Complementary resource use has often been claimed to explain positive effects of plant diversity on ecosystem functioning, but the underlying mechanisms of complementarity have rarely been directly quantified in forest systems. The aim of this study was to characterize spatial and temporal water uptake of subtropical tree saplings, to assess the role of interspecific interaction on water uptake patterns and to quantify species niche breadth and overlap. Our experiment comprised two deciduous (Castanea henryi, Quercus serrata) and two evergreen tree species (Elaeocarpus decipiens, Schima superba) that were each planted in monoculture and 4-species mixture. We used deuterium-enriched water to trace seasonal water uptake from 5 cm and 20 cm soil depth. Castanea exploited predominantly the upper soil (74% of overall tracer uptake across treatments) whereas Quercus (50%), Elaeocarpus (57%) and Schima (62%) tended to use both soil layers more equally. Species identity had an overall significant effect on isotopic enrichment in stem water. There was no effect of species richness: niche breadth and overlap of single species was not affected by interspecific interactions in mixtures. Niche overlap between deciduous species was lowest (74%) whereas the two evergreen species had similar water uptake patterns (91%). According to our results, interspecific competition did not alter water uptake patterns of the studied species during the early phase of forest establishment. Thus, soil water uptake complementarity could only occur through inherent (fundamental) specific differences in water uptake niches based on sapling specialization, while phenotypic adjustments to interspecific interaction or neighbor diversity are less important.

The experiment was set up on a former agricultural field where rice, rape and vegetables were previously grown. In March 2009, the field with a total area of 7900 m2 was ploughed, and four blocks of equal size were established. Each block was divided into 1 m2 sized plots separated by 20 cm deep and 20 cm wide trenches. Our experiment used 160 plots in total, with 40 plots randomly distributed in each block. We selected two deciduous species (Castanea henryi (Skan) Rehd. et Wils. and Quercus serrata Murray) and two evergreen species (Elaeocarpus decipiens Hemsley and Schima superba Gardn. et Champ.). In late March 2009, each plot was planted with 16 tree saplings, establishing model communities of one and four species (Fig. 1). Planting distance was about 25 cm, thus allowing fast formation of below- and aboveground interactions soon after planting. All plots were regularly weeded throughout the experiment.To test for spatio-temporal differences among species in water uptake, we injected deuterated water (D 2 O) into the mineral soilin full factorial combinations of two depths (5 cm and 20 cm) and four dates during the year: in summer (2009) when temperature is highest; in autumn (2009) before leaf shedding in the dry season; in winter (2010) when temperature is low and in spring (2010) during leaf flush. Further information on tracer application, plant sampling and analysis of d D in stem water is provided in Trogisch et al.(2016): http://dx.doi.org/10.1016/j.foreco.2016.02.018

Pilot Experiment, Xingangshan, China 29°06'29''N 117°55'28''

Castanea henryi, Quercus serrata, Schima suberba and Elaeocarpus decipiens

We tested for treatment effects on d D excess in stem water and propor- tional water uptake by fitting mixed-effects models including spe- cies richness (factor with 2 levels), species identity, soil depth (a factor with 2 levels), season and all two-way interaction terms as fixed effects, respectively. Root biomass was included as covariate to adjust indirectly for differences in plant size before assessing treatment effects.