Introduction

Plant–plant interactions such as competition and facilitation greatly influence plant community structure and dynamics, and are key topics in ecology (Armas and Pugnaire 2005; Callaway 2007). Tree–grass interactions, especially those between tree seedlings and grasses, have attracted community ecologists for a long time (Scholes and Archer 1997; Daly et al. 2000). In the past, many studies have considered the competitive effect of grasses on tree seedlings at later successional stages (White and Newton 1989; Caldwell et al. 1995; Noble 1980; Ball et al. 2002; Bloor et al. 2008). Grass decreased solar radiation reaching the canopy (Caldwell et al. 1995), depleted soil nutrients and water (Maina et al. 2002), and had allelopathic effects (Gruntman and Novoplansky 2004). Several studies reported that belowground competition was more important than aboveground competition in grass–tree interactions (Balandier et al. 2006; Bloor et al. 2008) because grasses developed dense and shallow root systems in the same soil horizon as the newly established tree seedlings (Collet et al. 2006). In recent years, however, perennial grass and annual grass have received substantial attention as nurse plants for improving the survival and growth of late-successional species, for reducing the extreme water stress induced by the vapor pressure deficit (Anthelme et al. 2007; Gasque and Garcia-Fayos 2004), or providing beneficial “fertile island effects” (soil organic matter content and water availability were higher in the soil under the grass) and beneficial canopy effects (photosynthetically active radiation and soil temperature were reduced under the grass canopy) (Maestre et al. 2001; Enkhtuya et al. 2005; Fernandez et al. 2006). However, more experimental evidence showed that grass species were poorer nurse plants for depletion of soil nutrients and weak light interception (Herrero and Gutiérrez 2006; Köchy and Wilson 2000; Peltzer and Köchy 2001). The interactions were concluded as the ultimate results of the balance of competition and facilitation (Berkowitz et al. 1995). These interactions have been studied in many degraded habitats, especially in savanna (Higgins et al. 2000; Riginos 2009), woodland (Jackson and Ash 1998), and semiarid steppe (Liang et al. 2003).

In China, degraded ecosystems occupy about 3.56 × 106 km2, representing 37 % of the entire Chinese land area (SEPA 2006). The problem of ecological degradation is also severe in South China (Peng et al. 2003), where industrialization, urbanization, and rapid increase in human population have destroyed native forest, especially subtropical monsoon evergreen broadleaved forest. This kind of climax vegetation has been replaced by large areas of degraded grassland and shrubland (Guan and Peart 2000). Compared with the remaining natural forests in South China, degraded ecosystems have lower biodiversity, simpler community structure, and lower productivity, and therefore provide poor ecosystem services (Peng et al. 2003; Ren et al. 2007a). Under natural conditions (without human intervention), the plant community in a degraded ecosystem returns to a climax vegetation community, and how to speed up these processes (biodiversity, and ecosystem services) is now a focus of ecological research and forest management (Ren et al. 2007a).

Another problem in South China is poor seedling establishment on degraded lands because of poor soil nutrition, high temperature, strong sun radiation, and the simultaneous occurrence of heavy rainfall and high temperatures during the monsoon season (Ren et al. 2007b). Although some ecological factors, including water and temperature, are apparently sufficient for local vegetation in lower subtropics of South China, water and temperature vary greatly throughout the year. In the dry season from October to March, plants grow slowly because of the low precipitation and temperature. However, in the monsoon season from April to September, heavy precipitation and high temperature result in root insolation after strong rainfall and plant damage.

Grassland in South China accounted for ca. 11.8 % of Chinese grassland area (Liao and Jia 1996). Grassland dominated by Miscanthus sinensis, a native pioneer perennial grass carrying out photosynthesis via the C4 metabolic pathway, is too difficult to develop into shrubland and forest. So, we conducted a field experiment to investigate the interaction (competition or facilitation) between dominant grass M. sinensis and tree seedlings.

