Gaia-Enceladus-Sausage: Milky Way Merger Remnant

• Gaia-Enceladus-Sausage (GES) is the remnant of the Galaxy’s last major merger, exhibiting strongly radial orbits, near-zero net rotation, and distinct low-α chemical signatures.
• Chemodynamical studies combine Gaia astrometry and APOGEE spectroscopy to distinguish GES from in-situ populations using orbital energy, angular momentum, and abundance ratios.
• Insights from GES inform our understanding of Milky Way evolution, including disk heating, globular cluster associations, and potential triggers for bar formation and warp dynamics.

Gaia–Enceladus–Sausage (GES), also written Gaia-Sausage–Enceladus, Gaia-Sausage/Enceladus, or GS/E, is the dominant accreted component of the Milky Way’s inner stellar halo and is widely interpreted as the debris of the Galaxy’s last major merger. In the contemporary literature it is identified through the conjunction of highly radial, near-zero-net-rotation orbits, characteristic low-$\alpha$ chemical sequences, and a substantial contribution to the inner-halo stellar population. Most reconstructions place the event roughly $8$–$11$ Gyr ago, with much work centering on its role in shaping the inner halo, heating the proto-disk, and leaving behind a large system of field stars, globular clusters, and spatially coherent halo overdensities (Perottoni et al., 2022, Limberg et al., 2022).

1. Definition, nomenclature, and place in Galactic archaeology

GES emerged from Gaia-era halo studies as the principal radially biased accreted structure in the Milky Way. Across the literature summarized here, it is consistently described as the remnant of the last major merger experienced by the Galaxy, a system that dominates the inner stellar halo and is associated with the canonical “sausage” kinematics: strong radial anisotropy, very eccentric orbits, and low or slightly retrograde $L_z$ (Perottoni et al., 2022). In cosmological context, it is treated as a relatively massive dwarf progenitor, with some studies characterizing it as comparable in mass to the present-day Large Magellanic Cloud, while others infer lower stellar masses depending on the adopted metallicity calibration, selection function, and dynamical modeling (Ernandes et al., 10 May 2025).

The dual name reflects two initially emphasized observational aspects of the same system. “Enceladus” foregrounded the chemically distinct accreted population in energy–angular-momentum space, whereas “Sausage” highlighted the elongated velocity-space morphology produced by strongly radial orbits. Subsequent work generally treats these as manifestations of one accretion remnant. The residual terminological variation is therefore historical rather than taxonomic.

GES matters because it is simultaneously a tracer of the Milky Way’s merger history and a laboratory for dwarf-galaxy chemical evolution under rapid disruption. Its debris has been used to reconstruct the progenitor’s stellar mass, star-formation history, metallicity gradient, globular-cluster system, and possible dynamical consequences for the Galactic disk. This breadth of use explains why GES is now central to Galactic archaeology rather than merely one substructure among many.

2. Chemodynamical identification and dynamical structure

Modern GES work is based on joint use of astrometry, spectroscopy, and orbit modeling. The common dynamical coordinates are the specific orbital energy $E$, the angular momentum component $L_z$, the radial action $J_R$, and the eccentricity

$$e=\frac{r_{\mathrm{apo}}-r_{\mathrm{peri}}}{r_{\mathrm{apo}}+r_{\mathrm{peri}}}.$$

A standard anisotropy diagnostic is

$\beta = 1-\frac{\sigma_t^2}{2\sigma_r^2},$

with high $\beta$ indicating radial bias. Chemistry is typically folded in through abundance ratios such as $8$0, $8$1, and $8$2, using

