{"id":"011a0db7-e166-4cf3-bc51-47b25be154f7","arxiv_id":"2411.19326","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"Radio-detected AGN in the LoTSS Deep Fields show a higher [O III] outflow detection rate (67.2±3.4%) than matched radio non-detected AGN (44.6±2.7%).","lead":"Using deep LOFAR 144 MHz images and SDSS spectra, this study compares ionized gas outflows in 198 AGN that are either detected or not detected in radio emission. Radio-detected AGN show outflow signatures about 23 percentage points more often than non-detected AGN, supporting a link between radio emission and outflows.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Matched comparison lacks SFR control for the radio non-detected AGN, so the 22.6-point outflow-rate gap may partly reflect host star formation rather than a direct radio-AGN connection.","rationale":"The reader's weakest assumption identifies the most load-bearing gap in the argument: the radio detected and radio non-detected samples are matched only in L6um and redshift, not in star formation rate or other host properties that could drive both radio emission and [O III] kinematics. This is directly stated in the manuscript ('we do not have access for reliable SFR measurements for these AGN'), and it affects the central quantitative claim, not a secondary detail. The proposed SED-fitting test is feasible with existing deep-field photometry and would settle whether the 22.6 percentage point outflow-rate difference persists after controlling for SFR. I agree with the reader's conditional verdict: the paper presents a statistically strong but not fully controlled comparison, and the missing SFR baseline justifies conditionality rather than full acceptance. Other concerns, such as the non-codified visual inspection of spectral fits and the abstract's slightly overstated stacked broad-component result, are real but secondary; the SFR confound is the single issue most capable of changing the headline conclusion.","tokens_in":23170,"tokens_out":8313,"duration_ms":76340,"concrete_test":"Obtain SFRs for the 83 radio non-detected AGN by running SED fits on the deep-field UV-to-FIR photometry in Kondapally et al. (2021) using the same AGN-plus-SFR decomposition employed by Best et al. (2023), then repeat the 1000-iteration L6um-z matching with an additional matching tolerance in log SFR (e.g., 0.3 dex) and recompute the outflow detection rates. If the radio-detected minus radio-non-detected rate remains at least 3 sigma after SFR matching, the confound is ruled out; if the difference drops below 2 sigma, the central claim is not robust to host star formation.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim rests on the L6um-redshift matched comparison described in Section 2.4. This matching controls AGN luminosity and redshift, but not star formation rate. The paper explicitly states in Section 2.1.1 and the Figure 9 caption that reliable SFRs are only available for the radio-detected AGN (from Best et al. 2023), and that no reliable SFR measurements are available for the radio non-detected AGN. The radio-excess diagnostic in Section 4.4.1 is computed only for the detected population, so it cannot establish that the two populations have similar SFR distributions. If radio-detected AGN are on average more star-forming or more gas-rich, they could exhibit both stronger 144 MHz emission and broader or more disturbed [O III] kinematics, raising the outflow classification rate without a direct causal link to radio jets or AGN winds. The headline result is the 67.2% versus 44.6% outflow detection rate (a 22.6 percentage point gap, quoted at >3 sigma). If that gap shrinks or disappears after matching in SFR, the interpretation as a connection between ionised outflows and radio emission becomes a host-property correlation rather than an AGN-driven effect. The paper's statement in Section 5.1 that star formation is 'unlikely' to be the driver is an interpretive argument, not a statistical control on the main matched comparison.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper selects 198 AGN in the LoTSS Deep Fields with SDSS spectroscopy, splits them into 115 radio-detected and 83 radio-non-detected sources at 144 MHz, fits the [O III] λ5007 and Hβ profiles with MCMC, and classifies outflows using either a second Gaussian component or W80 thresholds from Harrison et al. (2014). After matching the two populations in L6μm and redshift over 1000 random realizations, the authors report that radio-detected AGN show a higher total outflow detection rate (67.2±3.4%) than radio-non-detected AGN (44.6±2.7%), and that the stacked, narrow-component-normalized spectra of radio-detected AGN have a slightly brighter broad component. They interpret the