Understory bamboo removal impacts on woody seedling regeneration in forest ecosystems: a meta-analysis

Background Tree seedling regeneration in forests is often hindered by the competitive influence of dense under-story bamboo competition. While localized studies have investigated the ecological effects of understory bamboo removal on tree seedlings, a comprehensive analysis at a global scale is lacking. In this meta-analysis, we synthesized 497 observations from 32 experimental studies to assess the overall effects of understory bamboo removal on tree seedling regeneration. Results The results showed that understory bamboo removal enhanced tree seedling survival, emergence, and height growth. However, the response of tree seedlings to bamboo removal varied depending on regeneration characteristics, forest types, and bamboo removal methods. Specifically, understory bamboo removal increased the survival rate of deciduous seedlings but had no significant effect on evergreen seedlings. For regeneration stages, bamboo removal had a significant positive effect on the survival rate of saplings but not on seedlings. Regarding differences across forest types, bamboo removal significantly increased the emergence density of seedlings in deciduous broadleaf forests but had a significant negative influence in evergreen and mixed evergreen-deciduous forests. Additionally, natural removal of bamboo showed a greater positive effect on seedlings than bamboo removal by artificial or animal gnawing methods. Furthermore, we found that the duration of bamboo removal, mean annual temperature, precipitation of seasonality, and soil pH strongly influenced the response ratios of tree seedling regeneration. Conclusions Our meta-analysis demonstrates the significant effects of understory bamboo removal on multiple facets of tree seedling dynamics across different regeneration characteristics, forest types, and bamboo removal methods. In addition, our study emphasizes that the duration of bamboo removal, climate, and soil pH have a critical effect on tree seedling regeneration. Our findings elucidate the effects of understory bamboo removal on seedling regeneration, offering robust scientific insights for sustainable forest management.