Materials and methods

Study site

The study was conducted at a grass slope at the Heshan National Field Research Station of Forest Ecosystem (60.7 m a.s.l., 112°50′E, 22°34′N) in Heshan City, Guangdong, China. This station is one of the stations of the Chinese Ecological Research Network (CERN) and occupies 40 ha. The field site has a typical climate of south subtropical monsoon and laterite soil. The mean annual temperature is 22.6 °C, with a hot and humid summer but a cold and dry winter. The annual mean precipitation and evapotranspiration are 1700 and 1600 mm, respectively. The annual radiation is 4350.5 MJ m−2 a−1. The climax community is low subtropical monsoon evergreen broadleaved forest with representatives of the families Lauraceae, Euphorbiaceae, and Fagaceae. As a result of serious and long-term human disturbance, such as deforestation, however, soil erosion has been severe and the original vegetation has almost disappeared. Our experimental site occupies 2 ha and is the typical degraded, unburned shrub-slope community of the experiment station. Plant species diversity is low, including Miscanthus sinensis, Rhodomyrtus tomentosa, Ilex asprella, Melastoma candidum, Clerodendron fortuneatum, and others. M. sinensis is the dominant species. The current study assessed the potential of the native pioneer perennial C4 grass M. sinensis (Gramineae) to act as a canopy plant. In the lower subtropics of South China, M. sinensis commonly grows on different kinds of degraded slopes (below 1800 m altitude), where it is a dominant species in earlier successional periods. The stalk can reach 1–2 m in height, and the leaves are 40–100 cm long and 6–10 mm wide (Flora of China Editorial Committee 1997). Miscanthus is naturally distributed in the Pacific Islands, the Philippines, China, and Japan (up to 45°N) (Weng and Ueng 1997). It carries out photosynthesis via the Hatch–Slack metabolic pathway, which allows for high CO2 uptake when temperatures and incident light are high (Weng and Ueng 1997). The high water use efficiency (WUE) of C4 species also means that more carbon can be allocated to aboveground organs. Furthermore, Miscanthus reproduces by rhizomes, has a high growth rate and high rate of biomass production (Scebba et al. 2006), and is tolerant to aluminum, heavy metals, and oxidative stresses (Ezaki et al. 2008). In the last decade, Miscanthus cultivation has been promoted as an “energy plant” in Europe to replace fossil energy sources (Schwarz 1993; Himken et al. 1997; Ercoli et al. 1999). Miscanthus can adapt to the weather conditions in the Iberian Peninsula on both the Atlantic and Mediterranean coasts (Velasquez et al. 2003), and its growth could mitigate carbon release (Clifton-Brown et al. 2007).

Species selection

The selected native target plants, Rhodomyrtus tomentosa (shrub), Pinus massoniana (coniferous species), Schima superba (broadleaved tree), and Castanopsis chinensis (broadleaved tree) are all dominant species in different successional stages in the lower subtropics of China (Peng and Wang 1993), which are shrub land, conifer forest, and broadleaved forest (Fig. 1). Also, two noninvasive exotic species, P. elliottii (coniferous species) and Acacia mangium (broadleaved tree), were investigated; these have been extensively planted for restoration in South China. These species allowed us to compare the growth performance and physiology traits of native and exotic species. The characteristics of the target species are summarized in Table 1.

Fig. 1
figure 1

Successional stages and dominant species in South China

Full size image
Table 1 Characteristics of the six target species
Full size table

Nursery procedures and experimental design

Seeds of the native target species (R. tomentosa, P. massoniana, S. superba, and C. chinensis) were collected from the Dinghushan Nature Reserve, which is ca. 100 km north of Heshan and has similar climate and soil. Seeds of the exotic target species (P. elliottii and A. mangium) were purchased from Guangdong Seed Company. Seeds were germinated in a sand bed protected by a thin plastic mulch to prevent frost damage in nursery of South China Botanical Garden, with 21.4–21.9 °C mean annual temperature and 1623.6–1899.8 mm mean annual precipitation. When the seedlings germinated and the first cotyledon emerged, they were transplanted into columniform plastic cups (9 cm diameter × 10 cm height) containing local red soil from the field site supplemented with 10 g organic fertilizer (N:P:K = 1:1:1). Water and weed management followed conventional plantation methods. The seedlings were used for experimentation when they were 6 months old, about 10 cm in height with 3 well-developed leaves at least.

In March 2007, three plots were randomly designated at the 2-ha field site. All the sample plots were selected in the middle of the slope. Each plot was about 100 m2 and contained five patches colonized by M. sinensis (2 m height × 1.8 m canopy radius for each patch) and other areas without plants. The M. sinensis patches were at least 1.5 m apart and were distributed evenly within each plot to avoid spatial clumping. On rainy days in March 2007, seedlings were removed from the nutrition cups and their roots were dipped into mud made from field site soil, according to Castro et al. (2002).

