Effects of understory removal and tree girdling on soil microbial community composition and litter decomposition in two Eucalyptus plantations in South China

Authors

  • Jianping Wu,

    1. Key Laboratory of Vegetation Restoration and Management of Degraded Ecosystems, South China Botanical Garden, Chinese Academy of Sciences, Guangzhou 510650, China
    2. Graduate University of the Chinese Academy of Sciences, Beijing 100049, China
    Search for more papers by this author
    • The authors contributed equally to this work.

  • Zhanfeng Liu,

    1. Key Laboratory of Vegetation Restoration and Management of Degraded Ecosystems, South China Botanical Garden, Chinese Academy of Sciences, Guangzhou 510650, China
    Search for more papers by this author
    • The authors contributed equally to this work.

  • Xiaoling Wang,

    1. Key Laboratory of Vegetation Restoration and Management of Degraded Ecosystems, South China Botanical Garden, Chinese Academy of Sciences, Guangzhou 510650, China
    2. Graduate University of the Chinese Academy of Sciences, Beijing 100049, China
    Search for more papers by this author
  • Yuxin Sun,

    1. Graduate University of the Chinese Academy of Sciences, Beijing 100049, China
    2. State Key Laboratory of Organic Geochemistry, Guangzhou Institute of Geochemistry, Chinese Academy of Sciences, Guangzhou 510640, China
    Search for more papers by this author
  • Lixia Zhou,

    1. Key Laboratory of Vegetation Restoration and Management of Degraded Ecosystems, South China Botanical Garden, Chinese Academy of Sciences, Guangzhou 510650, China
    Search for more papers by this author
  • Yongbiao Lin,

    1. Key Laboratory of Vegetation Restoration and Management of Degraded Ecosystems, South China Botanical Garden, Chinese Academy of Sciences, Guangzhou 510650, China
    Search for more papers by this author
  • Shenglei Fu

    Corresponding author
    1. Key Laboratory of Vegetation Restoration and Management of Degraded Ecosystems, South China Botanical Garden, Chinese Academy of Sciences, Guangzhou 510650, China
    Search for more papers by this author

Correspondence author. E-mail: sfu@scbg.ac.cn

Summary

1. Soil micro-organisms play important roles in ecosystems and respond quickly to environmental changes. We examined how understory removal and tree girdling influence the composition of soil microbial community and the litter decomposition in two subtropical plantations.

2. Phospholipid fatty acids (PLFAs) analysis was used to characterize soil microbial community. Redundancy analysis and principal response curves (PRC) were used to investigate the relationships between soil microbial community and environmental factors.

3. Understory removal significantly reduced the amount of fungal PLFAs, the ratio of fungal to bacterial PLFAs, and the litter decomposition but did not affect bacterial PLFAs and total PLFAs. In contrast, tree girdling did not affect the soil microbial characteristics. The changes in soil microbial community caused by understory removal were mainly attributed to the indirect effects such as increased soil temperature and soil NO3-N availability. In addition, PRC analysis showed that the relative abundance of most PLFAs increased in response to understory removal in the 2-year-old plantation but decreased in the 24-year-old plantation.

4. We propose that understory plants are important components in subtropical forest ecosystems, and play different roles in maintaining soil microbial community and driving litter decomposition processes in young vs. old plantations. The functions of understory plants should be considered in forest management and restoration. The negligible effect of tree girdling on the soil micro-organisms can be attributed to the resprouting trait and mycorrhizal interactions of Eucalyptus.

Introduction

A key topic of ecological research involves predicting how human-induced losses of species and functional groups from real ecosystems influence ecosystem properties and functioning (Hooper et al. 2005; Wardle & Zackrisson 2005). There is a growing recognition that plant functional groups play an important role in driving ecosystem processes and functioning (Bardgett & Wardle 2010). When particular plant functional groups are deliberately removed from the ecosystem, it will cause apparent consequences on the ecosystem processes and functioning (Wardle et al. 2008). Since above-ground and below-ground communities are intimately linked and these linkages greatly affect ecosystem properties (Wardle et al. 2004; Bardgett et al. 2005), removal of plant functional groups from above-ground communities will exert a great impact on below-ground communities. However, the soil microbial community is often ignored in studies examining the effects of plant species and functional group losses on ecosystem functioning (Marshall, McLaren & Turkington 2011). Soil microbes play key roles in ecosystems and mediate many ecological processes that are central to ecosystem functioning, including nutrient cycling (Zeller et al. 2008), decomposition processes (Subke et al. 2004; Carney et al. 2007) and the regulation and maintenance of plant biodiversity (Zak et al. 2003; Wardle et al. 2004; Van Der Heijden, Bardgett & Van Straalen 2008). Further, biotic and environmental factors drive the activity, structure and diversity of soil microbial communities, which are controlled by many factors including plant species (Wardle et al. 1999; Hooper et al. 2005) and edaphic conditions (Bååth & Anderson 2003; Schimel, Balser & Wallenstein 2007, Fierer et al. 2009). Information concerning the influences of plant functional group losses on soil microbial community dynamics is valuable for both theoretical and management perspectives (Díaz et al. 2003).

