Appendix S1.
We sampled environmental variables that are likely related to different species requirements in a restinga forest. We quantified soil moisture, canopy opening, elevation, terrain slope, flooding height, and litterfall height in a 1-ha, permanent plot of restinga forest in southeastern Brazil. The plot was divided into 100 subplots of 10 m 10 m (Joly et al. , 2012). Soil moisture was determined as the volumetric soil water content measured at the centre of each subplot at 12-cm depth with a HS2 Hydrosense II sensor (Campbell Scientific, Logan, Utah, USA).
We obtained the percentage of canopy opening from hemispheric photographs taken at the centre of each subplot at two different moments, 2008 and 2017, as recent light environment should influence smaller plants and past light environment likely influenced large trees as they were growing (Poorter, 2007). In 2008, two digital cameras with similar settings were used, a Nikon Coolpix 950 and a Nikon Coolpix 5000, both with a fisheye lens Nikon FC E8 (Eisenlohr et al. , 2013). In 2017, we took the photographs using a Nikon D7200 digital camera with Sigma fisheye lens 4.5 mm f/2.8 EX DC and similar settings from 2008. At the two moments, the cameras were positioned on a tripod with the lens facing upwards at 1.3 m from the soil surface and parallel to it, with the magnetic north at the top of the photographs (Eisenlohret al. , 2013). The photographs were taken when the sky was uniform and showed high contrast with the canopy, between 8 h - 10 h and 15 h - 17 h (Rich, 1990). All photographs were processed with the software Gap Light Analyzer (Frazer et al. , 1999) in order to obtain the relative proportion of black and white (representing canopy opening) pixels.
We calculated the elevation of each subplot as the mean elevation of the four corners delimiting the subplots (data from the Functional Gradient Project Biota/FAPESP 03/12595-7 – available upon request). Terrain slope was calculated as the difference between the maximum and the minimum elevation of the four corners of each subplot (Rodrigueset al. , 2007). We considered flooding height as the height of the water column during the rainy season (January 2017). We measured water column and litterfall height at four diagonal points 3.5 m from each corner of the subplots using a graduated stake (Souza and Martins, 2005). Therefore, flooding height and litterfall height were measured at a 5 m 5 m scale, while soil moisture, canopy opening, elevation, and terrain slope were obtained at a 10 m 10 m scale. The values of the variables measured are presented in Figure A1.