2.2 | Study site and plots
The study was conducted in Herrontown Woods Preserve, in Princeton Township in suburban central New Jersey, USA (40.3792, -74.6469). The study site extends 20-45 m from the nearest forest edge, is 0.3 km to the nearest housing community, and is 3.3 km to a town center, Princeton Borough. A recent aerial, infrared drone survey in the region estimated deer density ranging from 35-39 deer/km2 . The preserve is a 136 ha, second-growth, deciduous forest stand estimated to be at least 150 years old, based on tree ring analysis of the cohort of largest trees in the study site (unpublished data). The most abundant tree species (in descending order) are Liriodendron tulipifera ,Fraxinus pennsylvanica , Nyssa sylvatica , Caryaspp., Quercus rubra , and Liquidambar styraciflua . Permanent plots are 4 x 4 m, and those designated for deer exclosure were surrounded since spring 2013 by 5 x 5 m of 2.3 m tall, black plastic fencing with a 4 x 4.5 cm mesh (obtained from deerbusters.com). This type of fencing does not alter light or wind . The fences were staked to the ground, but had three 10 x 30 cm gaps cut at ground level on each side to allow entry by small animals such as rabbits and voles, to ensure that the only excluded vertebrate herbivores were deer. The fences did prevent deer access; the rate of deer browse signs (measured as described above) on all woody species in unfenced plots in this forest was 9.7% (N = 6675 observations), compared to 0.5% (N = 5899) inside fences.