{"id":"76c7e9f9-e202-471c-912b-6f042a05e612","arxiv_id":"2412.15334","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"Nuclear star clusters that hierarchically merge in cosmological merger trees can grow supermassive black hole seeds and produce a local population of intermediate-mass black holes.","lead":"This paper models how dense star clusters at the centers of small early galaxies grow black hole seeds and then merge over cosmic time into supermassive black holes. It predicts a large population of intermediate-mass black holes, along with gravitational wave and stellar disruption rates that future observatories could test.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Seed-formation efficiency depends on an unconstrained compact tail of the high-redshift NSC radius distribution; if real NSCs are systematically larger, predicted IMBH densities drop by an order of magnitude, undermining the central claim.","rationale":"The reader's weakest_assumption identifies exactly the same load-bearing concern: the essential, unverified compactness and ubiquity of high-redshift NSCs. The paper is transparent about this assumption, explicitly labelling it essential in Sec. 4.1, and the model's outputs are demonstrably sensitive to it: Fig. 3 shows that seed formation is confined to the low-rh, high-N region of parameter space, and the log-uniform prior dictates how many NSCs fall in that region. No internal contradiction is present; the concern is empirical rather than logical. The proposed test is concrete and feasible with existing JWST data and a simple re-run of the public Nuce code, and it would settle whether the compact-tail assumption actually supports the predicted IMBH density. Other sensitivity axes, such as the gas parameters (Sec. 4.2.4-4.2.5), affect SMBH growth but not the seed-formation efficiency itself, so the compact-radius prior is the more fundamental uncertainty. Thus the reader's CONDITIONAL verdict remains appropriate, and no change is recommended.","tokens_in":25914,"tokens_out":11089,"duration_ms":100224,"concrete_test":"Re-run the fiducial model with an observationally motivated NSC initial radius distribution, for example a log-normal centered at 3 pc with scatter 0.3 dex (matching local NSC effective radii from Neumayer et al. 2020), instead of the log-uniform 0.1-10 pc prior, keeping all other hyperparameters fixed. If the predicted z=0.25 IMBH number density drops below ~0.02 Mpc^-3, the compact-tail assumption is quantitatively load-bearing and the central claim is weakened. Complementary check: compile the JWST lensed cluster sample (Adamo et al. 2024; Vanzella et al. 2023) to measure the fraction of high-z NSCs with rh<0.2 pc and compare it to the 15% assumed by the prior; a measured fraction below a few percent would directly lower the seed-formation efficiency by an order of magnitude.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The model's seed production is dominated by the low-radii tail of the assumed log-uniform prior (Sec. 2.2). Runaway stellar collisions require 0.2 tau_rh,0 < 3 Myr, which for N~1e6 implies rh_0 <~0.1-0.2 pc; repeated BH mergers require an escape velocity >300 km/s, which for N~1e7 implies rh_0 <~0.2-0.5 pc. The fiducial prior over 0.1-10 pc places ~15% of NSCs below 0.2 pc and ~35% below 0.5 pc, so a substantial subpopulation forms seeds. However, the true high-redshift NSC radius distribution is unconstrained: local NSCs have effective radii of a few pc, and the few JWST-detected compact clusters are strongly magnified and likely biased toward high surface brightness. If the true distribution is centered at a few pc with a small compact tail, the seed-forming fraction drops by roughly an order of magnitude, and the predicted z=0.25 IMBH number density of 0.13 Mpc^-3 falls below the lower end (0.02 Mpc^-3) of the observationally allowed range (Greene et al. 2020). The paper itself flags this in Sec. 4.1: 'Compact NSCs on sub-pc scales could theoretically form... This assumption is essential for assembling SMBH seeds in the NSC paradigm.' Because this assumption is upstream of all subsequent growth, it is the most load-bearing element of the argument.