As an effective research approach, tree girdling experiments have been widely used to examine how plant carbon allocation influences soil properties and processes in forest ecosystems (Högberg et al. 2001; Binkley et al. 2006). Tree girdling makes it possible to assess the effect of tree-root loss on soil microbial communities. Although it was shown that tree girdling significantly decreases plant below-ground carbon allocation and alters the availability and the quality of organic carbon sources for microbes (Subke et al. 2004; Högberg, Högberg & Myrold 2007), how tree girdling affects soil microbial communities is still unclear. It was reported that tree girdling generally caused a significant decline in the activity and biomass of soil microbes in boreal and temperate forests (Scott-Denton, Rosenstiel & Monson 2006, Weintraub et al. 2007), which was mainly due to the loss of ectomycorrhizal fungi (Högberg & Högberg 2002; Yarwood, Myrold & Högberg 2009; Pena et al. 2010). However, in several other studies, microbial activity and microbial biomass did not decrease or change immediately after girdling because of the specific resprouting trait (Chen et al. 2010), the availability of carbohydrates stored in root systems (Binkley et al. 2006), and the positive priming effect on SOM decomposition (Subke et al. 2004; Scott-Denton, Rosenstiel & Monson 2006).

In forest ecosystems, understory vegetation plays an important role in driving the processes and functions by affecting both the above-ground processes such as tree-seedling regeneration, forest succession, species diversity and stand productivity and the below-ground processes including decomposition, soil nutrient cycling and soil water conservation (Yarie 1980; Nilsson & Wardle 2005). Although it is of ecological importance, understory vegetation has usually been overlooked in forestry ecological studies. Removal experiments are especially useful for understanding the effects of loss of dominant species and functional group in forest ecosystems (Bardgett & Wardle 2010). In general, understory vegetation removal can facilitate the establishment and growth of tree seedlings (Hulme & Bremner 2006; Wang et al. 2009) and cause significant changes in soil microclimate (Matsushima & Chang 2007) and nutrient availability (Bret-Harte et al. 2004; Matsushima & Chang 2007). However, little is known about how understory vegetation removal affects the composition and function of soil microbial communities in forest ecosystems, especially for subtropical or tropical forests. Some removal experiments in forest ecosystems showed that understory vegetation removal could significantly reduce soil microbial biomass carbon (Xiong et al. 2008) and alter the composition of soil microbial communities (Ohtonen, Munson & Brand 1992), while some other studies demonstrated that understory removal had no significant impact on soil microbial properties (Urcelay et al. 2009). It was also pointed out that the effect of plant species and functional group removal on ecosystem properties was highly context-dependent and strongly influenced by soil-nutrient availability, temporal factors and spatial scale (Wardle & Zackrisson 2005).

Eucalyptus has been widely planted for reforestation in South China. By year 2010, the total area of Eucalyptus plantations in South China was about 2·6 million hectares (data from China Eucalyptus Research Centre, http://www.chinaeuc.com/show.asp?id=443). As a common practice in forest management, understory vegetation is usually removed from forest floor to prevent fire and to promote the growth and regeneration of tree-species seedling (Camprodon & Brotons 2006). However, little is known about how the loss of these plant functional groups affects the composition and function of the soil microbial communities. In this study, tree-girdling experiments and removal experiments were used to examine the effects of plant function group loss (tree girdling, understory vegetation removal and tree girdling plus understory removal) on the composition and function of the soil microbial communities in two Eucalyptus plantations of different ages (2 and 24-year-old). Our study also allowed assessment of the relative contribution of tree species and understory vegetation for the maintenance of soil microbial community composition and its decomposition function. We hypothesized that (i) plant functional group loss (tree girdling and understory removal) would affect the composition of soil microbial communities by altering the below-ground C input and soil properties; (ii) plant function group loss would also affect the litter decomposition by changing the composition of soil microbial communities; (iii) the effects of plant functional group loss on soil microbial community composition and its decomposition function would vary with the plantation age.

