Merger fraction in galaxy groups and clusters at z < 0.2: A non-parametric morphological study with Subaru Hyper Suprime-Cam

TL;DR

This study uses a novel non-parametric morphological method on Subaru HSC data to analyze galaxy merger fractions at low redshift, revealing environmental and redshift dependencies and potential links to AGN activity.

Contribution

Introduces a new non-parametric classification combining Gini-$M_{20}$ and shape asymmetry for robust merger detection in large galaxy samples.

Findings

Merger fraction increases with redshift across environments.

Higher merger activity observed toward cluster centers.

Potential link between mergers and AGN triggering in dense regions.

Abstract

We investigate the environmental dependence of galaxy mergers using high-resolution imaging data from the Hyper Suprime-Cam (HSC) Subaru Strategic Program. We focus on galaxy groups and clusters at $z < 0.2$ identified by the Sloan Digital Sky Survey as a laboratory of galaxy environment. We develop a new non-parametric classification scheme that combines the Gini-$M_{20}$ statistics with the shape asymmetry parameter, enabling robust identification of mergers with both central concentration and outer morphological disturbances. Applying this method to a sample of 33,320 galaxies at $0.075 \leq z < 0.2$ taken by the HSC, we identify 12,666 mergers, corresponding to a merger fraction of 38%. Our results are consistent with visual classifications from the GALAXY CRUISE project, validating the effectiveness of our method. We find that the merger fraction increases with redshift for all subsamples (field galaxies, galaxy pairs, and cluster members), and also shows a strong radial gradient within clusters, increasing toward the center. These trends suggest that merger activity is enhanced both at earlier cosmic times and in denser environments, particularly in galaxy groups. We also find tentative evidence that mergers may contribute to AGN triggering in cluster cores. Our study highlights the utility of combining non-parametric morphological diagnostics for large-scale merger identification and provides new insights into the role of environment in galaxy evolution.

Equations

$$S(G,M_{20})=0.139M_{20}+0.990G-0.327,$$

$$A_{\mathrm{shape}}=\frac{{\sum }_{i,j}|I_{ij}-I_{ij}^{180}|}{{\sum }_{i,j}|I_{ij}|}-A_{\mathrm{bkg}},$$

$$S(G,M_{20})>0\text{and}{A}_{\mathrm{shape}}>\mathrm{0.2.}$$

$$f_{\mathrm{merger}}=\frac{N_{\mathrm{merger}}}{N_{\mathrm{total}}}.$$

$$\delta f_{\mathrm{merger}}$$

$$\text{cutout\_smooth}\ge \text{ellip\_annulus\_mean\_flux}\cdot \frac{0.2}{\text{asym}+0.25}$$

$$\text{cutout\_smooth >= ellip\_annulus\_mean\_flux},$$

$$\text{cutout\_smooth}>=\text{ellip\_annulus\_mean\_flux}\cdot \frac{0.2}{\text{asym}+0.25}.$$

Full text

1 Introduction

Galaxy mergers are considered a crucial aspect for comprehending the evolution of galaxies (e.g., [Conselice (2014)], and references therein). These mergers often initiate star formation (SF) and/or active galactic nucleus (AGN) activity by increasing the inflow of material on a galactic scale into the vicinity of the nuclear region. Recent studies suggest that the occurrence of galaxy mergers may also be influenced by density environments (e.g., [Lin et al. (2004), McIntosh et al. (2008), Perez et al. (2009), Darg et al. (2010), Ellison et al. (2010), Alonso et al. (2012), Shibuya et al. (2022), Omori et al. (2023), Laishram et al. (2024), Pearson et al. (2024), Sureshkumar et al. (2024), Puskás et al. (2025)]) although whether lower- or higher-density environments are critical for galaxy mergers is a matter of debate (see table 6 in [Shibuya et al. (2025)]).

