Study Site

The study site was located within the Fubo plantation station forests (21°57′–22°19′N, 106°39′–106°59′E) of the Experimental Center of Tropical Forestry of the Chinese Academy of Forestry, Pingxiang City, Guangxi Zhuang Autonomous Region, China (Figure 1). The annual rainfall in this subtropical region is between 1,200 and 1,500 mm, the temperature varies between 20.5°C and 21.7°C, and the relative humidity is 80% to 84% (Jiang et al., 2015). The plots are located at an elevation of 430 to 680 m above the sea level. The local soil is classified as soils with little or no profile differentiation (IUSS Working Group WRB, 2015), derive from granite and are classified as red soil in the Chinese soil classification, with a pH value of 4.8 and 5.5 (Jiang et al., 2015). The study site was originally occupied by a subtropical evergreen forest that was cleared and planted with Masson pine (Pinus massoniana) in 1950. Several rotations were grown, and then the land was allowed to lie fallow.

Figure 1.

(a) to (c) Location of study sites in the subtropical monoculture plantation region, near Pingxiang City, Guangxi Zhuang Autonomous Region, China. (d) Experimental plot design: circular permanent plots of 400 m² with 11.2 m radius, and 5 m × 5 m square subplots.

The site was reforested by P. massoniana in 1993 at a density of 2,500 stems per hectare. This density has declined over time because of self-thinning, herbivory, and damage that was sustained during a snowstorm. However, the canopy continued to be dominated by P. massoniana, with certain herbs, such as Maesa japonica, Clerodendrum cyrtophyllum, and Psychotria rubra, present as minor components. Before the treatments, the density of the monoculture stands averaged 1,787 stems ha⁻¹ (SD = 26.7), the mean diameter of all the trees was 13.4 cm, their mean height was 12.7 m, and their basal area was 15.3 m² ha⁻¹ (SD = 8.6).

Two facilitator species were used. Castanopsis fissa (C. fissa), a member of the Fagaceae, is a rapidly growing, native tree species that is tolerant of shade and nutrient-poor environments. This species is a common pioneer in this region that fixes nitrogen and forms a heavy litter layer (Tang, Jeewon, & Hyde, 2005). Manglietia glauca, a member of the Magnoliaceae, is an exotic species introduced from Vietnam (Wang & Cai, 2008). This species is also fast-growing, drought resistant, and tolerates acidic soils. These two pioneer species are often planted in urban areas and used to reforest hillsides across the region (Pan & You, 1994). The two target species used are representative of a number of native species that might be established at these sites and are typical of reforestation trees in South China. Castanopsis hystrix (C. hystrix) is a long-lived, native species that can tolerate some shade and high moisture but not drought or low temperatures; this species grows best in acidic red soil, yellow soil, and brick-red soil developed from granite and sandstone and produces many seeds of high viability and rapid germination (Zhou, 2007). Erythrophloeum fordii (E. fordii) is a long-lived species, which prefers light and acidic soil and is capable of reaching a height of 37 to 45 m and a diameter of 200 to 250 cm, produces high-quality timber, and has important economic and ecological value (Sein & Mitlöhner, 2011).

Experimental Design

In the summer of 2007, four harvesting treatments that included unmanaged monoculture stands as controls (CT remaining after harvest density 1,500 tree·hm⁻²) and the following three levels of canopy retention were applied in the monoculture plantation: 20 (R20 remaining after harvest density 300 tree·hm⁻²), 50% (R50 remaining after harvest 750 tree·hm⁻²) and 75% retention (R75 density 1,150 tree·hm⁻²), respectively. The harvested trees were selected without regard for species or size, but as all compartments were monocultures (>85% P. massoniana) we assume that there is no systematic bias that could significantly affect the comparison between harvest retention levels. The canopy retention was calculated by the number of live trees present in the compartment before harvesting. The harvesting treatments were applied to 20 ha harvest units, which were each subdivided into four compartments. A thinning procedure was performed by poison girdling all of the nontimber tree species having a diameter at breast height (DBH) of 20 to 30 cm, which was measured 1.30 m above the ground, and systematically removing the postlogging competition. The harvesting equipment was restricted to the machine corridors but reached into the residual strips to remove the trees; thus, there was a minimal disturbance of the forest floor and soils in the retained strips.

