The effects of species-composition-oriented silviculture on timber value and carbon – a stand-level case study in subtropical China

ABSTRACT

Species selection and composition in afforestation affect forest characteristics, which, in turn, affect the quality and quantity of ecosystem services a forest provides to society. Trade-offs and synergies among the various forest goods and services are key issues in multipurpose forest management. In this study, we propose a stand-level integrated analysis framework applying the dynamic forest simulator PICUS v1.5 and techno-economic analysis to assess the effects of a range of species-composition-oriented silvicultural options on timber value and carbon sequestration in subtropical China. The aim is to inform stakeholders on the costs and benefits of the stand-level options. Taking Pinus massoniana and Castanopsis hystrix to represent dominant native conifers and broadleaves, respectively, in subtropical China, five typical silvicultural options were studied: P. massoniana monoculture with normal and high density (stems ha⁻¹); C. hystrix pure stands; and even-aged and uneven-aged mixtures of both species. Results indicate that the uneven-aged mixture performed better in carbon sequestration than the other four options. The even-aged mixture showed better combined benefits of timber production and carbon sequestration and additional advantages in balancing long- and short-term benefits over 50 years. Furthermore, the even-aged mixture had the strongest adaptability to market fluctuations. The uneven-aged mixture performed best economically among the five options in scenarios of future higher timber prices. The results will inform stakeholders about the performance of stand-level silvicultural options with respect to revenues and environmental benefits to society.

KEYWORDS:

• dynamic simulation
• timber value
• carbon sequestration
• techno-economic analysis

Introduction

China has achieved outstanding results in the expansion of planted forests and in reducing carbon emission due to deforestation and forest degradation (Liu et al. 2014; NASA 2019), but the country’s forestry development also faces several challenges. A shortage of domestic timber supply in total volume and large-dimension timber is one of these, resulting from dramatically increasing wood consumption coupled with low volumes of growing stock and a skewed forest age-class distribution. The imported timber volume currently accounts for up to 50% of timber demand, and the domestic timber supply shortage in both total volume and large-dimension timber is growing rapidly (He and Xu 2011; Tan 2011; State Forestry Administration of China 2014; Tang and Lei 2014). Given the current ban on commercial harvesting in natural forests, planted forests are playing an increasingly important role in China’s domestic timber supply. Revenues from timber are the major motivation for forest producers to undertake forest management. Most of China’s forests are located in underdeveloped rural areas, and the role of forestry as an income source for local people is substantial. Furthermore, climate change is among the most serious global concerns due to its risk potentiality (Grubb et al. 1997; Feng et al. 2006; Ramanathan and Xu 2010; UNFCCC 2015). Forest ecosystems play unique roles in combating climate change with their exceptional capacity to sequester carbon. Planted forests are important because of the total area they cover and their accessibility. China’s government is increasingly willing to tackle global climate change, and there is a commitment at the policy level to use plantations as a nature-based solution for mitigating climate change (Fang et al. 2001; Yin et al. 2010; Chen et al. 2014; Payn et al. 2015).

Several national-level programs and policies create substantial opportunities for the development of planted forests. The aim of the Fast-growing and High-yielding Timber Plantation Base Construction Program was to increase the domestic timber supply, with more than $10 billion (all dollar currencies in this paper are US dollars) in funding over the period 2001–2015. The National Timber Strategic Reserve Program expects to invest a total of around $60 billion between 2018 and 2035 with the aim of establishing a forest resource base for large-dimension and high-quality timber through the development of planted forests with multipurpose management. The aim of the Collective Forest Tenure Reform is to significantly promote the utilisation of 180 million ha of collective forest lands and to mobilise social capital to promote collective forest land development (State Forestry Administration of China and the Ministry of Finance of China 2017).

Forest management is currently high on the national agenda, with the publication of China’s revised national-level forest law at the end of 2019. Planted-forest management strategies are also in a transitional phase in China (Liu et al. 2015; State Forestry Administration of China 2016; Liu et al. 2018). The conventional timber-focused, single-purpose approach is shifting to multipurpose management, consistent with changes in social demands.