Introduction
Bamboo, a perennial flowering plant belonging to the subfamily Bambusoideae of the family Poaceae, predominantly thrives in tropical and semi-tropical habitats (Soderstrom and Calderon 1979).Its propagation relies primarily on asexual reproduction, specifically underground stems called bamboo whips, which contribute to its rapid growth, large biomass, and high litter production (Christanty et al. 1996).Previous studies have highlighted the negative effects of bamboo dominance in the understory on various aspects of forest ecosystems (Pearson et al. 1994;Itô and Hino 2004;Buziquia et al. 2019), including tree growth, tree seedling regeneration, and nutrient turnover in the understory (Taylor et al. 2004;Larpkern et al. 2011).In particular, understory bamboo has been posited as an ecological filter (Itô and Hino 2007), exerting robust competitive pressure on the emergence, establishment, and survival of tree seedlings in forests by effectively capturing light and vital resources (Takahashi et al. 2003;Doležal et al. 2009;Giordano et al. 2009;Montti et al. 2011a, b;Wijewickrama et al. 2022).
Tree regeneration plays a crucial role in forest succession, particularly during the seedling stage (Green et al. 2014;Marchin and Yuan 2023), as it directly and indirectly influences the composition, structure, and dynamics of forest communities (De Lombaerde et al. 2021).Nonetheless, the process of seedling regeneration within the forest understory is significantly influenced by the presence of dense bamboo, which alters light availability and soil nutrient dynamics (Narukawa and Yamamoto 2002;Qian et al. 2016).For instance, dense understory bamboo reduces seedling emergence by altering the microsite environment (Qian et al. 2019) and hindering the growth and establishment of tree seedlings in the forest (Itô and Hino 2007;Doležal et al. 2009).Conversely, areas without bamboo exhibit increased biomass, height, leaf count, and bud production in saplings (Marchesini et al. 2009).Consequently, effective management of dense understory bamboo is imperative to ensure the robust and sustainable development of forest communities.
However, the influence of understory bamboo on tree seedling regeneration and growth can be subject to alteration through bamboo die-off or removal.Firstly, most bamboo species exhibit a unique life history characterized by a singular reproductive event leading to massive flowering and subsequent bamboo die-off (Janzen 1976;Zheng et al. 2020), which has been shown to enhance tree seedling regeneration and growth (Montti et al. 2011a, b;Abe et al. 2005;André 2021).Secondly, understory bamboo may decline due to gnawing by wild animals (Yokoyama et al. 2000).In Japanese forests, the gnawing behavior of sika deer on dwarf bamboo promotes the survival of Abies homolepis seedlings (Itô and Hino 2004), and alters the emergence and growth of small coniferous seedlings (Kumar and Shibata 2007).Moreover, the artificial cutting of bamboo has a notable impact on tree seedling regeneration.Specifically, artificial bamboo harvesting can greatly enhance the survival rate of seedlings compared to uncut areas (Bitariho and McNeilage 2008;Doležal et al. 2009).In addition, the impact of bamboo removal on the understory environment is multifaceted, with direct consequences for tree seedling regeneration.Notably, bamboo removal leads to an increase in both the quantity of incident solar radiation and the red to far-red light ratio within the understory (Montti et al. 2011a, b;Giordano et al. 2009).These alterations in light availability can directly impact the photomorphogenic responses of seedlings, potentially influencing their survival and growth.Additionally, the removal of bamboo has been observed to mitigate the reduction rate of soil moisture following the rainfall events, potentially enhancing the water availability for seedling growth (Takahashi et al. 2003).Consequently, understanding how bamboo removal impacts the dynamics of tree seedlings through environmental changes is essential to formulating forest management strategies that balance ecological health and conservation objectives.
Over the past two decades, numerous single-site experimental studies have investigated the effects of bamboo removal on tree seedling regeneration, primarily in East Asia and South America (Takahashi et al. 2003;Taylor et al. 2004;Itô and Hino 2007;Giordano et al. 2009;Montti et al. 2011a, b).However, these studies are conducted across different climatic regions, influenced by diverse environmental factors and varying biomes (forest types, life form, regeneration characteristics), resulting in disparate conclusions.For instance, a study in the Patagonian mixed forest investigated the impact of bamboo die-off on tree seedling regeneration and uncovered both positive and negative effects on tree seedlings of different species within the gaps created by bamboo die-off (Caccia et al. 2015).Conversely, a study in Japan demonstrated that understory bamboo removal had a positive effect on seedling emergence (Itô and Hino 2007).Furthermore, the response of tree seedling regeneration to understory bamboo removal exhibits variability, which can be attributed to the diverse responses of different species to alterations in light and nutrient availability, particularly during the sensitive seedling stage (Hammond and Brown 1995;Ligot et al. 2014).Additionally, in contrast to the phenomenon of bamboo flowering and die-off, the underground stems of understory bamboo remain alive even after their aboveground parts are cut down or gnawed by animals.This leads to a competitive interaction between the underground bamboo stems and the roots of tree seedlings, thereby affecting the growth of tree seedlings (Mommer et al. 2016).In summary, although multiple case studies have examined the effects of bamboo removal on tree seedling regeneration (Itô and Hino 2007;Caccia et al. 2015), there is considerable variation among these studies.Furthermore, climatic conditions such as temperature and precipitation characteristics varied among study areas, particularly between Argentina, Chile, China, and Japan (Caccia et al. 2015;Itô and Hino 2007;Muñoz et al. 2012;Taylor et al. 2004).The effects of understory bamboo removal on seedling regeneration may vary among study areas depending on climatic conditions.Variations in temperature and precipitation lead to differences in seedling recruitment and height growth (Al-Hawija et al. 2015;Walbott et al. 2018).Consequently, it is crucial to synthesize and integrate the findings from these diverse studies in order to draw a comprehensive conclusion regarding the impact of understory bamboo removal on tree seedling regeneration at a global scale.
In this study, we conducted a meta-analysis to synthesize the differences observed in bamboo removal studies conducted across various climate regions and biomes.To achieve this, we performed a systematic literature search to identify studies investigating the effects of understory bamboo removal on various aspects of tree seedling regeneration, including emergence, survival, and growth in terms of height (De Lombaerde et al. 2021).We addressed the following research questions: (1) How do tree seedling regeneration processes, including emergence, survival, and growth, respond to understory bamboo removal?Our hypothesis posited that the removal of understory bamboo would benefit seedling establishment, survival, and growth of seedling by reducing competition.(2) What are the primary factors that significantly influence the effects of bamboo removal on tree seedling regeneration, and how do these key factors relate to the observed effects on seedling regeneration?We expected that regeneration characteristics, forest types, bamboo removal methods, and environmental factors (e.g., climate conditions, and soil factors) would play pivotal roles in shaping the response of seedling regeneration to bamboo removal.Moreover, we predicted that as the mean annual temperature increased, the response ratios of seedling emergence, survival, and height growth gradually decreased and eventually turn negative.

