Volume 110, Issue 10 p. 2522-2531
RESEARCH ARTICLE
Open Access

Species richness, functional traits and climate interactively affect tree survival in a large forest biodiversity experiment

Xiaojuan Liu

Xiaojuan Liu

State Key Laboratory of Vegetation and Environmental Change, Institute of Botany, The Chinese Academy of Sciences, Beijing, China

Search for more papers by this author
Yuanyuan Huang

Yuanyuan Huang

Department of Evolutionary Biology and Environmental Studies, University of Zürich, Zürich, Switzerland

Search for more papers by this author
Lei Chen

Lei Chen

State Key Laboratory of Vegetation and Environmental Change, Institute of Botany, The Chinese Academy of Sciences, Beijing, China

Search for more papers by this author
Shan Li

Shan Li

State Key Laboratory of Vegetation and Environmental Change, Institute of Botany, The Chinese Academy of Sciences, Beijing, China

Search for more papers by this author
Franca J. Bongers

Franca J. Bongers

State Key Laboratory of Vegetation and Environmental Change, Institute of Botany, The Chinese Academy of Sciences, Beijing, China

Search for more papers by this author
Nadia Castro-Izaguirre

Nadia Castro-Izaguirre

Department of Evolutionary Biology and Environmental Studies, University of Zürich, Zürich, Switzerland

Search for more papers by this author
Yu Liang

Yu Liang

State Key Laboratory of Vegetation and Environmental Change, Institute of Botany, The Chinese Academy of Sciences, Beijing, China

Search for more papers by this author
Bo Yang

Bo Yang

Jiangxi Key Laboratory of Plant Resources and Biodiversity, Jingdezhen University, Jingdezhen, China

Search for more papers by this author
Yuxin Chen

Yuxin Chen

Key Laboratory of the Coastal and Wetland Ecosystems (Ministry of Education), College of the Environment & Ecology, Xiamen University, Xiamen, China

Search for more papers by this author
Florian Schnabel

Florian Schnabel

German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig, Leipzig, Germany

Systematic Botany and Functional Biodiversity, Leipzig University, Leipzig, Germany

Search for more papers by this author
Ting Tang

Ting Tang

State Key Laboratory of Vegetation and Environmental Change, Institute of Botany, The Chinese Academy of Sciences, Beijing, China

College of Life Sciences, University of Chinese Academy of Sciences, Beijing, China

Search for more papers by this author
Yujie Xue

Yujie Xue

State Key Laboratory of Vegetation and Environmental Change, Institute of Botany, The Chinese Academy of Sciences, Beijing, China

School of Ecology and Nature Conservation, Beijing Forestry University, Beijing, China

Search for more papers by this author
Stefan Trogisch

Stefan Trogisch

German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig, Leipzig, Germany

Institute of Biology/ Geobotany and Botanical Garden, Martin Luther University Halle-Wittenberg, Halle (Saale), Germany

Search for more papers by this author
Michael Staab

Michael Staab

Ecological Networks, Technical University Darmstadt, Darmstadt, Germany

Search for more papers by this author
Helge Bruelheide

Helge Bruelheide

German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig, Leipzig, Germany

Institute of Biology/ Geobotany and Botanical Garden, Martin Luther University Halle-Wittenberg, Halle (Saale), Germany

Search for more papers by this author
Bernhard Schmid

Corresponding Author

Bernhard Schmid

Department of Geography, Remote Sensing Laboratories, University of Zürich, Zürich, Switzerland

Correspondence

Keping Ma

Email: [email protected]

Bernhard Schmid

Email: [email protected]

Search for more papers by this author
Keping Ma

Corresponding Author

Keping Ma

State Key Laboratory of Vegetation and Environmental Change, Institute of Botany, The Chinese Academy of Sciences, Beijing, China

College of Life Sciences, University of Chinese Academy of Sciences, Beijing, China

Correspondence

Keping Ma

Email: [email protected]

Bernhard Schmid

Email: [email protected]

Search for more papers by this author
First published: 21 July 2022
Citations: 3
Handling Editor Frank S Gilliam