To understand galaxy mergers and their environmental dependencies observationally, it is important to classify mergers accurately. Various methods have been used to classify mergers: (i) visual classification that is often collaborated with citizen science (e.g., [Lintott et al. (2008), Kartaltepe et al. (2015), Holincheck et al. (2016), Walmsley et al. (2022), Tanaka et al. (2023), Euclid Collaboration et al. (2025)]), (ii) parametric methods such as measuring Sérsic index (de Vaucouleurs, 1948; Sérsic, 1963; Aceves et al., 2006), and (iii) non-parametric methods such as concentration–asymmetry–smoothness (CAS) parameter (Abraham et al., 1994; Takamiya, 1999; Bershady et al., 2000; Conselice, 2003), Gini–$M_{\rm 20}$ classification (Abraham et al., 2003; Lotz et al., 2004, 2008b), the multimode–intensity–deviation (MID) statistics (Freeman et al., 2013; Peth et al., 2016), and the shape asymmetry ($A_{\rm shape}$: Pawlik et al. (2016)). More recently, machine-learning (ML) based merger identification (Dom´ınguez Sánchez et al., 2018; Ćiprijanović et al., 2020, 2021; Omori et al., 2023; Euclid Collaboration et al., 2024; Ferreira et al., 2024; Rose et al., 2024) has also been actively utilized, and such methods can predict even the merger stage (Bottrell et al., 2019; Chang et al., 2022).

To address environmental dependence of galaxy merger and its connection to AGN activity, the merger fraction ($f_{\rm merger}$) is a crucial parameter. Although many authors investigated how the merger (or galaxy pair) fraction depends on the environment (e.g., Jian et al. (2012); Duan et al. (2025); Dalmasso et al. (2024); Puskás et al. (2025)), their conclusions remain controversial.

It is now well established that $f_{\rm merger}$ rises with redshift out to $z\sim 1-2$ for field galaxies (e.g., Conselice (2003); Lotz et al. (2008b, 2011)); in quasar-host samples, $f_{\rm merger}$ shows a weak dependence on AGN luminosity (Tang et al., 2023). While several studies find that $f_{\rm merger}$ increases with redshift in massive field galaxies (Lin et al., 2004; Man et al., 2016; Kim et al., 2021), other works focusing on lower-mass systems or intermediate-density environments report a decline in $f_{\rm merger}$ at lower redshifts ($z\approx 0.03-0.2$; Nevin et al. (2023)), and some find no significant redshift trend at all in visually-classified AGN-host samples (covering $z=0-2$; Villforth (2023)). This low-redshift and environment-dependent discrepancy may stem from differences in image quality (e.g., Bickley et al. (2024b)), in the definition of environment (e.g., local density vs. halo mass), and in the adoption of ML-based classification methods (e.g., Margalef-Bentabol et al. (2024)).

In this work, we revisit the environmental dependence of galaxy merger and its connection to AGN activity in the local Universe ($z<0.2$) using Hyper Suprime-Cam (HSC: Miyazaki et al. (2018)), which is mounted on the Subaru Telescope. We particularly focus on galaxy groups and clusters as a laboratory of galaxy environment. The high image quality of HSC enables us to find tidal features (Kado-Fong et al., 2018), dual quasars (Silverman et al., 2020), and to reveal AGN host properties through the 2D-image decomposition (Li et al., 2021; Toba et al., 2022b). For instance, Tanaka et al. (2023) demonstrated that a non-negligible fraction of galaxies classified as non-merger based on the SDSS image shows merging features on the HSC image. We optimize a methodology for galaxy merger selection based on Gini–$M_{20}$ and $A_{\rm S}$ calibrated by GALAXY CRUISE, a citizen science project (Tanaka et al., 2023).

The remainder of this paper is structured as follows. We describe sample selection and the basic information of the sample in section 2. The morphological classification method optimized for galaxy mergers and evaluation of classification accuracy are presented in section 3. The obtained merger fraction in galaxy groups and clusters, and its dependence on cluster redshift and distance from cluster center are presented in section 4. We then discuss the environmental dependence of mergers and its connection to AGN activity in section 5. We summarize the results of this work in section 6. This work assumes a flat Universe with $H_{0}$ = 67.8 km s*-1* Mpc*-1*, $\Omega_{M}$ = 0.308, and $\Omega_{\Lambda}$ = 0.692 (Planck Collaboration et al., 2016), which are the same as those adopted in Tempel et al. (2017).

2 Data

2.1 Sample selection

A summary of the sample selection process and the number of galaxies retained at each stage is shown in figure 1. We constructed our galaxy sample based on the group and cluster catalog of Tempel et al. (2017), which identifies galaxy associations using the Friends-of-Friends (FoF) algorithm in the Sloan Digital Sky Survey (SDSS; York et al. (2000)) spectroscopic sample . The availability of spectroscopic redshift information in SDSS enables us to assume a high probability of physical association among galaxies classified as members.