Literature search
To investigate the relationship between understory bamboo removal and tree seedling regeneration, we conducted a comprehensive literature search until December 2023, using the Web of Science (https:// www.webof scien ce.com/ wos/ alldb/ basic-search) and the China National Knowledge Infrastructure Database (https:// www.cnki.net).We employed the following keyword combinations: [(bamboo) AND (seedling* OR sapling* OR regeneration* OR recruitment*) AND (effect*)].Our primary focus was on woody plant seedlings.Initially, we retrieved a total of 624 articles from many countries (Table A1), with each article subjected to careful screening to ensure it met our analysis criteria.These criteria were as follows: (1) involve removal methods of understory bamboo (e.g., cutting, gnawing, or flowering); (2) provide regeneration data, both with and without bamboo presence; (3) include seedlings or saplings with a height less than or equal to 100 cm; (4) report effects on seedling emergence, survival, height growth, diameter growth, or biomass growth, along with sample size and error estimates.Although we have collected numerous articles about understory bamboo die-off, being cutting or gnawing, most of them do not have data about bamboo before and after treatment.Finally, our meta-analysis incorporated a total of 32 selected articles, as detailed in Table A2.These articles collectively encompass data on six distinct regeneration indicators (Table 1), with each article contributing one or more of these indicators to our study.

Data extraction
We conducted data extraction by retrieving the means, standard deviations (SD), or standard errors (SE), along with their corresponding sample sizes, from figures using the software "GetData" or directly from the texts and tables of the original references.In cases where the SD was not reported, we calculated it using SE and sample size.A total of 497 observations were extracted, mainly from South America and East Asia (Fig. 1).These observations comprised 205 survival data, 51 growth data, and 241 emergence data.Due to the limited sample size, data on diameter and biomass were not included in our analysis.The regeneration indicators encompassed emergence rate, emergence density, survival rate, survival density, height relative growth rate, and height annual growth of tree seedlings.
To assess the potential impacts of bamboo removal on tree seedling regeneration, we incorporated three key categories of potential variables (Table 2).The first category was regeneration characteristics, including the method of seedling regeneration, seedling life form, and regeneration stage (Gibert et al. 2016).The second category encompassed forest types, consisting of seven types such as coniferous, deciduous, and mixed coniferous-broadleaf forests.The third category comprised experimental factors, primarily focusing on the bamboo removal methods and duration of bamboo removal.Furthermore, we extracted relevant important environmental factors from articles, the WorldClim dataset (https:// world clim.org), and the SoilGrids database (https:// soilg rids.org/) according to the geographic coordinates (latitude and longitude), including mean annual temperature (MAT), mean annual precipitation (MAP), precipitation of seasonality (Pre-season), and soil pH.