Materials and methods

Study site

The experiment was conducted at the Heshan National Field Research Station of Forest Ecosystem (112°50′ E, 22°34′ N), located in a subtropical hilly land region of South China. The climate in this region is subtropical monsoon. The mean annual precipitation is 1534 mm and the mean annual temperature is 22·5 °C from 2004 to 2009. The soil is an Acrisol (FAO 2006). Both experimental plantations used in this study were established on homogenous degraded hilly land. The older plantation was established in 1984 and the younger was established in 2006 (Fig. 1). In both plantations, the Eucalyptus saplings were planted with a spacing of 3 × 2 m. The most abundant understory species in both plantations is Dicranopteris dichotoma; other understory species include Rhodomyrtus tomentosa, Baeckea frutescens, Dianella ensifolia, Wikstroemia indica and Blechnum orientale. High-indigenous tree species are rare because of the seed resources and light limitation, as well as the influence of understory layer on the survival and growth of the indigenous seedlings in the study site (Wang et al. 2009). In March 2009, the mean understory biomass was 772 ± 92 g dry wt m−2 in the 2-year-old plantation and 2116 ± 61 g dry wt m−2 in the 24-year-old plantation. In April 2008, the mean basal diameter and d.b.h. of Eucalyptus were 6·8 ± 1·0 and 4·7 ± 0·6 cm in the 2-year-old plantation and 14·3 ± 0·3 and 10·2 ± 0·2 cm in the 24-year-old plantation. In both plantations, the vegetation was considered to consist of two functional groups: the Eucalyptus canopy and the D. dichotoma-dominated understory.

Figure 1.

 Photographs of 2-year-old (a) and 24-year-old (b) Eucalyptus plantations and associated understory vegetation in the Heshan National Field Research Station of Forest Ecosystem, Guangdong province, China. Photo credit: Zhanfeng Liu.

Experimental design

From December 2007 to January 2008, we established three experimental plots (10 × 10 m) in each plantation. Each plot was divided into four subplots; each of the four subplots corresponded to one treatment. The randomized block design had two levels for each of the two factors (±girdling and ±understory removal) to give four treatment combinations, which were ‘no girdling and no understory removal (C or control), girdling but no understory removal (G), no girdling but understory removal (UR) and girdling plus understory removal (GUR)’. Each subplot contained an average of six trees. A 40-cm deep trench was formed around each subplot to eliminate intrusion of roots from the other plots. On 27 March and 28 March 2008, we physically removed all understory species in the understory removal plots and girdled the trees (by cutting 10-cm bands around the stem at 50 cm above-ground) in the girdling plots. Because Eucalyptus can resprout after girdling and Dicranopteris can grow from remnant roots, new Eucalyptus and Dicranopteris growth (and growth of any other understory plant) were removed once per month in the girdling and removal plots.

Sampling collection and analyses

Litter bags were used to measure decomposition rate. We collected fresh Eucalyptus litter in September–October 2008 in litter traps (1 × 1 m nylon net) on the experimental plots. The fresh litter was dried and added to nylon litter bags (15 × 15 cm, 10 g dried leaf litter per bag). The mesh size was 1 mm2. Litterbags were placed on the surface in each subplot (six bags per subplot) of the four treatments in November 2008. For girdling plus understory removal treatment, the litterbags were placed in January 2009, because we did not collect enough fresh litter the first time. Litter bags (one per subplot) were collected every 2 months for 1 year between January 2009 and November 2009, three litter bags in total were collected for each treatment at one time. In the laboratory, adhering soil particles were removed from the litter by rinsing with deionized water. The litter was then dried to constant mass at 75 °C.

The mass of fine roots (diameter <2 mm) was determined by collecting five cores in the subplots of girdling plus understory removal and in the nearby region outside of the plot. The cores were taken to 20-cm depth with an 8-cm-diameter corer in March 2009. In the laboratory, fine roots were removed from the soil, cleaned by rinsing with deionized water and dried to constant mass at 75 °C.

Before the experiment was started, three soil cores (3-cm diameter, 20-cm depth) were collected from each subplot. Plant litter was removed from the soil surface before the cores were taken. The three cores were combined to form one sample per subplot. These samples were used for the determination of soil physico-chemical characteristics before treatment application. After treatment application in late March 2008, the soil was sampled again on 13 August 2008, 16 December 2008, 21 March 2009 and 16 August 2009. Samples collected after treatment application were divided into half; one part was used for determination of soil physico-chemical characteristics and the other part was used for PLFAs analysis. For determination of physico-chemical characteristics, soil was air dried and passed through a 2-mm sieve; remaining roots and stones were removed by hand. Soil to be used for PLFAs analysis was stored at −20 °C.

Soil samples collected for soil physico-chemical analyses were ground to pass through a 0·25-mm sieve. Total nitrogen (TN) concentration was measured after micro-Kjeldahl digestion using a flow injection autoanalyser (FIA, Lachat Instruments, USA). Soil organic carbon (SOC) was measured with the Walkley-Black method (Liu 1996). Soil C/N values were calculated as the ratio of SOC to TN. Soil pH was determined using a 1 : 2·5(wt/vol) ratio of soil to deionized water. Soil moisture content (SMC, g of water per 100 g dry soil) was examined gravimetrically by drying fresh soil at 105 °C to constant weight. Dissolved organic carbon (DOC) in filtered 0·5 m K2SO4-extracts of fresh soil sample was measured with a TOC analyser (TOC-VCPH Shimadzu Corp., Japan). NH4+-N and NO3-N in filtered 2 m KCL-extracts of fresh soil sample were measured with a flow injection autoanalyser (FIA, Lachat Instruments, USA). Soil temperature at 5-cm depth was measured at sampling time by a soil temperature probe (LI-COR Biosciences, Lincoln, NE, USA).

The soil microbial community was characterized using phospholipid fatty acids (PLFAs) analysis as described by Bossio & Scow (1998). For each sample, different PLFAs were considered to be the representative of different groups of soil micro-organisms. The abundance of individual fatty acids was determined as relative nmol per g of dry soil and standard nomenclature was used (Tunlid et al. 1989). Bacteria were considered to be represented by 10 PLFAs (i15:0, a15:0, 15:0, i16:0, 16: 1ω7, i17:0, a17:0, 17:0, cy17:0, cy19:0) and fungi were considered to be represented by the PLFAs 18:2ω6 (Frostegård & Bååth 1996; Bossio & Scow 1998; also see Table S1 Supporting information). Other PLFAs such as 16:1ω9c, 16:0, 17:1ω8c, 18:1ω9c and 18:3ω3c were also used to analyse the composition of microbial community. The ratio of 18:2ω6 to total bacterial PLFAs was used to estimate the ratio of fungal to bacterial biomass (F : B) in soils (Bardgett, Hobbs & Frostegård 1996; Frostegård & Bååth 1996). Taken together, all of the PLFAs indicated above were considered to be representative of the total PLFAs of soil microbial community.

Data analyses

We used one-way anova to analyse the soil physico-chemical characteristics. We used three-way anova to test for the effects of sample time, understory removal, tree girdling effects and their interactions on soil microbial characteristics and environmental factors. These statistical analyses were carried out with spss 15 (SPSS, Inc, Chicago, IL). Difference is significant at the 0·05 level.

We performed redundancy analysis (RDA) to determine which environmental factors were related to the composition of soil microbial community. The most discriminating soil variables were selected by ‘forward selection’ procedure of the programme. The principal response curves (PRC) method was used to determine the temporal trends of community composition for each of treatment. The PRC method is based on partial RDA. The PRC results show several sets of response curves with different line types. Control treatment is usually taken as reference line, which overlaps the horizontal axis line. Combined with the species weights, the PRC can be used to interpret the effect of treatments relative to the control based on the abundance of individual species (Van Den Brink & Ter Braak 1998, 1999; Lepš & Šmilauer 2003). The single PLFAs can be calculated by equation: the abundance of a species relative to the control = EXP (value in curve × value of species score on the first RDA axis). If the relative abundance is smaller than 1, it indicated the treatment decreased the abundance of the species or vice versa. For example, the abundance of PLFAs 18:2ω6 after 1 year in the girdling plus understory removal subplot in the 2-year-old plantation was EXP (−0·19 × 2·0) = 0·68 (Fig. 7a), indicating that the relative abundance of 18:2ω6 in this treatment was 68% of its relative abundance in the control. Statistical significance tests for RDA and PRC were run using canoco software for Windows 4·5 (Ithaca, NY, USA). Forward selection was based on Monte Carlo permutation (n = 499).

Figure 7.

 Principal response curves with species weights based on the PLFAs biomarker data in the 2-year-old plantation (a) and in the 24-year-old plantation (b). The proportion of variability explained by the first response curves was 6·8% (F = 4·063, P = 0·016) in the 2-year-old plantation, and 5·0% (F = 1·674, P = 0·498) in the 24-year-old plantation. The dotted line represents the girdling treatment, the dashed line represents the understory removal treatment, the solid line represents the girdling plus understory removal treatment, and the horizontal axis represents the control treatment. The vertical axis on the right shows the PLFAs biomarker weights.

Results

Effect of treatments on soil variables

Before treatment application and within each plantation, soil physico-chemical characteristics did not significantly differ among the treatment subplots (Table 1). DOC and pH were higher in the 2-year-old plantation than in the 24-year plantation, but TN was lower in the 2-year-old plantation (Table 1). Girdling did not change the soil environment except for NO3-N concentration in the 2-year-old plantation. From the mean values of four sampling times, however, girdling increased SMC by 6·1% and DOC by 17·6% in the 24-year-old plantation. Understory removal increased soil temperature (ST) by 1·24 °C in the 2-year-old plantation and by 0·93 °C in the 24-year-old plantation. The NO3-N concentration increased significantly in both plantations under the treatment of understory removal. There were no interactions between girdling and understory removal on soil environments except for NO3-N concentration in the 2-year-old plantation. Most of variables were significantly affected by sampling time (Table 2).