The four harvesting treatments were assigned to a total of 64 permanent, circular plots located within stands of the monoculture, with each treatment applied in 16 of the plots (Table 1). The plots each covered 400 m² (11.2 m radius) and were traced systematically (Figure 1(d)). The location of the plots within the area was selected according to the homogeneity and accessibility of the plots, which were randomly separated from one another. We initiated the experiment to examine the effects of planting additional species on the processes of succession and restoration within monoculture plantations.

Table 1.

Experimental Treatment and Number of Plots for Each Retention Level and Combination of Facilitator and Target Plant Species and for the CT.

In the spring of 2008, we planted saplings of the four tree species (C. hystrix, E. fordii, C. fissa, and M. glauca) across the entire treatment unit for the three retention harvesting treatments (Table 1). One- to 2-year-old seedlings of the facilitator species were planted at the same distance (3 m × 3 m) between individuals in all the plots, providing a density of approximately 1,200 seedlings ha⁻¹ in the treatments. The plots were divided into subplots, and a combination of each facilitator species (C. fissa or M. glauca) with each target species (C. hystrix or E. fordii) was planted in each subplot. The spacing between the planted trees was 3 m × 3 m. We alternated the identity of the species by planting row. The native target seedlings were watered and fertilized when transplanted and were then permitted to grow naturally. Sixty-four 5 m × 5 m subplots were thus established on the monoculture plantation for the four harvesting treatments (including the CT; Table 1). Forty-eight 5 m × 5 m subplots that received plantings and 16 unmanaged control subplots of similar dimensions that did not receive plantings composed the four harvesting treatments. The planted subplots were divided into four planting treatments (T1, C. fissa and C. hystrix; T2, C. fissa and E. fordii; T3, M. glauca and C. hystrix; and T4, M. glauca and E. fordii) according to the relations of the facilitator and target species, and there were four subplots of each planting treatment. In 2016, we again measured the facilitator and target species in all of the treatments.

Forest Surveys

The population structure of the monoculture plantation was studied from August to September 2007 in the Experimental Center of Tropical Forestry. The first survey was installed prior to harvest in September 2007. We surveyed once again before the harvest during the summer of 2007 and during the summers of 2008, 2012, and 2016. Within an 11.2 m radius from the center of each plot, all of the trees (DBH >5 cm) were botanically identified, individually marked, and counted, and their diameters were measured. The tree density was quantified in this way to account more precisely for the density of the retained trees at a specific location, given the inherent variability in the preharvest tree densities since the last harvest. The DBH, measured 1.3 m above the ground, was noted, and the tree height was measured using an optical height meter (PM-5/1520 P, Suunto, Vantaa, Finland). The mean height and diameter increments were calculated by dividing the height and diameter by the tree age. For each 5 m × 5 m subplot, we determined the number of the planted specimens and their leaf area index (LAI), DBH, height and live aboveground biomass (AGB) as well as the composition and the number of any naturally generated trees and shrubs. We assessed the AGB variation of the four transplanted species in the four planting treatments along with the variable retention harvesting of the tropical monoculture plantation. The DBH and the height of the planted individuals were measured in every permanent plot. Then, the two parameters were used to calculate the aboveground biomass using the following four biomass equations from the literature:

C. hystrix: $W=0.0641{(D^{2}H)}^{0.8699}+0.01050{(D^{2}H)}^{0.8246}$ (He et al., 2012)

E. fordii: $W=0.1957D^{2.0341}+0.0431D^{1.9442}$ (Ming et al., 2014)

C. fissa: $W=0.044{(D^{2}H)}^{0.9169}$ (H. K. Li & Lei, 2010)

M. glauca: $W_{=}0.022{(D^{2}H)}^{1.023}$ (J. Z. Li et al., 2011), where W is the individual aboveground biomass (kg), D is the DBH (cm), and H is the tree height (m).

