Spatial slope changes below 0.30 reported SE permit a half response in 30 years in 54 network summaries

We test spatial benchmark sensitivity in 54 published ecological network summaries. A common positive rate mixture has a 30-year step response of 0.500 at the published absolute score limit when each spatial slope can change by at most 0.29938 of its reported standard error. The root mean square change is 0.29692 standard errors. A convex relaxation excludes this half response when every change is below 0.0468 standard errors. Thus the minimum required box size is between 0.0468 and 0.2994 standard errors; the endpoints are not sharp. The spatial changes are chosen to improve the fit. They are not observed errors or a random-error model. This result shows sensitivity of the fixed-benchmark below-half claim to coordinated small changes. It does not show faster ecological adjustment or failure of a joint likelihood analysis. Spatial-temporal covariance, raw-data uncertainty, detection and habitat remain unresolved. Code, model witnesses, a lower-bound proof and independent numerical checks are included.