Habitat heterogeneity and biotic interactions mediate climate influences on seedling survival in a temperate forest

1.   Introduction

The mechanisms that accelerate species coexistence are a hot topic in the studies of plant community, yet they still remain a major challenge in community ecology (Chesson, 2000; Kobe and Vriesendorp, 2011). Species coexistence models developed for diverse communities usually emphasize conspecific negative density dependence (CNDD) and resource niche partitioning (Harms et al., 2000; Wright, 2002; Zhu et al., 2018). Because of resource competition and species-specific natural enemy attack, natural enemies can decrease seedling recruitment near conspecific trees and/or at high local conspecific seedling densities when surrounded by a high density of conspecific individuals under CNDD (Janzen, 1970; Connell, 1971). However, niche partitioning might be used by species for resource distribution and to prevent competitive exclusion (Silvertown, 2004; Zhu et al., 2018).

Growing evidence supports the idea that CNDD is widespread in forest community, and that conspecific neighbor effects (i.e., per neighbor CNDD) vary greatly among plant species (Comita et al., 2010; Kobe and Vriesendorp, 2011; Bai et al., 2012; Johnson et al., 2012; Liu et al., 2022). In the tropical forests of central Panama, it has been revealed that CNDD is the strongest at seedling stage, and that CNDD can vary among species according to their life history (Wright et al., 2010; Zhu et al., 2015). Prior studies had speculated that niche partitioning would enable each species to share available resource so as to prevent competitive exclusion (Johnson et al., 2017), yet it is doubtful to be the primary driver of vegetation structure diversity (Silvertown, 2004; Guo et al., 2020; Xu et al., 2022).

In comparison to CNDD, niche partitioning caused by habitat heterogeneity typically has a lesser importance in terms of abiotic factors such as light availability, nutrient resource availability, soil water content, and topographic circumstances (Kobe and Vriesendorp, 2011; Johnson et al., 2014). Researches on seedling survival within tree communities have demonstrated the presence of CNDD influenced by the local biotic neighborhood (Queenborough et al., 2009; Comita et al., 2010). In addition, other investigations have explored the correlation between growth and survival and the surrounding abiotic environment (Bai et al., 2012). Nonetheless, there has been a limited exploration of the concurrent impacts of both biotic and abiotic factors on seedling survival within a unified analysis (Comita and Hubbell, 2009; Pappas et al., 2020). Moreover, many biotic interactions are sensitive to environmental factors, and this sensitivity might have unpredictable effects on altering the structure of forest communities through influencing seedling mortality (Tylianakis et al., 2010; Chen et al., 2019). In actuality, biotic interactions might underestimate the direct influences in shaping how a particular species responds to climate variability (Alexander et al., 2015; Chu et al., 2016).

In forest ecosystems, interannual climate variation has the potential to be a major axis in niche partitioning and can significantly affect community dynamics (Clark et al., 2016). Up to now, however, little is known about how interannual temperature variations influence these ecosystems’ dynamics, particularly in the case of temperate deciduous forests. Recent studies have shown that interannual variations in climate variables, such as precipitation, temperature, and solar radiation, strongly constrain tree growth and productivity in the tropical forests, and high vapor pressure deficit (VPD) resulting from temperature and moisture conditions seems to be more critical than other climate factors for intra-annual tree growth during the growing season (Song et al., 2018; Uriarte et al., 2018; Li et al., 2022; Xu et al., 2022). Additionally, the responses of plant species to climatic variability vary significantly, and it seems that early successional species are more susceptible to drought stress than late successional species (Chazdon et al., 2005; Uriarte et al., 2016, 2018). Precipitation and air temperature have been well studied on seedling survival, however, soil temperature and VPD have been less explored as the major factors that shape regeneration dynamics of forest ecosystems (Chandra et al., 2022; Li et al., 2022).

The seedling stage over a plant's life cycle is the most susceptible and vulnerable, as seedlings have fewer stored resources, which makes them particularly sensitive to conditional stress (such as dry and low-light conditions) (Green et al., 2014; Clark et al., 2016). In addition, species distribution patterns and community succession are shaped by seedling establishment and survival (Johnson et al., 2017; Kuang et al., 2017). Consequently, disentangling the factors that influence tree seedling survival is essential to understanding, predicting, conserving and managing forests more effectively (Kuang et al., 2017). Recent researches have shown that numerous biotic interactions are very sensitive to environmental factors, with unpredictable impacts on tree seedling survival (Alexander et al., 2015; Chu et al., 2016; Suttle et al., 2007). In fact, the impact of abiotic factors such as climate variability on tree seedling survival is further complicated by the degree to which this heterogeneity affects biotic interactions (Blois et al., 2013; Uriarte et al., 2018).

