Within-site ozone and moisture terms gave no clear repeated injury link in four eastern US forest samples in 1994-2010

We tested whether a positive ozone and soil moisture term describes annual changes at the same forest sites. We used a verified published panel of US Forest Service injury reports, EPA ozone estimates and NOAA weather data. Four Northeast and North Central samples in 1994-2002 and 2003-2010 had 5,277 site-years after a three-visit rule. We separated the product of annual exposure deviations from products with site means. We then added curved exposure terms, plant and species counts, and region-year effects. The four annual product slopes were +0.049, +0.055, +0.113 and -0.039 injury probability per 10 ppm-h ozone per one soil moisture SD. All four two-way state and year cluster intervals included zero. The pooled eastern raw product slope was +0.042 (95 percent interval +0.018 to +0.066). The expanded pooled slope was +0.052 (-0.014 to +0.118). Point estimates did not show a common decrease. Thus the main change was lower precision, not proof that site differences caused the prior positive result. Checks with closer monitors and exposure before the visit also gave no clear repeated link. Few time clusters, selected indicator plants, estimated exposure, unknown missed injury and reuse of rounded source data limit inference. This result does not disprove a moisture effect on ozone uptake or measure biodiversity loss.

ozone · soil moisture · forest injury · within-site interaction · observation bias · regional replication · uncertainty