Table 1.   Soil characteristics of the two plantations before treatments were applied
Plantation age (years)TreatmentSOC (g kg−1)TN (g kg−1)DOC (μg C g−1)pH
  1. SOC, soil organic carbon; TN, total soil nitrogen; DOC, dissolved organic carbon; n = 3. Control, no tree girdling and no understory removal; G, tree girdling; UR, understory removal; GUR, tree girdling plus understory removal.

  2. Values are means (with 1SE in parentheses). Statistical significance was determined at P < 0·05, ns, no significant difference between treatments using anova.

2Control13·69 (0·82)0·70 (0·09)246·54 (46·55)3·86 (0·03)
G14·86 (1·54)0·87 (0·02)281·61 (42·96)3·84 (0·04)
UR14·04 (1·73)0·78 (0·04)264·30 (48·76)3·89 (0·02)
GUR15·21 (1·87)0·82 (0·06)299·54 (32,01)3·90 (0·07)
anovansnsnsns
24Control16·19 (1·37)1·41 (0·11)241·99 (15·72)3·80 (0·06)
G14·13 (0·85)1·87 (0·30)231·29 (13·56)3·76 (0·04)
UR14·02 (1·30)1·38 (0·03)208·39 (11·87)3·78 (0·06)
GUR13·84 (1·21)1·54 (0·09)242·37 (12·48)3·69 (0·05)
anovansnsnsns
Table 2.   Effects of time, tree girdling, understory removal and two-way interactions of tree girdling and understory removal on soil moisture content, soil temperature at 5-cm depth, and dissolved organic carbon in two plantations, n = 3
Plantation age (years)FactorsTimeGURG × UR
FPFPFPFP
  1. Time, sampling time; G, tree girdling; UR, understory removal; G × UR, interactions between tree girdling and understory removal; SMC, soil moisture content; ST, soil temperature at 5 cm depth; DOC, dissolved organic carbon.

  2. Results are from three-way factorial anova. The factors used for the anova were time (levels: August 2008, December 2008, March 2009 and August 2009), girdling (levels: girdled, not girdled), and understory removal (levels: understory removed, understory not removed).

2SMC9·46<0·0010·020·903·420·070·040·84
ST492·6<0·0010·100·7527·73<0·0010·090·76
DOC1·890·150·000·990·050·820·210·65
NH4+-N119·8<0·0010·330·573·550·0690·690·41
NO3-N0·260·857·250·0110·490·0035·250·03
24SMC62·39<0·0015·360·030·080·780·010·94
ST418·2<0·0010·810·3711·57<0·0020·010·93
DOC15·86<0·0019·000·012·350·140·140·71
NH4+-N144·14<0·0011·910·180·230·630·470·50
NO3-N10·37<0·0010·0020·9610·950·0020·060·81

One year after treatment, girdling plus understory removal reduced fine root biomass in both plantations, but not to statistical significance (Fig. 2). Relative to litter mass loss in the girdling and control treatments, litter mass loss was significantly reduced when understory plants were removed in the 2-year-old plantation (= 0·02) (Fig. 3a). Litter mass loss was also reduced by the removal of understory plants but not to statistical significance in the 24-year-old plantation (= 0·059) (Fig. 3b). Girdling did not significantly change the litter decomposition in either plantation (Fig. 3).

Figure 2.

 Fine root mass in ‘girdling plus understory removal’ subplots and control subplots in March 2009 (1 year after the girdling was started) in the 2 and 24-year-old plantations. Values are means ± 1SE. Within each group of four bars, values with a different letter are significantly different (< 0·05).

Figure 3.

 Mass of leaf litter remaining per litter bag in the 2 and 24-year-old plantations. Vertical bars represent SEs. UR, understory removal; G, girdling. The inserted P-values were from repeated measures anova.

Temporal and spatial dynamics of soil micro-organisms

According to the three-way factorial anova, sampling time significantly affected total PLFAs, bacterial PLFAs and fungal PLFAs in both plantations (Table 3). In the 2-year-old plantation, the quantity of bacterial PLFAs was higher in December 2008 than at other sampling times (Fig. 4a). A similar pattern was found for total PLFAs (Fig. 4c). Fungal PLFAs decreased significantly from December 2008 to March 2009 in control subplots (= 0·025). The F : B ratio was higher in August 2009 than at other sampling times (Fig. 4d). In the 24-year-old plantation, bacterial PLFAs were higher in December 2008 and March 2009 than in August 2008 and August 2009 (Fig. 5a). The total PLFAs had the similar pattern to the bacterial PLFAs (Fig. 5c). F : B was highest in August 2009 (Fig. 5d).

