Stand Characteristics Rather than Soil Properties Contribute More to the Expansion of Moso Bamboo (Phyllostachys edulis) into Its Neighboring Forests in Subtropical Region

Abstract

Moso bamboo (Phyllostachys edulis), once highly praised worldwide, has been found to be a problematic species due to its unconstrained expansion into adjacent woodlands and negative effects on the function services of forest ecosystems. To determine the major factors affecting bamboo expansion into neighbor woodlands, we investigated the expansion characteristics of moso bamboo and the properties of stand structure and soil for 58 bamboo–woodland interfaces (BWIs) across Jiangxi province in China. Then, we analyzed the relationships between the variables of bamboo expansion and the properties of interfaces through a redundancy analysis. The characteristics (the expansion distance and the number and size of new culms) of moso bamboo expansion into disturbed forests were more significant (p < 0.01) than those into non-disturbed forests. The bamboo expansion into deciduous broad-leaved forest was much faster (1.33 m/yr) than evergreen broad-leaved forest (0.82 m/yr) and needle-leaved forest (1.08 m/yr). The characteristics of stand structure had more direct explanatory power (58.8%) than soil properties (4.3%) and their interaction (10.0%) for the variations in bamboo expansion. The canopy closure of recipient forests was identified as the most significant factor negatively correlated to bamboo expansion. The number of parent culms and the ratio of deciduous to evergreen trees ranked in sequence, and both imposed positive effects on the expansion. Regarding soil properties, only the water content was identified for its explanatory power and negative influence on bamboo expansion. Our findings illustrated that the expansion of moso bamboo showed remarkable variations when facing different woodlands. Stand characteristics (canopy closure, canopy height, etc.) of good explanatory power were the major variables affecting the expansion of moso bamboo. In order to control the expansion of bamboo and protect woodlands, disturbances (extracting timber, girdling trunks) should be prevented in bamboo–woodland interfaces.

1. Introduction

As a diverse species group of ca 1200 species and 100 genera, bamboo (Poaceae: Bambusoideae) lives across the tropics, subtropics and temperate zones with a huge area of more than 20 million hm² [1,2]. Bamboo species play an important role in rural economics for their versatile usages as construction materials, edible shoots, pulp and for medical purposes [3,4,5,6]. Additionally, they are characterized by several unique features apart from common trees. For example, their new culms of rapid growth in height are successfully cloned every year [7,8]. For bamboos with leptomorphous rhizomes, their clonal reproduction helps the population of bamboo laterally expand into neighboring woodlands. This characteristic is frequently utilized to enlarge the area of bamboo forests, but it also makes “running” bamboo species invasive [9]. Their aggressive expansion has brought unexpected ecological problems [10,11,12,13,14,15]. As a result, the expansion of bamboo is regarded as one of the great threats to the conservation, especially for the nature reserves of south China [16,17].

Moso bamboo (Phyllostachys edulis) is one of the most popular bamboo species with multiple usages, and it lives widely throughout subtropical China [4]. Groves of this species approximately cover 3.76 million hm² across 12 provinces including Fujian, Jiangxi, Zhejiang, etc., [18]. It was also introduced to Japan as an important forest resource [11]. Because of its strong capability of spreading through clonal recruitment, moso bamboo exhibits an extreme overgrowth in population. It overcrowds other tree species quickly, and it even takes over their habitats in a relatively short time [19]. Consequently, it has been considered as a problematic species due to its overwhelming expansion into neighboring forests [10,15,17,20,21].

In recent years, numerous ecological effects caused by the expansion of moso bamboo have been evaluated, such as the simplification of stand structure [22,23] and the decline in plant species diversity [24,25] as well as changes in soil taxonomic diversity [26,27,28,29]. Cai et al. (2018) found that the fast expansion of moso bamboo may alter the chemical and physical properties of soil in adjacent forests [30]. Song et al. (2016) observed that bamboo expansion may modify the transformation processes of soil nitrogen (N) for evergreen broad-leaved forests [31]. Some researchers reported that the conversion of broad-leaved forest to bamboo-dominated forest caused by bamboo expansion reduces the ecosystem’s storage of carbon (C) [13,32] and N [33,34] and thus has potential impacts on regional cycles of C and N [33,35]. If the expansion of moso bamboo continues without any control, it could threaten the functions and services of woodlands, especially in natural reserves. In order to estimate the expansibility of moso bamboo and the invasibility of recipient ecosystems, investigations on the factors encouraging the expansion of bamboo into adjacent forest are urgently demanded before taking effective control.

The literature survey on these factors intimately related to bamboo expansion suggested that the characteristics of forests have been regarded as the main inducing factors responsible for the successful expansion of bamboo into adjacent communities [17,36]. Okutomi et al. (1996) [11] found that moso bamboo tends to spread into forests of a simple structure and low species diversity. Suzuki and Nakagoshi (2008) reported that the expanding rate of bamboo is related to the slope and aspects of neighbor forests [19]. Meanwhile, it was also realized that natural disturbances (i.e., wind, snowstorms, fire) can accelerate the clonal expansion rate of bamboo [37,38]. In addition, several studies have suggested soil properties to be relevant factors affecting the expansion of bamboo. For example, Dong (2003) [39] observed that both the expanding range and the number of new culms are positively associated with the fertility and porosity of soil. Cai et al. (2003) [40] declared that abundant soil moisture hinders the growth of expanding rhizomes. So, it can be yielded that the expansion of bamboo is linked to both the characteristics of the aboveground stand and the underground soil in the bamboo–woodland interface (BWI), a transition zone lying between bamboo forests and woodlands. Unfortunately, few studies have simultaneously focused on these multiple factors encouraging the expansion of moso bamboo so far. Furthermore, only a few parameters (i.e., expansion distance and area) have been employed to quantify the range and intensity of expansion in previous studies [41,42], although bamboo expansion is a complicated process characterized by many parameters including the spread distance, the number of new culms and the culm size, etc., [43,44].

