The Developmental Morphospace of Wildfire

Fire VASE changes the response being explained. Instead of reducing a wildfire to final area, duration, or peak growth, it treats the ordered accumulation of burned area as an observed developmental history that can be compared across events.

The current manuscript's central claim is precise: Fire VASE makes the ordered allocation of observed wildfire growth a robust, comparable developmental response. Its broad geometry persists across the tested observation-depth, temporal-order, and compositional-geometry comparisons. The method creates a shared response for asking why adequately observed fires follow different growth pathways; it does not claim that endpoint summaries, weather, or local neighborhood labels uniquely determine those pathways.

Read the assembled manuscript Read the assembled SI Browse every figure Read the guided figure walkthrough See how AI was used Inspect the corrected v2 evidence

The argument

The pathway matters, not only the endpoint

Two fires can reach similar final areas over similar durations through very different sequences: an early run followed by little change, persistent steady growth, a late surge, or repeated growth and quiescence. Fire VASE preserves those distinctions in a common geometry.

Relative developmental time runs upward through a VASE. Each observed slice adds a ring, and normalized cumulative burned area controls its width. The result is not a geographic perimeter or a fire-spread direction. It is a standardized record of when observed growth accumulated.

Developmental histories form a continuum

The manuscript describes gradients in the timing, concentration, persistence, and recurrence of observed growth. Named neighborhoods such as late surge, front-loaded plateau, or multi-pulse complex are interpretive landmarks, not discovered natural fire types. Intermediate and mixed pathways are part of the result.

The current 20-bin shape-only coordinate system uses 10,246 histories with at least three observations and no intervening calendar-day gaps. Its first axis explains 34.1% of standardized variance and the first five explain 89.4%. Positive, mass-conserving reference histories also compress, so compression by itself is not treated as evidence of biological restriction. The observed ordering and the meaning and stability of the broad axes carry the stronger developmental evidence.

Weather is an explanatory layer, not the definition

The developmental representation is constructed before weather is added. Weather is the first demonstration of a broader explanatory architecture in which fuels, terrain, active-edge conditions, ignition context, suppression, and the spatial geometry of progression can all be tested against the same developmental response.

The current evidence supports a restrained interpretation: weather associations are distributional and response-dependent. They do not uniquely specify an individual history, establish a universal growth response, or identify a causal mechanism.

State and mismatch generate better questions

Recent fire state explains substantially more subsequent observed growth than weather alone. Qualified weather-state interactions indicate that an exposure can have a different fitted association in different developmental contexts; they remain retrospective associations rather than operational or causal estimates.

Across 87,944 exact next-calendar-day transitions, the region-held-out autoregressive state baseline has R² about 0.448. Weather adds about 0.005 and the full set of weather-state interactions adds about 0.018 above that baseline. These increments are informative but small relative to recent-state predictability.

Likewise, fires that are close in weather space but distant in developmental space—and the reverse—are useful candidate comparisons. They show where the present representation is insufficient and where fuels, terrain, suppression, active-edge exposure, or observation error should be tested. They do not, by themselves, establish excess mismatch or its ecological prevalence.

Evidence boundary

The manuscript PDF articulates the project-level scientific argument across the full FIRED archive, then reports each result for its explicitly defined observation-supported cohort. The corrected v2 analysis is the repository authority for quantitative claims. It uses consecutively observed histories, a shape-only coordinate system, exact calendar-day transitions, day-specific newly burned-area exposure, and explicit null and stability tests. Its results narrow—but preserve—the central contribution: broad developmental gradients are useful and reproducible, while weather effects are weak, heterogeneous, and non-causal.

Submission document status

The supplied main-22.pdf and supplementary-4.pdf are the latest preserved editorial source exports. Their figure markers are provenance notes, not website instructions. The assembled PDFs replace those markers with validated assets and add full-size figure pages. Current claims, sources, and remaining human submission items are frozen in analysis/submission_freeze/.

Immediate research focus

The next step is not another endpoint model. It is to test the same developmental response against more process-specific, independently validated layers—especially active-edge fuel continuity and terrain aligned to dated observed growth—while improving observation-depth and burn-date uncertainty.