Gaia-Enceladus/Sausage: Milky Way Merger Remnant

• Gaia-Enceladus/Sausage is the remnant of a significant merger, characterized by strongly radial orbits and a dominant presence in the Milky Way’s inner halo.
• It exhibits distinct chemical signatures, including a narrow metallicity distribution around [Fe/H] ≃ -1.2 and low net Galactic rotation.
• Multiple selection methods and simulations reveal varied sample purity and contamination, highlighting the complexity of its composite merger history.

Gaia-Enceladus/Sausage, also written Gaia-Sausage-Enceladus and often abbreviated GSE or GS/E, is the accretion remnant associated with the Milky Way’s last major or last significant merger. In current usage, the term denotes both the disrupted progenitor dwarf galaxy and the present-day stellar debris that dominates much of the inner stellar halo. Observationally, the system is identified through strongly radial kinematics, low net Galactic rotation, and distinctive abundance patterns; dynamically, it is widely placed at a lookback time of roughly $8$–$11$ Gyr and linked to a major restructuring of the Galaxy’s halo and disk (Wang et al., 3 Jun 2026).

1. Definition and historical placement

GSE was first recognized in Gaia-era phase-space analyses as a nearly radially anisotropic component of the inner halo. Its canonical signature is a large radial-to-tangential velocity-dispersion ratio, usually summarized by the anisotropy parameter

$$\beta \equiv 1 - \frac{\sigma_\phi^2 + \sigma_\theta^2}{2\sigma_r^2},$$

with values around $\beta \approx 0.8$ in the Solar neighborhood in early characterizations, and often $\beta \approx 0.8$–$0.9$ for GSE-like debris in later summaries. A second common diagnostic is the orbital eccentricity

$$e \equiv \frac{r_{\rm apo} - r_{\rm peri}}{r_{\rm apo} + r_{\rm peri}},$$

with GSE selections frequently requiring $e \gtrsim 0.7$ (Folsom et al., 2024).

The system is usually interpreted as the remnant of the last major merger experienced by the Milky Way, and several studies place the event between $8$ and $11$ Gyr ago. In this picture, GSE contributed a large fraction of the inner halo, produced the characteristic “sausage” morphology in velocity space, and deposited stars on low-$11$0, high-$11$1, highly eccentric orbits. Orbit-superposition reconstructions further indicate that the GSE-like component is nearly spherical, with $11$2, and exhibits an outer ridge at $11$3 kpc, interpreted as an apocentric pile-up of the progenitor’s orbit (Sato et al., 2021).

2. Phase-space selection and present-day debris field

No single observational definition of GSE is universal. Common selections use eccentricity, energy–angular momentum space, radial-action–angular-momentum space, or combined chemical tagging. A widely used action-space box requires

$11$4

and

$11$5

with a closely related “clean” SkyMapper–Gaia RVS selection adopting $11$6 and the same $11$7 limits. That latter sample yielded $11$8 stars and a narrow MDF centered at $11$9 dex with dispersion $$\beta \equiv 1 - \frac{\sigma_\phi^2 + \sigma_\theta^2}{2\sigma_r^2},$$0 dex, and was argued to minimize contamination relative to velocity-space or $$\beta \equiv 1 - \frac{\sigma_\phi^2 + \sigma_\theta^2}{2\sigma_r^2},$$1–$$\beta \equiv 1 - \frac{\sigma_\phi^2 + \sigma_\theta^2}{2\sigma_r^2},$$2 selections (Feuillet et al., 2020).

Selection dependence remains a central methodological issue. A systematic comparison of five observational strategies—eccentricity, $$\beta \equiv 1 - \frac{\sigma_\phi^2 + \sigma_\theta^2}{2\sigma_r^2},$$3–$$\beta \equiv 1 - \frac{\sigma_\phi^2 + \sigma_\theta^2}{2\sigma_r^2},$$4, $$\beta \equiv 1 - \frac{\sigma_\phi^2 + \sigma_\theta^2}{2\sigma_r^2},$$5–$$\beta \equiv 1 - \frac{\sigma_\phi^2 + \sigma_\theta^2}{2\sigma_r^2},$$6, action diamond, and $$\beta \equiv 1 - \frac{\sigma_\phi^2 + \sigma_\theta^2}{2\sigma_r^2},$$7–$$\beta \equiv 1 - \frac{\sigma_\phi^2 + \sigma_\theta^2}{2\sigma_r^2},$$8—found samples ranging from $$\beta \equiv 1 - \frac{\sigma_\phi^2 + \sigma_\theta^2}{2\sigma_r^2},$$9 to $\beta \approx 0.8$0 stars, with markedly different contamination and completeness. In the Auriga benchmark, the $\beta \approx 0.8$1–$\beta \approx 0.8$2 method was best for sample purity, while the eccentricity method was best for completeness. The same study concluded that GSE selection cannot be treated as interchangeable across science cases (Carrillo et al., 2023).