In this paper, we tried to establish the relationships between the multiple influencing factors and characteristics of bamboo expansion as basic information for the management of moso bamboo. Our objectives were to: (1) quantify the variations in bamboo expansion among different BWIs; (2) assess the effects of stand structure (featured by canopy closure, canopy height, point density of stand trees, species diversity, etc.) and soil physicochemical properties (including depth, water content, content of organic C, total N and P, etc.) on the variations in bamboo expansion; (3) disentangle the relative influences of stand characteristics and soil properties on bamboo expansion. These findings will be essential to achieve a better understanding of the mechanism of bamboo expansion into neighboring forests and thus help to make pertinent decisions on bamboo management, especially in natural reserves.

2. Materials and Methods

2.1. Study Site

This study was conducted in Jiangxi province, located at 113°34′–118°28′ E and 24°29′–30°04′ N. This province has an area of 16.7 × 10⁴ km², 60% of which is mountainous and covered by the Wuyi mountain range in the east, Luoxiao mountain range in the west, Nanling range in the south and Poyang Lake in the north (Figure 1). It has a humid subtropical climate with annual mean precipitation of 1278.2–2734.0 mm/yr, a mean temperature of 14–19.7 °C and a frost-free period of 240–307 d/yr. The typical vegetation types are evergreen broad-leaved forests, mainly composed of Fagaceae, Lauraceae and Theaceae, etc. Unfortunately, a large fraction of the original forest has been disturbed throughout history and then transformed to secondary evergreen broad-leaved forests, evergreen deciduous broad-leaved mixed forests, needle-leaved forests and bamboo forests [45]. The total area of these woodlands in Jiangxi province is about 1001.8 × 10⁴ hm², and bamboo forest covers ca 99.9 × 10⁴ hm². It accounts for about 10% of the total area of bamboo forests and ranks the second largest in China [46]. More importantly, the bamboo in this forest has been found to have been expanding widespread in lots of reserves in Jiangxi province [13,21,47,48].

2.2. Experimental Design and Soil Sampling

In order to quantify the effects of stand characteristics and soil properties on the expansion of bamboo, we set up 58 bamboo–woodland interfaces (BWIs) in 13 localities across Jiangxi province in the summer (June to August) of 2017 and 2018 (Figure 1). The characteristics for the sampling sites are shown in

Table 1. The basic information of 13 localities for bamboo–woodland interfaces.

Site	Altitude (m)	Mean Annual Rainfall (mm)	Mean Annual Temperature (°C)	Soil Type
Dagang	300	1600	17	Red soil
Dagangshan	550	1590.9	16.8	Yellow-red soil
Guanshan	400	2000	16.2	Red soil
Guifeng	55	1878	17.9	Red soil
Jiulingshan	930	1600	15.8	Yellow soil
Jinggangshan	800–975	1889.8	14.2	Yellow-red soil
Junfengshan	500–700	2019.8	17	Yellow soil
Jinpengshan	300–500	1600	19.5	Yellow soil
Lingyunshan	640–700	1706	16.2	Yellow-red soil
Mingyueshan	900	1606	17.2	Yellow-red soil
Tongboshan	400	1815	15.4	Yellow soil
Qiyunshan	400–1100	1750	17	Yellow-red soil
Wuyishan	900	2583	14	Yellow soil

. There were 5 BWIs for each site in DG, GS, MYS, JGS, QYS, WYS (see Figure 1), and 4 BWIs for each one in DGS, GF, JLS, JFS, JPS, LYS, TBS. These BWIs had similar topographies (southwest slope of approximately 6–10°) without performing any harvest activity. Each BWI covered an area of 20 m × 20 m, spanning bamboo forest (10 m) and recipient woodland (10 m). The woodlands were identified as 3 types including secondary evergreen forests (SEF, 22 in total), secondary evergreen deciduous broad-leaved mixed forests (EDF, 20 in total) and needle-leaved forests (NLF, 16 in total). In terms of disturbance, the expanded woodlands were further classified into disturbed forests (32) and non-disturbed ones (26). The expanded woodlands were defined as disturbance forests according to these characteristics: the canopy closure was 0.1~0.5 and the density of stand was 0~20 ind/100 m²; the plant species was mainly dominated by deciduous trees with a number ratio of deciduous tree to evergreen tree of more than 1; and the tops of trees were broken down by human activities or natural disasters. If the forests had these features, we considered them as disturbed forests. Regarding the expansion orientations, the woodlands of contiguous BWIs were divided into uphill forests (23), horizontal forests (15) and downhill forests (20). These diverse BWI plots were selected to ensure the covering of various stand characteristics and soil properties.

