Introduction

Positive biodiversity-productivity relationships are a predominant pattern in global forests that are based on both observational and experimental studies (Zhang et al. 2012; Liang et al. 2016). Species richness, as a surrogate of biodiversity, has been demonstrated to increase ecosystem productivity (Diaz & Cabido 2001; Hooper et al. 2012). Traditionally, numerous biodiversity-ecosystem functioning experiments have been implemented through the manipulation of species richness to understand the mechanisms that drive this relationship (Tilman et al. 1996; Grossman et al. 2017; Huang et al. 2018). However, there has been controversy in regard to the relationship between species richness and productivity, being either positive (Grossman et al. 2017; Huang et al. 2018), insignificant, or even negative (Vila et al. 2003; Tobner et al. 2016) in forest ecosystems. Still, a deeper mechanistic elucidation of which ecological processes drive variant species richness-productivity relationships remains incomplete.
Experimental plant communities involve random assemblages of species from a species pool (Tilman et al. 1996; Grossman et al. 2017; Huang et al. 2018), which typically have distinct sets of functional traits (Leps 2004). The functional traits of individual species and their interactions can lead to different species mixtures outcomes (Diaz & Cabido 2001; Loreau et al. 2001; Tobner et al. 2016). Species with various functional traits lead to increased niche differentiation and positive interspecific interactions, more comprehensive resource use, and improved community-level biomass production, i.e., the so called complementarity effects (Tilman et al. 1997; Cardinale et al. 2011). Meanwhile, higher productivity in species mixtures can result from a selection effect, that is, species mixtures may be more productive in contrast to monocultures, due to the increased probability of particular productive species that dominate in mixtures (Grime 1997; Loreau & Hector 2001). However, the outcomes of various species mixtures might be influenced through the extent of trait variations, even though the species richness is constant (Tobner et al. 2016). Thus far, the functional significance of tree diversity on ecosystem productivity has been rarely tested, both across and within species richness levels.
Recent studies have confirmed that effects of biodiversity on ecosystem functioning may be predicted by the degree of functional differences between constituent species in mixtures (Heemsbergen et al. 2004; Chen et al. 2019). Functional differences might result in variable interactions between species (Heemsbergen et al. 2004). For instance, greater interspecific functional dissimilarities increase niche differentiation and facilitative interactions to enhance the usage of resources; thereby, increasing ecosystem productivity (Loreau et al. 2001; Wright et al. 2017). Functional trait dispersion (FDis) is theoretically associated with niche differentiation (Laliberte & Legendre 2010). Accordingly, we hypothesized that FDis in species mixtures would be positively associated with positive diversity effects on forest productivity, both across and within species richness levels. We expected that the positive effects of species richness on productivity were attributable to the positive association between species richness and functional dispersion. Furthermore, the effects of diversity on productivity increased with higher levels of functional dispersion, when the species richness was constant (Fig. 1).
The effects of plant mixtures on productivity are also driven by the functional identities of species mixtures, which represent the characterized functional strategies for resource acquisition of species assemblages (Mokany et al. 2008). Acquisitive species with significant production investments in their stems and leaves have higher efficiencies in terms of resource acquisition and utilization, than conservative species (Reich 2014; Diaz et al. 2016). High productivity associated with acquisitive traits in mixed communities have been matched with high productivity in corresponding monocultures (Mokany et al. 2008). Moreover, the functional characteristics of species determine the intensity of the interactions between constituent species in plant communities along abiotic stress gradients (Maestre et al. 2009). The varied intensities of species interactions, therefore, can lead to different outcomes in terms of community-level productivity (Lusk et al. 2008; Fichtner et al. 2017). Additional available resources allow for the improved realization of niche differentiation in communities dominated by acquisitive traits (Sterck et al. 2011; Baez & Homeier 2018). Alternatively, in stressed or resource-limited environments, conservative traits dominate and interspecific facilitation tends to be stronger, as predicted by a stress-gradient hypothesis (Prado-Junior et al. 2016). Therefore, it may be anticipated that the positive effects of species mixtures on productivity are contingent on the community-weighted means (CWM) of acquisitive traits (Fig. 1). Further, the trajectory of the influence of the CWM indirectly reflects the relative strength of its effects on niche differentiation versus interspecific facilitation.
Here, we aimed to investigate how functional differences and identity determine the various outcomes of tree mixture effects on ecosystem productivity, both across and within species richness levels. We conducted a global meta-analysis based on 210 paired observations of tree mixtures and corresponding monocultures from 59 tree diversity experiment studies (Fig. 1). We collected data on specific leaf area, leaf nitrogen content, and wood density of the selected species for each observation to determine whether FDis and the CWM of acquisitive traits of species mixtures might be positively associated with positive diversity effects on forest productivity, both across and within species richness levels.