Resource Measurements

To evaluate the effects of the experimental treatments on resource availability, we measured the light and nutrient availability at each of our sample points. The LAI is defined as the hemisurface area of green leaves (needles) per unit of horizontal ground surface area (Chen & Black, 1992). Thus, it is a measure of the amount of photosynthetically active tissue in a forest stand. We measured the light availability during 2008, 2012, and 2016. Photographs were captured in late August below the canopy in every plot at 1.5 m above the ground using a digital camera (Nikon) equipped with a fisheye lens. The measurements were made on cloudy days or early in the morning to obtain more contrast between the leaves and the gaps between them (sky or clouds) as well as to avoid direct sunlight. The photographs were exposed automatically because under- or overexposure did not improve the analysis. The LAI was obtained from the photographs by a canopy analysis system (Hemiview 2.1, Delta-T Devices Ltd., Cambridge, UK).

The soil sampling and measurements in 2008 were performed on 400 m² circular plots. For the nutrient analyses, the soil samples from the different retention harvesting levels were collected in the middle of each plot to evaluate the role of the facilitation species on the soil condition. Three punch-tube samples of the upper 15 cm of sandy soil were collected and mixed together for every plot. These bulk samples were then analyzed for total N by micro-Kjeldahl analysis, for soil acidity using the pH of water at a 1:2.5 soil to water ratio, and total organic matter by a modified Walkley–Black procedure (Holmgren et al.,1993).

Data Analysis

The cumulative volume growth was calculated for each year using the basal area multiplied by the corresponding tree height and a common form factor of 0.6 (Cannell, 1984), as shown in the following equation: $V_t=\pi \times {(\frac{DB{H}_t}{2})}^2\times h_t\times f$, where V_t is the volume at age t, DBH_t is the diameter at age t, h_t is the tree height at age t, and f is the form factor (the ratio of tree volume to the volume of a cylinder with the same basal diameter and height). The Shannon–Wiener index (H) was used to estimate the diversity among the observed individuals (Magurran, 1988). We compared the structural traits of the trees within the plots by means of one-way analysis of variance (ANOVA), with the retention harvesting level as a factor. The data were checked for normality and heterogeneity of variance before analysis and found to be acceptable for analysis without transformation. The four planting treatment levels were then compared using a Tukey’s test based on the overall effect. The measurements taken once were analyzed using ANOVA, and individual comparisons were based on a Tukey’s test. Indicator species analyses (ISAs, duleg function in the labdsv package; Roberts, 2007) were used to identify the tree species that were significantly associated with one stand origin category or another. Species with values > 0.3 were interpreted as species highly associated with the corresponding category. This test compares the relative frequency of occurrence and abundance of species in different treatments and identifies the tree species that vary more between treatments than would be expected by chance (Legendre & Legendre, 1998). The statistical significance of each indicator value was determined by a Monte Carlo test using 999 permutations. Finally, the variation of the soil elements in the different planting levels was analyzed and compared using a one-way ANOVA and a Tukey’s test. All of the statistical analyses were performed with R statistical software.

Forest Structure

The variable retention harvesting significantly changed most of the studied variables. Several forest attributes differed among the retention levels (R20, R50, and R75; Table 2). The number of trees and effective LAI decreased incrementally with declining retention, whereas the mean height displayed the opposite response, increasing with decreasing retention. These patterns were mostly attributable to the reduced cover and density of trees. There were also significant differences in the mean diameter and basal area. The mean diameter increased to 26.3 cm in R50 because of the selection of larger trees for retention. The basal area significantly increased in R50 compared with the other treatments. A lower basal area was observed in R20 because of the presence of larger gaps. Finally, the total volume was also higher in R50 than in the other treatments.

Table 2.

Forest Structure Results (Mean ± SE) for Different Retention Harvesting Types (R20, R50, and R75) and Primary Unmanaged Forests That Consider the Number of Trees (N) (n hm⁻¹), MD (cm), BA (m² hm⁻¹), MH (m), TV (m³ hm⁻¹), and Effective LAI.