$8$3

A key methodological result is that GES is not cleanly recoverable from any single projection of phase space. A self-consistent APOGEE DR17 + Gaia EDR3 analysis therefore used a novel set of chemodynamical criteria, combining $8$4, $8$5, eccentricity, and abundance patterns, and found that Sequoia substantially overlaps the GES chemical footprint while remaining offset toward lower metallicity; this makes joint chemo-kinematic selection preferable to chemistry alone (Limberg et al., 2022). Benchmarking against Auriga simulations reached a related conclusion: among simple observational cuts, the $8$6–$8$7 selection is best for purity, whereas a pure eccentricity cut is best for completeness, and different choices materially bias inferred progenitor properties (Carrillo et al., 2023). APOGEE DR16 + Gaia work likewise favored the near-non-rotating, high-$8$8 locus $8$9 and $11$0 as the cleanest GES window and showed that a chemically informed $11$1–$11$2 boundary efficiently removes in-situ contaminants (Feuillet et al., 2021).

Recent dynamical modeling has emphasized that the remnant is not described optimally by a constant-anisotropy distribution function. Using a chemically selected APOGEE DR17 × Gaia DR3 sample, one study found $11$3 beyond $11$4 kpc but only $11$5 at $11$6–$11$7 kpc, and showed that a two-component Osipkov–Merritt profile fits the radial trend better than a constant-$11$8 model (Lane et al., 4 Sep 2025). This strengthens the view that the central GES debris has undergone substantial dynamical processing even if its outer halo signature remains extremely radial.

3. Chemistry, metallicity distributions, and star-formation history

GES is chemically distinctive but not chemically singular. Different selection functions and surveys return different metallicity centroids, yet all place the system in the metal-poor, low-$11$9, accreted-halo regime. A chemodynamically selected APOGEE DR17 sample yielded a median $L_z$0 dex with $L_z$1 dex dispersion (Limberg et al., 2022). A SkyMapper–Gaia RVS analysis using a high-purity $L_z$2–$L_z$3 selection found $L_z$4 dex with $L_z$5 dex (Feuillet et al., 2020). APOGEE DR16 + Gaia selections gave a mean $L_z$6, shifting to mean/median $L_z$7 after chemically excising in-situ contamination (Feuillet et al., 2021). SEGUE/APOGEE analyses of high-eccentricity GSE-like stars in Hercules–Aquila Cloud and Virgo Overdensity found MDF peaks near $L_z$8, with the local prototypical GSE sample at $L_z$9 dex (Perottoni et al., 2022). These values are selection-dependent rather than mutually exclusive.

The abundance patterns are more stable than the absolute MDF peak. GES defines a single $E$0–$E$1 sequence with an $E$2-knee at $E$3 and uniformly low $E$4 to $E$5 dex in APOGEE-based work (Feuillet et al., 2021). Relative to surviving Milky Way satellites, GES stars show enhanced $E$6, $E$7, and $E$8, a pattern interpreted as evidence for higher star-formation efficiency and shorter enrichment timescales than in present-day low-mass dwarfs (Limberg et al., 2022). APOGEE DR17 comparisons across ten satellites further showed that a purely orbital GES sample falls almost entirely in the low-$E$9, high-$L_z$0 accreted locus, and that the metal-rich ends of the satellites sit at lower $L_z$1 than GES because their star formation continued longer (Fernandes et al., 2023).

Neutron-capture diagnostics extend this picture. A SAGA-based compilation of GSE stars found that $L_z$2 rises smoothly with $L_z$3, $L_z$4 stays flat at low metallicity and then rises above $L_z$5, and $L_z$6 does not turn down at the metal-rich end. That combination was interpreted as prolonged, low-efficiency star formation lasting more than $L_z$7 Gyr, followed by abrupt quenching associated with the merger (Ernandes et al., 10 May 2025). Independent CMD fitting of Gaia DR3 stars near the Sun recovered an almost linear age–metallicity relation with three dominant old populations and a smaller younger one: the bulk of star formation lasted for at least $L_z$8–$L_z$9 Gyr, ended about $J_R$0 Gyr ago, and spanned $J_R$1 to $J_R$2, while a younger population at $J_R$3 Gyr and $J_R$4 was found but left with unclear association to GSE (González-Koda et al., 27 Feb 2025). Taken together, these results suggest a dwarf galaxy that formed stars neither in a brief starburst mode nor in a long-lived Fornax-like mode, but in an extended low-efficiency mode truncated by accretion.