To obtain high-resolution morphological data, we cross-matched this catalog with HSC Subaru Strategic Program (HSC-SSP; Aihara et al. (2018b), \yearciteAihara18b, \yearciteAihara19, \yearciteAihara22)111We used s21a_wide data. The HSC-SSP is an optical-imaging survey covering approximately 1200 $\rm deg^{2}$ with five broadband filters and approximately 30 $\rm deg^{2}$ with five broadband and four narrowband filters (see Bosch et al. (2018); Coupon et al. (2018); Furusawa et al. (2018); Huang et al. (2018); Kawanomoto et al. (2018); Komiyama et al. (2018))222See also Schlafly et al. (2012), Tonry et al. (2012), Magnier et al. (2013), Chambers et al. (2016), Jurić et al. (2017), and Ivezić et al. (2019) for relevant papers.. While the SDSS images have a typical seeing of $\sim 1.1$ arcsec (Ross et al., 2011), the HSC offers significantly finer seeing of 0.6 arcsec. By focusing on galaxies present in both SDSS and the HSC-SSP region, we are able to utilize high-quality HSC imaging with associated redshift information, which is essential for accurate merger classification.

The cross-matching between the SDSS-based catalog and HSC data was performed with a positional tolerance of 1 arcsec. To mitigate contamination from bright nearby sources, we applied selection filters based on the bright-star mask (Coupon et al., 2018) provided in the HSC catalog (see also section 4.2 in Aihara et al. (2022)). The mask flags potential contamination based on the proximity and brightness of neighboring stars. In our analysis, we excluded objects flagged with _blooming, _ghost, or _halo in the $i$-band. We did not apply the broader _any mask or the _dip flag because the _dip condition tends to be triggered more frequently for non-merger galaxies with low interaction probability $P(\mathrm{interaction})$ in the GALAXY CRUISE sample (see section 3.2). Applying it would artificially lower the number of non-mergers and inflate the merger fraction.

We also excluded galaxies with redshifts below $z=0.075$ to avoid classification inaccuracies caused by excessive spatial resolution (see section 3.2.2 for details). Additionally, for galaxies belonging to groups or clusters, we retained only those systems for which all member galaxies had corresponding HSC imaging data, ensuring completeness. This criterion led us to exclude clusters and groups that are close to the edge of the HSC-SSP survey footprint. After applying these criteria, our final sample consisted of 33,320 galaxies, including both field galaxies and group/cluster members.

2.2 Sample properties

The final sample consists of 33,320 galaxies located within the HSC-SSP survey region, as shown in figure 2. These galaxies have spectroscopic redshifts in the range $0.075\leq z<0.2$, and their $i$-band magnitudes span from 14.0 to 21.2. The redshift and $i$-band magnitude histograms are illustrated in figures 3a and 3b.

We also confirmed that 4,827 galaxy groups and clusters are located in the HSC-SSP survey footprint. To characterize the galaxy environments, we also examine the histograms of galaxy richness ($N$, defined as the number of galaxies in a given group or cluster) and cluster mass ($M_{200}$, defined as the mass enclosed within a radius where the mean density is 200 times the critical density of the Universe), as presented in figures 3c and 3d. It is worth noting that the majority of our sample resides in group-scale systems with richness $N<18$.

3 Analysis

3.1 Calculating morphological diagnostics

To classify galaxy morphologies, we adopted a non-parametric approach by computing structural indicators directly from imaging data. For this purpose, we used the Python package statmorph333We used v0.5.7 (https://github.com/vrodgom/statmorph/releases/tag/v0.5.7). (Rodriguez-Gomez et al. (2019), \yearciteRGV2022), which is designed to calculate a range of morphological diagnostics for galaxies.

statmorph is primarily based on the methodology introduced by Lotz et al. (2004), along with subsequent improvements implemented in the IDL codes (Lotz et al. (2006), \yearciteLotz08a, \yearciteLotz08b). Given a galaxy cutout image, a corresponding weight map, and the point spread function (PSF), the code computes several diagnostic quantities, including the Gini–$M_{20}$ statistics (Lotz et al., 2004), the CAS parameters (Conselice, 2003), and the MID statistics (Freeman et al., 2013). It also performs two-dimensional Sérsic model fitting.

