{"id":"a14e898c-828a-4976-9862-90dc60696eb9","arxiv_id":"1908.02479","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"In a Lotka-Volterra model with one adaptive animal, exploiter switching can stabilize coexistence of strongly competing plants, while mutualists always specialize and allow coexistence only under weak competition.","lead":"Adaptive animal foraging preferences can change whether two competing plant species coexist, even when animal numbers are fixed. This paper fully classifies the possible outcomes when a shared exploiter or mutualist switches preferences to maximize its fitness.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Global-stability and completeness claims in Sections 3.1-3.4 are not backed by a proof excluding periodic orbits in the piecewise-smooth Filippov system.","rationale":"I read the paper as a mathematical classification of outcomes for a two-plant, one-animal module with fixed animal density and preference set by instantaneous fitness maximization. The model is a planar differential inclusion, and the authors introduce generalized isoclines to analyze it. I checked the derivations in Appendices A.1-A.2: the sector equilibria EI/EII, the switching-line equilibrium ES, the conditions (14),(18),(21),(22), and the stability conclusions for ES (sliding vs. repelling) are internally consistent. The classification in Table A.2 follows from the relative positions of the intersection points a,b,p,q and the boundary equilibria. The main gap I find is the step from local stability to global stability/completeness. The paper asserts global convergence for several cells without a proof excluding periodic orbits. In piecewise-smooth systems, local stability of all equilibria does not guarantee absence of stable limit cycles, and the linearity of the subsystems does not automatically exclude sliding cycles. Since the abstract and Section 3 sell a complete classification and global outcomes, this gap affects the strength of the central claim. I do not think the paper should be rejected: the local coexistence predictions (e.g., stable generalist ES under strong exploitation) are sufficient for the headline result that adaptive foraging can promote coexistence. But the verdict should be conditional on either supplying a rigorous global-stability argument or softening the global/completeness claims. I considered the reader's weaker assumption (instantaneous switching); that is a modeling limitation explicitly acknowledged in the text and in the discussion, and it delimits the scope rather than undermining the internal argument. The limit-cycle gap is not acknowledged and is internal to the mathematics.","tokens_in":28241,"tokens_out":22019,"duration_ms":230349,"concrete_test":"Implement the differential inclusion (or approximate it with the Hill preference (A.21) for steepness z=20 and z=50) and simulate from a dense grid of initial conditions for one representative parameter tuple in each non-empty cell of Table A.2, focusing on the cells where global stability is claimed (Figure 5c,f,h,i; Figure 6c,f,g,i; Figure 7e,f,h). Use a Poincaré section (e.g., the switching line) to identify any attracting periodic orbit not listed in the table. If every trajectory converges to the tabulated equilibrium, the global-stability claim is supported; if any stable limit cycle appears, the classification is incomplete.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The paper's central classification (Table A.2, Figures 5-8) asserts not just local equilibria but global dynamics: e.g., 'globally stable equilibrium EI' (Section 3.1, Figure 5f), 'ES is globally stable' under strong exploitation (Section 3.2, Figure 6i), and the 'complete classification of model outcomes'. Local stability of each equilibrium is rigorously derived in Appendices A.1-A.2. However, global stability in a planar Filippov system also requires ruling out periodic orbits that cross or slide along the switching line e1P1=e2P2. The subsystem on either side of the switching line is a linear Lotka-Volterra competition system, which alone has no cycles, but this does not rule out sliding cycles born from the coupling of two linear vector fields at the discontinuity; such cycles are known to occur in piecewise-linear systems. The paper offers no Dulac function, no Lyapunov function, and no explicit argument excluding limit cycles. Thus the 'complete classification' is, strictly, a classification of stable equilibria; the claim that all generic orbits converge to one of the listed equilibria is an unproven assumption. This does not invalidate the local coexistence results, but it is the most load-bearing gap in the paper's mathematical argument.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies a two-plant, one-animal model in which plant dynamics follow Lotka-Volterra competition, animal density A is fixed, and animal preferences u_i follow the stepwise optimal foraging rule (5), yielding a planar differential inclusion. The authors introduce 'generalized isoclines' by connecting the sector-wise plant isoclines with segments on the switching line e1P1 = e2P2. They derive explicit coexistence equilibria EI, EII, and ES (Eqs. 13, 17, 19), invasion thresholds γ_i and attraction thresholds τ_i (Eqs. 14, 18), and present a classification of stable equilibria for exploitation (s = -1) and mutualism (s = 1) under weak and strong competition (Figures 5-8, Table A.2). The central claims are that adaptive exploiters can stabilize coexistence even when c1c2 > 1, whereas adaptive mutualists always specialize at stable coexistence and preclude coexistence when c1c2 > 1.","tokens_in":28406,"tokens_out":9951,"duration_ms":111323,"significance":"The manuscript is a theory contribution with no free parameters and no fitted constants. Its