Representative quantitative determinations are summarized below.

Property	Representative determination	Source
Merger epoch	about $J_R$5–$J_R$6 Gyr ago	(Perottoni et al., 2022)
Field-star MDF	median $J_R$7 dex, dispersion $J_R$8 dex	(Limberg et al., 2022)
High-purity local MDF	$J_R$9 dex, $$e=\frac{r_{\mathrm{apo}}-r_{\mathrm{peri}}}{r_{\mathrm{apo}}+r_{\mathrm{peri}}}.$$0 dex	(Feuillet et al., 2020)
$$e=\frac{r_{\mathrm{apo}}-r_{\mathrm{peri}}}{r_{\mathrm{apo}}+r_{\mathrm{peri}}}.$$1-knee	$$e=\frac{r_{\mathrm{apo}}-r_{\mathrm{peri}}}{r_{\mathrm{apo}}+r_{\mathrm{peri}}}.$$2	(Feuillet et al., 2021)
Star-formation duration	$$e=\frac{r_{\mathrm{apo}}-r_{\mathrm{peri}}}{r_{\mathrm{apo}}+r_{\mathrm{peri}}}.$$3 Gyr from Eu/Ba/Mg clocks; $$e=\frac{r_{\mathrm{apo}}-r_{\mathrm{peri}}}{r_{\mathrm{apo}}+r_{\mathrm{peri}}}.$$4–$$e=\frac{r_{\mathrm{apo}}-r_{\mathrm{peri}}}{r_{\mathrm{apo}}+r_{\mathrm{peri}}}.$$5 Gyr from CMD fitting	(Ernandes et al., 10 May 2025, González-Koda et al., 27 Feb 2025)
Anisotropy profile	$$e=\frac{r_{\mathrm{apo}}-r_{\mathrm{peri}}}{r_{\mathrm{apo}}+r_{\mathrm{peri}}}.$$6 beyond $$e=\frac{r_{\mathrm{apo}}-r_{\mathrm{peri}}}{r_{\mathrm{apo}}+r_{\mathrm{peri}}}.$$7 kpc, $$e=\frac{r_{\mathrm{apo}}-r_{\mathrm{peri}}}{r_{\mathrm{apo}}+r_{\mathrm{peri}}}.$$8 at $$e=\frac{r_{\mathrm{apo}}-r_{\mathrm{peri}}}{r_{\mathrm{apo}}+r_{\mathrm{peri}}}.$$9–$\beta = 1-\frac{\sigma_t^2}{2\sigma_r^2},$0 kpc	(Lane et al., 4 Sep 2025)

The remnant also appears to preserve diluted memory of the progenitor’s internal metallicity gradient. A local APOGEE DR17 + Gaia DR3 study measured $\beta = 1-\frac{\sigma_t^2}{2\sigma_r^2},$1 and, using $\beta = 1-\frac{\sigma_t^2}{2\sigma_r^2},$2-body and HESTIA calibrations, inferred a progenitor gradient of $\beta = 1-\frac{\sigma_t^2}{2\sigma_r^2},$3 (Khoperskov et al., 2023). Auriga analogues support the same qualitative conclusion: all eight GES-like progenitors examined had negative infall gradients, which were strongly blurred by tidal stripping and phase mixing by $\beta = 1-\frac{\sigma_t^2}{2\sigma_r^2},$4 (Carrillo et al., 29 Sep 2025).

