The study area is in the Tiaohuapo Scenic Spot in the Baili Rhododendron National Nature Reserve, Guizhou Province, China (27°23′22′′N and 105°51′52′′E), near the Jiaozi Mountain, at an altitude of 1700–1900 m. The climate of the region can be characterized as subtropical humid monsoon climate, with an annual average temperature and precipitation of 11.8°C, and 1150.4 mm, respectively, a frost-free period of 257 days, and an annual sunshine duration of 1335.5 h. Zonal vegetation is dominated by the evergreen broad-leaved mountain forest. The existing vegetation is primarily Rhododendron shrub species exhibiting succession and transitional characteristics (Li and Chen, 2005; Jiang et al., 2015). Among Rhododendron shrubs, R. delavayi, R. annae, and Lyonia ovalifolia are the main dominant species. The soil type is primarily yellow soil with an acidic pH (4.61–5.32).

In August 2015, three 20 m × 20 m fixed quadrats were set in the study area to represent the control community (CK), and three 5 m × 5 m survey quadrats were set diagonally in each quadrat. The species, plant number, branch number, height, and canopy width of the shrubs as well as the species and number of the herbaceous plants in the quadrats were recorded. Each Rhododendron shrub in the fixed quadrat was labelled.

In conjunction with the guidelines of Management Office of the Baili Rhododendron National Nature Reserve, all plants in the fixed quadrat were deforested and recovered. The shrubs in the fixed quadrat were deforested in December 2015 and the stubble height was recorded as approximately 15 cm. The number of stubbles was also recorded. The age of shrubs was determined based on the rings of the main pile of shrubs. All other plants were deforested and eliminated.

The same method was used to survey and record the sprouting number, height, and canopy width of Rhododendron shrubs in August 2018 and August 2021. Moreover, the species number, height, and canopy width of other shrubs, as well as the species and number of herbaceous plants in the quadrats, were also surveyed. The communities surveyed in 2018 and 2021 were denoted as D_3yr (Community 3 years after clearcutting) and D_6yr (Community 6 years after clearcutting), respectively.

SPSS 26.0 software was used to measure the correlation (Pearson correlation) and linear regression analysis between the height and coverage as well as clearcutting time and age of Rhododendron. Analysis of variance was performed (p < 0.05 indicates significant differences) to determine the difference in IV of community species and biodiversity indices between the different stages of Rhododendron shrubs. All plots were prepared using SigmaPlot 14.0.

The three communities comprised 26 species, including nineteen species of shrubs and seven species of herbaceous plants, belonging to 14 families and 22 genera (Tables 1, 2). Ericaceae, Rosaceae, and Theaceae accounted for the highest number of species, specifically 19.23%, 15.38%, and 11.54% of the total, respectively. Three species of Rhododendron plants accounted for 11.54% of the total species.

CK comprised 13 species belonging to seven families (including Ericaceae, Lauraceae, Theaceae, and Poaceae) and 11 genera. Four species were identified in the herbaceous layer and nine in the shrub layer of which five viz., R. delavayi, L. ovalifolia, R. annae, and P. japonica of Ericaceae and L. pungens of Lauraceae exhibited IVs > 0.1 (0.31, 0.17, 0.12, 0.12, and 0.12, respectively). R. delavayi was the dominant species in the shrub layer of CK, and L. ovalifolia was the subdominant species.

D_3yr comprised the highest number of species, with 23 species from 12 families (including Ericaceae, Rosaceae, Theaceae, and Poaceae) and 19 genera. There were seven species in the herbaceous layer and 16 in the shrub layer of which two, R. delavayi and R. annae of Ericaceae, exhibited IVs of > 0.1 (0.53 and 0.19, respectively). R. delavayi was the dominant species in the shrub layer of D_3yr, and R. annae was the subdominant species.

D_6yr comprised 20 species belonging to nine families (including Ericaceae, Rosaceae, Theaceae, and Poaceae) and 16 genera. There were three species in the herbaceous layer and seventeen in the shrub layer. Consistent with D_3yr, two species in the shrub layer, R. delavayi and R. annae, exhibited IVs > 0.1 (0.46 and 0.18, respectively). R. delavayi was the dominant species in the shrub layer of D_6yr, and R. annae was the subdominant species.

