For each TSQ, the nearest surrounding trees were identified. The diversity and density were calculated.

As the distance between the planted trees and shrubs are relatively close, it seemed highly likely that the next row of shrubs and trees planted outside the TSQ would also interact and/or influence the focal shrub. Therefore, these shrubs and trees (from here on referred to as neighbouring shrubs and trees) were identified to species level. The neighbourhood scheme/design (see final sheet) was created, where Tree neighbour 1 (N1) is positioned in the South-West of the TSQ, and N2-N12 follow in a clockwise direction. Similarly for the shrubs, Shrub neighbour 1 (S1) is located South-West of the TSQ, and S2-S8 follow in a clockwise direction around the TSQ. An R script was created to calculate the coordinates of neighbouring shrubs and trees, given the coordinate of the focal shrub. The outputs were used in the field to easily find the trees and shrubs labelled with the coordinates. Non-living or absent individuals were classified as either “missing” (nothing present) or “dead” (dead trunk present). These are marked as "M" and "D", respectively in the data. Living individuals that could not be identified to species level were recorded as “unknown.” This was often the case for extremely small shrubs, particularly those with few remaining leaves. Unknown species were compared to the past years species identifications to aid with the identification. Species identifications were cross-checked against the species composition lists of the planted plots to detect possible errors and to assist in confirming uncertain identifications. However, some shrub species recorded were not part of the original planting design (e.g. Loropetalum chinense), indicating that they had likely established naturally. Because most small shrubs had lost their original tags, it was not possible to determine whether individuals matching the composition list were planted or self-established. Additionally, for Site A, multiple shrub individuals were often found at the same coordinate position. Consequently, the experimental neighbourhood does not necessarily reflect the realised neighbourhood.

BEF-China experiment, located near Xingangshang, Dexing City of Jiangxi Province, Site A

Trees; Castanea henryi, Schima superba, Choerospondias axillaris, Liquidambar formosana, Quercus serrata, Lithocarpus glaber, Castanopsis sclerophylla, Koelreuteria bipinnata, Sapindus mukorossi, Nyssa sinensis, Sapium sebiferum, Quercus acutissima

Based on these data, the neighbourhood diversity (species richness) and density (number of living individuals) surrounding each focal shrub were calculated separately for an original experimental design version and a realised version (Site A). For the experimental data, species which should not be there, according to the composition list, were excluded. When there was more than one shrub in a position, if it was unclear which was planted (due to missing tags), both were included in the diversity count, but not in the density count. Thus the density count for neighbouring shrubs could not exceed eight. Diversity and density were calculated manually, following these rules. For the trees, there were never more than one individual per location. However, species which should not occur there based on the species composition list, were excluded; one TSQ in which a shrub was planted instead of a tree.; this species was excluded from the diversity count. For the realised tree neighbourhoods, an R code was used to calculate density and diversity, as well as the diversity of the shrub neighbourhood. Only those coordinates for which there were more than one species present, the diversity was calculated manually. Species which should not be planted in the plot were still included in both the density and diversity counts. Unknown or uncertain species were still included in the density count, but not the diversity count (to prevent over-estimation). For the diversity and density of the shrubs, all species were included, regardless if they belonged there or not (i.e. /were unplanted). After both the realised and experimental versions of the neighbourhood for Site A (both shrubs and trees) was finalised, the two versions for the shrub density and diversity were compared: this is because the biggest differences in density and diversity for the different methods was biggest in shrubs, as there were so many unplanted shrubs and species occurring in plots they should not. This showed a significant difference in both diversity (t = -5.2787, p-value = 7.058e-07) and density (t = -6.0831, p-value = 1.86e-08) between the realised and experimental values. Therefore, the final data used for further analyses was the realised data.