The debris field is not confined to the local halo. HAC and VOD have orbital eccentricity distributions comparable to GSE, strongly peaked MDFs, and indistinguishable abundance distributions in the analyzed chemical spaces, supporting their interpretation as unmixed GSE debris in two major halo overdensities (Perottoni et al., 2022). At smaller Galactocentric radii, RR Lyrae mapping of the inner-central halo indicates a much lower GSE fraction than in the Solar neighborhood: only $\beta \approx 0.8$3–$\beta \approx 0.8$4 of the inner-central halo RR Lyrae population is consistent with a GES origin, and very few RR Lyrae stars with bulge-confined orbits are expected to have originated from GES (Kunder et al., 15 Jul 2025).

3. Metallicity, abundance ratios, and age structure

Published GSE metallicity distributions are similar in broad form but not identical in detail. A self-consistent APOGEE DR17 $\beta \approx 0.8$5 Gaia EDR3 characterization reported a median $\beta \approx 0.8$6 dex with $\beta \approx 0.8$7 dex dispersion for stars selected by novel chemodynamical criteria (Limberg et al., 2022). Other APOGEE-based selections found a mean $\beta \approx 0.8$8 with an approximately Gaussian width of $\beta \approx 0.8$9 dex, while high-eccentricity SEGUE-based fits for $\beta \approx 0.8$0 yielded $\beta \approx 0.8$1 and $\beta \approx 0.8$2 dex in one comparison sample (Feuillet et al., 2021).

In abundance space, GSE is typically described as a single $\beta \approx 0.8$3 sequence with the onset of SN Ia enrichment marked by an $\beta \approx 0.8$4-knee. Different selections place that knee at $\beta \approx 0.8$5 or $\beta \approx 0.8$6. APOGEE analyses also emphasize uniformly low $\beta \approx 0.8$7 dex relative to the Milky Way disc, and a tight accreted locus at high $\beta \approx 0.8$8 and low $\beta \approx 0.8$9; one empirical definition of the chemically unevolved accreted locus is $0.9$0 and $0.9$1 (Fernandes et al., 2023). At the same time, GSE stars show an excess of $0.9$2, $0.9$3, and $0.9$4 in comparison to surviving Milky Way dwarf satellites, a contrast interpreted as a difference in star-formation efficiencies and timescales (Limberg et al., 2022).

Neutron-capture abundances add a further temporal constraint. A SAGA-based compilation found that the joint behavior of $0.9$5, $0.9$6, and $0.9$7 indicates a prolonged period of slow star formation lasting over $0.9$8 Gyr, followed by abrupt quenching, with no super-solar $0.9$9 and a truncation near $$e \equiv \frac{r_{\rm apo} - r_{\rm peri}}{r_{\rm apo} + r_{\rm peri}},$$0. In that interpretation, GSE lies between Sculptor-like short-duration and Fornax-like very extended star-formation histories (Ernandes et al., 10 May 2025).

