Tree species and genetic diversity increase productivity via functional diversity and trophic feedbacks

  1. Ting Tang
  2. Naili Zhang
  3. Franca J Bongers
  4. Michael Staab
  5. Andreas Schuldt
  6. Felix Fornoff
  7. Hong Lin
  8. Jeannine Cavender-Bares
  9. Andrew L Hipp
  10. Shan Li
  11. Yu Liang
  12. Baocai Han
  13. Alexandra-Maria Klein
  14. Helge Bruelheide
  15. Walter Durka
  16. Bernhard Schmid  Is a corresponding author
  17. Keping Ma  Is a corresponding author
  18. Xiaojuan Liu  Is a corresponding author
  1. State Key Laboratory of Vegetation and Environmental Change, Institute of Botany, Chinese Academy of Sciences, China
  2. College of Life Sciences, University of Chinese Academy of Sciences, China
  3. College of Forestry, Beijing Forestry University, China
  4. Ecological Networks, Technical University Darmstadt, Germany
  5. Forest Nature Conservation, Georg-August-University Göttingen, Germany
  6. Nature Conservation and Landscape Ecology, University of Freiburg, Germany
  7. Institute of Applied Ecology, School of Food Science, Nanjing Xiaozhuang University, China
  8. Department of Ecology, Evolution, and Behavior, University of Minnesota, United States
  9. The Morton Arboretum, United States
  10. State Key Laboratory of Systematic and Evolutionary Botany, Institute of Botany, Chinese Academy of Sciences, China
  11. Chair of Nature Conservation and Landscape Ecology, Faculty of Environment and Natural Resources, University of Freiburg, Germany
  12. Institute of Biology/Geobotany and Botanical Garden, Martin Luther University Halle-Wittenberg, Germany
  13. German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig, Germany
  14. Department of Community Ecology, Helmholtz Centre for Environmental Research–UFZ, Germany
  15. Department of Geography, University of Zurich, Switzerland

Abstract

Addressing global biodiversity loss requires an expanded focus on multiple dimensions of biodiversity. While most studies have focused on the consequences of plant interspecific diversity, our mechanistic understanding of how genetic diversity within plant species affects plant productivity remains limited. Here, we use a tree species × genetic diversity experiment to disentangle the effects of species diversity and genetic diversity on tree productivity, and how they are related to tree functional diversity and trophic feedbacks. We found that tree species diversity increased tree productivity via increased tree functional diversity, reduced soil fungal diversity, and marginally reduced herbivory. The effects of tree genetic diversity on productivity via functional diversity and soil fungal diversity were negative in monocultures but positive in the mixture of the four tree species tested. Given the complexity of interactions between species and genetic diversity, tree functional diversity and trophic feedbacks on productivity, we suggest that both tree species and genetic diversity should be considered in afforestation.

Editor's evaluation

This study uses a landmark experiment to provide compelling evidence that two mechanisms (increased trait space and biological interaction through herbivores and soil fungi) interact with intra- and interspecific genetic diversity to promote forest productivity. These results will be important to foresters and molecular ecologists looking to improve their practices to increase or maintain forest ecosystem functions.

https://doi.org/10.7554/eLife.78703.sa0

eLife digest

Biodiversity, the richness of species in a given ecosystem, is essential for maintaining ecological functions. This is supported by many long-term biodiversity experiments where researchers manipulated the numbers of tree species they planted in a forest and then evaluated both its productivity (how much biological material the forest produced in a given timeframe) and the health of its trees. This work contributed to our understanding of forest ecology and paved the way for better reforestation approaches. The most important observation was that diverse forests, which contain several tree species, are more productive and healthier than monocultures where a single tree species dominates. However, it remained unclear what the role of genetic diversity within individual tree species is in determining productivity and health of forests.

Tang, Zhang et al. set out to improve on previous studies on tree genetic diversity and community productivity by looking at two possible mechanisms that might affect the productivity of a forest ecosystem using publicly available data. First, they looked at the diversity of traits found within a tree population, which determines what resources in the ecosystem the trees can exploit; for example, trees with varied specific leaf areas (that is the ratio between a leaf’s area and its dry mass) have more access to different intensities of sunlight for photosynthesis, allowing the whole forest to gain more biomass. Second, they considered interactions with other organisms such as herbivore animals and soil fungi that affect tree growth by either consuming their leaves or competing for the same resources.

Tang, Zhang et al. used a mathematical model to interpret a complex dataset that includes multiple parameters for each of four types of forest: a forest with a single tree species seeded from a single parent tree (which will have low species and genetic diversity), a forest with a single tree species seeded from several parent trees (low species diversity and high genetic diversity, due to the diversity of parents), a forest with four tree species each seeded from a single parent tree (high species diversity and low genetic diversity), and a forest with four tree species each seeded from several parent trees (high species and genetic diversity).

Using their model, Tang, Zhang et al. determined that species diversity promotes productivity because the increased diversity of traits allows trees to exploit more of the surrounding resources. Genetic diversity, on the other hand, did not seem to have a direct effect on overall productivity. However, greater genetic diversity did coincide with an increase in the diversity of traits in forests with a single tree species, which led to a decrease in damage to tree leaves by herbivores. This suggests that high genetic diversity in species-rich forests is likely also beneficial as herbivores are less able to damage tree foliage. As expected, in single-species forests with both low and high genetic diversity, higher soil fungi diversity was associated with a loss in productivity. Interestingly, in forests that had high species and genetic diversity, this effect was reversed, and higher genetic diversity reduced the loss of productivity caused by soil fungi, resulting in higher productivity overall.

These results should be considered in reforestation projects to promote genetic diversity of trees on top of species diversity when replanting. How genetic diversity leads to downstream mechanisms that benefit community productivity is not fully understood and future research could look at what specific genetic features matter most to help select the ideal mixture of trees to maximize productivity and increase the land’s ecological and economic value.

Introduction

Biodiversity is essential for maintaining ecosystem functioning and nature’s contributions to people (Cardinale et al., 2012; Diaz et al., 2019). Ongoing biodiversity loss has received widespread concern from the international community (Ceballos et al., 2015). Expanding our research focus to multiple dimensions of biodiversity helps us to better predict the consequences of biodiversity loss and prioritize the different dimensions of biodiversity in conservation efforts (Cardinale et al., 2012). Whereas many studies related to biodiversity–ecosystem functioning (BEF) have focused on how interspecific diversity (e.g., the number of species) affects key ecosystem functions such as plant productivity (Hector et al., 1999; Huang et al., 2018; Tilman et al., 2001), relatively few have addressed the effects of intraspecific diversity (such as genetic variation within a species). Furthermore, the effects of intraspecific diversity show an inconsistent picture: genetic diversity has promoted plant community productivity in herbaceous plant communities (Crutsinger et al., 2006; Kotowska et al., 2010) but not in forests (Bongers et al., 2020; Fischer et al., 2017). To get a better understanding of how genetic diversity influences plant productivity in forests and thereby help guiding afforestation priorities, we need to disentangle the underlying mechanisms.

Functional trait diversity, in short functional diversity, is expected to promote community productivity because different species or genotypes with diverse traits may use resources in complementary ways and then enhance the total utilization of resources in the whole community (Diaz and Cabido, 2001; Figure 1a). Thus, functional diversity, mostly quantified as the variation of species functional trait means in a plant community, has been used to explain how plant species diversity impacts plant productivity (Cadotte et al., 2011; Diaz et al., 2007; Hillebrand and Matthiessen, 2009). Although genetic diversity has been shown to cause substantial trait variation within species (Bongers et al., 2020), and intraspecific trait variation may have strong effects on plant productivity (Des Roches et al., 2018; Koricheva et al., 2018), the extent to which genetic diversity can influence tree productivity through increased functional diversity is still unclear.

Conceptual illustration of the effects of functional diversity (a) and trophic feedbacks on tree productivity (b) and the species × genetic diversity experimental design (c).

(a) shows resources for plant growth or other trophic groups in complementary ways due to functional diversity: the four hypothetical species/genotypes (A, B, C, D) with different functional traits (indicated by colored leaves) are able to use a heterogeneous resource (indicated by colored segments), thereby resulting in increased plant growth or providing niche opportunities for other trophic groups (Diaz and Cabido, 2001). (b) shows the mechanism of trophic feedbacks: with the increase in species diversity (SD) or genetic diversity (GD), negative feedbacks of enemies (e.g., herbivores) on tree productivity decrease due to diluted densities (Duffy, 2003) and positive feedbacks of mutualists on tree productivity increase due to increased diversity (e.g., mycorrhizal fungi; Semchenko et al., 2018). (c) We represent tree species and genetic diversity by the number of species and seed families (all seeds from the same mother tree are defined as a single seed family), respectively. Species diversity and genetic diversity per plot were both 1 or 4, resulting in a full factorial design of species × genetic diversity. We hypothesize that the positive effects of tree genetic diversity should be stronger in tree species monocultures (Sp-mono) than mixtures (Sp-mix).

Trophic feedbacks, which result from the interactions of plants of different species or genotypes with other trophic groups, have been suggested as an additional mechanism underpinning positive biodiversity effects (Laforest-Lapointe et al., 2017). Trophic feedbacks can enhance the performance of species or genotype mixtures either by reducing herbivore damage through enhancing the diversity of nutrient traits (Wetzel et al., 2016) and chemical traits (Bustos-Segura et al., 2017) or enhancing diversity of beneficial mutualists (e.g., mycorrhizal fungi; Semchenko et al., 2018; Figure 1b). These trophic feedbacks can be affected by plant functional diversity (Schuldt et al., 2019) and other factors (e.g., structural diversity; Schuldt et al., 2019), which may provide more niche opportunities for other trophic groups. However, whereas many studies have analyzed how plant diversity influences other trophic groups (Scherber et al., 2010; Schuldt et al., 2019) or how trophic interactions affect plant performance (Eisenhauer, 2012; Semchenko et al., 2018), the effects of plant diversity on other trophic groups and the feedbacks of these on productivity have rarely been analyzed in combination.

In real-world ecosystems, plant species diversity and genetic diversity can hardly be expected to influence ecosystems separately (Vellend and Geber, 2005). Previous studies of herbaceous plant communities have shown that the intensity of competition among species can be lowered by increased genetic diversity, which modifies the relationship between plant species diversity and plant productivity (Schöb et al., 2015). Likewise, the relative extent of plant intraspecific variation in functional traits, partly due to genetic diversity, has been shown to decrease with the increase in species diversity (Siefert et al., 2015). Although there are few forest experimental studies in which species and genetic diversity are simultaneously manipulated, most of them only compared their relative importance on ecosystem functions (Abdala‐Roberts et al., 2015; Koricheva et al., 2018), and we barely know their interactive effects via functional diversity and trophic feedbacks on plant productivity.

Here, we disentangle how tree species diversity and genetic diversity affect tree community productivity via the impact of tree functional diversity and trophic feedbacks. We use data from a long-term tree species × genetic diversity experiment in a subtropical forest (Bruelheide et al., 2014; Biodiversity–Ecosystem Functioning Experiment China Platform [BEF-China], https://www.bef-china.com). Tree species diversity (one or four species per plot) and genetic diversity (one or four seed families per species per plot) were manipulated in a factorial design to generate four plant diversity levels (Figure 1c). We measured five morphological and chemical leaf traits, which have been shown to relate to resource acquisition (Cornelissen et al., 2003) and can have substantial variation both among and within species (Albert et al., 2010). Functional diversity was calculated as the variation of these five traits among seed families (Laliberté and Legendre, 2010). We quantified trophic interactions either by direct measurements of interactions (i.e., herbivory) or using the diversity of the trophic group (i.e., soil fungi) as a proxy to capture unspecific interactions potentially underpinning BEF relationships (Delgado-Baquerizo et al., 2016). Specifically, we tested whether tree species and genetic diversity increased tree community productivity via increased functional diversity (Figure 1a) and trophic feedbacks (Figure 1b). Furthermore, we tested whether the effects of genetic diversity were more important in species monocultures than in species mixtures because in the latter case genetic diversity between species may compensate for genetic diversity within species (Figure 1c).

Results

Direct bivariate relationships between tree diversity, trophic interactions, and tree community productivity

Using linear mixed-model analyses, we tested the effects of species diversity and genetic diversity within species on trophic interactions and community productivity. Overall, tree community productivity was significantly higher in the four-species mixture than in the four-species monocultures (Figure 2a), while genetic richness had no main effect on tree productivity in the bivariate analyses (Figure 2a). Tree functional diversity was higher in the species mixture than in the species monocultures and was also higher in genetic mixtures than genetic monocultures (Figure 2b). The effects of genetic diversity on tree functional diversity, herbivore leaf damage, and soil fungal diversity differed between species monocultures and species mixtures. Tree functional diversity in four seed-family species monocultures was larger than in one seed-family species monocultures but did not differ between species mixtures with four or one seed family per species (Figure 2b). However, when we calculated functional diversity based on measurements taken on individual trees rather than based on seed-family means, only species diversity but not genetic diversity had effects on tree functional diversity (Figure 2—figure supplement 1), indicating additional within-seed-family variation masking some of the between-seed-family variation. Furthermore, both herbivore leaf damage and soil fungal diversity were similar in one and four seed-family species monocultures but lower in species mixtures with four than species mixtures with one seed family per species (Figure 2c and d). Due to the equal representation of seed families across tree diversity treatments (Appendix 1—table 1), we did not find any significant effects of tree species and genetic diversity effects on community-weighted means (CWMs) of tree functional traits (Appendix 2—table 1).

Figure 2 with 1 supplement see all
Tree community productivity, tree functional diversity, and trophic interactions in tree communities of low vs. high species and genetic richness.