strengths are the explicit threshold formulas, the fully worked local stability analysis in Appendices A.1-A.2, the classification table, and the new generalized-isocline construction that makes the preference discontinuity analytically tractable. The predictions—exploiter generalism promotes coexistence under strong competition; mutualist specialization leads to alternative stable states under weak competition—are clear, ecologically relevant, and falsifiable in principle. The main caveat is that several headline statements concern global asymptotic behavior, while the proof infrastructure supports local stability plus sector-wise linear dynamics. The significance is therefore high if the global claims can be proven or appropriately qualified.","major_comments":[{"comment":"The global-dynamics claims are not backed by a proof. Appendix A.1 establishes local stability of EI and EII from the Lotka-Volterra Jacobian, and Appendix A.2 establishes local attractivity of ES through the sliding-regime logistic equation (A.20), but these are local statements. The text nevertheless asserts global convergence, e.g., 'dynamics globally converge toward the monoculture equilibrium' in Section 3.1, 'ES is globally stable' in Section 3.2 (Figure 6i), and a 'complete classification of model outcomes' in Section 4. The system is a planar Filippov differential inclusion with switching line (6); in each sector the fields (A.1)-(A.2) are linear competition fields, but piecewise-linear Filippov systems can have closed orbits that extend across or slide along the switching line, and the absence of cycles in each linear subsystem does not exclude such orbits. No Dulac/Bendixson argument, Lyapunov function, or explicit cycle-exclusion proof for the full piecewise-smooth system is provided. Therefore the stated global stability and completeness results are, strictly, unproven; the proven content is a classification of locally stable equilibria. This does not affect the local coexistence conditions, but it is load-bearing for the abstract's and discussion's strongest claims.","section":"Sections 3.1-3.2, Appendix A.2, Table A.2"}],"minor_comments":[{"comment":"The caption says 'as a function of exploiter density' but the panel concerns adaptive mutualism (s = 1); it should read 'mutualist density'.","section":"Figure 4 caption"},{"comment":"Equation (5) defines U1 as a set-valued map, but later text uses u1 as if single-valued on the switching line; it would help to state explicitly that on the switching line the relevant u1 is the one selected by the sliding dynamics in Appendix A.2.","section":"Section 2.1"},{"comment":"Entries such as 'ba', 'pq', and 'bq' denote intervals on the switching line but are not defined in the table; please use notation like 'b-a' or define the abbreviated interval names in the caption.","section":"Table A.1"},{"comment":"The claim that gradual preferences 'well approximate' the step-like case is illustrated with only one Hill exponent (z = 20) and one parameter set; a brief quantitative statement or additional parameter exploration would strengthen the claim.","section":"Appendix D"}],"recommendation":"major_revision","confidential_remarks":"The only substantive obstacle is the global-stability proof gap. The local analysis and the qualitative predictions are valuable and likely correct, so I am not recommending rejection. The authors should either supply a rigorous global argument excluding limit cycles in the Filippov system, or carefully rephrase the claims as a classification of locally stable equilibria. If the latter route is chosen, the abstract and highlights should be revised so that 'complete classification' refers explicitly to stable equilibria under generic parameters."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, this is a real extension of the authors' earlier work: it adds explicit interspecific competition to the fixed-density adaptive-foraging model and produces a complete equilibrium map in terms of invasion and attraction thresholds. The generalized-isocline construction is a useful tool, and the main qualitative predictions—exploiter generalism can rescue coexistence under strong competition, mutualists specialize and give alternative stable states under weak competition—are biologically meaningful. Second, the paper is better on local stability than on global claims. The appendices derive equilibria and local stability correctly, and the sliding dynamics along the switching line (A.20) is a genuine contribution. But the text repeatedly says 'globally stable' and 'complete classification of model outcomes.' That global statement requires ruling out limit cycles in a piecewise-linear Filippov system, and no such argument appears. Each sector is a Lotka-Volterra competition system, which by itself has no cycles, but a trajectory can alternate sectors or slide along the switching line and produce a periodic orbit in a Filippov system. No Dulac function, Lyapunov function, or Poincaré-Bendixson argument is offered. This is not a fatal objection to the local coexistence results, but it is the load-bearing gap in the headline claim.\n\nThe weakest assumption—instantaneous, costless, perfect-information switching—is acknowledged in the Model assumptions section and discussed, so that is a scope limitation, not a flaw. The citations to the authors' own earlier papers are building blocks rather than fitted inputs, so the circularity burden is low.