppelman et al., 2019).

1. Naming, emergence, and conceptual status

The name “Sequoia” entered Galactic-archaeology usage through the rediscovery of the globular cluster FSR 1758, whose unusual appearance inspired the phrase “Sequoia in the garden.” Myeong et al. then adopted “Sequoia” for a putative progenitor dwarf galaxy or accretion event associated with retrograde halo debris, and the label persisted as subsequent work established that a distinct retrograde accretion event did exist (Romero-Colmenares et al., 2021).

From the outset, however, Sequoia was conceptually unstable in two ways. First, the name referred both to a dynamical population in the halo and, historically, to an object later shown probably not to belong to that population. Second, the “Sequoia region” of phase space overlaps with other retrograde debris, so the literature does not converge on a unique membership definition. The irony emphasized in later work is that FSR 1758, the cluster that inspired the name, is favored dynamically to be associated with Gaia-Enceladus-Sausage rather than Sequoia itself (Romero-Colmenares et al., 2021).

This history matters because Sequoia is not a purely nominal category. Its physical interpretation depends on whether the relevant retrograde stars are treated as a single progenitor, as a high-energy subset of a broader retrograde halo, or as one component within a more fragmented merger hierarchy.

2. Dynamical definition and orbital realization

Operational definitions of Sequoia vary substantially across studies. Some works define it in the $L_z$–$E$ plane, others in normalized-action space, others by clustering in action or energy–action space. The resulting samples are therefore related but not identical.

Study	Operational selection	Resulting picture
(Matsuno et al., 2021)	$L_z < -1600\ {\rm kpc\,km\,s^{-1}}$, $E > -1.3\times 10^5\ {\rm km^2\,s^{-2}}$	12 high-precision candidates
(Feuillet et al., 2021)	$J_{\phi}/J_{\mathrm{tot}} < -0.6$, $(J_z - J_R)/J_{\mathrm{tot}} < 0.1$	preferred low-contamination “Sequoia 06”
(Zhang et al., 2024)	$\eta < -0.15$, $J_\phi < -0.7\times 10^3\ {\rm kpc\,km\,s^{-1}}$, $E > -1.5\times 10^5\ {\rm km^2\,s^{-2}}$	high-energy retrograde population with 136 LR and 7 HR stars
(Deepak, 2024)	$$-1.5\times 10^5 < E < -0.9\times 10^5\ {\rm km^2\,s^{-2}}$$, $E$0	306 stars in the selection text, 421 in Table 1
(Koppelman et al., 2019)	$E$1, $E$2	Sequoia confined to high energy

Within a very metal-poor halo sample clustered directly in $E$3 with HDBSCAN, Sequoia was recovered as DTG-1 and DTG-5, totaling 19 stars with average confidence about 75%. In that realization, its mean properties were $E$4, $E$5, $E$6, $E$7, and $E$8, with median $E$9 (Limberg et al., 2020). This realization is particularly influential because it shows Sequoia emerging as a robust retrograde structure without being imposed a priori.

A larger very metal-poor LAMOST/Subaru analysis, based on friends-of-friends grouping in $L_z < -1600\ {\rm kpc\,km\,s^{-1}}$0, treated Sequoia as a high-energy retrograde population and reported median dynamical properties $L_z < -1600\ {\rm kpc\,km\,s^{-1}}$1, $L_z < -1600\ {\rm kpc\,km\,s^{-1}}$2, $L_z < -1600\ {\rm kpc\,km\,s^{-1}}$3, and $L_z < -1600\ {\rm kpc\,km\,s^{-1}}$4 (Zhang et al., 2024). A separate Gaia–LAMOST study, using an $L_z < -1600\ {\rm kpc\,km\,s^{-1}}$5–$L_z < -1600\ {\rm kpc\,km\,s^{-1}}$6 box, characterized Sequoia as one of the most highly retrograde halo components, with $L_z < -1600\ {\rm kpc\,km\,s^{-1}}$7, mean eccentricity $L_z < -1600\ {\rm kpc\,km\,s^{-1}}$8, median $L_z < -1600\ {\rm kpc\,km\,s^{-1}}$9 kpc, and a broad spread in vertical extent (Deepak, 2024).

These dynamical realizations agree on the core phenomenology: Sequoia is retrograde, dynamically hot, and less bound than lower-energy retrograde components. They disagree mainly on how much neighboring retrograde debris should be included under the same label.