Table 3.   Effects of time, tree girdling, understory removal and two-way interactions of tree girdling and understory removal on soil microbial PLFAs, n = 3
Plantation age (years)FactorsTimeGURG × UR
FPFPFPFP
  1. Time, sampling time; G, tree girdling; UR, understory removal; G × UR, interactions between tree girdling and understory removal. F : B indicates the ratio of fungal to bacterial PLFAs.

  2. Results are from three-way factorial anova for the soil microorganism variables. The factors used for the anova were time (levels: August 2008, December 2008, March 2009 and August 2009), girdling (levels: girdled, not girdled) and understory removal (levels: understory removed, understory not removed).

2Total PLFAs7·71<0·0010·600·441·080·300·280·60
Bacterial PLFAs8·30<0·0010·480·490·290·590·110·74
Fungal PLFAs7·41<0·0011·630·2112·16<0·0010·020·89
F : B2·270·0951·410·2418·02<0·0010·0050·94
24Total PLFAs32·50<0·0010·460·500·300·591·750·19
Bacterial PLFAs12·04<0·0012·000·160·230·641·720·20
Fungal PLFAs4·82<0·0061·030·3211·15<0·0022·400·13
F : B11·33<0·0010·010·9924·62<0·0010·360·55
Figure 4.

 Soil microbial PLFAs in the 2-year-old plantation. Control, no tree girdling and no understory removal; G, tree girdling; UR, understory removal; GUR, tree girdling and understory removal. F : B indicates the ratio of fungal to bacterial PLFAs. Values are means ± 1SE, n = 3. Within each group of four bars, values with a different letter are significantly different (< 0·05).

Figure 5.

 Soil microbial PLFAs in the 24-year-old plantation. Control, no tree girdling and no understory removal; G, tree girdling; UR, understory removal; GUR, tree girdling and understory removal. F : B indicates the ratio of fungal to bacterial PLFAs. Values are means ± 1SE, n = 3. Within each group of four bars, values with a different letter are significantly different (< 0·05).

Fungal PLFAs and the F : B ratios were significantly higher in the 2-year-old plantation than in the 24-year-old plantation for all treatments (Table 4). In contrast, bacterial PLFAs and total PLFAs did not differ between the two plantations (Table 4).

Table 4. F- and P- values derived from one-way anova comparing soil microbial characteristics as affected by tree girdling and understory removal in the 2 and 24-year-old plantation
VariableControlGURGUR
FPFPFPFP
  1. Control, no tree girdling and no understory removal; G, tree girdling; UR, understory removal; GUR, tree girdling plus understory removal. F : B indicates the ratio of fungal to bacterial PLFAs.

Total PLFAs0·530·482·590·120·670·420·030·88
Bacterial PLFAs0·220·652·720·111·230·281·740·20
Fungal PLFAs9·45<0·0069·48<0·00511·28<0·0039·37<0·006
F : B8·48<0·0087·760·0110·22<0·0044·500·045

Effect of treatments on soil microbial community

Based on one-way anova, total PLFAs and bacterial PLFAs were unaffected by girdling, understory removal or their combination in the 2-year-old plantation (Fig. 4a and c) and in the 24-year-old plantation (Fig. 5a and c). Fungal PLFAs and the F : B ratio, however, decreased significantly at several sampling times after treatments (Fig. 4 and 5). The three-way anova showed that understory removal significantly reduced fungal PLFAs and the F : B ratio in both plantations (Table 3). From the mean values of four samplings, understory removal reduced fungal PLFAs by 30% ± 0·12 in the 2-year-old plantation and by 43% ± 0·08 in the 24-year-old plantation, and reduced the F : B ratio by 34% ± 0·07 in the 2-year-old plantation and by 39% ± 0·09 in the 24-year-old plantation. The interaction between girdling and understory removal was not significant (Table 3).

Redundancy analysis showed that the composition of soil microbial community in the 2-year-old plantation was significantly related to DOC, NH4+-N, ST and NO3-N; together, all of the environmental data explained 29·1% of the variance, with axis 1 explaining 14·4% of the variance and axis 2 explaining another 9·4% (Fig. 6a). In the 24-year-old plantation, the composition of soil microbial community was significantly related to NH4+-N, SMC and ST; together, all of environmental data explained 33·3% of the variance, with axis 1 explaining 25·6% of the variance and axis 2 explaining another 4·7% (Fig. 6b).

Figure 6.

 Redundancy analysis of soil microbial PLFAs in the 2-year-old plantation (a) and in the 24-year-old plantation (b). Ordination diagrams presenting species scores and environmental factor scores (vectors). DOC, dissolved organic carbon; ST, soil temperature at 5-cm depth; SMC, soil moisture content.