The following effects were tested in linear mixed-effects models (LMMs) (n=92): species richness main effect (left vs. right pair of bars in each panel), genetic richness main effect (inset on upper left in each panel), genetic richness effect within each species richness level (arrows between bars within pairs). (a) tree community productivity, (b) tree functional diversity, (c) herbivore leaf damage, and (d) soil fungal diversity. The lower and upper hinges of the bars correspond to the first and third quartiles (the 25th and 75th percentiles); the lower and upper whisker extends from the hinge correspond to 1.5 * interquartile range (third quartiles - first quartiles). Asterisks indicate statistical significance (*** p<0.0001, ** p<0.001, * p<0.05); solid arrow indicates (p<0.05, without arrow indicates p>0.1). Details of the fitted models are given in Appendix 2—table 1.

Tree functional diversity calculated using either seed-family means or individual tree values had positive overall effects on community productivity, but this effect was mainly due to an increase in functional diversity from species monocultures to mixtures (Figure 3a, Figure 3—figure supplement 1). Herbivore leaf damage and soil fungal diversity showed negative overall effects on tree productivity (marginally significant for herbivory and significant for fungal diversity; Figure 3b and c). Furthermore, the effects of herbivore damage were different between genetic monocultures and genetic mixtures in species monocultures (Figure 3b), while the effects of soil fungal diversity were different between genetic monocultures and genetic mixtures in the species mixture (Figure 3c).

Figure 3 with 1 supplement see all
Bivariate relationships between tree community productivity and tree functional diversity (a), herbivory (b), and soil fungal diversity (c).

Green unfilled/dashed symbols represent genetic monocultures in species monocultures, green filled/solid symbols represent genetic monocultures in species mixture, orange unfilled/dashed symbols represent genetic mixtures in species monocultures, orange filled/solid symbols represent genetic mixture in species mixture. FDis, tree functional diversity; Herb, herbivore damage; Fungal, soil fungal diversity; Sp-mono, species monocultures; Sp-mix, species mixtures; SD, species diversity; GD, genetic diversity. ‘:’ indicates the interaction effects. Asterisks indicate statistical significance (*** p < 0.0001, ** p < 0.001, * p < 0.05, + p < 0.1, and ns p > 0.1).

Functional diversity and trophic feedbacks explain the effects of tree species and genetic diversity on tree productivity

Tree species and genetic diversity promoted tree community productivity as well as trophic interactions primarily indirectly through functional diversity (Figure 4). The increase in functional diversity was larger for increasing species diversity than for increasing genetic diversity (standardized path coefficient = 0.960 vs. 0.074, Figure 4). Herbivory and soil fungal diversity reduced tree community productivity (Figure 4, see also Figure 3b and c). Overall, tree diversity had contrasting effects on tree community productivity through different mechanisms: species and genetic diversity promoted tree functional diversity, which increased productivity directly but reduced it indirectly via negative feedbacks of herbivory and soil fungal diversity. However, species and genetic diversity also had positive indirect effects on community productivity via reduced soil fungal diversity (and genetic diversity additionally via reduced herbivory; Figure 4). Whereas tree functional diversity and trophic feedbacks explained all effects of tree species diversity on productivity, there remained a direct negative effect of tree genetic diversity on productivity, which could not be explained by the measured covariates (Figure 4). The analysis that functional diversity calculated from measurements on individual trees also showed that tree species diversity and genetic diversity affect community productivity via tree functional diversity and trophic feedbacks, although the effects of functional diversity were less pronounced (Figure 4—figure supplement 1), possibly because functional diversity calculated from individual trees included more response functional diversity (Sapijanskas et al., 2014) than did functional diversity calculated form seed-family means. Additionally, removing the path between genetic diversity and functional diversity did not change the remaining results we found by using functional diversity calculated from seed-family means (Figure 4, Appendix 3—figure 1).

Figure 4 with 2 supplements see all
Effects of tree diversity on higher trophic levels and tree community productivity (global Fisher’s C = 1.677, DF = 4, p = 0.795).

Positive and negative paths are indicated in green and orange, respectively. The standardized path coefficients are indicated by the numbers, statistical significance is indicated by asterisks (*** p < 0.0001, ** p< 0.001, * p < 0.05, and + p < 0.1), and the explained variance of dependent variables is indicated by the percentage values. The gray dashed line indicates a nonsignificant (p > 0.1) pathway in the final model. The direct effect of tree species diversity on tree community productivity was removed in the model because it was not significant (p > 0.5) and the removal reduced the AICc by more than 2 (ΔAICc = 3.269).

Effects of tree genetic diversity in species monocultures and species mixtures

When the above analysis was split into two (Figure 5), in contrast to our hypothesis, we found that tree genetic diversity negatively affected community productivity via functional diversity and soil fungal diversity in species monocultures and had positive effects via soil fungal diversity in the species mixture (see also Figures 2b and 3a). The results obtained with functional traits calculated from measurements on individual trees showed weaker effects of genetic diversity on functional diversity (path coefficient = 0.193 vs. 0.883) but did not change the significance and direction of the effects of genetic diversity on productivity via functional diversity in species monocultures. Additionally, the effects of functional diversity on tree productivity in species mixtures were positive when using functional diversity calculated from measurements on individual trees but were nonsignificant when using functional diversity calculated from seed-family means (Figure 5, Figure 5—figure supplement 1). Positive indirect effects through herbivory (resulting from two negative paths from genetic diversity to herbivory and form herbivory to community productivity) were similar in both species monocultures and mixtures. Using functional diversity calculated from measurements on individual trees did not change the effects of genetic diversity via trophic feedbacks, except that the effects of herbivory on productivity became nonsignificant from marginally significant. When we excluded the effects of genetic diversity on functional diversity in the analyses using functional diversity calculated from seed-family means, the remaining path coefficients did not change (Figure 5, Appendix 3—figure 2). The negative indirect effect of genetic diversity on community productivity via functional diversity in species monocultures, which contrasts with the combined analysis, was counterbalanced by a positive direct effect of genetic diversity on productivity, indicating that other aspects than those included with the five functional traits measured were important.

Figure 5 with 2 supplements see all
Effects of tree genetic diversity on higher trophic levels and tree community productivity in tree species monocultures (a) and the mixture of the four tree species (b).

The results were obtained by a multigroup structural equation models (SEM) (global Fisher’s C = 3.416, DF = 4, p = 0.491). Positive and negative paths are indicated in green and orange, respectively. The standardized path coefficients are indicated by the numbers, and statistical significance is indicated by asterisks (*** p < 0.0001, ** p < 0.001, * p < 0.05, and + p < 0.1). Gray dashed lines indicate nonsignificant (p > 0.1) pathways in the final model. The nonsignificant path from tree functional diversity to soil fungal diversity was removed because the removal decreased the AICc by more than 2 (ΔAICc = 2.176). Multigroup SEM analyses first test the interaction (explanatory variable × groups) in the whole model using the full dataset and then estimate the local coefficient for each path by using different datasets (the full dataset or group sub-datasets [species richness = 1 or 4, respectively]) depending on the significance of explanatory variable × groups interactions. Thus, we could not get the percentage of the explained variance in the local multi-group SEM model. All the paths were allowed to be different between species monocultures and mixtures (none of the paths was constrained manually beforehand); the interaction statistics of the multigroup model, and the explained variance of the whole model for each response is shown in Appendix 2—table 5.

Discussion

Our study demonstrates that manipulating tree species and genetic diversity in a factorial design can reveal effects of both as well as their interaction on measured ecosystem variables. Regarding our first hypothesis, we found that tree species diversity and genetic diversity can increase tree community productivity via increased functional diversity and trophic feedbacks as predicted. This suggests complementary resource-use and biotic niches, respectively, as mechanisms underpinning the biodiversity effects (Turnbull et al., 2016). Nevertheless, compared with the effects of species diversity, the effects of genetic diversity on tree community productivity through functional diversity were weaker, whereas the effects of genetic diversity on trophic interactions were strong (see Figure 4, Figure 4—figure supplement 1), indicating that the mechanisms underpinning the effects of genetic diversity may in part differ from those underpinning the effects of species diversity, as we will discuss below. Regarding our second hypothesis, we found that the effects of tree genetic diversity on productivity via functional diversity and soil fungal diversity were negative in tree species monocultures but positive in the species mixture, which differed from our predictions. In the following, we discuss these results in more detail.

Tree species and genetic diversity drive tree community productivity mainly via functional diversity and trophic feedbacks

Although only species diversity but not genetic diversity was found to affect tree productivity in binary analyses, both kinds of diversity positively affected tree community productivity and trophic interactions via functional diversity according to our structural equation models (SEMs) depicted in the corresponding path-analysis diagrams (see Figure 4). Tree functional diversity appeared to enhance complementary resource acquisition at community level (Kahmen et al., 2006; Marquard et al., 2009; Williams et al., 2017), which consequently enhanced tree community productivity. Meanwhile, tree functional diversity also provided more niche opportunities to benefit generalist herbivores and soil fungi, which reduced tree community productivity, as has been found for these tree species in a parallel field study nearby (Brezzi et al., 2017). It is expected that herbivory has negative effects on plant productivity via the reduction of leaf area (Zvereva et al., 2012) and photosynthesis of remaining leaves (Nabity et al., 2009), and via trade-offs between growth and herbivore defense (Züst and Agrawal, 2017). The negative effects of soil fungal diversity on productivity correspond with the finding that the majority of these fungi were saprophytes (Appendix 2—figure 1), competing with plants for resources (Kaye and Hart, 1997; van der Heijden et al., 2008). Indeed, in a related study in the same region, the diversity of saprophytic fungi had been found to decrease ecosystem multifunctionality (Schuldt et al., 2018).

Indirect positive effects of species and genetic diversity – remaining after accounting for paths via functional diversity – via reduced herbivory and soil fungal diversity further increased community productivity (see Figure 4). This finding corresponds to previous findings that plant diversity may reduce negative feedbacks of other trophic groups by decreasing the density and diversity of specialist enemies (e.g., Duffy, 2003; Jactel and Brockerhoff, 2007).

To account for possible effects of functional diversity within seed families, we also calculated functional diversity based on measurements of individual trees (see ‘Materials and methods’). Overall, the results from this novel method still support our hypotheses that tree species diversity and genetic diversity affect community productivity via tree functional diversity and multi-trophic feedbacks (Figure 4—figure supplement 1), although compared with the typically used ‘mean’ method, the novel method includes more variation among individuals, which partly reflects responses of traits to the particular local environment (Sapijanskas et al., 2014); and this may have blurred the mean effects of tree genetic diversity and species diversity (Figure 4, Figure 4—figure supplement 1). At the same time, the results indicate that the seed-family means method may bring an artifact to the effect of genetic diversity on functional diversity because of the zero value of functional diversity in genetic monocultures of single species (1.1 communities). However, excluding the path between genetic diversity and functional diversity did not affect remaining paths, indicating that the partly artificial relationship between genetic diversity and functional diversity did not distort the path model in general (Figure 4, Appendix 3—figure 1).

Even after accounting for tree functional diversity and trophic feedbacks, we still detected a direct negative effect of tree genetic diversity on tree productivity, while the direct effect of tree species diversity was fully explained by functional diversity and trophic feedbacks. This suggests that aspects of genetic diversity that do not contribute to functional diversity or trophic interactions as measured in this study may reduce ecosystem functioning, for example, due to trade-offs between genetic diversity and species diversity. For example, it has been shown that in species-diverse grassland ecosystems, niche-complementarity between species can increase at the expense of reduced variation within species (van Moorsel et al., 2018; van Moorsel et al., 2019; Zuppinger-Dingley et al., 2014; Zvereva et al., 2012). Thus, our experiment simulating high genetic diversity within species in mixtures might have reduced the positive effects of high species diversity. This interpretation would be compatible with the observation that in the separate path-analyses diagrams direct negative effects of genetic diversity on productivity were only found in species mixtures, whereas in the species monocultures these effects were positive (see next section). Independent of this interpretation, our finding could also imply that partly different mechanisms underpin effects of species vs. genetic diversity on ecosystem functioning (Barantal et al., 2019; Des Roches et al., 2018).

Effects of tree genetic diversity differ between tree species monocultures and mixtures

In contrast to our second hypothesis, we found that the effects of genetic diversity via functional diversity and soil fungal diversity were negative in species monocultures but not significant via functional diversity and positive via soil fungal diversity in the species mixture (Figure 5). We found that genetic diversity had positive effects on tree functional diversity and soil fungal diversity in species monocultures but negative effects in the species mixture, which supports the trade-offs between genetic and species diversity discussed in the previous section. However, the hypothesized positive effects of tree functional diversity on productivity turned negative in species monoculture. This result indicates that functional diversity may not have positive effects on the ecosystem functioning under low environmental heterogeneity, that is, species monocultures in our study (Hillebrand and Matthiessen, 2009). Moreover, other aspects of tree genetic diversity seem to play an important role not only for productivity in tree species mixtures (see previous section) but also for productivity in tree species monocultures. These may include unmeasured functional traits such as root traits (Bardgett et al., 2014) or unknown mechanisms underpinning effects of tree genetic diversity.

The two methods of calculating functional diversity either from seed-family means or from trait values of individual trees yielded different results regarding the indirect effects of genetic diversity on tree productivity via functional diversity. The method based on seed-family means has the advantage to be less circular, whereas the method based on trait values of individuals has the advantage of producing functional diversity values >0 also for genetic monocultures of single species (1.1 communities; see ‘Materials and methods’). The weaker indirect effects of genetic diversity on tree productivity via functional diversity in the method using trait values of individuals suggest that the zero value of functional diversity in 1.1 communities in the method using seed-family means may lead to an overestimation of these indirect effects of genetic diversity in species monocultures. Nevertheless, the method using seed-family means is still useful for species monocultures with multiple seed families and for species mixtures.