To understand the expansion process of moso bamboos, we measured their growth and reproduction in each BWI in terms of the number of parent culms (ind/100 m²) and the number, diameter and height of new culms. We calculated the expansion distance (ExD, m/yr) by determining the horizontal moving distance of new culms along the direction of bamboo forest to woodland. As the positions of new bamboo culms were random in the field, as shown in Figure 2, when the new bamboo was located at front of the nearest bamboo (such as new bamboo 1 and old bamboo 2) along the expansion direction, we calculated the horizontal distance between them. The new bamboo was located at the back of the nearest old culm (such as new bamboo 3 and old bamboo 2); in this case, we ignored the new bamboo to calculate the ExD. As the number of new culms in front of nearby old bamboos increased, we estimated ExD by averaging the horizontal distance between all of the new culms and the front of the old bamboos. The ExD is caculated as

$$ExD=\frac{{\sum }_i^n{d}_i}{n}$$

(1)

In addition, we determined the age of each living culm from the residual of its sheath and its external colour as well as the developing status of its branches and leaves [49,50] (Institute of Nanjing Forestry University, 1974; Yen, 2016). Considering the biennial cycle of the production of new shoots, a parent culm was considered to be 2 years older than its new culm.

The indicators of stand structure for recipient forests included the species identity, a diameter at breast height of 1.3 m DBH (cm), the tree height (m), the stand density (ind/100 m²) and the growth form (deciduous or evergreen) for each tree with either a height ≥ 5 m or a DBH ≥ 5 cm. Additionally, we also directly recorded the canopy closure (–) and canopy height (m) for each stand.

In each recipient forest, we dug 3 profiles to measure the average depth of soil and evaluate the bare-rock ratio (%). Meanwhile, in each profile, we sampled ca 30 g of soil at a depth of 0–30 cm using an aluminium box to measure the soil water content, and then we packed ca 500 g of soil in a cloth bag to determine the organic C of the soil (SOC, %), the total content of N (TN, %) and the total content of P (TP, %) in our laboratory. The SOC was measured by the oil bath K₂CrO₇ titration method [51]. The concentrations of TN and TP were determined by the indophenol-blue method [52] and the Mo-Sb colorimetry method [53], respectively.

2.3. Data Processing and Analysis

We calculated the expansion distance by averaging the distances of new culms to their nearest parent culms. We also used the number of parent culms to describe the potential expansibility of the bamboo forest. According to the measured data of expanded stands in field, we chose 6 variables to characterize the expanded woodland, including canopy closure, canopy height, stand density, the height ratio of moso bamboo to trees (BTR, –), the density ratio of deciduous tree to evergreen tree (DER, –) and the Shannon–Wiener index (SW, bit). SW is calculated as

$$SW=-{\displaystyle \sum }_{i=1}^s\frac{n_i}{N}\ln (\frac{n_i}{N})$$

(2)

where s is the total number of tree species, N is the number of trees individuals and n_i is the abundance of the ith species.

2.4. Statistical Analyses

In the first part, we used a descriptive analysis to quantify the variables of bamboo expansion, stand characteristics and soil properties through SPSS22. Regarding the non-quantified environmental variables, we applied the independent T-test to detect the differences in the variables of bamboo expansion and the characteristics of stand structures as well as the soil properties (significance level p = 0.05) between the disturbed forests and non-disturbed ones. Moreover, we employed the one-way ANOVA and Duncan’s post hoc test to check the differences in these characteristics among the various forest types and forest positions.

Furthermore, we investigated the relationships between the characteristics of bamboo expansion (response variables) and the attributes of stands and soil (explanatory variables) by a redundancy analysis (RDA) in CANOCO 4.5 [54]. Preliminarily, the response data were processed by a detrended correspondence analysis (DCA) to select the appropriate response model (linear or unimodal). Subsequently, the performed DCA gave a gradient length of 0.682, which was lower than 2 times the standard deviation. This meant that the expansion variables exhibited linear response to environmental gradients. Moreover, each data item (x) quantifying the bamboo expansion was log-transformed through [log(x + 1)] before RDA to prevent extreme values (outliers) from overly influencing the ordination [54]. Because any inflation factor higher than 5 has been identified as an indicator of collinearity in multivariate analysis [55], environmental variables with an inflation factor greater than 5 were removed in the preliminary analysis. During the RDA, the species scores were post transformed and divided by the standard deviation to standardize the ordination diagram for the data of bamboo expansion.

Finally, we quantitatively partitioned the influences of stand attributes and soil properties as well as their interaction on the variations in bamboo expansion by the partial RDA [56]. The multiple linear regression analysis with a forward selection procedure was employed to link the response variables and selected environmental variables by the RDA. The unrestricted permutation test of Monte Carlo (499 replicates) [54] was used to assess the statistical significance of the association between bamboo expansion and the attributes of stands and soil for expanded woodlands. In this way, we could not only quantify the relative contributions of stand attributes and soil properties to the variations in bamboo expansion but also detect the significant causal factors.

3. Results

3.1. Variations in Bamboo Expansion

The expansion characteristics of moso bamboo showed remarkable variations among the different bamboo–woodland interfaces. In

Table 2. The statistics for the variables of bamboo expansion, stand characteristic and soil properties at 58 bamboo–woodland interface plots in Jiangxi province, China.