3. Stellar populations, metallicity, age, and the abundance sequence

Chemical characterization is where Sequoia becomes more than a kinematic overdensity. In a differential high-precision analysis of 12 candidates selected in $E > -1.3\times 10^5\ {\rm km^2\,s^{-2}}$0–$E > -1.3\times 10^5\ {\rm km^2\,s^{-2}}$1 space, Sequoia stars at $E > -1.3\times 10^5\ {\rm km^2\,s^{-2}}$2 were found to have significantly lower $E > -1.3\times 10^5\ {\rm km^2\,s^{-2}}$3, $E > -1.3\times 10^5\ {\rm km^2\,s^{-2}}$4, $E > -1.3\times 10^5\ {\rm km^2\,s^{-2}}$5, $E > -1.3\times 10^5\ {\rm km^2\,s^{-2}}$6, $E > -1.3\times 10^5\ {\rm km^2\,s^{-2}}$7, and $E > -1.3\times 10^5\ {\rm km^2\,s^{-2}}$8 than Gaia-Enceladus stars, with representative mean offsets $E > -1.3\times 10^5\ {\rm km^2\,s^{-2}}$9 dex in Na, $J_{\phi}/J_{\mathrm{tot}} < -0.6$0 dex in Mg, and $J_{\phi}/J_{\mathrm{tot}} < -0.6$1 dex in Ca. The same study argued that Sequoia can be chemically separated from Gaia-Enceladus or in situ stars only when abundance precision reaches $J_{\phi}/J_{\mathrm{tot}} < -0.6$2 dex (Matsuno et al., 2021).

This low-$J_{\phi}/J_{\mathrm{tot}} < -0.6$3, low-Na sequence is interpreted as evidence that Type Ia supernovae affected Sequoia at lower metallicity than Gaia-Enceladus. A plausible implication is that the Sequoia progenitor formed stars less efficiently, or had a lower mass, so that Fe from delayed channels became important earlier in its chemical evolution (Matsuno et al., 2021).

At the same time, other work shows that Sequoia’s inferred metallicity and age are strongly selection-dependent. In APOGEE+Gaia, a conservative normalized-action cut, $J_{\phi}/J_{\mathrm{tot}} < -0.6$4 and $J_{\phi}/J_{\mathrm{tot}} < -0.6$5, yielded a “Sequoia 06” sample with mean $J_{\phi}/J_{\mathrm{tot}} < -0.6$6 and a dominant age peak at $J_{\phi}/J_{\mathrm{tot}} < -0.6$7–$J_{\phi}/J_{\mathrm{tot}} < -0.6$8 Gyr; after chemical cleaning, the remaining 45 stars had mean/median $J_{\phi}/J_{\mathrm{tot}} < -0.6$9 (Feuillet et al., 2021). By contrast, a large Gaia–LAMOST study reported mean $(J_z - J_R)/J_{\mathrm{tot}} < 0.1$0, median $(J_z - J_R)/J_{\mathrm{tot}} < 0.1$1, mode $(J_z - J_R)/J_{\mathrm{tot}} < 0.1$2, and age statistics based on 66 stars with mean $(J_z - J_R)/J_{\mathrm{tot}} < 0.1$3 Gyr, median $(J_z - J_R)/J_{\mathrm{tot}} < 0.1$4 Gyr, and peak $(J_z - J_R)/J_{\mathrm{tot}} < 0.1$5 Gyr (Deepak, 2024).

The common ground across these analyses is that Sequoia is old and metal poor. The disagreement lies in whether its representative metallicity is closer to $(J_z - J_R)/J_{\mathrm{tot}} < 0.1$6, $(J_z - J_R)/J_{\mathrm{tot}} < 0.1$7, or the very metal-poor tail near $(J_z - J_R)/J_{\mathrm{tot}} < 0.1$8, and that disagreement tracks the adopted membership definition.

4. Neutron-capture chemistry and $(J_z - J_R)/J_{\mathrm{tot}} < 0.1$9-process enrichment

Neutron-capture abundances have become a second major axis of Sequoia research. In a combined Gaia Sausage/Sequoia study based on UVES spectroscopy plus archival data, the accreted sample showed clear Eu enhancement, $\eta < -0.15$0–$\eta < -0.15$1, with a tight low-metallicity sequence at $\eta < -0.15$2 and a later upturn interpreted as the onset of asymptotic giant branch contamination. Because that analysis combined Gaia Sausage and Sequoia members, it constrains an accreted-halo heavy-element history more than an isolated pure Sequoia sequence (Aguado et al., 2020).

A Sequoia-specific very metal-poor analysis with LAMOST/Subaru found the clearest Sequoia heavy-element signature to be “moderately high $\eta < -0.15$3-process abundances.” Of six Sequoia stars with robust metallicities, four had reliable Eu measurements and three of those four were $\eta < -0.15$4-I stars; in the higher-energy Sequoia subset, all three stars were $\eta < -0.15$5-I, with $\eta < -0.15$6, together with higher Sr and Ba than the full very metal-poor sample (Zhang et al., 2024). This places Sequoia above the typical very metal-poor halo level in Eu, but below the extreme $\eta < -0.15$7-II systems.

