Seedling surveys. In 2011, individual northern red oak seedlings present were marked within a 2000 m2 area belt transect centered on 20 long-term plots at HBEF. A belt-transect running 100 m East and West of the plot center and 5m to the North and South of the center line was searched by at least two people by walking slowly along one side of the center line with one person searching the inner 2.5 m area and the second person scanning the outer 2.5 m area. All oak seedlings (<1.5m in height) were given a unique number and mapped with distance and direction in reference to the center transect line. Seedlings were then measured for vertical height from the ground, number of leaves, age, and presence of an attached acorn if two years or younger. The transects were surveyed in the same manner every year (2012-ongoing) in late summer. All marked individuals were assessed for survival, height, and leaf number; new recruits were marked, mapped, and measured in the same manner. Starting in 2015, seedlings were also inspected for leaf damage on a five point scale: less than 5% damage, 6-25% damage, 26-50% damage, 27-75% damage and greater than 75% damage, and branch dieback as the number of dead and live branches.
Plot environment. In 2019 and 2020, environmental variables were measured by dividing the study transects into 80 contiguous plots that were 5 m2 in area. For all such plots containing at least one marked oak seedling, data were collected at the center of these plots (2.5m along (E-W) and from (N-S) the transect line). Therefore, the number of plots sampled per transect varied. Tree composition was measured with a prism sweep (2.5 factor metric basal area) at plot center; shrub cover in the plot was estimated visually. We quantified understory light availability, slope percent, and slope-aspect at 20-meter intervals along the transects. For analysis, slope-aspect was decomposed into Eastness and Southness, which are unitless and derived using the sine and cosine, respectively. This transformation creates a linear relationship for Eastness where East is +1 and West is -1 with North and South set to 0, and for Southness where South +1 and North -1 with East and West set to 0.
Understory light availability at each point was characterized with hemispherical photographs. All photos were taken at 1 m height under diffuse light conditions using a scientific grade photographic lens (Nikon 8 mm) and a digital single-lens reflex camera (Nikon digital Df). Understory foliage or branches within 1 m of the lens were moved out of the viewshed. We used the Gap Light Analyzer software to compute the fraction of total transmitted radiation reaching each photo point during the growing season (Canham 1988, Frazer et al. 1999). The above-canopy radiation model was based on long-term results from the headquarters weather station at HBEF (USDA 2019). Light levels were reported as the percent of above canopy total transmitted radiation (%TT) received during the growing season (May 15 - August 31). Precision error associated with photographic analyses was 7.6% of relative root mean squared error (rRMSE). For the low light conditions in these forests, this error represents +/- 0.44 percentage unit precision of %TT.