Seedlings of the six target species were subjected to one of two treatments. In the open site (OS) treatment, seedlings were transplanted into the patches without vegetation and outside of the M. sinensis canopies. In the M. sinensis canopy (MS) treatment, the seedlings were transplanted beneath the canopy of M. sinensis, such that the seedlings were located ca. 50 % of the distance between the grass stem (at 0 %) and the outer edge of the canopy (at 100 %). For each of the six target species, 40 seedlings were planted in the MS treatment (eight per M. sinensis patch) and 40 in the OS treatment (eight per open patch). Thus, each plot initially contained 240 seedlings in the MS treatment (6 target species × 8 seedlings per species per patch × 5 patches) and 240 in the OS treatment (6 target species × 40 seedlings per species). Seedlings in both treatments were about 50 cm apart but not grouped by species to avoid intraspecific competition. Transplanted seedlings not surviving to 1 week were removed and replaced by the same number of same-sized seedlings. After transplanting, plants were not watered or fertilized.

Measurement of morphological traits in seedlings

Five individuals (one sample from each patch) were randomly selected at the end of the experiment in March 2009. After measuring the basal diameter (stem diameter at soil surface) and shoot height (maximum height from soil surface to terminal bud) of a target seedling, we harvested them for each treatment and each species. Plant tissues of the individuals were divided into 3 parts: root, stem, and leaf. Root and stem were chopped into small pieces and determined by electronic balance after drying at 70 °C for 96 h. Leaf biomass was determined after drying for 48 h at 70 °C.

Measurement of chlorophyll fluorescence in seedlings

Chlorophyll fluorescence characteristics in leaves, especially rapid light curve (RLC), can be a useful technique for obtaining rapid qualitative and quantitative information on photosynthesis (White and Critchley 1999; Ralph and Gademann 2005), and RLC is also more convenient and nondestructive (Ralph and Gademann 2005; Belshe et al. 2007). For each species, one fully developed leaf per seedling (1–2 seedlings per plot in each treatment) at the top and center of the crown was selected for rapid light-response curves (RLC) measurement from 9:30 am to 11:30 am on three clear days in August 2008. RLC of chlorophyll a fluorescence in the six different species were measured using a portable pulse-amplitude modulated photosynthesis yield analyzer (PAM-2100; Heinz Walz, Effeltrich, Germany). When the RLC was measured, actinic light illumination was adjusted from low to high radiation between 90 and 1800 μmol m−2 s−1 in eight steps and measured by recording the chlorophyll fluorescence parameters, including electron transport rate (ETR) and actual photochemical efficiency of PS II in the light (∆F/F m′) with a standard 2030-B leaf clip holder. ∆F/F m′, which can be calculated as ∆F/F m′ = 1 − F s/F m′ (where F s is the steady-state fluorescence yield and F m′ is the maximum fluorescence yield after light adaptation), decreases with increasing quantum flux density. The relative ETR can be obtained as 0.84 × 0.5 × ∆F/F m′ × PAR, where 0.84 is the coefficient of absorption of the leaves, 0.5 is the fraction of electrons involved in the photoexcitation produced by one quantum, as two photosystems are involved, and PAR is the actinic photosynthetically active radiation generated by the internal halogen lamp of the PAM-2100.