Planted Species Among Harvesting Scenarios

There was a significant influence of the retention level on the composition of the planted saplings in all of the analyses (Table 3). The number of trees per hectare declined with the retention level, but all of the contrasts were significant (Table 3, p < .05 based on Turkey’s test.). Although the number of individual trees increased over time in all harvest plots (Table 2, with Turkey’s test p < .05.), the number of individuals of the four planted species declined with increasing harvest retention level (Table 3, with Turkey’s test p < .05.). In addition, the number of all of the planted species was higher in the R20 treatment than in the other two retention treatments. For all of the treatments, the plots planted with C. hystrix showed significantly more individuals (877 stems hm⁻²) than the other plots (Table 3, with Turkey’s test p < .05.). The mean number of dead plants was quite variable among the retention levels, and all of the contrasts were significant (Table 3, with Turkey’s test p < .05.). The mortality of the four species of saplings was significantly higher for the R75 treatment than for the other two retention levels. For the R75 treatment, E. fordii showed the largest number of dead trees (414 stems hm⁻²), followed by M. glauca (408 stems hm⁻²), C. hystrix (394 stems hm⁻²), and finally C. fissa (382 stems hm⁻²), with the smallest number. In addition, the planted species with different preferences responded differently to the retention levels after harvesting. The ISA indicates that the planted tree species were associated with particular retention levels (Table 3, with Turkey’s test p < .05.). C. hystrix was significantly associated with R20 and R50 (with values of 0.354 and 0.413, respectively), whereas E. fordii showed a positive relation with R50 (0.420).

Table 3.

Characteristics After a 5-Year Period for Four Planted Species at Three Different Harvesting Retention Levels and for an Unmanaged CT That Included the Number of Live Trees Per Hectare (stems hm⁻²) (N), the MN_d (stems hm⁻²), and, Except for the Control, the Results of an ISA for the Tree Species.

Facilitator Species Performance

C. fissa and M. glauca both performed as facilitators where the planted trees were present. For the total DBH and the sapling height of the four planting treatments, the saplings grew taller and greater in diameter when the treatments included C. fissa (T1 and T2) in the retention harvesting regimen (Figure 2). Most of the saplings had their greatest DBH in the treatments that included C. fissa, except for T3 with R20 (Figure 2). The height of the saplings was also greater in the company of the planted species, especially C. fissa (Figure 2). The maximum sapling height was 3.8 m, achieved in T1 with R50, whereas the maximum height for the M. glauca treatments was 3.2 m, lower than that for the C. fissa treatments and achieved in T3 with R20.

Figure 2.

Sapling DBH and sapling height of four planting treatments (T1.CaFi-CaHy= C. fissa + C. hystrix; T2.CaFi-ErFo = C. fissa + E. fordii; T3.MaGl-CaHy= M. glauca + C. hystrix; T4. MaGl-ErFo= M. glauca + E. fordii) in three variable retention harvesting (R20%, R50% and R75%) in 2012. Bars represent one standard error. DBH = diameter at breast height.

The mean annual DBH growth of the tree species differed significantly among the four planting treatments in the variable retention harvesting regimens (Table 4, with Turkey’s test p < .05.). C. hystrix showed the greatest annual DBH growth in the treatment combination of T1 (CaFi-CaHy; Table 4) with R50 (0.74 cm year⁻¹), followed by E. fordii in the combination of T2 (CaFi-ErFo) with R20 (0.71 cm year⁻¹). A similar but even stronger trend was evident for the mean growth in height of C. hystrix and E. fordii in the presence of C. fissa, with both C. hystrix and E. fordii showing the greatest growth in the plots with 50% retention harvesting (0.38 m year⁻¹ and 0.35 m year⁻¹, respectively). This result clearly demonstrated the facilitator effect of the two plant species, with positive effects on the growth of the two target species.

Table 4.

Results of Analyses Examining the Influence on the Target Species Saplings (C. hystrix and E. fordii) of the Species Planting and Retention Harvesting Treatments. The Variables Examined in the Target Species Included the MDG (cm year⁻¹), MHG (m year⁻¹), and Total AGB (t hm⁻²).