In the partial RDA with canoco, the first canonical axis was significant in the 2-year-old plantation (F = 4·063, = 0·016) but not in the 24-year-old plantation (F = 1·674, = 0·498) (Fig. 7). The PRC analysis showed that the eigenvalues of first canonical axis were 6·8% in the 2-year-old plantation and 5·0% in the 24-year-old plantation. PRC with species weights showed that understory removal and girdling plus understory removal changed the soil microbial community more than girdling alone. Generally, the relative abundance of 13 out of 17 major soil PLFAs increased in response to the understory removal treatment in the 2-year-old plantation (Fig. 7a). However, the trend was nearly the opposite in the 24-year-old plantation. The relative abundance of 13 out of 17 major soil PLFAs decreased in response to the understory removal treatment in the 24-year-old plantation (Fig. 7b).

Discussion

Understory removal effects

In the present study, understory removal decreased fungal PLFAs and the F : B ratio in both Eucalyptus plantations. It was reported that the removal of understory vegetation directly resulted in a reduction of net primary production (NPP) and induced a ‘bottom-up limitation’ on soil microbial biomass and microbial activity due to the loss of plant biomass and the reduced input of labile C (Wardle et al. 1999; Wardle & Zackrisson 2005). However, the DOC remained unaffected by understory removal 1 year after treatment in our study, so the ‘bottom-up limitation’ on soil microbes seemed unlikely in this case. There must be some other reasons causing the reduced fungal PLFAs and the F : B ratios.

The indirect effects of understory removal on soil microclimate and nutrient availability have been well-acknowledged (Wardle & Zackrisson 2005; Liu et al. 2010; Marshall, McLaren & Turkington 2011). RDA showed that all of the environmental variables explained 29·1–33·3% of the variance of soil microbial community. In the present study, understory removal significantly increased soil temperature and NO3-N supply and subsequently suppressed the fungal biomass. Our result was consistent with the findings by Nylund (1988) and Cox et al. (2010), they found that the increase of N availability reduced fungal biomass and diversity, and caused shifts in the microbial community composition.

In addition, our results showed that understory removal reduced the rate of litter decomposition. We considered that priming effect was not major in this case because both soil microbial biomass (i.e. total PLFAs) and soil C did not change significantly. In contrast, we considered that the decrease of litter decomposition rate under treatment of understory removal was mainly caused by a shift of soil microbial community composition. In fact, the fungal PLFAs and F : B decreased greatly under the treatment of understory removal. Wardle & Zackrisson’s (2005) study also showed that removal of plants negatively affected the decomposition function of soil microflora.

Tree girdling effects

Tree girdling did not significantly affect the soil microbial characteristics in either plantation. Girdling, which stops the flow of current photosynthates from leaves to roots and soil was expected to affect the composition and decomposition function of soil microbial communities. Previous study in boreal forests reported that tree girdling caused a 45% reduction in the fungal biomarker mainly due to a reduction in ectomycorrhizal fungi (Högberg, Högberg & Myrold 2007). In the present study, tree girdling did not significantly decrease the fungal PLFAs 18:2ω6 and F : B in either plantation. Similarly, another girdling study also reported the small effects of girdling on soil microbial biomass in Eucalyptus plantations (Chen et al. 2010). It was also reported that the response of bacterial abundance and biomass was marginal to tree girdling in boreal forests (Högberg, Högberg & Myrold 2007; Yarwood, Myrold & Högberg 2009) and subtropical evergreen broad-leaved forests (Li et al. 2009).

The small effect of girdling on soil micro-organisms may be explained by the regeneration of fine roots and the continuous supply of carbohydrates to soil from starch reserve of roots. One year after ‘girdling plus understory removal’ treatment, we found that 51% and 62% of the fine roots remained alive in the 2-year-old plantation and the 24-year-old plantation respectively. It was reported that Eucalyptus species were able to resprout from below the girdling tissue and could still produce new roots and root exudates (Binkley et al. 2006; Chen et al. 2010). Hence, soil microbial growth and activity could be sustained by starch reserves in the surviving roots and the decomposition of freshly dead roots after girdling treatment (Zeller et al. 2008).

