{"id":"2a0f4452-dfbc-4654-ac49-b209bdcee178","arxiv_id":"2507.04066","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"M dwarf stars with high metallicity are the only ones in this sample found with large planets, and very short-period planets orbit more metal-rich M dwarfs, with a transition near 4.3 days.","lead":"This paper measures the metal content of 48 M dwarf stars that host planets and compares those measurements to the sizes and orbits of 246 planets. The results suggest that bigger planets and very close-in planets are found around more metal-rich M dwarfs, a pattern that may reflect how and where planets form around low-mass stars.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 4.3-day period–metallicity threshold is selected as the minimum of an uncorrected K-S p-value scan; the reported 99% confidence is not statistically valid.","rationale":"The reader's verdict identified sampling representativeness as the weakest assumption. That concern is real but hard to adjudicate without constructing the full APOGEE/TESS selection function. The sharper, internally testable flaw is the statistical procedure behind the 4.3-day threshold. The paper's own text shows the threshold was chosen as the minimum p-value of a scan, and the reported confidence ignores that selection. This is not a disagreement with consensus; it is a correctness risk in the paper's own significance calculations. The radius-metallicity result (large planets only around [M/H]>=0) is less affected by the scan issue and is supported by consistency with FGK studies and by the benchmark-star metallicity scale, which is independently checked. The multiplicity difference (singles vs multis) also relies on K-S p=0.002 from planet-level data, but the main novel quantitative claim—the 4.3-day transition and its physical explanation—rests on the invalid scan p-value. Therefore the paper should be accepted only conditionally on a corrected, multiplicity-aware multiple-testing analysis; if the corrected significance fails, the period-threshold conclusion should be removed. This agrees in part with the reader, who noted the uncorrected scan and clustering in the rationale but placed sampling first.","tokens_in":19100,"tokens_out":5351,"duration_ms":56918,"concrete_test":"Recompute the threshold scan using a multiplicity-aware permutation test: for each of ~1000 threshold values, compute the K-S statistic on [M/H] between the two period groups; then permute host-star [M/H] values across systems (keeping each system's planets together) to obtain the null distribution of the minimum p-value over all thresholds. The observed minimum of 6.4e-5 is significant only if it falls below the 5th percentile of that null. Also repeat the scan using only the largest planet per system. If the corrected global p-value is above 0.05, the 4.3-day transition and the associated dust-sublimation interpretation should be withdrawn or reframed as an upper limit.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In Section 3.2 (Figure 6, top panel), the authors compute a K-S p-value for every period threshold in 0.1-day steps, comparing the [M/H] distributions of 'short-period' and 'long-period' planets, and then quote the minimum p-value, 6.4e-5 at P_orb=4.3 days, as a 99%-confidence detection of a transition. This is a classic look-elsewhere effect: the quoted value is the minimum over many correlated tests, and no multiple-testing correction or permutation calibration is applied. In addition, the tests treat planets as independent observations even though multiple planets from the same host share one [M/H] value; with 65 planets around 45 stars, correlated points inflate the apparent significance. Because the 4.3-day threshold and its dust-sublimation interpretation are headline conclusions, the current analysis does not establish them. A corrected global p-value could easily exceed 0.05, in which case the transition period is an artifact of the scanning procedure.