{"id":"71acd78d-7421-4272-9227-87bc6ee4d68f","arxiv_id":"2607.17687","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"Under an intermediate level of chromosome redundancy, populations with additive traits under structured inheritance balance speed, exploitation, and exploration, and this matches observed ploidy levels.","lead":"A minimal computer model of evolution shows that carrying extra chromosome copies can help populations explore new traits when environments change abruptly. The model also predicts an optimal intermediate number of copies under stable selection, matching copy numbers seen in some plants and bacteria.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Free-N selection drives ploidy to low N, contradicting the claimed optimal range N≈15–30","rationale":"The paper's central claim is the emergence of an optimal ploidy range N≈15–30 in the mean-set model and its quantitative consistency with natural ploidy distributions. The most load-bearing requirement for this claim is that the model actually predicts selection in favor of that range. The free-N simulations are the direct test: when ploidy is heritable, selection should push N toward the optimum. Instead, Figure 4a shows the opposite — low N progressively prevails. The authors' attempt to reconcile this ('while low N values did not confer the best mean phenotype and highest variance, they provided the faster convergence') actually exposes the problem: the fixed-N 'optimum' is not a single point that maximizes all three objectives; the three peaks (N≈22, 16, 30) are for different metrics, and when N is free, the convergence advantage of low N dominates. Thus the claimed prediction is not an evolutionary outcome. This internal inconsistency is more fundamental than parameter sensitivity: even a perfect sensitivity analysis cannot fix the fact that the model's own selection dynamics do not favor N≈15–30. The quantitative match to empirical ploidy distributions is therefore coincidental, not explained by the model. A concrete check is to quantify the stationary N distribution in the mixed-N simulation; the paper's own text indicates it will be low. If so, the central claim must be withdrawn or substantially revised. The reader's weakest assumption about parameter sensitivity is valid but secondary; the free-N outcome is a direct falsification.","tokens_in":12121,"tokens_out":9690,"duration_ms":102178,"concrete_test":"Quantify the stationary N distribution in the mean-set smooth-fitness mixed-N simulation: compute the mean and mode over 100 repetitions. If the mean/mode are below N=10 (as Figure 4a suggests), the claimed optimal range is not selected for. Also run a control with a small mutation rate on N (e.g., N→N±1 with probability 10^-4) to verify the low-N attractor is robust. Report the distribution and compare to the fixed-N peaks at N≈22,16,30.","verdict_should_be":"REJECT","load_bearing_attack":"The central claim (Conclusions, point ii) that the mean-set model predicts an optimal ploidy range N≈15–30 is contradicted by the paper's own free-N simulation. In the smooth fitness landscape, Figure 4a shows that starting from a uniform distribution over N∈[1,100], 'individuals with low N progressively prevail after a few rounds of reproduction.' Thus the stationary distribution of N is concentrated at low values, not at the claimed optimum. The authors call this 'consistent' with the fixed-N results, but it is actually inconsistent: if low N out-compete intermediate N when ploidy is heritable, then the fixed-N 'optimum' is not an evolutionary attractor. Consequently, the claimed quantitative match to natural ploidy distributions (bacteria, cyanobacteria, seed plants) is unexplained: selection on N in the model drives ploidy down, so the model cannot account for observed N≈15–30. This is not a parameter-sensitivity issue; it is an internal contradiction between the fixed-N optimum and the free-N evolutionary outcome.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces a minimal one-dimensional evolutionary model to study how ploidy level N, inheritance mode (set vs. random), and genotype-phenotype mapping (mean vs. max) jointly shape adaptation. Four model variants (mean-set, mean-random, max-set, max-random) are compared under neutral, smooth, and rough fitness landscapes, and in mixed-N populations where N is inherited. The central claims are that (i) max-random maximizes both exploitation and exploration; (ii) the mean-set model displays an optimal ploidy range N≈15–30 under stabilizing selection, quantitatively consistent with empirical ploidy distributions; (iii) polyploidy facilitates escape from fitness valleys; and (iv) free evolution of N reflects a balance between exploitation, exploration, and convergence speed. The paper includes 100 independent repetitions for each condition and a PCA-based analysis linking fixed-N features to success in mixed-N populations.","tokens_in":12332,"tokens_out":7339,"duration_ms":76084,"significance":"If the central claims held, the model would offer a tractable framework for understanding when polyploidy is favored, connecting ploidy level to ecological selection pressures and making testable