Conclusions

In this study, we tried to disentangle the effects of tree species and genetic diversity via functional diversity and trophic feedbacks on tree community productivity in a simple experimental system with four species and multiple seed families per species. Even though this was already challenging to set up, manage, and assess by measurements on trees and soil samples, larger studies will be required to generalize results. Nevertheless, our results suggest that both partitioning of resource-use and enemy niches (Turnbull et al., 2016) between and among genotypes within tree species played a role in affecting tree community productivity. Although both tree species and genetic diversity contributed to productivity, the underpinning mechanisms differed and were harder to explain for tree genetic diversity. We suggest that trade-offs between tree species and genetic diversity may cause the latter to switch strength and direction between species monocultures and mixtures. We were not able to definitively report causality between trophic feedbacks and tree productivity because we did not experimentally manipulate herbivore leaf damage and soil fungi. However, our results do support the hypothesis that trophic feedbacks affect plant community productivity. Given the importance of afforestation projects to mitigate carbon loss and provide ecological and economic benefits (Brockerhoff et al., 2008; Lamb et al., 2005), we strongly recommend that both tree species and genetic diversity should be considered in afforestation projects.

Materials and methods

Study site and experimental design

This study was carried out in the species × genetic diversity experiment of the BEF-China (https://www.bef-china.com; Bruelheide et al., 2014; Hahn et al., 2017). BEF-China is located close to Xingangshan, Dexing City, Jiangxi Province, China. The mean annual temperature is 16.7 °C, and the mean annual precipitation is 1821 mm. The species × genetic diversity experiment was established in 2010 and comprises 24 plots of 25.8 × 25.8 m equal to one Chinese unit of ‘mu’. Each plot was planted with 400 individual trees from a pool of four species (Alniphyllum fortunei, Cinnamomum camphora, Daphniphyllum oldhamii, and Idesia polycarpa) with the mother trees of all tree individuals known (Appendix 1—figure 1). We defined the offspring from the same mother tree as a seed family and assumed that the genetic variation was larger among seed families than within a seed family (Bongers et al., 2020; Hahn et al., 2017). Since the offspring of a single mother tree could have been sired by different father trees, they represented anything between full- and half-sib families. Thus, in this study, we used the number of seed families per species as a measure of genetic diversity (Bruelheide et al., 2014). Across the 24 plots, we combined species diversity (one or four species) and genetic diversity (one or four seed families per species), which resulted in four tree diversity levels: one species with one seed family (1.1), one species with four seed families (1.4), four species with one seed family per species (4.1), and four species with four seed families per species (4.4) (Appendix 1—figure 1; Bongers et al., 2020).

For each of the four species, we collected seeds from eight mother trees to allow for two replications of four-family mixtures per species. Furthermore, to avoid the effects of unequal representation of particular seed families and correlations between seed family presence and diversity treatments, we made sure that every seed family occurred the same number of times at each diversity level (see Appendix 1—table 1, small deviations from the rule were required where not enough seeds from a seed family could be obtained). Due to budget limitations and the number of replicates required per single seed family, the 1.1 and 1.4 diversity treatments were applied at subplot level (0.25 mu) and replicated 32 and 8 times, respectively. The 4.1 and 4.4 diversity treatments were applied at plot level (1 mu) and were replicated eight and six times, respectively (Appendix 1—figure 1; see also Figure 1 in Bongers et al., 2020). To allow for simpler analysis, we obtained most community measures at subplot level also for the 4.1 and 4.4 diversity treatments and thereafter used the subplots for all tests of diversity effects on these community measures, including plots as error (i.e., random-effects) term for testing the diversity effects in the corresponding mixed models. In total, because one 1-mu plot could not be established due to logistic constraints, the number of subplots used was 92 (32 subplots of 1.1, 8 subplots of 1.4, 28 subplots of 4.1, and 24 subplots of 4.4 diversity treatment). Note that in biodiversity experiments lower richness levels represent more different communities and thus require more plots. For the highest richness level, where there is typically only one species composition, this same community is typically replicated multiple times, as we did here for the 4.4 diversity treatment.

Tree functional traits and functional diversity

Five leaf functional traits were measured in 2017 and 2018, including leaf area (LA), specific leaf area (SLA), chlorophyll content (CHL), leaf nitrogen content (LN), and leaf carbon content (LC). These traits can reflect the resource acquisition ability of plants and may show substantial variation not only among species but also within species (Albert et al., 2010; Cornelissen et al., 2003). We collected these traits on 547 individuals of all the seed families of the four species across all the species ×genetic diversity combinations (Appendix 1—table 2), with details described in Bongers et al., 2020.

Functional leaf-trait diversity was expressed as multivariate functional dispersion (FDis), which in our case corresponds to the mean distance of individual seed families to the centroid of all seed families in the community (Laliberté and Legendre, 2010). To reduce circularity, we used the seed-family means across all species × genetic diversity combinations to calculate FDis values per subplot that did not only depend on the functional trait measures obtained in that particular subplot. Using traits measured in a particular subplot to calculate FDis for that subplot bears the risk that the measured traits reflect a response to the local environment, yet we want to use FDis as a predictor variable for the performance of that subplot. In every mixture, trait values were weighted equally across seed families and species because these were planted in equal numbers in each subplot. The mean value of FDis per species × genetic diversity level was used to fill in missing values in a few subplots with families lacking trait data (Appendix 1—table 2). We also calculated another frequently used functional diversity index, Rao’s Q (Rao, 1982). However, a strong positive correlation was detected between FDis and Rao’s Q in simulated data (Laliberté and Legendre, 2010) and in our study (Appendix 2—figure 2). Moreover, in the case of equal weighting, FDis should perform better than Rao’s Q (Laliberté and Legendre, 2010). Therefore, we only used FDis in the analyses presented in this study. The calculations of FDis and Rao’s Q were done with the ‘dbFD’ function of the ‘FD’ package (versions 1.0–12.1) in R (Laliberté et al., 2014, https://www.r-project.org). We further calculated FDis using traits measured on individual trees across all tree diversity treatment combinations. This alternative FDis had the advantage that it could also be calculated for subplots planted with trees of a single seed family (which had FDis values of zero when calculated with seed-family means), reflecting within seed-family functional trait diversity. The disadvantage is that this measure likely includes more response variation because every individual tree responds to a number of unknown factors in its local environment. We also calculated CWMs for the five functional traits. To obtain a multivariate equivalent, we subjected the individual traits to a varimax rotation principal component analysis (PCA) to obtain two orthogonal axes as principal-component CWMs. The two principal components captured together 64% variation of trait variation (Appendix 2—table 2). PC1 indicated the functional traits directly connected with growth, and PC2 indicated the functional traits connected with photosynthesis (Appendix 2—figure 3). The varimax rotation PCA was done usinh ‘psych’ R package version 2.1.9 (Makowski, 2018).

Trophic interactions

Herbivory

Herbivory results from the interaction between plants and herbivores and can be recorded as leaf damage. For every individual tree, four or five damaged leaves were randomly collected and herbivory visually estimated (Johnson et al., 2016) (same 547 trees as for the traits, see above) in 2017. Thus, in this study herbivory represents the percentage of damaged area per leaf attacked by herbivores. The herbivory caused by chewers, gall formers, leaf miners, and rollers were collectively counted. Because we only collected damaged leaves in this study, we might have overestimated the herbivory per individual tree. We therefore used data from other plots of the BEF-China experiment (Schuldt et al., 2015), which did not exclude nondamaged leaves to correct the potential bias. This former study assessed herbivore damage by visually inspecting 21 leaves (7 leaves per branch) on three random branches from different parts of the canopy (Schuldt et al., 2015). They used the mean percentage damage value as the overall leaf damage for each individual. We related leaf damage of corresponding tree individuals from this former study (total leaf damage) to leaf damage excluding nondamaged leaves (damage per damaged leaf) for all four species by linear regression (Pearson’s correlation = 0.86–0.96, p < 0.001) (Appendix 2—table 3). With these regression models, we got the predicted values of herbivory for our study and used these predicted values in the final analyses. The mean value of herbivore damage per species × genetic diversity level was used to fill in missing values in a few subplots with tree individuals lacking herbivory data (Appendix 1—table 2).

Soil fungal diversity

Soil fungal diversity was used as a proxy of unspecified trophic interactions. To be consistent with the species and genetic diversity treatment design, soil samples were taken on subplot level for the 1.1 and 1.4 diversity treatments, but, due to feasibility constraints, on plot level for the 4.1 and 4.4 diversity treatments in 2017. In each subplot or plot, five soil samples from the top 0–5 cm soil layer were collected from the four corners and the center of each subplot or plot. The five samples were then mixed together. Each soil sample was packed with dry ice and transferred to the laboratory for storage at −80°C until DNA extraction. The total genomic DNA of the subsample was extracted using Soil Genomic DNA Kit (Tiangen Biotech Co., Beijing, China), following the manufacturer’s protocol. The DNA was extracted to perform PCR amplification. We amplified the nuclear rDNA internal transcribed spacer 2 (ITS2) region using primers ITS3F (GCATCGATGAAGAACGCAGC) and ITS4R (TCCTCCGCTTATTGATATGC). We processed the raw sequences with the QIIME 2 pipeline (https://docs.qiime2.org/) to cluster and assign operational taxonomic units (OTU). The fungal OTU tables were rarefied to 10,975 reads to account for the different sequencing depths. We then assigned the sequences to taxonomic groups using the UNITE database (Nilsson et al., 2019). Based on the taxonomic and abundance information of every subplot or plot, the Chao1 diversity index (Chao, 1984) was used to quantify soil fungal diversity, because most fungal species in our study were relatively rare and the Chao1 index can account well for rare species (Chao, 1984). The calculation of diversity of soil fungi was done using the ‘vegan’ package version 2.5–7 in R (Oksanen et al., 2019).

Tree community productivity

We measured the basal area (BA) and the height (H) of all trees in the species × genetic diversity plots in 2018 (Bongers et al., 2020). Individual tree biomass (kg) was calculated using the biomass equation (H × BA × CV) of the BEF-China experiment (Huang et al., 2018) in which CV is a correction factor for stem shape and wood density. More details about the biomass equation can be found in Huang et al., 2018. We summed the biomass of individual trees to subplot level to calculate tree community productivity (Mg ha–1).

Statistical analysis

First, we evaluated the bivariate relationships between tree diversity, trophic interactions, and tree community productivity. To determine how species and genetic diversity and their interaction affected tree functional diversity and trophic interactions, linear mixed-effects models (LMMs) were fitted with two types of contrast coding. In the first, we used the ordinary two-way analysis of variance with interaction and in the second we replaced the genetic diversity main effect and the interaction with separate genetic diversity effects for species monocultures and the species mixture (Appendix 2—table 4). Note that as our design was orthogonal, fitting sequence did not matter in either of the codings. However, we focused our major analysis on the second type of coding to make it consistent with our hypotheses. Main effects of genetic diversity are presented in inset panels in Figure 2. Our second contrast coding ensured that we tested the effects of genetic diversity separately in species monocultures and species mixture, but within the same analysis. For all LMMs, we used ‘plot’ as a random variable since subplots were nested in plots. This also ensured that fixed terms whose levels did not vary within plots among subplots (specifically the four-species diversity treatments) were correctly tested against the variation among plots rather than the residual variation among subplots. LMMs were fitted in R with the ‘lmer’ function of the lme4 package version 1.1.27.1 (Bates et al., 2015) using Kenward–Roger’s method to calculate denominator degrees of freedom and F-statistics with the lmerTest-package version 3.1.3 (Kuznetsova et al., 2017). To meet the assumptions of linear mixed models, the proportion of leaf damage caused by herbivores was angular transformed (Snedecor and Cochran, 1989). For the display of regression lines in Figure 3, we used linear models relating tree functional diversity, herbivore leaf damage, and soil fungal diversity for the four diversity-treatment combinations to tree community productivity (‘lm’ function in R).

Second, we fitted SEMs and displayed the results in path-analysis diagrams (Grace, 2006) with the ‘piecewiseSEM’ package version 2.1.2 in R (Lefcheck, 2016) to assess causal hypotheses about how the effects of tree species and genetic diversity on community productivity could have been mediated via tree functional diversity and trophic interactions. The initial model was constructed by the most relevant pathways derived from theoretical assumptions (Figure 4—figure supplement 2). Additionally, we used separate linear regressions to assess the relationships between variables hypothesized to be related in cause–effect relationships in the SEMs. We assumed that both tree genetic diversity and species diversity could influence trophic interactions and community productivity directly or indirectly, that is, mediated via functional diversity (Müller et al., 2018; Scherber et al., 2010; Schuldt et al., 2019). Moreover, we hypothesized that tree functional diversity, herbivore leaf damage, and soil fungal diversity have direct feedbacks on community productivity (Eisenhauer, 2012; Semchenko et al., 2018). We sequentially dropped noninformative pathways if their removal reduced the AICc of the SEMs by more than 2 (Grace, 2006). To detect potential distorting effects of the relationship between genetic diversity and functional diversity calculated from seed-family means, we also calculated a SEM model without the path between genetic diversity and functional diversity.

Thirdly, separate multigroup SEMs were fitted for species monocultures and mixtures since significant interactions between species and genetic diversity in the ANOVAs indicated that genetic diversity had different effects between species monocultures and the species mixtures. The initial multigroup path diagram is shown in Figure 5—figure supplement 2. We simplified the multigroup initial model with the same procedure as described above by comparing AICc values. For the multigroup models, we also calculated an additional one in which the path between genetic diversity and functional diversity was excluded.

Finally, to detect the robustness of our results, we used the same paths as in the above final single and multipath models to analyze the data with FDis calculated with the trait measures of individual trees. All the analyses were carried out in R 4.0.5.

Appendix 1

Appendix 1—figure 1
Diagram of the seed families planted in the species × genetic diversity experiment.