Abstract

  1. Tree survival affects forest biodiversity, structure and functioning. However, little is known about feedback effects of biodiversity on survival and its dependence on functional traits and interannual climatic variability.
  2. With an individual-based dataset from a large subtropical forest biodiversity experiment, we evaluated how species richness, functional traits and time-dependent covariates affected annual tree survival rates from age 3–12 (years) after planting 39 species across a diversity gradient from 1 to 2, 4, 8 and 16 tree species.
  3. We found that overall survival rates marginally increased with diversity at the plot level, with large variation among plots within diversity levels. Significant variation among species in survival responses to diversity and changes in these responses with age were related to species functional traits and climatic conditions. Generally, survival rates of conservative species (evergreen, late-successional species with thick leaves and high carbon to nitrogen ratio but low specific leaf area, leaf phosphorus and hydraulic conductivity) increased with diversity, age and yearly precipitation, whereas acquisitive species showed opposite responses.
  4. Synthesis. Our results indicate that interactions between diversity, species functional traits and yearly climatic conditions can balance survival among species in diverse forests. Planting mixtures of species that differ in functional traits in afforestation projects may lead to a positive feedback loop where biodiversity maintains biodiversity, together with its previously reported beneficial effects on ecosystem functioning.

1 INTRODUCTION

Stand structure and species composition of forests are shaped by demographic rates of individual trees (Clark, 2010; Hart et al., 2016; Iida et al., 2014). Among these, survival provides the most important link between ecosystem functions and community assembly (Leibold et al., 2017). The often-observed positive biodiversity–ecosystem functioning (BEF) relationship with regard to productivity can be due to increased growth or increased survival of individuals in more diverse plant communities (Barrufol et al., 2013; Marquard et al., 2009; Tilman et al., 2014). However, while a large number of BEF studies have explored biodiversity effects on individual growth, very few have studied such effects on individual survival (but see Kirui et al., 2008; Van de Peer et al., 2016; Yang et al., 2017), potentially due to the short-term nature of most of these studies.

Biodiversity is expected to promote plant survival and thus species coexistence (Grossman et al., 2018; Isbell et al., 2011; Neuner et al., 2015) due to three mechanisms. In diverse stands under environmental fluctuations, a ‘buffering effect’ could result from asynchronous responses of different species to these fluctuations, so that rarely all individuals are simultaneously exposed to the same mortality risk. A ‘performance-enhancing effect’ could result when species with high performance under particular environmental conditions at each time have greater survival, so that the mean community performance increases (Yachi & Loreau, 1999). In the absence of environmental fluctuations, trees in mixtures may still survive better due to lower inter- than intra-specific competition, which is expected when trees have complementary niches, also referred to as ‘complementarity effect’ (Tilman et al., 2014). However, so far these three mechanisms have mostly been studied for productivity (Ammer, 2019), especially in grassland experiments. Survival and its relation to buffering, performance-enhancing and complementarity effects have far less been studied (but see Luo & Chen, 2011).

Differential survival responses among species to biodiversity over time might be interactively affected by functional traits and interannual variation in climatic conditions. Survival rates under particular environmental conditions may be predictable to a certain extent based on functional traits regulating the acquisition of light, water or nutrient resources (Van de Peer et al., 2016). Previous studies have reported that species adapted to less favourable environmental conditions (e.g. low nutrient or water availability) generally have enhanced resource-holding capacity (e.g. higher specific leaf area, higher stomatal density, higher hydraulic conductivity) and lower biomass loss than species adapted to better environmental conditions (Laughlin et al., 2018; Maire et al., 2015; O'Brien et al., 2014; O'Brien et al., 2017; Pu et al., 2020; Sterck et al., 2003; Sterck et al., 2006). In addition, functional traits are often used to place species along an acquisitive to conservative strategy gradient, ranging from deciduous, early-successional species with thin leaves, high specific leaf area (SLA) and high hydraulic conductivity to evergreen, late-successional species with thick leaves, low SLA, low hydraulic conductivity and high leaf nutrient concentration (Anderegg et al., 2016; Fichtner et al., 2017; Schnabel et al., 2021; Wright et al., 2004). As a consequence, if the functional traits of a species do not match the corresponding environmental conditions, its survival rate may be reduced. However, because environmental conditions, in particular climatic conditions, can vary between years, it is necessary to study effects of functional traits and environmental conditions on individual survival in a time-dependent manner (Adler et al., 2006; Ammer, 2019). This requires the study of survival rates obtained from interval-censored data (Sparling et al., 2006), because it is more difficult to accommodate time-dependent covariates such as yearly precipitation in studies of survival times (Egli & Schmid, 2001).

