LEGA-C Galaxy Survey

Updated 13 December 2025

LEGA-C Survey is a large public spectroscopic program using ESO's VLT that provides high-resolution optical spectra of approximately 4000 galaxies between 0.6 and 1.0 to study galaxy evolution.
It employs deep integration and advanced spectral fitting techniques to accurately determine stellar ages, metallicities, absorption-line indices, and kinematic properties.
The survey delivers actionable insights into galaxy assembly, quenching mechanisms, and scaling relations, serving as a benchmark for comparison with cosmological simulations and theoretical models.

The Large Early Galaxy Astrophysics Census (LEGA-C) Survey is a 130-night public spectroscopic program on the ESO Very Large Telescope, acquiring deep, high-resolution continuum spectra for a mass-complete sample of ∼4000 galaxies at $0.6 < z < 1.0$ in the COSMOS field. Targeting primarily $K_s$ -selected galaxies ( $K_s < 20.7 - 7.5 \log[(1+z)/1.8]$ ), LEGA-C is the first survey to deliver high signal-to-noise, rest-frame optical spectroscopy sufficient to robustly determine stellar ages, metallicities, absorption-line indices, and dynamical properties for galaxies near the peak epoch of cosmic star formation. The survey’s unprecedented combination of sample size, S/N (median ∼20–30 Å $^{-1}$ ), and structural ancillary data enables quantitative fossil record analysis, transforming constraints on galaxy assembly, quenching, kinematic evolution, and chemical enrichment at intermediate redshift ( $z\sim0.7$ ).

1. Survey Design, Data Acquisition, and Sample Definition

LEGA-C employs the VIMOS spectrograph at the 8 m ESO VLT (Unit Telescope 3, Melipal), with multi-object masks and $1''\times8''$ slits. Each mask has ∼20 hr of on-source integration, achieving $R\sim3500$ (∼35 km s $^{-1}$ instrumental resolution) over 6300–8800 Å ( $\lambda_{\mathrm{rest}}\sim3500$ –5500 Å for $z\sim0.8$ ). The sample is drawn from the $K_s$ -band-selected UltraVISTA catalog and is >90% complete above $M_*\sim10^{10.4} M_\odot$ .

The primary target selection is stellar-mass limited and largely model-independent. Redshift constraints ensure coverage of critical absorption and emission features for absorption-line and emission-line analysis. Full spectroscopic and imaging (HST/ACS F814W) coverage facilitates robust measurements of effective radii ( $R_e$ ), axis ratios, Sérsic indices ( $n$ ), and precise dynamical analyses.

DR2 released 1988 spectra, with DR3 expanding to over 4000 spectra and providing value-added catalogs of kinematic, structural, and stellar population parameters, Lick indices, and completeness weights. Quality cuts on S/N, redshift, and spectral coverage enable construction of high-fidelity subsamples for stellar population, dynamical, and scaling relation studies (Straatman et al., 2018, Gallazzi et al., 8 Dec 2025).

2. Methodologies for Stellar Population and Dynamical Analysis

Spectral analysis employs both classic absorption-line (Lick/IDS) index measurements and full-spectral fitting techniques:

Key absorption indices: D $_n$ 4000, H $\delta_\mathrm{A}$ , H $\gamma_\mathrm{F}$ , H $\beta$ , Mgb, Fe5270, Fe5335, Fe4383. These are measured on emission-line–subtracted spectra, with emission component fits via pPXF (Wu et al., 2018).
Stellar population fitting: Bayesian frameworks such as BaStA, full-spectrum fitters (pPXF, Prospector, Bagpipes, etc.), and stochastic SFH libraries. Models account for dust attenuation (e.g., Charlot & Fall 2000), chemical enrichment, and diverse star formation and burst histories (Gallazzi et al., 8 Dec 2025).
Dynamical modeling: Velocity dispersions ( $\sigma_*$ ) and resolved stellar kinematics are extracted using pPXF and axisymmetric Jeans models (JAM), with stellar light profiles from HST decomposed into Multi-Gaussian Expansions. Observed kinematics are corrected for beam smearing, slit orientation, and seeing (Houdt et al., 2021, Bezanson et al., 2018).

Sample classifications separate quiescent and star-forming populations via UVJ color–color cuts, specific SFR thresholds, and emission-line equivalent width criteria (Wu et al., 2018, Gallazzi et al., 14 Nov 2025).

3. Scaling Relations: Ages, Metallicities, and the Mass–Size Plane

LEGA-C uniquely constrains the scaling of stellar population properties with mass and structure at $z\sim0.7$ :

Age–Mass and Metallicity–Mass Relations: Light-weighted mean ages and metallicities exhibit a bimodal age distribution (young star-forming, old quiescent), with transition near $M_*\sim10^{11} M_\odot$ (Gallazzi et al., 8 Dec 2025, Gallazzi et al., 14 Nov 2025). Metallicity exhibits a single steep-to-flat trend, with flattening above $M_*\sim6\times10^{10} M_\odot$ ; no age bimodality is seen in metallicity.
Correlations with Velocity Dispersion: Stellar age is more tightly correlated with $\sigma_*$ than with $M_*$ , with a sharp young-to-old transition at $\log\sigma \sim 2.3$ ( $\sim$ 200 km s $^{-1}$ ), while metallicity correlates similarly with both $M_*$ and $\sigma_*$ .
Downsizing: The scaling relations at $z\sim0.7$ mirror those at $z\sim0$ , indicating that archeological downsizing trends were already established 6 Gyr ago (Gallazzi et al., 8 Dec 2025).
Mass–Metallicity Relation (MZR): At $z\sim0.7$ , the gas-phase MZR extends to $\log(M_*/M_\odot)\sim11.1$ and shows a mild 0.05–0.13 dex offset to lower metallicities than $z\sim0$ , with a flattening above $\sim10^{10.6} M_\odot$ and strong sensitivity to AGN feedback prescriptions in theoretical models (Lewis et al., 2023).
Fundamental Plane (FP): Both quiescent and star-forming massive galaxies lie on a common mass FP with similar scatter, indicating stable baryon-to-dark-matter ratios within $R_e$ and constraining the coevolution of mass, size, and velocity dispersion (Graaff et al., 2021).

