MAGPI Survey: Galaxy Evolution at z~0.3

• MAGPI Survey is an advanced VLT/MUSE program using adaptive-optics-assisted IFS to spatially resolve galaxy structures and key properties at z~0.3.
• The survey utilizes deep integrations and robust data pipelines to extract stellar and gas kinematics, metallicity gradients, and star-formation distributions.
• MAGPI bridges local legacy surveys and high-redshift studies, providing critical calibration for models of galaxy assembly, quenching, and morphological transformation.

The Middle-Ages Galaxy Properties with Integral field spectroscopy (MAGPI) Survey is a major VLT/MUSE program targeting the spatially resolved structure and evolution of galaxies at intermediate redshift ($z\sim0.3$), corresponding to a lookback time of 3–4 Gyr. MAGPI delivers adaptive-optics-assisted integral field spectroscopy for a statistically robust, mass-complete sample of central (“primary”) galaxies and a large set of satellite companions, enabling detailed studies of stellar and gas kinematics, chemical abundance gradients, star-formation distributions, and morphological transformations in a wide range of environments. MAGPI is positioned to bridge the gap between local IFS legacy surveys (e.g., SAMI, MaNGA) and both high-redshift and theoretical studies, thereby anchoring models of late-epoch galaxy assembly and quenching.

1. Survey Goals, Design, and Sample Selection

MAGPI was conceived to address key outstanding questions in galaxy evolution: How do the internal dynamical, chemical, and star-formation properties of galaxies evolve as feedback and environmental quenching become dominant? At $z\sim0.3$, simulations predict maximal diversity in evolutionary pathways due to the interplay between secular processes, gas accretion, mergers, tidal interactions, and AGN feedback (Foster et al., 2020).

The primary sample consists of $\sim$60 massive central galaxies (M$_* > 7\times10^{10}$ M$_\odot$) drawn from the GAMA group catalog, spanning a representative range of environments from field to cluster ($$11.35 \leq \log(M_\mathrm{halo}/M_\odot) \leq 15.35$$). These are complemented by secondary satellites ($M_* \gtrsim 10^9$–$10^{10}$ M$_\odot$) within the 1$'$×1$z\sim0.3$0 MUSE field. Deep integration (4.4 hr per field), combined with Ground-Layer Adaptive Optics (GLAO), delivers spatial resolution of 0.6–0.8 arcsec ($z\sim0.3$12.5–3.5 kpc at $z\sim0.3$2), sufficient to resolve the internal structure of both large and compact systems (Foster et al., 2020, Foster et al., 14 Jan 2025, Mai et al., 8 Dec 2025).

Comprehensive ancillary datasets from the GAMA survey (21-band photometry, environmental measures, group memberships) and HSC/SUBARU imaging are available for SED fitting, stellar mass determination, and cross-survey comparisons (Foster et al., 2020, Mukherjee et al., 2024).

2. Data Products, Reduction, and Sample Characterization

Observations were conducted with MUSE in Wide Field Mode, providing continuous spectral coverage (4650–9350 Å) at $z\sim0.3$3, and 0.2″/pixel sampling. Fields were observed with six × 2 × 1320 s exposures, with large ($z\sim0.3$45″) dithers and position angle rotations to mitigate systematics (Foster et al., 2020).

Data cubes are processed with the ESO MUSE pipeline, applying ZAP for advanced sky subtraction and PROFOUND for source segmentation and extraction of “mini-cubes” for each galaxy (Foster et al., 14 Jan 2025). Spectral energy distributions are fit with ProSpect (broadband $z\sim0.3$5–$z\sim0.3$6 photometry from KiDS/VIKING) to derive robust stellar masses, star-formation histories (SFHs), and stellar ages using a skewed normal SFH and Chabrier IMF (Foster et al., 14 Jan 2025). High-throughput pipelines (pPXF, GIST) extract stellar and ionized-gas kinematics (velocity $z\sim0.3$7, dispersion $z\sim0.3$8, Gauss–Hermite coefficients $z\sim0.3$9, $\sim$0), as well as emission line fluxes for SFR and chemical abundance diagnostics (Foster et al., 14 Jan 2025, Mai et al., 8 Dec 2025).

Sample selection imposes stringent criteria on S/N, spatial filling factor, and structural properties (e.g., $\sim$1/FWHM $\sim$2 1, $\sim$3 mag), yielding 77 central and satellite galaxies for dynamical analysis (Foster et al., 14 Jan 2025), 70 star-forming systems with high-quality metallicity gradients (Mai et al., 8 Dec 2025), and sub-populations optimized for kinematical-morphological mapping and resolved star formation studies (Foster et al., 23 Feb 2025, Mun et al., 2024, Mun et al., 2024).

3. Kinematics, Orbital Structure, and Mass Profiles

MAGPI delivers high-fidelity maps of LOSVD, stellar spin proxy $\sim$4, higher-order moments ($\sim$5, $\sim$6), and kinematic asymmetries. It applies both axisymmetric Jeans and full triaxial Schwarzschild dynamical models (DYNAMITE) to derive enclosed mass profiles, intrinsic shapes, and orbital distributions (Santucci et al., 2024, Derkenne et al., 2024).