In the CK→D_3yr→D_6yr community sequence, the height and coverage of Rhododendron shrubs first decreased and then increased (Figure 1). The height decreased significantly from 191.87 cm in CK to 75.61 cm in D_3yr and then recovered to 116.71 cm in D_6yr. The coverage decreased significantly from 68.46% in CK to 37.89% in D_3yr and then increased significantly to 73.28% in D_6yr. The relative height increased from 43.79% in CK to 65.40% in D_3yr and then decreased to 58.54% in D_6yr. The relative coverage increased from 49.05% in CK to 83.24% in D_3yr and then decreased to 77.32% in D_6yr. The height of Rhododendron species in the three communities were significantly different (p < 0.05). The coverage in D_3yr was significantly different from that in CK and D_6yr (p< 0.05). The relative height and coverage in CK were both significantly different from those in D_3yr and D_6yr (p < 0.05).

The Margalef richness index values of species in the shrub layers of the three communities were significantly different (p < 0.05) (Figure 2). However, the differences in the Simpson diversity, Shannon–Wiener diversity, and Pielou evenness indices were not significant (p > 0.05). The values of all four indices were the highest in CK and lowest in D_3yr. The values of the Margalef richness, Simpson diversity, Shannon–Wiener diversity, and Pielou evenness indices of species in D_3yr were 61.15%, 80.30%, 84.79%, and 96.61% of that of species in CK, respectively.

As CK→D_3yr→D_6yr progressed, the dissimilarity coefficients of the three communities showed a decreasing trend (Table 3). Both Cody and Whittaker indices were the highest in CK and D_3yr (3.83 and 1.65, respectively) and the lowest in D_3yr and D_6yr (0.83 and 1.20, respectively). The similarity indices, the Jaccard and Sorenson indices, were the highest in D_3yr and D_6yr (0.87 and 0.93, respectively) and the lowest in CK and D_3yr (0.38 and 0.55, respectively). The similarity between CK and D_3yr was low, that between CK and D_6yr was moderate, and that between D_3yr and D_6yr was high.

The height and coverage as well as the sprouting time and age of Rhododendron shrubs were significantly and positively correlated (p < 0.01). The correlation degree of sprouting time was higher than that of the age of Rhododendron shrubs (Table 4).

The sprouting time (t) and age (a) are the two main factors that affect clearcutting sprouting of Rhododendron shrubs. The linear relationships between sprouting height (h), sprouting coverage (c), sprouting time (t), and age (a) of Rhododendron shrubs after clearcutting were well fitted (R² = 0.65, p < 0.01 and R² = 0.50, p < 0.01, respectively) (Table 5).

The height of Rhododendron shrubs in CK, D_3yr, and D_6yr showed a gradual increasing trend with age (Figure 3). Compared with CK, the height of Rhododendron shrubs in D_3yr and D_6yr was lesser and increased slowly. The coverage rates of Rhododendron shrubs in the three communities also increased gradually with increasing age. The coverage rates of Rhododendron shrubs in D_3yr and D_6yr at the early stage were higher than those in CK.

To determine the sprouting ability at different ages, the age of Rhododendron shrubs was divided into three classes (I, II, and III). The sprouting ability of Rhododendron shrubs first increased and then decreased with age (Figure 4). Class II Rhododendron shrubs had the strongest sprouting ability. Moreover, the sprouting ability of class III Rhododendron shrubs was significantly different from that of classes I and II shrubs (p < 0.05).

Species composition reflects the structure, dynamic changes, and succession characteristics of a community and plays a decisive role in community diversity (Gilliam, 2007; Liu et al., 2020). In this study, 26 species belonging to 14 families and 22 genera were identified in the three communities studied (Table 1). The number of species in CK was the lowest probably because the canopy density of Rhododendron shrubs was high and the litter layer was thick, which reduced the migration and colonisation of Rubus trianthus, Cotoneaster horizontalis, Eurya loquaiana, Populus adenopoda, belonging to the Rosaceae, Theaceae, Salicaceae, as well as herbaceous plants. However, clearcutting weakened interspecies competition by reducing canopy density (Jones et al., 2019). Consequently, a large number of herbaceous plants, and species belonging to Rosaceae, Theaceae, and Salicaceae migrated; thus, D_3yr had the highest number of species. Subsequently, Rhododendron shrubs gradually occupied the upper space, the canopy density increased, and herbaceous plants gradually disappeared. Therefore, the number of species in D_6yr was lower compared with that in D_3yr.