predictions about natural ploidy distributions. The systematic comparison of inheritance modes and genotype-phenotype mappings is a strength, and the simulation protocol is clearly specified with an unusually large number of replicates. The model's qualitative results, particularly the advantage of stochastic inheritance with max-mapping, are plausible and could inform empirical work. However, the paper's headline quantitative claim—the optimal ploidy range in the mean-set model—is undermined by an internal inconsistency with the free-N simulations and by the absence of sensitivity analysis, so the significance of the current version is limited.","major_comments":[{"comment":"The free-N simulation in the smooth fitness landscape (Figure 4a) shows that low N values progressively prevail, while the fixed-N analysis (Figure 2d) identifies an optimal ploidy range N≈15–30. The authors call this 'consistent' because low N provides faster convergence, but this is not a reconciliation: if heritable N evolves to low values, the fixed-N optimum is not an evolutionary attractor. Moreover, the stationarity criterion used to stop the simulation is based on the mean phenotype, not on the N distribution; the N distribution may still be transient at the stopping time. Thus the coloring of points in Figure 4b and the PCA-based claims about 'successful N' may be artifacts of the stopping rule. As it stands, the paper contains a direct contradiction between Conclusion (ii) and the free-N result. The authors should either demonstrate that the N≈15–30 range is an attractor under","section":"§4 and Conclusions (ii)"},{"comment":"The predicted optimal ploidy range and its quantitative comparison with empirical ploidy distributions rest on a single, hand-picked parameter set: mutation rate λ=0.00032, Beta(1,1) mutation increments, initial gene value 0.2, fitness amplitudes A=0.5, A_v=0.3, A_p=0.3, and peak widths σ=0.09/0.05. No sensitivity analysis is provided. The peaks in exploitation (N≈22), exploration (N≈16), and convergence speed (N≈30) are numerical outcomes; without varying λ, the mutation distribution, or the fitness curvature, one cannot determine whether these peaks are robust or whether the match to PloiDB data is coincidental. For a claim of 'quantitative consistency' (Conclusions, point ii), the authors should show how the optimal range shifts under plausible parameter variations, or at minimum provide a phase diagram over key parameters.","section":"Methods and Figure 2d–e"},{"comment":"The text states that 'the mean-set model is confirmed to be the least adaptable model: in the peak landscape it fails to reach the global maximum for any value of N' yet later claims that 'polyploidy can confer an adaptive advantage during abrupt environmental changes in all four models.' For the mean-set model under the peak fitness, polyploidy may increase phenotypic variance, but it does not lead to escape from the suboptimal peak. The conclusion (iii) that polyploidy 'facilitates escape from suboptimal fitness states in all four models' is therefore unsupported. The authors should either soften the claim to 'increases phenotypic variability in all models, but does not guarantee escape in all cases' or provide evidence that mean-set populations with larger N move closer to the global optimum.","section":"§3 and Conclusions (iii)"}],"minor_comments":[{"comment":"The 'neutral fitness' f(y)=y is not neutral; it is directional selection favoring higher phenotypes. This is misleading because the 'baseline' environment is not selection-free. The authors should rename it (e.g., 'unimodal directional fitness' or 'no-optimum fitness') and adjust the interpretation of the 'neutral' results throughout, particularly in the derivation of the acceleration Ac.","section":"§2"},{"comment":"The mutation distribution is described as Beta(1,β) with β=1/x̄−1 and x̄=0.5, which gives Beta(1,1), i.e., the uniform distribution on [0,1]. The text states that this distribution 'explores the full range [0,1] with a bias toward intermediate values.' A uniform distribution is not biased toward intermediate values; please correct this description.","section":"Methods"},{"comment":"The claim that the simulated optimal range 'closely matches' the PloiDB histogram would be strengthened by a formal statistical comparison (e.g., overlap coefficient, Kolmogorov–Smirnov test) and by reporting confidence intervals on the simulated peak locations. As presented, the visual match is subjective.","section":"Figure 2e"},{"comment":"In several figures (e.g., Figures 1–3), error bars or confidence bands from the 100 independent repetitions are not shown, making it difficult to assess the statistical significance of the reported peak and plateau differences. The shaded regions in Figure 4a are welcome; similar treatment elsewhere would improve the paper.","section":"Figures 1–3"}],"recommendation":"major_revision","confidential_remarks":"The paper has a clear model and some convincing qualitative results, but the central quantitative claim—the optimal ploidy range in the mean-set model—is internally inconsistent with the free-N simulation, which selects for low N. This is a