1.1: species diversity = 1, genetic diversity = 1; 1.4: species diversity = 1, genetic diversity = 4; 4.1: species diversity = 4, genetic diversity = 1; 4.4: species diversity = 4, genetic diversity = 4. The uppercase letters indicate four tree species (A: Alniphyllum fortunei; B: Cinnamomum camphora; C: Daphniphyllum oldhamii; D: Idesia polycarpa), the number after ‘_’ indicates the seed family tag of a given species, two numbers indicate both of the seed families were used in this plot due to not enough designed seedlings.

Appendix 1—table 1
The designed and planted occurrence times of each seed family per species in the four diversity treatment combinations.

1.1: species diversity = 1, genetic diversity = 1; 1.4: species diversity = 1, genetic diversity = 4; 4.1: species diversity = 4, genetic diversity = 1; 4.4: species diversity = 4, genetic diversity = 4. ‘SP’ is the species name and ‘SF’ is the tag of seed family.The experiment was designed to use eight seed families per species, but additional or repeated seed families were used to complement the lack of enough individuals in some seed families. The numbers in brackets indicate the seed family tags that were used to complement. AlFo (A): Alniphyllum fortune’, CiCa (B): Cinnamomum camphora; DaOl (C): Daphniphyllum oldhamii; IdPo (D): Idesia polycarpa. ‘x’ represents the number of individuals per subplot, and the number of the ‘x’ represents the number of subplots.

Tree diversity1.1 (x = 100)1.4 (x = 25)4.1 (x = 100)4.4 (x = 25)
SPSFTree individualsTree individualsTree individualsTree individuals
AlFo1xxxx + x + x
(A)2xxxx + x + x
3xx(1)xx + x + x
4xxxx + x + x
5xx(2)xx + x + x
6xx(4)xx + x + x(9)
7xxxx + x + x
8xx(7)xx + x + x
CiCa1xxxx + x + x
(B)2xxxx + x + x
3xxx(6)x + x + x
4xxx(6/9)x + x + x
5xxxx + x + x(3)
6xxxx + x + x
7xxxx + x + x
8xxxx + x + x
DaOl1xxxx + x + x
(C)2xxx(7)x + x + x
3xxx(7)x(11) + x(11) + x(9)
4xxxx + x(10) + x(12)
5xxxx + x + x(13)
6xxxx + x + x
7xxxx + x + x(14)
8xxx(8/9)x + x + x
IdPo1xxxx + x + x
(D)2xxxx + x + x
3xxxx + x + x
4xxxx + x + x
5xx(4)xx + x + x
6xxxx + x + x
7x(5)x(8)xx + x + x
8xxxx + x + x
Appendix 1—table 2
Data description of multi-trophic levels.
Data typeData descriptionSubplotsYear
Plant traitLA, SLA, CHL, LN, LC772017
Herbivore damageVisually estimated772017
Soil fungiMainly composed of saprophytes532017
Community productivitySum of the biomass per subplot/area of subplot922018
  1. LA, leaf area; SLA, specific leaf area; CHL, chlorophyll content; LN, leaf nitrogen content; LC, leaf carbon content.

Appendix 2

Appendix 2—figure 1
Trophic composition of soil fungi in this study.

All fungi from this study were pooled together to calculate the relative abundance of each trophic group.

Appendix 2—figure 2
Relationship between functional dispersion (FDis) and Rao’s Q (RaoQ).
Appendix 2—figure 3
Varimax rotation principal component analysis (PCA) biplot for the five functional traits.
Appendix 2—table 1
Summary of linear mixed-effects models (LMMs) of species diversity (SD), genetic diversity (GD), and their interactions on tree productivity, tree functional diversity, trophic interactions, and community-weighted mean (CWM) of functional traits.

Expressed values are Df representing degree of freedom and F-values with related significances, *** p < 0.001; * *p < 0.01; * p < 0.05, + p < 0.1. Note that the very small F-values for CWMs are due to the equal representation of seed families across all tree diversity treatments (Appendix 1—table 1).

Tree productivityFD(Mean)FD (Individual)HerbivorySoil fungal diversityCWM (RC1)CWM (RC2)
FactorsDfRandomF-valueF-valueF-valueF-valueF-valueF-valueF-value
SD1Plot6.16*44.80***20.60***0.081.060.0040.002
GD1Plot0.513.66+0.115.86*1.570.0000.033
SD × GD1Plot0.237.44*0.051.445.29*0.0030.011
SD1Plot6.16*44.80***20.60***0.081.060.0040.002
GD (Sp-mono)1Plot0.0011.09**0.150.171.350.0020.000
GD (Sp-mix)1Plot0.740.000.027.13*5.51*0.0010.044
Appendix 2—table 2
Dimension reduction of community-weighted mean trait values (CWMs) by varimax rotation principal component analysis (PCA).

Loadings and eigenvalues of rotation principal components (RC) selected from a varimax rotation PCA on the CWM of leaf traits (most influential variables in bold).

RC1RC2
LA0.490.11
SLA0.17–0.34
CHL0.050.86
LN–0.200.16
LC0.55–0.01
Explained41%23%
Cumulative explained41%64%
  1. LA, leaf area; SLA, specific leaf area; CHL, chlorophyll content; LN, leaf nitrogen content; LC, leaf carbon content.

Appendix 2—table 3
Results of linear models of leaf damage excluding undamaged leaves (this study) – leaf damage including undamaged leaves (from other plots of the BEF-China experiment) for the four species used in this study.

These models were used to correct the potential bias of herbivory estimates as a result of only collecting damaged leaves.

SpeciesSlopeInterceptR2Pearson’s correlation
Alniphyllum fortunei0.899702.834830.860.93
Cinnamomum camphora0.944652.074840.730.86
Daphniphyllum oldhamii0.883872.540320.920.96
Idesia polycarpa0.924061.775230.860.93
Appendix 2—table 4
Contrast coding of genetic diversity in species monocultures and species mixtures separately.

Sp-mono presents species monocultures, and Sp-mix presents species mixtures.

Species diversityGenetic diversityGenetic diversity in Sp-monoGenetic diversity in Sp-mix
11-10
410-1
1410
4401
Appendix 2—table 5
The interaction of significant results and the explained variance of the whole model of the multigroup structural equation models (SEM) shown in Figure 5.

*** p < 0.001; ** p < 0.01; * p < 0.05.

ResponsePredictorDFTest.StatExplained variance %
Tree functional diversityGD:SD10.5*11
HerbivoryGD:SD10.011
HerbivoryTree functional diversity:SD10.0*
Soil fungal diversityGD:SD152982.7***6
Tree community productivityGD:SD1327.6***24
Tree community productivityHerbivory:SD1327.6
Tree community productivitySoil fungal diversity:SD1327.6
Tree community productivityTree functional diversity:SD1327.6*
  1. SD, species diversity; GD, genetic diversity.

Appendix 3

To investigate the possible consequences of the zero functional diversity of genotype monocultures of single species (diversity treatment = 1.1), we further added SEM without the path between genetic diversity and functional diversity. We found that excluding the path between genetic diversity and functional diversity in the SEM models did not change the direction and significance of other paths. These results indicate possible artifacts brought in by the zero functional diversity in 1.1 communities do not affect the remaining effects that we found in the analyses.

Appendix 3—figure 1
Effects of tree diversity on higher trophic levels and tree community productivity without the path between genetic diversity and functional diversity (global Fisher’s C = 16.766, DF = 6, p = 0.01).

Positive and negative paths are indicated in green and orange, respectively. The standardized path coefficients are indicated by the numbers, statistical significance is indicated by asterisks (*** p < 0.0001, ** p < 0.001, * p < 0.05, and + p <0.1), and the explained variance of dependent variables is indicated by the percentage values. The gray dashed line indicates a nonsignificant (p > 0.1) pathway in the final model. To allow comparison, the same structural equation model (SEM) as for Figure 4 was used except excluding the path between genetic diversity and functional diversity.

Appendix 3—figure 2
Effects of tree genetic diversity on higher trophic levels and tree community productivity in tree species monocultures (a) and mixtures of four tree species (b) without the paths between genetic diversity and functional diversity.

The results were obtained by a multigroup structural equation model (SEM) (global Fisher’s C = 3.485, DF = 4, p = 0.480). Positive and negative paths are indicated in green and orange, respectively. The standardized path coefficients are indicated by the numbers, and statistical significance is indicated by asterisks (*** p < 0.0001, ** p < 0.001, * p < 0.05, and + p <0.1). Gray dashed lines indicate nonsignificant (p > 0.1) pathways in the final model. Here, tree functional diversity was calculated from traits measured on individual trees. To allow comparison, the same SEM as for Figure 5 was used. Multigroup SEM analyses first test the interaction (explanatory variable × groups) in the whole model using the full dataset and then estimate the local coefficient for each path by using different datasets (the full dataset or group sub-datasets [species richness = 1 or 4, respectively]) depending on the significance of explanatory variable × groups interactions. Thus, we could not get the percentage of the explained variance in the local multigroup SEM model.

Data availability

All numerical data were used to generate the figures that have been deposited in Dryad.

The following data sets were generated
    1. Liu X
    (2022) Dryad Digital Repository
    Gata from: Tree species and genetic diversity increase productivity via functional diversity and trophic feedbacks.
    https://doi.org/10.5061/dryadgf1vhhmqx

References

    1. Chao A
    (1984)
    Non-parametric estimation of the number of classes in a population
    Scandinavian Journal of Statistics 11:265–270.
  1. Software
    1. Laliberté E
    2. Legendre P
    3. Shipley B
    (2014)
    FD: measuring functional diversity from multiple traits, and other tools for functional ecology
    Package ‘FD.’.
  2. Book
    1. Snedecor WG
    2. Cochran GW
    (1989)
    Statistical Methods
    Iowa State University Press.

Decision letter

  1. David A Donoso
    Reviewing Editor; Escuela Politécnica Nacional, Ecuador
  2. Detlef Weigel
    Senior Editor; Max Planck Institute for Biology Tübingen, Germany

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Tree species and genetic diversity increase productivity via functional diversity and trophic feedbacks" for consideration by eLife. Your article has been reviewed by 2 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Detlef Weigel as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

Both reviewers provide a large set of questions on the experimental design, data collection, and data analyses that need to be fully addressed in a new version and a response letter. Please notice that formal acceptance depends on the quality of the new version and therefore not guaranteed at the moment.

Reviewer #1 (Recommendations for the authors):

Figure 3 shows that for monocultures with one genotype there is no functional diversity, which does of course not fully reflect reality. One way around this would be to know the functional diversity in these monocultures, e.g. if several individuals were measured in each monoculture. However, based on the Methods description (L291), traits were not measured in each plot but randomly across plots. This is a shortcoming, given that plant traits are known to change dependent on their plant neighbourhood. Alternatively, I think these monocultures should be removed from this figure.

Beyond functional diversity, the functional traits per se might be relevant and would also allow to test potential explanations for productivity differences between monocultures of single genotypes. For example, Community Weighted Means of traits could be calculated and used as explanatory variables for ecosystem functioning. Multivariate CWMs are also possible, e.g. using ordination.

L175-176: This tends to be true based on Figure 4, but not based on Figure 5. These differences suggest to me that the differences in the impact of genetic diversity on different ecosystem functions are probably not very consistent and clear. I would therefore be cautious with the interpretation of these results.

Figure 5: Regarding the multi-group SEM not declared as such in the figure legend, but in the Methods (L381): I would assume that the relationships between herbivory and productivity as well as between soil fungal diversity and productivity would be the same independent of species monocultures and mixtures, meaning that those relationships could be fixed among the different groups (i.e. species monoculture vs species mixture), while all other relationships could vary between groups.

L186-187: This title is a bit bold and does not reflect all the results, namely the direct effect of genetic diversity on productivity.

Reviewer #2 (Recommendations for the authors):

Lines 80-84: The impact of enemies can also be reduced in a mixed community by the effects of plant trait variation, for example, higher trait variation among plants can affect herbivore performance and herbivory see: Wetzel W, et al. 2016. Nature 539 (7629), 425-427; Bustos-Segura C. 2017. Ecology Letters, 20(1), 87-97

Line 99: Cook-Patton et al. (2011) used herbaceous plants, not trees.

Lines 174-176: This is not clear, actually in Figures 2 and 5 we observe a clear effect of genetic diversity on functional diversity in monocultures but weak in mixed species.

Lines 188-190: Please rewrite this sentence, it is not clear what type of diversity it is referring to and some statements do not match the results. The effects of genetic and species diversity were different. For example, SD has a positive indirect effect (mediated through fungal diversity) on productivity and GD has positive indirect effects (mediated through herbivory and fungal diversity) and a negative direct effect.

Line 190: It is also not clear which positive direct effect this is referring to.

Lines 193: Also there could be trade-offs between defense induction and growth.

Lines 254-256: Afforestation or reforestation? In the abstract "reforestation" is used.

Line 264: What was the separation among plots? Is it the field as in Bongers, et al. 2020? It would be good to have the diagram here as well.

Line 268: How many families were used per plant species? If more than four were used, how were they selected for each plot?

Lines 276-279: The design is not clear. What was the replication unit, plot, or subplot? It cannot be a different one among treatments, since the scale of the replicate has to be the same to be comparable. In Bongers et al. 2020, it is shown that treatment 1.1 has the same species within the plot, and treatment 1.4 has different species within the plot. This would be problematic since treatment plots of 1.4 have already a mix of species not just genotypes. If 1.1 has the same species within the subplot but is different among subplots, then it is clear that the replication unit is the subplot, but this has to be explicitly explained.

Line 277: The replication for 1.4 does not seem good enough, how is this justified? How reliable are the results for this treatment with only two plots? In addition, in Bongers et al. 2020, these two plots are shown on the same side of the land, which increases the likelihood that this is a biased result.