MINCE II reached a related but more tentative conclusion from only three Sequoia stars. In that sample, Sequoia showed $\eta < -0.15$8 systematically higher than Gaia Sausage-Enceladus at a given $\eta < -0.15$9, and median $J_\phi < -0.7\times 10^3\ {\rm kpc\,km\,s^{-1}}$0 dex versus $J_\phi < -0.7\times 10^3\ {\rm kpc\,km\,s^{-1}}$1 dex for GSE, suggesting a different neutron-capture history, but the authors explicitly treated the result as suggestive because of the very limited sample size (François et al., 2024).

A purely dynamical cross-match of $J_\phi < -0.7\times 10^3\ {\rm kpc\,km\,s^{-1}}$2-process-enhanced stars added one further clue: exactly one $J_\phi < -0.7\times 10^3\ {\rm kpc\,km\,s^{-1}}$3-I star, 2MASS J11444086$J_\phi < -0.7\times 10^3\ {\rm kpc\,km\,s^{-1}}$40409511, was tentatively associated with Sequoia DTG-1 at 20% membership confidence, with $J_\phi < -0.7\times 10^3\ {\rm kpc\,km\,s^{-1}}$5, $J_\phi < -0.7\times 10^3\ {\rm kpc\,km\,s^{-1}}$6, and $J_\phi < -0.7\times 10^3\ {\rm kpc\,km\,s^{-1}}$7 (Limberg et al., 2020). Taken together, these results support the view that Sequoia participated in early $J_\phi < -0.7\times 10^3\ {\rm kpc\,km\,s^{-1}}$8-process enrichment, though its detailed heavy-element evolution is not yet mapped as well as its orbital structure.

5. Relation to Gaia-Enceladus, Thamnos, and retrograde halo complexity

The central controversy in the Sequoia literature is whether it is a distinct merger remnant or a label imposed on part of a broader retrograde continuum. A major retrograde-halo analysis argued that Sequoia has a smaller range in orbital energies than previously suggested and is confined to high energy, with selection $J_\phi < -0.7\times 10^3\ {\rm kpc\,km\,s^{-1}}$9 and $E > -1.5\times 10^5\ {\rm km^2\,s^{-2}}$0. In that reading, Sequoia could be a small galaxy in itself, but it overlaps in both integrals-of-motion space and chemical-abundance space with the less bound debris of Gaia-Enceladus, so its nature cannot be fully settled (Koppelman et al., 2019).

A nearby-halo clustering study sharpened the point by splitting the conventional “Sequoia region” into three components, clusters #62, #63, and #64. In that framework, cluster #63 alone was identified as Sequoia proper, while #62 and #64 were treated as additional retrograde debris, likely remnants of separate accretion events or, at minimum, not securely part of the same progenitor (Ruiz-Lara et al., 2022). This result does not deny Sequoia; it narrows it.

Chemically, APOGEE-based cross-comparisons reinforce the ambiguity. When three literature-based Sequoia samples were reconstructed in APOGEE DR17, the M19 and N20 realizations were statistically consistent with Gaia-Enceladus, whereas the K19 realization was not. The same study therefore left open several possibilities: Sequoia as a detached lower-mass galaxy, Sequoia as a higher-angular-momentum or less-bound component of Gaia-Enceladus, or Sequoia as a mixed retrograde region whose chemical distinctness depends on the membership definition (Horta et al., 2022).

The controversy is thus not about whether retrograde high-energy debris exists. It is about whether the debris assigned the name “Sequoia” should be treated as one progenitor, one subset, or one historically important but intrinsically composite region of halo phase space.

6. Tracers, candidate members, and contested associations

Several objects have become important proxies in the Sequoia debate precisely because they sit near its phase-space boundaries. FSR 1758 is the classic example. Although that cluster helped inspire the name “Sequoia,” a chemo-dynamical analysis of APOGEE-2 giants concluded that it is a normal globular cluster with multiple populations and that its orbit is more consistent with Gaia-Enceladus-Sausage than with Sequoia. The paper emphasized the paradox directly: the cluster that gave rise to the name does not appear to belong to the event so named (Romero-Colmenares et al., 2021).

Antaeus provides a different kind of probe. It was identified as a wide retrograde moving group in the Galactic disk plane whose mean action values $E > -1.5\times 10^5\ {\rm km^2\,s^{-2}}$1, mean energy $E > -1.5\times 10^5\ {\rm km^2\,s^{-2}}$2, metallicity $E > -1.5\times 10^5\ {\rm km^2\,s^{-2}}$3, and age $E > -1.5\times 10^5\ {\rm km^2\,s^{-2}}$4 Gyr resemble Sequoia-like debris, but with extraordinarily low $E > -1.5\times 10^5\ {\rm km^2\,s^{-2}}$5. On that basis it was proposed, cautiously, as possible debris from the dense core of the Sequoia progenitor, driven to low vertical action by dynamical friction before final disruption (Oria et al., 2022).