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper derives stellar metallicities ([M/H]) and oxygen abundances ([O/H]) for 48 M dwarf planet hosts using LTE spectrum synthesis of APOGEE near-infrared spectra, and analyzes exoplanet radii and orbital periods for a larger sample of 246 planets around 188 M dwarfs from Kepler, K2, and TESS. The main findings are that small planets (<3 R_Earth) orbit stars spanning a wide metallicity range (-0.6<[M/H]<+0.3), while larger planets in the sample are found only around stars with [M/H]>=0.0; that short-period planets (P_orb<4.3 days) orbit more metal-rich stars than long-period planets; that large planets (R_p>4 R_Earth) pile up at orbital periods near 2-5 days; and that multi-planet systems are more metal-poor and host only smaller planets. The paper also places the sample in Galactic context using the [O/M]-[M/H] relation.","tokens_in":19302,"tokens_out":4694,"duration_ms":52329,"significance":"If the trends are robust, the paper offers valuable evidence that planet formation around M dwarfs has metallicity-dependent channels that are analogous to, but quantitatively different from, FGK stars, particularly the shorter period threshold of 4.3 days. The metallicity scale is a strength: it is validated against binary benchmark M dwarfs with FGK companions (median offset +0.01+-0.04 dex) and against external FGK samples, and the radius-metallicity trend is qualitatively visible in Figure 5 and consistent with previous literature. The paper also benefits from explicit comparisons with FGK host populations. However, the statistical treatment of the period threshold currently overstates the significance, and the lack of a selection-function analysis weakens the strong 'only around metal-rich stars' claim.","major_comments":[{"comment":"The minimum K-S p-value of 6.4e-5 at P_orb=4.3 days is quoted as a 99% confidence detection of a transition period, but this value is the minimum of a scan over thresholds in 0.1-day steps with no multiple-testing correction. Because the threshold tests are highly correlated, the reported p-value is a look-elsewhere effect and the global significance is likely much weaker. The authors should present a permutation or bootstrap calibration over the full threshold grid, or clearly downgrade the claim to an exploratory finding.","section":"Section 3.2, Figure 6 (top panel)"},{"comment":"The sample combines Kepler, K2, TESS, and APOGEE targets without applying a selection function or completeness correction. The central claims that planets with R_p>3 R_Earth occur only around stars with [M/H]>=0, and that there is a sharp period threshold at 4.3 days, are vulnerable to selection effects: if metal-poor M dwarfs with large planets or short-period planets are systematically missing from APOGEE or from the transit surveys, the trends would be artifacts. The authors should either quantify the selection function and show that the conclusions survive, or soften the 'only' language in the Conclusions and abstract.","section":"Section 2 and Sections 3.2-3.3"},{"comment":"The K-S tests and the Kendall tau and Spearman rho correlations treat individual planets as independent data points, but multiple planets from the same host star share a single [M/H] value. With 65 planets around 45 stars, and 13 multi-planet systems, the effective independent sample size is smaller than the number of planets, making the p-values anti-conservative. The analyses should be repeated using one planet per system, or with a bootstrap that resamples by host star, to verify that the reported significances (e.g., p=0.002 for singles vs. multis) remain.","section":"Section 3.2 (K-S and rank-correlation tests)"},{"comment":"The quadratic fit R_p = 2.83 + 4.20 [M/H] + 2.42 [M/H]^2 is derived using only the largest planets of multiple systems (red filled circles), and the statement that 'there is an absence of exoplanets with R_p > 2.5 R_Earth around stars with [M/H]<0' is based on a small and heterogeneous sample. The upper envelope is suggestive, but the phrase 'threshold in the maximum planet size' overstates the evidence, especially without quantifying detection completeness in the metal-poor regime.","section":"Section 3.2 (quadratic fit and upper envelope)"}],"minor_comments":[{"comment":"The black dashed line in Figure 4 is described in the text as the input abundance pattern of the MARCS model grid and is labeled only as 'schematic relation' in the figure; labeling it directly in the figure as 'MARCS input/schematic' would avoid confusion with an independent Galactic chemical evolution track.","section":"Section 3.1, Figure 4"},{"comment":"There is a typo in the text: 'R oplus' appears where 'R_Earth' or 'R_sun' is intended in the sentence 'most Hot Neptunes (2 R_Earth < R_p < 6 R_oplus)'.","section":"Section 3.3"},{"comment":"The Conclusion states 'larger exoplanets, with R_p > 3 R_Earth, are found only around the more metal-rich M dwarfs, with [M/H] > +0.0', while the Abstract and Figure 5 use [M/H] >= 0.0; these notations should be reconciled.","section":"Conclusions and Abstract"},{"comment":"The