{"id":"8ee6bae4-aee0-406e-880c-d14996b61980","arxiv_id":"2607.29001","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For a Fisher–KPP population in a habitat whose beneficial region moves at speed β, the paper proves the invasion front stays within O(1) of the moving habitat interface in several parameter regimes, with Bramson-type log t corrections when the interface is slow.","lead":"This mathematics paper surveys how shifting climate boundaries control the speed and shape of invading populations, and proves new estimates for where the invasion front sits when the habitat ahead of the front is worse. It packages a mature Hamilton–Jacobi toolbox for reaction–diffusion ecologies and adds logarithmic-correction bounds in the shifting-habitat setting.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 11.8 bridges from small δ to all δ∈(0,1−a) via the unproved assertion ζ_{δ1}(t)−ζ_{δ2}(t)=O(1); without it Theorem 11.3(iii) and Corollary 11.4 lack full support.","rationale":"The reader's weakest_assumption identifies exactly the unsupported O(1) level-set difference used at the end of Lemma 11.8. This is the most load-bearing concern because Theorem 11.3(iii) and Corollary 11.4 depend on the lower bound for every δ∈(0,1−a), and the only proof offered for that extension is a one-sentence assertion. The concern is substantive, not a matter of presentation: the sub-solution construction in Lemma 11.8 can only produce a plateau of height at most σ = 1−β²/4−1/ℓ², which may be much smaller than 1−a; without a separate argument the theorem does not cover thresholds above σ. No explicit counterexample is known, and the assertion may be true, but the paper as written leaves a genuine gap in the proof of its central new result. I agree with the reader's conditional assessment: the manuscript should be accepted only after the authors supply a proof or citation for the O(1) level-set difference, or otherwise repair the extension to arbitrary δ. I do not see a reason to strengthen or weaken the verdict beyond the reader's CONDITIONAL.","tokens_in":38749,"tokens_out":6968,"duration_ms":68337,"concrete_test":"For parameters such that β²/4 > a (e.g. a=0.5, β=1.9), run a numerical simulation of (74) with compactly supported initial data satisfying liminf_{x→−∞}u0>0 and track ζ_{0.05}(t) and ζ_{0.8}(t) for large t. If the difference ζ_{0.05}(t)−ζ_{0.8}(t) appears unbounded, Theorem 11.3(iii) is false. If it remains bounded, that supports the missing assertion, though a rigorous proof would require showing convergence of v(t,·) to a translate of the forced wave from [12] in the moving frame.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 11.3(iii) is the core lower bound in the forced-wave regime (2√(1−a)<β<2, η≥0): it claims ξδ(t)≥βt−ηlog(t+1)−C for every δ∈(0,1−a). The proof of Lemma 11.8 constructs a stationary subsolution v(y) with amplitude κφ(y*) and explicitly concludes only for δ∈(0,κφ(y*)). The final sentence of the lemma then states: 'In fact, (84) still holds for any δ∈(0,1−a) (with T=T(δ)) due to the fact that ζ_{δ1}(t)−ζ_{δ2}(t)=O(1) for any δ1,δ2∈(0,1−a).' No proof, citation, or argument is supplied. This is not a minor gap: the direct construction caps the reachable δ by σ = 1−β²/4−1/ℓ², which can be far below 1−a (e.g. if β²/4>a, then σ<1−a regardless of ℓ). The O(1) level-set difference is precisely what must be proved to extend the lower bound. It is not an immediate consequence of the comparison argument already used, since the solution has not been shown to converge to a single translate of a forced wave. If ζ_{δ1}(t)−ζ_{δ2}(t) were unbounded, the lower bound would hold only for small thresholds and the headline conclusion that the front stays within bounded distance of X(t) would fail. The same unproved fact is needed for Corollary 11.4, whose proof invokes a 'suitably modified' version of (ii).","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript is a survey of spreading phenomena in reaction-diffusion equations in shifting habitats, organized around the Hamilton-Jacobi approach. It reviews the Shigesada-Kawasaki conjecture, the reduction of competition systems to scalar shifting-habitat models, local versus nonlocal selection of spreading speeds, flux-limited viscosity solutions for non-monotone environments, spreading in competition and predator-prey systems, and entire solutions. The paper's original contribution is in Section 11.2: Theorem 11.3 gives Bramson-type logarithmic corrections for the level set xi_delta(t) when the region ahead