result as evidence for a connection between ionized outflows and low-frequency radio emission, possibly from low-powered jets or AGN-wind shocks, while noting that most sources are not radio excess and hence not dominated by powerful jets.","tokens_in":23457,"tokens_out":4986,"duration_ms":43308,"significance":"If the central result holds, the paper gives one of the first statistical demonstrations at 144 MHz that AGN with low-frequency radio detections show higher [O III] outflow rates after matching in AGN luminosity and redshift, extending earlier work to low radio luminosities. The 1000-iteration matching bootstrap, the use of the public QubeSpec fitting module, the transparent outflow definitions, and the comparison with previous literature are all strengths. However, the headline interpretation that radio presence is directly connected to ionized outflows is weakened by the lack of SFR control for the radio-non-detected population and by a partly subjective model-selection step; the result is more securely described as a correlation between radio presence and outflow classification that survives L6μm-z matching.","major_comments":[{"comment":"The headline comparison controls for L6μm and redshift but not for star formation rate. Section 2.1.1 states that SFRs from Best et al. (2023) are available only for radio-detected AGN, and the Figure 9 caption explicitly says no reliable SFRs are available for the radio non-detected AGN. The matched 67.2% vs 44.6% outflow rate gap (Section 4.1) could therefore be partly driven by higher SFR or gas content among radio-detected sources, producing both 144 MHz emission and broader [O III] kinematics. The argument in Section 5.1 that star formation is 'unlikely' is a physical argument, not a statistical control. Please either obtain SFR estimates (even with upper limits) for the non-detected population, match in SFR or stellar mass in addition to L6μm and redshift, or present a sensitivity analysis showing how large an SFR difference would be needed to erase the outflow-rate gap.","section":"§2.4, §4.4.1, §5.1"},{"comment":"The matching procedure's effective sample size is unclear. In the 'closest matched population' of Figure 1, only 59 of the 83 radio non-detected AGN are matched (118 total), yet Section 4.1 reports outflow detection rates averaged over 1000 random matching runs without stating how many sources are matched in each run. If the number and composition of matched subsets vary, the reported rates mix different populations and the uncertainties from bootstrapping over runs may not reflect the true sampling variance. Please report the distribution of matched sample sizes across the 1000 runs, state how outflow rates are computed when a non-detected AGN has no matched counterpart, and verify that the matched subset is representative of the full non-detected sample.","section":"§2.4, §4.1"},{"comment":"The outflow classification depends on a model-selection step that is partly subjective: the authors state that the lowest-BIC fit was overridden by visual inspection for 'several spectra' in order to prioritize the [O III] fit. Since the central result is a difference in outflow detection rates, the number and nature of these overrides should be quantified, and the visual-inspection criteria should be specified more concretely (e.g., residual thresholds, priority rules). Without this, readers cannot assess whether the classification is reproducible or whether the BIC overrides could systematically favor radio-detected AGN.","section":"§3.1"}],"minor_comments":[{"comment":"Typo: 'catalouges' should be 'catalogues'.","section":"§2.2"},{"comment":"The sentence 'Above these lines, sources are considered to be radio excess, where their radio emission is dominated by star formation' is internally inconsistent with the next paragraph, which says radio emission in sources above the radio excess line can be attributed to radio jets; please clarify the definition of radio excess.","section":"§4.4.1"},{"comment":"The 'best fit with a slope of 0.58' is not described: please state the fitting method, the quantity being fit, and the uncertainty on the slope.","section":"§4.3"},{"comment":"Typo: 'Zakamska & Greene (2014) suggests that utflows trigger shocks' should read 'outflows'.","section":"§5.1"},{"comment":"It would help to state explicitly how many of the eight resolved sources show jet-like morphology, since only two are shown in Figure 10.","section":"§4.4.2"}],"recommendation":"major_revision","confidential_remarks":"This is a solid observational study, but the central causal claim is somewhat stronger than the controls support. The missing SFR control for the radio-non-detected population is the main blocker; if the authors can