# Study 000D Results

## Summary

The corrected full program supported a `selective adaptation model` more strongly than a `broad normalization model`.

All six evidence families favored the selective-expression model.

That means the data supports something stronger than "adaptation happened." It supports a specific form of adaptation:

`successful running expression became narrower, more stabilized, more turnover-dependent, and only partially less costly`

## Evidence Matrix

| evidence_family | observed_result | aligned_with |
|---|---|---|
| ecological breadth | `2025` running share `21.00%` to `2026` `84.18%`; treadmill running share `92.78%` to `98.74%` | selective_expression |
| mechanical gain pathway | cadence change `21.15%`; stride change `12.24%`; cadence share of speed gain `62.42%` | selective_expression |
| higher-demand probe behavior | later outdoor cadence residual `+11.15%`; stride residual `-10.58%`; cadence above expected `5/5`; stride below expected `5/5` | selective_expression |
| constraint persistence | vertical ratio lowest-variability candidate in `000A`; above expected in all `8/8` outdoor runs and `5/5` later specialized outdoor runs | selective_expression |
| internal cost | speed-per-HR improved from `0.01559` to `0.01768`, but HR residual shifted from `-1.59691 bpm` to `3.09257 bpm`; later outdoor mean HR residual `8.67%` | selective_expression |
| ecological correction robustness | `000C` corrected hidden hybrid cardio without overturning phase structure, turnover-led gain, conserved mechanics, or mixed burden | selective_expression |

## Environmental narrowing

The later system did not show broader successful environmental expression.

Instead:

- `2025` treadmill running share: `92.781%`
- `2026` treadmill running share: `98.736%`
- `2025` outdoor running share: `7.219%`
- `2026` outdoor running share: `1.264%`

The phase table shows the same general pattern at the later program level:

- `P4` treadmill share: `93.13%`
- `P5` treadmill share: `98.333%`

This is environmental concentration, not broadening.

## Accessible mechanical lever

The later gain pathway stayed asymmetric:

- speed gain: `36.10775%`
- cadence change: `21.14932%`
- stride change: `12.24394%`
- cadence share of cadence-stride gain: `62.42026%`

That is more consistent with selective use of an accessible lever than with evenly distributed mechanical normalization.

## Higher-demand probe pattern

The later higher-demand probe behavior was especially important:

- cadence residual above expected in `5/5` later specialized outdoor runs
- stride residual below expected in `5/5`
- vertical ratio above expected in `5/5`

This is the opposite of what a simple normalization story would predict.

## Burden pattern

Later adaptation improved efficiency but did not erase burden:

- early speed-per-HR: `0.01559`
- late speed-per-HR: `0.01768`
- early HR residual: `-1.59691 bpm`
- late HR residual: `3.09257 bpm`
- later outdoor mean HR residual: `8.67%`

The program therefore supports `mixed internal cost`, not full burden disappearance.

## Study Answer

The strongest complete answer is:

`the six-year program supports a selective adaptation model rather than a broad normalization model`