statement that conclusions would not change if the sample were split by mission is only qualitative; providing a brief quantitative breakdown (e.g., the period threshold and radius-metallicity slope for each mission subsample) would strengthen this robustness check.","section":"Section 3.3"}],"recommendation":"major_revision","confidential_remarks":"The paper's core qualitative results are interesting and likely correct, but the 4.3-day transition period is a headline claim that rests on an uncorrected p-value scan. The requested reanalysis (permutation-based p-values, host-star-resampled statistics, and a selection-function discussion) is essential; without it, the abstract's period-threshold statement should be revised. The qualitative radius-metallicity trend and the benchmark-validated metallicity scale are solid contributions worth preserving."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, the metallicity scale is the real product: 48 M dwarfs, APOGEE spectra, LTE synthesis, benchmarked against binary companions with a median offset of +0.01±0.04 dex. That is solid and useful. Second, the headline 4.3-day period–metallicity threshold is not supported by the statistics as presented. The paper takes the minimum of a K-S p-value scan over 0.1-day period thresholds and quotes it as a 99% detection, with no multiple-testing correction. That is a textbook look-elsewhere effect, and the claim is fragile.\n\nThe radius–metallicity trend is the strongest science here. Larger planets (Rp>3R⊕) only appear around stars with [M/H]>0, and the quadratic fit quantifies a positive correlation. This is visually clear in Figure 5 and consistent with independent FGK results. The single-vs-multi metallicity offset, already reported by Rodriguez Martinez et al. (2023) with heterogeneous values, is confirmed here on a homogeneous scale, which is worth having. The concentration of large planets at 2–5 days for M dwarfs, compared with broader FGK distributions, is also a reasonable observational claim.\n\nThe soft spots are real but localized. The period-threshold analysis ignores host-star clustering: 65 planets around 45 stars share the same [M/H], which inflates the significance of both K-S and rank-correlation tests. A host-star-level resampling or mixed-effects model is the right check. The [O/M]–[M/H] thin-disk relation is partly circular, as the authors themselves note: the dashed line they compare to is the MARCS grid input abundance pattern. The sample combines Kepler, K2, and TESS with APOGEE target selection and no completeness correction, so the absence of large planets around metal-poor stars could in principle be a selection artifact. I do not think it is, given the benchmark scale and the agreement with FGK results, but the paper should acknowledge that limitation more explicitly.\n\nBottom line: this deserves a serious referee. The right referee will push for a permutation-based global p-value for the period threshold and a host-star-level analysis. The qualitative results will likely survive. I would not cite the 4.3-day value, but I would cite the metallicity scale and the radius–metallicity trend.","headline":"The homogeneous APOGEE metallicity scale and the radius-metallicity trend are solid, but the 4.3-day period threshold is a minimum-p scan artifact and should not be taken at face value.","tokens_in":19860,"tokens_out":2792,"would_cite":true,"duration_ms":26243,"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":"Metal-poor M dwarfs never host planets larger than about three Earths, and planets with orbits shorter than 4.3 days orbit metal-rich stars instead.","keywords":["M dwarf stars","exoplanet host metallicity","APOGEE spectra","spectral synthesis","exoplanet radii","orbital period-metallicity relation","planet multiplicity","Galactic chemical evolution"],"falsifier":"Find one well-measured M dwarf with $[\\mathrm{M/H}] < 0.0$ hosting a transiting planet with radius above $3\\,R_\\oplus$ in a sample free of selection bias, and the metallicity threshold for large planets fails. Likewise, a volume-complete or selection-corrected survey of M dwarf hosts that shows the $[\\mathrm{M/H}]$ step at 4.3 days disappears -- for example, because metal-poor short-period hosts were missed -- would falsify the period-metallicity transition.","tokens_in":18904,"feed_emoji":"🪐","tokens_out":7750,"duration_ms":72822,"temperature":0.7,"pith_summary":"This paper tries to show that the metal content of M dwarf stars shapes the planets they host: planets larger than about three Earth