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper presents a semi-analytic model, NSC-tree, that couples the Nuce nuclear star cluster evolution code to galaxy merger trees extracted from the NewHorizon simulation. The model assumes that each protogalaxy above a critical stellar mass forms a nuclear star cluster with specified initial mass, gas fraction, and a log-uniform initial half-mass radius prior, and that stellar-mass black holes in these clusters grow through runaway collisions, repeated black-hole mergers, gas accretion, and stellar consumption to form massive black hole seeds. These seeds then grow and merge hierarchically during major galaxy mergers, producing predictions for the local MBH-Mstar relation, black hole spins, occupation fractions, number densities of IMBHs and SMBHs, NSC properties, and rates of massive black hole mergers, stellar-mass black hole captures, extreme-mass-ratio inspirals, and tidal disruption events. The headline claim is that the seeds of supermassive black holes formed in nuclear star clusters via stellar black hole mergers at early epochs, and that this scenario can reproduce local scaling relations and yields a LISA-detectable massive black hole merger rate.","tokens_in":26281,"tokens_out":5421,"duration_ms":52749,"significance":"If the central scenario is correct, the paper offers a concrete pathway from ordinary stellar-mass black holes to SMBH seeds without invoking exotic massive seed formation, and it produces a rich set of falsifiable predictions: local IMBH number densities, occupation fractions, TDE rates, EMRI rates, and LISA-detectable merger rates. The manuscript is transparent about its simplifications, provides a public code, and makes an explicit connection to a high-resolution cosmological simulation via merger trees. However, the model's headline outputs are not fully independent predictions: the essential initial conditions for high-redshift NSCs are unconstrained, and the authors state in the Conclusions that many free parameters were chosen to give a crude fit to local scaling laws. These issues must be addressed before the claim that the NSC paradigm explains SMBH seeding can be regarded as robust.","major_comments":[{"comment":"The log-uniform prior on the initial half-mass radius, rh,0 in [0.1, 10] pc, is load-bearing but observationally unconstrained at high redshift. The seed-formation thresholds in §2.3.1 (0.2 tau_rh,0 < 3 Myr for runaway collisions and v_esc > 300 km/s for repeated BH mergers) imply that only the compact tail, roughly rh,0 < 0.2-0.5 pc for N ~ 1e6-1e7, contributes to seed formation. Under the fiducial prior, this tail contains roughly 15-35% of the clusters, but if the true high-redshift NSC radius distribution is centered at a few pc with only a small compact tail, the seed-forming fraction drops by about an order of magnitude, and the predicted z=0.25 IMBH density of 0.13 Mpc^-3 falls below the 0.02 Mpc^-3 lower end of the observationally allowed range from Greene et al. (2020). The authors themselves flag this in §4.1: 'Compact NSCs on sub-pc scales could theoretically form... This assumption is essential for assembling SMBH seeds in the NSC paradigm.' I request a quantitative sensitivity test that reweights or replaces the radius prior with a distribution informed by local NSC radii and by the few available high-z cluster observations, and that reports how the IMBH density, occupation fraction, and merger rates change.","section":"§5 and §4.2"},{"comment":"The Conclusions state explicitly that 'Our code has many free parameters, that we have chosen to give a crude fit to local scaling laws involving NSCs.' Consequently, headline outputs such as the MBH-Mstar slope in Eq. (1), the occupation fractions in Fig. 6, and the IMBH number density in Table 2 are partly calibrated rather than independent predictions. The hyperparameter tests in §4.2 show that the slope of the MBH-Mstar relation varies from about 0.35 (fg,max = 0.1) to about 1.13 (fstar,max = 0.1), so the fiducial slope of 1.10 is not robust to plausible parameter choices. Please specify which hyperparameters were tuned, against which observables and with what tolerance, and then separate calibrated statements from genuinely predicted quantities. A small grid or emulator-based exploration of the parameter region that still matches the calibration targets would greatly strengthen the paper.","section":"§2.2 and §3.1.3"},{"comment":"The assumption that every protogalaxy forms one NSC in a single burst when its stellar mass reaches Mcr = 1e7 Msun directly produces the near-unity NSC and IMBH occupation fractions at stellar masses below about 1e8 Msun. The authors acknowledge in §3.1.3 that 'Our predictions for high occupation fractions are impacted by our choice to form an NSC in every galaxy with stellar mass larger than 1e7 Msun.' Because the occupation fraction is a key comparison with Greene et al. (2020) and Nguyen et al. (2019), this assumption needs to be relaxed or bracketed, for example by introducing an NSC formation efficiency with a physically motivated dependence on galaxy mass or environment, and by modeling the ex-situ contribution from inspiraling globular