Golden Camellia as a Driver of Forest Regeneration and Conservation: A Case Study of Value-Chain Forestry with Camellia quephongensis in Que Phong, Nghe An, North-Central Vietnam

Abstract

Golden camellia is a highly valued commercial plant owing to its flowers’ medicinal substances. One species of golden camellia, Camellia quephongensis Hakoda et Ninh, is used as a non-timber forest product for value-chain forestry (VCF) in Que Phong District, Nghe An Province, North-Central Vietnam; its production and sale is an important business activity which contributes substantially to local livelihood improvement. According to previous studies and in situ observation, golden camellias, including C. quephongensis, inhabit regenerated forests, especially along streams and rivers. This encourages VCF with C. quephongensis to function as a driver of forest regeneration and conservation. However, the ecological aspect of this species in regenerated forests is poorly understood. Thus, this study aimed to determine this species’ growth patterns and population expansion during forest regeneration. We surveyed the morphological characteristics of this species at different ages in several populations, and revealed that the plants reached a tree height of 150 cm and started blooming eight years after germination. Comparing C. quephongensis-populations inhabiting different succession-staged forests, we observed that the population in long-term regenerated forests was significantly larger and had a pyramid-shaped age distribution pattern to possibly expand its population size. Based on this, we conclude that C. quephongensis is a good indicator of forest regeneration and contributes to local livelihoods and forest regeneration and conservation.

1. Introduction

Value-chain is a concept which regards business activities as a series of values from production, processing, and distribution to consumption. It is a methodology to maximize benefits by analyzing the values derived from each process of business activities [1]. Value-chain forestry (VCF), in this study, is defined as forestry, which is not a production activity but a business activity starting from production to sale in markets. VCF can provide requisite benefits to the regions and boost forest values from an economic perspective.

The trial of the VCF with Camellia quephongensis Hakoda et Ninh (Figure 1) [2,3,4] was implemented in Que Phong District, Nghe An Province, North-Central Vietnam. The central government designated Nghe An Province as a VCF and agriculture-promoted province in Vietnam [5]. Under this trial, residents of Que Phong try to manage a non-timber forest product of golden camellia, that is, Camellia quephongensis.

C. quephongensis is a short-sized shrub tree, growing under canopies and blooming yellow flowers annually from December to February [4]. Golden camellia contains medicinal compounds, such as antioxidants: polyphenols, carotenoids [6,7,8], and saponins [3]; therefore, it has been used as a health-tea, named golden camellia tea. It is traded at a high price, and is sold at hundreds of USD per kilogram in retail markets [3].

In Que Phong, at least three processing factories have been established in recent decades (Figure 1). They are the critical actors in VCF, which involves procuring fresh flower buds from villagers, processing them to commoditize golden camellia tea, and shipping them to the markets or directly selling them to consumers. Before they were established, villagers used to collect the matured flower buds from the naturally growing trees in the forests to sell the intermediaries for 1–2 USD/kg; subsequently, they were brought to the processing factories in other regions. However, the processing factories directly purchased the fresh materials at a 4–5 USD/kg, increasing total revenue of the villagers from 10,000–20,000 USD/year to 50,000–70,000 USD/year according to the processing company. Recently, several villagers have transplanted saplings into their plantations. In addition, VCF can feed back demands of golden camellia tea in the markets to the collectors; thus, it prevents over-harvesting, contributing to the sustainable use of natural resources of golden camellia.

Camellia in Que Phong consists of not only C. quephongensis but also two other species: C. ngheanensis Do et al., sp. nov. [9] and C. puhoatensis Ly et al., sp. nov. [10]. The major habitat of these three species was reported to be riparian forests [4,9,10], where plant growth is restricted and plant morphology is affected by flooding [11,12,13]. They also grow in secondary forests on mountain slopes, which were once naturally or anthropogenically disturbed. Other golden camellias include C. impressinervis Chang et Liang [14], C. kirinoi Ninh [15], and C. tuyenquangensis Luong et al., sp. nov. [16], primarily distributed in the secondary forests in North Vietnam [17]. Based on these habitat characteristics, many golden camellias could be characterized as disturbance-adapted species.