Measurement of chloroplast ultrastructure in seedlings

Chloroplast ultrastructure in leaves, especially the thylakoid membranes in chloroplasts, is associated with chlorophyll a molecules, whose fluorescence emission pattern is a sensitive indicator of the integrity and functionality of the photosynthesis apparatus (Lage-Pinto et al. 2008). Live individuals were selected randomly from the OS and MS treatments. The youngest, fully developed leaves from the individuals were measured as the leaf samples. Leaf samples of six target plants were fixed in aqueous solution containing 2.5 % (v/v) glutaraldehyde and 4.0 % (v/v) paraformaldehyde in 0.1 M sodium phosphate buffer, under low vacuum for more than 24 h. The samples were then rinsed with 0.1 M sodium phosphate buffer for six times and post-fixed in stationary liquid of 1.0 % osmium tetroxide with 0.1 M sodium phosphate buffer for 8 h. After rinsing with 0.1 M sodium phosphate buffer for six times, the samples were dehydrated in a graded series of ethanol [30, 50, 70, 80, 90, 100 % (v/v) for three times per grade, 20 min each time]. After transition in propylene oxide three times (20 min each time), the samples were infiltrated with mixed solution of propylene oxide and spur (1:1 in volume) for 0.5 h and with pure spur for overnight. Subsequently, the samples were cut into sections using a Leica Ultracut S ultramicrotome after being embedded in epoxy resin and polymerized for 8 h at 70 °C. The ultrathin sections were collected on copper grids (300 mesh), and stained with 2.0 % (w/v) uranyl acetate for 1 h followed by 6.0 % (w/v) lead citrate for 20 min. Sections were observed using a JEM-1010 transmission electron microscope (JEOL Ltd., Tokyo, Japan).

Microenvironment characteristics

Air temperature (T A, °C), relative air humidity (RHA, %) 10 cm above the soil, soil temperature (T S, °C) at 5 cm depth, and total radiation (μmol m−2 s−1) were recorded every 1 h for 3 continuous days without clouds in August 2008. Measurements were made by microclimatic sensors connected to a HOBO Pro data logger (HOBO H8; Onset Computers, Pocasset, MA, USA). Two series of sensors were randomly placed in the two treatments (MS and OS), respectively.

To determine the chemical characteristics of the soil, we collected one sample per treatment per plot in June 2008, about 15 months after the experiment began. Each sample consisted of five randomly selected subsamples collected using a 5-cm-diameter soil corer to depth of 20 cm after surface litter had been removed. The subsamples were mixed in a plastic bag and air-dried in the laboratory. Soils were passed through a 2-mm mesh sieve. Soil organic matter (SOM) was determined by titration with FeSO4 after digestion in K2Cr2O7 and H2SO4. The pH of soil slurry (1.0/2.5 w/w of soil/water) was determined using a pH meter (HANNA Instruments, Italy). Total soil nitrogen (TN) was determined by the semimicro-Kjeldahl method. Total soil phosphorus (TP) was determined colorimetrically after HClO4–H2SO4 digestion. Hydrolyzed nitrogen (HN) was determined by titration with HCl after hydrolyzation by NaOH. Available phosphorus (AP) was determined colorimetrically after NaHCO3 extraction. Exchangeable K, Na, Ca, and Mg were determined by GBC932AA atomic absorption spectroscopy (GBC Ltd., Australia) after extraction with 1.0 M NH4OAc (pH 7.0).

To determine soil physical characteristics, one intact soil core (6 cm diameter, 20 cm deep) per treatment per plot was taken using ring knives after the litter and humus layer had been removed. Soil bulk density (SBD), soil moisture at saturation (SSM), and soil capillary moisture (SCM) were measured, using three cores per determination. For SSM, intact soil cores were placed horizontally on a wet filter paper until the soil was saturated; the weights of the moist soil and dry soil were determined; SSM was expressed as a percentage (g of water per g of dry soil × 100). For SCM, intact soil cores were put in salvers to absorb water through capillary action via filter paper until they reached a steady weight, usually 8 h later; the weights of the moist soil and dry soil were determined, and SCM was also expressed as a percentage (g of water per g of dry soil × 100). SBD (g of soil/cm3 of soil) was determined from intact soil cores after removing soil moisture by oven-drying at 105 °C to constant weight. Soil chemical and physical characteristics were analyzed by standard methods at the South China Botanical Garden and Guangdong Institute of Eco-Environment and Soil Sciences (Liu 1996; Duan et al. 2008).