In a previous study, girdling reduced litter decomposition due to elimination of rhizosphere C input (Subke et al. 2004). However, the effect of tree girdling on litter decomposition was much weaker than understory removal treatment in the present study. Mycorrhizal fungi were reported to contribute to the subtle responses of litter decomposition to tree girdling treatment (Subke et al. 2004). Eucalyptus urophylla (tree species) can form two types of mycorrhizal association: ectomycorrhiza (ECM) and arbuscular mycorrhiza (AM) (dos Santos et al. 2001), however, D. dichotoma (the dominant species of understory vegetation) can only form AM (Zhang, Guo & Liu 2004). It was reported that ECM usually negatively interacts with AM at the same root system (Lodge & Wentworth 1990), which indicated that the potential effects of tree girdling on one kind of mycorrhizal fungi could be compensated by another kind associated with E. urophylla. However, no such compensation effects occur regarding the understory plant –D. dichotoma. Therefore, fungal communities and its decomposition function could be suppressed by understory removal but not by tree girdling.

Effects of sampling time and plantation age

Sampling time substantially affected the abundance of soil micro-organisms and the composition of soil microbial community in the present study, probably because of the seasonal changes in temperature and moisture, which should be considered when conducting ‘before-after-control-impact’ (BACI) experiment in the monsoon climate regime. Summers in the study area are hot and wet and winters are relatively cold and dry. The wet season is from April to September and the dry season is from October to March. The rainfall and heat input during the wet season account for 83% and 66% respectively, of the input for the entire year (Zhou, Peng & Yu 1995). The temporal changes in soil microbial community could be ascribed to the changes in moisture, soil temperature and substrate availability, as indicated by other studies (Lipson, Schadt & Schmidt 2002, Bardgett et al. 2005, Waldrop & Firestone 2006).

The age of forest plantation also seemed to be closely linked to microbial community. We found that fungal PLFAs and the F : B ratio were significantly higher in the 2-year-old plantation than that of the 24-year-old plantation. However, the bacterial PLFAs were not significantly different between the two plantations. This can be directly associated with the C inputs from plants and soil N status of the plantations. DOC was significantly higher in 2-year-old plantation, which might be in favour of fungal growth rather than bacteria when soil N was limited because fungi demand less N than bacteria (Table 1). In fact, NPP, only measured in control plot, was much higher in the 2-year-old plantation, which intensified the competition for N between plants and soil microbes and resulted in soil N limitation (Bell, Ter-Mikaelian & Wagner 2000).

The age of plantation might confound the effects of understory removal on soil microbial community. We found that 13 individual PLFA biomarkers generally responded positively to understory removal in the 2-year-old plantation. In contrast, 13 individual PLFA biomarkers responded negatively in the 24-year-old plantation. The NPP was higher in the 2-year-old plantation than in the 24-year-old plantation; consequently, the competition for soil nutrients among understory plants, trees and soil microbes would be stronger in the 2-year-old plantation. This was probably why understory vegetation affected soil microbial communities and decomposition function differently between the two plantations.

We also observed an increase in soil DOC in girdling subplots in the 24-year-old plantation, but girdling did not affect the DOC in the 2-year-old plantation, which was supported by another study in our field station (Chen et al. 2010). Our results were contrary to those Zeller et al. (2008) found, they reported DOC increased in O horizon in young spruce forest but decreased in old forest. We postulated that the change of DOC is sampling time-dependent and is also related to plantation age. DOC would decrease at first after girdling because the flow of current photosynthates (carbohydrates, amino and organic acids) from leaves to roots and soil was stopped by girdling (Dannenmann et al. 2009; Kaiser et al. 2010) but it would increase later when root dieback acted as a new source of C (Zeller et al. 2008). In the present study, we might have missed the decreasing phase of DOC for the 24-year-old plantation.

Conclusion

Our study has several implications. First, we separated the relative contribution of tree species and understory vegetation to driving microbial community composition and its decomposition function in terms of below-ground C input in the field experiments, which was usually overlooked in the previous girdling experiment in the forest ecosystems. Secondly, we provided evidence that fungal communities and litter decomposition rate were greatly suppressed by understory vegetation removal, suggesting that understory vegetation plays an important role in maintaining soil microbial community and driving litter decomposition processes. Finally, we demonstrated that the influences of plant functional group loss on soil microbial communities and decomposition were highly depended on the sampling time and plantation age, revealing the importance of seasonality and temporal scale in the regions of monsoon climate.

Acknowledgements

This work was funded by National Basic Research Programme of China (No. 2011CB403200 and No. 2009CB421101), National Science Foundation of China for Distinguished Young Scholars (No. 30925010), Guangdong Natural Science Foundation (9451065005003254), and the Knowledge Innovation Programme of the Chinese Academy of Sciences (KSCX2-YW-G-074-02 and KSCX2-EW-J-28). We thank Prof. Weixing Zhu, Prof. Hua Chen, Dr. Weixin Zhang and Dr. Yuanhu Shao for their comments and discussions on the early version of the manuscript. We are grateful to Prof. Bruce Jaffee and Mr Guomin Huang for improving the manuscript. We also thank Jie Zhao and Xiaoli Wang for their assistant on sample analysis and data collecting.

Ancillary