In this study, we applied statmorph to the HSC $i$-band images using cutouts with a radius eight times the Petrosian radius444The Petrosian radius was obtained from the SDSS DR10 PhotoObjAll table, where we adopted the petroR90_i parameter., following the procedure described in Khanday et al. (2022). We also extracted the PSF image for each galaxy using the PSF picker tool. Among the outputs produced by statmorph, we focused on two key morphological indicators for merger classification: the Gini–$M_{20}$ statistics and the shape_asymmetry parameter. These parameters were selected for their sensitivity to morphological disturbances, which are characteristic of ongoing or recent mergers. Figure 4 illustrates an example of the outputs by statmorph, including the fitted Sérsic model, residual images, segmentation maps, and detection masks used in the diagnostic calculations. For further details, see Rodriguez-Gomez et al. (2019).

3.1.1 Gini–$M_{20}$ statistics

The Gini–$M_{20}$ method (Lotz et al., 2004) provides a quantitative framework for identifying galaxy mergers based on the concentration and distribution of light in a galaxy image. The Gini coefficient ($G$) measures the inequality of the pixel flux distribution, with higher values indicating stronger central concentration. The $M_{20}$ parameter quantifies the second-order moment of the brightest 20% of the galaxy’s light and is sensitive to spatial extent and substructure.

These two quantities are combined into a linear statistic defined as:

[TABLE]

following Snyder et al. (2015). Galaxies with $S(G,M_{20})>0$ are considered merger candidates. This method is widely used due to its sensitivity to asymmetric features and multiple light concentrations that often indicate merging activity.

3.1.2 Shape asymmetry

The shape_asymmetry parameter measures the degree of morphological asymmetry in a galaxy, based on its segmentation map rather than its flux image (Pawlik et al., 2016). This makes it particularly effective for detecting faint tidal features and extended disturbances in low surface-brightness regions.

The parameter is calculated by rotating the segmentation map 180 degrees around its centroid and comparing it to the original:

[TABLE]

where $I_{ij}$ and $I_{ij}^{180}$ are the original and rotated pixel values, and $A_{\mathrm{bkg}}$ is the background asymmetry correction. Unlike traditional asymmetry measures, shape_asymmetry is more sensitive to outer, low-signal morphological disturbances and complements the Gini–$M_{20}$ criteria. If a galaxy has a shape_asymmetry value greater than 0.2, it is identified as a merger in 95% of cases with tidal features, according to the benchmark analysis by Pawlik et al. (2016).

3.1.3 A new method for galaxy merger classification

To improve merger identification accuracy, we propose a combined classification criterion using both $S(G,M_{20})$ and $A_{\mathrm{shape}}$. Each of these indicators is sensitive to different morphological signatures of mergers: $S(G,M_{20})$ captures concentrated and asymmetric light distributions, while $A_{\mathrm{shape}}$ excels at detecting faint tidal features and extended disturbances.

We define a galaxy as a merger candidate if it satisfies both of the following conditions:

[TABLE]

Because both metrics are computed within the segmentation map associated with each galaxy, this dual criterion enables us to detect merger features manifested in either brightness structure or geometric irregularity. The combined use of these indicators enhances robustness, particularly for galaxies with low surface brightness features that might be missed by traditional methods. This classification scheme forms the foundation for our subsequent analysis of merger fractions and environmental trends. The novelty of our method is calibrating combined Gini-$M_{20}$ and $A_{\mathrm{shape}}$ thresholds using the GALAXY CRUISE interaction parameter. Section 3.2.3 and figure 5 demonstrate this yields a higher correlation and improves both completeness and purity compared to single-metric approaches.

3.2 Evaluation of classification accuracy using GALAXY CRUISE

GALAXY CRUISE is a large-scale citizen science project conducted in the same HSC-SSP region as this study (Tanaka et al., 2023). Over two million independent visual classifications were collected for 20,686 galaxies with redshifts $z\leq 0.2$. These visual classifications are treated as the ground truth in our evaluation of the statmorph-based merger classification.

3.2.1 Definition of merger fraction

We define the merger fraction $f_{\mathrm{merger}}$ as the proportion of mergers within a given bin:

[TABLE]

The associated uncertainty is calculated using standard error propagation. Assuming Poisson statistics, the error is given by:

[TABLE]

3.2.2 Accuracy comparison with visual classification

Each galaxy in the GALAXY CRUISE sample is assigned two classification probabilities: $P(\mathrm{spiral})$ and $P(\mathrm{interaction})$, both ranging from 0 to 1. A higher $P(\mathrm{interaction})$ indicates a greater likelihood of the galaxy being visually classified as a merger (see Tanaka et al. (2023), for more detail). We use $P(\mathrm{interaction})$ as the horizontal axis binning parameter and compute the merger fraction in each bin using our automated classification.