Lines 315-318: How non-damaged leaves were included? It is not clear how this regression was performed and how each variable was collected.

Line 326: Were the soil samples collected at the subplot or plot level? If they were collected at different scales between treatments, it would not be something comparable.

Line 355: If explanatory terms were analysed sequentially, it means that genetic diversity was analysed after the effects of species diversity were taken into account. How would an ANOVA type II differ from a sequential analysis? This could be justified in the methods, but also emphasize in the results/discussion, to allow for correct interpretation.

Figure 1c. Better name mother trees 1, 2, 3, and 4 to match the seed families naming.

Figure 2. Why the direct link between species diversity and productivity is not shown? Please, clarify this in the figure or methods. In addition, the paths could be changed to more contrasting colors to facilitate the reading for color-blind people.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Tree species and genetic diversity increase productivity via functional diversity and trophic feedbacks" for further consideration by eLife. Your revised article has been evaluated by Detlef Weigel (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

Both reviewers find that using species family means for trait values is problematic (for example, the assigning of zero values in Rao's Q calculations seems puzzling). While you have made a case for continuing to use family means in your R1 response, we find that a new opportunity must be given to sustain using these data, and explain possible artifacts when using zero values in, for example, your SEM paths. Possible alternatives (such as a SEM without the tests/paths between genotypic diversity and FDis due to colinearity issues when using family means) should be discussed (in an Appendix?), and the Discussion should also be extended, to make it clear this is a shortcoming of your analysis.

Reviewer #1 (Recommendations for the authors):

I have reviewed this revised version of the manuscript after already reviewing the original version (reviewer 1). In my opinion, the manuscript significantly improved and is much clear now. In particular, the detailed comments of reviewer 2 must have helped to clarify the methods, which in turn helped to better understand the results and discussion. I am overall happy with the analyses and also appreciate very much the inclusion of the results with traits measured at the individual level. I understand the reason of the authors with respect to circularity, but also maintain my concerns with using species family means, that do not only eliminate the potential responses of the genotypes to the treatments (which was the goal of the authors with using species family means) but also the potential effects of the genotypes. It is basically impossible to eliminate the response of a trait without also eliminating its effect. And in this study, I think this is particularly tricky because the results turn out to be significantly different when using species family means versus individual values, with important implications for the objectives of this study. And here I am in serious doubt whether the way the authors decided to go forward is the most appropriate one. I explain this based on the key result affected:

L263ff: I find the results of negative genotype diversity effects on productivity in species monocultures puzzling. And it, in fact, only appears in the SEM and only when using the species family means of trait values but not the measured trait values in each subplot. I can't really make a conclusion out of this, but it seems weird to me, and I really wonder to what extent this is due to the assumed 0 functional diversity of genotype monocultures (1.1 communities). I can't help but think this is an artifact. Actually, the SEM with the functional traits measured on individual trees matches much better the results observed in the binary analyses, where genotype diversity doesn't have an impact on functional diversity of species monocultures nor community productivity of species monocultures (Figure 2 and its supplement 1). It seems rather that increased functional diversity in species monocultures goes along with reduced productivity (Figure 3), which is, however independent of tree genetic diversity (as there is no clear relationship between tree genetic diversity and functional diversity of species monocultures as stated above – when individual trait values are used). So, it seems to me that the negative effect of tree genetic diversity on productivity in species monocultures, as claimed by the authors, is not a genetic diversity effect but a functional diversity effect independent of genetic diversity. That would at least be my interpretation of the results.

Reviewer #2 (Recommendations for the authors):

I thank the authors for thoroughly answering all the reviewers' questions and comments. I think the present version clarifies all the points raised and shows more clarity in the design and discussion.

I am glad to read this revised version and congratulate the authors for such an impressive and interesting study that represent a step forward in our understanding of the effects of plant diversity.

https://doi.org/10.7554/eLife.78703.sa1

Author response

Reviewer #1 (Recommendations for the authors):

Figure 3 shows that for monocultures with one genotype there is no functional diversity, which does of course not fully reflect reality. One way around this would be to know the functional diversity in these monocultures, e.g. if several individuals were measured in each monoculture. However, based on the Methods description (L291), traits were not measured in each plot but randomly across plots. This is a shortcoming, given that plant traits are known to change dependent on their plant neighbourhood. Alternatively, I think these monocultures should be removed from this figure.

Beyond functional diversity, the functional traits per se might be relevant and would also allow to test potential explanations for productivity differences between monocultures of single genotypes. For example, Community Weighted Means of traits could be calculated and used as explanatory variables for ecosystem functioning. Multivariate CWMs are also possible, e.g. using ordination.

Thank you for the valuable comments. We fully understand plant traits will change depending on their neighborhood. However, in our experiment, we hypothesized that functional diversity would be an explanatory variable that was used to explain how species and genetic diversity will affect multi-trophic interactions and community productivity. In this case, if we calculate functional diversity only based on the individuals planted in a particular subplot, trait plasticity in response to the neighborhood will be involved (see Sapijankas et al., 2014; Niklaus et al., 2017), which makes the analyses partly circular. In theory, we should use the mean trait value from common gardens to calculate functional diversity, but here we use the mean value across all the plots in our experiment to represent the realistic forest ecosystem scenery.

Lines 339–344: “To reduce circularity, we used the seed-family means across all species × genetic diversity combinations to calculate FDis values per subplot that did not only depend on the functional trait measures obtained in that particular subplot. Using traits measured in a particular subplot to calculate FDis for that subplot bears the risk that the measured traits reflect a response to the local environment, yet we want to use FDis as a predictor variable for the performance of that subplot.”

The zero functional diversity in species and genetic monoculture comes from using the mean trait value of each seed family of each species to calculate functional diversity. This method has been typically used in other studies (e.g. Craven et al., 2018; Williams et al., 2017), and we believe that it provides a proper estimation of functional diversity. However, we agree that the “real” functional diversity would not be zero even in the species and genetic monoculture, so now we have added functional diversity calculated by traits measured on individual trees across all subplots and put the results into the supplementary figures (Figure 2—figure supplement 1, Figure 3—figure supplement 1, Figure 4—figure supplement 1, Figure 5—figure supplement 1). With the “individual” functional diversity method we could calculate FDis also for subplots planted with trees of a single seed family (which had FDis values of zero when calculated with seed-family means). Overall, the results from the “individual” method still support our hypotheses that tree species diversity and genetic diversity affect community productivity via tree functional diversity and multi-trophic feedbacks (Figure 4—figure supplement 1, Figure 5—figure supplement 1). Although some effects of functional diversity disappeared or were weaker with FDis based on individuals than on seed-family means, this is understandable because FDis based on individuals includes more environmental response variation among individuals which can blur the effects of tree genetic diversity and species diversity. We added this FDis calculation to the Methods, reported the different results from the two FDis methods, and justified the weaker effects of functional diversity by this analysis in the Discussion.

Lines 354–359: “We further calculated FDis using traits measured on individual trees across all tree diversity treatment combinations. This alternative FDis had the advantage that it could also be calculated for subplots planted with trees of a single seed family (which had FDis values of zero when calculated with seed-family means), reflecting within seed-family functional trait diversity. The disadvantage is that this measure likely includes more response variation because every individual tree responds to a number of unknown factors in its local environment.”

Lines 130–133: “However, when we calculated functional diversity based on measurements taken on individual trees rather than based on seed-family means, only species diversity but not genetic diversity had effects on tree functional diversity (Figure 2—figure supplement 1), indicating additional within seed-family variation masking some of the between seed-family variation.”

Lines 140–143: “Tree functional diversity calculated using either seed-family means or individual-tree values had positive overall effects on community productivity but this effect was mainly due to an increase in functional diversity from species monocultures to mixtures (Figure 3a, Figure 3—figure supplement 1).”

Lines 164–170: “The analysis which functional diversity calculated from measurements on individual trees also showed that tree species diversity and genetic diversity affect community productivity via tree functional diversity and trophic feedbacks, although the effects of functional diversity were less pronounced (Figure 4—figure supplement 1), possibly because functional diversity calculated from individual trees included more response functional diversity (Sapijanskas et al. 2014) than did functional diversity calculated form seed-family means.”

Lines 176–178: “Similar results were found in the analyses using functional traits calculated from measurements on individual trees, except that the effects of functional diversity on tree productivity became positive (Figure 5—figure supplement 1).”

Lines 226–233: “To account for possible effects of functional diversity within seed families, we also calculated functional diversity based on measurements of individual trees (see Methods). Overall, the results from this novel method still support our hypotheses that tree species diversity and genetic diversity affect community productivity via tree functional diversity and multi-trophic feedbacks (Figure 4—figure supplement 1), although compared with the typically-used “mean” method, the novel method includes more variation among individuals which partly reflects responses of traits to the particular local environment (Sapijanskas et al. 2014); and this may have blurred the mean effects of tree genetic diversity and species diversity (Figure 4, Figure 4—figure supplement 1).”

We agree that community weighted means (CWM) of traits could also be an explanatory variable for ecosystem functioning. However, in our experiment, due to the equal representation of seed families across tree diversity treatments (Appendix 1—table 1), we assume that there must be no effects of tree species and genetic diversity on community weighted means (CWMs) of tree functional traits. Now we have calculated CWM and statistically proved that there are no significant effects of tree diversity on CWM (Appendix 2—table 1; actually, the corresponding F-values are so small that they can only be explained with the assumed constancy due to equal representation of seed families across diversity treatments).

Lines 359–366: “We also calculated community weighted means (CWMs) for the five functional traits. To obtain a multivariate equivalent we subjected the individual traits to a varimax rotation principal component analysis (PCA) to obtain two orthogonal axes as principal-component CWMs. The two principal components captured together 64% variation of trait variation (Appendix 2—table 2). PC1 indicated the functional traits directly connected with growth and PC2 indicated the functional traits connected with photosynthesis (Appendix 2—figure 3). The varimax rotation principal component analysis was done by ‘psych’ R package version 2.1.9 (Makowski, 2018).”

Lines 136–139: “Due to the equal representation of seed families across tree diversity treatments (Appendix 1—table 1), we did not find any significant effects of tree species and genetic diversity effects on community weighted means (CWMs) of tree functional traits (Appendix 2—table 1).”

L175-176: This tends to be true based on Figure 4, but not based on Figure 5. These differences suggest to me that the differences in the impact of genetic diversity on different ecosystem functions are probably not very consistent and clear. I would therefore be cautious with the interpretation of these results.

In this sentence, we only wanted to state the relative strength of effects of species diversity and genetic diversity which could be derived from Figure 4, which we now pointed out more clearly in the text. Meanwhile, we have further discussed the different effects of genetic diversity via functional diversity on productivity between species monocultures and mixtures in the following discussion.

Lines 190–194: “Nevertheless, compared with effects of species diversity, effects of genetic diversity on tree community productivity through functional diversity were weaker, whereas effects of genetic diversity on trophic interactions were strong (see Figure 4), indicating that mechanisms underpinning effects of genetic diversity may in part differ from those underpinning effects of species diversity, as we will discuss below.”

Lines 256–264: “However, the hypothesized positive effects of tree functional diversity on productivity turned negative in species monoculture. This result indicates that functional diversity may not have positive effects on the ecosystem functioning under low environmental heterogeneity, i.e. species monocultures in our study (Hillebrand and Matthiessen 2009). Therefore, our findings show that the different effects of genetic diversity on tree productivity between species monocultures and mixtures not only depend on the different effects of genetic diversity on functional diversity and trophic interaction, but also on the varied tree productivity consequences from functional diversity and trophic interaction on tree productivity between species monocultures and mixtures.”

Figure 5: Regarding the multi-group SEM not declared as such in the figure legend, but in the Methods (L381): I would assume that the relationships between herbivory and productivity as well as between soil fungal diversity and productivity would be the same independent of species monocultures and mixtures, meaning that those relationships could be fixed among the different groups (i.e. species monoculture vs species mixture), while all other relationships could vary between groups.

In our study, we hypothesized that all the paths are possible to be different between species monocultures and species mixtures. So, we did not constrain any path beforehand, the SEM results showed the differences between species monocultures and species mixtures according to the data itself. Now we have declared this in the figure legend and added a table in the Appendix (Appendix 2—table 5) to show the interaction results tested from the multi-group SEM model.

Lines 822–825: “All the paths were allowed to be different between species monocultures and mixtures (none of the paths was constrained manually beforehand); the interaction statistics of the multi-group model are shown in Appendix 2—table 5.”

L186-187: This title is a bit bold and does not reflect all the results, namely the direct effect of genetic diversity on productivity.

We did not include the “direct effects” of genetic diversity on productivity in this title because we could not test if the “direct” effects in the SEM are real direct effects or just because of some other mediating factors which we did not include in this study. Thus, we interpreted the direct effects as the variation of plant community productivity we failed to explain by functional diversity and trophic feedbacks we included in this study instead of a clear result. However, we agree that the title of the section was a bit bold and thus inserted the word “mainly” in front of “via functional diversity and trophic feedbacks” (line 203).

Lines 234–242: “Even after accounting for tree functional diversity and trophic feedbacks, we still detected a direct negative effect of tree genetic diversity on tree productivity, while the direct effect of tree species diversity was fully explained by functional diversity and trophic feedbacks. This suggests that aspects of genetic diversity that do not contribute to functional diversity or trophic interactions as measured in this study may reduce ecosystem functioning, e.g. due to trade-offs between genetic diversity and species diversity. For example, it has been shown that in species-diverse grassland ecosystems niche-complementarity between species can increase at the expense of reduced variation within species (van Moorsel et al., 2018; van Moorsel et al., 2019; Zuppinger-Dingley et al., 2014; Zvereva et al., 2012).”

