{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp05_complete_proposal_portfolio20_20260803","cell_id":"invariant_mode_decomposition_design__biology_ecology","arm":"COMPLETE_PROPOSAL_PORTFOLIO","candidate_id":"cand_bioeco_amphibian_modal_sentinel_003","proposal_index":3,"version":0,"title":"Mode-Aligned Sentinel Allocation for Amphibian Pathogen Surveillance","problem":"A wildlife-health program allocates pathogen assays according to recent positive counts, pond accessibility, or individual host abundance. However, pathogen burden can move jointly among larval, juvenile, and adult hosts, environmental reservoirs, and connected ponds. Aggregate detections may remain steady while a coordinated host-stage–reservoir–pond direction grows. Sampling the largest visible coordinate can consequently leave the action-relevant transmission pattern poorly observed.","actors":["Wildlife-health surveillance authority","Amphibian conservation managers","Field sampling crews","Diagnostic laboratory personnel","Animal-welfare reviewer","Managers and landholders responsible for sampled ponds","Amphibian hosts represented in the surveillance state"],"observable_state":"At fixed intervals within a declared breeding-season phase, the program records quantitative assay results or detection probabilities for a focal amphibian pathogen in larval, juvenile, and adult host groups and in water or sediment samples at each pond, together with specified connectivity variables. An estimated local transition matrix maps deviations in this host-stage–reservoir–pond vector from one interval to the next. Decision signals include modal composition, gain, stability class, node loading, sampling sensitivity, modal-coordinate uncertainty, eigenvector conditioning, held-out reconstruction residual, spectral separation, and mode drift.","consequence":"A count-led surveillance schedule could repeatedly assay accessible ponds or currently positive adult hosts while undersampling the combination of environmental reservoir, larval host stage, and connected ponds along which the pathogen state is changing. The authority could then lack a defensible basis for deciding whether a signal is localized, fading, persistent, or spreading through the monitored system.","affected_objective":"Allocate a fixed field-handling and assay budget so that any reproducible growing or weakly damped pathogen mode is observable with bounded uncertainty while maintaining minimum geographic coverage, animal-handling limits, laboratory quality controls, and explicit model-validity checks.","intervention":"Implement a mode-gated sentinel-allocation protocol. Begin with uniformly structured pilot sampling over a small, predefined pond network; define the host-stage–reservoir state and estimate its local transition matrix; compute the complete eigenbasis and gain spectrum; classify modes by stability; and map the entries of action-relevant eigenvectors back to ponds, host stages, and environmental reservoirs. A modal sensitivity sweep reallocates one sampling slot at a time among water assays, sediment assays, and capped nonlethal host swabs, measuring the resulting change in uncertainty and reconstruction of the targeted modal coordinate while logging losses of coverage elsewhere. The selected schedule preserves a fixed minimum sample allocation at every pond and assigns only the remaining budget according to modal observability. It is suspended when the basis is ill-conditioned, the relevant spectral gap is inadequate, retained modes fail held-out reconstruction, mode identity drifts, assay missingness becomes informative, or observations leave the stated seasonal regime.","structural_mapping":[{"archetype_element":"Transformation Scope","domain_realization":"The transformation is the fixed-interval local transition of pathogen measurements among amphibian life stages, environmental reservoirs, and connected ponds during a bounded breeding-season phase."},{"archetype_element":"State-Vector Definition","domain_realization":"Each coordinate identifies a pond and either a host stage or environmental reservoir, preventing aggregate positives from substituting for the coupled surveillance state."},{"archetype_element":"Invariant Mode Basis and Modal Gain Spectrum","domain_realization":"Eigenvectors describe coupled host-stage–reservoir–pond patterns approximately preserved as directions by the fitted transition, while eigenvalues describe their estimated interval-to-interval growth, decay, persistence, reversal, or oscillation."},{"archetype_element":"Stable/Unstable Mode Partition","domain_realization":"Modes are classified relative to the discrete-time stability boundary within the declared seasonal window, with uncertain near-boundary modes retained as marginal."},{"archetype_element":"Dominant Mode Selection Rule","domain_realization":"A mode becomes sampling-relevant only when it is reproducible, sufficiently separated and conditioned, relevant to a predefined surveillance consequence, and inadequately observed by the minimum-coverage schedule."},{"archetype_element":"Modal Intervention Map","domain_realization":"Sampling slots are treated as controllable inputs whose placement changes the uncertainty and reconstructability of modal coordinates rather than changing the pathogen process itself."},{"archetype_element":"Reconstruction Residual and Mode-Coupling Checks","domain_realization":"Held-out dense-reference observations test whether retained modes reconstruct the pathogen state, while the coupling register records near-degenerate modes and sampling reallocations that improve one mode but obscure another."},{"archetype_element":"Spectral Gap, Drift, and Interpretation Scope","domain_realization":"The protocol tracks retained-versus-omitted spectral separation and mode rotation across intervals and restricts interpretation to the sampled ponds, host stages, assay types, and breeding-season conditions."}],"mechanism_mapping":[{"mechanism_slug":"eigendecomposition_workflow","role":"Factor the explicit host-stage–reservoir–pond transition matrix into a complete local mode basis and gain spectrum traceable to surveillance coordinates.","counterfactual_removal":"Without this mechanism, the program could fit multivariable trends but could not identify the coupled directions whose observability should guide sentinel allocation."