Variables	Mean	Standard Deviation	Minimum	Maximum	Range	Coefficient of Variance
Bamboo expansion variables
ExD (m)	1.08	0.66	0.23	2.50	2.27	61.67
NNC (ind/100 m2)	5.10	3.31	1.20	16.38	15.18	64.90
DNC (cm)	8.22	1.91	4.72	12.65	7.93	23.21
HNC (m)	13.15	3.54	6.34	19.96	13.62	26.93
Stand characteristics
NPC (ind/100 m2)	12.05	5.44	3.94	27.45	23.51	45.11
CaC (–)	0.66	0.16	0.30	0.90	0.60	24.81
CaH (m)	16.46	5.91	6.50	28.00	21.50	35.90
StD (ind/100 m2)	16.71	7.90	3.00	31.00	28.00	47.28
BTR (–)	0.95	0.47	0.25	2.26	2.01	49.52
DER (–)	0.63	0.53	0.10	1.98	1.88	83.72
SW (bits)	2.53	0.87	0.94	4.08	3.14	34.32
Soil properties
SoD (cm)	52.33	23.19	30.00	120.00	90.00	44.32
SoW (%)	28.93	6.11	18.00	50.20	32.20	21.14
BR (%)	10.59	15.15	0.00	50.00	50.00	143.03
SOC (%)	3.00	0.70	1.12	4.67	3.55	23.37
TN (%)	0.29	0.08	0.10	0.45	0.35	26.25
TP (%)	0.02	0.01	0.01	0.05	0.04	31.42

The variables of bamboo expansion included ExD (mean expansion distance, m), NNC (number of new culms, ind/100 m²), DNC (diameter of new culms at 1.3 m, cm) and HNC (height of new culms, m). Stand variables included NPC (number of parent culms, ind/100 m²), CaC (canopy closure, –), CaH (canopy height, m), StD (stand point density, ind/100 m²), BTR (height ratio of bamboo to tree, –), DER (number ratio of deciduous tree to evergreen tree) and SW (Shannon–Wiener index, bit). Soil variables included SoD (soil depth, cm), SoW (soil water content, %), BR (bare-rock ratio, %), SOC (soil organic carbon content, %), TN (total nitrogen content, %) and TP (total phosphorus content, %).

, the distance of expansion varied from 0.23 m/yr to 2.50 m/yr, with a mean value of 1.08 m/yr. The mean, minimum and maximum values for the number of new culms were 5, 1 and 16 ind/100 m², respectively. The mean values for the diameter and height of new culms were 8.22 cm and 13.15 m, respectively. These four parameters of bamboo expansion, particularly the expansion distance and number of new culms, showed a significant dispersion, with coefficients of variation (CVs) reaching as much as 64.90%.

Considering the effects of disturbance, the distance of bamboo expansion into disturbed forest (1.50 m/yr) was significantly higher (p < 0.01) than that for non-disturbed forest (0.54 m/yr). The number, diameter and height of new culms behaved in similar manner (p < 0.01). Regarding the distance of expansion for the different expanded forests, it was 1.33 m/yr for EDF, which was significantly higher (p < 0.01) than that for SEF (0.82 m/yr) and NLF (1.08 m/yr) by 62.2% and 23.1%, respectively, while the number and diameter of new culms showed little difference among the three types of forests (p > 0.05) (

Table 3. Statistics of expansion variables of moso bamboo in the different disturbances, forest types and positions.

	Forests	ExD (m)	NNC (ind/100 m2)	DNC (cm)	HNC (m)
Disturbance	DTF (32)	1.51 ± 0.56 a	6.89 ± 3.41 a	8.7 ± 1.8 a	15.01 ± 2.16 a
	NDF (26)	0.54 ± 0.28 b	2.91 ± 1.22 b	7.6 ± 1.9 b	10.86 ± 3.60 b
Forest type	SEF (22)	0.82 ± 0.56 b	7.80 ± 1.76 a	12.1 ± 4.0 a	3.84 ± 2.48 b
	EDF (22)	1.33 ± 0.63 a	8.56 ± 1.65 a	14.5 ± 2.7 a	5.05 ± 3.01 a,b
	NLF (14)	1.08 ± 0.74 a,b	8.35 ± 2.45 a	12.7 ± 3.6 a	6.39 ± 4.25 a
Position	UHF (23)	1.34 ± 0.73 a	6.24 ± 3.89 a	8.0 ± 2.0 a	13.56 ± 3.47 a
	HZF (15)	1.08 ± 0.58 a,b	5.44 ± 2.78 a	8.8 ± 1.6 a	13.67 ± 3.42 a
	DHF (20)	0.77 ± 0.52 b	3.53 ± 2.33 a	8.2 ± 2.0 a	12.31 ± 3.73 b

Values are means ± standard errors. Numbers in brackets mean the number of interface plots. Different lowercase letters in the same column indicate significant variations in expansion variables under different disturbances, forest types and positions (p < 0.05). The abbreviations of expansion variables are as described in Table 1. DTF and NDF are abbreviations of the disturbed forests and non-disturbed forests. Three types of expanded forest included SEF (secondary evergreen broad-leaved forest), EDF (evergreen deciduous broad-leaved mixed forest) and NLF (needle-leaved forest). Three positions of recipient forests included UHF (uphill of bamboo forest), HZF (at the left or right of bamboo forest) and DHF (downhill of bamboo forest).

). When looking at the positions of the recipient forests, the distance of expansion and the number of new culms for uphill forests (1.34 m/yr, 6.24 ind/100 m²) were comparable to those of horizontal forests and higher than those of downhill forests (0.77 m/yr, 3.53 ind/100 m²), while the diameter and height of new culms were not affected by the positions of the expanded forests.