of the shifting interface is less favorable, with Lemmas 11.6-11.8 providing the proof. The claimed result pins the level set to within O(1) of the interface in the intermediate regime 2*sqrt(1-a) < beta < 2 and gives the classical (3/(2*lambda_min)) log t delay when beta < 2*sqrt(1-a).","tokens_in":39171,"tokens_out":6051,"duration_ms":56977,"significance":"The survey is well-organized and serves a useful purpose in consolidating the recent literature, with clear statements of the Hamilton-Jacobi framework, nonlocal pulling, and open questions. If Theorem 11.3 is fully established, it is a significant new result: it extends Bramson's logarithmic correction to shifting habitats with a worse region ahead and shows that in the forced-wave regime the invasion front is pinned within bounded distance of X(t). The proof strategy is natural, combining comparison with homogeneous KPP dynamics, a forced-wave supersolution, and a stationary subsolution, and Lemmas 11.6 and 11.7 are convincing. However, the manuscript has a load-bearing gap in Lemma 11.8 that prevents Theorem 11.3(iii) and Corollary 11.4 from being fully supported as written.","major_comments":[{"comment":"The proof of Lemma 11.8 constructs a stationary subsolution v with amplitude kappa*phi(y*) and explicitly yields the lower bound zeta_delta(t) >= -C only for delta < kappa*phi(y*). The final sentence then asserts (84) for every delta in (0,1-a) 'due to the fact that zeta_{delta_1}(t)-zeta_{delta_2}(t)=O(1)', but no proof, citation, or argument is supplied. This fact is exactly what is needed to pass from the small-threshold construction to the full statement of Theorem 11.3(iii). It is not an immediate consequence of the comparison principle already used, because the solution has not been shown to converge to a single translate of a forced wave. Since (iii) is the core lower bound in the regime 2*sqrt(1-a)<beta<2, the theorem currently lacks support for all delta. Please provide a proof or a precise reference.","section":"Section 11.2, Lemma 11.8 (final sentence)"},{"comment":"The proof of Corollary 11.4 says that 'the argument of (ii) in Theorem 11.3 can be suitably modified' to cover beta=2*sqrt(1-a), eta=0. This modification is not shown. Lemma 11.7 requires beta>2*sqrt(1-a) to obtain lambda_0>0 and a non-minimal forced wave U_beta with exponential decay; at beta=2*sqrt(1-a), lambda_0=0 and the construction as written breaks down. Since the corollary is the sharp 'bounded distance from X(t)' statement at the boundary of the forced-wave regime, the required modification needs to be spelled out in detail.","section":"Section 11.2, Corollary 11.4"}],"minor_comments":[{"comment":"The running title contains a typo: 'INV ASION' should be 'INVASION'.","section":"Title/header"},{"comment":"The lower-case letter v is used both for the moving-frame solution v(t,y) and for the stationary subsolution v(y). This is confusing; consider renaming the subsolution (e.g., w or phi).","section":"Section 11.2, Lemma 11.8"},{"comment":"The support condition 'y in (-infinity, -L_0 + pi*ell) subset (-infinity, 0)' relies on L_0 > pi*ell; this is chosen, but the sentence 'By the choice of L_0' could be more explicit about the ordering of the constants.","section":"Section 11.2, Lemma 11.8"},{"comment":"The notation for the logarithmic correction in the transition case beta=2*(sqrt(a)+sqrt(1-a)) is not fully consistent: Theorem 11.2(iii) defines m_q(t) with q=-3/2+eta*sqrt(a), but the display after the theorem gives the special case eta=0 with coefficient -5/(4*lambda_min). Unifying these formulas would improve readability.","section":"Section 11.1, Theorem 11.2 and following display"}],"recommendation":"major_revision","confidential_remarks":"The survey portion is useful and the new result, if the gap is repaired, is a nice contribution to the subject. The main issue is the unproved level-set difference assertion in Lemma 11.8, which is used to extend the lower bound to all delta in (0,1-a). I would ask the authors to supply a complete proof or a precise reference for that assertion, and to spell out the 'suitably modified' argument in Corollary 11.4. The survey relies heavily on the authors' own previous works, which is acceptable but should be kept in proportion for a survey."