},{"mechanism_slug":"modal_stability_analysis","role":"Classify each pathogen mode as growing, decaying, marginal, or oscillatory within the declared seasonal window.","counterfactual_removal":"Without stability classification, a strongly observed but fading pattern could be prioritized over a less visible direction requiring closer surveillance."},{"mechanism_slug":"network_spectral_centrality_analysis","role":"Read the pond, host-stage, and reservoir entries of an action-relevant mode as structural loadings for constructing a candidate sentinel map, subject to localization and coverage checks.","counterfactual_removal":"Without node-level mode loadings, the decomposition would not translate into identifiable sampling locations and biological compartments."},{"mechanism_slug":"modal_sensitivity_sweep","role":"Reallocate sampling slots in turn and measure their effect on uncertainty and reconstruction of the targeted mode while registering cross-mode losses.","counterfactual_removal":"Without the sweep, high-loading nodes might be oversampled even when an additional assay there contributes little observability or erodes coverage of another mode."},{"mechanism_slug":"residual_reconstruction_test","role":"Compare states reconstructed from the retained modes and proposed sentinel subset with independent dense-reference assay results.","counterfactual_removal":"Without reconstruction testing, a compact sentinel schedule could appear efficient while systematically omitting a low-amplitude host stage or reservoir pathway."},{"mechanism_slug":"spectral_gap_monitor","role":"Track separation, rotation, and reordering of retained modes across successive sampling intervals.","counterfactual_removal":"Without gap and drift monitoring, sampling could remain concentrated on a mode after its identity becomes unstable or secondary modes become equally relevant."},{"mechanism_slug":"spectral_decomposition_report","role":"Communicate modal loadings, gains, uncertainty, couplings, residuals, coverage constraints, animal-handling limits, and prohibited interpretations.","counterfactual_removal":"Without a bounded report, descriptive surveillance modes could be mistaken for proven transmission routes or direct authorization for disease-control actions."}],"causal_chain":["Pathogen measurements change jointly among amphibian host stages, environmental reservoirs, and connected ponds.","A locally estimated transition matrix represents those coupled interval-to-interval changes within a declared seasonal regime.","Decomposition exposes combinations that approximately preserve their direction while growing, decaying, persisting, or oscillating.","Stability, conditioning, consequence relevance, and spectral-separation rules identify whether any mode warrants additional observability.","Node loadings locate the ponds and biological compartments carrying that mode, while sensitivity sweeps determine where an additional assay most changes modal uncertainty.","A fixed minimum-coverage layer protects against tunnel vision, and only the discretionary sampling budget is shifted toward the mode-aligned sentinel set.","Dense-reference reconstruction, assay quality, welfare observations, spectral drift, and residual structure determine whether the allocation is retained, revised, or withdrawn."],"baseline":"The baseline is a fixed stratified schedule with additional samples assigned in proportion to recent positives, host counts, and field accessibility. It maintains routine coverage but does not estimate coupled invariant directions, classify their dynamics, measure modal observability, or condition allocation on spectral separation and reconstruction residuals.","nearest_rivals":["A habitat-risk score that ranks ponds from environmental covariates and recent detections without modeling the transition among host stages and reservoirs.","A fixed stratified design that samples every pond and host stage at constant frequency regardless of changing coupled dynamics.","An adaptive design that assigns the next assay where the predicted probability of a positive result is highest, without distinguishing modal growth from current burden.","A connectivity-degree or centrality ranking based only on pond links, without incorporating pathogen-state transitions or host-stage coordinates.","A spatiotemporal predictor optimized for aggregate held-out assay accuracy without a modal interpretation, minimum-coverage layer, or spectral-drift gate."],"remaining_contrastive_claim":"The candidate's testable contrast is procedural: it allocates only the discretionary portion of a fixed sampling budget according to the observability of a reproducible action-relevant transition mode, while conditioning use on stability, spectral separation, coupling, reconstruction, drift, assay quality, and minimum coverage. Rivals allocate from current positives, habitat risk, connectivity, fixed strata, or aggregate predictive accuracy. This contrast does not establish earlier detection or superior surveillance performance.","authority_safety":{"decision_authority":"The wildlife-health surveillance authority owns sampling allocation. The animal-welfare reviewer sets nonlethal handling limits, and the diagnostic laboratory controls assay acceptance. Separate conservation or public-health authorities retain all power over closures, treatment, movement restrictions, translocation, or public notification.","authorized_first_step":"Run a bounded observational pilot across a predefined pond set. During each interval, collect a minimum-coverage panel plus blinded duplicate environmental samples and capped nonlethal swabs sufficient to create a dense reference. Use the first phase for operator identification, freeze the allocation rule, and evaluate the baseline and mode-aligned assay subsets against reference results from the later phase.","excluded_actions":["No culling, treatment, habitat alteration, pond closure, translocation, or