Although seedling mortality is one of the most investigated processes of forest ecosystems, we still know very little about the mechanisms that prevent seedling mortality in the species-rich temperate forests (Liu et al., 2022). The fact that the process may alter during the course of an organism's life history is one possible explanation for the diversity of hypotheses on the underlying species coexistence mechanisms (Bai et al., 2012; Zhu et al., 2018). Changes in abiotic and biotic factors have different effects on species performance, and life history strategies may also have an impact along environmental niche axes on growth, distribution, and resource uptake and utilization (Comita et al., 2014; Guo et al., 2020).

Furthermore, survival rates between species can differ significantly depending on their traits (McCarthy-Neumann and Kobe, 2008; Kobe and Vriesendorp, 2011; Xu et al., 2022). Usually, local biotic neighborhoods can increase the survival of rare species compared to common species in a community, while the neighborhoods of common species will be more likely to contain a conspecific individual, and common species may experience stronger negative density dependence effects (Connell, 1971; Bai et al., 2012). Moreover, species traits (such as trees and shrubs) can determine their ability to protect themselves from abiotic stress and natural enemies (Squinzani et al., 2022; Xu et al., 2022). Recent researches found significant negative density-dependence effects on survival of tree seedlings, and limited effects of habitat heterogeneity on survival of shrub seedlings (Bai et al., 2012). Attempts to ascertain species attributes that are responsible for CNDD variation have yielded inconsistent results (Kobe and Vriesendorp, 2011; Zhu et al., 2015). As a result, relevant stage-species functional features should be included to study the differences in the importance of abiotic and biotic variables throughout plant life cycle. This is a vital step in understanding how seedling survival responds to climate variability in forest communities (Fisher et al., 2010; Uriarte et al., 2018).

Up to now, however, it still remains unclear how important the direct and indirect pathways are in influencing seedling dynamics in changing environments. Our main goal of this study is to determine how seedling survival is affected by biotic interactions (i.e., CNDD) and abiotic factors (i.e., interannual climate variability and habitat conditions) within communities and within different species groups. In order to quantify these issues, we examined the relative importance of biotic interactions (i.e., CNDD), interannual climate variation and habitat heterogeneity for seedling survival over a 5-year period in a temperate forest in north-central China. Particularly, we asked the following questions.

(1) How is seedling survival in the temperate forest affected by interannual climate variation, biotic interactions, and habitat conditions? We hypothesized that interannual climate variability would have stronger effects than biotic interactions (i.e., CNDD) and habitat conditions.

(2) Do differences in age classes of different tree species have an impact on how biotic interactions, interannual climate variability, and habitat factors affect the survival of seedlings? We expected that seedling survival would strongly correlate with interannual climate variability at all life stages, and the strength of the correlation increased with seedling age.

(3) How does the strength of CNDD vary with climate and habitat conditions? We hypothesized that seedlings would experience stronger CNDD under higher maximum air temperature and vapor pressure deficits in growing season since low soil moisture controls the activity of soil fungal pathogens (Swinfield et al., 2012).

(4) Whether various species groups exhibit the same behavior in the presence of biotic interactions, interannual climate variation and habitat conditions? We expected common species and shrub species should have experienced fewer conspecific neighbor effects than rare species and tree species in a community, probably due to soil pathogens' strong influence on the rare species and tree species (Marden et al., 2017; Zhu et al., 2018).

2.   Materials and methods

2.1   Study site

This study was conducted in the Foping National Nature Reserve (33°33′–33°46′ N, 107°41′–107°55′ E), which is located in the south-facing slope of the Qinling Mountains in the north-central China. The Qinling Mountains serve as a significant geographic demarcation line and the most vital border for climate and vegetation distribution in mainland China due to its east to west layout (Shi et al., 2019). Under the subtropical monsoon climate, the canopy primarily consists of deciduous broad-leaved Quercus species (980–2,300 ​m), mixed deciduous broad-leaved forest and coniferous forest (2,300–2,500 ​m), and coniferous forest (2,500–2,900 ​m). The growing season extends from March to October. The mean annual precipitation is around 1,100 ​mm, most of which falls between July and September. The average annual air temperature is about 13 ​℃ (27 ​℃ in July and −2 ​℃ in January) (Shi et al., 2019).

