{"id":"b030eb94-1a0b-4a25-8323-91511701ea08","arxiv_id":"2411.16960","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":14,"one_line_summary":"The paper re-draws the sub-Jovian and Neptune desert using incident stellar flux instead of orbital period, yielding power-law boundaries in flux-radius and flux-mass space.","lead":"The authors map the sub-Jovian and Neptune desert using the stellar flux received by planets instead of their orbital periods. The new maps show that most previously identified hot Neptunes are not actually inside the desert when stellar radiation is accounted for, which could change how planetary formation and evaporation models are tested.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Boundary power-law exponents are conditional on post-hoc choices (z-score, percentile threshold, clipping); no sensitivity test is provided, so the quantitative claim is not yet established.","rationale":"The reader's weakest-assumption identification is exactly the load-bearing issue: the boundary points are constructed using z-scores, percentile thresholds, and clipping levels selected a posteriori from the same dataset, with no sensitivity analysis for those choices. I agree that this is where the quantitative claim is least secure. The paper itself is honest about the arbitrariness (Sect. 2.2: 'The decision to define the boundary points as the points with a z-score equal to 2 in each distribution is arbitrary, but was made in light of an a posteriori visual analysis'; Sect. 2.3: 'the choice of the proper z score was based on an a posteriori visual inspection'), which turns the issue into a missing robustness demonstration rather than a hidden assumption. The qualitative result, that a deficit of highly irradiated Neptunes persists in incident-flux space and that many period-defined hot Neptunes fall outside the irradiation desert, is supported by the figures and by the independent F75th Monte Carlo exercise. However, the headline slopes of -0.27, 0.11, -0.70, and 0.47 are not stable under alternative reasonable boundary definitions unless demonstrated otherwise. The proposed test varies only the z-score while keeping everything else fixed, which directly targets the most explicit arbitrary choice in Eqs. (3)-(6). Because the reader already recommended conditional acceptance with a request for robustness tests, my read does not change that verdict; it reinforces it.","tokens_in":20191,"tokens_out":3490,"duration_ms":38037,"concrete_test":"Recompute the lower and upper boundary points and refit Eq. (8) for the F-Rp plane using z=1 and z=3 instead of z=2, and for the F-Mp plane using z=0.5 and z=2 instead of z=1, holding all other choices (flux grids, 2-sigma clipping, F75th thresholds, sample cuts) fixed. Compare the resulting alpha and ln beta with Table 3 and recompute the Bayes factors against the linear model. If any slope shifts by more than the 68% credible interval, or if the power law is no longer strongly favoured (Delta ln E < 5), then the fitted exponents are not robust to the explicitly arbitrary z-score choice and the quantitative boundary claim should be downgraded.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim is that the desert edges are power laws with the slopes in Table 3. Those slopes are not measured from an independently defined physical boundary: they are outputs of a pipeline whose key ingredients are chosen after inspecting the same data. In Sect. 2.2 the radius boundary points are defined by R_low,i = mu_AR,i + 2 sigma_AR,i and R_up,j = mu_BR,j - 2 sigma_BR,j (Eqs. 3-4), with the z-score explicitly described as 'arbitrary' and selected 'in light of an a posteriori visual analysis.' In Sect. 2.3 the mass boundaries use z=1 (Eqs. 5-6), also after visual inspection. The flux thresholds F75th = 200 F_sun and 550 F_sun are set as 75th percentiles of the same sample (Sect. 2.4), the 2-sigma clipping of contaminants is applied in every slice, and the Mp < 600 M_sun cut is also arbitrary. The paper reports one robustness test for bin size (20 vs. 10 objects per bin) and a Monte Carlo study of F75th, but it never varies z, the percentile defining F75th, or the clipping level. Because the fitted exponents describe points that are themselves defined by these choices, the strong Bayesian preference for the power-law model (Table 2) only shows that a power law fits the constructed points, not that the desert boundary is intrinsically a power law. The mass-radius relations in Eqs. (14)-(15) are derived by combining the same fitted power laws, so they inherit this arbitrariness and are not independent support for the boundary picture. The qualitative paucity of highly irradiated Neptunes in flux space is well supported, but the specific slopes, and therefore the analytic boundary expressions, are not yet robustly established.