Figure 5a compares the results from statmorph using default settings (black) and our adjusted settings (red). The ideal one-to-one relation is indicated by the yellow line. Our adjusted method shows significantly better alignment with the visual classification trend, indicating improved accuracy. Figure 5b presents a comparison of different classification methods applied to the same dataset. CAS (green), Gini–$M_{20}$ (blue), and our combined method (red, using equation (3)) are shown. Our method identifies more mergers across a wider range of $P(\mathrm{interaction})$, demonstrating improved sensitivity to faint or subtle merger signatures.

To evaluate redshift dependence, figure 5c shows the same analysis as (a), broken down by redshift bins. At lower redshifts ($z<0.075$), the agreement between $P(\mathrm{interaction})$ and the merger fraction decreases, suggesting an increased rate of false positives. We therefore excluded galaxies with $z<0.075$ from our final analysis.

3.2.3 Final thresholds for detection

Based on the evaluation above, we adopted the following detection threshold in our modified statmorph setup:

[TABLE]

(see Appendix A for further details). This empirically tuned criterion improves the robustness of morphological measurements against noise, especially in low surface-brightness galaxies.

4 Results

4.1 Merger classification

We applied the merger classification method defined in equation (3) to our galaxy sample using the adjusted statmorph configuration. We note that statmorph outputs a flag that indicates the quality of the basic morphological measurements. It takes an integer from 0 to 4. If it has 2 or more, it means the calculation was bad (see Rodriguez-Gomez et al. (2019) for further details). Hence, galaxies with flag $\geq$ 2 were considered “unclassified” in this work.

Figure 6 shows the distribution of galaxies in the Gini–$M_{20}$ plane. The solid line, given by $G=-0.14\,M_{20}+0.33$, defines the empirical separation boundary, above which galaxies are considered mergers based on Gini–$M_{20}$ statistics alone (Lotz et al., 2008b). The color scale represents the shape_asymmetry values. Most galaxies with high $A_{\mathrm{shape}}$ ($>0.2$) lie above this line, indicating a strong correlation between the two diagnostics. Notably, some galaxies with low $A_{\mathrm{shape}}$ ($\leq 0.2$) are also found above the boundary, underscoring the need to combine both indicators to avoid false positives and ensure robust classification.

Using the combined criterion ($S(G,M_{20})>0$ and $A_{\mathrm{shape}}>0.2$), we classified 12,666 galaxies (38% of the full sample) as mergers. The full classification breakdown by redshift bin is summarized in table 4.1.

4.2 Redshift distribution

We investigated how the merger fraction evolves with redshift across three subsamples: cluster member galaxies, galaxy pairs, and field galaxies. Figure 7a shows the merger fraction $f_{\mathrm{merger}}$ in three redshift bins: $0.075\leq z<0.10$, $0.10\leq z<0.15$, and $0.15\leq z<0.20$.

All subsamples exhibit an increasing trend in merger fraction with redshift. The measured fractions for member galaxies in each redshift bin are $0.352\pm 0.012$, $0.390\pm 0.012$, and $0.46\pm 0.03$, respectively. Galaxy pairs show similar values: $0.371\pm 0.016$, $0.392\pm 0.014$, and $0.43\pm 0.02$. Field galaxies follow with $0.352\pm 0.009$, $0.368\pm 0.007$, and $0.415\pm 0.010$. The correlation coefficients for these trends are 0.987, 0.991, and 0.959, respectively.

Although the differences are within statistical uncertainties, member galaxies and galaxy pairs tend to show slightly higher merger fractions than field galaxies. This suggests that mergers are more common at higher redshifts and in denser environments, consistent with hierarchical structure formation scenarios.

4.3 Galaxy cluster radial distribution

We further examined the environmental dependence of the merger fraction within galaxy clusters by analyzing its variation with projected radial distance from the cluster center. The radial distance $R$ was normalized by the virial radius $R_{200}$, defined as the radius within which the average density is 200 times the critical density of the Universe. Both the cluster center positions and $R_{200}$ values were taken from the group and cluster catalog of Tempel et al. (2017).