clusters, which is currently neglected.","section":"§2.4 and §3.3.1"},{"comment":"The model ignores time delays between galaxy mergers, NSC mergers, and massive BH binary mergers. The estimate in §4.1, tau_har + tau_gw ~ 44 Myr, addresses only the hardening and gravitational-wave phase and explicitly drops the dynamical-friction delay tau_df. Since the predicted LISA merger-rate redshift distribution in Fig. 10 and the detectability statement in §3.3.1 depend on when massive BH binaries actually coalesce, the instantaneous-merger approximation could bias the z-distribution of events. A simple log-normal delay prescription, or at least a comparison of the 44 Myr estimate with the median inter-merger time in the NewHorizon trees, would show whether this simplification is benign for the paper's LISA predictions.","section":""}],"minor_comments":[{"comment":"The phrase 'strong HHEII emission' contains a typo and should read 'strong He II emission.'","section":"§1"},{"comment":"The sentence 'Most of these NSCs have half-mass radii larger than 50M⊙' should read 'larger than 50 pc,' consistent with the detectability threshold defined in §3.2.","section":"§3.2.1"},{"comment":"The word 'super-Eddigton' should be 'super-Eddington.'","section":"§4.1"},{"comment":"The description of the major/minor merger classification is confusing: 'if the asymmetry in the galaxy masses is smaller' appears to describe a minor merger, but the sentence then states the NSC becomes an ultra-compact dwarf. Please rephrase to make the threshold Qth and the two outcomes unambiguous.","section":"§2.4"}],"recommendation":"major_revision","confidential_remarks":"I agree with the stress-test assessment: the compact high-redshift NSC radius distribution is the most load-bearing assumption, and the paper itself labels it essential. The requested sensitivity analysis should be a condition of acceptance. The authors' transparency about their tuned parameters is commendable, but the conclusions currently overstate the predictive status of several headline results. No concerns about novelty or scope; the paper is a reasonable contribution to the semi-analytic SMBH-seeding literature once the robustness of the compact-tail assumption is quantified."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my take on Kritos et al. It's a transparent, code-backed semi-analytic model that makes a plausible case for a specific SMBH seeding channel: stellar-mass BHs in nuclear star clusters growing via mergers and accretion, then being assembled hierarchically. The genuinely new piece is coupling the Nuce single-cluster code to NewHorizon cosmological merger trees and generating population-level predictions: IMBH occupation fractions, number densities, a LISA SNR catalog, and transient rates. That synthesis hasn't been done before, and the code is public. I believe the authors on their own limitations; they list them clearly and don't hide the circularity.\n\nThe soft spots are real but not fatal. The most load-bearing assumption is the initial NSC radius distribution. The model produces seeds predominantly from the low-radius tail (roughly 0.1-0.5 pc) of the log-uniform prior, and that prior is essentially unconstrained at high redshift. Local NSCs are a few pc, and the JWST compact clusters are a biased, magnified sample. If the true high-z distribution has only a small compact tail, the seed-forming fraction drops by about an order of magnitude, and the predicted IMBH number density falls below the lower end of the observationally allowed range. The authors themselves flag this in Sec. 4.1. So the central claim should be read as a conditional plausibility argument, not a robust prediction. That's a fair reading, and the stress-test note gets it right.\n\nSecond, the model's free parameters are chosen to give a crude fit to local NSC scaling laws. That means the good agreement on some headline relations (occupation fractions, MBH-Mstar slope) is partly by construction. The MBH-Mstar relation isn't directly fitted, but the parameter choices affect it. Out-of-sample predictions, like the redshift evolution of the merger rate or the IMBH occupation in dwarfs below 1e9 Msun, would be more convincing. They do vary parameters one at a time, which helps.\n\nThird, the comparisons to observations are qualitative; they state that selection effects are omitted. Minor in context, but it limits the quantitative claims.