Direct age inference from Gaia CMD fitting yields a more structured chronology. ChronoGal recovered three main populations and a fourth smaller one arranged along an almost linear age–$$e \equiv \frac{r_{\rm apo} - r_{\rm peri}}{r_{\rm apo} + r_{\rm peri}},$$1 relation. The three oldest populations correspond to the bulk of star formation lasting for at least $$e \equiv \frac{r_{\rm apo} - r_{\rm peri}}{r_{\rm apo} + r_{\rm peri}},$$2–$$e \equiv \frac{r_{\rm apo} - r_{\rm peri}}{r_{\rm apo} + r_{\rm peri}},$$3 Gyr and ending about $$e \equiv \frac{r_{\rm apo} - r_{\rm peri}}{r_{\rm apo} + r_{\rm peri}},$$4 Gyr ago, with metallicities from $$e \equiv \frac{r_{\rm apo} - r_{\rm peri}}{r_{\rm apo} + r_{\rm peri}},$$5 to $$e \equiv \frac{r_{\rm apo} - r_{\rm peri}}{r_{\rm apo} + r_{\rm peri}},$$6. These were grouped into an “isolation epoch” from $$e \equiv \frac{r_{\rm apo} - r_{\rm peri}}{r_{\rm apo} + r_{\rm peri}},$$7 to $$e \equiv \frac{r_{\rm apo} - r_{\rm peri}}{r_{\rm apo} + r_{\rm peri}},$$8 Gyr and a “merger-event epoch” from $$e \equiv \frac{r_{\rm apo} - r_{\rm peri}}{r_{\rm apo} + r_{\rm peri}},$$9 to $e \gtrsim 0.7$0 Gyr. The fourth population, at $e \gtrsim 0.7$1 Gyr and $e \gtrsim 0.7$2, has an unclear link to GSE (González-Koda et al., 27 Feb 2025).

4. Progenitor mass, geometry, globular clusters, and internal gradients

Mass estimates for the progenitor vary substantially with methodology. Redshift-dependent mass–metallicity inversions using $e \gtrsim 0.7$3 and a mean stellar age of $e \gtrsim 0.7$4 Gyr gave $e \gtrsim 0.7$5 in one study (Feuillet et al., 2020). Related estimates gave $e \gtrsim 0.7$6, and a benchmark across several observational selections yielded an average $e \gtrsim 0.7$7 of $e \gtrsim 0.7$8 from the $e \gtrsim 0.7$9 mass–metallicity relation (Carrillo et al., 2023).

Direct density-modeling of high-purity samples produces lower stellar masses. A distribution-function-corrected fit to APOGEE DR16 $8$0 Gaia found that GS/E has a shallow density profile in the inner Galaxy, a break between $8$1 and $8$2 kpc, and a triaxial shape with axis ratios $8$3. The major axis is oriented about $8$4 from the Sun–Galactic centre line and $8$5 above the plane. The inferred stellar mass was

$8$6

implying that GS/E could make up as little as $8$7–$8$8 of the Milky Way stellar halo (Lane et al., 2023). A later re-analysis using a radially varying anisotropy profile increased that mass estimate to

$8$9

while retaining GS/E on the low end of the literature range (Lane et al., 4 Sep 2025).

The globular-cluster connection is especially important. In a joint analysis of stars and probable GCs, APOGEE observations showed no evidence for atypical $11$0 spreads among probable GSE globular clusters, with the exception of $11$1 Centauri. Under the assumption that $11$2 Cen is a stripped nuclear star cluster, the stellar mass of its progenitor was estimated as $11$3, described as well within literature expectations for GSE, leading to the proposal that GSE is the best available candidate for the original host galaxy of $11$4 Cen (Limberg et al., 2022).

Internal chemical structure is another active line of reconstruction. Auriga analogues of GES-like progenitors show negative stellar metallicity gradients at infall, ranging from $11$5 to $11$6 dex/kpc against radius, which become much shallower by $11$7, with percentage changes of $11$8–$11$9 in radial metallicity gradient. This suggests that present-day debris may retain only a blurred version of the progenitor’s original chemical cartography (Carrillo et al., 29 Sep 2025).

5. Dynamical and chemo-dynamical impact on the Milky Way

Simulations consistently assign a major dynamical role to the GSE merger, but the detailed mechanism differs by observable. In Auriga cosmological MHD simulations, the GES merger is likely to have been gas-rich and to contribute $11$00–$11$01 of the gas to a merger-induced centrally concentrated starburst. That event rapidly formed a compact, rotationally supported thick disc and heated the proto-disc, with local $11$02 increases of $11$03–$11$04. The same simulations linked the merger mass and timing to the fraction and age distribution of local counter-rotating stars (Grand et al., 2020).

A separate Auriga analysis examined bar formation. In the Au-18 realization, a GES-like merger with peak stellar mass $11$05 reached first pericentre at $11$06 Gyr and final coalescence at $11$07 Gyr. Bar seeds appeared immediately after first pericentre, and a permanent bar was in place by $11$08 Gyr, whereas removing the GES-18 progenitor delayed stable bar formation by $11$09 Gyr (Merrow et al., 2023).