Figure 7b shows the merger fraction as a function of $R/R_{200}$. The values increase steadily from the outskirts to the central regions: 0.33 $\pm$ 0.02, 0.37 $\pm$ 0.02, 0.38 $\pm$ 0.02, 0.43 $\pm$ 0.02 and 0.43 $\pm$ 0.03, from outer to inner bins. The Spearman’s rank correlation coefficient is $-0.886$ (with p-value 0.019), indicating a strong negative radial gradient. These results imply that mergers are more likely to occur near the cluster center, where the local galaxy density is higher, which is in good agreement with Shibuya et al. (2025). This supports the scenario that dynamical interactions and tidal perturbations are enhanced in dense cluster cores.

5 Discussion

5.1 Evaluation of classification accuracy

As the galaxies in our sample do not have associated $P(\mathrm{interaction})$ values, a direct comparison with GALAXY CRUISE classifications is not possible. Instead, we compared overall classification trends and performed a visual inspection of selected galaxies.

Table 5.1 summarizes the number of mergers identified using three classification methods: CAS, Gini–$M_{20}$, and our proposed method (equation (3)). The difference in merger fractions between our sample and GALAXY CRUISE is within 3%, suggesting comparable accuracy. To further validate our classification, we visually inspected randomly selected samples from both the merger and non-merger categories. Representative examples are shown in figure 8. The galaxies classified as mergers exhibit clear signs of morphological disturbance, such as asymmetries or tidal features, while the non-merger galaxies appear morphologically regular. This supports the overall reliability of our automated classification scheme.

5.2 Redshift dependence of merger fraction

Our results confirm a positive correlation between merger fraction and redshift, consistent with previous findings such as those by Conselice et al. (2009). However, our measured merger fractions are approximately three times higher than those reported in earlier studies. This discrepancy may be attributed to several factors:

• •

Differences in redshift range and sample selection: the previous studies focused on galaxies at $0.2\leq z<1.2$ observed with e.g., the the Hubble Space Telescope (HST), while our sample lies at $0.075\leq z<0.2$ in the HSC-SSP region.

• •

Higher spatial resolution of the HSC data compared to SDSS, allowing for more sensitive detection of morphological disturbances.

• •

Adoption of a new classification scheme combining multiple non-parametric indicators.

Taken together, these factors suggest that our methodology enables the detection of a greater number of mergers, particularly those with subtle features that might be missed in previous analyses.

Although our sample spans a relatively narrow redshift range and benefits from the high resolution of HSC imaging, it is important to carefully consider the effect of decreasing angular resolution on physical scales at higher redshifts. This effect is particularly significant for faint and small galaxies, which are more susceptible to resolution loss than their brighter counterparts. If this resolution effect were corrected for, the observed redshift dependence of the merger fraction would likely become even more pronounced.

5.3 Environmental dependence of merger activity

We found that the merger fraction increases toward the centers of galaxy clusters and in higher-density environments. This contrasts with the result from Tanaka et al. (2023) (see also Omori et al. (2023)), which suggested a decrease in merger activity in high-density regions.

One possible explanation is the difference in the dominant halo mass scale. Our sample primarily consists of galaxy groups with richness $N<50$, where relative velocities are lower and mergers can occur more efficiently. In contrast, the Tanaka et al. sample likely includes a higher fraction of massive clusters, where high velocity dispersions suppress merger rates. This interpretation is supported by numerical simulations (e.g., Jian et al. (2012)).

Furthermore, when comparing local density (within to the 5th nearest neighbor) dependence using the same definition as Tanaka et al., as shown in figure 9, we find that for galaxies with $N\geq 10$, the merger fraction increases with density, while for $N<10$, the trend is weaker or ambiguous. This suggests that density-enhanced merger activity is most apparent in moderately rich systems. The comparison and specific derivation with the local density presented in Tanaka et al. (2023) (see figure 17 in Tanaka et al. (2023)) are discussed in Appendix B.

5.4 Connection between mergers and AGN activity

We explored whether galaxy mergers contribute to AGN activity in galaxy groups and clusters. AGN were classified based on the BPT diagnostics (Baldwin et al., 1981), using the criteria proposed by Kewley et al. (2001), Kauffmann et al. (2003), and Schawinski et al. (2007), as implemented in the SDSS DR10 emissionLinesPort table. Galaxies that exceed both the Kewley et al. (2001) and Schawinski et al. (2007) thresholds were selected, resulting in a total of 4492 AGN-host galaxies.