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know about this paper if you work on shifting-habitat reaction-diffusion models: it's a broad survey of the Hamilton-Jacobi approach, and it contains a new result (Theorem 11.3) on front-location bounds in the harder 'worse region ahead' case. Previous work only gave spreading speeds; this upgrades them to O(1) level-set bounds in some regimes. The proof is a standard comparison argument, and the main inequalities in Lemmas 11.6 and 11.7 check out. That part is worth taking seriously.\n\nThe soft spot is in Lemma 11.8. The direct subsolution construction only establishes the lower bound for small δ. The final sentence then asserts the bound for every δ∈(0,1−a) using ζ_{δ1}(t)−ζ_{δ2}(t)=O(1), with no proof or citation. That fact is not an immediate consequence of what came before, and it's needed for both Theorem 11.3(iii) and Corollary 11.4. If it fails, the lower bound only holds for small thresholds and the headline claim about bounded distance from X(t) doesn't follow. This is a genuine gap, though probably fixable. Corollary 11.4's 'suitably modified' argument for the boundary case β=2√(1−a) is similarly thin.\n\nThe survey's reliance on the authors' own prior work is heavy, but that's a feature of this niche; the new theorem doesn't depend on the speed formulas being proved in this paper. The open questions about β≥2 and the critical case are honestly stated.\n\nIf I were the editor, I'd send this to peer review with a request that the authors either prove the level-set difference or restrict the statement to small δ. The survey is useful, and the new theorem is plausible; it just needs to be fully supported.\n\nAll best.","headline":"Useful survey with a genuinely new but modest front-location theorem; the proof has a real gap in Lemma 11.8 that should be fixed before publication.","tokens_in":39678,"tokens_out":3638,"would_cite":true,"duration_ms":35674,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35K57","92D25","35F21","35B40","35Q92"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper surveys Hamilton–Jacobi methods for invasion spreads in shifting habitats and proves a new result: at intermediate habitat-shift speeds the invasion front is pinned to within bounded distance of the moving habitat edge, while a sl","keywords":["reaction-diffusion equations","invasion fronts","shifting habitats","Hamilton-Jacobi equations","spreading speeds","logarithmic correction","nonlocal pulling","Fisher-KPP equation"],"falsifier":"Solve the moving-frame equation v_t=v_yy+(β−η/(t+1))v_y+v(g(y)−v) with g(y)=1 for y≤0 and g(y)=1−a for y>0, using compactly supported initial data satisfying (62), for a fixed β∈(0,2) and η≥0. Track ζ_{δ1}(t)−ζ_{δ2}(t) for two levels δ1,δ2∈(0,1−a) up to large t. If this difference is unbounded as t→∞, the level-set gap assertion fails and the lower bound in Theorem 11.3(iii) does not hold for all δ.","tokens_in":38620,"feed_emoji":"🌱","tokens_out":7642,"duration_ms":63545,"temperature":0.7,"pith_summary":"Drawing together recent work on spreading phenomena in reaction–diffusion models of ecological invasion, this paper develops and applies the Hamilton–Jacobi (WKB) approach to environments whose quality shifts in space and time. Its new contribution concerns a Fisher–KPP population in which the habitat ahead of the front is less favorable (growth rate 1−a) and the favorable region retreats along a curve X(t)=βt−η log(t+1). The authors prove that the δ-level set of the invasion front, ξδ(t), satisfies sharp two-sided bounds: for slow habitat shifts (β<2√(1−a)) the front lags by (3/(2λ_min)) log t, the classical logarithmic delay; for intermediate shifts (2√(1−a)<β<2) the front stays within O(1) of the habitat edge; and for any β<2 with η≥0 it never lags behind the habitat edge. This pins down the precise front location in a regime where only speeds were known.","feed_headline":"Invasion fronts pin to shifting habitat edge","feed_subtitle":"New bounds show a log-t delay for slow habitat shifts and O(1) pinning when the shift is faster.","key_machinery":"The key object is the invasion front's level set ξδ(t) together with the Hamilton–Jacobi rate function w(t,x)=lim_{ε→0} −ε log u(t/ε,x/ε), whose zero set gives the occupied region. The new proof works in the moving frame y=x−βt+η log(t+1), where the equation becomes v_t=v_yy+(β−η/(t+1))v_y+v(g(y)−v) with g(y)=1 for y≤0 and g(y)=1−a for y>0. The upper bound in (ii) is obtained by gluing a non-minimal forced wave for the shifted problem to an exponential tail, producing a generalized supersolution; the lower bound in (iii) uses a compactly supported sinusoidal subsolution in the region where the growth rate is the larger one. These comparisons convert the shifting-habitat problem