Morphological Bias: How Ellipticals and Spirals Trace the Cosmic Web Differently

Technical Summary: Morphological Bias: How Ellipticals and Spirals Trace the Cosmic Web Differently

Problem Statement
Observational cosmology has long established that galaxies are biased tracers of the underlying dark matter distribution, with clustering strength dependent on intrinsic properties such as luminosity and color. While the Halo Occupation Distribution (HOD) framework successfully links galaxy populations to halo masses, recent evidence suggests that the assumption of mass-only dependence is insufficient. Phenomena such as halo assembly bias—where clustering depends on secondary halo properties like formation time and concentration—may introduce systematic errors in standard models. Furthermore, while previous studies have noted differences in clustering between early-type (elliptical) and late-type (spiral) galaxies, a rigorous, data-driven measurement of the scale-dependent relative bias between these morphological types using angular power spectra has been lacking. This paper addresses the need to empirically test whether galaxy morphology imprints a distinct scale dependence on galaxy bias, potentially distinguishing between mass-dependent selection effects and genuine assembly bias signatures.

Methodology
The authors utilize angular auto- and cross-power spectra derived from two large-scale structure surveys: the Dark Energy Survey (DES) and the DESI Legacy Imaging Surveys Data Release 8 (DR8).

• Data and Classification: The study employs robust morphological classifications (elliptical vs. spiral) from DES (J. Vega-Ferrero et al. 2021) and DESI. To mitigate systematics and classification uncertainty, a novel per-galaxy weighting scheme ($w_g$) is introduced. This weight is derived from a linear regression predicting normalized redshift based on morphological probabilities, photometric redshift errors, and systematic parameters (e.g., shape parameters, flux radius). Galaxies with anomalous properties are down-weighted, improving sample purity without significantly altering the redshift distribution.
• Power Spectrum Estimation: Angular power spectra ($C_\ell$) and cross-spectra are estimated using the pseudo-$C_\ell$ method implemented in the NaMaster library. The analysis focuses on the multipole range $20 < \ell < 200$, where the Limber approximation and linear perturbation theory are valid.
• Cross-Tracer Clustering Ratio (CTCR): A key methodological innovation is the introduction of the CTCR, defined as the ratio of the auto-power spectrum of one tracer to the cross-power spectrum between two tracers ($R^\ell_{ij} = C^\ell_{ii} / C^\ell_{ij}$). This ratio isolates the relative bias ($b_i/b_j$) as a function of scale ($\ell$), effectively canceling out common cosmological dependencies.
• Modeling: The authors fit a parametric sigmoid model to the CTCR to characterize the transition in bias from large to small scales. Additionally, a standard linear bias analysis is performed for each tracer independently. The robustness of results is tested using two covariance estimators (theoretical Gaussian and bootstrap resampling), contamination tests (swapping 10% of morphological types), luminosity splits, and comparisons with N-body mock catalogs (Uchuu and Euclid Flagship 2).

Key Contributions

1. CTCR Observable: The paper establishes the Cross-Tracer Clustering Ratio as a robust observable for testing halo occupation and assembly-bias models, specifically designed to isolate relative bias without requiring absolute cosmological parameter constraints.
2. Scale-Dependent Morphological Bias: The study provides empirical evidence that the bias difference between ellipticals and spirals is not constant but scale-dependent.
3. Weighting Scheme: A data-driven weighting method is developed to enhance the purity of morphological samples in the presence of photometric uncertainties and systematics.
4. Simulation Validation: The findings are cross-validated against two distinct N-body simulations (Uchuu and Euclid Flagship 2), confirming that mixing galaxy populations (e.g., broad elliptical samples vs. narrow LRG selections) systematically alters measured bias amplitudes.

Results

• Relative Clustering: Across both surveys, elliptical galaxies are consistently more strongly clustered than spiral galaxies. The CTCR is greater than unity at all measured scales.
• Scale Dependence: The relative bias is close to unity on large angular scales ($\ell \lesssim 50$) but increases significantly toward smaller scales. In the DESI sample, where redshift distributions of the two morphological types are well-matched, the difference reaches an average separation of $2.6\sigma$ at $\ell \sim 150\text{--}200$.
• Bias Evolution:
  • Ellipticals: Exhibit a positive bias relative to the matter field that increases with multipole $\ell$. The bias shows little redshift evolution in the DES sample, consistent with a broad halo-mass distribution rather than a narrow, high-mass selection.
  • Spirals: Show a bias consistent with unity or weak anti-bias on large scales, declining toward smaller scales.
• Robustness: The results are stable across different covariance estimators, contamination levels, and luminosity splits. The DESI sample yields cleaner constraints than DES due to better-matched redshift distributions and higher statistics.
• Simulation Consistency: Mock catalogs confirm that the observed bias levels for "broad" morphological samples lie between those of pure LRG samples (high bias) and low-mass spiral samples (low bias), supporting the interpretation that the DES/DESI elliptical samples trace a broader range of halo masses than narrow LRG selections.

Significance and Claims
The authors claim that these findings provide empirical evidence that galaxy morphology imprints both the amplitude and the scale dependence of galaxy bias. The observed trend is qualitatively consistent with expectations from HOD and assembly-bias models, where ellipticals preferentially occupy older, more concentrated, and more strongly clustered halos, while spirals reside in younger halos with recent mass accretion.

However, the paper maintains a modest stance regarding the interpretation of assembly bias. The authors explicitly state that because their analysis does not fix the halo-mass distributions of the two morphological samples, the observed scale dependence cannot definitively distinguish between genuine assembly bias and mass-dependent selection effects (i.e., ellipticals simply tracing more massive halos on average). They conclude that disentangling these contributions requires joint clustering and weak-lensing analyses to constrain halo mass distributions.

Ultimately, the paper establishes the CTCR as a useful tool for future Stage-IV imaging surveys (such as CSST, Euclid, Roman, and Rubin/LSST) to test galaxy formation models with finer morphological classes and reduced statistical uncertainties.