Here, we used a long-term, interval-censored dataset from a large forest BEF experiment in subtropical China (BEF-China) to analyse annual survival rates of individuals trees from age 3 to 12 (years) among 39 subtropical tree species with contrasting functional traits as a function of stand diversity, age and yearly precipitation (Figure 1a). Stand diversity was represented by tree species richness in the experiment and varied from monocultures to 2-, 4-, 8- and 16-species mixtures. Each diversity level was represented by different species compositions, with each species occurring at each richness level. There were 469 plots (25.8 × 25.8 m), each planted with 400 trees of which the central 16 were yearly censused (Figure S1). We tested if stand diversity, that is species richness, could promote tree survival and how survival rates were modified by species functional traits and yearly climatic conditions. Our specific hypotheses were (Figure 1b): H1, stand diversity increases annual survival rate and this effect strengthens with age. The second part of H1 was based on our previous observation that positive diversity effects on stand biomass and productivity increased with stand age due to species complementarity (Huang et al., 2018). H2, different species show differential survival responses to stand diversity and these responses change with age. H3, differential survival responses of species with age may depend on functional traits and annual climatic conditions such as precipitation.

Details are in the caption following the image
Conceptual figure. Illustration of (a) how the survival rates were estimated at annual intervals with an example of an 8-species mixture plot. Different tree shapes represent the eight species A to H. different colours represent functional traits. In each plot, the central 16 trees were censused in each year. We use A1, C1 and H1 as example trees from species A, C and H to show how annual survival for each individual tree was censused. The number of survivors among the central 16 trees is shown at the end of each arrow. The individual binomial survival data were used in generalized linear mixed models (GLMM) to estimate survival rates, which can be calculated for particular groups of individuals such as those of a single species or per age as the number of survivors at the end of an interval divided by the numbers alive at the beginning of the interval (often expressed in percent and then called survival probability instead of survival rate); (b) central hypotheses explaining complex diversity–survival rate relationships and how they are modified by stand age, functional traits of species and annual climatic conditions. H: Overall, stand diversity may promote tree survival rate. However, H1: The diversity effects may only develop over time, becoming more positive as stands mature. H2: Species may respond differentially to stand diversity and these responses may change with age (A, C and H are three example species). H3: These different responses of species with age may depend on functional trait differences among species (Trait × Age × Diversity). Furthermore, the changing responses may also be affected by annual climatic conditions (Trait × Climate × Diversity).

2 MATERIALS AND METHODS

2.1 Study site

Our study was conducted in the Biodiversity–Ecosystem Functioning Experiment China (BEF-China, www.bef-china.com) located at Xingangshan, Dexing, Jiangxi, China (29°08′–29°11′N, 117°90′–117°93′E; Bruelheide et al., 2014). This large forest BEF experiment was set up in 2009/2010 in an area previously used for Cunninghamia lanceolata plantations, which were harvested about every 20 years and then allowed to regrow vegetatively. After removal of these trees, we established 492 plots with 400 tree seedlings each at two sites 5 km apart from each other (Huang et al., 2018). Here, we focus on 469 plots (Figure S1) planted with evergreen and deciduous broad-leaved tree species (Table S1), excluding 23 plots with only shrub monocultures and plots with unsuccessful tree establishment. The local climate at the site is characterized as a seasonal monsoon. Over the study period the mean annual temperature was 16.7°C, ranging from 17.8°C to 19.3°C, while the mean annual precipitation was 1821 mm, ranging from 1399 mm to 2766 mm.

2.2 Target species and individual trees

The 469 plots of 25.8 × 25.8 m (1 mu in Chinese units, Figure S1) analysed here contained 1, 2, 4, 8 or 16 tree species. These came from a pool of 40 native tree species. Species in plots at lower diversities were subsets of plots at higher diversity. Detailed descriptions of the experimental design can be found in Bruelheide et al. (2014) and Huang et al. (2018). In this study, we analysed annual survival rates of individual trees from age 3 after planting, when the seedling were well established, to age 12 after planting for 39 of the 40 species (see Table S1). One species, Castanopsis carlesii, had to be excluded due to unsuccessful planting with very few survivors. In each of the 469 plots, the survival status (alive, dead or missing) of all trees from the 16 central planting positions was recorded yearly in October. In total our dataset contained 48,067 data points from 7046 trees, covering the years 2011–2020 for site A and 2012–2021 for site B (site B was established 1 year later that site A; Table S2).