Figure 10 shows the merger fraction as a function of cluster-centric radius for galaxies with and without hosting AGN. We observed a slight increase in merger fraction toward cluster centers among AGN-host galaxies, although the overall merger fractions for AGN-host and non-AGN galaxies were statistically consistent. Conversely, as shown in figure 11, when examining AGN fraction ($f_{\rm AGN}$) as a function of cluster-centric distance for mergers and non-mergers separately, we found a mild increasing trend only for merger galaxies.

We found tentative evidence that galaxy mergers could enhance AGN activity in denser environments, albeit with low statistical significance. This enhancement of AGN in denser environments has also been found by Hashiguchi et al. (2023) who reported the AGN fraction shows a significant excess in the cluster center (but see e.g., Pimbblet et al. (2013) for counterargument). Recently, Drigga et al. (2025) also reported that the merger fraction in AGN associated with denser environments is higher than that in field environments in which they used HSC images to classify galaxy mergers. Those results suggest that galaxy mergers may play an important role in triggering AGN activity in dense environments. It should be noted, however, that the merger-AGN connection may depend on several factors, including cluster mass (e.g., Toba et al. (2024)), AGN selection methods (e.g., Galametz et al. (2009)), and AGN luminosity (e.g., Yuk et al. (2025)). Moreover, the occurrence of AGN activity also depends on the merger stage (e.g., Toba et al. (2022a); Bickley et al. (2024a); Ellison et al. (2025)), which should be kept in mind in interpreting these results.

6 Summary

In this study, we investigated the environmental dependence of galaxy mergers using high-resolution imaging data from the Hyper Suprime-Cam Subaru Strategic Program (HSC-SSP). Our main findings are summarized as follows:

• •

We developed a new morphological classification method that combines Gini–$M_{20}$ statistics with the shape_asymmetry parameter, enhancing the detection of both centrally concentrated and morphologically disturbed galaxies.

• •

Applying this method to a sample of 33,320 galaxies at $0.075\leq z<0.2$, we identified 12,666 mergers, corresponding to a merger fraction of 38%. Our classification results show good agreement with GALAXY CRUISE visual annotations.

• •

The merger fraction increases with redshift for all subsamples (member galaxies, galaxy pairs, and field galaxies), consistent with hierarchical formation scenarios.

• •

A clear radial dependence was observed in clusters, with merger fractions increasing toward the cluster center. This trend supports enhanced merger activity in denser environments, especially within galaxy groups.

• •

We also found tentative evidence that mergers may contribute to AGN triggering in high-density regions, although the statistical significance is limited.

These results demonstrate the effectiveness of combining non-parametric diagnostics for large-scale merger identification and highlight the importance of both cosmic time and environment in galaxy interaction processes.

{ack}

The Hyper Suprime-Cam (HSC) collaboration includes the astronomical communities of Japan and Taiwan, and Princeton University. The HSC instrumentation and software were developed by the National Astronomical Observatory of Japan (NAOJ), the Kavli Institute for the Physics and Mathematics of the Universe (Kavli IPMU), the University of Tokyo, the High Energy Accelerator Research Organization (KEK), the Academia Sinica Institute for Astronomy and Astrophysics in Taiwan (ASIAA), and Princeton University. Funding was contributed by the FIRST program from the Japanese Cabinet Office, the Ministry of Education, Culture, Sports, Science and Technology (MEXT), the Japan Society for the Promotion of Science (JSPS), Japan Science and Technology Agency (JST), the Toray Science Foundation, NAOJ, Kavli IPMU, KEK, ASIAA, and Princeton University.

This paper is based on data collected at the Subaru Telescope and retrieved from the HSC data archive system, which is operated by Subaru Telescope and Astronomy Data Center (ADC) at NAOJ. Data analysis was in part carried out with the cooperation of Center for Computational Astrophysics (CfCA) at NAOJ. We are honored and grateful for the opportunity of observing the Universe from Maunakea, which has the cultural, historical and natural significance in Hawaii.

This paper makes use of software developed for Vera C. Rubin Observatory. We thank the Rubin Observatory for making their code available as free software at http://pipelines.lsst.io/.

