SP 11 studied the interaction between tree layer diversity and herb layer diversity in highly diverse subtropical forests. Apart from species richness alone, a high functional diversity of resident communities influences ecosystem functioning. Effects of functional group losses in herb layer communities of the Main Experiment were studied by species removal. Removing species or functional groups allows studying within-community (biomass) compensation and effects on the tree individuals. We applied a functional group removal experiment in the Main experiment in which most dominant functional groups (herbs, grasses, ferns and woody weeds) were experimentally removed by selective weeding. By comparing the removal subplots with control plots containing the entire suite of species, we assess the role of functional groups for herb layer feedback effects on young trees and for evaluating community responses to such a loss at the same time.

The removal experiment was established on Site A and B in June 2011 and was conducted within each 32 plots on Site A and on Site B, thus representing the full tree diversity gradient of the Main Experiment, including the tree diversity levels one (n = 16), two (n = 8), four (n = 4) eight (n = 2), 16 (n =1) and 24 (n = 1) at both sites. Within the northern four rows of each plot, five subplots (5 m × 5 m) containing each 16 trees, were established in June 2011 and randomly assigned to one of five removal treatments. Thus, the removal experiment was based on a total of 320 (2 × 32 × 5) subplots. Five levels of removal treatments were applied, i.e. woody species (W), fern species (F), grass species (G) and herbaceous species (H) were removed. One additional subplot served as a control without removal. The initial removal weeding was carried out in June 2011, where the aboveground biomass of each function group was entirely removed by selective cutting. Additional weeding took place in 2012, in June and September each, and in June 2013. The dry weight of the removed aboveground biomass was determined subsequent to the weeding campaigns. In August 2012, a vegetati+A30on monitoring was conducted for all 320 subplots. In September 2012, a subsample of each subplot was harvested for productivity assessment. In VIPs of 1- and 2-species mixtures, the total aboveground biomass was harvested within each of the removal subplots as a pooled sample of three 0.5 m × 0.5 m replicates and separated by functional group: fern, grass, woody species and herbs. In VIPs of 4-, 8- 16- and 24-species mixtures, five replicates were harvested and pooled by subplot.

Main experiment, Site A, Site B, no shrub VIPs

Initial weeding of removal treatments was performed in June 2011

For each subplot, the coverage, maximum and mean height of each functional group is given; for details see data group description.