4. Quenching Pathways, Structural Evolution, and Environmental Dependencies

LEGA-C results reveal multiple physical paths to quiescence:

Slow Quenching: Star formation ceases gradually ( $\gtrsim1$ Gyr), yielding quiescent galaxies structurally similar to progenitor star-formers, with size growth attributable to “progenitor bias.”
Fast Quenching: Rapid shutdown (few $\times10^8$ yr) via gas-rich mergers or disk instabilities triggers central starbursts; this produces compact, post-starburst (PSB) galaxies with inverse age gradients (younger centers), small $R_e$ , and high merger rates (∼40%), contrasting with inside-out growth of normal quiescent systems (Wu et al., 2018, D'Eugenio et al., 2020).
Size–Age Relations: Larger quiescent galaxies at fixed $M_*$ are slightly younger (Δage ≲ 500 Myr), while PSBs are compact and the youngest subset (Wu et al., 2018).

Environmental analysis demonstrates that the most massive slow-rotator elliptical galaxies are preferentially assembled earlier in overdense regions, consistent with accelerated minor merging; at these masses, only the densest environments have a high ( $\sim90$ \%) slow-rotator fraction (Cole et al., 2020).

5. Stellar and Dynamical Structure: Kinematics and Rotational Support

LEGA-C provides the first statistical, resolved kinematic census for galaxies at $z\sim0.8$ :

Rotational Support: Nearly all massive star-forming galaxies display rotation-dominated kinematics ( $V/\sigma\sim1$ –2), whereas $\sim$ 50% of quiescent galaxies are rotation-supported, with the highest-mass quiescents ( $M_*\gtrsim2\times10^{11}M_\odot$ ) being exclusively dispersion dominated (Houdt et al., 2021, Bezanson et al., 2018).
Evolution of Kinematics: At fixed $\sigma$ , $z\sim0.8$ quiescent galaxies show nearly twice the rotational support as local counterparts; thus, quiescent galaxies lose angular momentum between $z\sim1$ and today, most likely via dissipationless (dry) merging (Bezanson et al., 2018, Houdt et al., 2021).
Jeans Modeling: JAM analysis yields dynamical masses within 1, 5, 10 kpc, $R_e$ , and $2R_e$ , finding robust constraints on total mass and $V/\sigma$ evolution. Dark matter fractions and IMF variations can be mapped through residuals on the FP and mass FP (Houdt et al., 2021, Graaff et al., 2021).

6. Stellar Population Evolution and Cosmological Applications

Archaeological Trends: Comparison to SDSS DR7, incorporating aperture bias corrections, shows that the metallicity–mass relation for quiescents evolves little from $z\sim0.7$ to $z\sim0.1$ , with most metallicity evolution happening in lower-mass star-forming galaxies ( $M_*<10^{10.8}M_\odot$ ). Median ages for both populations have increased by only 1–2 Gyr in 5 Gyr of cosmic time, implying ongoing rejuvenation, progenitor bias, and minor merging (Gallazzi et al., 14 Nov 2025).
Cosmic Chronometers and $H(z)$ : LEGA-C cosmic chronometer samples, based on full-spectrum or Lick-index fitting, yield $H(z=0.8)=113.1\pm15.1^{+29.1}_{-11.3}$ km s $^{-1}$ Mpc $^{-1}$ (statistical and systematic), anchoring model-independent measures of expansion history (Jiao et al., 2022).
Simulation Constraints: LEGA-C’s absorption-index distributions (e.g., $D_n4000$ , EW(H $\delta$ )) at $z\sim1$ provide stringent benchmarks for cosmological hydrodynamical simulations such as IllustrisTNG and SIMBA, identifying discrepancies in recent rejuvenation, AGN feedback efficiency, and the shape of the high-mass MZR plateau (Wu et al., 2021, Lewis et al., 2023).

7. Data Releases, Impact, and Prospects

LEGA-C DR2 and DR3 supply FITS/ASCII catalogs of spectra, redshifts, kinematics, structural parameters, and revised index measurements. Bayesian stellar population parameters and completeness weights enable robust demographic analyses and comparison to theoretical models. The survey bridges the gap between low-redshift IFU surveys and high- $z$ photometric studies, setting the gold standard for detailed spectroscopic studies of galaxy evolution at cosmic noon (Straatman et al., 2018, Gallazzi et al., 8 Dec 2025).

Future work includes extending these constraints to higher redshift with JWST and 30-m class telescopes, nuanced mapping of dark-matter fractions, and leveraging archaeological scaling relations to refine models of quenching, rejuvenation, and chemical enrichment in galaxy evolution.

Selected Reference arXiv IDs: