2.1 SubhaloFlag parameter

The SUBFIND algorithm (Springel et al. Reference Springel, White, Tormen and Kauffmann2001) is widely used to identify gravitationally bound structures known as ‘subhaloes,’ which typically represent galaxies. However, as previously mentioned, not all these objects qualify as traditional galaxies formed due to structure formation and collapse. Special care is necessary for the original Illustris simulation, as it is impossible to differentiate between these genuine galaxies and potentially suspicious identified structures, mainly when dealing with lower-mass objects. Adding a new classifierFootnote ^b called SubhaloFlag has vastly improved the analysis process in IllustrisTNG. With this additional parameter, the team has simplified the filtering of objects to be more reliable.

The classifier can have two values: 0 (False) and 1 (True). The subhalo is flagged as False (or of non-cosmological nature) if all of the following criteria are satisfied:

• • At its formation time, the subhalo is a satellite.

• • The subhalo forms within the virial radius of its host.

• • The dark matter fraction of the subhalo at formation is less than $0.8$ .

The Illustris team emphasises that these are relatively conservative criteria adopted to ensure a low false-positive rate. As a result, some spurious subhaloes may not be flagged as False (these would be false negatives), and a more rigorous approach may require checking the history of the subhalo. It is essential to highlight that flagged subhaloes should not necessarily be treated as numerical artefacts. Strictly speaking, numerical artefacts stem from issues related to the code and other numerical recipes and solutions. Although some flagged subhaloes can fall under this category, others may represent genuine structures, although not galaxies (e.g. tidal debris or kinematically distinct substructures of larger galaxies). Another physically grounded possibility is that some of these flagged subhaloes represent genuine but exotic types of dark matter-deficient galaxies formed in the aftermath of a high-velocity head-on collision between two gas-rich galaxies (Silk Reference Silk2019; Shin et al. Reference Kim, Jeong, Rey, Lee, Lee, Joo and Kim2020; Lee, Shin, & Kim Reference Lee, Shin and Kim2021; van Dokkum et al. Reference van Dokkum2022; Lee et al. Reference Lee, Shin, Kim, Shapiro and Chung2024). In our qualitative analysis, we must acknowledge all possibilities when interpreting the results and avoid the hasty dismissal of flagged subhaloes (Section 4).

Since our sample includes subhaloes with both values of SubhaloFlag parameter, we show the colour-coded number of subhaloes in the stellar mass-(stellar) metallicity plane in Figure 2 for the whole sample (upper panel) and additionally filtered True subhaloes (lower panel). The Cloudlet population is discernible in the upper panel, albeit appearing differently than in the previous work (i.e. there is a distinct second prong above the main sequence, but there is no noticeable, more significant gap between the two at lower masses). The additional filter clearly shows that this population predominantly comprises False subhaloes. However, even if we consider only genuine galaxies (lower panel), some scatter above the main sequence is still noticeable, although less significant.

Figure 2.

Colour-coded number of subhaloes in the stellar mass-metallicity plane for the whole sample (upper panel) and the filtered sample for $\texttt{SubhaloFlag}=1$ (lower panel).

As one of the primary aims of the present work is to revisit the previous result and reassess the habitability of genuine metal-rich dwarf galaxies in more detail, we will focus on this sub-population of the Cloudlet in the following. Naturally, we will also qualitatively discuss the possible habitability-related implications of the whole Cloudlet, including False subhaloes. A more detailed assessment of the habitability of the entire Cloudlet population requires a robust and precise classification of False subhaloes, which is beyond the scope of this work.