2.3 Quantification of survival rates

Tree survival was directly estimated from the binary individual data (1 for survival, 0 for death). Only trees still alive at the beginning of an interval were used to calculate annual survival. This corresponds to a conditional probability of survival (number of trees alive at the end of an interval divided by number of individuals alive at the beginning of an interval; Egli & Schmid, 2001). For example, if an individual tree in 2009, 2010 and 2011 was alive, but from 2012 on was dead, then the survival from 2009 to 2019 was 1, 1, 1, 0, NA, NA, … NA, where NA indicates a ‘missing value’ not included in subsequent analysis. Furthermore, individual tree survival rates were aggregated to survival rates of groups of trees such as those coming from the same species, age and diversity, depending on the hypotheses we were testing (Figure 1a).

2.4 Functional traits

Nine species-level functional traits measured in our experiment were used in our study (Table S3). These traits are related to species succession (leaf habit and successional stage), leaf structure (leaf thickness), leaf economic spectrum [specific leaf area (SLA), leaf carbon to nitrogen ratio (Leaf CN), leaf phosphorus concentration (Leaf P)], leaf stomatal conductance and control of conductance (stomatal density and stomatal area) and stem hydraulics [maximum xylem hydraulic conductivity (Max hydraulic conductivity)]. The detailed description and biological meaning of each trait can be found in Table S3 and our previous related studies (Bongers et al., 2021; Kröber et al., 2014; Kröber, Heklau, & Bruelheide, 2015; Ma et al., 2021; Schnabel et al., 2021).

We further summarized the functional trait variation among species with a principal component analysis (PCA) of the nine traits. The first and second PCA axes explained 64% of the total trait variation among species. After varimax rotation we obtained two variables associated with trait groups that could be interpreted as plant strategies. Specifically, we consider PC1 to represent the leaf economic spectrum or a strategy gradient from ‘acquisitive’ to ‘conservative’ species (also related to early- vs. late-successional species). Acquisitive (early-successional) species were characterized by high SLA, leaf P and maximum hydraulic conductivity, while conservative (late-successional) species were characterized as evergreen and late-successional, with thick leaves and high leaf CN. We consider PC2 to represent a stomatal control gradient or drought-tolerance strategy gradient from ‘water savers’, who close their stomata fast under dry conditions, to ‘water spenders’, who do not close their stomata fast and continue to transpire water (McDowell et al., 2008). Water savers are characterized by high stomatal density and large stomatal area (Figure 4a) and these traits are associated with a fast down-regulation of stomatal conductance under increasing vapour pressure deficits in our study system (see our previous study Schnabel et al., 2021 for a detailed description). The PCA was calculated with the ‘rda’ function in the vegan package version 2.5-6 of R (Oksanen et al., 2019).

2.5 Climatic conditions

The annual precipitation (1399–2766 mm during the studied years) was included as climatic condition in the study (Table S2). The data were provided by the climate bureau of Dexing, where the experimental sites are located. Annual temperature was quite constant during the observation time (Table S2) and thus not used in the subsequent analyses.

2.6 Statistical analysis

All analyses were aiming to determine the influence of diversity, that is, tree species richness as the main design factor in the BEF-China experiment, on tree survival rates. To test H1, we first looked at the main effect of diversity on annual survival rate at plot level regardless of species and how it changed with age. We used a generalized linear mixed model (GLMM) with complementary log–log link function and binomial error distribution (Egli & Schmid, 2001) to predict survival rates as a function of tree species richness (log2-transformed, logSR) and age (log-transformed age, logage, corresponding to Weibull survivorship curves). Fixed terms were fitted sequentially (type-I ANOVA). The random terms in the model were site, plot and the interaction plot × logage.

To test H2, we added species identity and its interaction with species richness as explanatory terms to the statistical model. Thus this GLMM model had the fixed effects species identity (species), age (logage), species richness (logSR) and their 2- and 3-way interactions; the random terms were again site, plot and plot × logage (see Table S5). Furthermore, to look at the relationship between diversity and annual survival rates for each species separately, we used a third GLMM model were we replaced the continuous variable logage with a factor AGE with 10 levels. Four ages were selected to show the diversity effects on annual survival rates among species (used in Figure 3).