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes replacing the orbital period with the incident stellar flux F as the key variable for defining the sub-Jovian and Neptune desert. Using 1527 confirmed exoplanets with precise period, radius, and stellar parameters (sample S1) and a mass subsample of 512 planets (S3), the authors apply a two-component Gaussian mixture model to separate small/rocky planets from giants, then define lower and upper desert boundaries in the (F,Rp) and (F,Mp) planes from mean-plus/minus-z-sigma points in flux slices. Power-law models are fitted to these boundary points with the nested-sampling code Diamonds, yielding the slopes reported in Table 3. The paper also derives mass-radius relations for the two boundaries by eliminating F between the fits, and reports that about 87% of period-defined hot Neptunes fall outside the new irradiation desert.","tokens_in":20663,"tokens_out":4940,"duration_ms":49094,"significance":"If robust, the paper would provide a useful re-framing of the Neptune/sub-Jovian desert using a physically motivated stellar property, and the analytical boundary expressions would be directly testable with future transit samples. The qualitative message—that the dearth of highly irradiated Neptune-sized planets persists in flux space—is well supported by the figures and by the KS tests. The authors also make good use of Bayesian model comparison and include a Monte Carlo investigation of the F75th threshold. However, the central quantitative claim (the power-law slopes in Table 3) is currently conditional on a chain of a posteriori choices that are not stress-tested, and the mass-plane boundary rests on very few points. The paper is publishable only after the boundary-definition procedure is shown to be robust or its limitations are explicitly quantified.","major_comments":[{"comment":"The boundary points that drive the entire analysis are defined with arbitrarily chosen z-scores (z=2 for radius, z=1 for mass), a 2-sigma clipping procedure applied only to the lower-edge groups, and flux thresholds set to the 75th percentile of the same sample. The paper explicitly states that these choices were made after visual inspection (e.g., §2.2: 'The decision to define the boundary points ... is arbitrary, but was made in light of an a posteriori visual analysis'). No sensitivity test is provided for the z-score, the percentile defining F75th, the clipping level, or the mass cutoff of 600 M⊕. Because the fitted exponents in Table 3 describe points that are themselves defined by these choices, the abstract's claim that the boundaries are 'well described by a power-law model' is not yet established as a property of the desert rather than a property of the chosen boundary construction. I request a systematic robustness test (e.g., z=1 and z=3 for radius, 60th/90th percentiles for F75th, with/without clipping) and a report of how the exponents change.","section":"§2.2, Eqs. (3)-(4); §2.3, Eqs. (5)-(6)"},{"comment":"The strong Bayesian preference for the power-law model over the linear and quadratic models is computed for the constructed boundary points, not for an independently defined boundary. The model comparison therefore shows only that a power law fits the constructed points, not that the boundary is intrinsically a power law. A more convincing test would be to define the boundary with one procedure (e.g., a density contour in the F-Rp plane) and check whether the power-law form predicts the location of a hold-out sample or a different flux range. Without such an independent check, the evidence ratios in Table 2 are circular with respect to the boundary definition.","section":"§2.6, Table 2"},{"comment":"In the F-Mp plane the quoted power laws come from only 3 lower-edge points and 4 upper-edge points. The fitted parameters have very large uncertainties (e.g., lnβ = 1.78 ± 1.77 for the upper boundary), and the confidence bands of the two boundaries overlap in Fig. 11. The paper acknowledges the overlap but still presents the mass-plane constraints as if they support the power-law conclusion. With such a small number of points, the large Bayes factors in Table 2 are not compelling by themselves; the authors should either use more flux slices (with a lower minimum count or a bootstrap resampling) or explicitly demote the mass-plane power-law slopes to tentative values.","section":"§3.2, Table 3, Fig. 11"},{"comment":"The mass-radius relations are obtained by eliminating F between two fitted power laws, so their exponents are ratios of the fitted slopes and their uncertainties are not independent. Moreover, the radius-plane fits use sample S1 over [200,4000] F⊕, while the mass-plane fits use the smaller sample S3 over [550,4000] F⊕, and the boundary points are defined with different z-scores in the two planes. Combining these two sets of fits into a single mass-radius relation without accounting for the different samples and thresholds is questionable. This should at least be stated as an inconsistency, or the relation should be derived from a common subsample with a consistent boundary definition.","section":"§3.3, Eqs. (14)-(15)"}],"minor_comments":[{"comment":"The quantity F is called the 'insulation flux' throughout; the standard terms are 'incident flux' or 'instellation flux'. Please use one consistent term.","section":"Eq. (1) and throughout"},{"comment":"The sentence 'the errors on the fitted parameters are relatively larger because the statistics in the flux-mass diagram are larger' appears to mean 'sparser' or 'poorer'; as written it contradicts the surrounding discussion and the description of sample S3.","section":"§2.3"},{"comment":"There is a typo in 'we only considered exoplanets that received at lest 550 F⊕' ('at least').","section":"§2.3"},{"comment":"Typo: 'the the Rp distributions of group AR' has a duplicated 'the'; also the unusual spacing in 'D iamonds' should be regularized to 'DIAMONDS' or 'Diamonds' throughout.","section":"§2.2"},{"comment":"The figure caption calls the 110 period-defined hot Neptunes 'yellow dots', while the text refers to them as 'yellow triangles'; the caption and text should agree.","section":"Fig. 11 caption vs §3.2"},{"comment":"The cross-reference 'as shown in Sects. 2.6 and 2.6' should cite the correct sections for the radius-plane and mass-plane fits.","section":"§3, first paragraph"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of A&A and the qualitative result is interesting. The main issue is not the qualitative dearth but the quantitative boundary slopes, which are insufficiently robust to the many a posteriori choices in the boundary definition. I recommend major revision: the authors should add sensitivity tests and either strengthen or appropriately weaken the claims about the specific power-law exponents. I do not see grounds for rejection, as the central idea is sound and the necessary tests are feasible within the manuscript's scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the paper shows convincingly that the Neptune/sub-Jovian desert persists when you plot against incident flux instead of period, and that most period-defined hot Neptunes scatter outside the flux-defined desert. That qualitative result is worth having. The quantitative boundary fits, however, are only as good as the ad hoc definitions that produce the boundary points, and the paper doesn't test how much the slopes move when those choices move.\n\nWhat's new: Lundkvist et al. used flux for the hot super-Earth desert, and Szabó et al. showed stellar-parameter dependence of period-based edges, but nobody had fitted analytic power-law edges in (F,Rp) and (F,Mp) for this desert. The paper does that, gives Table 3 slopes, and derives boundary mass-radius relations. That is a concrete, citable contribution.\n\nWhat's good: the sample is clean (10% uncertainties), the GMM separation into two populations is sensible, the KS tests support distinct populations, and the Monte Carlo on F75th for the radius plane is a reasonable check. The writing is clear and the Bayesian model comparison is standard. The authors are honest that the z-scores are arbitrary and chosen after looking at the data.\n\nWhere it's soft: the central quantitative claim is conditional on choices that are not varied. z=2 for radius, z=1 for mass, the 75th percentile threshold, the 2-sigma clipping, and the 600 Earth-mass cut are all post-hoc. The paper tests bin size and does the F75th Monte Carlo, but never sweeps z or the percentile or the clipping. For a fitted edge, the exponents are exactly the kind of thing that shifts with those choices. The mass-radius relations in Eqs (14)-(15) are just algebra from the same fits, so they don't independently support the boundary picture. Also, no data/code release is mentioned, which matters for a fitting paper. The claim that ~87% of hot Neptunes move outside the irradiation desert is descriptive and robust to some of these choices, so that part holds up.