\n\nWho gets value: anyone working on trait-mediated indirect interactions, adaptive foraging, or coexistence theory. The paper deserves a serious referee. I would ask for a proof (or a stated conjecture) for global convergence, or a rewriting that replaces 'globally stable' with 'locally stable and numerically observed to be globally attracting.' As written, I would not trust the global statements without that fix, but the local classification and the isocline tool are solid and worth publishing.","headline":"A careful equilibrium classification of adaptive foraging in a two-plant, one-animal module; the local results are solid, but the global-stability claims outrun the proof.","tokens_in":29008,"tokens_out":3408,"would_cite":true,"duration_ms":41801,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["34A60","37N25","92D25"],"pacs":[],"model":"deepseek-v4-flash","headline":"Adaptive animal preferences determine whether two competing plants coexist, and the direction depends on whether the shared animal is an exploiter or a mutualist.","keywords":["behaviorally-mediated interactions","competition for preference","differential inclusion","generalized isocline","switching","sliding and repelling regimes","plant coexistence","adaptive foraging"],"falsifier":"Take a system of two plants with measured competition coefficients satisfying $c_1c_2>1$, add a fixed-density herbivore that is observed to forage adaptively, and choose carrying capacities so that both invasion thresholds $\\gamma_i$ are met; the paper predicts that the plants converge to a unique coexistence equilibrium $E_S$ at which the herbivore is a generalist, for all initial conditions. A controlled mesocosm, or a simulation with the same parameters, that shows persistent exclusion of one plant or stable coexistence while the herbivore remains a specialist at that parameter set would falsify the claim.","tokens_in":27986,"feed_emoji":"🌱","tokens_out":9245,"duration_ms":94914,"temperature":0.7,"pith_summary":"Two plants that compete for space can coexist through a shared animal even when their direct competition is too strong, provided the animal is an exploiter, such as a herbivore, that switches its preference to the more profitable plant. With the animal's density fixed, the paper shows that these behavioral, trait-mediated effects alone are enough: adaptive exploiters can make a strongly competing pair coexist at a globally stable equilibrium where the exploiter is a generalist, and can even let each plant survive at animal densities that would kill either one alone. Adaptive mutualists, such as pollinators, have the opposite effect: at any locally stable coexistence equilibrium the mutualist specializes on one plant, and when the plants compete strongly, coexistence is impossible. The full classification is obtained through generalized isoclines, which combine the usual Lotka–Volterra isoclines in the two specialization sectors with the switching line where both plants are equally profitable.","feed_headline":"Flexible foraging lets strong plant rivals coexist","feed_subtitle":"Only with exploiters: adaptive mutualists always specialize and never stabilize strong competition.","key_machinery":"The central object is the generalized isocline, a continuous piecewise-linear curve obtained by taking the classical Lotka–Volterra isoclines inside each preference sector and adding the segment of the switching line $e_1P_1=e_2P_2$ that connects them. Along the switching line the animal's payoff is equal for the two plants and optimal preference is not unique, so the population dynamics form a differential inclusion whose Filippov regularization produces sliding regimes under exploitation and repelling regimes under mutualism. The generalized isoclines are classified by two thresholds each: invasion thresholds $\\gamma_i$, which decide whether the missing plant can invade the other's monoculture, and attraction thresholds $\\tau_i$, which decide whether the equilibrium $E_S$ on the switching line attracts or repels nearby orbits. Comparing $K_1/K_2$ with $\\gamma_1,\\tau_1$ and $K_2/K_1$ with $\\gamma_2,\\tau_2$ yields the complete set of stable equilibria.","core_discovery":"Under the step-like optimal-preference rule, the plant phase plane splits into two sectors in which animals are specialists, separated by the switching line $e_1P_1=e_2P_2$ where both plants are equally profitable and animal preference is not unique. The central discovery is that existence and stability of an interior equilibrium on this switching line, $E_S$, depend on two pairs of thresholds: invasion thresholds $\\gamma_i$ and new attraction thresholds $\\tau_i$. For exploiters ($s=-1$) and strong plant competition ($c_1c_2>1$), $E_S$ is stable and, when both plants can invade each other's monoculture, globally stable, so adaptive generalist exploiters promote plant coexistence that the Lotka–Volterra model forbids. For mutualists ($s=1$), $E_S$ is always unstable, so any locally stable coexistence has the mutualist as a specialist; with $c_1c_2>1$ no coexistence equilibrium exists at all. These statements rest on a complete enumeration of generic isocline configurations, which reduces all possible outcomes to 56 feasible cases and 11 possible sets of stable equilibria.","pith_inferences":["Beyond the paper: if animal density were allowed to vary numerically, apparent competition should reappear for exploiters and resource depletion should push mutualists toward generalism, so the predictions here should not be imported into models with numeric responses.","Beyond the paper: because gradual Hill-function preferences converge to the step-like rule as steepness grows, the listed stable equilibria are the limits of soft-switching models; a testable next step is whether finite switching costs or perceptual error shrink the basin of $E_S$ under strong competition without destroying it.","Beyond the paper: the classification suggests an empirical asymmetry—at fixed or managed animal densities, herbivores should more often appear as generalists at stable mixed plant stands, while pollinators should appear