Reviewer #2 (Recommendations for the authors):

Lines 80-84: The impact of enemies can also be reduced in a mixed community by the effects of plant trait variation, for example, higher trait variation among plants can affect herbivore performance and herbivory see: Wetzel W, et al. 2016. Nature 539 (7629), 425-427; Bustos-Segura C. 2017. Ecology Letters, 20(1), 87-97

Thank you very much for recommending these two important papers, now we have added this mechanism and the references to the main text.

Lines 77–81: “Trophic feedbacks can enhance the performance of species or genotype mixtures either by reducing herbivore damage though enhancing the diversity of nutrient traits (Wetzel et al., 2016) and chemical traits (Bustos-Segura et al., 2017) or enhancing diversity of beneficial mutualists (e.g. mycorrhizal fungi (Semchenko et al., 2018)) (Figure 1b).”

Line 99: Cook-Patton et al. (2011) used herbaceous plants, not trees.

Thank you for pointing out this reference mistake, now we have replaced the reference with an experimental study (Abdala-Roberts et al., 2015) that used trees to compare the importance of species diversity and genetic diversity.

Lines 94–98: “Although there are few forest experimental studies in which species and genetic diversity are simultaneously manipulated, most of them only compared their relative importance on ecosystem functions (Abdala-Roberts et al., 2015; Koricheva and Hayes, 2018), and we barely know their interactive effects via functional diversity and trophic feedbacks on plant productivity.”

Lines 174-176: This is not clear, actually in Figures 2 and 5 we observe a clear effect of genetic diversity on functional diversity in monocultures but weak in mixed species.

Thank you for pointing out the unclear description. Now we have rewritten this sentence and made clear that here compared with the effects of species diversity, the effects of genetic diversity on tree community productivity through functional diversity were weaker. Additionally, we have discussed the different effects of genetic diversity via functional diversity between species monocultures and mixtures in the following discussion.

Lines 190–194: “Nevertheless, compared with effects of species diversity, effects of genetic diversity on tree community productivity through functional diversity were weaker, whereas effects of genetic diversity on trophic interactions were strong (see Figure 4), indicating that mechanisms underpinning effects of genetic diversity may in part differ from those underpinning effects of species diversity, as we will discuss below.”

Lines 256–264: “However, the hypothesized positive effects of tree functional diversity on productivity turned negative in species monoculture. This result indicates that functional diversity may not have positive effects on the ecosystem functioning under low environmental heterogeneity, i.e. species monocultures in our study (Hillebrand and Matthiessen 2009). Therefore, our findings show that the different effects of genetic diversity on tree productivity between species monocultures and mixtures not only depend on the different effects of genetic diversity on functional diversity and trophic interaction, but also on the varied tree productivity consequences from functional diversity and trophic interaction on tree productivity between species monocultures and mixtures.”

Lines 188-190: Please rewrite this sentence, it is not clear what type of diversity it is referring to and some statements do not match the results. The effects of genetic and species diversity were different. For example, SD has a positive indirect effect (mediated through fungal diversity) on productivity and GD has positive indirect effects (mediated through herbivory and fungal diversity) and a negative direct effect.

Thank you for pointing out the unclear place. Here we only discuss the effects of species diversity and genetic diversity on productivity and trophic interactions via functional diversity; we have now rewritten this sentence.

Lines 205–213: “Although only species diversity but not genetic diversity was found to affect tree productivity in binary analyses, both kinds of diversity positively affected tree community productivity and trophic interactions via functional diversity according to our structural equation models (SEMs) depicted in the corresponding path-analysis diagrams (see Figure 4). Tree functional diversity appeared to enhance complementary resource acquisition at community level (Kahmen et al., 2006; Marquard et al., 2009; Williams et al., 2017), which consequently enhanced tree community productivity. Meanwhile, tree functional diversity also provided more niche opportunities to benefit generalist herbivores and soil fungi which reduced tree community productivity, as has been found for these tree species in a parallel field study nearby (Brezzi et al., 2017).”

Line 190: It is also not clear which positive direct effect this is referring to.

We now have specified the “positive direct effect” is the effect of tree functional diversity on tree community productivity.

Lines 208–211: “Tree functional diversity appeared to enhance complementary resource acquisition at community level (Kahmen et al., 2006; Marquard et al., 2009; Williams et al., 2017), which consequently enhanced tree community productivity.”

Lines 193: Also there could be trade-offs between defense induction and growth.

Thank you for this useful comment. Now we have added this statement in this part of the Discussion.

Lines 213–216: “It is expected that herbivory has negative effects on plant productivity via the reduction of leaf area (Zvereva et al., 2012) and photosynthesis of remaining leaves (Nabity et al., 2009), and via trade-offs between growth and herbivore defense (Züst and Agrawal, 2017).”

Lines 254-256: Afforestation or reforestation? In the abstract "reforestation" is used.

Thank you very much for pointing out the inconsistent words. Now we have changed all of them into “afforestation” since we do not refer to replanting of a forest as it has been before in this study.

Line 264: What was the separation among plots? Is it the field as in Bongers, et al. 2020? It would be good to have the diagram here as well.

Thank you very much for this suggestion. Now we have added a plot design diagram in the Appendix (Appendix 1—figure 1) which shows the species and seed families used in each plot. Additionally, we have added a table to better describe how many individuals of how many seed families we planted in our experiment (Appendix 1—table 1). The field is as in Bongers et al. (2020), Figure 1, which we now cite specifically in the Methods (line 314) and include here for convenience:

The side length or each plot is about 26 m (square root of 1 mu=1/15 ha), thus form the figure it can be seen that distances between plots of the same treatment were mostly more than 50 m, but distance was not controlled because treatments were allocated randomly.

Line 268: How many families were used per plant species? If more than four were used, how were they selected for each plot?

In total, we used eight seed families for every species and we randomly selected the seed-family combination for different tree-diversity combinations, and made sure that every seed family occurred the same number of times at each diversity level. Now we have added this information to the Methods and a table (Appendix 1—table 1) to explain that.

Lines 309–314: “For each of the four species, we collected seeds from eight mother trees to allow for two replications of four-family mixtures per species. Furthermore, to avoid the effects of unequal representation of particular seed families and correlations between seed family presence and diversity treatments we made sure that every seed family occurred the same number of times at each diversity level (see Appendix 1—table 1, small deviations from the rule were required where not enough seeds from a seed family could be obtained).”

Lines 276-279: The design is not clear. What was the replication unit, plot, or subplot? It cannot be a different one among treatments, since the scale of the replicate has to be the same to be comparable. In Bongers et al. 2020, it is shown that treatment 1.1 has the same species within the plot, and treatment 1.4 has different species within the plot. This would be problematic since treatment plots of 1.4 have already a mix of species not just genotypes. If 1.1 has the same species within the subplot but is different among subplots, then it is clear that the replication unit is the subplot, but this has to be explicitly explained.

As the answer to the above two questions, now we have added the plot design in the Appendix to better present the experimental design. We have different units for species monocultures and species mixtures due to budget limitations and the number of replicates required per single seed family. For treatment 1.4, the replication unit is subplot (0.25 mu).

We agree that ideally we would have wanted to grow all experimental communities in plots of 400 trees, but due to the difficulty to obtain enough seeds from a single family we used subplots. To ensure that diversity treatments were still the same at plot level, we grouped different family identities of the same species and different species identities for treatments 1.1 and 1.4, respectively. The first was to make sure we still had species monocultures per plot, which was not necessary for the second. Admittedly, it would have been better to spread diversity treatment 1.4 over more plots, but that would have required to plant subplots among the plot grid or to combine them with other treatments in neighboring subplots, and still would then not have allowed us to have diversity treatment 1.4 at plot level, a requirement which we have achieved with the current design. For testing effects of diversity treatments we of course used plot as error (random-effects term in the mixed models), so these subplots of diversity treatments 1.1 and 1.4 were not used as pseudoreplicates.

We have added more detail to the Methods, but not as much as provided in this reply as we believe it would make the description unneccessarily complicated. However, we could of course add more if requested.

Lines 314–322: “Due to budget limitations and the number of replicates required per single seed family, the 1.1 and 1.4 diversity treatments were applied at subplot level (0.25 mu) and replicated 32 and 8 times, respectively. The 4.1 and 4.4 diversity treatments were applied at plot level (1 mu) and were replicated 8 and 6 times, respectively (Appendix 1—figure 1; see also Figure 1 in Bongers et al., 2020). To allow for simpler analysis, we obtained most community measures at subplot level also for the 4.1 and 4.4 diversity treatments and thereafter used the subplots for all tests of diversity effects on these community measures, including plots as error (i.e. random-effects) term for testing the diversity effects in the corresponding mixed models.”

Lines 429–433 “For all linear mixed-effects models, we used ‘plot’ as a random variable since subplots were nested in plots. This also ensured that fixed terms whose levels did not vary within plots among subplots (specifically the four diversity treatments) were correctly tested against the variation among plots rather than the residual variation among subplots.”

Line 277: The replication for 1.4 does not seem good enough, how is this justified? How reliable are the results for this treatment with only two plots? In addition, in Bongers et al. 2020, these two plots are shown on the same side of the land, which increases the likelihood that this is a biased result.

As we mentioned in the answer to the above question, due to budget limitations and the number of replicates required per single seed family, we used subplot as a replication unit for the 1.4 treatment (Appendix 1—figure 1), so here we have 8 replications for the 1.4 treatment. As we mentioned before, the plot distribution is designed to be completely random in the experiment site which is independent of the diversity treatments. Although the two 1.4 plots by chance were located only a bit more than 100 m apart, there was nothing we could do about this, and we believe that the distance was large enough to avoid biased results. As explained above, the reason for grouping the 1.4 subplots of different species in two single plots was to have this treatment also represented at plot scale.

Lines 314–316: “Due to budget limitations and the number of replicates required per single seed family, the 1.1 and 1.4 diversity treatments were applied at subplot level (0.25 mu) and replicated 32 and 8 times, respectively.”

Lines 315-318: How non-damaged leaves were included? It is not clear how this regression was performed and how each variable was collected.

We used a data set in which all the leaves were sampled (both damaged leaves and non-damaged leaves) from the same experimental site. Then we established linear models from this dataset to relate the herbivore damage sampled from both damaged leaves and non-damaged leaves with the herbivore damage only sampled from damaged leaves. We found strong correlations of the herbivore damage results from the two sampling methods, so we used the linear models to correct the potential bias of our data which only sampled the damaged leaves. Now we have rewritten the description in the Methods section.

Lines 375–385: “Because we only collected damaged leaves in this study, we might have overestimated the herbivory per individual tree. We therefore used data from other plots of the BEF-China experiment (Schuldt et al., 2015) which did not exclude non-damaged leaves to correct the potential bias. This former study assessed herbivore damage by visually inspecting 21 leaves (7 leaves per branch) on three random branches from different parts of the canopy (Schuldt et al., 2015). They used the mean percentage damage value as the overall leaf damage for each individual. We related leaf damage of corresponding tree individuals from this former study (total leaf damage) to leaf damage excluding non-damaged leaves (damage per damaged leaf) for all four species by linear regression (Pearson's correlation = 0.86–0.96, P < 0.001) (Appendix 2—table 3). With these regression models, we got the predicted values of herbivory for our study and used these predicted values in the final analyses.”

Line 326: Were the soil samples collected at the subplot or plot level? If they were collected at different scales between treatments, it would not be something comparable.

As mentioned in the answer to the plot design, we collected soil samples based on the species and genetic diversity treatment-replication unit and we also used “plot” as a random factor when we tested the effects of species diversity and genetic diversity on soil fungal diversity (see above answers) to solve this problem from sampling at different scales between treatments. Now we have clarified this in the Methods section.

Lines 390–393: “Soil fungal diversity was used as a proxy of unspecified trophic interactions. To be consistent with the species and genetic diversity treatment design, soil samples were taken on subplot level for the 1.1 and 1.4 diversity treatments, but on plot level for the 4.1 and 4.4 diversity treatments in 2017.”

Lines 429–433: “For all linear mixed-effects models, we used ‘plot’ as a random variable since subplots were nested in plots. This also ensured that fixed terms whose levels did not vary within plots among subplots (specifically the four diversity treatments) were correctly tested against the variation among plots rather than the residual variation among subplots.”

Line 355: If explanatory terms were analysed sequentially, it means that genetic diversity was analysed after the effects of species diversity were taken into account. How would an ANOVA type II differ from a sequential analysis? This could be justified in the methods, but also emphasize in the results/discussion, to allow for correct interpretation.

Because this was a designed experiment with orthogonally crossed species and genetic diversity treatments, ANOVA type I and II yield the same results (very slight differences due to missing values). However, we used a different approach, as indicated above, to the typical two-way ANOVA with interaction. Instead of fitting genetic diversity and interaction with fitted genetic diversity separately for species monocultures and the species mixture (note that this uses exactly the same sum of squares and degrees of freedom—for further detail see above). In Figure 2 we still show the main effect of genetic richness as well. However, as we found our alternative contrast coding more biologically meaningful and used it also for the separate SEMs, we focus the analysis on this contrast coding. Of course, also these contrasts are orthogonal and thus again independent of fitting sequence. In the revised version we tried to better explain the use of this alternative contrast coding.

Lines 420–428: “To determine how species and genetic diversity and their interaction affected tree functional diversity and trophic interactions, linear mixed-effects models (LMMs) were fitted with two types of contrast coding. In the first, we used the ordinary 2-way analysis of variance with interaction and in the second we replaced the genetic diversity main effect and the interaction with separate genetic diversity effects for species monocultures and the species mixture (Appendix 2—table 4). Note that as our design was orthogonal, fitting sequence did not matter in either of the codings. However, we focused our major analysis on the second type of coding to make it consistent with our hypotheses. Main effects of genetic diversity are presented in inset panels in Figure 2.”

Lines 120–121: “Using linear mixed-model analyses, we tested the effects of species diversity and genetic diversity within species on trophic interactions and community productivity.”