\n\nWho is it for: exoplanet demographics people, photoevaporation modelers, anyone writing a survey paper on the desert. It deserves peer review, but the referee should ask for a sensitivity analysis of the slopes to z-score, threshold, and clipping before publication.","headline":"Qualitative desert in flux space is real, but the fitted power-law boundaries rest on post-hoc choices and need a robustness pass before the slopes are quoted.","tokens_in":21249,"tokens_out":1895,"would_cite":true,"duration_ms":17923,"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 exoplanet desert is carved by starlight, not by orbital period, and its edges follow power laws in incident flux.","keywords":["sub-Jovian desert","Neptune desert","incident stellar flux","irradiation desert","exoplanet demographics","power-law boundaries","Gaussian mixture model","hot Neptunes"],"falsifier":"Recompute the boundary points in the same samples using $z=1$ and $z=3$ in the flux-radius plane and $z=2$ in the flux-mass plane, and recompute the flux thresholds from the 60th and 90th percentiles of the same radius and mass selections; if the fitted power-law slopes shift by more than the quoted 68.3% credible intervals, or the power-law model no longer wins the Bayes-factor comparison, the reported desert boundaries are an artifact of those choices.","tokens_in":19962,"feed_emoji":"🪐","tokens_out":8400,"duration_ms":71811,"temperature":0.7,"pith_summary":"This paper argues that the sub-Jovian and Neptune desert, the scarcity of Neptune- and Saturn-sized planets on very short orbits, is better traced by the stellar radiation a planet receives than by its orbital period. Working with 1,527 confirmed transiting planets with precisely measured radii and 512 with masses, the authors show that the desert appears as a triangular gap in the incident-flux versus radius and incident-flux versus mass planes. They fit the upper and lower edges of that gap and find both are power laws in flux, with slopes that differ between the radius and mass planes. If the claim holds, the period-based 'hot Neptune' population is largely a coordinate artifact: about 87% of hot Neptunes defined by period fall outside the flux-defined desert in the radius plane, and 63 of 110 do so in the mass plane.","feed_headline":"87% of hot Neptunes are not in the true desert","feed_subtitle":"Counting planets by stellar flux instead of orbital period puts 87% outside the desert and gives simple power-law edges.","key_machinery":"The load-bearing object is the incident stellar flux $F = L_*/(4\\pi d^2)$, or in solar units $F/F_\\oplus = (\\rho_*/\\rho_\\odot)^{-2/3}(P/1\\,\\mathrm{yr})^{-4/3}(T_*/T_\\odot)^4$, which folds orbital distance, stellar temperature, and stellar density into a single number. To place the desert edges, the paper runs a two-component Gaussian mixture model in $\\log F$ versus $\\log R_p$ (and $\\log F$ versus $\\log M_p$) to separate small planets from giants, then defines boundary points in flux slices as $\\mu \\pm z\\sigma$ after removing contaminants, with $z=2$ in the radius plane and $z=1$ in the mass plane. A Bayesian nested-sampling fit then selects a pure power-law model over linear and quadratic alternatives by Bayes factors exceeding the strong-evidence threshold, producing the analytic boundary curves.","core_discovery":"On the paper's own terms, the discovery is that the boundaries of the sub-Jovian and Neptune desert are power-law relations in the incident stellar flux $F$, not in orbital period: $R_p/R_\\oplus = \\beta (F/F_\\oplus)^\\alpha$ with $\\alpha = -0.27^{+0.02}_{-0.02}$ for the lower radius edge and $\\alpha = 0.11^{+0.02}_{-0.03}$ for the upper radius edge, and $M_p/M_\\oplus = \\beta (F/F_\\oplus)^\\alpha$ with $\\alpha = -0.70^{+0.16}_{-0.13}$ and $0.47^{+0.19}_{-0.22}$ in the flux-mass plane. The paper calls the depleted region the 'irradiation desert.' Combining the two planes yields mass-radius relations for the desert edges, with the upper edge consistent with very low-density gas giants around $0.1\\,\\mathrm{g/cm^3}$ and the lower edge spanning denser objects that could be silicate worlds or stripped cores. A direct consequence is that 194 of the 221 planets classified as hot Neptunes by the standard period-based boundaries lie outside the irradiation desert in the flux-radius plane, while only three planets previously outside the period desert move into it.","pith_inferences":["A testable extension not in the paper: convert survey completeness maps from (period, radius) to (flux, radius) and check whether the irradiation-desert boundaries persist after selection-bias correction; if they weaken, part of the desert may be an artifact of transit detectability.","The z-score choices ($z=2$ in radius, $z=1$ in mass) and the 75th-percentile flux thresholds are the fragile link; re-running the boundary construction with $z=1$ and $z=3$ and with thresholds from the 60th and 90th percentiles would show whether the quoted exponents are stable.","If the flux-based desert is the physically meaningful one, theoretical models of photoevaporation and high-eccentricity migration should predict boundaries that move with stellar spectral type in flux space; comparing the power-law slopes with population-synthesis predictions would test the mechanism.","The paper's