as specialists even when both plants are present.","Beyond the paper: adding a second animal species creates two switching lines, three sectors, and five-segment generalized isoclines, so the complete classification will not scale; simulation studies of larger networks could test whether the exploiter-generalist and mutualist-specialist asymmetry persists in multispecies mixtures."],"forward_implications":["If adaptive exploiters have fixed density, coexistence of two strongly competing plants is possible without any density-mediated apparent competition; the effect is purely behavioral.","At exploiter densities above a threshold, the only stable coexistence state has the exploiter as a generalist, and increasing animal density then decreases both plant equilibrium densities together.","Under adaptive mutualism, a pollinator that maximizes fitness will not remain a generalist at a stable coexistence equilibrium, so stable coexistence of weakly competing plants occurs as alternative states with the mutualist favoring one plant or the other.","With strong plant competition, adaptive mutualists cannot rescue coexistence, in sharp contrast to adaptive exploiters; the sign of the interaction determines whether flexible preferences buffer or amplify competitive exclusion.","The generalized-isocline construction provides a complete classification of all generic outcomes, reducing them to 56 feasible configurations and 11 possible sets of stable equilibria."],"supporting_citations":[{"why":"Supplies the optimal-foraging ideal-free-distribution framework with population dynamics that this model extends to two competing plants and to mutualism.","marker":"(Křivan, 2003b)"},{"why":"Provides the pollinator mutualism counterpart and the result that slower adaptation can change the coexistence predictions.","marker":"(Revilla and Křivan, 2016)"},{"why":"Provides the regularization of discontinuous vector fields used to define sliding and repelling regimes on the switching line.","marker":"(Filippov, 1988)"},{"why":"Supplies the differential-inclusion formulation for optimal foraging when animal preference is not unique.","marker":"(Colombo and Křivan, 1993)"},{"why":"Defines apparent competition, the density-mediated alternative that the fixed-density assumption deliberately removes.","marker":"(Holt, 1977)"},{"why":"Documents positive indirect effects between prey sharing a switching predator, a precursor of the exploiter-generalist facilitation shown here.","marker":"(Abrams and Matsuda, 1996)"},{"why":"Provides the standard Lotka–Volterra competition classification used as the within-sector baseline.","marker":"(Case, 2000)"}],"fun_headline_variants":["Exploiters with flexible tastes let plant rivals coexist","Generalist exploiters promote plant coexistence","Mutualist specialists can't fix strong competition","Adaptive foragers change plant coexistence odds"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument rides on the animal being an ideal, costless, instantaneous optimizer: preferences follow the step rule $u_1=1$ when $e_1P_1>e_2P_2$, $u_1=0$ when $e_1P_1<e_2P_2$, and are arbitrary on the switching line $e_1P_1=e_2P_2$, with animal density fixed. If real animals learn slowly, perceive profitability imperfectly, or pay switching costs, the sharp switching line and the sliding and repelling regimes, and with them the specialist and generalist coexistence predictions, need not hold.","fun_headline_variants_meta":{"raw":{"variants":["Exploiters with flexible tastes let plant rivals coexist","Generalist exploiters promote plant coexistence","Mutualist specialists can't fix strong competition","Adaptive foragers change plant coexistence odds"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000824,"raw_usage":{"total_tokens":3635,"prompt_tokens":1008,"completion_tokens":2627,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":624,"completion_tokens_details":{"reasoning_tokens":2570}},"tokens_in":624,"tokens_out":2627,"duration_ms":20606,"temperature":1.0,"reasoning_tokens":2570,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:43:31.136116+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a system of two plants with measured competition coefficients satisfying $c_1c_2>1$, add a fixed-density herbivore that is observed to forage adaptively, and choose carrying capacities so that both invasion thresholds $\\gamma_i$ are met; the paper predicts that the plants converge to a unique coexistence equilibrium $E_S$ at which the herbivore is a generalist, for all initial conditions. A controlled mesocosm, or a simulation with the same parameters, that shows persistent exclusion of one plant or stable coexistence while the herbivore remains a specialist at that parameter set would falsify the claim.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the pollinator mutualism counterpart and the result that slower adaptation can change the coexistence predictions."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the regularization of discontinuous vector fields used to define sliding and repelling regimes on the switching line."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the differential-inclusion formulation for optimal foraging when animal preference is not unique."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines apparent competition, the density-mediated alternative that the fixed-density assumption deliberately removes."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Documents positive indirect effects between prey sharing a switching predator, a precursor of the exploiter-generalist facilitation shown here."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the standard Lotka–Volterra competition classification used as the within-sector baseline."}],"review_version":1}