Finally, to test H3, we replaced species with their functional traits and added annual precipitation as an alternative to age to the analysis. The corresponding GLMM models had the fixed effects functional trait (PC1, PC2 or each of the 9 specific traits), precipitation (or logage), species richness (logSR) and their 2-way interactions. Random terms were site, plot, species, species × precipitation (or species × logage), plot × precipitation (or plot × logage) and plot × trait (see Table S6). To account for any confounding relationship between precipitation and logage we repeated the analyses with covariate adjustment, using logage as covariate in the analyses of precipitation effects and precipitation as covariate in the analyses of logage (see Table S7).

All analyses were done in R 3.6.2 and ASReml-R 4.1.0 (Butler, 2019).

3 RESULTS

3.1 Survival tends to decrease with age and increase with diversity (H1)

Overall annual survival rate declined from age 3 to 12 but was marginally higher in more diverse than less diverse stands (Table S4; Figure 2; also see the cumulative survival from Figure S2). The interaction between the two was not significant (p = 0.143). These overall effects were tested against the variation in survival rates among plots with different species compositions and their changes with age as random-effects terms, which themselves were quite large (variance component for plots 0.103 ± 0.009 and for plot × logage interaction 0.204 ± 0.022).

Details are in the caption following the image
Diversity effects on annual survival rate from age 3–12 (years) after the experimental forest stands were established. The grey line is the average survival rate across all ages. The x-axis was log2-transformed.

3.2 Different species show differential survival responses to diversity and with age (H2)

The main effects of species identity and age (including their interaction) were significant, as was the 3-way interaction of the two with species richness (Table S5). This 3-way interaction demonstrated that species within communities showed different survival responses to richness and that these response trends diverged over time. We further found that although the number of species with a positive and negative relationship was different among ages, eight among the 10 ages had more positive than negative relationships (Figure 3; Figure S3).

Details are in the caption following the image
Annual diversity–survival rate relationships for each of the 39 species of the experiment for four distinct ages. The colour of the lines indicates different species. The four ages represent the early age after establishment (Age3), intermediate ages (Age6 and Age9) and the latest age included in the analysis (Age12). The observed and predicted survival rates for the other six ages are show in Figure S4. Predicted survival rates are back-transformed from the generalized linear mixed model described in the text with significance values given in Table S5

3.3 Differential survival responses of species to diversity depend on functional traits and vary with age and annual precipitation (H3)

Overall variation among species in annual survival rate could not well be explained by differences among species in functional traits. Only SLA and Leaf P had significant overall negative effects on survival rates (Table S6a). There were many interactions between functional traits and age, reflecting in particular that relative to acquisitive species, conservative species had increased survival with age (Table S6a). Furthermore, the many interactions between functional traits and diversity indicated that the differential diversity responses of species described in the previous section were due to their differences in functional traits (see following paragraph; Figure 4b; Figure S4a).

Details are in the caption following the image
(a) Principal component analysis (PCA) biplot for the nine functional traits defining two gradients of plant strategies; and significant functional trait (PCs)-dependent relationships between (b) species richness and (c) precipitation and annual survival rates. PC1 is a gradient that can be considered to represent the leaf economic spectrum or a strategy gradient from ‘acquisitive’ to ‘conservative’ species (also related to early vs. late successional species). Acquisitive (or early-successional) species are characterized by high leaf specific area (SLA), leaf phosphorus (Leaf_P) and maximum xylem hydraulic conductivity (Max_hydraulic_conductivity), while conservative (or late-successional) species are characterized as evergreen (Leaf_habit) and late successional (Successional stage), with thick leaves (Leaf_thickness) and high leaf C:N ratio (Leaf_CN). PC2 is a gradient that can be considered to represent stomatal control or a drought-tolerance strategy gradient from ‘water savers’ to ‘water spenders’. Water savers are characterized by high stomatal density (Stomatal_density) and stomatal area per leaf area (Stomatal_area). All traits are species-level mean values measured on site. For species full names see Table S1. Deciduous and evergreen species are coloured as red dots and blue triangles, respectively. The first and second PCA axis explained 64% of the total variation among species in the nine traits. In (b and c), fitted lines are shown for PC scores covering the range from the species with lowest to highest score. Predicted annual survival rates were back-transformed from the generalized linear mixed models described in the text with significance values given in Table S6b.