provide any SFR constraints for that population or a convincing sensitivity analysis, and also clarify the matched sample size and the BIC-override step, I would be happy to see it accepted after revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The new result here is a matched L6μm-redshift comparison of [O III] outflow detection rates between LoTSS Deep Field radio-detected and radio non-detected AGN at 144 MHz. That specific comparison hasn't been done before at these low radio luminosities and redshifts, and it passes the first test: radio-detected AGN show a higher outflow rate (67.2±3.4% vs 44.6±2.7%), and the gap survives when you restrict to definite outflows (55.3±3.2% vs 34.3±2.3%). The matching is done with 1000 random realizations and bootstrap uncertainties, and the 2D KS test on the matched samples gives p=0.61, so the luminosity control is solid.\n\nThe paper does several things well. The sample selection is carefully described, including the removal of FIRST-flag quasars and two sources below the Boötes flux limit. The outflow classification follows Harrison et al. thresholds, and the authors are transparent about the BIC being overridden by visual inspection—though they don't codify that override, which is a soft spot. They also compare to Mullaney et al. and Calistro Rivera et al. and are honest that the majority of their radio sources are unresolved at 6\", so they cannot pin down the radio emission origin.\n\nThe main soft spot is the SFR control. The radio-excess diagnostic is only computed for the radio-detected sources; the paper states in the Figure 9 caption that reliable SFRs are not available for the radio non-detected AGN. So the matched comparison cannot rule out that radio-detected AGN are on average more star-forming or more gas-rich, and that both the 144 MHz emission and the broader [O III] kinematics trace the same host property rather than a direct radio-AGN connection. The paper argues star formation is unlikely, but that's an interpretive argument, not a statistical control. This is the load-bearing caveat.\n\nA second, smaller issue: the abstract says the stacked broad component is 'enhanced' in radio-detected AGN, but the body states the area difference is not significant (the confidence intervals overlap by ~5 km2 s-2). The abstract overstates that point.\n\nThe central correlation likely holds—the outflow rate difference is big and robust—but the physical interpretation as AGN-driven winds/jets rather than host star formation is not yet settled. That's a fair conditional-acceptance situation.\n\nWorth sending to a serious referee. The authors should add an SFR baseline or explicitly caveat the interpretation, fix the abstract, and ideally share the fitting code and per-source products.","headline":"A solid, carefully matched comparison showing radio-detected AGN have higher [O III] outflow rates, though the missing SFR control for the radio-undetected sample leaves the physical interpretation open.","tokens_in":24078,"tokens_out":2895,"would_cite":true,"duration_ms":24282,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Radio-detected AGN host ionised-gas outflows more often than matched radio-undetected AGN, after removing AGN luminosity as a driver.","keywords":["galaxies: active","quasars: emission lines","galaxies: kinematics and dynamics","galaxies: jets","radio continuum: galaxies","ISM: jets and outflows","AGN feedback","LOFAR"],"falsifier":"Give the 83 radio non-detected AGN reliable SED-based star-formation rates (or deep far-infrared limits) and re-match the two populations in SFR as well as $L_{\\mathrm{6\\mu m}}$ and redshift; if the outflow-detection gap of roughly 23 percentage points disappears or drops below significance, the claimed connection between radio emission and outflows would be shown to be a byproduct of star formation. Alternatively, sub-arcsecond radio imaging that resolves the radio emission into regions coincident with star-forming clumps rather than AGN-related structures would directly falsify the low-powered-jet/wind-shock interpretation.","tokens_in":22962,"feed_emoji":"📡","tokens_out":9391,"duration_ms":72307,"temperature":0.7,"pith_summary":"This paper asks whether the radio emission seen in many active galactic nuclei is physically connected to the ionised-gas outflows that can reshape a host galaxy. The authors examine 198 AGN with SDSS spectra in the LOFAR Two-metre Sky Survey Deep Fields, 115 of them detected at 144 MHz, and match the radio-detected and radio-undetected populations in mid-infrared luminosity ($L_{\\mathrm{6\\mu m}}$) and redshift to remove AGN luminosity biases. Using two-Gaussian fits to the [O iii] $\\lambda$5007 line plus the non-parametric line-width measure $W_{80}$, they find that 67.2$\\pm$3.4 per cent of radio-detected AGN show signs of an outflow, versus 44.6$\\pm$2.7 per cent of radio-undetected AGN. They also report that, in stacked spectra normalised by the narrow [O iii] component, the broad outflowing component is brighter in the radio-detected population. If correct, this establishes a statistical link between low-frequency radio emission and ionised outflows that does not reduce to AGN luminosity, and points toward low-powered jets or AGN-wind shocks as the common driver.","feed_headline":"Radio-detected AGN host outflows 67% of the time vs 45%","feed_subtitle":"Matching for AGN luminosity and redshift, the gap points to jets or wind shocks rather than star formation.","key_machinery":"The analysis rests on three linked tools. First, the [O iii] $\\lambda$5007 emission line is fitted simultaneously with H$\\beta$ using one or two Gaussian components (with Fe II templates when a broad H$\\beta$ is present), and outflows are identified either by a second, blueshifted Gaussian or by the non-parametric width $W_{80}$ (the velocity range containing 80 per cent of the line flux), using thresholds of 600 and 800 km s$^{-1}$ to define 'likely' and 'definite' single-component outflows. Second, the radio-detected and radio non-detected populations are matched in $L_{\\mathrm{6\\mu m}}$ and redshift with a tolerance of $\\Delta\\log L_{\\mathrm{6\\mu m}}=0.3$ and $\\Delta z=0.06$, repeated 1000 times to bootstrap the outflow rates; $L_{\\mathrm{6\\mu m}}$ serves as an orientation-free proxy for AGN luminosity. Third, median stacked spectra are normalised by the continuum and by the peak of the narrow [O iii] component, so that the relative strength of the broad outflow component can be compared between the two populations. The pairing of the spectral decomposition with the luminosity matching is what allows the authors to attribute differences in outflow incidence to the presence of radio emission rather than to AGN luminosity.","core_discovery":"The central claim is that, after matching radio-detected and radio non-detected AGN in $L_{\\mathrm{6\\mu m}}$ and redshift, the radio-detected AGN have a higher rate of [O iii] outflow classification (67.2$\\pm$3.4 per cent across all outflow categories) than the radio non-detected AGN (44.6$\\pm$2.7 per cent), with the difference driven mainly by AGN in which a second, blueshifted Gaussian component is required by the spectral fit (52.3$\\pm$2.8 per cent versus 34.3$\\pm$2.3 per cent). The authors further show that in median-stacked spectra normalised by the peak of the narrow [O iii] component, the area of the broad outflow component is larger for radio-detected AGN, which they interpret as a sign that these AGN host outflows with more gas. Because the majority of the radio-detected sources fall below the radio-excess threshold defined by the star-formation–radio luminosity relation, the radio emission is not attributable to powerful jets; instead the authors argue the combination of results is consistent with low-powered jets or with radio emission generated by shocks within AGN-driven winds.","pith_inferences":["Because reliable star-formation rates are only available for the radio-detected half of the sample, the cleanest test of the paper's interpretation is to obtain SED-based SFRs (or deep far-infrared upper limits) for the radio non-detected AGN and re-match in SFR; if the gap in outflow rates collapses, star formation rather than radio emission would be the true driver.","The normalised-stack result predicts a quantitative relation: if outflows and radio emission share a physical origin, then in a larger matched sample the area of the broad [O iii] component should correlate with 144 MHz luminosity even after controlling for $L_{\\mathrm{6\\mu m}}$ and redshift; this is directly testable with the forthcoming sub-arcsecond LOFAR imaging of the same fields.","An alternative interpretation the authors do not fully explore is that the radio-detected AGN may be at a different evolutionary stage (e.g., recently triggered), where both the outflow and the radio emission are enhanced; testing this would require comparing outflow rates at fixed Eddington ratio or black-hole mass.","The threshold-based outflow definition means the gap could in part reflect a shift in the narrow-component FWHM rather than in the broad component; the authors show radio-detected AGN have broader narrow components, so a re-analysis with a continuous outflow metric (e.g., flux fraction above a velocity threshold) could confirm whether the difference is genuinely in the outflowing gas."],"forward_implications":["If the radio–outflow connection is real, deep low-frequency radio surveys such as LoTSS can be used to