serious issue that the authors may attempt to patch by reinterpreting the stopping criterion or by adding longer simulations. I suspect the fixed-N 'optimum' is a transient artifact of the convergence-speed metric, and the empirical match may be coincidental. The paper also lacks sensitivity analysis, which is essential for a claim of quantitative consistency with natural ploidy distributions. These issues are fixable with additional computational experiments, but they are substantial enough to require thorough revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague, here's my take on arXiv:2607.17687. The paper is a genuinely useful simulation study of how inheritance mode and genotype-phenotype mapping interact with ploidy. The systematic comparison across neutral, smooth, valley, and peak landscapes is clean, and the result that max-random maximizes both exploitation and exploration across all environments is robust. The extension to heterogeneous chromosome numbers and the PCA linking fixed-N features to selective success is a nice addition.\n\nWhere it falls down is the central empirical claim. The mean-set model under smooth fitness shows fixed-N peaks in mean phenotype, variance, and convergence speed at N≈22, 16, and 30, which the authors claim matches natural ploidy distributions in bacteria and seed plants. But when N is allowed to evolve (free-N simulations), low N rapidly take over in that same landscape. The authors call this 'consistent' with the fixed-N results, but it is not: if low N outcompete intermediate N when ploidy is heritable, then the fixed-N 'optimum' is not an evolutionary attractor. The model's own dynamics do not generate the observed ploidy distributions; they predict a collapse to low N. That is load-bearing for conclusion (ii), and the stress-test note is right.\n\nAlso worth flagging: (1) the 'neutral' fitness f(y)=y is directional selection, not neutral; this mislabels all the neutral-landscape results. (2) The conclusion that polyploidy facilitates escape from fitness valleys 'in all four models' is contradicted by the mean-set model failing to reach the global maximum under peak fitness. (3) No sensitivity analysis for the parameters that produce the N≈15–30 window, and no code or deposited data.\n\nThese are fixable. The core qualitative message — that stochastic inheritance plus a nonlinear mapping is best for exploration, and that different ploidies trade off exploitation, exploration, and speed — survives. But the quantitative match to natural ploidy distributions needs to be rethought or heavily qualified.\n\nWho is this for? Anyone modeling polyploidy or evolvability. It deserves a serious referee; the issues are addressable and the framework is useful. My recommendation: send it to review, but push hard on the free-N contradiction and the mislabeled neutral case.","headline":"The mean-set optimal ploidy range is not evolutionarily stable under the model's own free-N dynamics, which undermines the headline empirical match; the rest of the paper is a solid simulation study worth engaging.","tokens_in":12869,"tokens_out":3307,"would_cite":false,"duration_ms":35704,"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":"This paper argues that polyploidy's evolutionary value is context-dependent, with an optimal chromosome-copy range near N≈15–30 under stabilizing selection, matching ploidy distributions observed in some plants and bacteria.","keywords":["polyploidy","whole-genome duplication","genotype-phenotype mapping","inheritance mode","fitness landscape","exploration-exploitation tradeoff","evolutionary innovation","ploidy evolution"],"falsifier":"Run a parameter sweep over mutation rate λ ∈ [10^-5, 10^-2] and fitness amplitudes in [0.2, 0.8] for the mean-set model; if the peaks in mean phenotype, variance, and acceleration do not cluster in a low-to-intermediate N range for a substantial fraction of the sweep, the claimed optimal ploidy range is not generic.","tokens_in":11947,"feed_emoji":"🧬","tokens_out":6410,"duration_ms":57732,"temperature":0.7,"pith_summary":"This paper sets out to determine whether the number of chromosome sets (ploidy) is an evolutionary asset or a dead end, and under what conditions. Using a minimal model in which individuals carry N copies of a single gene, with phenotype defined either as the average or the maximum of the copies and inheritance either structured or random, it finds that polyploidy can boost adaptation, especially during abrupt environmental changes. The most striking result is that in the 'mean-set' version—additive gene action with structured inheritance—performance peaks in the low-to-intermediate ploidy range N≈15–30, matching ploidy distributions seen in real plants and bacteria. The paper thus argues that genome redundancy per se is not disadvantageous; rather, its value depends on how phenotypes are built and how chromosomes are inherited.","feed_headline":"Model finds optimal ploidy: 15–30 chromosome copies","feed_subtitle":"Simulations tie intermediate chromosome numbers to a balance of exploitation, exploration, and speed that matches real plant and bacterial d","key_machinery":"The model's central object is