3.2. Variations in Stand and Soil Properties

Statistically, the indictors of stand properties among the 58 BWIs showed notable variations. The number of parent culms shifted from 4 ind/100 m² to 27 ind/100 m² with a mean value of 12 ind/100 m². The average value of canopy closure for the expanded forests was 0.66. Moreover, the stand characteristics also showed high CVs, ranging from 24.8% to 83.7%. Among the indicators of stand properties, DER varied by the most dramatic degree (Table 2).

In terms of disturbance, all the variables of stand properties except for the Shannon–Wiener index exhibited significant differences (

Table 4. Statistics of stand variables on bamboo–woodland interfaces in various disturbances, forest types and positions.

	Forests	NPC (ind/100 m2)	CaC (–)	CaH (m)	StD (ind/100 m2)	BTR (–)	DER (–)	SW (bit)
Disturbance	DTF (26)	14.09 ± 5.88 a	0.57 ± 0.16 b	14.12 ± 5.16 b	13.66 ± 7.49 b	1.19 ± 0.40 a	0.75 ± 0.60 a	2.59 ± 0.93 a
	NDF (32)	9.55 ± 3.57 b	0.76 ± 0.08 a	19.35 ± 5.56 a	20.46 ± 6.80 a	0.65 ± 0.37 b	0.47 ± 0.38 b	2.47 ± 0.81 a
Forest type	SEF (22)	9.81 ± 3.52 b	0.68 ± 0.17 a	18.00 ± 5.47 a	19.00 ± 6.89 a	0.80 ± 0.49 b	0.41 ± 0.22 b	2.50 ± 0.76 a
	EDF (22)	13.82 ± 5.19 a	0.62 ± 0.17 a	14.64 ± 5.87 a	12.82 ± 7.94 b	1.15 ± 0.44 a	0.79 ± 0.67 a	2.64 ± 0.88 a
	NLF (14)	12.80 ± 7.19 a,b	0.69 ± 0.13 a	16.89 ± 6.27 a	19.21 ± 7.32 a	0.87 ± 0.39 a,b	0.71 ± 0.53 a,b	2.43 ± 1.05 a
Position	UHF (23)	13.19 ± 5.47 a	0.61 ± 0.18 a	16.32 ± 5.62 a	15.00 ± 7.48 a	0.97 ± 0.45 a	0.51 ± 0.56 b	2.63 ± 0.79 a
	HZF (15)	12.98 ± 5.27 a	0.67 ± 0.17 a	17.41 ± 6.44 a	17.07 ± 8.54 a	0.93 ± 0.45 a	0.96 ± 0.61 a	2.38 ± 0.89 a
	DHF (20)	10.06 ± 5.21 a	0.71 ± 0.12 a	15.92 ± 6.05 a	18.40 ± 7.88 a	0.93 ± 0.53 a	0.51 ± 0.27 b	2.53 ± 0.97 a

Values are means ± standard errors. Numbers in brackets mean the number of bamboo–woodland interface plots. Different lowercase letters in the same column indicate significant variations in stand variables in different conditions (disturbance, forest types and positions) (p < 0.05). The abbreviations of stand variables are described as per Table 1, and those of forest types are described as per Table 2.

). The stand variables, including canopy closure, canopy height and stand density, for the disturbed forests were remarkably lower than those for the non-disturbed ones (p < 0.01). In contrast, another three variables, including the number of parent culms, BTR and DER, showed an opposite trend. The value of each of these three variables for the disturbed forests was higher than that for the non-disturbed ones.

From the viewpoint of the effects of forest type, canopy closure and canopy height, the Shannon–Wiener index varied a little among different forests, but the number of parent culms, BTR and DER for EDF was obviously higher (p < 0.01) than the corresponding value for SEF and was comparable to that for NLF. Instead, the stand density of EDF was lower than that for SEF and NLF. Furthermore, although DER attained its highest value in the horizontal forests, all the other stand variables were free from the positions of the expanded forests (Table 4).

Although the indictors of soil properties exhibited significant variations with considerable CVs higher than 100% (Table 2), none of the soil variables (soil depth, soil water, bare-rock ratio, SOC, TN and TP) were statistically affected by the disturbance, forest type and expansion position (p > 0.05) (

Table 5. Statistics of soil variables in bamboo–woodland interfaces in various disturbances, forest types and positions.

	Forests	SoD (cm)	SoW (%)	BR (%)	SOC (%)	TN (%)	TP (%)
Disturbance	DTF (26)	52.81 ± 24.39 a	27.77 ± 4.67 a	8.75 ± 15.50 a	3.14 ± 0.69 a	0.30 ± 0.08 a	0.02 ± 0.01 a
	NDF (32)	51.73 ± 22.09 a	30.35 ± 7.37 a	12.87 ± 14.69 a	2.83 ± 0.69 a	0.28 ± 0.08 a	0.02 ± 0.01 a
Forest type	SEF (22)	50.00 ± 24.64 a	28.29 ± 4.90 a	7.95 ± 11.51 a	2.93 ± 0.81 a	0.28 ± 0.10 a	0.02 ± 0.01 a
	EDF (22)	54.09 ± 25.94 a	29.25 ± 6.36 a	10.20 ± 14.97 a	3.03 ± 0.62 a	0.30 ± 0.06 a	0.02 ± 0.01 a
	NLF (14)	53.21 ± 16.60 a	29.41 ± 7.67 a	15.36 ± 19.85 a	3.06 ± 0.68 a	0.30 ± 0.07 a	0.02 ± 0.00 a
Position	UHF (23)	58.26 ± 26.09 a	27.76 ± 2.40 a	9.55 ± 15.72 a	3.13 ± 0.82 a	0.31 ± 0.08 a	0.02 ± 0.01 a
	HZF (15)	49.33 ± 23.06 a	28.96 ± 6.33 a	8.33 ± 11.13 a	2.96 ± 0.49 a	0.28 ± 0.07 a	0.02 ± 0.01 a
	DHF (20)	47.75 ± 19.02 a	30.25 ± 8.51 a	13.49 ± 17.25 a	2.89 ± 0.70 a	0.28 ± 0.08 a	0.02 ± 0.01 a