Stellar systems other than streams and clusters have also been linked to Sequoia. A high-resolution study of halo wide binaries argued that WB2, the pair HD134439/HD134440, is more likely related to the Sequoia event than to other halo components because it combines low $E > -1.5\times 10^5\ {\rm km^2\,s^{-2}}$6 with orbital properties matching Sequoia stars and Sequoia-associated globular clusters (Lim et al., 2021). At the same time, not every chemically remarkable retrograde star securely diagnoses Sequoia alone. BPM 3066, for example, has exceptional $E > -1.5\times 10^5\ {\rm km^2\,s^{-2}}$7Li and $E > -1.5\times 10^5\ {\rm km^2\,s^{-2}}$8Be abundances, $E > -1.5\times 10^5\ {\rm km^2\,s^{-2}}$9 and $$-1.5\times 10^5 < E < -0.9\times 10^5\ {\rm km^2\,s^{-2}}$$0, on an eccentric strongly retrograde orbit confined below $$-1.5\times 10^5 < E < -0.9\times 10^5\ {\rm km^2\,s^{-2}}$$1 kpc from the plane, but its exact classification depends on whether low-energy retrograde stars are labeled Sequoia or Thamnos (Monaco et al., 30 Apr 2025).

These cases show that Sequoia is not inferred from a single tracer type. It is reconstructed from a network of halo stars, wide binaries, streams, and candidate associations, each of which illuminates a slightly different part of the underlying merger debris.

7. Origin models and the current interpretive landscape

A fully cosmological simulation containing analogues of Kraken, Gaia-Enceladus, and Sequoia provides one coherent origin model. In that simulation, the Sequoia analogue has $$-1.5\times 10^5 < E < -0.9\times 10^5\ {\rm km^2\,s^{-2}}$$2, $$-1.5\times 10^5 < E < -0.9\times 10^5\ {\rm km^2\,s^{-2}}$$3, merger ratio $$-1.5\times 10^5 < E < -0.9\times 10^5\ {\rm km^2\,s^{-2}}$$4, $$-1.5\times 10^5 < E < -0.9\times 10^5\ {\rm km^2\,s^{-2}}$$5, $$-1.5\times 10^5 < E < -0.9\times 10^5\ {\rm km^2\,s^{-2}}$$6, and approximate mixing redshift $$-1.5\times 10^5 < E < -0.9\times 10^5\ {\rm km^2\,s^{-2}}$$7. It is accreted slightly after Gaia-Enceladus, from the opposite side of the main galaxy along the same dominant filament, and leaves debris with $$-1.5\times 10^5 < E < -0.9\times 10^5\ {\rm km^2\,s^{-2}}$$8 and mean $$-1.5\times 10^5 < E < -0.9\times 10^5\ {\rm km^2\,s^{-2}}$$9, contributing preferentially to the outer halo. The paper’s main claim is that Gaia-Enceladus and Sequoia can be separate entities and still reproduce the observed present-day chemo-dynamical signatures (García-Bethencourt et al., 2023).

A more recent and more revisionist proposal treats Sequoia not as an unrelated merger remnant, but as the outermost debris of a common disrupted progenitor whose surviving nucleus is $E$00 Centauri. In this “$E$01 Dwarf” scenario, Sequoia is the earliest-stripped outer component, Thamnos is a later-stripped inner component, and Gaia-Enceladus may or may not belong to the same progenitor. The Sequoia sample in that analysis comprised 123 APOGEE stars and 85 GALAH stars; it was characterized as high-energy retrograde debris dominated by the primordial P1 population, lacking the centrally concentrated P2 population, less $E$02-enhanced than the inner components, and part of Eu-rich, $E$03-process-dominated outskirts with $E$04. The paper explicitly presented this as a plausible but not conclusive interpretive framework, and it stated that the connection with Gaia-Enceladus remains unsure (Souza et al., 24 Mar 2026).

The current landscape is therefore structurally plural. One line of work supports Sequoia as a distinct lower-mass retrograde merger, chemically offset from Gaia-Enceladus and naturally reproduced in cosmological simulations. Another line treats the Sequoia region as intrinsically composite. A third, explicitly interpretive line places Sequoia within an outside-in stripping sequence around $E$05 Centauri. What remains stable across these revisions is the empirical core: Sequoia denotes a high-energy retrograde accreted-halo population whose chemistry, orbit distribution, and relation to neighboring substructures continue to refine the merger history of the Milky Way.