A permanent forest dynamic plot with a size of 25 ​ha (500 ​m ​× ​500 ​m) was set up in the deciduous broad-leaved forest in the Foping National Nature Reserve in 2014 in accordance with the measurement protocols from the Forest Global Earth Observatory (ForestGEO, http://www.forestgeo.si.edu). The 25-ha plot was divided into 625 subplots of 20 ​m ​× ​20 ​m. Each subplot was then sub-divided into 16 quadrats of 5 ​m ​× ​5 ​m. Within each quadrat, all woody individuals (including tree and shrub species) with DBH (diameter at breast height) ≥1 ​cm were completely mapped, recorded, and identified. The elevation of the 25-ha forest plot varies from 1,715 to 1,836 ​m above sea level (Fig. S1).

2.2   Seedling census

To monitor seed rainfall and litterfall, 135 seed traps were established in the 25-ha plot in 2015 (Fig. S2). 3 seedling plots (1 ​m ​× ​1 ​m) were set up 2 ​m away from three sides (west, north, east) of each seed trap (Fig. S3). The first seedling census was finished in September 2015. Woody plants with DBH <1 ​cm were treated as seedlings within each seedling plot. Seedlings in these plots had been fully mapped, identified to species, and their stem height was measured and the number of their leaves was also counted. The age of each seedling was estimated by counting annual bud scale scars (Bai et al., 2012). In total, 11,408 seedlings from 69 woody species were collected from 2015 to 2019, including 39 shrub species and 30 tree species (Tables S1 and S2).

2.3   Climate data and habitat variables

Using the nearest meteorological station (distance ~3 ​km), we measured interannual climate variation as the change in yearly climatic factors. Specifically, soil temperatures at 10 ​cm in depth were automatically monitored hourly using HOBO Prov 2 (HOBO Data Loggers-Onset Corporation of USA). Four climate factors were calculated to examine the effects of interannual climate variation on seedling survival: growing season maximum air temperature (GT), growing season vapor pressure deficit (VPD), non-growing season maximum precipitation (NGP), and non-growing season maximum soil temperature (NSGT) through correlation analysis (Tables S3 and S5). To avoid the inherent collinearity among climate variables, non-growing season maximum air temperature, growing season maximum precipitation and growing season maximum soil temperature were excluded by calculating variance inflation factors. A VIF value exceeding 5 was considered indicative of high multicollinearity, indicating a potential issue with excessive correlation between predictors.

Topography and soil properties were identified as the seedling habitat variables. The elevation of each of the 625 subplots was determined by averaging the elevations of the four corners of the subplot. Our previous analyses indicated that slope, aspect and convexity had no significant influence on seedling survival (Liu et al., 2022), so we excluded these variables from analyses in this study. Six soil characteristics (i.e., soil total phosphorus, soil total nitrogen, soil water content, soil pH value, soil volumetric weight, and organic carbon) were measured in the 25-ha plot every 20 ​m on a regular grid to form the soil-resource gradient. Using variogram models with conventional kriging, soil variables were predicted spatially for each of the seedling plots (Song et al., 2020). The principal component analysis (PCA) of the six soil variables was conducted because these abiotic variables were strongly correlated. The first two components (PCA1 and PCA2) accounted for 61.9% of the total soil variables. High levels of soil total organic carbon and soil total phosphorus as well as soil total nitrogen were linked to the PCA1 axis, and high soil water content and high soil pH were related to the PCA2 axis (Table S4).

2.4   Calculation of neighboring densities

In order to identify the seedling and adult neighbors for each focal seedling, we calculated the densities of conspecific seedlings (Scon) and heterospecific seedlings (Shet) in every seedling plot as well as the densities of conspecific trees (Acon) and heterospecific trees (Ahet) based on the total distance-weighted basal area (BA) of conspecific and heterospecific trees within a 10-m radius of each seedling plot (Kuang et al., 2017). According to the report of the models at five distances (5, 10, 15, 20 and 25 ​m) (Liu et al., 2022), the 10-m radius was caught since it had the lowest Akaike's information criterion (AIC) value. Each individual's basal area was divided by the distance (DIST) from its center to the seed trap.

where $i$ is a conspecific individual adult (Acon) or a heterospecific individual adult (Ahet). To correct the edge effect, only those seedlings that had a distance of at least 10-m from the 25-ha plot boundary were included.