• Spin and LOSVD: The spin parameter proxy $\sim$7 follows the Emsellem et al. (2007) definition, applying seeing corrections for robust comparison to MaNGA, SAMI, and simulations (Derkenne et al., 2024, Foster et al., 14 Jan 2025). At $\sim$8, the slow rotator fraction ($\sim$9) matches mass-matched MaNGA descendants, indicating the massive slow rotator population was already established by $_* > 7\times10^{10}$04 Gyr ago, with tentative evidence for mild spin-down in the fastest rotators (Derkenne et al., 2024).
• Kinematic Morphology–Density Relation: Bayesian consensus visual classification shows the Hubble sequence and dichotomy between rotating/non-rotating systems are present by $_* > 7\times10^{10}$1, but the kinematic morphology-density relation observed locally has not yet emerged. Non-rotators are not preferentially in dense environments, suggesting early “dynamical pre-processing” in low-mass groups before cluster assembly (Foster et al., 23 Feb 2025).
• Orbital Structure: Schwarzschild models reveal a broad mix of oblate, triaxial, and prolate intrinsic shapes, with a strong correlation between triaxiality and hot-orbit fraction. The hot-orbit fraction and radial anisotropy are enhanced in triaxial systems, consistent with a merger-driven assembly. Comparisons to EAGLE “MAGPI-like” mock galaxies confirm model reliability and quantify biases (e.g., NFW halos underestimate enclosed dark matter within $_* > 7\times10^{10}$2 by $_* > 7\times10^{10}$350%) (Santucci et al., 2024).
• Mass Profiles and DM Coupling: The median total density slope is $_* > 7\times10^{10}$4, with significant intrinsic scatter. There is no evidence for a “bulge–halo conspiracy”: the total-density profile does not maintain low scatter due to compensating stellar and DM components. Instead, $_* > 7\times10^{10}$5 is tightly coupled to $_* > 7\times10^{10}$6, following $_* > 7\times10^{10}$7. This stochasticity points to merger-driven diversity rather than fine-tuning (Derkenne et al., 2024).

4. Star Formation, Quenching, and Radial Trends

Integral-field mapping of SFR from H$_* > 7\times10^{10}$8 (and D4000 where emission is weak) enables resolved study of star-formation main sequence (SFMS) relations and quenching mechanisms:

• Resolved SFMS: At $_* > 7\times10^{10}$9, resolved SFMS ($_\odot$0) shows a slope $_\odot$1 (for OLS fits), with local stellar mass density as the principal driver of both SF and metallicity, and a weak secondary (inverse) dependence on $_\odot$2 (Koller et al., 2024). Global SFMS slope ($_\odot$3) matches cosmological simulations, but resolved SFMS is systematically steeper than EAGLE/TNG/Magneticum [$_\odot$4 vs $_\odot$5], highlighting discrepancies in subgrid feedback (Mun et al., 2024).
• Radial Profiles and Quenching: Galaxies above the SFMS show flat or mildly rising SFR profiles out to $_\odot$6, indicating disc-wide enhancement at this epoch. Main-sequence galaxies show modest positive gradients, while systems below the SFMS exhibit inside-out quenching signatures, with SFR increasingly suppressed toward the centers (Mun et al., 2024, Mun et al., 2024).
• AGN and Environmental Effects: Simulations find that centrally enhanced suppression of SFR (inside-out quenching) is strongly dependent on AGN mode (thermal/kinetic): TNG kinetic-mode produces strongest central suppression, EAGLE only weak central dips. Satellites show outside-in suppression tied to increasing halo mass, tracing environmental quenching (Mun et al., 2024).

5. Metallicity Gradients, ISM Structure, and Gas Turbulence

Resolved gas-phase metallicity gradient ($_\odot$7[O/H]) measurements using Blobby3D Bayesian forward modeling rigorously account for seeing, substructure, and kinematics (Mai et al., 8 Dec 2025):