Values are means ± standard errors. Numbers in brackets mean the number of interface plots for each forest. Superscript letters in the same column indicate significant variations in soil variables under different conditions (disturbance, forest types and positions) at significant level p = 0.05. The abbreviations of soil variables are as described Table 1, and those of forests are shown in Table 2.

).

3.3. The Relationship between Expansion and Influence Variables

The ordination in Figure 3 quantified the relationships between the properties of bamboo expansion and the characteristics of stands and soil in BWIs. The permutation test of significance showed that AXIS 1 (F = 58.00, p = 0.002) and all canonical axes (F = 9.19, p = 0.002) were significant, clearly indicating the potential effects of the variables of stands and soil on the expansion of moso bamboo.

AXIS 1 explained the variability in bamboo expansion by 77.8%, while AXIS 2 only explained it by 17.4%. Moreover, AXIS 1 was remarkably and positively linked with the number of parent culms, BTR and DER, but it was negatively linked with canopy closure, canopy height and stand density. AXIS 2 was significantly and positively correlated with DER and soil water but negatively correlated with the number of parent culms (Figure 3 and

Table 6. The correlation coefficients of axes, seven stand factors and six soil factors at bamboo–woodland interface plots in Jiangxi province, China.

	AXIS 1	AXIS 2	NPC	CaC	CaH	StD	BTR	DER	SW	SoD	SoW	BR	SOC	TN	TP
AXIS 1	1
AXIS 2	0.14	1
NPC	0.60 **	−0.37 **	1
CaC	−0.73 **	−0.08	−0.14	1
CaH	−0.48 **	0.02	−0.23	0.53 **	1
StD	−0.63 **	−0.02	−0.24	0.69 **	0.72 **	1
BTR	0.60 **	0.14	0.21	0.66 **	−0.90 **	−0.75 **	1
DER	0.35 **	0.31 *	0.13	−0.02	−0.07	−0.18	0.1	1
SW	0.08	0.05	0.05	0.03	0.13	0.19	−0.09	0.05	1
SoD	0.16	0.01	0.12	−0.23	−0.07	−0.12	0.17	−0.12	−0.22	1
SoW	−0.29	0.38 **	−0.40 **	0.12	0.06	0.14	−0.12	0.07	−0.04	−0.08	1
BR	−0.03	0.04	0.03	0.11	−0.01	0.08	−0.1	0.28	0.08	0.03	0.1	1
SOC	0.09	0	0.02	−0.01	−0.05	−0.05	0.06	0.08	0.15	−0.02	0.01	0.08	1
TN	0.03	0.1	0.03	0.17	0.1	0.07	−0.11	0.18	0.14	−0.07	0.01	0.21	0.80 **	1
TP	−0.11	0.02	0.07	0.29	0.15	0.18	−0.18	0.01	0.16	−0.13	−0.05	0	0.52 **	0.61 **	1

The abbreviations for the variables of stands and soil are as described in Table 1. Level of significance are * p < 0.05; ** p < 0.01.

).

Overall, the lengths of the stand properties (number of parent culms, BTR, DER, canopy closure, canopy height and stand density) were longer than all the soil factors except for soil water (Figure 3). This obviously meant that the stand variables possessed more power than the soil characteristics in explaining the variations in bamboo expansion. Specifically, among the stand variables, canopy closure, canopy height and stand density were all negatively related with the indexes of bamboo expansion (see Figure 3). In contrast, the number of parent culms, BTR and DER were positively linked with bamboo expansion. As a property of soil, soil water imposed negative effects on bamboo expansion.

3.4. Explanatory Power for the Variables of Bamboo Expansion

Through the partial RDA, the stand characteristics together with soil properties explained the variations in the expansion characteristics by 73.1%. In detail, the stand features explained it by 58.8%, which was much higher than the power of the soil properties (4.3%) and their interactions (10%) (Figure 4). Meanwhile, the remaining 26.9% of the variations were left unexplained, indicating the existence of other undetected factors playing certain roles in the expansion of moso bamboo, perhaps those such as altitude, slope aspect, pH and other nutrients.

The principal factors affecting the characteristics of expansion were determined by forward selection, as listed in

Table 7. The marginal and conditional effects obtained from the forward selection for moso bamboo expansion.

Variables	Marginal Effects			Variables	Conditional Effects
	Explanatory Power	F	p		Explanatory Power	F	p
CaC	0.34	28.83	0.002	CaC	0.34	28.83	0.002
NPC	0.27	20.66	0.002	NPC	0.2	24.57	0.002
StD	0.25	18.26	0.002	DER	0.09	12.73	0.002
BT R	0.24	17.72	0.002	CaH	0.03	6.16	0.006
CaH	0.14	9.35	0.002	SoW	0.02	3.27	0.042
DER	0.11	6.8	0.002	BTR	0.02	2.27	0.114
SoW	0.09	5.6	0.008	TN	0.01	1.06	0.366
SoD	0.02	1.04	0.37	SOC	0.01	1.95	0.136
TP	0.01	0.43	0.714	SW	0.01	0.45	0.676
SOC	0.01	0.37	0.75	TP	0	0.58	0.636
SW	0	0.24	0.786	SoD	0	0.1	0.984
TN	0	0.22	0.83	StD	0	0.27	0.816

The abbreviations for the variables of stand and soil are as described in Table 1.