2.5   Statistical analyses

We used the generalized linear mixed-effects model (GLMM) to simulate survival probability of the collected seedlings between 2015 and 2019 as a function of three different fixed factors: (1) climate variable, (2) biotic neighborhood, and (3) habitat condition. The GLMM in this paper was essentially a logistic regression, with the response variable as a logit transformation of seedling state: 1 (alive) or 0 (dead). By subtracting the mean and then being divided by the standard deviation, all the values of the continuous explanatory variables were standardized. This was done to run correlation analysis on various variables through putting different variables on the same scale. Seed-trap station, census year, and species were included as the random effects within our models (Chen et al., 2010). We also used the point method (Camarero et al., 2006) to estimate light availability (i.e., cover) through counting understory plants by placing metal rods (diameter ​= ​2 ​mm) every 4 ​cm.

To assess the relative contribution of the biotic and abiotic variables at the community-level, the following five alternative models were compared: (1) a null model only with random effect; (2) a biotic model with fixed effect of adult and seedling neighbors included in the null model; (3) a habitat model with fixed effect of topography and soil included in the null model; (4) a climate model with fixed effect of climate variables included in the null model; and (5) a full model with fixed effect of all variables included in the null model. Using the Akaike's information criterion (AIC), all of the models were compared (Table S6). Each model's fit was examined, and the interactions between conspecific neighborhoods and habitat variability and interannual climatic variability were included. Three subsets of the data were measured at: (1) community level (all the seedlings were pooled); (2) different age class (all the seedlings were pooled); and (3) species level (i.e., common and rare species, tree and shrub species) (Bai et al., 2012; Ye et al., 2014; Jiang et al., 2022). All the predictors were Z-scored before analysis, so the ratio for each coefficient can be used to quantify each predictor's relative importance (i.e., the exponential of the estimate of each coefficient) (Tables S7 and S8) (Gross et al., 2017).

All the analyses were conducted using the statistical environment R 3.6.2 (https://www.r-project.org/). The glmer function in the lme4 package was used to fit the models (Bates et al., 2015).

3.   Results

3.1   Effects of biotic and abiotic variables on seedling survival

Compared among the four models (i.e., null, biotic, habitat, and full), the AIC result showed that the full model was the best fit model at the community level (Table S6). Climatic factors displayed remarkable influences on seedling survival (47.22%): non-growing season soil maximum temperature (NSGT, 3.13%), growing season vapor pressure deficit (VPD, 20.09%) and growing season maximum air temperature (GT, 20.96%) showed significant positive effects on seedling survival, while non-growing season maximum precipitation (NGP, 3.05%) illustrated significant negative relationships with seedling survival (Fig. 1; Tables S7 and S8). Within the biotic factors, conspecific adult density (Acon, 6.37%) and conspecific seedling density (Scon, 14.86%) had the most strongly negative influence on seedling survival (Fig. 1; Table S8). Elevation had a positive influence on seedling survival (9.18%), while a significant negative effect of soil PCA1 (6.33%) was detected. In addition, light availability (6.7%) as well as biotic and abiotic factors showed a strongly positive impact on seedling mortality (Fig. 1; Table S8).

Figure  1.  Community-wide estimates of interannual climate variability, habitat conditions, and biotic neighborhoods on seedling survival (a). Relative importance of each factor, expressed as the percentage of explained variance (b). Odds ratios >1 indicate positive effects on seedling survival, while ratios <1 indicate negative effects on seedling survival, with 95% confidence limits (CL) indicated by horizontal lines. Confidence intervals that do not cross the zero line indicate that the variables considered have significant effects on seedling survival. ^*P ​< ​0.05; ^**P ​< ​0.01; ^***P ​< ​0.001.