Figure 1c. Better name mother trees 1, 2, 3, and 4 to match the seed families naming.

Thank you for the suggestion. We have named the mother trees as “1, 2, 3, 4” to match with the names of seed families (see Figure 1c).

Figure 2. Why the direct link between species diversity and productivity is not shown? Please, clarify this in the figure or methods. In addition, the paths could be changed to more contrasting colors to facilitate the reading for color-blind people.

We removed the direct link between species diversity and productivity because it was a non-informative pathway and its removal reduced AICc of the path models. Now we have justified this in the Methods and the figure legend.

Lines 452–454: “We sequentially dropped non-informative pathways, if their removal reduced the AICc of the path models by more than 2 (Grace, 2006).”

Lines 794–797: “The direct effect of tree species diversity on tree community productivity was removed in the model because it was not significant (P > 0.5) and the removal reduced the AICc by more than 2 (ΔAICc = 3.269).”

Thank you for your suggestion for the color, now we have changed the colors of the paths and we used an online tool (https://coolors.co/ca5100-3f979a) to make sure the current colors are viewer-friendly.

References:

Abdala-Roberts, L., Mooney, K. A., Quijano-Medina, T., Campos-Navarrete, M. J., Gonzalez-Moreno, A., and Parra-Tabla, V. (2015). Comparison of tree genotypic diversity and species diversity effects on different guilds of insect herbivores. Oikos, 124(11), 1527-1535. https://doi.org/10.1111/oik.02033

Bardgett, R. D., Mommer, L., and De Vries, F. T. (2014). Going underground: root traits as drivers of ecosystem processes. Trends in Ecology and Evolution, 29(12), 692-699. https://doi.org/10.1016/j.tree.2014.10.006

Bates, D., Mächler, M., Bolker, B., and Walker, S. (2015). Fitting Linear Mixed-Effects Models Using lme4. Journal of Statistical Software, 67(1), 1 – 48. https://doi.org/10.18637/jss.v067.i01

Bruelheide, H., Nadrowski, K., Assmann, T., Bauhus, J., Both, S., Buscot, F., Chen, X. Y., Ding, B. Y., Durka, W., Erfmeier, A., Gutknecht, J. L. M., Guo, D. L., Guo, L. D., Hardtle, W., He, J. S., Klein, A. M., Kuhn, P., Liang, Y., Liu, X. J., Michalski, S., Niklaus, P. A., Pei, K. Q., Scherer-Lorenzen, M., Scholten, T., Schuldt, A., Seidler, G., Trogisch, S., von Oheimb, G., Welk, E., Wirth, C., Wubet, T., Yang, X. F., Yu, M. J., Zhang, S. R., Zhou, H. Z., Fischer, M., Ma, K. P., and Schmid, B. (2014). Designing forest biodiversity experiments: general considerations illustrated by a new large experiment in subtropical China. Methods in Ecology and Evolution, 5(1), 74-89. https://doi.org/Doi 10.1111/2041-210x.12126

Bustos-Segura, C., Poelman, E. H., Reichelt, M., Gershenzon, J., and Gols, R. (2017). Intraspecific chemical diversity among neighbouring plants correlates positively with plant size and herbivore load but negatively with herbivore damage. Ecology Letters, 20(1), 87-97. https://doi.org/https://doi.org/10.1111/ele.12713

Craven, D., Eisenhauer, N., Pearse, W. D., Hautier, Y., Isbell, F., Roscher, C., Bahn, M., Beierkuhnlein, C., Bönisch, G., Buchmann, N., Byun, C., Catford, J. A., Cerabolini, B. E. L., Cornelissen, J. H. C., Craine, J. M., De Luca, E., Ebeling, A., Griffin, J. N., Hector, A., Hines, J., Jentsch, A., Kattge, J., Kreyling, J., Lanta, V., Lemoine, N., Meyer, S. T., Minden, V., Onipchenko, V., Polley, H. W., Reich, P. B., van Ruijven, J., Schamp, B., Smith, M. D., Soudzilovskaia, N. A., Tilman, D., Weigelt, A., Wilsey, B., and Manning, P. (2018). Multiple facets of biodiversity drive the diversity–stability relationship. Nature Ecology and Evolution, 2(10), 1579-1587. https://doi.org/10.1038/s41559-018-0647-7

Grace, J. B. (2006). Structural Equation Modeling and Natural Systems. Cambridge University Press. https://doi.org/DOI: 10.1017/CBO9780511617799

Hillebrand, H., and Matthiessen, B. (2009). Biodiversity in a complex world: consolidation and progress in functional biodiversity research. Ecology Letters, 12(12), 1405-1419. https://doi.org/10.1111/j.1461-0248.2009.01388.x

Kahmen, A., Renker, C., Unsicker, S. B., and Buchmann, N. (2006). Niche complementarity for nitrogen: An explanation for the biodiversity and ecosystem functioning relationship? Ecology, 87(5), 1244-1255. https://doi.org/10.1890/0012-9658(2006)87[1244:Ncfnae]2.0.Co;2

Koricheva, J., and Hayes, D. (2018). The relative importance of plant intraspecific diversity in structuring arthropod communities: A meta-analysis. Functional Ecology, 32(7), 1704-1717. https://doi.org/10.1111/1365-2435.13062

Kuznetsova, A., Brockhoff, P. B., and Christensen, R. H. B. (2017). lmerTest Package: Tests in Linear Mixed Effects Models. Journal of Statistical Software, 82(13), 1-26. https://doi.org/10.18637/jss.v082.i13

Marquard, E., Weigelt, A., Temperton, V. M., Roscher, C., Schumacher, J., Buchmann, N., Fischer, M., Weisser, W. W., and Schmid, B. (2009). Plant species richness and functional composition drive overyielding in a six-year grassland experiment. Ecology, 90(12), 3290-3302.https://doi.org/10.1890/09-0069.1

Nabity, P. D., Zavala, J. A., and DeLucia, E. H. (2009). Indirect suppression of photosynthesis on individual leaves by arthropod herbivory. Annals of Botany, 103(4), 655-663. https://doi.org/10.1093/aob/mcn127

Niklaus, P. A., Baruffol, M., He, J.-S., Ma, K., and Schmid, B. (2017). Can niche plasticity promote biodiversity–productivity relationships through increased complementarity? Ecology, 98(4), 1104-1116.https://doi.org/10.1002/ecy.1748

Sapijanskas, J., Paquette, A., Potvin, C., Kunert, N., and Loreau, M. (2014). Tropical tree diversity enhances light capture through crown plasticity and spatial and temporal niche differences. Ecology, 95(9), 2479-2492. https://doi.org/10.1890/13-1366.1

Schuldt, A., Bruelheide, H., Härdtle, W., Assmann, T., Li, Y., Ma, K., von Oheimb, G., and Zhang, J. (2015). Early positive effects of tree species richness on herbivory in a large-scale forest biodiversity experiment influence tree growth. Journal of Ecology, 103(3), 563-571. https://doi.org/10.1111/1365-2745.12396

Semchenko, M., Leff, J. W., Lozano, Y. M., Saar, S., Davison, J., Wilkinson, A., Jackson, B. G., Pritchard, W. J., De Long, J. R., Oakley, S., Mason, K. E., Ostle, N. J., Baggs, E. M., Johnson, D., Fierer, N., and Bardgett, R. D. (2018). Fungal diversity regulates plant-soil feedbacks in temperate grassland. Science Advances, 4(11), eaau4578. https://doi.org/10.1126/sciadv.aau4578

van Moorsel, S. J., Hahl, T., Wagg, C., De Deyn, G. B., Flynn, D. F. B., Zuppinger-Dingley, D., and Schmid, B. (2018). Community evolution increases plant productivity at low diversity. Ecology Letters, 21(1), 128-137. https://doi.org/10.1111/ele.12879

van Moorsel, S. J., Schmid, M. W., Wagemaker, N. C. A. M., van Gurp, T., Schmid, B., and Vergeer, P. (2019). Evidence for rapid evolution in a grassland biodiversity experiment. Molecular Ecology, 28(17), 4097-4117.https://doi.org/10.1111/mec.15191

Wetzel, W. C., Kharouba, H. M., Robinson, M., Holyoak, M., and Karban, R. (2016). Variability in plant nutrients reduces insect herbivore performance. Nature, 539(7629), 425-427. https://doi.org/10.1038/nature20140

Williams, L. J., Paquette, A., Cavender-Bares, J., Messier, C., and Reich, P. B. (2017). Spatial complementarity in tree crowns explains overyielding in species mixtures. Nature Ecology and Evolution, 1(4), 0063. https://doi.org/10.1038/s41559-016-0063

Zuppinger-Dingley, D., Schmid, B., Petermann, J. S., Yadav, V., De Deyn, G. B., and Flynn, D. F. B. (2014). Selection for niche differentiation in plant communities increases biodiversity effects. Nature, 515(7525), 108-111. https://doi.org/10.1038/nature13869

Züst, T., and Agrawal, A. A. (201). Trade-Offs Between Plant Growth and Defense Against Insect Herbivory: An Emerging Mechanistic Synthesis. Annual Review of Plant Biology, 68(1), 513-534. https://doi.org/10.1146/annurev-arplant-042916-040856

Zvereva, E. L., Zverev, V., and Kozlov, M. V. (2012). Little strokes fell great oaks: minor but chronic herbivory substantially reduces birch growth. Oikos, 121(12), 2036-2043.https://doi.org/10.1111/j.1600-0706.2012.20688.x

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

Both reviewers find that using species family means for trait values is problematic (for example, the assigning of zero values in Rao's Q calculations seems puzzling). While you have made a case for continuing to use family means in your R1 response, we find that a new opportunity must be given to sustain using these data, and explain possible artifacts when using zero values in, for example, your SEM paths. Possible alternatives (such as a SEM without the tests/paths between genotypic diversity and FDis due to colinearity issues when using family means) should be discussed (in an Appendix?), and the Discussion should also be extended, to make it clear this is a shortcoming of your analysis.

Thanks for the positive comments on the revision. We agree that the zero values of functional diversity in genetic monocultures of single species (1.1 diversity level) can be criticized, but we also believe the mean method has the advantages we mention in cases where it is not zero by definition. We now discuss the caveats and increase focus on the new method using individual trait values to calculate functional diversity as:

“We found that tree species diversity increased tree productivity via increased tree functional diversity, reduced soil fungal diversity and marginally reduced herbivory. The effects of tree genetic diversity on productivity via functional diversity and soil fungal diversity were negative in monocultures but positive in the mixture of the four tree species tested.” (lines 41-45)

“The results obtained with functional traits calculated from measurements on individual trees showed weaker effects of genetic diversity on functional diversity (path coefficient = 0.193 vs 0.883) but did not change the significance and direction of the effects of genetic diversity on productivity via functional diversity in species monoculture. Additionally, the effects of functional diversity on tree productivity in species mixtures were positive when using functional diversity calculated from measurements on individual trees but were non-significant when using functional diversity calculated from seed-family means (Figure 5, Figure 5—figure supplement 1).” (lines 178-185)

“Using functional diversity calculated from measurements on individual trees did not change the effects of genetic diversity via trophic feedbacks, except that the effects of herbivory on productivity became non-significant from marginally significant.” (lines 188-190)

“Regarding our first hypothesis, we found that tree species diversity and genetic diversity can increase tree community productivity via increased functional diversity and trophic feedbacks as predicted. This suggests complementary resource-use and biotic niches, respectively, as mechanisms underpinning the biodiversity effects (Turnbull et al., 2016). Nevertheless, compared with effects of species diversity, effects of genetic diversity on tree community productivity through functional diversity were weaker, whereas effects of genetic diversity on trophic interactions were strong (see Figure 4, Figure 4—figure supplement 1), indicating that mechanisms underpinning effects of genetic diversity may in part differ from those underpinning effects of species diversity, as we will discuss below. Regarding our second hypothesis, we found that the effects of tree genetic diversity on productivity via functional diversity and soil fungal diversity were negative in tree species monocultures but positive in the species mixture, which differed from our predictions. In the following, we discuss these results in more detail.” (lines 200-212)

“At the same time, the results indicate that the seed-family means method may bring an artifact to the effect of genetic diversity on functional diversity because of the zero value of functional diversity in genetic monocultures of single species (1.1 communities). However, excluding the path between genetic diversity and functional diversity did not affect remaining paths, indicating that the partly artificial relationship between genetic diversity and functional diversity did not distort the path model in general (Figure 4, Appendix 3—figure 1).” (lines 244-250)

“In contrast of our second hypothesis, we found that the effects of genetic diversity via functional diversity and soil fungal diversity were negative in species monocultures but not significant via functional diversity and positive via soil fungal diversity in the species mixture (Figure 5). We found genetic diversity had positive effects on tree functional diversity and soil fungal diversity in species monocultures but negative effects in the species mixture, which supports the trade-offs between genetic and species diversity discussed in the previous section.” (lines 269-274)

“The two methods of calculating functional diversity either from seed-family means or from trait values of individual trees yielded different results regarding the indirect effects of genetic diversity on tree productivity via functional diversity. The method based on seed-family means has the advantage to be less circular whereas the method based on trait values of individuals has the advantage of producing functional diversity values >0 also for genetic monocultures of single species (1.1 communities; see Methods). The weaker indirect effects of genetic diversity on tree productivity via functional diversity in the method using trait values of individuals suggests that the zero value of functional diversity in 1.1 communities in the method using seed-family means may lead to an overestimation of these indirect effects of genetic diversity in species monocultures. Nevertheless, the method using seed-family means is still useful for species monocultures with multiple seed families and for species mixtures.” (lines 283-293)

We have added SEMs without the paths between genetic diversity and functional diversity as Appendix 3 (Appendix 3—figure 1, Appendix 3—figure 2). Fortunately, the remaining path coefficients were not distorted by functional diversity and changed only little if the path from genetic diversity to functional diversity was omitted. We also added multi-group SEMs in which the path between genetic diversity and functional diversity calculated from seed-family means was excluded, the coefficients of the remaining paths also were not changed (Appendix 3—figure 2). We have included these analyses in the new version (lines 170-172, lines 190-193, lines 237 -250, lines 479-481 and lines 487-488).