Quantifying effect size
To quantify the effect size of understory bamboo removal on tree seedling regeneration, we calculated the log response ratio (lnRR) for treatment group from each study (Hedges et al. 1999).The lnRR was calculated as follows: where X t represents the mean of the seedling regen- eration indicator without bamboo (cutting, gnawing, or flowering), and X c represents the mean of the seedling regeneration indicator with bamboo.
Given the variability in data quality across studies, we evaluated data quality based on sample size and weighted the effect ratio accordingly: where n t is the sample size of treatment group, and n c is the sample size of control group.
The relationship between standard deviation (SD), standard error (SE), and sample size is described as follows (Zhang et al. 2022):

Data analysis
To quantitatively assess the impact of bamboo removal on seedling regeneration indicators, we employed mixedeffects models, which incorporate random effects to account for non-independence in the data (Kisamore and Brannick 2008).These models were specifically chosen to allow for the exploration of heterogeneity across studies, enabling us to test the influence of potential covariates on the observed effects.The equation for our mixed-effects model is detailed as follows: (1) Table 2 Variables potentially influencing the effects of bamboo removal on tree seedling regeneration In this model, µ represents the fixed intercept, α j represents the random effect for the j th study, which is assumed to be normally distributed with a mean of zero and variance σ 2 α (i.e., α j ~N (0, σ 2 α ), β is the fixed effect coefficient for the bamboo removal treatment X ji , and E ji represents the sampling error for the i th observation (4) within the j th study, assumed to be normally distributed with a mean of zero and constant variance σ 2 ǫ .Furthermore, we conducted subgroup analyses for categorical variables (regeneration characteristics, forest types, and bamboo removal methods), to estimate their mean effect sizes and to assess the significant differences among subgroups in the impact of understory To determine the influence of covariates on the effect of bamboo removal, we performed model selection analysis using Akaike's information criterion (AIC) with the 'MuMIn' package in R (Bartoń 2018).For each possible model, we calculated the AICc values (corrected for small sample sizes), ΔAICc (difference in AICc compared to the top-ranked model), and Akaike weights (probability that a model is the best possible model).The relative contribution for a particular covariate was then calculated as the sum of Akaike weights (W) derived for all the models through model averaging (Anderson 2008).A threshold W value of 0.8 was set to distinguish important and nonimportant covariates (Terrer et al. 2016;Chen et al. 2020a, b).The relationship between the sum of Akaike weight (W) and ΔAIC is described as follows (Si et al. 2014): where AIC i = AIC i − AIC min , AIC i is the AIC value for model i, and AIC min is minimum AIC value in the model set.
Additionally, a linear regression model was constructed to explore the relationship between the regeneration response ratio and covariates, including mean annual temperature (MAT), precipitation of seasonality (Pre-season), soil pH and duration of bamboo removal (period).The mixed-effect model was implemented using the R package 'lme4' , and the linear regression was performed using the R package 'stats' (Collyer and Adams 2018), with a significance level set at p < 0.05.All figures were generated using the R packages 'ggplot2' and 'ggpmisc' .

The impacts of regeneration characteristics, forest types, and removal methods on tree seedling regeneration
To further clarify the response of seedling emergence, growth, and survival to bamboo removal under different conditions, we performed subgroup analysis.We observed significant disparities in seedling responses to understory bamboo removal across various regeneration characteristics (seedling life forms, regeneration stages, and regeneration methods).Specifically, we found that understory bamboo removal had a significant positive effect on the annual height growth of both deciduous and evergreen seedlings (Fig. 3B).Furthermore, understory bamboo removal led to an increase in the survival rate of deciduous seedlings (leaf_d), while no significant effect on survival rate of evergreen seedlings (leaf_e) was observed (Fig. 3C).In terms of regeneration stages (e.g., saplings and seedlings), bamboo removal increased the survival rate of saplings but not seedlings (Fig. 3C).Conversely, bamboo removal improved the annual height growth of seedlings but not saplings (Fig. 3B).In addition, we found that understory bamboo removal had a positive effect on emergence density under natural removal conditions, but it had no effect under unnatural removal conditions (Fig. 3A).Furthermore, our results demonstrated that the response of tree seedling regeneration differs (Fig. 3).In terms of emergence density, bamboo removal increased seedling density in deciduous broadleaf forest (DBF), while it had a significant negative influence in evergreen forest (EF) and mixed evergreen-deciduous forest (EDF_ mixed), with no significant effect observed in other forest types (Fig. 3A).Additionally, the survival rate exhibited a significant positive effect in deciduous forest (DF) and mixed coniferous broad-leaved forest (CBF_mixed), while a negative effect was observed in evergreen broadleaved forest (EBF) (Fig. 3C).
Moreover, we found that the effect of unnatural removal group on emergence density was significantly positive, while no significant effect was observed in natural removal group (Fig. 3A).Furthermore, the natural removal method had a significant and positive effect on the annual height growth of tree seedlings, whereas the unnatural removal group exhibited the opposite trend (Fig. 3B).Bamboo removal had a positive effect on the survival rate under both natural and unnatural removal, with the effect being stronger in the natural removal group (Fig. 3C).