Note: Cover, light availability; Scon, conspecific seedling density; Shet, heterospecific seedling density; Acon, conspecific adult density; Ahet, heterospecific adult density; PCA1, soil-resource gradient for seedling plot; PCA2, soil-resource gradient for seedling plot; meanelev, elevation of seedling plot; GT, growing season maximum air temperature; VPD, growing season vapor pressure deficit; NGP, non-growing season maximum precipitation; NSGT, non-growing season maximum soil temperature.

DownLoad: Full-Size Img PowerPoint

3.2   Effects of climate variability and habitat heterogeneity on CNDD

At the community level, CNDD varied considerably with climate variability and habitat heterogeneity. Increased VPD and GT enhanced the detrimental effects of conspecifics on seedling survival (P ​< ​0.001; Fig. 2a and b), whereas NSGT and soil PCA2 (higher soil moisture) decreased the detrimental effects of CNDD on seedling survival (P ​< ​0.001) (Fig. 2c and d). Models of seedling survival for all species displayed significant interactions between climate variability and conspecific density (Fig. S6). The negative effects of conspecific density were intensified by VPD conditions for shrub species and rare species (Fig. S6b1, d1).

Figure  2.  Response of seedling survival to conspecific density for various climate and habitat variables at the community level. We calculated the 5% (solid lines), 95% (dashed lines) quantiles and the mean values (dotted lines) of each variable to analyze the variation of CNDD with different levels of climate and habitat variables. The strength of CNDD at different level is: 95% ​> ​MEAN ​> ​5%. ^***P ​< ​0.001

Note: VPD, growing season vapor pressure deficit; GT, growing season maximum air temperature; NSGT, non-growing season maximum soil temperature; PCA2, soilresource gradient for seedling plot.

DownLoad: Full-Size Img PowerPoint

3.3   Effects of biotic and abiotic variables at age stage level

At the community-level, biotic and abiotic variables changed drastically with seedling age classes. For all age levels, climate variability showed a significant influence on seedling survival probability, and the correlation between interannual climate variation and seedling survival increased with seedling age (Fig. 3; Table S7). The strongest negative correlation was found between NSGT (71.44%) and focal seedling survival in the age class of 1–2 years (Fig. 3a; Table S8), whereas a significant positive correlation was also detected in the age classes of 3–4 years and ≥5 years (Fig. 3b and c; Table S8). NGP and VPD consistently tended to display a positive relationship with the survival of the seedlings throughout their entire lifetime (Fig. 3). The effect of GT on the survival probability varied with life stage: seedlings of 3–4 years old had a significantly higher survival rate (33.7%), but seedlings of ≥5 years old showed a relatively lower survival rate (27.03%) (Fig. 3b and c; Table S8).

Figure  3.  Odds ratios of seedling survival at different age stage in the temperate deciduous broad-leaved forest. A closed symbol indicates that the variables considered have a significant effect on seedling survival, while an symbol indicates that they do not have a significant effect.

Note: Cover, light availability; Scon, conspecific seedling density; Shet, heterospecific seedling density; Acon, conspecific adult density; Ahet, heterospecific adult density; PCA1, soil-resource gradient for seedling plot; PCA2, soil-resource gradient for seedling plot; meanelev, elevation of seedling plot; GT, growing season maximum air temperature; VPD, growing season vapor pressure deficit; NGP, non-growing season maximum precipitation; NSGT, non-growing season maximum soil temperature. Same as below.

DownLoad: Full-Size Img PowerPoint

For the tree seedlings at the age class of 1–2 years, soil PCA2 (8.66%) was significantly and positively correlated with their survival (Fig. 3a; Table S8), whereas for the tree seedlings at the age class of 3–4 years, soil PCA1 (4.11%) showed a significant negative influence on their survival (Fig. 3b; Table S8). As to the understory light availability (2.38%), the significant effect was only detected for the seedlings of ≥5 years old (Fig. 3c; Table S8), which was in line with that at the community level. Seedlings which were surrounded by more heterospecific neighbors usually displayed a strong positive survival effect at the age classes of 1–2 years (8.20%) and 3–4 years (7.87%) (Fig. 3b; Table S8), whereas conspecific neighbors tended to display a negative influence on seedling survival at the age classes of 3–4 years (3.77%) and ≥5 years (1.22%) (Fig. 3b and c; Table S8).