The height and coverage of plants in communities are important indicators for evaluating the plant growth status (Chen et al., 2014). In this study, the height and coverage of plants in communities after clearcutting were significantly higher than those of plants in the seed renewal communities (Figure 1), primarily because the large number of sprouting branches produced by Rhododendron shrubs could occupy more space by utilizing the resources, hence promoting rapid formation of community hierarchy. Post clearcutting, the sprouting coverage of Rhododendron shrubs in D_6yr was higher than that of shrubs in CK, primarily because clearcutting promoted shrub tillering and expanded individual canopy width. After clearcutting, the relative height of Rhododendron shrubs in D_3yr and D_6yr also increased significantly, probably because Rhododendron shrubs occupied the upper space of the whole shrub community after clearcutting, playing a dominant role in light absorption and utilization. The absence of other tall trees and shrub vegetation enabled Rhododendron plants roots to utilize soil nutrients more effectively. Therefore, clearcutting resulted in the domination of Rhododendron plants in the community, with vigorous population renewal and a stronger self-sustaining ability of the population.

As a basic feature of a community, biodiversity is a key factor driving the dynamics and processes of an ecosystem and can indicate the habitat status, composition structure, and distribution pattern of a community (Liu and Bra, 1998; Tilman et al., 2012). In this study, the evenness index of the communities changed after clearcutting (Figure 2). The index value was relatively large, indicating that the heterogeneity of the study area was poor. This was also determined by the characteristics of the sprouting branches of Rhododendron shrubs. Although clearcutting increased the number of species, the α-diversity was still lower than that in CK before clearcutting. This may be because Rhododendron plants has a strong sprouting ability. After clearcutting, Rhododendron plants is likely to occupy space in the form of sprouting branches to inhibit the migration and colonisation of foreign species (Grant and Loneragan, 1999; Kruger and Midgley, 2001). Therefore, Rhododendron plants with its strong sprouting abilities can reduce species diversity and species turnover in the community to a certain extent.

β-Diversity is usually expressed as the turnover rate of biological species in different habitats. A high β-diversity index indicates a low level of species similarity between different habitats or ecosystems (Yang et al., 2020b). The results showed that the similarity between CK and D_3yr was low, that between CK and D_6yr was moderate, and that between D_3yr and D_6yr was high (Table 3). The Cody and Whittaker indices were higher before and after clearcutting, whereas the Jaccard and Sorenson indices were lower, indicating that clearcutting changed the species composition and community structure of Rhododendron. The similarity between CK and D_3yr was low but that between CK and D_6yr was moderate, indicating that with the progression of CK→D_3yr→D_6yr, the number of common plant species first decreased and then increased. Subsequently, the similarity among species composition also first decreased and then increased, indicating that Rhododendron plants could recover to the pre-clearcutting community level in a short time through clearcutting–sprouting renewal.

In Rhododendron shrub communities, both seedlings and sprouting seedlings can undergo population renewal. Nevertheless, sprouting seedlings are an important source of population dynamics (Vesk and Westoby, 2004). Due to the strong sprouting ability of Rhododendron plants, deforested communities rapidly self-substitute, which not only affects the community structure but also has an important impact on population dynamics. Clearcutting–sprouting renewal avoids the problems of very low sprouting and survival rates as well as weak seedling competitiveness. The results of this study similarly showed that the dominance of the Rhododendron population in D_3yr and D_6yr after sprouting was greater than that in CK before clearcutting. The growth recovery of Rhododendron plants after clearcutting is completely dependent on the amount of stubble, root tiller, and rhizome, whereas the number of sprouts is a common index to characterize the strength of sprouting (Chi et al., 2019). Although the sprouting ability of Rhododendron plants was strong in this study, the recovery ability first increased and then decreased with the age of the Rhododendron shrubs. The sprouting ability of Rhododendron shrubs of age 10–12 years was the strongest (Figure 4). This may be associated with the aging of plants, an excessively large root system, and a lack of rhizome growth, which led to partial necrosis of the underground root system.

The community succession of Rhododendron plants does not deviate far from the original direction because of its strong sprouting ability and the effect of the ecological niche after clearcutting (Ahmad et al., 2021). However, the communities in this study had a high number and proportion of Rhododendron shrubs initially, which were in situ substituted with a similarly high number, accelerating the recovery to the original structure and succession rate of the community. Therefore, the shrub layer of Rhododendron communities after clearcutting was primarily composed of sprouting branches of Rhododendron, and the final direction of succession was subtropical evergreen broad-leaved forest. Compared with seed renewal and succession, the clearcutting–sprouting succession of Rhododendron plants takes a shorter time to achieve maturity (Loucks, 1970; Chen et al., 2019).