movement restriction based on the pilot.","No intentional pathogen exposure or movement of animals, water, sediment, or biological material among ponds.","No host handling beyond approved species-, stage-, and interval-specific caps.","No removal of the minimum geographic coverage layer to concentrate the entire budget on high-loading nodes.","No extrapolation to unsampled ponds, host species, life stages, pathogens, assays, or seasonal regimes.","No public disease declaration or operational control trigger based solely on a modal estimate.","No suppression of assay failures, informative missingness, ill-conditioning, near-degenerate modes, structured residuals, welfare incidents, or null results."],"halt_rollback":"Immediately stop host sampling at an affected site after an injury, abnormal mortality, handling-limit breach, biosecurity breach, or unexpected animal-stress signal; revert to approved environmental sampling or observation-only status. Withdraw the modal allocation and restore the fixed stratified baseline if quality controls fail, missingness is associated with modal loadings, eigenvectors are ill-conditioned, the targeted mode is not reproducible, the spectral gap falls below its preregistered threshold, mode drift exceeds tolerance, held-out residuals exceed budget or retain biological structure, or observations leave the interpretation window."},"negative_tests":{"strongest_counterevidence":"In the blinded validation phase, the mode-aligned subset does not reconstruct dense-reference modal coordinates or their direction of change more reliably than equal-budget fixed stratification, recent-positive allocation, or a direct spatiotemporal predictor, while the inferred modes remain unstable across resampling or intervals.","problem_falsifier":"Dense-reference observations show that the surveillance consequence is adequately represented by a single pond or host-stage measurement, with no reproducible coupled host-stage–reservoir–pond direction that changes independently of aggregate positives.","intervention_falsifier":"At the same assay and handling budget, the preregistered mode-aligned schedule does not reduce uncertainty or improve reference concordance for the targeted modal coordinate relative to the fixed-stratified baseline, or it misses a reference signal, violates minimum coverage, increases welfare events, or causes another protected mode to become unobservable.","risks":["Sparse detections and assay censoring can make transition estimates unstable.","Detection probability may change independently of pathogen burden across host stages, ponds, or weather conditions.","A non-normal operator can produce transient amplification not summarized by eigenvalue stability alone.","Closely spaced modes can localize, rotate, or exchange identity and create unstable sentinel rankings.","Sampling high-loading host groups can alter welfare burden even when total sample count is fixed.","Incomplete connectivity data can assign structural importance to the wrong ponds.","A dense reference may itself remain an imperfect representation of the underlying pathogen state.","Mode-focused allocation can miss an emerging direction absent from the identification window.","Modal descriptions may be misread as causal transmission pathways rather than bounded surveillance constructs."]},"next_evidence_step":"Conduct one preregistered breeding-season pilot with an identification phase and a temporally held-out validation phase. Collect the approved minimum-coverage panel at every interval and preserve blinded duplicate samples so the laboratory can later construct a dense reference without changing field actions. Before validation assays are opened, freeze the state vector, transition interval, scaling, mode-matching rule, conditioning limit, stability boundary, spectral-gap threshold, residual budget, welfare caps, minimum-coverage layer, modal sensitivity objective, and competing equal-budget schedules. Compare each schedule's reconstruction and uncertainty for the targeted modal coordinate against the dense reference and record all missed signals, assay failures, welfare events, and mode drift. The first evidence question is whether a reproducible mode exists and whether its observability changes as predicted under reallocation; no pathogen-control or field-effect claim follows.","prior_art_status":"UNSEARCHED","diversity_from_prior_proposals":"Proposal 1 manages coupled bloom and hypoxia dynamics in a shallow lake by changing nutrient input and mechanical mixing. Proposal 2 suppresses an invasive annual plant's seedbank-replenishment mode by scheduling seed capture, seedling removal, and pre-seed-set clipping. This proposal does not manipulate an ecosystem process or remove an organism. It reallocates a fixed diagnostic sampling budget to observe a coupled wildlife-pathogen mode across host stages, reservoirs, and ponds. Relative to proposal 1, it has a disease-surveillance objective, host-reservoir transition state, assay-allocation intervention, minimum-coverage architecture, animal-handling safeguards, and blinded reference-panel evidence path rather than water-quality control. Relative to proposal 2, it targets observability rather than population suppression, uses diagnostic samples rather than manual lifecycle treatments, and is governed by wildlife-health, laboratory-quality, and animal-welfare authorities rather than invasive-plant control and seed containment. It is independently adoptable as a surveillance protocol and is not a feature or implementation variant of either earlier proposal.","revision_record":{"parent_version":null,"progress_targets_addressed":["Created a third complete candidate addressing a materially different biological and ecological problem.","Specified a surveillance-allocation intervention and observability causal path distinct from the two earlier control interventions.","Explained candidate-level diversity from proposals 1 and 2.","Included authority boundaries, welfare and biosecurity safeguards, rivals, falsifiers, and bounded evidence."],"conceptual_changes":[],"operational_changes":[],"evidence_changes":[],"claim_changes":[]}}