Relative importance covariates on tree seedling regenerations
Model-averaged relative importance of covariates regulating the effects of bamboo removal on tree seedling regenerations (Fig. 4).Specifically, we found the effects of bamboo removal were mainly explained by changes in seedling emergency density across different soil pH (Sum of Akaike weights = 0.98), forest types (Sum of Akaike weights = 0.98), mean annual temperature (MAT, Sum of Akaike weights = 0.98), duration of bamboo removal (Period, Sum of Akaike weights = 0.92), bamboo removal method (Sum of Akaike weights = 0.85), and mean annual precipitation (MAP, Sum of Akaike weights = 0.81) (Fig. 4A).Moreover, for annual height growth, life forms (Sum of Akaike weights = 0.4) and bamboo removal methods (Sum of Akaike weights = 0.4) were identified as the primary factors, although their sum of Akaike weights did not exceed 0.8 (Fig. 4B).Furthermore, the analysis showed that the regeneration methods (Sum of Akaike weights = 0.95) and forest types (Sum of Akaike weights = 1.0) played crucial roles in explaining the effects of bamboo removal on survival density and survival rate, respectively (Fig. 4D, E).

Effects of period and environmental factors on tree seedling regeneration
Our results showed the influence of mean annual temperature (MAT), precipitation of seasonality (Pre-season), soil pH, and duration of bamboo removal (Period) on the natural logarithm of the response ratios (lnRR) of tree seedling regenerations (Table 3 and Figs. 5, 6).Specifically, the lnRR for emergence density (lnRR.ed)was negatively correlated with MAT and positively correlated with soil pH (Table 3 and Fig. 5).In addition, the lnRR for height growth (lnRR.gh)showed a significant decline with both Per-season and soil pH, contrasting with a positive correlation with period (Table 3 and Fig. 5).Moreover, the lnRR for survival rate (lnRR.sr)was negatively correlated with MAT, showing no significant response to Pre-season and soil pH (Table 3 and Fig. 6).

Discussion
Understanding the impacts of understory bamboo removal on tree seedling regeneration in forests is crucial.In this study, we conducted a comprehensive metaanalysis of literature primarily sourced from Asia and South America.Specifically, we investigated the response of tree seedling regeneration to understory bamboo removal across various regeneration characteristics, forest types, bamboo removal methods, bamboo removal durations, and environmental factors.Our findings shed light on the importance and effects of understory bamboo removal on bamboo forest management.

Positive effects of bamboo removal on tree seedlings' emergence, survival, and growth
Our findings revealed that understory bamboo removal improved tree regeneration, likely attributed to the alleviation of competition for light, water, and nutrient resources between seedlings and bamboo, which is consistent with previous studies (Abe et al. 2002;González et al. 2002;Giordano et al. 2009;Montti et al. 2011a, b).The alteration of the microsite environment in the understory after bamboo removal affects tree seedling regeneration (Marchesini et al. 2009;Montti et al. 2011a, b).Specifically, the removal of understory bamboo would create a favorable environment for seedling height growth by alleviating resource competition, particularly for light (Gratzer et al. 1999;Colleta et al. 2001).Additionally, understory bamboo removal enhances tree seedling emergence by increasing understory openness and reducing leaf litter cover, which are likely the main factors promoting this process (Nakagawa et al. 2023;Sayer 2006).Moreover, fluctuations in soil temperature resulting from understory bamboo removal also play a role in facilitating the emergence of pioneer seedling species (Abe et al. 2002).Our findings align with the previous experiments demonstrating that understory bamboo removal promotes seedling survival (Larpkern et al. 2011).