radii appear only around metal-rich M dwarfs, while smaller planets form across the whole metallicity range. It also claims that planets on short orbits tend to orbit metal-rich M dwarfs, with the transition between two distinct host-metallicity populations at an orbital period of roughly 4.3 days. The authors derive metallicities and oxygen abundances for 48 M dwarf planet hosts from high-resolution near-infrared APOGEE spectra using spectral synthesis, and combine them with radii and periods for 246 planets around 188 M dwarfs from Kepler, K2, and TESS. Establishing these trends matters because M dwarfs are the most numerous stars and a prime target for future planet characterization; if the trends are physical, they constrain where and how planets form around low-mass stars.","feed_headline":"Big M dwarf planets need metal-rich stars; 4.3-day period splits the pattern","feed_subtitle":"Small planets appear around metal-poor and metal-rich M dwarfs alike; large ones need metal-rich hosts.","key_machinery":"The argument rests on a metallicity scale for M dwarfs built from LTE spectral synthesis of APOGEE near-infrared spectra using a radiative transfer code and standard cool-star model atmospheres, fitting water and OH molecular features to derive $T_{\\rm eff}$, $\\log g$, $[\\mathrm{M/H}]$, and oxygen abundances. Validation against benchmark M dwarfs in binaries with FGK companions gives a median offset of $+0.01 \\pm 0.04$ dex, anchoring the scale. On the planet side, radii come from transit depths and an $M_{K_s}$-based stellar radius relation. The period threshold is located by a period-scanning procedure: for each 0.1-day cutoff, the host-metallicity distributions of shorter- versus longer-period planets are compared with two-sample distribution tests, and the minimum $p$-value ($6.4\\times10^{-5}$) selects $P_{\\rm orb} = 4.3$ days; kernel regression and median metallicities confirm the step. The dust sublimation radius $R_{\\rm sub}$, scaling as the square root of stellar luminosity, is the proposed physical mechanism for why the threshold is shorter than for FGK stars.","core_discovery":"The central claim is a set of metallicity-dependent thresholds in M dwarf planetary systems. In the sample, every planet with radius $R_p > 3\\,R_\\oplus$ orbits an M dwarf with $[\\mathrm{M/H}] \\geq 0.0$, while planets smaller than $3\\,R_\\oplus$ are found around hosts spanning $[\\mathrm{M/H}]$ from $-0.6$ to $+0.3$. Host metallicity also anti-correlates with orbital period: stars hosting planets with $P_{\\rm orb} < 4.3$ days are statistically more metal-rich than stars hosting longer-period planets, the sharpest division found by scanning period thresholds and applying two-sample distribution tests. For the largest planets ($R_p > 4\\,R_\\oplus$), the orbital-period distribution peaks sharply between 2 and 5 days, which the authors attribute to inward migration. Multi-planet systems stand apart: their hosts are more metal-poor than single-planet hosts, and, with one exception, all their planets have $R_p < 3\\,R_\\oplus$. These trends mirror those seen around hotter FGK stars but with a shorter period threshold (4.3 days versus 8--10 days), which the paper suggests may reflect the smaller dust sublimation radius around the lower-luminosity M dwarfs.","pith_inferences":["If the trends are physical, the frequency of large planets around M dwarfs should be a steep function of $[\\mathrm{M/H}]$; a straightforward test is to compare planet occurrence rates in a metallicity-complete M dwarf sample, split at $[\\mathrm{M/H}] = 0$.","The single-versus-multi metallicity offset may partly be a detection bias: metal-rich disks are predicted to produce more mutually inclined multi-planet systems, making more of their planets transit-undetectable, so the 'single' population around metal-rich stars could hide unseen companions; the paper raises this but does not resolve it.","The dust-sublimation explanation is testable at the population level: if $R_{\\rm sub}$ sets the threshold, stars of similar luminosity should show the same period cutoff regardless of other properties, and the cutoff should scale with $L^{1/2}$ across spectral types."],"forward_implications":["Targeted searches for giant planets around M dwarfs should concentrate on metal-rich stars, since the sample finds no $R_p > 3\\,R_\\oplus$ planets around $[\\mathrm{M/H}] < 0$ hosts.","The 4.3-day host-metallicity step predicts that surveys of even later, lower-luminosity M dwarfs should find the transition at shorter periods if