Because the stand is the fundamental unit of forest ecosystems, studies are mostly being developed at the stand level to support the development of the new adapted management regimes. The close-to-nature forest management approach, which originated in Germany and Central Europe, is considered a potentially feasible practice for future large-scale expansion (Bürgi 2015; Wu et al. 2015; State Forestry Administration of China 2016). Its application has facilitated the development and formation of a range of silvicultural options. The concept and specific techniques of close-to-nature forest management in the context of China have been discussed and studied widely (Lu et al. 2009; Jiang et al. 2015; Wu et al. 2015; Lan et al. 2016). Some specific effects of silvicultural options on the development of planted forests are also being studied at the stand level, which indicates the promotion of ecosystem services (Lei et al. 2005; Kang et al. 2009; Zhang et al. 2010; Liu et al. 2012; Luo et al. 2013; Li et al. 2015; Ming et al. 2019). These studies facilitate planted-forest management and are further driving the policy transition in China (Liu et al. 2014).

However, knowledge gaps still exist. First, studies on the effects of species-composition-oriented silvicultural options on stand-level ecosystem service provisioning are very limited. This is of great importance in China, where afforestation plays a considerable role. Second, assessments of ecosystem service provisioning from planted forests with changing silvicultural strategies are rare at the stand level, with recent multiple ecosystem-services assessments focusing mainly on planted forests under conventional rotation-based clear-cut management (Chen et al. 2008; Bai et al. 2011; Su et al. 2012). This lack of consideration of other silvicultural regimes is due mainly to the availability of insufficient empirical data because field experiments focused on multipurpose forest management were initiated only in recent years.

Simulation-based ex-ante assessments are a potential means for overcoming the lack of long-term field data (Seidl et al. 2007). The forest ecosystem simulator PICUS v1.5 has been employed successfully to analyse stand-level forest growth dynamics and forest management strategies in Europe (Lexer and Seidl 2009; Seidl et al. 2011; Maroschek et al. 2015; Irauschek et al. 2017). Recently, the PICUS model has been parameterised for forests in Asia (Chen et al. 2013; Zhang et al. 2020).

The aim of this paper is to analyse the effects of various species-composition-oriented silvicultural options on timber value and carbon sequestration at the stand level, focusing on Pinus massoniana Lamb. and Castanopsis hystrix Miq., based on an integrated assessment framework.

Materials and methods

An integrated interdisciplinary assessment framework, including field trials, model simulations and economic analysis, was employed to analyse the effects of a range of species-composition-oriented silvicultural options on timber value and carbon sequestration in subtropical China. The study framework is illustrated in Figure 1.

Figure 1. Flowchart of the analysis framework for the current study

Field trials

The field trials were conducted at the Experimental Centre of Tropical Forestry (ECTF), Chinese Academy of Forestry, Pingxiang City, Guangxi Zhuang Autonomous Region, in southwest China (106°51′–106°53′E, 22°02′–22°04′N). The area has a semi-humid to humid southern subtropical monsoon climate, and it receives adequate sunshine (1200–1600 h y⁻¹) and abundant precipitation (1200–1500 mm y⁻¹). The soil is dominated by the ferralsol soil type according to the World Reference Base for Soil Resources, and the average annual temperature is 20.5–21.7°C. The ECTF has been conducting field experiments on the effects of silvicultural strategies for around 50 years.

Pinus massoniana and C. hystrix were used in this study because they are dominant native conifer and broadleaved (respectively) species in subtropical China. Pinus massoniana is a typical fast-growing species, which quickly provides a large volume of timber, and C. hystrix produces valuable timber. Both are locally popular species in afforestation activities, fulfiling distinct demands for large-quantity (P. massoniana) and high-quality (C. hystrix) timber (Wang and Jiang 2002; Zhang et al. 2019). Pinus massoniana monoculture is dominant in terms of area and makes a substantial contribution to the timber supply; the area of C. hystrix monoculture is growing rapidly, with the increasing importance of high-value timber and ecologically preferable native tree species (Harrington 1999; Cai et al. 2007). Mixtures of these two species are attracting attention in China due to the expectation of the provision of multiple ecosystem services (Guo et al. 2010).