4. Globular clusters, $\beta = 1-\frac{\sigma_t^2}{2\sigma_r^2},$5 Centauri, and associated halo overdensities

A notable aspect of GES is the coherent globular-cluster system that appears to accompany its field-star debris. A combined APOGEE DR17 + Gaia EDR3 study identified eight globular clusters with membership probability $\beta = 1-\frac{\sigma_t^2}{2\sigma_r^2},$6 as likely GES members and determined their ages and metallicities through statistical isochrone fitting using Gaia photometry with APOGEE metallicity priors (Limberg et al., 2022). In that analysis, no probable GES cluster showed evidence for atypical internal $\beta = 1-\frac{\sigma_t^2}{2\sigma_r^2},$7 spreads except $\beta = 1-\frac{\sigma_t^2}{2\sigma_r^2},$8 Centauri.

$\beta = 1-\frac{\sigma_t^2}{2\sigma_r^2},$9 Centauri is therefore treated as a special case rather than an ordinary member of the cluster population. Under the assumption that it is a stripped nuclear star cluster, its mass and chemical complexity imply a host-galaxy stellar mass of $\beta$0, a value regarded there as consistent with literature expectations for GES. On that basis, the study proposed GES as the best available candidate for the original host of $\beta$1 Centauri (Limberg et al., 2022). This is a host-association argument, not a general statement that all GES mass estimates converge near $\beta$2.

Beyond globular clusters, large halo overdensities have been tied directly to GES debris. By combining SEGUE, APOGEE DR17, Gaia EDR3, and StarHorse distances, one study showed that the Hercules–Aquila Cloud and Virgo Overdensity have eccentricity distributions, MDFs, and abundance-plane loci indistinguishable from a prototypical GSE sample. About $\beta$3 of their stars have $\beta$4, and their high-eccentricity MDFs are strongly peaked around $\beta$5, leading to the interpretation that both structures are “unmixed debris” from the same merger, plausibly accumulated near apocenter (Perottoni et al., 2022).

RR Lyrae provide a complementary view in the inner-central Galaxy. Using BRAVA-RR/APOGEE RR Lyrae in the bulge region and calibrating against the Auriga Au-18 GES analogue, one study inferred that only $\beta$6–$\beta$7 of the inner-central halo RR Lyrae population originated from GES, markedly less than in the solar neighborhood. The same work found very few simulated GES particles with true bulge-confined orbits, implying that few if any bulge RR Lyrae should come from GES (Kunder et al., 15 Jul 2025). This indicates that GES is dominant in the nearby accreted halo but not necessarily at the smallest Galactocentric radii or in every tracer population.

5. Proposed role in shaping the Milky Way

GES is frequently invoked not only as a halo-building event but as a driver of early Milky Way disk evolution. In Auriga simulations selected for Sausage-like inner halos, the merger is gas-rich and contributes $\beta$8–$\beta$9 of the gas consumed by a merger-induced centrally concentrated starburst. That starburst rapidly forms a compact, rotationally supported thick-disk component, while the same merger heats the proto-disk and scatters older in-situ stars onto less-circular orbits, producing a metal-rich in-situ halo or “Splash”-like component that connects the thick disk to the inner halo (Grand et al., 2020). This supports a dual-origin picture in which GES contributed both dissipative star formation and non-dissipative disk heating.

Bar formation has also been linked to the event in simulation. In the Auriga analogue Au-18, a GES-like merger with stellar mass ratio $8$00 triggers strong tidal forcing, gas inflows, and a starburst, and a long-lived bar appears within $8$01–$8$02 Gyr of first pericenter. A rerun with the GES progenitors removed delays the bar by $8$03 Gyr and produces a weaker, more secularly formed structure. Across the Auriga suite, systems with GES-like mergers tend to form bars earlier, particularly when the stellar mass ratio is $8$04 (Merrow et al., 2023). This is a proposed formation pathway rather than an observationally settled causal claim for the Milky Way bar.