C. quephongensis is a highly valued economic shrub. If villagers recognize its commercial value, they will conserve the forests to create VCF with this species, and consequently, it could promote forest regeneration and conservation. However, the propagation of this species in regenerated forests remains poorly understood, and whether this species functions as a driver of forest regeneration has not yet been elucidated. The present study aimed to understand the (1) growth patterns, (2) density, (3) stand structure, and (4) age distribution of C. quephongensis, and speculate its (5) population dynamics in regenerated forests to consider the possibility of C. quephongensis as a driver of forest regeneration and conservation.

2. Materials and Methods

2.1. Study Area and Sites

Que Phong District is located in the westernmost part of Nghe An Province, adjacent to the boundary of Laos. It is surrounded by high mountain ranges on three sides, the west, north, and east, from which many streams flow down to meet in the district’s center, forming the Hieu River. C. quephongensis is scattered along these streams and rivers. Since the district is rich in forests, it is also designated as a biosphere reserve by UNESCO (Figure 2).

Among the different C. quephongensis growing areas, four populations of C. quephongensis were selected for this study in the following areas: Co Muong (CM: 19°34′36″ N, 104°52′21″ E); Chau Kim Commune, Na Sanh (NS: 19°42′11″ N, 105°00′16″ E); and Phuong Tien (PT1: 19°38′25″ N, 104°59′18″ E, and PT2: 19°38′33″ N, 104°59′07″ E); Tien Phong Commune (Figure 2). The NS population was developed anthropogenically, and all plants were individually planted.

Table 1. Change in forest distribution in the site of CM population and its surroundings (Co Muong) from 1995 to 2022. Figures in the table were estimated using Google Earth and Google Earth Engine Timelapse.

Category		1995 (ha)	2022 (ha)
Forests	Tall-seized trees	364.3	33.8
	Middle-sized trees		419.7
	Plantations		17.9
Logged areas	Shrubs	190.2	25.2
	Grasslands, bare lands		46.1
Others	Arable lands	45.5	25.6
	Water bodies		6.6
	Artifacts, such as houses		25.0
Total		600.0	600.0
Ratio of logged lands (%)		31.7	11.9

compares CM’s present state, including forests and surroundings (2022), with the past state (1995). Slash-and-burn farming started in this area in 1988 and continued until 1992; afterward, logging was prohibited by the local government in 1996. Forests included tall-sized trees, middle-sized trees, and plantations, and logged lands included shrubs, grasslands, and bare lands in 2022, but these categories could not be identified in 1995, according to Google Earth Engine Timelapse. The logged lands decreased from 31.7% to 11.9% in 25 years. Because the site for CM growing was once bare land, the age of the earliest plants was estimated to be 25 years.

The southwest and northeast monsoons influence the region from May to September and October to March, respectively. During summer, from April to October, it is hot and humid, and the temperature peaks at 34 °C on average, while during winter, from November to March, it drops to 9 °C on average. The rainy and dry seasons are from May to October and November to April, respectively. The mean annual rainfall is 1800 mm [18].

2.2. Methods

2.2.1. Surveys

Pre-field surveys for selecting the C. quephongensis populations were conducted on 30–31 October 2022, followed by field surveys on growth patterns, density, stand structure, and age distribution on 5–9 December 2022; 9–11 January 2023; and 12–14 February 2023.

2.2.2. Growth Patterns

A plantation of C. quephongensis trees of several ages growing under the matured Cinnamon tree canopies existed at the foot of the CM population; hereafter referred to as CMP. A 20 m × 20 m quadrat was set up, and tree height (TH), stem diameter at the ground level (SD), and crown width (CW) were measured (

Table 2. Quadrat features. Height from the river bed was the mean of five individuals of C. quephongensis growing in the quadrat.

Population	Census Area	Quadrat Features
		Size	Altitude	Land	Gradient	Vegetation	Note
	(m × m)	(m × m)	(m)		(°)	(Dominant Tree)
CMP	–	20 × 20	200	Slope	15	Plantation (Cinnamon)	–
CM	275 × 10	20 × 20	330	Slope	15	Evergreen broadleaved forest (Vernicia)	–
		20 × 20	300		30		–
PT1	220 × 10	15 × 20	140	Slope	43		–
PT2	580 × 10	10 × 20	130	Riparian	35	Riparian forest (Fraxinus)	107 cm from the river bed.
NS	85 × 10	7.5 × 15	160	Slope	42	Mixed bamboo	–
		7 × 15	160		43	and Acacia

). If the stem was multiplied, each SD was measured, and summed to represent the SD of the tree.