Statistical analysis

To quantitatively compare RLCs, nonlinear estimation in Statistica 6.0 software (StatSoft Inc., Tulsa, USA) was used to fit the RLC curves by the least-squares method. Data were mathematically fitted to a double exponential decay function using a Marquardt–Levenberg regression algorithm (Platt et al. 1980; Ralph and Gademann 2005):

$$ P = P_{\text{s}} \left( {1 - {\text{e}}^{{ - (\alpha \times {\text{PAR}}/P_{\text{s}} )}} } \right){\text{e}}^{{ - (\beta \times {\text{PAR}}/P_{\text{s}} )}}. $$
(1)

In the absence of photoinhibition (β = 0), Eq. (1) can be simplified to

$$ P = P_{\text{m}} \left( {1 - \text{e}^{{ - (\alpha \times {\text{PAR}}/P_{\text{m}} )}} } \right), $$
(2)

where P is the photosynthetic capacity (rETR), P s is a scaling factor defined as the maximum potential rETR, P m is the photosynthetic capacity at saturating light, PAR is the actinic light intensity, α is the initial slope of the RLC, and β is the slope of the RLC where PS II declines. The initial slope of a light curve estimated by fitting this model was affected by the shape of the RLC under saturating irradiance, and α was estimated by directly calculating the slope of the initial part (first four data points) of the RLC, using linear regression (João et al. 2006).

Results are presented as mean ± standard deviation (SD). The general linear model (GLM) was used to determine the effect of treatment on all measured variables. SPSS 13.0 (SPSS Software Inc., USA) was used for statistical analysis.

Results

Morphological performance

The morphological performance of the six target species in the MS and OS treatments is shown in Figs. 2 (biomass) and 3 (shoot height and basal diameter). Biomass of every organ (root, shoot, and leaf) of all the species was obviously higher in OS treatment than in MS treatment (p < 0.001). Compared with MS treatment, the values of shoot height were significantly higher in OS treatment for C. chinensis (p = 0.004), P. elliottii (p < 0.001), and A. mangium (p < 0.001). There was no difference of shoot height in other species between the two treatments (p > 0.05). All the values of basal diameter were significantly higher in OS treatment than in MS treatment, especially for two exotic species, P. elliottii (p < 0.001) and A. mangium (p < 0.001).

Fig. 2
figure 2

Biomass of six target species: R. tomentosa, P. massoniana, S. superba, C. chinensis, P. elliottii, and A. mangium in the MS (under the canopy of M. sinensis) and OS (open site) treatments. Asterisks indicate difference between treatments: *p < 0.05, **p < 0.01, ***p < 0.001

Full size image
Fig. 3
figure 3

Shoot height and basal diameter of six target species in the MS (under the canopy of M. sinensis) and OS (open site) treatments: RT (R. tomentosa), PM (P. massoniana), SS (S. superba), CC (C. chinensis), PE (P. elliottii), and AM (A. mangium). Asterisks indicate difference between treatments: *p < 0.05, **p < 0.01, ***p < 0.001

Full size image

Rapid light curve of chlorophyll fluorescence

ETR and ∆F/F m′ of RLC of chlorophyll fluorescence were measured for different species. The ETR values increased and ∆F/F m′ values decreased with elevated actinic light intensity (Fig. 4). All the ETR and ∆F/F m′ values were higher in the OS treatment than in the MS treatment. The initial slopes of the RLC (α) and the values of photosynthetic capacity at saturating light (P m) were higher in the OS treatment than in the MS treatment for all target species except P. massoniana (p > 0.05) (Table 2). P m values of the two exotic species, P. elliottii and A. mangium, were remarkably higher in the OS treatment than in the MS treatment.

Fig. 4
figure 4

Electron transport rate (ETR) and photochemical efficiency of PS II in the light (∆F/F m′) of rapid light response curves (RLC) of six target species as affected by actinic light illumination in the MS (under the canopy of M. sinensis) and OS (open site) treatments: RT (R. tomentosa), PM (P. massoniana), CC (C. chinensis), SS (S. superba), PE (P. elliottii), and AM (A. mangium). Values are means ± standard error (SE) of five replicates

Full size image
Table 2 Chlorophyll fluorescence parameters of RLCs for six plant species (R. tomentosa, P. massoniana, C. chinensis, S. superba, P. elliottii, and A. mangium)
Full size table