3.4   Effects of biotic and abiotic variables with species traits

Instead of using each species-trait type as a categorical predictor, we modeled the species groups separately in our model. Although the relative importance of climate factors on seedling survival changed significantly among species groups, they all contributed significantly to variation in seedling survival for all species (Figs. 4 and 5; Table S7). Climate variables showed a large relative contribution (48.44%) to the change in the survival of tree seedling, while biotic factors (36.34%) showed a slight effect on tree seedling survival. Similarly, climate variability also showed a larger relative contribution than biotic interaction for shrub species (47.81% vs. 20.58%), common species (45.58% vs. 33.65%), and rare species (43.9% vs. 20.74%) (Figs. 4 and 5; Table S7). Specifically, GT was significantly and positively associated with seedling survival of all species, while NGP only showed a significant positive influence on shrub seedling survival and a significant negative effect on tree seedling survival (Figs. 4 and 5; Table S8).

Figure  4.  Odds ratios of survival for tree and shrub seedlings in the temperate deciduous broad-leaved forest. A closed symbol indicates that the variables considered have a significant effect on seedling survival, while an symbol indicates that they do not have a significant effect.

DownLoad: Full-Size Img PowerPoint

Figure  5.  Odds ratios of survival for the seedlings of common species and for the seedlings of rare species in the temperate deciduous broad-leaved forest. A closed symbol indicates that the variables considered have a significant effect on seedling survival, while an symbol indicates that they do not have a significant effect.

DownLoad: Full-Size Img PowerPoint

It was found that biotic interactions showed a relatively larger contribution (36.34%) within tree species when compared to other trait groups (Table S7). Particularly, the influences of biotic factors on survival of seedling (mainly tree species and common species) were most strongly predicted by Scon, whereas Acon had a pronounced negative effect on shrub and rare species survival. Importantly, seedling survival of focal shrub as well as common and rare species was significantly positively influenced by light availability (Figs. 4b and 5).

4.   Discussion

This study found that interannual climate variation and biotic neighborhoods were the most strongly correlated with seedling survival. In particular, increased NSGT and light availability had a positive effect on seedling survival. Furthermore, we discovered that CNDD was stronger when GT and VPD increased and NSGT decreased, which is the projected outcome of climate warming. Species’ life-history strategies explaining biotic and abiotic factors varied according to the age stage of tree seedlings (Bai et al., 2012). Across species traits, the response of seedling mortality to biotic and abiotic variables was different. These results reinforce the idea that the studies of the influences of climate change on forest ecosystems could benefit greatly from a shift away from the viewpoint that community structure and population regulation are primarily shaped by environment-mediated changes in local biotic interactions (Uriarte et al., 2018; Xu et al., 2022).

4.1   Seedling survival was most sensitive to climate variability

Conspecific neighbor density was first raised to explain why the tropical forests maintain such a high level of tree diversity (Janzen, 1970; Connell, 1971), but its negative effects on plant survival have been well documented throughout the world (Comita et al., 2014; Zhu et al., 2018). According to the recent researches, biotic neighborhoods usually have a greater impact on seedling survival than abiotic conditions, and neighborhood interactions can help mitigate the influences of interannual environmental change in tropical and subtropical forests (Johnson et al., 2014; Uriarte et al., 2018; Yao et al., 2020). Our study found that neighborhood competition had a negative association with seedling survival at the community level (Fig. 1), being in line with previous researches in the tropical and subtropical forests (Comita et al., 2014; Chen et al., 2019; Xu et al., 2022). Our study indicates, however, that abiotic variables such as interannual climate variation are more predictive of seedling survival than biotic neighborhoods (48.22% vs. 24.46%) (Fig. 1; Table S7), suggesting that temporal variations in climate may make it easier to predict seedling survival dynamics in the temperate deciduous forest (Suttle et al., 2007; Alexander et al., 2015).