• Metallicity Gradients: The median gradient is mildly negative ($_\odot$8 dex/kpc), with $_\odot$933% of galaxies displaying robustly negative slopes, 10% positive, and 57% consistent with flat. The $$11.35 \leq \log(M_\mathrm{halo}/M_\odot) \leq 15.35$$0[O/H]–$$11.35 \leq \log(M_\mathrm{halo}/M_\odot) \leq 15.35$$1 relation displays a U-shape, reflecting the balance between (i) accretion/outflow-driven mixing flattening $$11.35 \leq \log(M_\mathrm{halo}/M_\odot) \leq 15.35$$2 at low $$11.35 \leq \log(M_\mathrm{halo}/M_\odot) \leq 15.35$$3, (ii) in-situ enrichment steepening gradients at intermediate $$11.35 \leq \log(M_\mathrm{halo}/M_\odot) \leq 15.35$$4, and (iii) renewed mixing/flattening in high-$$11.35 \leq \log(M_\mathrm{halo}/M_\odot) \leq 15.35$$5, large systems (Mai et al., 8 Dec 2025).
• Gas Turbulence Coupling: Gas velocity dispersion ($$11.35 \leq \log(M_\mathrm{halo}/M_\odot) \leq 15.35$$6, from forward modeling) correlates positively with flattened or inverted metallicity gradients. Stronger turbulence, associated with enhanced feedback or gas transport, mixes metals more efficiently across discs. A negative correlation between $$11.35 \leq \log(M_\mathrm{halo}/M_\odot) \leq 15.35$$7[O/H] and effective radius further supports a size-dependent mixing efficiency (Mai et al., 8 Dec 2025, Mai et al., 2024).
• Drivers of Gas Turbulence: Across MAGPI and SAMI, $$11.35 \leq \log(M_\mathrm{halo}/M_\odot) \leq 15.35$$8 at fixed $$11.35 \leq \log(M_\mathrm{halo}/M_\odot) \leq 15.35$$9 is invariant with redshift out to $M_* \gtrsim 10^9$0, indicating feedback-regulated turbulence dominates through $M_* \gtrsim 10^9$1, with secondary contributions from gas transport/accretion (Mai et al., 2024). Residuals above the $M_* \gtrsim 10^9$2–$M_* \gtrsim 10^9$3 relation correlate with large-scale non-rotational motions, tied to clumpy accretion or streaming (Mai et al., 2024).

6. Kinematic Asymmetries, Morphology, and Environmental Imprints

Systematic comparison of kinematic asymmetries in stellar and gas velocity fields with local studies yields:

• Stellar vs Gas Asymmetry: Stellar kinematic asymmetry (Fourier-normalized by $M_* \gtrsim 10^9$4) anti-correlates with $M_* \gtrsim 10^9$5, while gas asymmetry shows no such trend, consistent with stars retaining dynamical memory of perturbations, whereas gas rapidly dissipates asymmetries (Bagge et al., 2024). In both MAGPI and SAMI, old-star-forming systems have higher gas asymmetry than stellar, interpreted as ongoing, clumpy external gas accretion, while young systems show the opposite (larger stellar asymmetry) (Bagge et al., 2024).
• Environmental Role in Kinematics: The only stellar dynamical parameter with a significant direct correlation to environment (group dynamical mass) is the mean $M_* \gtrsim 10^9$6 (LOSVD kurtosis); $M_* \gtrsim 10^9$7 is primarily determined by age and mass. For satellites, $M_* \gtrsim 10^9$8 rises with group mass, indicating enhanced dynamical heating (hot orbit fraction) in denser environments, while the spin–age anti-correlation is already established by $M_* \gtrsim 10^9$9 (Foster et al., 14 Jan 2025).
• Morphology and the Hubble Sequence: Visual and kinematic classifications confirm a fully developed Hubble sequence at $10^{10}$0. However, the local kinematic morphology–density relation (excess of non-rotators in dense environments) is not yet established—supporting a scenario in which galaxies undergo spin-down (dynamical pre-processing) in less massive groups prior to cluster assembly (Foster et al., 23 Feb 2025).

7. High-Redshift Probes: Ly$10^{10}$1 Emitters and the CGM

MAGPI’s survey depth enables the study of hundreds of LAEs up to $10^{10}$2, yielding new insight into circum/intergalactic HI distribution and LyC escape:

• Double-Peaked LAEs: Among 417 LAEs at $10^{10}$3, 108 exhibit double peaks. At $10^{10}$4, the double-peak fraction is $10^{10}$537%, declining to $10^{10}$614% at $10^{10}$7 due to IGM absorption. The blue-to-total flux ratio is anti-correlated with Ly$10^{10}$8 luminosity, with blue-dominated sources indicating gas inflow. Residual flux in the trough, peak separations, and red-peak asymmetry map neutral column densities and LyC escape channels (Mukherjee et al., 21 Oct 2025, Mukherjee et al., 2023).
• Spatial Variations in Ly$10^{10}$9 Halos: Extended LAEs display increasing blue-to-total flux ratio and decreasing peak separation with radius, mapping spatial variations in outflow velocity and $_\odot$0; anisotropies reflect non-spherical flows. Diagnostic combinations identify strong LyC-leaker candidates, with implied $_\odot$110-30% (Mukherjee et al., 21 Oct 2025).
• Reionization Era Bubble Sizes: In 22 LAEs at $_\odot$2–$_\odot$3, Ly$_\odot$4 line width correlates with estimated ionized bubble radii ($_\odot$5), with broad ($_\odot$6260 km/s) lines marking galaxies inside $_\odot$70.6 pMpc Strömgren zones. A clear positive correlation between FWHM and $_\odot$8 supports a patchy, inside-out reionization topology seeded by UV-bright LAEs (Mukherjee et al., 2024).

---

MAGPI establishes the epoch at $_\odot$9 as the critical bridge between today's quiescent, morphologically diverse population and the dynamically hot, rapidly evolving galaxies at cosmic noon. It offers an unprecedented calibration set for the subgrid physics of cosmological simulations and a fertile dataset for dissecting the co-evolution of mass assembly, dynamical heating, quenching, and chemical enrichment in the "middle ages" of galaxy evolution.