Variation in the light environment
Light penetration within our old-field site ranged from 0.3-72.4% (n = 440, μ = 16.6% ± 12; Fig. S1A). The highest variance in light penetration occurred between plots (n = 3, median interplot variance=77.3), followed by variance across months (n = 49, median temporal variance = 57.4) with within-plot light penetration varying the least (n = 147, median intraplot variance = 26.2; Fig. 3). Intraplot variance in light penetration ranged from 0.07-1546.5 (n = 147, μ = 88.3 ± 15.3; Fig. S1B). Mean light penetration was highest in June (20.2%), lowest in July (13.6%), and slightly higher in August (15.9%). Light penetration differed significantly across months (repeated measures ANOVA,F 2,96 = 9.2, P < 0.001; Fig. 4); with a posthoc test revealing that light penetration in June was significantly different than in both July (Bonferroni,P adj<0.001; Fig. 4A) and August (Bonferroni, P adj = 0.036; Fig. 4B), but that the last two months did not significantly differ (Bonferroni,P adj = 0.4; Fig. 4C). High variance in light penetration among sample plots further justified our plot-level approach to hypothesis testing.