The results also indicated significant differences among the planting treatments in the aboveground biomass (Figure 3), with the highest biomass produced by the treatment combination of R50 with T1, followed by T2. Both T3 and T4 exhibited lower aboveground biomasses than those of T1 and T2 at similar retention harvesting levels, except for T2 at R75. The aboveground biomass of the planted species differed significantly among the three retention levels and four planting treatments. For the retention levels, the R50 harvesting treatment presented a higher biomass (T1 and T2 were 129.2 t hm⁻² and 122.1 t hm⁻², respectively) than those of the other two levels (Table 4). The aboveground biomass was higher in the C. fissa than in the M. glauca treatments (Table 4) and, on average, was highest in the plots with C. fissa combined with R50 (129.2 t hm⁻²). The biomass in T4 was lowest (94.3 t hm⁻²) at the R75 level because E. fordii grows very slowly. For the planting treatments, C. fissa performed better than C. hystrix as a facilitator at all of the retention levels (Table 4).

Figure 3.

Box plots showing AGB (t hm⁻²) of planted species for four planted treatments in three variable retention harvests. The horizontal line in each box is the median. Boxes enclose the 75th and 25th percentiles, and error bars enclose the 90th and 10th percentiles. Significant differences (p < .05) between treatments are indicated by different letters above the boxes. AGB = aboveground biomass.

The abundance and the diversity of the naturally regenerated saplings differed significantly among the four planting treatments managed with variable retention harvesting (Table 5). Furthermore, the average abundance (649 N hm⁻²) and diversity (H: 1.3) in T1 were significantly higher for R50 than for the other treatments. At the same retention harvesting levels, the abundance and diversity were higher in the plots planted with C. fissa than in those planted with M. glauca. The plots planted with C. fissa also showed a higher recruitment of species than those planted with M. glauca. The abundance (498 N hm⁻²) and the diversity (H: 0.64) were lowest for the plots containing the T4 with R20 and the T3 with R20 treatment combinations, respectively. The LAIs differed significantly among the four planting treatments with Turkey’s test (p < .05). Overall, the LAI decreased because of thinning. The average LAI (1.44) was significantly higher in the T3 with R75 plots planted with M. glauca than in the other planting treatment plots (Table 5). T1 with R75 showed the second highest LAI (1.38), but the differences between C. hystrix and E. fordii were not significant when the data were pooled by species within R75 (e.g., T1, T2 and T3, T4). Finally, the mean LAI for the R75 treatments was larger than the means for the other two retention levels. Across the tree retention treatments, the plots with less tree retention showed higher light availability.

Table 5.

The N_R (stems hm⁻²), Shannon–Wiener Index (H), and LAI of Naturally Regenerated Saplings for Each Treatment Over a 5-Year Period (Mean ± SE).

Many soil properties in the monoculture were altered by the different planting treatments 5 years postharvest (Table 6). The different treatments affected the soil properties (N, pH, and organic matter) very differently (Table 6). The average N totals of T1, which was planted with C. fissa, were significantly higher than those of the other planting treatments. The total N increased significantly over the study period for the C. fissa planting treatments (T1 and T2), whose values were clearly higher than those of the other treatments. The total N in the CT plots changed the least (0.05 g kg⁻¹). The differences in soil pH among the treatments were found to be just statistically significantly. The average soil pH (4.5) in T4, which was planted with M. glauca, was higher than those in the other treatments, with T3 having the second highest value. The change in soil pH for the C. fissa treatments (T1 and T2) was significantly lower than that for the M. glauca treatments (T3 and T4) during the 5-year period. The average soil organic matter differed significantly among the four treatments (p < .05). The T1 plots had the highest soil organic matter (23.33 g kg⁻¹) when compared with the T2, T3, and T4 plots. The soil organic matter in the plots planted with C. fissa was higher than that in the plots planted with M. glauca. The soil organic matter was lowest, on average, for T4, but increased significantly beneath C. fissa in T1 and T2 (p < .05) and decreased by 0.11 g kg⁻¹ in the CT plots.

Table 6.

Soil Total N, pH, and Organic Matter (Mean ± SE) in Four Treatments.