Ultrastructural characteristics

Figure 5 shows the ultrastructural characteristics of the mesophyll cell chloroplast found by transmission electron microscopy (TEM) of the six target species, R. tomentosa (a), P. massoniana (b), S. superba (c), C. chinensis (d), P. elliottii (e), and A. mangium (f) under the canopy of M. sinensis and on the open site. It is clear that R. tomentosa (a), P. massoniana (b), S. superba (c), C. chinensis (d), and A. mangium (f) showed better performance on the open site, such as clear thylakoid structure, and compact grana and stroma lamella. More elaioplasts were found near the starch in the chloroplast of S. superba (Fig. 5, c3) and two starch grains in the chloroplast were in three target species, P. massoniana, P. elliottii, and A. mangium on the open site (Fig. 5, b3, e3, and f3). Compared with the open treatment, all species exhibited grana reduction with disorganization of thylakoids, with little dilation of chloroplast thylakoids in MS treatment. The two exotic species, P. elliottii and A. mangium, exhibited chloroplasts with transmutative forms, and there were no or small starch grains in chloroplast in MS treatment (Fig. 5, e1 and f1). Also, clear mitochondria were seen for P. massoniana and P. elliottii in MS treatment (Fig. 5, b2 and e1).

Fig. 5
figure 5

Transmission electron microscopy (TEM) of the mesophyll cell chloroplast (1 and 3) and stroma (2 and 4) of target species, R. tomentosa (a), P. massoniana (b), S. superba (c), C. chinensis (d), P. elliottii (e), and A. mangium (f) under the canopy of M. sinensis (1 and 2) and on the open site (3 and 4). Starch grains (S). Mitochondria (M). Elaioplast (arrow). a3, b1, b3, c1, c3, d1, d3, e3, and f3 (×12000, Bar 0.5 μm). a1, c2, e1, f1 and f4 (×30000, Bar 0.2 μm). a2, b2, b4, c4, e2, e4 and f2 (×80000, Bar 0.1 μm). a4, d2 and d4 (×100000, Bar 0.05 μm)

Full size image

Microenvironment characteristics

Air temperature and total radiation were significantly lower in the MS treatment than in the OS treatment (p < 0.001 and p = 0.002, respectively) (Fig. 6). Soil temperature and relative humidity did not differ between the MS and OS treatments (p > 0.05).

Fig. 6
figure 6

Air temperature, soil temperature, total radiation, and relative humidity in MS (under the canopy of M. sinensis) and OS (open site) plots from August 21 to 23, 2009

Full size image

With regard to soil physical and chemical characteristics (Table 3), SBD was significantly higher in the OS than in the MS plots (p = 0.040). Soil moisture content at saturation (SSM) and SCM did not differ between the two treatments (p = 0.320 and p = 0.602, respectively). Total soil phosphorus (TP) was significantly higher in the MS than in the OS plots (p = 0.002). The other soil chemical characteristics, pH, total nitrogen (TN), hydrolyzed nitrogen (HN), available phosphorus (AP), exchangeable potassium (K), exchangeable calcium (Ca), exchangeable sodium (Na), and exchangeable magnesium (Mg), did not differ between MS and OS plots (p > 0.05).

Table 3 Soil physical and chemical characteristics in MS (under the canopy of M. sinensis) and OS (open site) plots
Full size table

Discussion

Positive and negative effects take place simultaneously, and the combined effect depends on the balance between the two (Armas and Pugnaire 2005). Nurse plants frequently appear in extreme habitats, but a recent study showed that facilitation existed in mild environments (Holmgren and Scheffer 2010), as also proved by our previous study in which increased shade, soil porosity, and soil moisture under the canopy of pioneer shrub R. tomentosa increased the performance of some seedlings of the climax tree species in South China (Yang et al. 2010). In our study, however, M. sinensis reduced the biomass and basal diameter of all seedlings, as well as the shoot height of C. chinensis, P. elliottii, and A. mangium, indicating that in M. sinensis the negative effects may be stronger than the positive effects. M. sinensis reduced SBD and its canopy buffered the soil temperature compared with the open site. Amelioration of soil physical characteristics often appears in nurse effect syndrome. Fulbright et al. (1995) found that presence of the shrub Prosopis glandulosa could improve seed germination and seedling growth of Celtis pallida by buffering the maximum temperature in southern Texas. In our results, M. sinensis also increased soil nutrients, especially total phosphorus content, compared with the open site. Enkhtuya et al. (2005) found that grass Calamagrostis epigejos can be a nurse plant and facilitate phosphorus uptake for Salix purpurea seedlings through arbuscular mycorrhizal fungi.

