Active galactic nucleus activity in nearby disk galaxies: The roles of bar strength and spiral arm morphology

Summary

This paper investigates the connection between active galactic nucleus (AGN) activity and large-scale non-axisymmetric structures—specifically stellar bars and spiral arms—in nearby disk galaxies. The study uses a sample of 843 morphologically undisturbed Sa–Sd galaxies from the Mapping Nearby Galaxies at Apache Point Observatory (MaNGA) survey, within the redshift range 0.026 ≤ z ≤ 0.1 and inclination i ≤ 70°. Bar strengths, stellar masses, colors, and AGN classifications were adopted from the literature, while spiral-arm classes (flocculent [FL], multi-armed [MA], and grand-design [GD]) were visually determined for the entire sample using optical imaging together with bulge–disk decomposition residual maps.

The main results are as follows:

1. Bar strength and AGN activity: Before controlling for stellar mass and color, the AGN fraction increases systematically with bar strength, from 0.10+0.02−0.02 in unbarred galaxies to 0.18+0.02−0.02 in weakly barred systems and to 0.34+0.03−0.03 in strongly barred galaxies. Both Fisher’s exact and Barnard’s tests indicate that these differences are statistically significant, particularly for comparisons involving strongly barred galaxies.

2. Controlling for stellar mass and color: After controlling for stellar mass and color, a statistically significant enhancement of AGN activity in barred galaxies remains detectable primarily at intermediate stellar masses (10.5 ≲ log(M⋆/M⊙) ≲ 11.0). In this mass range, strongly barred galaxies exhibit systematically higher AGN fractions than unbarred systems. However, no significant dependence on bar class is found at higher masses (log(M⋆/M⊙) > 11.0), suggesting that bar-driven gas inflow becomes less efficient in massive galaxies, likely due to the stabilizing influence of massive bulges and dynamically hot stellar components.

3. Spiral-arm morphology: Spiral-arm morphology alone does not show a robust independent connection with AGN activity once stellar mass and color are controlled. Before controlling for these parameters, GM (GD + MA) galaxies exhibit systematically higher AGN fractions than FL systems. However, after restricting the analysis to subsamples with statistically consistent stellar-mass and color distributions, most of these apparent differences become statistically insignificant. Only a weak and localized enhancement of the AGN fraction in GM versus FL galaxies remains detectable within the intermediate-mass regime (e.g., in the combined E + F bin, with PB = 0.038).

4. Combined effects of bars and spiral arms: When bar strength and spiral AC are considered jointly, the highest AGN fraction in the complete sample is observed in strongly barred GM galaxies (fAGN = 0.39+0.04−0.04), while the lowest AGN fraction is found in unbarred FL systems (fAGN = 0.09+0.03−0.02). After controlling for stellar mass and color, the statistically most robust enhancement is obtained in the intermediate-mass bin, where strongly barred GM galaxies show significantly higher AGN fractions than unbarred FL systems (fAGN = 0.28+0.07−0.05 vs. 0.04+0.03−0.02). Within the GM population itself, strongly barred galaxies also exhibit significantly enhanced AGN fractions compared to both unbarred and weakly barred GM systems.

The authors conclude that stellar mass and color drive the primary trend in AGN activity in nearby disk galaxies, while stellar bars and spiral arms act as secondary structural drivers. The effect of spiral-arm morphology is weaker and becomes most apparent when considered jointly with bars. The findings highlight the importance of simultaneously accounting for stellar mass, color, and the combined large-scale morphological structure of galaxies when investigating the triggering of nuclear activity. The results suggest that bars and spiral structures may act in a dynamically coupled manner, jointly contributing to the redistribution of angular momentum and gas transport toward the central regions of galaxies.

The conclusions remain robust against reasonable variations in the adopted stellar-mass and color thresholds, as well as when the stricter criterion of EW(Hα) > 3 Å is applied. The alternative stellar-mass- and color-weighting procedure of Garland et al. (2024) yields consistent results for the bar, spiral-arm, and combined arm–bar analyses. One of the main limitations of the study is the decrease in statistical power when the galaxy sample is simultaneously subdivided according to stellar mass, color, bar strength, and spiral AC.

Improvements to AI Systems Based on This Paper:

1. Causal Inference with Confounder Control

• Improvement: Implement a machine-learning pipeline that automatically controls for confounding variables (e.g., stellar mass, color) when testing correlations between structural features (bars, spiral arms) and AGN activity.

• Capability: The AI can distinguish true structural drivers from spurious correlations, reducing false positives in astrophysical surveys.

2. Multi-Feature Interaction Modeling

• Improvement: Develop a model that jointly analyzes bar strength and spiral-arm class (flocculent, multi-armed, grand-design) as interacting features, rather than independently.

• Capability: The AI can identify combined morphological signatures that predict AGN activity (e.g., strongly barred + grand-design galaxies) with higher accuracy than single-feature models.

3. Mass-Dependent Predictive Segmentation

• Improvement: Train a classifier that dynamically partitions galaxies by stellar mass (e.g., intermediate vs. high mass) and applies different feature weights for each regime.

• Capability: The AI can adapt its predictions to physical regimes, correctly identifying that bars matter for AGN triggering only in intermediate-mass galaxies, while ignoring them in massive systems.

4. Uncertainty-Aware Classification

• Improvement: Integrate Bayesian or ensemble methods that output confidence intervals for AGN fraction predictions, especially when sample sizes shrink after multi-way subdivision.

• Capability: The AI can flag low-confidence predictions (e.g., when few galaxies exist in a combined mass–color–bar–arm bin), preventing overinterpretation of sparse data.

5. Automated Morphological Feature Extraction

• Improvement: Use the paper’s visual classification criteria (bulge–disk residual maps, spiral-arm patterns) to train a convolutional neural network for automated bar-strength and spiral-arm classification.

• Capability: The AI can process large imaging surveys (e.g., Euclid, Roman) without human labeling, enabling scalable studies of galaxy structure–activity connections.

6. Simulation-to-Observation Transfer Learning

• Improvement: Use the paper’s empirical trends (e.g., bar-driven inflow efficiency decreasing with mass) to calibrate physics-based simulations of gas inflow and AGN feedback.

• Capability: The AI can generate synthetic galaxy catalogs that match observed AGN fractions as a function of mass and morphology, improving theoretical models of black hole growth.

7. Dynamic Feature Selection for Survey Design

• Improvement: Build an AI that recommends optimal sample selection criteria (e.g., mass ranges, bar-strength thresholds) to maximize statistical power for detecting AGN–structure links.

• Capability: The AI can guide future observational campaigns (e.g., JWST or VLT) to target galaxies where structural effects are most detectable, saving telescope time.

8. Cross-Dataset Generalization

• Improvement: Train a meta-model on MaNGA data that can predict AGN activity in other surveys (e.g., SDSS, DESI) by learning the mass–color–morphology relationship, with domain adaptation for different photometric bands.

• Capability: The AI can transfer knowledge across surveys, providing robust AGN predictions even when detailed morphological data are missing.