Reviewer #1 (Recommendations for the authors):

I have reviewed this revised version of the manuscript after already reviewing the original version (reviewer 1). In my opinion, the manuscript significantly improved and is much clear now. In particular, the detailed comments of reviewer 2 must have helped to clarify the methods, which in turn helped to better understand the results and discussion. I am overall happy with the analyses and also appreciate very much the inclusion of the results with traits measured at the individual level. I understand the reason of the authors with respect to circularity, but also maintain my concerns with using species family means, that do not only eliminate the potential responses of the genotypes to the treatments (which was the goal of the authors with using species family means) but also the potential effects of the genotypes. It is basically impossible to eliminate the response of a trait without also eliminating its effect. And in this study, I think this is particularly tricky because the results turn out to be significantly different when using species family means versus individual values, with important implications for the objectives of this study. And here I am in serious doubt whether the way the authors decided to go forward is the most appropriate one. I explain this based on the key result affected:

Thanks for the constructive comments on the revision. We agree that the seed-family means method is not a perfect way to show the functional variation among genotypes, especially for the 1.1 communities. However, because this is the common method used in the literature and because it has some advantages when it is not zero by definition, we would still like to use it, but now with the increased focus on the new method using trait values of individual trees to calculate functional diversity. We have extended our discussion and adjusted the interpretation by comparing the results obtained with the two methods (see the above text pieces taken from the main text in the response to the editor’s comments).

L263ff: I find the results of negative genotype diversity effects on productivity in species monocultures puzzling. And it, in fact, only appears in the SEM and only when using the species family means of trait values but not the measured trait values in each subplot. I can't really make a conclusion out of this, but it seems weird to me, and I really wonder to what extent this is due to the assumed 0 functional diversity of genotype monocultures (1.1 communities). I can't help but think this is an artifact. Actually, the SEM with the functional traits measured on individual trees matches much better the results observed in the binary analyses, where genotype diversity doesn't have an impact on functional diversity of species monocultures nor community productivity of species monocultures (Figure 2 and its supplement 1). It seems rather that increased functional diversity in species monocultures goes along with reduced productivity (Figure 3), which is, however independent of tree genetic diversity (as there is no clear relationship between tree genetic diversity and functional diversity of species monocultures as stated above – when individual trait values are used). So, it seems to me that the negative effect of tree genetic diversity on productivity in species monocultures, as claimed by the authors, is not a genetic diversity effect but a functional diversity effect independent of genetic diversity. That would at least be my interpretation of the results.

In binary analyses, we did not find the overall negative effects of genetic diversity on productivity, neither in monoculture nor mixture (Figure 2a). However, we found genetic diversity had overall positive effects on functional diversity calculated from seed-family means and showed different effects between species monocultures and mixtures (Figure 2b), in part possibly because of 1.1 communities forced to have zero functional diversity. We also found that functional diversity had significant effects on tree productivity by both when calculated from seed-family means or from trait values of individual trees (Figure 3a, Figure 3a—supplementary 1). According to these binary results, we hypothesized that genetic diversity has the potential to affect productivity via functional diversity and the effects might differ between species monocultures and mixtures. So, we carried out multi-group structural equation models to test this hypothesis (Figure 5, Figure 5—supplementary 1). To show the effects of the zero value in 1.1 communities, we also calculated the structural equation model based on the functional diversity calculated by individual values. We found that the negative effects of genetic diversity on productivity in species monoculture were consistent in both methods (Figure 5 vs Figure 5—supplementary 1). According to these results, we concluded that although we did not find genetic diversity had significant effects on productivity in binary analysis, we found that it negatively affected productivity in species monocultures when we considered the indirect effect via functional diversity on productivity. Now we extended the discussion and adjusted the interpretation to make the results clearer and state the potential effects of zero values in 1.1 communities (see the above text pieces taken from the main text in the response to the editor’s comments).

Reviewer #2 (Recommendations for the authors):

I thank the authors for thoroughly answering all the reviewers' questions and comments. I think the present version clarifies all the points raised and shows more clarity in the design and discussion.

I am glad to read this revised version and congratulate the authors for such an impressive and interesting study that represent a step forward in our understanding of the effects of plant diversity.

Thanks for the positive comments on the revised manuscript.

https://doi.org/10.7554/eLife.78703.sa2

Article and author information

Author details

  1. Ting Tang

    1. State Key Laboratory of Vegetation and Environmental Change, Institute of Botany, Chinese Academy of Sciences, Beijing, China
    2. College of Life Sciences, University of Chinese Academy of Sciences, Beijing, China
    Contribution
    Conceptualization, Data curation, Software, Formal analysis, Visualization, Methodology, Writing - original draft, Writing - review and editing
    Contributed equally with
    Naili Zhang
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1145-0723
  2. Naili Zhang

    College of Forestry, Beijing Forestry University, Beijing, China
    Contribution
    Data curation, Methodology
    Contributed equally with
    Ting Tang
    Competing interests
    No competing interests declared
  3. Franca J Bongers

    State Key Laboratory of Vegetation and Environmental Change, Institute of Botany, Chinese Academy of Sciences, Beijing, China
    Contribution
    Visualization, Writing - review and editing
    Competing interests
    No competing interests declared
  4. Michael Staab

    Ecological Networks, Technical University Darmstadt, Darmstadt, Germany
    Contribution
    Data curation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
  5. Andreas Schuldt

    Forest Nature Conservation, Georg-August-University Göttingen, Göttingen, Germany
    Contribution
    Data curation, Investigation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
  6. Felix Fornoff

    Nature Conservation and Landscape Ecology, University of Freiburg, Freiburg, Germany
    Contribution
    Data curation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0446-7153
  7. Hong Lin

    Institute of Applied Ecology, School of Food Science, Nanjing Xiaozhuang University, Nanjing, China
    Contribution
    Data curation
    Competing interests
    No competing interests declared
  8. Jeannine Cavender-Bares

    Department of Ecology, Evolution, and Behavior, University of Minnesota, St. Paul, United States
    Contribution
    Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
  9. Andrew L Hipp

    The Morton Arboretum, Lisle, United States
    Contribution
    Visualization, Writing - review and editing
    Competing interests
    No competing interests declared
  10. Shan Li

    State Key Laboratory of Vegetation and Environmental Change, Institute of Botany, Chinese Academy of Sciences, Beijing, China
    Contribution
    Resources, Data curation, Investigation
    Competing interests
    No competing interests declared
  11. Yu Liang

    State Key Laboratory of Vegetation and Environmental Change, Institute of Botany, Chinese Academy of Sciences, Beijing, China
    Contribution
    Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4259-6028
  12. Baocai Han

    State Key Laboratory of Systematic and Evolutionary Botany, Institute of Botany, Chinese Academy of Sciences, Beijing, China
    Contribution
    Data curation
    Competing interests
    No competing interests declared
  13. Alexandra-Maria Klein

    Chair of Nature Conservation and Landscape Ecology, Faculty of Environment and Natural Resources, University of Freiburg, Freiburg, Germany
    Contribution
    Data curation, Writing - review and editing
    Competing interests
    No competing interests declared
  14. Helge Bruelheide

    1. Institute of Biology/Geobotany and Botanical Garden, Martin Luther University Halle-Wittenberg, Halle, Germany
    2. German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig, Leipzig, Germany
    Contribution
    Data curation, Writing - review and editing
    Competing interests
    No competing interests declared
  15. Walter Durka

    1. German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig, Leipzig, Germany
    2. Department of Community Ecology, Helmholtz Centre for Environmental Research–UFZ, Halle, Germany
    Contribution
    Data curation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
  16. Bernhard Schmid

    Department of Geography, University of Zurich, Zurich, Switzerland
    Contribution
    Conceptualization, Formal analysis, Methodology, Writing - review and editing
    For correspondence
    bernhard.schmid@geo.uzh.ch
    Competing interests
    Reviewing editor, eLife
  17. Keping Ma

    1. State Key Laboratory of Vegetation and Environmental Change, Institute of Botany, Chinese Academy of Sciences, Beijing, China
    2. College of Life Sciences, University of Chinese Academy of Sciences, Beijing, China
    Contribution
    Conceptualization, Resources, Supervision, Funding acquisition, Validation, Project administration, Writing - review and editing
    For correspondence
    kpma@ibcas.ac.cn
    Competing interests
    No competing interests declared
  18. Xiaojuan Liu

    State Key Laboratory of Vegetation and Environmental Change, Institute of Botany, Chinese Academy of Sciences, Beijing, China
    Contribution
    Conceptualization, Resources, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Methodology, Writing - original draft, Project administration, Writing - review and editing
    For correspondence
    liuxiaojuan06@ibcas.ac.cn
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9292-4432

Funding

National Natural Science Foundation of China (31870409)

  • Ting Tang
  • Franca J Bongers
  • Xiaojuan Liu

Strategic Priority Research Program of the Chinese Academy of Sciences (XDB31000000)

  • Naili Zhang
  • Xiaojuan Liu

National Natural Science Foundation of China (32161123003)

  • Naili Zhang
  • Yu Liang
  • Keping Ma
  • Xiaojuan Liu

Younth Innovation Promotion Association CAS (2019082)

  • Xiaojuan Liu

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We acknowledge the support of the BEF-China platform of the Zhejiang Qianjiangyuan Forest Biodiversity National Observation and Research Station. This study was financially supported by the National Natural Science Foundation of China (31870409), Strategic Priority Research Program of the Chinese Academy of Sciences (XDB31000000), and National Natural Science Foundation of China (32161123003). XL was supported by the Youth Innovation Promotion Association CAS (2019082). BS was supported by the University Research Priority Program Global Change and Biodiversity of the University of Zurich.

Senior Editor

  1. Detlef Weigel, Max Planck Institute for Biology Tübingen, Germany

Reviewing Editor

  1. David A Donoso, Escuela Politécnica Nacional, Ecuador

Version history

  1. Received: March 17, 2022
  2. Preprint posted: March 28, 2022 (view preprint)
  3. Accepted: November 14, 2022
  4. Accepted Manuscript published: November 29, 2022 (version 1)
  5. Version of Record published: December 15, 2022 (version 2)

Copyright

© 2022, Tang, Zhang et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,394
    Page views
  • 338
    Downloads
  • 4
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Ting Tang
  2. Naili Zhang
  3. Franca J Bongers
  4. Michael Staab
  5. Andreas Schuldt
  6. Felix Fornoff
  7. Hong Lin
  8. Jeannine Cavender-Bares
  9. Andrew L Hipp
  10. Shan Li
  11. Yu Liang
  12. Baocai Han
  13. Alexandra-Maria Klein
  14. Helge Bruelheide
  15. Walter Durka
  16. Bernhard Schmid
  17. Keping Ma
  18. Xiaojuan Liu
(2022)
Tree species and genetic diversity increase productivity via functional diversity and trophic feedbacks
eLife 11:e78703.
https://doi.org/10.7554/eLife.78703

Further reading

    1. Ecology
    2. Plant Biology
    Daniel Fuks, Yoel Melamed ... Ehud Weiss
    Research Article

    Global agro-biodiversity has resulted from processes of plant migration and agricultural adoption. Although critically affecting current diversity, crop diffusion from Classical antiquity to the Middle Ages is poorly researched, overshadowed by studies on that of prehistoric periods. A new archaeobotanical dataset from three Negev Highland desert sites demonstrates the first millennium CE&'s significance for long-term agricultural change in southwest Asia. This enables evaluation of the 'Islamic Green Revolution' (IGR) thesis compared to 'Roman Agricultural Diffusion' (RAD), and both versus crop diffusion during and since the Neolithic. Among the finds, some of the earliest aubergine (Solanum melongena) seeds in the Levant represent the proposed IGR. Several other identified economic plants, including two unprecedented in Levantine archaeobotany-jujube (Ziziphus jujuba/mauritiana) and white lupine (Lupinus albus)-implicate RAD as the greater force for crop migrations. Altogether the evidence supports a gradualist model for Holocene-wide crop diffusion, within which the first millennium CE contributed more to global agricultural diversity than any earlier period.

    1. Ecology
    2. Evolutionary Biology
    Songdou Zhang, Jianying Li ... Xiaoxia Liu
    Research Article

    Temperature determines the geographical distribution of organisms and affects the outbreak and damage of pests. Insects seasonal polyphenism is a successful strategy adopted by some species to adapt the changeable external environment. Cacopsylla chinensis (Yang & Li) showed two seasonal morphotypes, summer-form and winter-form, with significant differences in morphological characteristics. Low temperature is the key environmental factor to induce its transition from summer-form to winter-form. However, the detailed molecular mechanism remains unknown. Here, we firstly confirmed that low temperature of 10 °C induced the transition from summer-form to winter-form by affecting the cuticle thickness and chitin content. Subsequently, we demonstrated that CcTRPM functions as a temperature receptor to regulate this transition. In addition, miR-252 was identified to mediate the expression of CcTRPM to involve in this morphological transition. Finally, we found CcTre1 and CcCHS1, two rate-limiting enzymes of insect chitin biosyntheis, act as the critical down-stream signal of CcTRPM in mediating this behavioral transition. Taken together, our results revealed that a signal transduction cascade mediates the seasonal polyphenism in C. chinensis. These findings not only lay a solid foundation for fully clarifying the ecological adaptation mechanism of C. chinensis outbreak, but also broaden our understanding about insect polymorphism.