Using precipitation instead of age in the analysis showed that there was no overall main effect of annual precipitation on survival, but several interactions between precipitation and functional traits (Table S6b). The annual survival rates of evergreen, late successional and conservative species increased, while that of deciduous, early successional and acquisitive species decreased with increasing precipitation (Figure 4c; Figure S4b). PC2 and six traits had significant interactions with diversity, which confirmed the trait-dependent diversity–survival rate relationships (Table S6b; Figure 4b; Figure S4a). Generally, diversity increased the survival rates of evergreen, late successional and conservative species or water spenders, but not of deciduous, early successional and acquisitive species or water savers (Figure 4b).

Adjusting the above analyses for covariates did not change the pattern of results, indicating that these were not affected by a potential confounding between age and precipitation (Table S7).

4 DISCUSSION

Using an individual-based dataset from a large subtropical forest biodiversity experiment, we evaluated how species richness, functional traits and time-dependent covariates affected annual tree survival rates 3–12 years after planting in 39 species across a diversity gradient from 1, 2, 4, 8 to 16 tree species. We found that annual survival rates were influenced by interactions between diversity, age, species identity, functional traits and time-dependent covariates, in particular annual precipitation. These interactions likely contributed to species coexistence within stands as predicted under the assumption of buffering and performance-enhancing effects in diverse stands under environmental fluctuations (Yachi & Loreau, 1999).

Strong variation within diversity levels among plots and among plots over time were indicative of variation among different species compositions in plot-level survival rates. This large variation within diversity levels reduced the positive main effect of diversity on survival rate to marginal significance and caused its interaction with age to remain insignificant. We therefore could not support our first hypothesis (H1) of positive diversity effects on annual survival increasing with age. In a previous study in established forest plots near our experimental site, positive effects of tree species richness on tree density have been found from about 20–100 years for similar species compositions as used here (Barrufol et al., 2013). Perhaps our time series from age 3 to 12 years was too short for significant main effects of diversity to develop. During early stages of forest development competitive hierarchies between species tend to be more important than niche differences (Luo et al., 2020). Positive diversity effects on tree survival in the early years of stand development were also absent in forest experiments set up in temperate (Van de Peer et al., 2016) and tropical regions (Healy et al., 2008; Potvin & Gotelli, 2008; Tuck et al., 2016). Over time, different species in mixtures can establish different resource-uptake strategies, thus potentially reducing inter-specific relative to intra-specific competition and competition-induced mortality (Adams et al., 2017; Anderegg et al., 2016; Anderegg et al., 2018; McDowell et al., 2018). Indeed, we found in previous studies that, at least during the first 10 years of our experiment, complementarity effects among species increased with stand development (Bongers et al., 2021; Huang et al., 2018), which would be consistent with reduced inter-specific competition.

In contrast to the marginal positive effect of diversity on plot-level annual survival rates, individual tree species significantly varied in their survival responses to diversity and age, supporting our second hypothesis (H2). Among the 39 tested species more than half, on average across age, responded positively to diversity and less than half responded negatively (see Figure S3). These differential survival responses among species to stand diversity may have been related to different relations between inter- and intra-specific competitive abilities. Dominant species that rank high in competitive hierarchy generally benefit in mixtures whereas subdominant species that rank low in competitive hierarchy may actually benefit in monoculture (Stoll & Prati, 2001; Vogt et al., 2010).

Our third hypothesis (H3) thus was that differential and asynchronous survival responses among species to diversity were due to differences in functional traits among species (Bongers et al., 2020; Kröber, Li, et al., 2015) and environmental variation in time (O'Brien et al., 2017). This was confirmed in the corresponding analysis with species identities replaced by functional traits and by using annual precipitation as time-dependent explanatory variable (see Figure 4; Figure S4). Overall, species with traits characteristic of conservative (late-successional) species increased their annual survival rate relative to acquisitive (early-successional) species with age. Furthermore, survival responses to diversity were positive for conservative species (in particular with low SLA, low hydraulic conductivity and high Leaf CN). In contrast, species with high SLA, high hydraulic conductivity and low Leaf CN, usually represented by deciduous and early successional species, over time may be increasingly suppressed by inter-specific competition in more diverse mixtures (McDowell et al., 2018; Russo et al., 2010; Tilman, 1988), explaining their neutral or even negative survival responses to diversity. Finally, survival responses to diversity were positive for species with a water-spender strategy and negative for species with a water-saver strategy, potentially as the former profited from the faster stomatal closure and thus the lower water consumption of the latter in diverse stands (Forrester, 2017). In contrast, water savers may have experienced higher interspecific than intraspecific competition for water.