In this plantation, trees of ca. 40 cm in TH were transplanted from the CM population, and flower buds were harvested five years after transplantation, according to the plantation owner. Because it took 6–8 years to set flower buds in the case of a similar species, C. chrythanta [19], 40 cm transplanted trees were regarded as three-year-old saplings, and the age of transplanted trees was estimated by adding three years to the growing duration in the plantation. Regarding trees growing in the CM population, individuals with heights similar to transplanted trees in the plantation were considered saplings. In addition, the plants with heights significantly lower than the mean and less than 5% of the possible density of tree-stage aggregation (n = 30) were regarded as second or more filial generations, while others were regarded as 25 years old. The growth curve was simulated using these data.

2.2.3. Tree Density

Trees growing in the populations were counted by measuring the TH, SD and CW. Population growing areas were estimated by census distance multiplied by 10 m in width (5 m on both sides from the census route); the distances were 275 m, 220 m, 580 m, and 85 m for CM, PT1, PT2, and NS, respectively (Table 2).

2.2.4. Stand Structure

Quadrats were set up at the represented sites of the populations (Table 2). Number of stand layer, trees consisting of the canopy layer, and trees whose crowns were next to the canopy layer, hereafter referred to as the second layer, were recorded with their height and diameter at breast height (DBH:130 cm from the ground level), and basal area was calculated with trees consisting of the canopy layer and second layer. Hemispherical photographs were taken (Nikon D90, Tokyo with fisheye lens EX, SIGMA, Kawasaki, Japan) to understand the light condition inside the quadrat, and canopy openness was estimated using a software (free software, CanopOn 2 version 2.03c, http://takenaka-akio.org/etc/canopon2/ (accessed on 4 April 2023)) in the laboratory. The number of quadrats was one or two for each population.

2.2.5. Age Distribution

The growth curve divided the growing stage into four categories: seedling, sapling, young tree, and mature tree. Subsequently, the age distribution patterns were clarified.

2.2.6. Statistical analysis

Correlation coefficients of age with tree height, stem diameter, and crown width were calculated to understand the growth patterns of the transplanted trees in the plantation. We used an ANOVA to test the effects of populations on the following tree variables: TH, SD, CW, number of stems, and ratio of TH to CW (TH/CW) using BellCurve for Excel 2015 (Social Survey Research Information Co., Ltd., Tokyo, Japan). An effect was considered significant when the probability was p < 0.05. In case of a significant effect, a Tukey–Kramer test was consecutively performed to compare the tree variables between populations.

3. Results

3.1. Growth Patterns

Age of the transplanted trees in the plantation, termed CMP, correlated with TH, SD, and CW, with correlation coefficients (r) of 0.9787 (p < 0.01), 0.9806 (p < 0.01), and 0.9817 (p < 0.01), respectively. (

Table 3. Correlation coefficients of age with tree height, stem diameter, and crown width. Double asterisks indicate a difference at p < 0.01. TH, SD, and CW stand for tree height, stem diameter, and crown width, respectively.

		TH	SD	CW
Age	correlation coefficient (r)	0.9787	0.9806	0.9817
	probability (p)	**	**	**

).

Figure 3 shows the growth curve fitted by the Gompertz model (r² = 0.9439). The growth curve showed that this species grew to a maximum height of approximately 367 cm. The flower buds were harvested five years after transplantation, according to the plantation owner. The transplanted saplings were approximately 40 cm in TH, equivalent to the three-year-old saplings; thus, C. quephongensis grew to the tree stage, bloomed eight years after germination, and its height was more than 150 cm. We defined seedling, sapling, young tree, and matured tree, as height ≤25.0 cm (age < two-year-old), ≤150.0 cm (age < eight-year-old), ≤280.0 cm (age < thirteen-year-old), and >280.0 cm (age ≥ thirteen-year-old), respectively.

$$TH=K{b}^e{}^{-ct}$$

(1)

where, TH is tree height, t is age, and K, b, and c are constants: 366.941, 0.01157, and 0.21058, respectively (r² = 0.9439).

Figure 4 shows the relationship between TH and CW in the CMP and CM populations (r² = 0.8041). C. quephongensis showed a CW of ca. 300 cm, while its TH was ca. 350 cm; thereafter, it tended to decrease.