The Pan-STARRS1 Surveys (PS1) and the PS1 public science archive have been made possible through contributions by the Institute for Astronomy, the University of Hawaii, the Pan-STARRS Project Office, the Max Planck Society and its participating institutes, the Max Planck Institute for Astronomy, Heidelberg, and the Max Planck Institute for Extraterrestrial Physics, Garching, The Johns Hopkins University, Durham University, the University of Edinburgh, the Queen’s University Belfast, the Harvard-Smithsonian Center for Astrophysics, the Las Cumbres Observatory Global Telescope Network Incorporated, the National Central University of Taiwan, the Space Telescope Science Institute, the National Aeronautics and Space Administration under grant No. NNX08AR22G issued through the Planetary Science Division of the NASA Science Mission Directorate, the National Science Foundation grant No. AST-1238877, the University of Maryland, Eotvos Lorand University (ELTE), the Los Alamos National Laboratory, and the Gordon and Betty Moore Foundation. Data analysis was in part carried out on the Multi-wavelength Data Analysis System operated by the Astronomy Data Center (ADC) and the Large-scale data analysis system co-operated by the Astronomy Data Center and Subaru Telescope, National Astronomical Observatory of Japan.

This work is supported by Japan Science and Technology Agency (JST) Support for Pioneering Research Initiated by the Next Generation (SPRING), Grant Number JPMJSP2115. This work is also supported by the Japan Society for the Promotion of Science (JSPS) KAKENHI Grant Numbers 20K04027 (NO), 23K22537 (YT), 21K03632, and 25K07359 (MI). NO acknowledges partial support by the Organization for the Promotion of Gender Equality at Nara Women’s University.

Appendix A Adjustments to statmorph

In its default configuration, statmorph allows the user to input a mask image as a two-dimensional array of the same size as the science image. However, the masking information in the HSC images used in this study was insufficient to fully exclude neighboring sources. To address this, we modified the pipeline to focus only on the central object in each image, masking all other sources.

In addition, we revised the object segmentation process to improve the accuracy of morphological measurements. Specifically, we adjusted the segmentation map threshold used in the calculation of Gini and $M_{20}$ statistics. This adjustment incorporates the shape_asymmetry value as a control variable.Originally, shape_asymmetry was computed after the Gini–$M_{20}$ statistics. In our modified version, it is calculated immediately beforehand so that it can be used to dynamically set the segmentation threshold.

We updated the segmentation criterion equation (7) to equation (8). Segmentation maps generated using equation (7) tend to be too small for merger galaxies and too large for non-mergers, which can compromise morphological accuracy. This is because equation (7) selects pixels above ellip_annulus_mean_flux for inclusion in the segmentation area. The smaller this flux threshold is, the fainter the structure that can be captured—important for detecting tidal features in mergers. Therefore, a lower threshold is desirable for mergers, while a higher threshold is more appropriate for non-mergers.

We found that shape_asymmetry is an effective parameter for adjusting the segmentation threshold because mergers tend to exhibit higher shape_asymmetry values than non-mergers. The new formulation in equation (8) improves robustness to background noise and enables more reliable morphological diagnostics, particularly in faint or low-surface-brightness galaxies.

[TABLE]

[TABLE]

Appendix B Estimation of Local Galaxy Density

To facilitate comparison with figure 17 in Tanaka et al. (2023), we calculated local galaxy density using a similar method. The procedure was as follows:

Compute the line-of-sight velocity ($v_{\rm rad}$) from the spectroscopic redshift. 2. 2.

Select galaxies within $\pm$1000 km/s of the target galaxy in radial velocity space. 3. 3.

Compute the angular distance to the 5th nearest galaxy in this projected sample. 4. 4.

Convert this angular distance to physical distance ($r_{5}$) using the angular diameter distance. 5. 5.

Calculate the local surface density as $\Sigma=\frac{5}{\pi r_{5}^{2}}$ (in units of galaxies per Mpc2).

Note that this density estimate does not account for masked regions or incompleteness due to observational effects, and should therefore be interpreted as a simplified approximation.

To compare classification performance under different merger identification schemes, we cross-matched our sample with the GALAXY CRUISE sample. Figure 12 shows the merger fraction as a function of local density under three different criteria:

• •

Our method (light blue circles)

• •

GALAXY CRUISE with $P(\mathrm{interaction})>0.79$ (blue triangles)

• •

GALAXY CRUISE with relaxed threshold $P(\mathrm{interaction})>0.5$ (orange squares)

For reference, we also plot the values reported in Tanaka et al. (2023) (red crosses). Our method yields a higher merger fraction than the stricter GALAXY CRUISE threshold, and is broadly consistent with the relaxed version. This suggests that our approach is sensitive to intermediate merger candidates that may be missed by stricter visual thresholds.