In addition to the different survival responses to diversity for species with contrasting strategies, they also showed different survival responses to variation in annual precipitation. In dry years, the survival rates of conservative species with thick leaves, low Leaf P and low hydraulic conductivity were lower than in wet years. This seems counter-intuitive but first the dry years were not extremely dry. And second herbivory and other diseases may have reduced a competitive disadvantage for these species relative to others in wet years. In contrast, the higher survival of acquisitive species in dry years may partly be due to generally lower herbivory in dry years. These trait-driven, asynchronous survival responses may be one mechanism contributing to the positive effect of diversity on community stability, which we reported in a previous study (Schnabel et al., 2021).

Survival responses to precipitation might interact with survival responses to diversity, yet three-way interactions were not significant for most trait-based analyses. The increased precipitation could have accelerated the rates of soil erosion and nutrient leaching, especially in monocultures and in low-diversity mixtures more than in high-diversity mixtures (Aerts & Chapin, 2000; Santiago et al., 2004), potentially reducing soil nutrient availability in wet years. Poor nutrient availability should affect the tree survival rates of acquisitive species more so than that of conservative species.

5 CONCLUSIONS

The manifold interactions between the effects of stand diversity and age, functional traits and identity of species and time-dependent covariates on annual survival rates indicate that generalizations for this demographic variable are perhaps more difficult to be obtained than for ecosystem functions such as biomass production. Thus, even though the species traits affecting survival are suggestive for successional (Devaney et al., 2020; Tilman, 1988) or plant-economic strategies (Wright et al., 2004), none of them seems to present a consistent advantage across environmental fluctuations between years. While the differential responses to temporal climatic fluctuations between years increase coexistence and community stability due to buffering and performance-enhancing effects of biodiversity (Schnabel et al., 2021; Yachi & Loreau, 1999), reduced inter-specific relative to intra-specific competition in diverse stands may increase coexistence due to complementarity effects within years (Tilman et al., 2014). The marginally positive effect of diversity on plot-level survival is consistent with this assumption. Considering all these results together, we recommend that in reforestation projects foresters should plant mixtures including species of contrasting plant strategies, such as acquisitive and conservative species or water-savers and water-spenders. Likely, this will not only benefit ecosystem functioning but at the same time help secure the maintenance of species richness itself.

AUTHOR CONTRIBUTIONS

Xiaojuan Liu, Keping Ma and Bernhard Schmid conceived the study; Yuanyuan Huang, Nadia Castro-Izaguirre, Shan Li, Bo Yang, Ting Tang, Yujie Xue and Helge Bruelheide collected the data; Xiaojuan Liu and Bernhard Schmid performed data analyses with the contribution from Lei Chen, Yuanyuan Huang and Yuxin Chen; Franca J. Bongers, Yu Liang, Florian Schnabel, Stefan Trogisch, Michael Staab, Helge Bruelheide and Keping Ma joined the discussion of analyses and results; Xiaojuan Liu and Bernhard Schmid wrote the paper. All authors contributed to the final preparation of the manuscript.

ACKNOWLEDGEMENTS

We acknowledge the support of the BEF-China platform. This study was financially supported by the Strategic Priority Research Program of the Chinese Academy of Sciences (XDB31000000), the National Natural Science Foundation of China (31870409) and CAS Interdisciplinary Innovation Team (JCTD-2018-06). X.L. was supported by the Youth Innovation Promotion Association CAS (2019082). B.S. was supported by the University of Zurich Research Priority Program on Global Change and Biodiversity (URPP GCB). F.S. was supported by the International Research Training Group TreeDì funded by the Deutsche Forschungsgemeinschaft (DFG, 319936945/GRK2324). Open access funding provided by Universitat Zurich.

    CONFLICT OF INTEREST

    The authors have no conflict of interest to declare.

    PEER REVIEW

    The peer review history for this article is available at https://publons.com/publon/10.1111/1365-2745.13970.

    DATA AVAILABILITY STATEMENT

    Data are available on the BEF-China data portal: https://data.botanik.uni-halle.de/bef-china/datasets/655.