. When regarding the marginal effects (the independent effect of each variable), canopy closure ranked the first among all the factors. The number of parent culms, stand density and BTR took second place and were followed by canopy height, DER, etc. Since some variables operated simultaneously, such as canopy closure and stand density, number of parent culms and BTR, the explanatory power of stand density and BTR declined strikingly according to the conditional effects. Consequently, canopy closure and the number of parent culms were identified as the top two factors influencing the ordination of expansion characteristics. They together explained the variations in expansion by 54.0%. The rest of the factors only accounted for 18.0%.

3.5. Multiple Linear Regression Analysis

The multiple linear regression analysis of bamboo expansion on the variables of interface (number of parent culms, canopy closure, DER, canopy height and soil water) yielded the following Equations (2)–(5). It is obvious that canopy closure was significantly and negatively linked with all the expansion parameters. The influences of the number of parent culms on the expansion distance and the number of new culms were remarkably positive. DER was identified as a positive factor relating to the expansion distance, diameter and height of new culms. However, canopy height and soil water were not related to bamboo expansion.

$$ExD=1.791-2.521CaC+0.064NPC+0.256DER \left(R^2=0.909; F=85.43; P=0.00\right)$$

(3)

$$NNC=7.667-11.475CaC+0.412NPC \left(R^2=0.939;F=204.07;P=0.00\right)$$

(4)

$$DNC=10.378-4.886CaC+1.665DER (R^2=0.627;F=17.78;P=0.00)$$

(5)

$$HNC=18.919-11.018CaC+2.329DER (R^2=0.619;F=17.08;P=0.00)$$

(6)

where ExD is the mean expansion distance, NNC is the number of new culms, DNC is the diameter of new culms at 1.3 m, HNC is the height of new culms, CaC is canopy closure, NPC is number of parent culms and DER is number ratio of deciduous trees to evergreen trees.

4. Discussion

In recent years, bamboo expansion into neighboring woodlands has been considered as a worrying problem and has received much attention from ecologists [11,16,17,30]. Identifying habitats sensitive or immune to bamboo expansion and the factors encouraging expansion are important for forest management before adopting effective measures to prevent expansion. Disentangling these questions may provide unique insights into the fundamental ecological theory on the control of bamboo expansion.

4.1. The Characteristics of Moso Bamboo Expansion

Our study clearly pointed out quite significant differences in the characteristics of bamboo expansion among the different BWIs (Table 2), which were similar to the findings of Kobayashi et al. [36] in that there were large variations in the expansion rates among the different bamboo–woodland transects. The expansion distance (0.23~2.50 m/yr) in this study was well consistent with the results of Bai et al. [41], who reported bamboo expansion into adjacent forests at a rate of 0.57~1.80 m/yr. Okutomi et al. [11] also observed that the distances of bamboo expansion into shrubs, plantations and deciduous broad-leaved forests were 3.0 m/yr, 2.5 m/yr and 1.8 m/yr, respectively. In another aspect, the number of new culms for the different expanded forests varied from 1~16 ind/100 m², which exceeded the previously observed rates for SEF (4.37 ind/100 m²) and Chinese fir forests (5.33 ind/100 m²) [41]. The great variations in the expansion distance and the new number of culms in our study may be attributed to the fact that we investigated numerous BWIs plots with wide distribution ranges, and the measurement methods are different in different studies.

4.2. The Effects of Disturbance on Bamboo Expansion

From the viewpoint of disturbance, we found that not only was the expansion distance of moso bamboo into the disturbed forests significantly greater than that into the non-disturbed ones but also that the diameter and height of new culms in the disturbed forests were larger (Table 3). Our findings agreed with several reported observations, in that disturbance facilitated the expansion of bamboo-dominated forests [37,57,58,59,60]. For instance, Gagnon et al. [57] found that the expansion rate of Arundinaria gigantea increased nearly by 100% when neighboring forests were disturbed by a typhoon. Regarding forest types, the expansion distance into EDF was greater than that into SEF and NLE, which was consistent with previous findings by early explorers. Okutomi et al. [11] reported that the distance of moso bamboo expansion into deciduous broad-leaved forest was 1.8 m/yr in the southwest of Tokyo, which was obviously higher than the rates into evergreen broad-leaved forest (1.04 m/yr) and needle-leaved forest (1.28 m/yr) [41]. Considering the expansion orientation, the expansion toward uphill woodland (1.34 m/yr) was faster than that toward downhill (0.77 m/yr) (Table 3). That is to say that the expansion of moso bamboo exhibited considerable variations when facing different neighboring stands.