A similar simulation-based argument has been advanced for the Galactic warp. A gas-rich GSE major-merger model constructed with GIZMO reproduces an S-shaped stellar and gaseous warp, its asymmetry, age dependence, and a long-lived, nonsteady precession pattern that persists for more than $8$05 Gyr after merger completion. In that scenario, the warp is sustained by angular-momentum reorientation in the gas and by a tilted, oblate, retrograde dark halo left by the merger (Deng et al., 2024). Again, this is presented as a viable dynamical origin, not an exclusive explanation.

Chemical evolution models have further suggested that the trajectory of the merger matters for the disk’s $8$06 bimodality. In a multi-zone Galactic chemical-evolution model with angular-momentum dilution and radial gas flows, radial or retrograde GSE-like encounters drive a strong inward gas “sinking event,” lower the gas surface density across much of the disk, and allow ongoing SNe Ia to push $8$07 rapidly downward. Under those assumptions, a retrograde $8$08 GSE event can amplify the low-$8$09 sequence at fixed $8$10 and thereby help produce the Milky Way’s observed bimodality (Johnson et al., 9 Oct 2025). This suggests that GES may matter not only because it deposited stars, but because it changed the disk’s gas-dynamical state.

6. Debates, methodological sensitivities, and the evolving picture

Although the existence of a major radially biased accreted component is not in doubt, several aspects of GES remain under active revision. The first is methodological. Selection strategy changes inferred metallicity, sample purity, and progenitor mass. A SkyMapper–Gaia RVS study inferred $8$11 from a clean local MDF and a redshift-dependent mass–metallicity relation (Feuillet et al., 2020). A broader APOGEE DR17 benchmarking analysis found $8$12 across common GES selections and an average $8$13, while also showing that eccentricity cuts can overestimate the true stellar mass by $8$14 and $8$15–$8$16 cuts can underestimate it by $8$17 in Auriga analogues (Carrillo et al., 2023). By contrast, the $8$18 Centauri stripped-NSC argument yields $8$19 (Limberg et al., 2022). These values are not simple contradictions; they arise from different observables and modeling assumptions.

The second debate concerns internal structure. IllustrisTNG50 analogues show that about one third of GSE-like cases are built from two mergers rather than one, with single-merger cases typically accreted later than the two-merger cases; kinematics alone often cannot distinguish the two, while ages and chemistry are more informative (Folsom et al., 2024). More recent observational work has pushed this further. APOGEE DR17 analysis with the CREEK framework found two statistically distinct GES populations, a lower-energy Pop 1 and a higher-energy Pop 2, with different abundance distributions and chemical-evolution fits consistent with an inside-out progenitor and at least two stripping passages through the Milky Way disk (Berni et al., 30 Jan 2026). DESI-based GS$8$20 Hunter analysis then identified four distinct substructures within the traditional GSE region—GSE-GSH1, GSE-GSH2, GSE-GSH3, and GSE-GSH4—with characteristic ages of about $8$21, $8$22, $8$23, and $8$24 Gyr and distinct action-space and abundance patterns (Wang et al., 3 Jun 2026). This suggests that the observational “GSE box” may contain either multiple stripping phases of one progenitor or multiple early radial mergers that have not been fully disentangled.

A third issue is spatial extent. In Illustris, GE/GS-like mergers in the correct mass and epoch are common, but compact, dominant analogues are rare; the median radius of ancient radial-merger debris is $8$25 kpc, whereas the observed GE/GS is usually regarded as concentrated within $8$26 kpc. The best analog in that study also deposited $8$27 into the outer halo, raising the possibility that a substantial outer GES component remains to be identified (Elias et al., 2020). This is consistent with the idea that present-day local and inner-halo samples may trace only the most conspicuous part of a larger remnant.

Despite these complications, the core picture is robust. GES remains the name for the Milky Way’s dominant radially biased accreted inner-halo component, associated with a major ancient merger around $8$28, a metal-poor but relatively massive progenitor, and a star-formation history more efficient than that of most surviving dwarf satellites. What has changed is not the existence of GES, but the level of internal complexity now attributed to the debris field and to the progenitor’s pre-disruption structure.