A significant positive association of GT and NSGT with seedling survival was found at the community level (Fig. 1). Additionally, a stronger influence of GT on seedling survival than NSGT was also observed (20.96% vs. 3.13%) (Table S8). Consequently, forest community responses to climate change are most closely related to microclimate change and localized microclimatic variation may enable additional temperature niches that exclude particular plant species assemblages, allowing for a higher number of plants to inhabit a given area (Blonder et al., 2015; Le Bagousse-Pinguet et al., 2017; Ohler et al., 2020; Zellweger et al., 2020). It was found that NGP was significantly and negatively associated with seedling survival, being in accordance with previous studies conducted in the tropical and temperate forests (Johnson et al., 2017; Xu et al., 2022) where drier climates and less frequent rainfall may decrease the power of fungal pathogens to affect density-dependent mortality (Swinfield et al., 2012). Numerous recent studies have demonstrated that the increasing VPD is increasingly restricting plant growth and recruitment in many forest ecosystems, sometimes to a larger degree than temperature and soil available water (Babst et al., 2019; Trotsiuk et al., 2021). In this study, however, it indicates that VPD has a greater impact on forest functioning than previously assumed, as a significant positive relationship consistently occurs between VPD and seedling survival (20.09%) (Fig. 1; Table S8). To our knowledge, this is the first study to demonstrate a direct link between seedling survival and VPD. It is important to keep in mind, however, that VPD impacts are not likely isolated from other environment factors. In certain conditions, such as in the case of fungal diseases, high VPD can create an unfavorable environment for pathogens (Duursma et al., 2019; Grossiord et al., 2020; Pappas et al., 2020; Trotsiuk et al., 2021; Sapak et al., 2023). At the community level model, we also observed a positive influence of canopy shade on seedling mortality (Fig. 1), indicating that a significant amount of light availability interception could result in a reduction in the mean amount and variation of solar radiation (Canham et al., 1990; Walker, 1991).

4.2   Interactions between environmental factors and conspecific density on seedlings survival

A changing environment and habitat heterogeneity may be contributing factors to CNDD effects on community dynamics (Song et al., 2018; Uriarte et al., 2018). There is a link between CNDD and climate, which has become increasingly evident in recent years because of intense climate change and extreme precipitation and temperature events (Clark et al., 2016; Romero et al., 2018; Uriarte et al., 2018). This present study indicates that in the temperate forest, interactions between CNDD and interannual climate variation have a significant impact on seedling survival (Fig. 2). Influences of conspecific neighbor density were intensified with increased GT and VPD (Fig. 2a and b; Table S8), in part because that the activity of soil fungal pathogen is controlled by low level of soil moisture, which physiologically restricts their capacity to adapt (Gadgil, 1974; Swinfield et al., 2012). Contrarily, we found that the influences of conspecific densities on seedling survival could be reduced by NSGT (Fig. 2c), which might be attributable to the promoted plant growth and photosynthetic rates induced by increased energy availability (Pauses and Austin, 2001). Furthermore, soil moisture would reduce the impacts of CNDD, which have been widely reported in both the tropical and temperate forests (LaManna et al., 2017; Xu et al., 2022), and might be interpreted by the habitat effect hypotheses (Comita and Hubbell, 2009). Overall, our results highlight that the neighborhood-mediated environmental effects can update out or even reverse the influences of changing abiotic environments on seedling survival in the temperate forest (Ettinger and HilleRisLambers, 2013; Chen et al., 2019).

4.3   Interannual climate variation was correlated with seedling survival at all life stages

Our findings revealed the significant non-random mortality at the early life stage in constructing heterogeneous communities by indicating that the influences of CNDD on seedling mortality were the strongest at the earlier life stages (13.96%) (Table S7). In fact, with growing seedling age, the impacts of biotic and abiotic variables tended to be more varied (Fig. 3) (Green et al., 2014; Zhu et al., 2018). The strength of CNDD was prevalent in plant communities at the earlier life stage (Zhu et al., 2018; Guo et al., 2020) partially because seedlings were more vulnerable to competition, environmental filtration, natural enemy harm, and/or negative density dependence at the earlier life stages (Kobe and Vriesendorp, 2011; Zhu et al., 2015; Song et al., 2020). Surprisingly, in our studied temperate deciduous broad-leaved forest, interannual climate variation was significantly correlated with seedling survival at all the life stages, and the impacts of the interannual climate variation on seedling survival increased with seedling age (71.44% at the age class of 1–2 years, 76.73% at the age class of 3–4 years, and 88.30% at the age class of ≥5 years) (Fig. 3; Table S7). Seedlings seem to be the most susceptible to herbivores and pathogens which could be altered when climate fluctuates (e.g., soil temperature) (Huang et al., 2008). We also found that seedlings at the later life stages were more susceptible to light availability (Fig. 3). Given the low level of light availability that most understory plants experience and the patterns that represent life-history variation in response to light (Clark et al., 1996; Wright et al., 2010; Johnson et al., 2014), we speculate that mature seedlings increased tolerance to environmental factors (e.g., light availability) with the developed tissues and organs, and these seedlings would occur in a suitable environment following passage through density dependence and environmental filtering during the early life stage (Guo et al., 2020; Liu et al., 2022).