2.2 Habitability: Bimodal distribution of galaxies?

Previous work by Stojkovic et al. (Reference Stojkovic, Vukotic, Martinovic, Cirkovic and Micic2019) has found and named the Cloudlet population of metal-rich low-mass galaxies, implicating that it might be significant for its possible habitability aspect. Until then, studies of galactic habitability were indicating that large elliptical (and spiral) galaxies are most suitable for hosting habitable planets (e.g. Dayal et al. Reference Dayal, Cockell, Rice and Mazumdar2015; Zackrisson et al. Reference Zackrisson, Calissendorff, González, Benson, Johansen and Janson2016), and the Cloudlet population presented a new and interesting possibility for life at the other end of the morphological spectrum. The significance of this hypothetical population of metal-rich dwarfs stemmed from their low SFR coupled with high (above-MZR) metallicity. The authors used the indicator given by Dayal et al. (Reference Dayal, Cockell, Rice and Mazumdar2015), the quantity $N_\mathrm{p}$ , which is widely used to assess the number of potentially habitable planets in different types of galaxies. This quantity scales with metallicity, which in general amplifies the probability of a habitable planet or moon forming, and is inversely proportional to SFR. Therefore, at a given (high) metallicity, Cloudlet galaxies have the distinct advantage of a lower SFR than their high-mass counterparts. The low mass of the Cloudlet galaxies decreases their overall number of potentially habitable planets or moons. Still, the expected high frequency of this type of galaxy was thought to compensate for the small number of habitable planets or moons hosted by each individual system.

While this earlier work presented a valuable concept, we revisit and expand upon their analysis using the updated IllustrisTNG simulation, which, as previously mentioned, features improved hydrodynamics, feedback models, and sub-grid prescriptions. Moreover, it offers improved agreement with observational constraints across a wide mass range. However, previous work contains several methodological limitations that significantly affect the interpretation of their results. Although the SubhaloFlag classifier, described previously, was not available at the time and was only introduced in the newer IllustrisTNG simulations, Cloudlet galaxies could have been systematically assessed for their physical nature or environmental conditions, which was not included in the study. For example, tracking the history of Cloudlet galaxies could have revealed that the majority of those objects formed extremely late during the simulation, perhaps even in the last few snapshots, which would point to the conclusion that those objects are not necessarily galaxies in a traditional sense. Examining the dark matter content and distances to larger and more massive galaxies could have also proven useful. Such metrics could have revealed whether a Cloudlet galaxy might in fact be a satellite, a substructure, or even a transient tidal feature. In summary, while the identification of a bimodality in galactic habitability was an intriguing result, we argue that the conclusions of Stojkovic et al. (Reference Stojkovic, Vukotic, Martinovic, Cirkovic and Micic2019) cannot be validated and should have been subjected to more rigorous scrutiny. In particular, the lack of an environmental, structural, and evolutionary context for Cloudlet galaxies undermines the robustness of the inferred bimodality. Our current study revisits this issue using a higher-fidelity simulation and a more stringent classification of galaxies.

3.1 Kernel density smoothing

We applied kernel density smoothing to investigate the statistical significance of the separation between the high metallicity population and the bulk of the sample. Using a simple product Gaussian 2D kernel, the optimal smoothing bandwidths are calculated from the maximum likelihood cross-validation (Duin Reference Duin1976).

Figure 3 shows the resulting probability density distributions. The total sample and sub-sample cases show a similar distribution in the bulk of the sample. At the same time, the high-metallicity segment is separated from the rest of the sample for $\log (M_\star/M_\odot)\gt 8.5$ and $12+\log Z_\star \gt 10$ . The separation is visible at the first contour level of $0.01$ for the subhalo sub-sample, while the whole sample indicates the separation even at the $0.05$ contour level. The subhalo objects, therefore, have less pronounced separation compared to the rest of the data points from the sample.

Figure 3.

Kernel density smoothing and the resulting probability density distribution for the selected sample in the metallicity-mass plane. Contour levels are at 0.01, 0.05, 0.5, and 1.5. Black – all data points (kernel widths $h_{\log M}=0.029$ , $h_{\log Z}=0.016$ ), grey (kernel widths $h_{\log M}=0.029$ , $h_{\log Z}=0.017$ ) – points with subhalo flag value 1.

Given that the highest contour value is $1.5$ , the separation occurs at best at the few per cent level for the whole sample and up to the 1% level for the subhalo objects. In the $\log (M_\star/M_\odot)\lt 8.5$ and $12+\log Z_\star \lt 10$ region, there is no clear separation, and the high-metallicity population can rather be considered as the extended tail of the main population. Given that there are no other visible separations within the sample, it is evident that apart from the main branch of the sample, there is a significantly smaller branch, yet visible at a per cent level, in the $\log (M_\star/M_\odot)\gt 8.5$ and $12+\log Z_\star \gt 10$ region.

3.2 Density-based clustering

To identify metal-rich dwarf galaxies scattered above the main sequence of the mass-metallicity relation, we used a density-based clustering algorithm called DBSCAN (Ester et al. Reference Ester, Kriegel, Sander and Xu1996; Schubert et al. Reference Schubert, Sander, Ester, Kriegel and Xu2017) included in the scikit-learn packageFootnote ^c (Pedregosa et al. Reference Pedregosa2011). The algorithm has two crucial input parameters: eps and min_samples. The former defines the maximum distance between two samples that should be considered neighbouring points. The latter sets a threshold for the minimum number of samples in a neighbourhood. Although eps is considered the most critical parameter in determining the outcome for a given dataset, min_samples controls the average density of the resulting detected clusters (the higher the value of min_samples, the denser the detected clusters will be, and vice versa). This algorithm is most effective for datasets with distinct clusters of similar density, which is not the case with our sample. However, alternative algorithms that can detect clusters of various densities and sizes failed to detect the sub-sample of metal-rich dwarf galaxies as a single cluster, which was our desired outcome. We describe our procedure in the following and show the results in Figure 4.

Figure 4.

Results of the DBSCAN algorithm application in the stellar mass-metallicity plane. Contours represent the main sequence, while the scatter points represent detected clusters, whose labels are indicated in the legend.

First, we apply the DBSCAN algorithm using $\texttt{eps}=0.1$ and $\texttt{min\_samples}=50$ as input parameters. This allows us to isolate the main sequence of the stellar mass–(stellar) metallicity relation, MZ $_\star$ R. The results revealed a single cluster (covering almost the entire main sequence) and outliers. Objects in the sparse regions of the main sequence (i.e. its higher and lower ends) are recognised as outliers. We then apply the algorithm to a sub-sample of outliers, with $\texttt{eps}=0.2$ and $\texttt{min\_samples}=10$ , which were adjusted to the new data. This revealed new data clusters: higher and lower ends of the main sequence, metal-rich dwarf galaxies scattered above, and true outliers. The results are shown in Figure 4, and the sub-sample of metal-rich dwarf galaxies contains 616 subhaloes, about 1% of the whole sample.

To ensure that the sub-sample is composed of genuine metal-rich dwarf galaxies, we also trace the history of each subhalo and record the snapshot in which the subhalo was formed (i.e. when it first appeared). The results, with the subhalo counts per snapshot bin, are listed in Table 1. Interestingly, most of the subhaloes in this sub-sample appear for the first time in the last snapshot at redshift $z=0$ . This makes them false negatives of the procedure that assigns the SubhaloFlag described in Section 2.1. Only 97 subhaloes have formed sufficiently early, at redshifts $z \geq 7.6$ . We consider them genuine metal-rich dwarf galaxies and will explore their relevant basic properties in the following.

Table 1.

Snapshot number (i.e. SnapNum) and redshift z at formation time of each subhalo in the sub-sample of metal-rich dwarfs. The total number of subhaloes (i.e. Count) is given per bin.

3.3 Basic properties of Metal-Rich dwarfs

One of the crucial conditions for habitability is the continuity of habitable-friendly conditions, the so-called ‘habitable time’ defined by Vukotic et al. (Reference Vukotic, Steinhauser, Martinez-Aviles, Cirkovic, Micic and Schindler2016). Close stellar encounters can disrupt this continuity if they are catastrophic enough to alter the orbits of planets located in circumstellar habitable zones. The rate of close encounters $\Gamma$ is calculated as $\Gamma = n_\star \langle v \rangle \langle \sigma \rangle$ , where $n_\star$ is stellar number density, $\langle v \rangle$ relative stellar velocity, and $\langle \sigma \rangle$ cross-section of the encounter having adverse consequences. Although the formula is quite simple, its application to numerical simulations is not entirely straightforward. This is mainly due to the stellar number density, which is not readily available and requires assumptions about the average stellar mass (as discussed by Mitrašinovic et al. Reference Mitrašinovic, Vukotic, Micic and Cirkovic2023). We can thus roughly estimate the average global rate of close encounters for each metal-rich dwarf galaxy if we make the appropriate assumptions. Alternatively, we can first check how compact these galaxies are. One solution would be to use one of the well-known and observationally determined criteria (e.g. Barro et al. Reference Barro2013; Damjanov et al. Reference Damjanov, Geller, Jabran Zahid and Hwang2015). However, these observationally-based criteria may be too restrictive, and an alternative approach would be more appropriate. This approach is based on the idea that compact galaxies are outliers on the mass-size relation (used by, e.g. Lohmann et al. Reference Lohmann, Schnorr-Müller, Trevisan, Ricci and Slodkowski Clerici2023), the relation which is well reproduced by IllustrisTNG (Genel et al. Reference Genel2018).

The mass-size relation in the TNG100 simulation box, along with its boundaries and metal-rich dwarf galaxies, is shown in Figure 5. With a few exceptions, most of the early-formed metal-rich dwarf galaxies are indeed compact. Such systems are densely populated, which in turn should give a high rate of close encounters $\Gamma$ . What is expected to affect and increase the rate of close encounters even further is the fact that these systems are metal-rich, making average stellar mass most likely sub-solar, which would result in higher stellar number density. Hence, it is likely that, in these galaxies, the rate of close encounters is too high.

Figure 5.

Mass-size relation in the TNG100 simulation box, determined considering all the galaxies in the stellar mass range $7 \leq \log (M_\star/\mathrm{M}_\odot) \leq 11$ . The expected boundaries of the relation are represented with black dashed lines, while early-formed metal-rich dwarf galaxies are marked with green diamonds.

Since most of the galaxies in our clean sub-sample of metal-rich dwarfs are compact, the next question is how these systems evolved to become compact. To answer this, we start by looking into the mass history of each galaxy in this clean sub-sample. In Figure 6, we show present-day total mass (thus, including the gaseous and dark matter components) as a percentage of the maximum total mass and the redshift when the galaxy has reached its maximum mass. By redshift $z=0$ , all of the galaxies in this sub-sample have retained very little of their maximum acquired mass. This suggests that the whole sub-sample is composed of compact, or almost compact, dwarf galaxies that have formed through the tidal stripping process, a highly plausible formation pathway, as demonstrated by numerous studies (e.g. Bekki et al. Reference Bekki, Couch, Drinkwater and Gregg2001; Reference Bekki, Couch, Drinkwater and Shioya2003; Pfeffer & Baumgardt Reference Pfeffer and Baumgardt2013; Pfeffer et al. Reference Pfeffer, Griffen, Baumgardt and Hilker2014; Martinovic & Micic Reference Martinovic and Micic2017; Ferré-Mateu et al. Reference Ferré-Mateu, Forbes, Romanowsky, Janz and Dixon2018; Kim et al. Reference Kim, Jeong, Rey, Lee, Lee, Joo and Kim2020; Deeley et al. Reference Deeley, Drinkwater, Sweet, Bekki, Couch and Forbes2023; Lohmann et al. Reference Lohmann, Schnorr-Müller, Trevisan, Ricci and Slodkowski Clerici2023; Moura et al. Reference Moura, Chies-Santos, Furlanetto, Zhu and Canossa-Gosteinski2024). The stripping formation pathway implies that the compact dwarf galaxies are tidally stripped remnants of much more massive galaxies processed by dense environments (e.g. galaxy clusters). Moreover, since the tidal stripping is an outside-in process (e.g. Diemand, Kuhlen, & Madau Reference Diemand, Kuhlen and Madau2007; Choi, Weinberg, & Katz Reference Choi, Weinberg and Katz2009), it primarily affects the extended dark matter halo component of galaxies, resulting in lower dark matter fractions in these galaxies in general. In the clean sub-sample, the median dark matter fraction is about $f_\mathrm{DM} \simeq 0.5$ , and as many as 92 out of 97 galaxies have dark matter fractions lower than the cosmic average of about $f_\mathrm{DM} \simeq 0.84$ (Planck Collaboration et al. Reference Planck2016). Hence, the fact that the galaxies in the clean sub-sample are outliers on the mass-metallicity relation can be interpreted from an evolutionary point of view. The galaxies were, most likely, not always outliers but moved out of the main sequence as they lost a portion of their mass while retaining generally higher metallicities. According to Paul et al. (Reference Paul, John, Gupta and Kumar2017), clusters of galaxies are groups more massive than $8\times 10^{13} \; \mathrm{M_\odot}$ . Following that definition and setting a lower mass threshold for a galaxy group of $5\times 10^{12} \; \mathrm{M_\odot}$ , we find that only 4 galaxies in our sample are located in clusters, 17 are members of groups, while the rest (majority of the sample, 76 galaxies) are satellites of more massive galaxies.

Figure 6.

History of the total mass of metal-rich dwarf galaxies, where the x-axis represents the redshift when the galaxy reached its maximum total mass, and the y-axis represents present-day total mass as a percentage of the maximum total mass. Histogram counts are given for each axis, with the corresponding probability density function.

The present-day stellar photometric data for the SDSS bands, particularly of interest g and r, are available in the supplementary catalogue (Nelson et al. Reference Nelson2015). The median colour of the clean sub-sample is quite blue $(g-r) \simeq 0.2$ , while 78 out of 97 galaxies can be classified as blue with $(g-r)\lt0.4$ , indicating a significant population of younger stars. Morphological information is also available in one of the supplementary catalogues (Genel et al. Reference Genel, Michael Fall, Hernquist, Vogelsberger, Snyder, Rodriguez-Gomez, Sijacki and Springel2015), allowing at least a rough classification corresponding to the Hubble sequence. However, for the TNG100 simulation box, the catalogue is restricted to galaxies that are sufficiently well resolved for a task, with stellar mass $M_\star \gt 3.4 \times 10^8 \; \mathrm{M}_\odot$ . The information on morphological decomposition is, thus, only available for 45 out of 97 galaxies in the clean sub-sample. Most of them can be classified as elliptical as the median fraction of the spherical component is high, $f_\mathrm{spherical} \simeq 0.74$ , while 6 of them can be classified as irregular due to a relatively high fraction of the stellar mass that is not accounted for (i.e. it is neither part of the spherical components nor a thin disc). We emphasise that most galaxies are elliptical, not necessarily perfectly spherical, since their flatness, calculated based on the provided eigenvalues of the mass tensor, ranges between $\sim 0.68$ and $\sim 1$ , where perfectly spherical galaxies have flatness equal to 1. This implies that various sub-classes of elliptical galaxies are present in the sub-sample. One of the galaxies, which has a sufficiently high thin disc fraction, $f_\mathrm{thin} \simeq 0.19$ , and the lowest flatness, can arguably be classified as lenticular, depending on the definition Figure 7 shows where the metal-rich dwarf galaxies are positioned in the stellar mass – SFR plane, concerning the observationally determined star-forming main sequence (Renzini & Peng Reference Renzini and Peng2015). The galaxies are colour-coded based on their baryonic fraction of gas, most of which lie within the main sequence (within the boundaries of $\Delta=\pm 0.5$ ). A fraction of mostly gas-rich galaxies are barely above, while the quenched galaxies scattered below the main sequence are typically gas-poor.

Figure 7.

Star-forming main sequence (Renzini & Peng Reference Renzini and Peng2015, black dashed line, where the dotted lines represent the boundaries $\Delta=\pm 0.5$ ), and the metal-rich dwarf galaxies in the stellar mass – SFR plane, colour-coded for the baryonic fraction of gas.

4.1 The complexity of galactic habitability

Galactic habitability is a multifaceted concept, shaped by both intrinsic properties of galaxies, essentially unknowable initial conditions at their formation, and their evolutionary history. Typically, the evolution of galaxies strongly depends on their immediate environment and external factors. Thus, galactic habitability depends on a number of properties that may bolster or hinder habitable conditions, and these effects are often strongly interrelated.

External factors often influence crucial internal properties of galaxies. This is the most evident in dense environments, such as galaxy clusters, where close interactions and mergers are the most frequent, disrupting the stability of stellar systems, causing starbursts, multiple reignitions of the nuclear engine, and other perturbing phenomena. However, even in a quiescent environment, stellar orbits in a Milky Way-like spiral galaxies are subject to perturbations within the broad stellar neighbourhood; and even stars with nearly circular and stable orbits are partly coming from populations that have radially migrated from other parts of the galactic disk (Mitrašinovic et al. Reference Mitrašinovic, Vukotic, Micic and Cirkovic2023). Tidal stripping and dynamical changes due to galactic flybys and interactions, which also occur in isolated environments, although less frequently, can produce a wealth of substructures (e.g. Kim et al. Reference Kim, Peirani, Kim, Ann, An and Yoon2014; Pettitt & Wadsley Reference Pettitt and Wadsley2018; Mitrašinovic & Micic Reference Mitrašinovic and Micic2023), making the habitability assessment much more complex than the original zonal habitability concept (Gonzalez, Brownlee, & Ward Reference Gonzalez, Brownlee and Ward2001). Similarly, strong tidal stripping (associated with dense environments, interactions, or massive neighbouring galaxies) can indicate a dynamically unstable environment, further reducing the potential for sustained habitability.

As mentioned previously, tidal stripping can lead to the formation of ultra-compact dwarf galaxies. In such densely packed stellar systems (i.e. compact dwarf galaxies in this case), close interactions between individual stars occur frequently, significantly impacting the potential habitability of planets. These interactions make it less likely for planets to maintain stable orbits within the habitable zones around their host stars. On the other hand, a high density of the stellar system offers potential for interplanetary transport of biological material (lithopanspermia, Wesson Reference Wesson2010), as argued by Balbi, Hami, & Kovacevic (Reference Balbi, Hami and Kovacevic2020) and interstellar panspermia in general (Vukotic, Seckbach, & Gordo 2021).

While proxies such as high stellar metallicity and low SFR have been historically used to infer favourable conditions for planet formation and long-term stability (e.g. Lineweaver Reference Lineweaver2001; Lineweaver, Fenner, & Gibson Reference Lineweaver, Fenner and Gibson2004; Dayal et al. Reference Dayal, Cockell, Rice and Mazumdar2015), such an interpretation is perhaps overly simplified. On the one hand, the high SFR could be considered a habitability risk factor: a higher SFR would imply a higher rate of supernova and gamma-ray burst explosions, which could be detrimental to the development and survival of life on a planet’s surface. Additionally, a high SFR coupled with generally low gas content might imply that the gas reservoir in these galaxies will soon be depleted, resulting in a smaller number of newly formed habitable systems. However, the effects of a high SFR could be positive: high-energy radiation, stemming from supernova explosions, while potentially dangerous, is highly unlikely to completely wipe out all life on the surface of a planet (Sloan, Alves Batista, & Loeb Reference Sloan, Alves Batista and Loeb2017); however, it could provide the necessary energy for complex prebiotic chemistry, or fuel photosynthesis (Lingam & Loeb Reference Lingam and Loeb2019), or even trigger genetic mutations in surface life that would allow more complex life forms to arise. Moreover, studies of mass extinction events on Earth (e.g. McElwain & Punyasena Reference McElwain and Punyasena2007; Alroy Reference Alroy2008; Filipovic et al. Reference Filipovic, Horner, Crawford, Tothill and White2013) have led to the conclusion that these catastrophic events can lead to even greater biodiversity and may be necessary for the development of complex life.

Ultimately, any claim about habitability advantages must account for this multi-dimensional complexity. However, we have to specify what we are considering when discussing habitability: the emergence of life and development of complex life forms, or their sustainability and persistence. At first glance, these concepts appear very similar, but they can be entirely different. Factors that are considered detrimental to life development and sustainability are sometimes the same ones that can lead to the formation of complex life forms and the transportation of biological material across planetary or stellar systems.

4.2 Re-evaluating the habitability of metal-rich dwarfs

While the original Cloudlet population appeared promising due to its high metallicity and relatively low SFR, our refined analysis reveals a more nuanced picture. In light of the discussion above, we revisit the potential habitability of galaxies identified in our final filtered sample. Genuine dwarf galaxies examined in this work are metal-rich, which could imply superior habitability conditions, and the effects of their star-formation rates suggest considerable habitable potential. However, these galaxies are overwhelmingly compact, exhibit signs of strong mass loss over time, and typically represent remnants of formerly more massive systems that experienced tidal stripping. In many cases, they lie near massive host galaxies, suggesting that environmental processes played a dominant role in shaping their present-day properties. Although such galaxies may, on average, show favourable values in habitability proxies such as high metallicity and low SFR, these alone do not guarantee suitability for life. The stellar encounter rate, which scales with stellar density and internal dynamics, is generally high for these compact systems. This points to potentially disruptive environments, where planetary orbits may be unstable on longer timescales, which may inhibit long-term habitability. The existence of high metallicity alone is not sufficient to conclude a habitable advantage, especially in dynamically violent environments.

As discussed previously, dynamically challenging environments are not necessarily inhospitable to life. Various factors that affect habitability are intertwined, but they can also have opposing effects individually. Further studies of metal-rich dwarf galaxies are required to closely examine these opposing effects and their joint effect on habitability. Unfortunately, this is beyond the scope and possibilities of the present work. Due to resolution limitations, the explored galaxies contain only a few hundred stellar particles at the very most, making robust orbital perturbation analysis unfeasible. Future work exploring this interesting avenue should be done using detailed analysis based on higher-resolution data or tailored zoom-in simulations.

For the sake of argument, we will assume that the sample of Cloudlet galaxies examined in this work is indeed superior in their habitable potential. Compared to the original Cloudlet population, which probably included many young or dynamically inconsistent objects, these galaxies represent a much smaller and more physically meaningful subset. Importantly, even if these systems offered favourable conditions, their relative abundance is extremely low. Only 97 early-forming, filtered metal-rich dwarf galaxies remain in our final sample out of tens of thousands of subhaloes. This suggests that, even under optimistic assumptions about planetary formation, the global contribution of such systems to cosmic habitability is negligible.

These results stand in contrast to the conclusions of the previous study (Stojkovic et al. Reference Stojkovic, Vukotic, Martinovic, Cirkovic and Micic2019), which suggested that our search for extraterrestrial life should not be limited to galaxies similar to the Milky Way, but should also include these high-metallicity dwarfs. Genuine dwarf galaxies, as they are conventionally understood and examined in the present study, constitute only a small fraction of the Cloudlet population. This could lead us to the conclusion that previous assumptions about these objects may have been overly optimistic. Instead, we find that the Cloudlet is not a distinct secondary peak in the galactic habitability landscape, but a sparse tail shaped by extreme evolutionary pathways. We therefore conclude that the bimodality of galactic habitability reported in earlier work does not persist when evaluated in a more physically consistent simulation framework. Although some systems may retain marginally habitable properties, they do not constitute a distinct or significant population.

4.3 Perspective beyond conventional hosts

In this study, we focused on clearly defined and dynamically coherent galaxies. Our results suggest that the previously identified Cloudlet population, when carefully filtered for reliability and formation history, does not represent a robust or dominant component of the habitable galaxy population. However, this naturally raises the question of what kinds of structures were excluded by our conservative criteria. In particular, the unfiltered Cloudlet of metal-rich low-mass objects is predominantly composed of structures that are not considered galaxies in a traditional sense. It is quite possible that a significant portion of this unfiltered population, much larger in numbers than our analysed sample, represents distinct substructures of larger, more massive galaxies. As such, it provides an additional strong argument for the complex habitability assessment that differs from the pioneering annular zone concept (Gonzalez, Brownlee, & Ward Reference Gonzalez, Brownlee and Ward2001). The complexity should be intuitively understandable, given that galaxies are not as centrally influenced (in terms of radiation and motion) as individual planetary systems and their host stars. More detailed models of galactic habitability, especially focusing on galaxy dynamics and taking into account different sources of high-energy radiation, coupled with exoplanet hunt observations in nearby Milky Way satellites and other members of our Local Group, are crucial to provide better insights in future research on this subject. The localised nature of habitable regions within galaxies is already accepted and commonly discussed qualitatively, although the traditional view of the GHZ as a simple annular region (Gonzalez, Brownlee, & Ward Reference Gonzalez, Brownlee and Ward2001; Lineweaver, Fenner, & Gibson Reference Lineweaver, Fenner and Gibson2004) remains a norm in quantifiable terms (e.g. when the study aims to constrain habitable regions of a galaxy). Thus, it should not be controversial that distinct substructures of larger galaxies may show more habitable-friendly conditions. In fact, this could be an additional argument that challenges the traditional model, urging us to re-examine our approach to large-scale habitability.

At the same time, the remaining Cloudlet population is composed of diverse objects, not just substructures of larger, more massive galaxies. Although some of these objects may be numerical artefacts, it is highly plausible that most of them are either dense, weakly bound tidal debris or previously mentioned, kinematically distinct substructures of larger galaxies. Both possibilities offer exciting and fresh insights, providing a new perspective on the holistic approach to habitability. Regarding tidal structures, their habitable potential was already speculated, although not thoroughly investigated, by Forgan et al. (Reference Forgan, Dayal, Cockell and Libeskind2017). It should be noted that all of the above-mentioned objects are dynamically and evolutionarily challenging, which could affect the habitable potential by influencing the long-term stability of orbits. However, as we discussed previously, it should not be considered detrimental to life as it might appear at first glance. There is clearly a balance between the long-term stability required for life to develop and thrive and the dynamic, maybe even violent, nature of an environment required for spreading biological material across the galaxy and over larger scales in general.

Finally, these structures merit further investigation, and a broader follow-up will be needed to explore whether these ambiguous objects and structures might nonetheless host conditions conducive to life. In follow-up work, we plan to analyse these ambiguous or transitional structures and objects in more detail, incorporating, perhaps, if possible, particle-level dynamics and environmental analysis. This will help us better understand their origin, nature, degree of perturbation, and survival. It might very well be that, instead of genuine metal-rich dwarf galaxies, these ambiguous or transient structures are the ones that the previous study identified as possible cradles of life. Even if they prove to be unstable environments for life to persist and develop, they could hold a key to spreading life across large scales.