The stand features of adjacent woodlands varied greatly when suffering from enforced disturbance (Table 4). Firstly, the intensive disturbances from human and natural activities not only strongly converted dense and high canopies into sparse and short ones (with lower canopy closure, canopy height and stand density and a higher BTR) but also increased the proportion of deciduous trees in the stands (with a higher DER, see Figure 4), leading to an improved canopy opening. Since deciduous trees acquire more light, they are more vulnerable to the expansion of shading bamboos. Secondly, the number of parent culms in bamboo forests became greater under disturbance, indicating a higher possibility of bamboo invasion into disturbed forests. The higher rate of bamboo expansion into the uphill woodland than that of downhill woodland may be attributed to the fact that the forests on an upslope showed a lower canopy closure and stand density and comprised more relatively shorter deciduous trees (Table 4).

Considering ecological traits such as the great reproduction and quick spread of moso bamboo, which is similar to some invasive species, the fact that the severe expansion of bamboo into disturbed neighboring woodlands with a high speed, great quantity and large size was in well accordance with the empty niche hypothesis of invasion [61,62,63]. The hypothesis proposed that the increasing opportunity of invasion is attributed to empty niches (categorized as excess resources, i.e., water, light and nutrients) after the suffering of recipient ecosystems from disturbance. In a given community, the more niches that are available and unused by native species, the more likely the community will be invaded [64]. In the case of serious disturbance (i.e., cutting, girdling, pruning or snowstorms), trees will be damaged or even die, releasing their original niches and leaving relatively empty niches. Bamboo can arrive at those empty niches faster and settle down more successfully than tree species, since their new culms will asexually emerge every year and quickly grow in the form of sprouts [41]. This phenomenon may be defined as a disturbance effect, suggesting that when a forest suffers from disturbance, its ecological resistance against foreign plants will decline [65,66].

4.3. The Effects of the Characteristics of Stands and Soil on Bamboo Expansion

The characteristics of aboveground stands, especially canopy closure, were found to be more powerful in terms of explaining bamboo expansion than the properties of soil (Figure 4, Table 7). Moso bamboo is considered to be light-loving as well as shade-tolerant [67]. A large and heavy canopy of a stand would decrease the photosynthesis of bamboo despite the independent growth of new shoots in sunlight [35,36]. Consequently, similar to many other bamboos (Chusquea, Merostachys and Guadua genus) [68,69], moso bamboo in the field tends to bring up new culms at available sites such as forest gaps and sparse canopies, where the available light is enough to maintain its survival and growth. In turn, the living new culms can then help to support their parent culms through their joint rhizome system [17]. In the field, a few or even only one big evergreen tree with a close and wide crown (called a “tyrant” tree) can completely prevent bamboo shoots from sprouting and growing beneath their shades [17]. Moreover, it has been frequently suggested that foresters should cut off dominating trees to enhance the production of bamboo [70]. This management indirectly supports the negative effect of canopy closure on bamboo expansion. Nevertheless, some other factors of stands may be not negligible, such as dense stands leaving insufficient space for bamboo growth [70].

Meanwhile, our findings about the positive effects of the number of parent culms on expansion distance (Figure 3, Equations (2) and (3)) supported the propagule pressure hypothesis proposed by Lockwood et al. [71]. This hypothesis assumed that the probability of invasion success might increase with the propagule pressure. As the number of released individuals rises, the propagule pressure will also increase. Consequently, the increasing amount of bamboo would then inhibit the growth of other trees through casting a deep shade, slowly overcrowding them to eventually occupy the available growing space, thus augmenting the expansion range. Evidently, the features of adjacent woodlands enforced the profound effects on the success of bamboo expansion. It is worth pointing out that we mainly emphasized the associations between bamboo expansion and the influencing factors investigated herein. The mechanisms underlying these associations need further exploration in the near future.

Surprisingly, the relationships between the expansion of moso bamboo and all the soil properties except soil water were not significant. This does not mean the soil variables (soil depth, SOC, TN, TP, etc.) were not responsible for the expansion of bamboo. The insignificant soil properties among the different disturbances, forest types or positions possibly imposed little effect on the expansion of bamboo (Table 4). However, there was a notable exception in that soil water was found to be negatively linked with the expansion of bamboo. This may be attributed to the fact that excessive soil water is a disadvantage to the growth of the rhizome, which thus inhibits bamboo expansion [40]. Moreover, although bamboo is a nutrient-liking species favoring fertile habitats, it can still satisfy its nutrient demand in poor sites by adjusting itself in several other manners, such as by improving the accessibility of nutrients by its massive root system [72,73], raising the availability of nutrients through modifying their transformation processes [74] and increasing its nutrient-use efficiency through clonal physiological integration [75]. As a result, the soil properties were less important than the aboveground stand features for bamboo expansion.

5. Conclusions

According to our censuses, we illustrated the remarkably different expansion characteristics of moso bamboo in various bamboo–woodland interfaces (BWIs). Bamboo expansion into disturbed forests was of a significantly higher speed, greater abundance and larger size. Although many factors affected bamboo expansion into neighboring woodlands, the stand factors of BWIs, particularly canopy closure, DER and the number of parent culms, were revealed to be the dominating variables, whereas the properties of soil (except soil water) were found to contribute little to the success of bamboo expansion. Regardless of the forest type, moso bamboo preferred to expand into woodlands with a small and sparse crown and a low density. Three hypotheses, including the disturbance enhancement hypothesis, empty niche hypothesis and propagule pressure hypothesis, were proposed to explain the capability of moso bamboo to form monotypic stands. Therefore, in order to effectively control bamboo expansion into preserved woodlands, we should not only stop damaging the trees with particularly large crowns in the recipient ecosystem but also increase the harvesting of shoots and young bamboo to suppress the propagules of bamboo.