reclassification suggests that reported 'oases' or 'savannas' of hot Neptunes may depend on the coordinate choice, and a homogeneous re-analysis of new transit candidates in flux space would clarify whether the remaining desert occupants are a distinct sub-population or contaminants."],"forward_implications":["If the flux-based picture is right, about 87% of planets previously classified as hot Neptunes in the radius plane are ordinary planets once stellar luminosity and distance are folded in, so occurrence-rate studies based on period alone will misclassify the desert population.","Only 27 of the 221 period-defined hot Neptunes remain inside the flux-defined desert, and three new planets enter it, giving a different census of the desert's occupants.","The mass-radius relations for the two edges, $R_p \\propto M_p^{0.23}$ on the upper edge and $R_p \\propto M_p^{0.39}$ on the lower edge, associate the upper edge with tenuous gas giants and the lower edge with denser, partially stripped planets, connecting the desert to photoevaporation and migration scenarios.","The difference between the flux thresholds in the radius and mass planes (about $200\\,F_\\oplus$ versus $550\\,F_\\oplus$) survives Monte Carlo tests that equalize sample sizes, which the paper interprets as a physical rather than purely statistical difference.","Future surveys that measure masses for many more short-period Neptunes should shrink the uncertainties on the mass-plane power laws and test whether the two planes still agree."],"supporting_citations":[{"why":"Supplies the period-based desert boundaries and the set of 221 'hot Neptunes' whose reclassification is the paper's headline result.","marker":"Mazeh et al. 2016"},{"why":"Introduced the incident-flux perspective for the hot super-Earth desert, which the paper extends to sub-Jovians and Neptunes.","marker":"Lundkvist et al. 2016"},{"why":"Showed the desert edge depends on stellar effective temperature, metallicity, and log g; the paper replicates this on a larger sample.","marker":"Szabó & Kálmán 2019"},{"why":"Updated the stellar-parameter dependence using uniformly retrieved stellar parameters; the paper confirms these findings in the period-mass plane.","marker":"Szabó et al. 2023"},{"why":"Provides the DIAMONDS nested-sampling tool used for the Bayesian parameter estimates and model comparison.","marker":"Corsaro & De Ridder 2014"},{"why":"Provides the forecaster mass-radius relation used to augment the mass sample in the threshold-flux robustness check.","marker":"Chen & Kipping 2016"},{"why":"Defines the Neptunian ridge separating the desert from the savanna, which motivates the sigma-clipping of the lower boundary.","marker":"Castro-González et al. 2024"}],"fun_headline_variants":["Flux, not period, exposes true Neptune desert","87% of hot Neptunes are not in the desert","Power-law flux edges map the real Neptune desert","Stellar flux redraws Neptune desert as power laws"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the desert edges can be located by the paper's own statistical conventions, namely the 75th-percentile flux thresholds of $200\\,F_\\oplus$ and $550\\,F_\\oplus$, the $z=2$ and $z=1$ boundary definitions, and the 2-$\\sigma$ clipping, so if a different reasonable choice of those conventions moves the edges, the power laws are not a stable physical feature.","fun_headline_variants_meta":{"raw":{"variants":["Flux, not period, exposes true Neptune desert","87% of hot Neptunes are not in the desert","Power-law flux edges map the real Neptune desert","Stellar flux redraws Neptune desert as power laws"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000675,"raw_usage":{"total_tokens":3175,"prompt_tokens":1149,"completion_tokens":2026,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":765,"completion_tokens_details":{"reasoning_tokens":1961}},"tokens_in":765,"tokens_out":2026,"duration_ms":14723,"temperature":1.0,"reasoning_tokens":1961,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T12:42:44.383773+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the boundary points in the same samples using $z=1$ and $z=3$ in the flux-radius plane and $z=2$ in the flux-mass plane, and recompute the flux thresholds from the 60th and 90th percentiles of the same radius and mass selections; if the fitted power-law slopes shift by more than the quoted 68.3% credible intervals, or the power-law model no longer wins the Bayes-factor comparison, the reported desert boundaries are an artifact of those choices.","supporting_citations":[{"cited_title":"S., Kjeldsen, H., Albrecht, S., et al","cited_arxiv_id":null,"evidence_quote":"Introduced the incident-flux perspective for the hot super-Earth desert, which the paper extends to sub-Jovians and Neptunes."}],"review_version":1}