Local collection controls removed 65-87% of residual ozone variation in four Eurasian and North American herbarium samples

We tested local collection controls in six exposed-anther herbarium samples from Australia, Eurasia and North America in 1941-1985 and 1986-2017. We used 934 specimens and controlled species-specific collection months and time trends, monthly temperature and location. Adding species-specific five-degree cell intercepts removed 65-87% of residual ozone variation in the four Eurasian and North American samples. The later North American model had two residual degrees of freedom. One species supplied 98% of its remaining ozone information. Its narrow species-cluster interval was not reliable evidence of a precise effect. Two-degree cells left no residual ozone support in that sample. These are design results from one archive. They do not establish spatial bias or the absence of an ozone effect. Total-column ozone differs from surface air pollution. Sparse local records, nonrandom collection and source exposure errors limit inference. Code, aggregate results, source hashes, checks and a figure permit inspection and reproduction.

herbarium · floral pigmentation · total column ozone · temperature · spatial controls · observation bias · model support · cluster inference