Effects of regeneration characteristics, forest types, and removal methods on tree seedling regeneration
Our analysis indicated significant variations in seedling responses to understory bamboo removal across various regeneration characteristics (seedling life forms, regeneration stages, and regeneration methods), forest types, and removal methods.Regarding life forms, our findings revealed that the removal of understory bamboo had no effect on the survival rate of evergreen seedlings but showed a positive effect on deciduous seedlings.This effect on deciduous species may be attributed to their shade-tolerance and their ability to regenerate Fig. 3 Effects of bamboo removal on seedling emergence density (A), height growth (B), and survival rate (C) in different conditions.The horizontal lines on either side of the point represent 95% confidence intervals.Text on the left side is the sample size of each covariate.The black line in the middle indicates no effect.H.Growth: annual growth in height; rege_plant: artificially transplanted; rege_colonization: natural regeneration; rege_seed: artificially sow; leaf_e: evergreen seedling; leaf_d: deciduous seedling; leaf_both: both evergreen and deciduous seedlings; EF: evergreen forest; DF: deciduous forest; CF: coniferous forest; DBF: deciduous broadleaf forest; EDF_mixed: mixed evergreen-deciduous forest; CBF_mixed: mixed coniferous and broadleaf forest; CDF_mixed: mixed coniferous-deciduous forest aggressively in gaps, as they are typically shade-intolerant or intermediate (Zhu et al. 2014).For instance, a recent study conducted in secondary forests in northern China showed that the abundance and density of shade-intolerant species were higher than that of shade-tolerant species in forest gaps during natural regeneration (Lu et al. 2019).Conversely, evergreen species exhibit a more conservative resource-use strategy, resulting in less variability in their growth and survival across different habitats (Reich 1998;Jin et al. 2017).Moreover, deciduous seedlings exhibit greater morphological and physiological plasticity, enabling them to benefit more from bright light conditions compared to evergreen species (Ke and Werger 1999).These findings align with previous studies, Fig. 4 The relative importance of covariates was assessed in their influence on various responses to bamboo removal, including seedling emergence density (A), height growth (B), relative growth rate of height (C), survival density (D), and survival rate (E).The variables include forest types, environmental factors (e.g., MAT and MAP), regeneration characteristics (e.g., life form and regeneration method), and experimental factors (e.g., method and duration of bamboo removal).The cutoff was set to 0.8 to differentiate between important and nonessential variables.Leaf: life form; Regeneration: regeneration method; Removal: removal method; Period: duration of bamboo removal; MAT: mean annual temperature; MAP: mean annual precipitation; Pre-season: precipitation of seasonality suggesting that shade-tolerant species have limited sensitivity to strong light environment (Yoshida and Kamitani 2000;Lin et al. 2002).However, there was no significant difference in emergence density among different life forms.Studies have shown that seedling emergence is more influenced by soil conditions than life form (Iijima et al. 2007), especially soil temperature (Qian et al. 2019).
In addition, our findings revealed distinct responses among seedlings at different regeneration stages to understory bamboo removal.Specifically, we observed that understory bamboo removal increased the survival rate of saplings but had no effect on the seedling survival rate.We speculated that although the increased light conditions following understory bamboo removal enhanced seedling survival, for small individuals, the removal of large amounts of bamboo from the understory increases their vulnerability to herbivores, thereby offsetting the positive effects and potentially leading to negative outcomes (Nakashizuka 2001;Lusk 2004).However, we did not find a significant effect of understory bamboo removal on sapling height growth, but a significantly positive effect on height growth of seedlings.Height growth appears to be encouraged by bamboo removal only at a very early growth stage (Sumida et al. 1997).
Furthermore, previous studies have demonstrated notable variations in understory light conditions across various forest types (Kabakoff and Chazdon 1996;Van Der Meer et al. 1998;Montgomery and Chazdon 2001).
Our study showed that bamboo removal promotes survival rates in deciduous forests and mixed coniferousbroad leaved forests, but inhibits survival in mixed evergreen-deciduous forests.This discrepancy can be attributed to the higher light availability in the former two forest types during the deciduous stages, which potentially enables more seedlings to survive (Tanouchi 1996;Zhou et al. 2020).Interestingly, our findings indicated negative responses in emergence density to bamboo removal in evergreen forests, mixed evergreendeciduous forests, and deciduous forests, aligning with previous studies (Abe et al. 2002;González et al. 2002;Holz and Veblen 2006;Muñoz et al. 2012).A previous case study found that the emergence of Aucuba japonica, Cercidiphyllum japonicum, and Sorbus commixta was negatively affected by the removal of bamboo (Abe et al. 2002).However, another separate experiment in Chile has demonstrated that the establishment of Araucaria araucana, Nothofagus dombeyi and Laureliopsis philippiana understory was not promoted by bamboo removal (Muñoz et al. 2012).Additionally, a recent study has demonstrated that a strong and positive relationship between microsite light availability and seedling height and diameter (Sangsupan et al. 2021).However, our findings showed that forest type has no significant effect on annual height growth.We suspected that the increase in light resulting from understory bamboo removal in these forest types is not sufficient to significantly affect the height growth of seedlings compared to the original light intensity (Liu et al. 2018).
Additionally, our findings showed that natural removal significantly enhanced the survival and height growth of seedlings, whereas unnatural removal had no such effect.Compared with natural removal, unnatural removal (artificial cutting and animal gnawing) bamboo will leave the underground parts (bamboo whips) and sometimes aboveground parts (Qian et al. 2019), which will compete with understory seedlings for resources, affecting their survival and growth of seedlings (Narukawa and Yamamoto 2002;Tripathi et al. 2005).Interestingly, natural removal had no impact on seedling emergence density, while unnatural removal significantly promoted it.The thick litter layer produced after bamboo flowering is a hindrance to seedling emergence (Cavieres and Arroyo 2000;Hastwell and Facelli 2000;Larpkern et al. 2011), and there is no litter production in unnatural removal.