a one-dimensional population-genetic simulation: M individuals, each with N homologous gene copies whose activities x_ij in [0,1] determine a phenotype y_i by either the mean or the maximum of the copies. Inheritance occurs in two extreme modes: 'set' (each parent's copy set is duplicated and transmitted together) and 'random' (copies are sampled independently, approximating polysomic segregation). Mutations occur at rate λ per copy per generation, with new values drawn from a Beta distribution. The key machinery is the comparison of the four model variants across four fitness landscapes (neutral, smooth, valley, peak), and the identification of N-dependent peak","core_discovery":"The central claim is that polyploidy's evolutionary value is context-dependent and can be traced to a tradeoff among three measurable population-level properties: exploitation (proximity to the fitness optimum), exploration (phenotypic variance), and convergence speed (generations to stationarity). In the mean-set model, which the authors argue approximates structured inheritance with additive gene dosage, the tradeoff yields a non-monotonic dependence of all three properties on N, with a coherent optimal ploidy range of roughly 15–30 copies. The paper further claims that these N values are quantitatively consistent with the ploidy distributions observed in certain polyploid bacteria, cyanob","pith_inferences":["The model implies that species with dosage-sensitive traits and structured inheritance should show a modal ploidy near 15–30; this can be tested across independent phylogenetic contrasts, rather than just the taxa cited.","The framework suggests that polyploidy's role in cancer—where cells undergo whole-genome duplication during therapy—may be an instance of the same exploration-exploitation tradeoff under abrupt environmental change; this is an extension the authors mention as a practical motivation.","Because the optimal range is derived from a single parameter set, an immediate test is to sweep the mutation rate and fitness curvature; if the peaks are robust, the 15–30 window becomes a strong quantitative prediction, if not, the match with natural ploidies may be coincidental.","The model's 'max' mapping could represent threshold or dominant gene effects; real organisms with such regulatory architectures should show even stronger advantages from polyploidy than those with purely additive dosage."],"forward_implications":["Under stabilizing selection, populations exhibiting structured inheritance and additive gene action are expected to evolve toward roughly 15–30 chromosome copies, a window the model finds consistent with natural ploidy distributions in polyploid bacteria, cyanobacteria, and seed plants.","Abrupt environmental changes favor polyploidy in all four model variants, because extra chromosome copies raise phenotypic variance and thus the chance of escaping fitness valleys or local optima.","The best combination for both exploitation and exploration is random (stochastic) inheritance paired with a maximum-based phenotype map, indicating that redundancy is most beneficial when it can generate diversity through segregation noise and nonlinear readouts.","When chromosome number is an inherited trait, selection does not maximize any single objective; it tunes ploidy to balance exploitation, exploration, and convergence speed, with the balance determined by the environment."],"fun_headline_variants":["Polyploidy's sweet spot: 15-30 copies via tradeoff","Evolutionary tradeoff puts optimal ploidy at 15-30","Why 15-30 chromosomes? Model says tradeoff","Optimal ploidy: 15-30 copies from three-way balance","Model: ploidy 15-30 balances exploration and speed"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that the chosen parameters (λ=0.00032, Beta(1,1) mutation increments, initial value 0.2, fitness amplitudes and widths) faithfully represent dosage-sensitive traits under stabilizing selection; if the parameter set is unrepresentative or lacks sensitivity, the 15–30 optimal window may be an artifact.","fun_headline_variants_meta":{"raw":{"variants":["Polyploidy's sweet spot: 15-30 copies via tradeoff","Evolutionary tradeoff puts optimal ploidy at 15-30","Why 15-30 chromosomes? Model says tradeoff","Optimal ploidy: 15-30 copies from three-way balance","Model: ploidy 15-30 balances exploration and speed"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000418,"raw_usage":{"total_tokens":1986,"prompt_tokens":736,"completion_tokens":1250,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":480,"completion_tokens_details":{"reasoning_tokens":1170}},"tokens_in":480,"tokens_out":1250,"duration_ms":8690,"temperature":1.0,"reasoning_tokens":1170,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T17:15:21.660263+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run a parameter sweep over mutation rate λ ∈ [10^-5, 10^-2] and fitness amplitudes in [0.2, 0.8] for the mean-set model; if the peaks in mean phenotype, variance, and acceleration do not cluster in a low-to-intermediate N range for a substantial fraction of the sweep, the claimed optimal ploidy range is not generic.","supporting_citations":[],"review_version":1}