Experimental design
The experiment was established in 2013 at the Ontario Forest Research Institute Nursery and Arboretum in Sault Ste. Marie, ON, Canada (46.5465°N, 84.4556°W, MAT: 4.7°C, MAP: 898 mm) (Belluau et al. 2021). The soil is a well-drained Eluviated Dystric Brunisol with sandy loam texture. Before planting, calcitic lime (3 t/ha) was incorporated into the upper soil profile and a slow-release fertilizer (Nutricote® 14-14-14, 1.5 t/ha) was applied to the soil surface. The experiment is part of the International Diversity Experiment Network with Trees (IDENT) that uses high density plantations with orthogonal gradients in species richness and functional diversity to examine the effects of diversity on ecosystem functioning (Tobneret al. 2014). Six native tree species (Acer saccharumMarsh. (ACSA), Betula papyrifera Marsh., (BEPA), Larix laricina K.Koch, (LALA), Pinus strobus L. (PIST), Picea glauca (Moench) Voss (PIGL) and Quercus rubra L. (QURU)) were selected to create a broad range of functional diversity among tree communities. The experiment consists of 22 communities including monocultures (n = 6), and mixtures of two (n = 9), four (n = 6), and six species (n = 1) replicated across 8 blocks.
Trees were planted in May 2013 with seven-by-seven trees at 40 cm spacing in 10.2 m2 square plots surrounded by a 60 cm unplanted perimeter. In each plot, the central five-by-five group of trees was used for analysis and the outer row was considered a plot-level buffer to reduce edge effects. In each block, species planting locations within each mixed community and the location of each community were randomly assigned. In each mixed species community, all species were planted at equal abundance in the central and buffer area and no single tree was surrounded by four seedlings of the same species. Two rows of trees with an alternating species sequence were planted at 1 m spacing around each block to serve as block-level buffers.
In 2014, four of the eight blocks were randomly assigned to either a low or high soil water availability treatment created using a 25% rainfall exclusion apparatus and weekly overhead irrigation with 25 mm of water from June to August, respectively. Basal stem diameter was measured in the fall of 2019 and 2020 to estimate total community level above-ground productivity. Above-ground biomass of each tree was estimated from species-specific allometric equations to predict above-ground biomass from diameter (Belluauet al. 2021). Above-ground productivity for the 2020 growing season was calculated at the tree and plot levels.
Functional diversity of each tree community was estimated using nine functional traits reflecting species differences in resource acquisition strategy and life-history characteristics: maximum leaf area-based net photosynthetic rate (Amax), leaf life span (LL), leaf mass per unit area (LMA), leaf nitrogen content (Nmass), wood density (WD), seed mass (SM), and shade, drought, and waterlogging tolerance (Table S1). All traits were assigned a weight of one, except for two groups of closely related traits (i.e., Nmass and Amax, and drought and shade tolerance) that were each assigned a weight of 0.5 due to their physiological redundancy. Functional diversity was calculated as the functional dispersion index based on the planted relative abundance of the species in a given community using the dbFD function in the FD package in R (Laliberté & Legendre 2010; Laliberté et al. 2014). The functional dispersion index varies from 0 to 1. Monocultures were assigned functional diversity values of 0.
The functional identity of each community, indicative of the average functional trait composition, was estimated as the first axis of a principal components analysis (PCA) on the community-weighted mean values of six functional traits: Amax, LL, LMA, Nmass, WD, and SM (Fig. S1). This axis explained 69.3% of the variation in traits among species weighted by the species planted relative abundance using the PCAfunction in the FactoMineR package in R (Lavorelet al. 2008; Lê et al. 2008).
The net diversity effect (NE), complementarity effect (CE), and selection effect (SE) on total above-ground productivity during the 2020 growing season were calculated for each mixed community by block based on the method of Loreau and Hector (2001). The NE for a given mixture was calculated as,
\begin{equation} NE=\sum{Y_{o}-\overset{\overline{}}{M}},\nonumber \\ \end{equation}
where Yo is the observed yield (productivity) of the mixture (plot-level) and \(\overset{\overline{}}{M}\) is the mean yield of the species in monoculture.
The CE for each mixture was calculated as:
\begin{equation} CE=N\times\Delta\overset{\overline{}}{R}Y\times\overset{\overline{}}{M},\nonumber \\ \end{equation}
where N is the number of species in the mixture and\(\Delta\overset{\overline{}}{R}Y\) is the mean difference between the observed relative yield (RYo = Yo/M) and the expected relative yield (RYe = 1/N).
The SE was calculated as:
\begin{equation} SE=N\times Cov\left(M,\Delta RY\right).\nonumber \\ \end{equation}
T-tests were used to determine if NE, CE, and SE were different from zero in the high and low soil water availability treatments.