Factors influencing the effect of bamboo removal
Seedling establishment and survival, representing the most climate-sensitive phase of tree development, are influenced by various environmental factors (Petrie et al. 2016;Bailey et al. 2021).Among these factors, temperature emerged as a crucial determinant.Notably, the effect of bamboo removal on emergence density and height growth shifted to negative when the mean annual temperature exceeds 12 °C.Our findings indicated that the removal of understory bamboo results in the creation of gaps and an increase in light intensity, which in turn can cause a rise in temperature (Giordano et al. 2009;Caccia et al. 2015).This can lead to elevated temperatures and reduced soil moisture content (Marchesini et al. 2009), particularly in areas with high mean annual temperatures, thereby impacting the growth of seedlings.In addition, the successful emergence of seedlings often requires low temperatures to break dormancy, making areas with lower average annual temperatures more favorable for seedling emergence (Junttila 1973;Chen et al. 2020a, b;Song et al. 2023).Moreover, water availability is another critical factor for seedling regeneration (Delerue et al. 2015;Verhoeven et al. 2018).The coefficient of variation of precipitation seasonality (Pre-season) reflects the extent of seasonal fluctuation in precipitation.High precipitation seasonality means heavy precipitation concentrates on multiple months (Liu et al. 2021).We observed that areas with more stable precipitation patterns were more conducive to seedling establishment and growth.This suggests that the greater the value of precipitation seasonality (Pre-season), the greater the possibility and frequency of drought and flooding events in the area, which are detrimental to seedling growth and survival (Galiano et al. 2013;Maxwell et al. 2016).
In addition, the chemical properties of the soil, which serves as the substrate for seedling regeneration, play a crucial role in the regeneration process (Martini et al. 2020;Puhlick et al. 2021).We observed a negative correlation between the response ratio of seedling height growth and soil pH.Forest soils in many regions are acidic (Gilliam 1991;Valentin et al. 2018), and the species living in them have adapted to acidic soil.Therefore, acidic soil is more suitable for survival and growth than neutral soil for seedlings of these species.
Moreover, the effects of understory bamboo removal on emergence density and annual height growth were found to be strengthened as the bamboo removal period increased.This delayed response can be attributed to the necessary time for the understory environment to adapt, primarily due to the slow decomposition of bamboo litter, leading to a prolonged release of nutrient resources that support seedling growth (Singh and Singh 1999;Marchesini et al. 2009).Over time, more nutrient resources are released for seedling growth.However, we found that long treatment times had a negative impact on seedling survival.We speculated that with a longer interval between bamboo removal and the start of the experiment, new bamboo seedlings may emerge in the bamboo flowering areas, influencing seedling survival of other species in the understory (Marchesini et al. 2009;Montti et al. 2011a, b;Montti et al. 2011a, b).These findings underscore the complexity of the interactions between understory bamboo removal and seedling responses, emphasizing the importance of considering environmental factors and temporal dynamics in bamboo forest management and conservation efforts.
While our study provides valuable insights into the impacts of bamboo removal on tree seedling regeneration across varying conditions, it is essential to acknowledge several limitations.Our dataset predominantly originates from East Asia and South America, potentially limiting the generalizability of our findings to other regions.In addition, due to constraints in sample size, we were unable to collect environmental data after bamboo removal, hindering our ability to elucidate direct influences on seedling regeneration dynamics.Moreover, while our study focused on tree seedling regeneration, the regeneration of herb and shrub communities after bamboo removal remains unclear.Therefore, future research should focus on extensive data collection across diverse ecosystems and environmental conditions to enhance our understanding of forest regeneration dynamics.