3.2. Population size and morphological characteristics

Table 4. Density, tree height, stem diameter, number of stems, crown width, and ratio of tree height to crown width. Sap and seed indicate sapling and seedling, respectively. En-dash and N/A indicate none and not applicable, respectively.

Variables	Density	Seedling				Sapling				Tree
	Upper: Total
	(Middle: Tree)	Mean	±	SE	n	Mean	±	SE	n	Mean	±	SE	n
	(Lower: Sap + Seed)
	(Indv./100 m2)	(cm)				(cm)				(cm)
CM
TH	1.1	–	±	–	0	92.0	±	N/A	1	351.0	±	8.9	30
DB	(1.06)	–	±	–		1.5	±	N/A		13.7	±	1.4
Number of stems	(0.04)	–	±	–		1.0	±	N/A		5.3	±	0.5
CW	Area: 2750 m2	–	±	–		59.0	±	N/A		314.5	±	14.6
TH/CW		–	±	–		1.6	±	N/A		1.2	±	0.1
PT1
TH	1.4	–	±	–	0	101.4	±	13.1	14	244.0	±	18.6	17
DB	(0.8)	–	±	–		1.3	±	0.2		3.9	±	0.8
Number of stems	(0.6)	–	±	–		1.1	±	0.1		1.8	±	0.3
CW	Area: 2200 m2	–	±	–		65.1	±	8.5		119.9	±	15.7
TH/CW		–	±	–		1.7	±	0.2		2.5	±	0.3
PT2
TH	0.3	25.0	±	N/A	1	104.0	±	20.7	5	205.2	±	12.1	11
DB	(0.2)	0.7	±	N/A		3.0	±	1.4		6.7	±	1.8
Number of stems	(0.1)	1.0	±	N/A		1.4	±	0.4		2.7	±	0.6
CW	Area: 5800 m2	N/A	±	N/A		64.0	±	14.3		190.8	±	32.1
TH/CW		N/A	±	N/A		1.8	±	0.4		1.4	±	0.3
NS
TH	1.4	–	±	–	0	98.5	±	21.1	4	248.9	±	13.7	8
DB	(0.9)	–	±	–		1.8	±	0.7		2.8	±	0.6
Number of stems	(0.5)	–	±	–		1.5	±	0.5		1.4	±	0.2
CW	Area: 850 m2	–	±	–		46.8	±	8.5		107.8	±	13.6
TH/CW		–	±	–		2.1	±	0.1		3.0	±	0.9

shows the tree density and the tree variables of the four populations. PT1 and NS had the highest total density (1.4 individual/100 m²), followed by CM (1.1 individual/100 m²), whose density of seedlings and saplings was significantly lower than the two other populations, because of its removal by the forest owner to transplant into the plantation. PT2’s density was the lowest (0.3 individual/100 m²), and trees were scattered in the relatively large area of the habitat. NS was an anthropogenically originated population, and the initial density when planted was 2.4 individual/100 m², according to the forest owner.

In terms of the tree variables, TH, SD, number of stems, and CW of tree-phase were significantly higher in the CM population compared to other populations (p < 0.01); the means were 351.0 ± 8.9 cm, 13.7 ± 1.4 cm, 5.3 ± 0.5, 314.5 ± 14.6 cm, respectively. It was followed by the TH of NS: 248.9 ± 13.7 cm and SD, number of stems, and CW of PT2: 6.7 ± 1.8 cm, 2.7 ± 0.6, and 190.8 ± 32.1 cm, respectively (

Table 5. Differences in TH, SD, number of stems, CW and TH/CW of tree-phase among the populations tested using ANOVA, Tukey–Kramer. Variables with a same letter are not significantly different.

Variables (Tree-Phase)	CM	PT1	PT2	NS	p
TH (cm)	351.0 ± 8.9 a	244.0 ± 18.6 b	205.2 ± 12.1 b	248.9 ± 13.7 b	<0.01
SD (cm)	13.7 ± 1.4 a	3.9 ± 0.8 b	6.7 ± 1.8 b	2.8 ± 0.6 b	<0.01
Number of stems	5.3 ± 0.5 a	1.8 ± 0.3 b	2.7 ± 0.6 b	1.4 ± 0.2 b	<0.01
CW (cm)	314.5 ± 14.6 a	119.9 ± 15.7 b	190.8 ± 32.1 b	107.8 ± 13.6 b	<0.01
TH/CW	1.2 ± 0.1 c	2.5 ± 0.3 ab	1.4 ± 0.3 bc	3.0 ± 0.9 a	a–c, <0.01
					ab–c, <0.01
					a–bc, 0.0348

). There were no differences in any variables of seedlings and saplings.