dust sublimation sets it.","Because multi-planet M dwarf systems are metal-poor and host only small planets, transit statistics that count single versus multiple systems will be skewed by metallicity; interpretations of multiplicity rates need this offset.","The 2--5 day pile-up of large planets implies inward migration is efficient around M dwarfs, so wide-orbit giant planets around M dwarfs should be rare relative to hot giants."],"supporting_citations":[{"why":"Supplies the FGK comparison threshold of 8--10 days for the short-period/metallicity effect and the kernel regression method for mean metallicity versus period.","marker":"Wilson et al. 2018"},{"why":"Provides the FGK host metallicity-versus-planet-size/period comparison showing hot planets orbit more metal-rich stars, the reference the M dwarf results are measured against.","marker":"Ghezzi et al. 2021"},{"why":"Benchmark M dwarfs with FGK companions used to validate the derived metallicity scale, with median offset +0.01±0.04.","marker":"Souto et al. 2022"},{"why":"Independent compilation of M dwarf host metallicities that already showed large planets preferring metal-rich hosts, providing external support for the radius-metallicity trend.","marker":"Hirano et al. 2018"},{"why":"Supplies the inward-migration mechanism invoked to explain the pile-up of large planets at 2--5 day periods.","marker":"Lin et al. 1996"},{"why":"Establishes the dust sublimation radius scaling $R_{\\rm sub} \\propto L^{1/2}$, the proposed explanation for the 4.3-day period threshold.","marker":"Millan-Gabet & Monnier 2002"},{"why":"Model linking metal-rich disks to larger mutual orbital inclinations, used to interpret the single/multi metallicity difference.","marker":"Pan et al. 2025"}],"fun_headline_variants":["Big M dwarf planets need metal-rich stars; small ones don't","Metal-rich M dwarfs host larger planets and shorter orbits","M dwarf planet radius and period tied to host metallicity","4.3-day orbit splits M dwarf planets by host metallicity","Large M dwarf planets orbit only metal-rich stars"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The sample combines stars from Kepler, K2, TESS, and APOGEE target selections without applying a selection function or completeness correction, so the trends are interpreted as physical only if the observed M dwarf planet hosts represent the full M dwarf planet population.","fun_headline_variants_meta":{"raw":{"variants":["Big M dwarf planets need metal-rich stars; small ones don't","Metal-rich M dwarfs host larger planets and shorter orbits","M dwarf planet radius and period tied to host metallicity","4.3-day orbit splits M dwarf planets by host metallicity","Large M dwarf planets orbit only metal-rich stars"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000593,"raw_usage":{"total_tokens":2898,"prompt_tokens":1182,"completion_tokens":1716,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":798,"completion_tokens_details":{"reasoning_tokens":1633}},"tokens_in":798,"tokens_out":1716,"duration_ms":14299,"temperature":1.0,"reasoning_tokens":1633,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T19:57:16.154993+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Find one well-measured M dwarf with $[\\mathrm{M/H}] < 0.0$ hosting a transiting planet with radius above $3\\,R_\\oplus$ in a sample free of selection bias, and the metallicity threshold for large planets fails. Likewise, a volume-complete or selection-corrected survey of M dwarf hosts that shows the $[\\mathrm{M/H}]$ step at 4.3 days disappears -- for example, because metal-poor short-period hosts were missed -- would falsify the period-metallicity transition.","supporting_citations":[{"cited_title":"F., Teske, J., Majewski, S","cited_arxiv_id":null,"evidence_quote":"Supplies the FGK comparison threshold of 8--10 days for the short-period/metallicity effect and the kernel regression method for mean metallicity versus period."},{"cited_title":"2018, AJ, 155, 127","cited_arxiv_id":null,"evidence_quote":"Independent compilation of M dwarf host metallicities that already showed large planets preferring metal-rich hosts, providing external support for the radius-metallicity trend."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the inward-migration mechanism invoked to explain the pile-up of large planets at 2--5 day periods."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the dust sublimation radius scaling $R_{\\rm sub} \\propto L^{1/2}$, the proposed explanation for the 4.3-day period threshold."}],"review_version":1}