into a sequen","core_discovery":"The paper's central new result, Theorem 11.3, establishes that for the reaction–diffusion equation u_t = u_xx + u(r(t,x)−u) with r(t,x)=1 for x ≤ X(t) and r(t,x)=1−a for x > X(t), X(t)=βt−η log(t+1), the level set ξδ(t)=sup{x:u(t,x)≥δ} satisfies: (i) if β<2√(1−a), then ξδ(t)=2√(1−a)t − (3/(2λ_min)) log t + O(1) with λ_min=√(1−a); (ii) if 2√(1−a)<β<2, then ξδ(t)≤βt−η log(t+1)+O(1); and (iii) if 0<β<2 and η≥0, then ξδ(t)≥βt−η log(t+1)−O(1). In words, when the favorable region retreats slowly the classical logarithmic delay persists; when it retreats at an intermediate speed the invasion front is confined to a bounded neighbourhood of the moving habitat edge; and a logarithmic retarding of the","pith_inferences":["A natural extension would be to prove profile convergence: in the intermediate regime the solution should approach a forced traveling wave with a bounded (possibly t-dependent) phase, analogous to the classical logarithmic convergence result.","The same comparison strategy may work for smooth monotone growth profiles r(x−c1t) instead of a step function, but the level-set gap assertion used in Lemma 11.8 would need to be re-established in that generality.","The paper's open case β=2, 1/2≤η<3/2 could be probed numerically: if the supercritical formula continues to hold, the front should exhibit the coefficient (1/λ∗)(3/2−√a η) log t; a direct simulation of the level set would test this.","If the unproved assertion that δ-level sets in the moving frame remain a bounded distance apart fails, Theorem 11.3(iii) would only be valid for small δ; this is checkable by computing ζ_{δ1}(t)−ζ_{δ2}(t) for two different levels."],"forward_implications":["If the habitat shift speed is below the minimal KPP speed of the less favorable region, the invasion front's position is 2√(1−a)t − (3/(2λ_min)) log t + O(1): the classical logarithmic delay survives the moving environment.","If the habitat shift speed lies strictly between 2√(1−a) and 2, the front cannot outrun the habitat edge: ξδ(t) is bounded above by X(t)+O(1).","If the habitat edge is retarding (η≥0) and shifts at any speed β<2, the front cannot lag behind it: ξδ(t)≥X(t)−O(1). Together with (ii) this pins the front to within O(1) of the edge when η=0 and 2√(1−a)≤β<2.","When the habitat shift is linear (η=0) and β∈[2√(1−a),2), the front position is βt+O(1): the invasion speed equals the environmental shift speed exactly.","These bounds refine Hamilton–Jacobi speed results into front-location asymptotics, connecting the spreading-speed picture to the logarithmic-correction literature."],"fun_headline_variants":["Habitat shift speed decides invasion front lag or pinning","Slow shifts cause log delay; fast shifts pin front to edge","Invasion front: log delay for slow retreat, pinning for fast","Habitat retreat speed sets front lag or edge pinning","Front lags with log delay or pins to edge as shift speed varies"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that, for the solution in the moving frame, the distance between any two δ-level sets ζ_{δ1}(t) and ζ_{δ2}(t) (with δ1,δ2∈(0,1−a)) remains bounded uniformly in time; this is stated without proof in Lemma 11.8 and is needed to pass from small δ to all δ in the lower-bound statement.","fun_headline_variants_meta":{"raw":{"variants":["Habitat shift speed decides invasion front lag or pinning","Slow shifts cause log delay; fast shifts pin front to edge","Invasion front: log delay for slow retreat, pinning for fast","Habitat retreat speed sets front lag or edge pinning","Front lags with log delay or pins to edge as shift speed varies"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000223,"raw_usage":{"total_tokens":1325,"prompt_tokens":807,"completion_tokens":518,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":551,"completion_tokens_details":{"reasoning_tokens":430}},"tokens_in":551,"tokens_out":518,"duration_ms":5273,"temperature":1.0,"reasoning_tokens":430,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T15:36:46.263439+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Solve the moving-frame equation v_t=v_yy+(β−η/(t+1))v_y+v(g(y)−v) with g(y)=1 for y≤0 and g(y)=1−a for y>0, using compactly supported initial data satisfying (62), for a fixed β∈(0,2) and η≥0. Track ζ_{δ1}(t)−ζ_{δ2}(t) for two levels δ1,δ2∈(0,1−a) up to large t. If this difference is unbounded as t→∞, the level-set gap assertion fails and the lower bound in Theorem 11.3(iii) does not hold for all δ.","supporting_citations":[],"review_version":1}