Ratio of TH to CW (TH/CW) indicates the growth form; the balance of apical and lateral bud growth, and CM had the lowest CW/TH: 1.2 ± 0.1, followed by PT2: 1.4 ± 0.3. These values of CW/TH were significantly smaller than those of PT1 and NS (p < 0.01), and that of NS (p = 0.0348), respectively.

3.3. Stand Structure

The stands containing C. quephongensis populations were different in canopy forming species and anthropogenic impacts (

Table 6. Stand structure.

Land Morphology		Riparian	Slope
Anthropogenic perspective		partially natural	planted after seven years		regenerated for one/two decade/s	regenerated for 25 years
Disturbances		grazing flooding	trampling, grazing		Harvesting	saplings removed for transplanting
C. quephongensis population		PT2	NS		PT1	CM
			(planted)
Forest type in the quadrat (Dominant tree)		native forest (Fraxinus)	bamboo with half planted (Acacia)		secondary forest: evergreen broadleaved forest (Vernicia)
Canopy height	(m)	16.0	12.0	14.0	22.0	20.0	21.0
Number of forest layer	(n)	5	3	3	5	4	4
Tree density	(indiv./100 m2)	3.0	7.1	7.6	2.7	3.3	2.3
Basal area	(cm2/100 m2)	891	111,669	138,273	5733	1248	918
Canopy openness	(%)	20.2	28.0	28.3	36.5	22.7	18.8

). The PT2-included stand was the native riparian forest dominated by Fraxinus. The NS-included stands consisted of bamboos and Acacia, which was planted seven years ago. The stands including CM and PT1 were Vernicia-dominated secondary forests. The former has been regenerating for 25 years after discontinuity of slash-and-burn farming, while it has been one or two decade/s since the latter was logged, which was confirmed using the Google Earth Engine Timelapse.

The canopy height of the stands containing CM and PT1 reached 20 m or more, and they developed four or five layers. The NS-included stands with bamboos and Acacia had high tree densities, which were 7.1–7.6 individual/100 m², and basal areas were 111,669–138,273 cm²/100 m². Conversely, the CM-included stands had lower tree densities and basal areas, which were 2.3–3.3 individual/100 m² and 918–1248 cm²/100 m², respectively, and the PT1-included stand had tree density and basal area of 2.7 individual/100 m² and 5733 cm²/100 m², respectively. The native riparian forest including PT2 had 16.0 m canopy height, five layers, 3.0 individual/100 m² tree density, with a less basal area of 891 cm²/100 m². All forest stands had relatively closed canopies, varying from 20.2% canopy openness for the native riparian forest (PT2 stand) to 36.5% for the long-term regenerated forest (PT1 stand).

3.4. Age Distribution

Figure 5 compares the age distribution of PT2, NS, and PT1 to explore population size expansion with forest stand structure development. These populations were affected by different disturbances; PT2 could be influenced by yearly flooding owing to the growing site being almost at the same level as the river bed (107 cm). In addition, these populations were affected by grazing because the stream in the dry season is used for buffalo transfer. Anthropogenically developed NS was affected by disturbances, such as trampling and grazing by cows. PT1 was not disturbed, except for irregular flower bud harvesting by the residents.

The population size of PT2 was 1/5-fold that of the other two populations, and the age distribution pattern also differed among the three populations: PT1 had a peak in the category of seedling and sapling (45.2%), which had a pyramid-like shape, while NS peaked at the young tree stage; PT2 had the same trend as NS but without a matured tree. The ratio of trees categorized as young trees in the population of NS and PT2 was 58.3% and 58.8%, respectively.

Regarding stand structure, the forest’s basal area and tree density differed; the stands including NS were significantly higher in the basal area, which was two and three digits greater than those of the stands including PT1 and PT2, respectively. The canopy height of the stand with PT1 was the highest (22.0 m), followed by those of stand with PT2 (16.0 m) and stand with NS (13.0 m).