The shade produced by the dense canopy of M. sinensis poses a negative effect, as it limits the growth of undercanopy species. Moderate shade from other canopy species can facilitate undercanopy species by reducing photoinhibition that inhibits the repair of PS II and by improving carbon assimilation and plant growth (Werner et al. 2001); For example, shrubs facilitated survival of Quercus and Pinus species under their canopies because of a combined “canopy effect” and “soil effect,” but the canopy effect was stronger than the soil effect in Mediterranean montane ecosystems in southeast Spain (Gomez-Aparicio et al. 2005). Establishment of Sabina vulgaris seedlings was facilitated under Salix cheilophila canopy in Mu Us Sandy Land of northern China because S. cheilophila changed lutein and beta-carotene levels in Sabina vulgaris so that the latter plant was protected against high-light and low-temperature photoinhibition (Ishii et al. 2006). In our study, the daily mean radiation on the open site was 511 ± 90 μmol m−2 s−1, almost 5.2 times greater than under the canopy of M. sinensis (98 ± 10 μmol m−2 s−1). The maximum value of total radiation at midday was 956 ± 107 μmol m−2 s−1 on the open site but only 152 ± 23 μmol m−2 s−1 under the canopy of M. sinensis. The shading generated by M. sinensis seriously restrained light assimilation and photosynthesis for all target species under the canopy of M. sinensis compared with the open site, as indicated by the poor chloroplast ultrastructure, i.e., dilation of chloroplast thylakoids, and lower values of chlorophyll fluorescence parameters, including ETR and actual photochemical efficiency of PS II in the light (∆F/F m′) of RLC. Other limiting factors could also have contributed to the lack of a nursing effect by M. sinensis, including root competition and allelopathy (Chou and Chung 1974; Chou and Lee 1991), which we intend to examine in the future.

On the other hand, the response of later successional species under the canopy of pioneer species may be positive (Gomez-Aparicio et al. 2004; Padilla and Pugnaire 2006) or negative (Ball et al. 2002; Bloor et al. 2008), depending on the species characteristics, abiotic or biotic environment, and other temporal or spatial traits. In our study, there was no clear evidence indicating that the interspecific interactions were related to the successional sequence of the target species under the canopy of M. sinensis. Rather than being related to successional characteristics, our results reflect species-specific characteristics and especially shade tolerance. Shade tolerance can greatly influence interactions between plants (Kerstiens 2001; Yang et al. 2010); For example, the magnitude of the facilitation or nurse effect of the pioneer shrub Rhodomyrtus tomentosa was positively correlated with the shade tolerance of the target species in degraded land of South China (Yang et al. 2010). Seedlings of the two exotic species in our study, P. elliottii and A. mangium, showed poorer shade tolerance and greater sensitivity to competition from M. sinensis than the native species. Both exotic species have been introduced as afforestation trees to restore degraded land in South China because they can grow rapidly and provide essential ecological or socioeconomic services, i.e., preventing water and soil loss, improving soil nutrients, and fixing more carbon as pioneer trees. In our study, however, M. sinensis had a strong negative effect on the growth of both exotic species, probably because of their heliophilous traits. Under high light intensity, a higher transpiration rate and stomatal conductance increased the photosynthetic rate and carbon gain in P. elliottii (Zeng et al. 1999; Gong et al. 2006) and A. mangium (Yu and Bee 2002). In lower light intensity, heliophilous species had lower light use efficiency (LUE) and reduced photosynthetic electronic flow plus a higher dark respiration rate that consumed more assimilation products (Walters and Reich 1999; Zhang et al. 2005). The dense canopy of M. sinensis prevented undercanopy species from performing photosynthesis at lower photosynthetic active radiation (PAR), which greatly reduced the photosynthetic rate of P. elliottii and A. mangium.

In conclusion, negative effects may be stronger in perennial C4 grass M. sinensis. Also, seedlings of dominant species in different successional stages have different positive and negative effect under the canopy of M. sinensis. Seedlings of exotic species used in restoration engineering cannot be directly planted under the canopy of M. sinensis. For establishment of target species in M. sinensis communities in degraded lands of South China, restoration strategies may include clearing of aboveground vegetation, planting of target seedlings in openings to reduce the effects of canopy shading, or selection of competition-tolerant target species. These reasons could partially explain why earlier efforts to restore M. sinensis grassland by directly planting seedlings have failed, and why postfire afforestation has been used before planting the seedlings of native species in Southern China.