\n\nI don't see a fatal flaw or internal contradiction. The paper is honest about what it can and cannot claim. This is a paper for people working on BH seed formation, LISA source predictions, and dwarf galaxy nuclei. It deserves a serious referee. My recommendation: send it to peer review, and ask the authors to add an exploration of the sensitivity to the NSC radius distribution, quantify uncertainties, and present at least one untuned out-of-sample prediction. That would move it from a plausible scenario to a testable one.","headline":"A transparent, code-backed plausibility argument for NSC-born SMBH seeds, with the main caveat being an unverified compact-radius assumption and some tuned parameters.","tokens_in":26832,"tokens_out":3249,"would_cite":true,"duration_ms":21259,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper claims that supermassive black hole seeds formed in compact nuclear star clusters at high redshift, and that growing them along galaxy merger histories reproduces observed black hole–galaxy scaling relations.","keywords":["supermassive black hole seeds","nuclear star clusters","intermediate-mass black holes","black hole mergers","galaxy merger trees","gravitational wave backgrounds","tidal disruption events","dwarf galaxies"],"falsifier":"A high-redshift census of star-forming galaxies at redshifts 6 to 10 that finds nuclear star clusters are rarely as compact as about one parsec would undercut the model's seed channel; the compact clusters JWST has already found would need to be representative rather than rare exceptions.","tokens_in":1753,"feed_emoji":"🕳️","tokens_out":2419,"duration_ms":87452,"temperature":0.7,"pith_summary":"This paper argues that the seeds of supermassive black holes did not need exotic massive objects: they could have grown from ordinary stellar-mass black holes in nuclear star clusters, the dense stellar systems found at galaxy centers. The authors couple a semianalytic model of nuclear cluster evolution to realistic galaxy merger trees and show that runaway collisions, repeated black hole mergers, gas accretion, and star consumption can grow ten-solar-mass black holes into seeds of thousands to hundreds of thousands of solar masses within a gigayear. What makes the claim worth attention is that this ordinary-seed scenario reproduces observed black hole–galaxy scaling relations and predicts testable populations: intermediate-mass black holes in dwarf galaxies, ejected and wandering black holes, tidal disruption events, and a gravitational-wave merger rate accessible to space-based observatories.","feed_headline":"Merging star clusters can grow supermassive black hole seeds","feed_subtitle":"Run through realistic galaxy merger histories, the model matches local scaling laws and predicts a LISA-visible merger rate.","key_machinery":"The carrying object is Nuce, a semianalytic code for nuclear cluster evolution that relies on Hénon's principle, the balanced evolution by which energy generation in the cluster core drives gradual expansion. Nuce evolves cluster structure and the central black hole through four growth channels: runaway stellar collisions, Bondi gas accretion capped at the Eddington limit, repeated mergers of stellar-mass black holes in a core-collapsed subsystem, and loss-cone consumption of stars. Coupled to galaxy merger trees from the NewHorizon simulation, the code decides which nuclear star clusters coalesce after major galaxy mergers and which black hole binaries merge, using published fitting formulae for remnant mass, spin, and gravitational-wave recoil; whether the merger remnant stays in the nucleus is set by comparing the kick velocity to the cluster escape velocity.","core_discovery":"At the paper's center is the claim that galactic nuclei assemble their supermassive black holes hierarchically from light seeds. Each protogalaxy above a critical stellar mass forms a nuclear star cluster; inside compact clusters, stellar-mass black holes formed from massive stars undergo repeated mergers, with an escape-velocity threshold of roughly 300 kilometers per second, and are boosted by gas accretion to produce seeds of intermediate mass. When the seeding model is evolved along galaxy merger trees, the local population contains about 0.13 intermediate-mass black holes per cubic megaparsec, an occupation fraction near 80 percent in dwarf galaxies below one billion solar masses, and a black hole mass versus galaxy stellar mass relation with slope around 1.1. Supermassive black holes grow further during major galaxy mergers, those that grow by coherent gas accretion emerge with near-maximal spins, and gravitational-wave recoil occasionally ejects them into off-nuclear wandering or isolated states.","pith_inferences":["A test the paper does not run is to compare its predicted near-maximal spins for supermassive black holes against X-ray reflection measurements with selection effects accounted for; if accretion is chaotic rather than