identify AGN that are likely driving ionised outflows, even without high-resolution radio imaging.","The outflow-rate gap implies that a substantial fraction of so-called radio-quiet AGN still produce synchrotron emission connected to their winds, so 'radio-quiet' does not mean 'radio-inactive' at 144 MHz.","The combination of a higher outflow detection rate with mostly unresolved radio morphologies strengthens the case for compact, low-powered jets or wind shocks as the origin of radio emission in the majority of AGN, rather than large-scale jets or star formation.","The paper's consistency check with the Mullaney et al. (2013) flux-weighted FWHM method indicates that the same trend holds when the sample is restricted to their luminosity and redshift range, extending their result to deeper radio luminosities.","A direct corollary is that AGN feedback via outflows may produce observable synchrotron radiation, so radio observations can trace the same gas that carries kinetic energy out of the galaxy."],"supporting_citations":[{"why":"SDSS DR16 quasar catalogue that supplies the higher-redshift half of the parent AGN sample.","marker":"Lyke et al. (2020)"},{"why":"Broad-line AGN catalogue that supplies the low-redshift AGN with SDSS DR7 spectra.","marker":"Liu et al. (2019)"},{"why":"Multi-wavelength catalogues for the LoTSS Deep Fields used for cross-matching, MIR fluxes, and radio information.","marker":"Kondapally et al. (2021)"},{"why":"Large statistical study of [O iii] kinematics versus radio luminosity that this work extends and checks consistency against.","marker":"Mullaney et al. (2013)"},{"why":"Source of the W80 outflow classification thresholds (600 and 800 km/s).","marker":"Harrison et al. (2014)"},{"why":"Original [O iii]-radio luminosity correlation that the paper re-derives at 144 MHz.","marker":"Rawlings et al. (1989)"},{"why":"Proposed wind-shock mechanism for radio emission in outflows, the main interpretive alternative discussed.","marker":"Zakamska & Greene (2014)"},{"why":"SED-fitting catalogue that gives star-formation rates and stellar masses for the radio-detected sources, used for the radio excess analysis.","marker":"Best et al. (2023)"},{"why":"LoTSS Deep Fields data release providing the 144 MHz images and catalogues.","marker":"Tasse et al. (2021)"}],"fun_headline_variants":["Radio-bright AGN show outflows 67% vs 45% in matched sample","AGN with radio emission host more ionized outflows, study finds","Outflows more common in radio-detected AGN: 67% vs 45%","Radio detection predicts higher AGN outflow rates after matching","Low-frequency radio signals correlate with AGN outflows, new data"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Everything depends on the assumption that the higher outflow rate in radio-detected AGN is caused by the radio emission itself, rather than by higher star formation or other host-galaxy properties, because reliable star-formation rates are only available for the radio-detected half of the sample and the $L_{\\mathrm{6\\mu m}}$–redshift matching cannot control for SFR.","fun_headline_variants_meta":{"raw":{"variants":["Radio-bright AGN show outflows 67% vs 45% in matched sample","AGN with radio emission host more ionized outflows, study finds","Outflows more common in radio-detected AGN: 67% vs 45%","Radio detection predicts higher AGN outflow rates after matching","Low-frequency radio signals correlate with AGN outflows, new data"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000489,"raw_usage":{"total_tokens":2505,"prompt_tokens":1144,"completion_tokens":1361,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":760,"completion_tokens_details":{"reasoning_tokens":1262}},"tokens_in":760,"tokens_out":1361,"duration_ms":9241,"temperature":1.0,"reasoning_tokens":1262,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T10:17:21.834584+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Give the 83 radio non-detected AGN reliable SED-based star-formation rates (or deep far-infrared limits) and re-match the two populations in SFR as well as $L_{\\mathrm{6\\mu m}}$ and redshift; if the outflow-detection gap of roughly 23 percentage points disappears or drops below significance, the claimed connection between radio emission and outflows would be shown to be a byproduct of star formation. Alternatively, sub-arcsecond radio imaging that resolves the radio emission into regions coincident with star-forming clumps rather than AGN-related structures would directly falsify the low-powered-jet/wind-shock interpretation.","supporting_citations":[{"cited_title":"A., Mackay C","cited_arxiv_id":null,"evidence_quote":"Original [O iii]-radio luminosity correlation that the paper re-derives at 144 MHz."}],"review_version":1}