4.4   Effects of biotic and abiotic factors varied depending on species type

Different species traits (i.e., tree and shrub; common and rare) showed potential niche partitioning in plant communities in response to biotic and environmental parameters (Figs. 4 and 5; Table S7). Shrub seedlings, for instance, did not show significant CNDD effects when compared with tree seedlings (Fig. 4), probably because there were more conspecific tree seedlings in the studied temperate forest (Comita and Hubbell, 2009; Chen et al., 2010). Additionally, this is in accordance with previous studies performed in the temperate and tropical forests, showing that the neighboring heterospecific seedlings had a slight significant positive influence on tree seedling survival (Comita and Hubbell, 2009; Chen et al., 2010; Bai et al., 2012; Xu et al., 2022). We also discovered that shrub seedlings were much more vulnerable to light availability than tree seedlings (Fig. 4a and b), suggesting that shrub seedlings’ lower mortality rate was probably due to the influence of their biotic neighbors. The relative abundance of species has also been linked to the differences in the strength of conspecific neighbor effects between species. This result is in accordance with previous studies performed in the tropical forest, showing that the common species experienced less conspecific neighbor effects (Fig. 5) when compared to the rare species in a community, probably because of the result of the strong influences induced by soil pathogens on the rare species (Mangan et al., 2010; Marden et al., 2017; Zhu et al., 2018). Thus, our results elucidate the vital role of local biotic interactions in regulating the responses of species traits to varying environments in their early life stage.

5.   Conclusion

In conclusion, the responses of seedling survival to biotic interactions and interannual climate variation as well as to habitat conditions were highly variable in the temperate forest. Our analyses indicated that interannual climate variation, followed by biotic interactions and then by habitat conditions, had the strongest correlation with seedling survival. Understory light showed a positive impact on seedling mortality and seedling survival responded differently to soil and air temperature at the community level. Influences of conspecific neighbor density were strengthened by increased maximum air temperature and vapor pressure deficits in the growing season, but were weakened by increased maximum soil temperature and maximum precipitation in the non-growing season. Analysis of life history strategy and its interactions with biotic and abiotic variables at different life stages is necessary to comprehend the mechanisms of plant species coexistence. Considering that the influences of CNDD on seedling mortality occurred early in life and that interannual climate variation was the most significantly correlated with seedling survival at all life stages, it is possible that species coexistence mechanisms may change along with ontogeny, according to the importance and strength of these processes varying with different life stage. We also found that the influences of the biotic and abiotic factors analyzed in this study on seedling survival changed significantly among various species traits, which could further support the framework for developing hypotheses related to the processes and mechanisms underlying species’ responses to temporal environmental change. These findings suggest that efforts to comprehend, conserve, and manage temperate forests should explicitly take into account the relative importance of intrinsic and extrinsic factors on forest dynamics, even though the implications for species dynamics and coexistence have yet to be fully investigated. These findings also reinforce the notion that, in a changing environment, it is crucial to take into account changes in local biotic interactions that are mediated by the environment when predicting shifts in forest communities. Further studies need to be conducted to examine how functional traits of species combine with biotic and abiotic variables to affect seedling survival differently among species, as well as other fundamental axes of niche differentiation, so as to obtain a mechanism framework to predict the shifts in forest communities in a changing environment.

Authors' contributions

Haikun Liu and Haishan Dang contributed the central idea, analyzed most of the data, and wrote the initial draft of the paper. The remaining authors contributed to refining the ideas, carrying out additional analyses and finalizing this paper.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

We appreciate their help of Mr. Zhenhai Wu and Mr. Gaodi Dang in identifying the species. We are grateful to the numerous field workers who helped develop and count the 25-ha permanent forest area. We appreciate the Foping National Nature Reserve's assistance with fieldwork logistics. We also thank the anonymous reviewers for their useful comments on an early version of the manuscript.

Abbreviations

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.fecs.2023.100138.