4. Discussion

4.1. Growth Patterns of C. quephongensis

Several age–tree height models have been proposed [20,21]. In this study, the Gompertz model was selected and fitted well to the growth curve of C. quephongensisits; the coefficient of decision (r²) was 0.9439. Equation (1) predicted that the maximum height was approximately 367 cm, which was shorter than the figures previously reported: 400–500 cm [4]. Some individuals of CM had more than 400 cm TH. This growth model was able to predict the average height of C. quephongensis and could be sufficient for monitoring the growth of the population.

It is vital to understand productivity to perform VCF successfully; in terms of VCF with C. quephongensis, flower bud yield influences revenues for residents involved in this activity, and TH and CW frequently determine it [22]. Figure 4 indicates the relationship between TH and CW; CW reached approximately 300 cm when TH grew to 350 cm 20–22 years after germination. A similar species of C. euphlebia, a species with natural distribution in southern China and Vietnam, yields 3 kg flower buds when its TH reaches 250–300 cm and CW reaches 220–270 cm [3]; thus, C. quephongensis could also produce 3 kg flower buds in this stage of growth. For the CM population, TH reached 350 cm, and CW decreased slightly. This species developed multi-stems at the tree base to form the crown, and the mean number of stems and total SD were 5.3 ± 0.5 and 13.7 ± 1.4 cm, respectively, indicating that less than 3 cm of one stem supported 350 cm of CW. As the stems were potentially bowed owing to their weights, it decreased the CW.

4.2. C. quephongensis as a Driver of Forest Regeneration and Conservation

In this study, forest regeneration referred to the recovery of the forest through natural processes after significant human or natural disturbances [23]. The forest including CM was once used for slash-and-burn farming; 25 years have passed since the last logging. The size of trees in the CM population was the highest in all parameters: TH, SD, number of stems, and CW among the four populations surveyed. However, it was strongly affected by the collection of saplings for transplantation to the plantation; thus, it lacked the earlier stage of trees.

To understand the adaptivity of C. quephongensis to the regenerated forests, the age distribution and population size of three forests, which were at different succession stages, were compared (Figure 5). They were developed on similar slopes of 35°–43° along the stream or rivers, and their canopy openness did not significantly differ, ranging from 20.2% to 36.5% (Table 6). However, their canopy height and basal area were significantly different. Regarding canopy height, the stand with PT1, which has been regenerated for possibly one or two decades without distinguished disturbances, was the highest, followed by the stands with PT2 and NS. The stand including PT2 was the native riparian forest, and it has been facing frequent flooding and undergoing regeneration; this could be affecting growth of trees. Generally, excessive disturbances of flooding cause leaning, downsizing, and multi-stemming of trees [12]. Thus, significantly smaller TH/WC of PT2 population could be due to frequent disturbances by flooding. The stand containing NS had an extremely high basal area with a low canopy height, and it was a bush-like forest. Based on these speculations, the stand with PT1 progressed more in succession than the other two stands.

Eliminating NS, which was the anthropogenically originated C. quephongensis population, PT1 was compared with PT2. PT1 had approximately 5-fold the population size of PT2 and an appropriate age structure to expand the population size [24,25] in regenerating forest without distinguished negative impacts. This implied that C. quephongensis was well-adapted to the once-disturbed forest and has been expanding its population with a steady forest recovery process. Thus, this species could be an indicator and driver of forest succession from significant disturbances. Conserving C. quephongensis would contribute to the local livelihood and promote forest regeneration and conservation.

5. Conclusions

In this study growth patterns and population expansion of C. qupephongensis were clarified. In terms of growth patterns, growth curve of this species was fitted by the Gompertz model, showing that it grew to the tree stage, bloomed eight years after germination, and its height was more than 150 cm, and finally reached a height of 367 cm on average. Using the growth model, age distributions of different succession-staged populations were elucidated, revealing that C. quephongensis population, which grew in the long-term regenerated forests without significant disturbances, expanded more in size with an appropriate age structure compared to one in the earlier succession stage affected by frequent disturbances of flooding and other impacts. Based on this, we concluded that C. quephongensis could be an indicator of forest regeneration, and conserving VCF with this species could promote forest regeneration.