The Galactic warp has likewise been modeled as a GSE consequence. A GIZMO gas-rich merger with host virial mass $11$10, satellite virial mass $11$11, $11$12, and retrograde bias produced an S-shaped warp in both the stellar and H I disks after the merger. The fitted onset radius was $11$13 kpc, the best-fit slope was $11$14, and the young-star warp amplitude at $11$15 kpc reached $11$16–$11$17 kpc. In that simulation the warp persisted from $11$18 Gyr post-merger to $11$19 Gyr, i.e. for $11$20 Gyr (Deng et al., 2024).

Chemo-dynamical models further propose that the retrograde or radial nature of the merger trajectory matters for disk abundance bimodality. In a numerical GCE framework with radial gas flows, a retrograde GSE merger near $11$21 was found to drive a major sinking event, lower gas surface density across much of the disk, and accelerate the decline in $11$22 through continuing SN Ia enrichment. In that scenario, retrograde trajectories increase the number of low-$11$23 stellar populations relative to prograde trajectories and can help generate the observed bimodality when a preexisting high-$11$24 population is already substantial (Johnson et al., 9 Oct 2025).

6. Debates, composite scenarios, and current synthesis

The classical picture treats GSE as the debris of a single ancient merger, but recent work has raised two distinct complications: multiple stripping episodes within one progenitor, and genuinely composite origins involving more than one progenitor.

Evidence for a multi-passage history comes from APOGEE DR17 analyses that separate GES into a low-energy population of $11$25 field stars and a high-energy population of $11$26 field stars. The low-energy component is chemically more enriched, the high-energy component more metal-poor, and the difference is reproduced by inside-out chemical-evolution models with a former metallicity gradient of $11$27. This work explicitly interprets GES as a multi-passage event through the Milky Way disc, with a first passage $11$28–$11$29 Gyr ago stripping outer stars and a second around $11$30 Gyr ago stripping inner stars (Berni et al., 30 Jan 2026).

A stronger revision appears in a DESI-based unsupervised clustering study using GS$11$31 Hunter. That analysis identified $11$32 halo structures and argued that the GSE region itself contains four distinct substructures: GSE-GSH1 ($11$33 Gyr), GSE-GSH2 ($11$34 Gyr), GSE-GSH3 ($11$35 Gyr), and GSE-GSH4 ($11$36 Gyr). Although all four broadly match the overall GSE phase-space distribution and abundance patterns, they display distinct orbital actions and chemical abundances. The authors therefore argue that GSE is not the remnant of a single accretion event, but rather a composite structure assembled through multiple, sequential merger episodes during the early Milky Way (Wang et al., 3 Jun 2026).

Cosmological simulations also weaken any presumption that high radial anisotropy implies a single progenitor. In IllustrisTNG50, GSE-like debris was found in $11$37 of $11$38 Milky Way analogues, and two-merger GSEs accounted for a third of those cases. Kinematics alone could not robustly distinguish single-merger from two-merger GSEs; chemistry and star-formation histories were more informative, because single-merger GSEs tended to be more metal-rich and younger, with median infall times $11$39 Gyr ago, while the galaxies in two-merger GSEs had median infall times around $11$40 Gyr ago (Folsom et al., 2024).

A further complication is overlap with other accreted structures. In the APOGEE DR17 $11$41 Gaia EDR3 reconstruction, stars assigned to Sequoia essentially overlap the GSE footprint in all analyzed chemical-abundance spaces, but present lower metallicities (Limberg et al., 2022). This makes chemical separation non-trivial and reinforces the broader conclusion that neither kinematics alone nor chemistry alone is sufficient in every regime.

Taken together, the current literature supports a stable core picture—an old, radially biased, low-$11$42, inner-halo-dominant accretion remnant with a major role in Milky Way evolution—while leaving open whether the observed GSE locus is the debris of one progenitor, one progenitor stripped in multiple passages, or a chemically and chronologically composite structure. The newer results suggest that “Gaia-Enceladus/Sausage” may function less as a single-object label than as the name for the dominant region of phase space in which several early accretion histories overlap.