Two or three specimens supplied half of adjusted ozone information in three herbarium samples from 1941 to 2017

We tested how much specimen support remains for ozone slopes after collection-month controls. We used 934 exposed-anther flower records in three regions and two periods. In both Australian samples and later North America, two or three specimens supplied half of the residual ozone sum of squares. The other samples needed 14 to 38 specimens. With exposures and controls fixed, bounded changes of 0.73 to 4.31 percentage points in UV-absorbing petal area could bring each of the six fitted slopes to zero. A separate linear program confirmed these exact worst-case bounds. These bounds do not estimate actual error. All six species-cluster uncertainty intervals included zero. The study adds a support and measurement-error audit to prior slope studies. Nonrandom collection, sparse taxa, one archive and uncertain source exposures limit interpretation. Total-column ozone is a UV proxy, not surface air pollution.

floral pigmentation · total column ozone · temperature · observation bias · measurement error · exposure support · herbarium