coherent, spins would be lower and the growth channel would need revision.","Because the simulated volume is small and the merger trees stop at redshift 0.25, the model says little about the rare, highly luminous quasars seen at redshift above 6; applying the same seeding prescription to larger-volume trees would show whether an additional massive-seed channel is needed.","The predicted black hole occupation fraction of about 80 percent in dwarf galaxies is effectively an upper bound under the assumption that every galaxy above the mass threshold forms a nuclear cluster; patchy cluster formation would lower the implied number densities proportionally."],"forward_implications":["If the scenario is right, supermassive black hole seeds can form from ordinary stellar-mass black holes, removing the need for massive Population III or primordial seeds.","The local universe should contain roughly 0.13 intermediate-mass black holes per cubic megaparsec, about half nuclear and half satellite, concentrated in dwarf galaxies.","Space-based gravitational-wave observatories should see a massive black hole merger rate of about 5.3 events per year out to redshift 5, dominated by intermediate-mass black hole binaries.","Nuclear star clusters would naturally produce a high tidal disruption event rate density of 10,000 to 100,000 per year per cubic gigaparsec at redshifts below 5.","Nuclear supermassive black holes grown by coherent gas accretion should appear highly spinning, while intermediate-mass black holes should show a bimodal spin distribution depending on formation channel."],"supporting_citations":[{"why":"Supplies the Nuce code and semianalytic nuclear cluster evolution model on which all seed growth is based.","marker":"Kritos et al. (2024a)"},{"why":"Provides the NewHorizon simulation and merger trees that drive galaxy and nuclear cluster assembly histories.","marker":"Dubois et al. (2021)"},{"why":"Sets the observational scaling relations, occupation fraction limits, and mass function lower limits the model is compared against.","marker":"Greene et al. (2020)"},{"why":"Establishes the roughly 300 kilometers per second escape-velocity threshold for massive black hole formation through repeated mergers.","marker":"Antonini et al. (2019)"},{"why":"Supports the runaway stellar collision channel that forms very massive stars and their direct-collapse seed black holes.","marker":"Portegies Zwart & McMillan (2002)"},{"why":"Motivates the assumption that protogalactic nuclear star clusters can form with half-mass radii below half a parsec.","marker":"Devecchi et al. (2010)"},{"why":"Documents compact parsec-scale young massive clusters at redshift around 10, used to justify sub-parsec initial cluster radii.","marker":"Adamo et al. (2024)"},{"why":"Supplies the fitting formulae for remnant mass, spin, and gravitational-wave recoil that decide whether merged black holes are retained or ejected.","marker":"Gerosa & Kesden (2016)"}],"fun_headline_variants":["Star cluster mergers seed supermassive black holes","Hierarchical merging grows black hole seeds in clusters","Nuclear star clusters spawn galaxy giants","Cosmic mergers birth black hole seeds in clusters","Merging clusters forge supermassive black hole seeds"],"cache_read_input_tokens":28800,"weakest_assumption_plain":"The model assumes that almost every small protogalaxy above about ten million solar masses formed a dense nuclear star cluster at high redshift with sub-parsec scale sizes, and that such compact clusters are common enough; if they are not, the seed-assembly engine loses its fuel.","fun_headline_variants_meta":{"raw":{"variants":["Star cluster mergers seed supermassive black holes","Hierarchical merging grows black hole seeds in clusters","Nuclear star clusters spawn galaxy giants","Cosmic mergers birth black hole seeds in clusters","Merging clusters forge supermassive black hole seeds"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00052,"raw_usage":{"total_tokens":2476,"prompt_tokens":858,"completion_tokens":1618,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":474,"completion_tokens_details":{"reasoning_tokens":1548}},"tokens_in":474,"tokens_out":1618,"duration_ms":11616,"temperature":1.0,"reasoning_tokens":1548,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T11:31:24.772003+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A high-redshift census of star-forming galaxies at redshifts 6 to 10 that finds nuclear star clusters are rarely as compact as about one parsec would undercut the model's seed channel; the compact clusters JWST has already found would need to be representative rather than rare exceptions.","supporting_citations":[],"review_version":1}