Conclusions
Our meta-analysis demonstrates the significant positive effects of understory bamboo removal on multiple facets of tree seedling dynamics, encompassing survival rate, density, emergence characteristics, and height growth indices.Notably, the influence of bamboo removal on tree seedling regeneration varied greatly across different regeneration characteristics, forest types, and bamboo removal methods.Furthermore, the duration of bamboo removal, mean annual temperature, precipitation of seasonality, and soil pH exerted strong impacts on the response ratios of tree seedling regeneration.These findings not only enhance our understanding of the ecological implications of bamboo removal across diverse forest ecosystems, but also provide valuable insights for seedling management and conservation of post-clearance.

Fig. 1
Fig. 1 Geographical distribution of the 32 studies identified through our systematic literature search.Panel a gives an overview of the studies on a global scale, whereas Panel b on the studies from South America and Panel c on those from East Asia

Fig. 2
Fig. 2 Forest plot showing the mean weighted response ratios of bamboo removal on seedlings regeneration indicators.The horizontal lines on either side of the point represent 95% confidence intervals.Numbers on the left side are the observation sample size of each indicator.The black line in the middle indicates no effect.H.RGR: the relative growth rate of height.H.Growth: annual growth in height

Fig. 5
Fig. 5 Regression models of emergence density and annual height growth response ratio to key drivers.MAT: mean annual temperature; Pre-season: precipitation of seasonality; pH: soil pH; Period: duration of bamboo removal

Fig. 6
Fig. 6 Regression models of survival rate to drivers.MAT: mean annual temperature; Pre-season: precipitation of seasonality; pH: soil pH; Period: duration of bamboo removal

Table 1
Seedling regeneration indicators and calculation formulas 0 = height in the previous year H 1 = current year height H.Growth (H 1 -H 0 )/Time H 0 = height in the previous year H 1 = current year height Time = years of experiment

Table 3
Parameters of linear regression of environmental factors and period with the natural logarithm of the response ratios (lnRR) for seedling regeneration indicators MAT mean annual temperature, Pre-season precipitation of seasonality, pH soil pH, Period duration of bamboo removal.k: slope of the regression equation. *