The Gaia Sausage Enigma: The Milky Way Survived Not One, But Multiple Major Mergers ✦
The Gaia-Sausage-Enceladus (GSE) structure has traditionally been considered the remnant of the last major merger that shaped the inner halo of the Milky Way. However, analysis of DESI survey data using a new unsupervised clustering algorithm, GS³ Hunter, identified 17 structures in the local halo, including GSE and previously unknown streams. A detailed investigation of chronological, dynamical, and chemical properties revealed four distinct subgroups within the GSE region: GSE−GSH1 (age 12 Gyr), GSE−GSH2 (10 Gyr), GSE−GSH3 (8 Gyr), and GSE−GSH4 (7 Gyr). Despite overall similarities in phase-space distribution and elemental abundances, these subgroups exhibit notably different orbital actions and chemical compositions. These findings point to a complex formation history for GSE, which is not the product of a single event but a composite structure assembled through several successive mergers in the early evolution of the Galaxy.
Context
Understanding how our Galaxy grew and formed is one of the key challenges in modern astrophysics. The stellar halo of the Milky Way preserves fossilized evidence of past mergers with smaller satellite galaxies, much like geological strata on Earth. The Gaia Sausage/Enceladus (GSE) structure holds a special place—it was considered the 'last major merger' that shaped the inner halo. However, the simplicity of this picture raised doubts: how could a single object leave such diverse chemical and dynamical traces? A new study, armed with big data and machine learning, allows us to peer into the Galaxy's past with unprecedented detail.
Methods
Astronomers applied a new unsupervised clustering algorithm, GS3 Hunter, to a sample of 86,945 stars obtained with the DESI (Dark Energy Spectroscopic Instrument) spectrograph. This algorithm, like a seasoned detective, searches for groups of stars with similar characteristics in a multidimensional parameter space: chemical composition (derived using the 'Payne' method, reminiscent of the work of Cecilia Payne-Gaposchkin), orbital integrals of motion, and age determined from photometry using PARSEC isochrones. Additionally, Kullback–Leibler divergence was used to automatically link candidate groups. This allowed the identification of 17 distinct structures in the local halo, including previously known ones, but within GSE, four distinct components emerged.
Results
The four substructures, named GSE-GSH1–4, exhibit systematic differences in binding energy, angular momentum, and age: from ~12 billion years for the oldest to ~7 billion years for the youngest. Their orbits are radial, with high eccentricity, typical of stars captured from outside. Chemical analysis revealed that, despite a common high abundance of oxygen and carbon (signs of rapid star formation), each group has unique ratios of elements such as aluminum and magnesium. For example, in GSE-GSH2 and GSE-GSH3, a lower [Al/Fe] ratio is observed at the same metallicity, indicating a different enrichment rate by supernovae. The spread in [C/N] and [Mg/Fe] ratios suggests that the stars were born from gas that was unevenly mixed, in different star-forming regions. Of particular interest are Type Ia supernovae, arising from the explosion of white dwarfs (predicted by Chandrasekhar), which contribute decisively to the iron and manganese content. As Fred Hoyle and colleagues showed, different types of supernovae produce different sets of elements, which is exactly what is observed in the chemical inhomogeneity of the GSE components.
Implications
The discovery of GSE's multiple nature transforms it from a monolithic relic into a complex 'galactic cocktail.' This means that the inner halo of the Milky Way was assembled not in a single catastrophic event, but through a series of accretions, possibly from satellites of different masses and with different star formation histories. This conclusion supports the hierarchical model of galaxy assembly, where large structures grow by absorbing many smaller ones. The work also demonstrates the power of modern spectroscopic surveys combined with machine learning algorithms—the era of true 'galactic archaeology' is opening. This fundamentally changes our understanding of how galaxies grow.
Future development
In the near future, a similar analysis will be extended to the whole sky thanks to next-generation surveys such as the spectroscopic projects 4MOST and WEAVE, as well as the photometric LSST survey. The James Webb Space Telescope will allow us to study the chemical composition of stars in the infrared with unprecedented precision. Simulations of multiple mergers in cosmological simulations like Illustris will test how typical such a complex assembly history is for galaxies like ours.
Impact
This research will impact several fields: from the physics of supernovae (refining nucleosynthesis models) to cosmology, since the dynamics of stellar streams are sensitive to the distribution of dark matter in the Milky Way's halo.
Next steps
The next step will be a detailed chemical analysis of each substructure using high-resolution spectroscopy on next-generation telescopes to definitively confirm their distinct origins.
Key open problems
The results are directly linked to the key unsolved problem of galaxy formation: how exactly hierarchical accretion creates the observed diversity of structures. They could also shed light on the nature of dark matter, as the dynamics of stellar streams are sensitive to the mass distribution in the halo, including possible dark matter clumps.
🎯 The 'Gaia Sausage' structure got its name from its characteristic elongated shape in velocity space, resembling a sausage, while 'Enceladus' refers to the ancient Greek giant, symbolizing its enormous contribution to the growth of the Milky Way.
🎬 This story echoes the idea of galactic empires in Isaac Asimov's 'Foundation,' where civilizations grow by absorbing neighbors, but here gravity replaces politics, and stellar islands stand in for worlds.
Key numbers
- number of structures detected: 17
- number of GSE components: 4
- age range: 7–12 billion years
- typical [Mg/Fe] scatter: <0.07 dex
- DESI sample size: 86,945 stars