Five species-composition-oriented silviculture options involving the two species in subtropical China were studied in this paper: (1) P. massoniana (Pm) monoculture with normal (Pm-pure-st) and (2) higher density (Pm-pure-hd), (3) C. hystrix (Ch) pure stands (Ch-pure), and (4) even-aged mixture (Mea) and (5) uneven-aged mixture (Mua) of both species. The related preliminary field trials have been developed at the ECTF, and the low-risk treatments have been generated in practice (Guo et al. 2010). The trials were established no more than 30 years ago and thus are too short for decision-making on long-term forest management based on direct field results. For this reason, forest simulation modelling was applied to explore the mid- to long-term development of the stands using the five options.

The forest ecosystem model PICUS v1.5

PICUS v1.5 (henceforth referred to as PICUS) is a hybrid of classical gap model components PICUS v1.2 (Lexer and Hönninger 2001) and a process-based stand-level net primary production model 3PG (Landsberg and Waring 1997). In PICUS, the growth, regeneration and mortality of individual trees are simulated using a grid of 10 m × 10 m patches and layers of crown cells of 5 m height each. A three-dimensional light model, used to estimate absorbed radiation for each tree, is a key component to connect the gap model elements to the production model. For a detailed description of core model components, see Seidl et al. (2005) and Irauschek et al. (2017). The model provides accurate tree-growth projections and captures key ecosystem processes such as deadwood decay and tree regeneration, including browsing by ungulates. The PICUS model approach aims to combine the abilities of gap models with regard to inter- and intraspecific competition, multispecies and multilayered stand structure, and the benefits of a widely used, robust stand-level forest production model based on the concept of radiation-use efficiency (Seidl et al. 2007). The model can simulate spatially explicit silvicultural activities, which are implemented with a PICUS-specific scripting language. It requires monthly weather data (mean temperature, precipitation, vapour pressure deficit, global radiation) of the study area as an input.

Confidence in the suitability of the model was based on several aspects. First, the underlying mathematical models, including the forest gap model and 3PG model, have been examined and verified for a long time (Bugmann et al. 1997; Landsberg and Waring 1997; Bugmann 2001; Sands and Landsberg 2002; Almeida et al. 2004; Stape et al. 2004; Yan and Shugart 2005). Their integration in the PICUS model has also been examined as an effective way of overcoming the limitations of the individual models (Mäkelä et al. 2000; Peng 2000). Second, PICUS was successfully tested and applied in various studies in European forest ecosystems in temperate-continental, maritime and Mediterranean biomes (Seidl et al. 2007; Didion et al. 2009; Seidl et al. 2011; Huber et al. 2013; Maroschek et al. 2015; Pardos et al. 2015; Irauschek et al. 2017). Third, the parameterisation process for the whole model is well established and the model is available with a set of reference parameters. In China, the parameters of several local species have been further calibrated based on local field data (Seidl et al. 2005; Chen et al. 2013; Zhang et al. 2020). Based on the well-calibrated parameters of previous studies, the final evaluation of the model for the current study was implemented by a qualitative comparison of emerging average diameter at breast height (DBH), average height and the number of stems, based on simulations with expert knowledge (Zhang et al. 2020).

Simulation and options

PICUS was applied to simulate the development of stands representing the five options in this study for 50 years. Model output variables included annual values for the volume of standing trees, the volume of harvested wood, and the biomass of standing trees, including stems, branches, leaves and roots. Biomass was converted to whole-tree carbon by assuming a carbon content of dry biomass of 50%.

The selected silvicultural treatments followed the field trials (Table 1). The treatments are the most adapted, mature and low-risk for each species-composition option, and their feasibility and reliability have been examined in practice. Specifically, the initial planting densities were the outcome of the local experts’ experience. In the P. massoniana pure forest, thinning was applied four times because the species is shade-intolerant, allowing the remaining P. massoniana more space for normal growth. The planting of relatively high-density (stems ha⁻¹) trees was also applied in reality to examine whether it could increase economic revenue by increasing the number of stems. In the C. hystrix pure forest, thinning was applied six times to provide more space and further facilitate DBH growth for large-sized wood supply. In Mea, thinning was applied five times, with higher thinning intensities for P. massoniana. This can generate short-term revenue because P. massoniana usually grows faster and its thinning allows more space for C. hystrix growth for the long-term high-value wood supply. In the Mua, high-intensity thinning was used three times, providing more space for the planting of C. hystrix in the midterm and generating a short-term wood supply of P. massoniana. The conventional regime applied locally is a final cut at 26–28 years for P. massoniana and 30–35 years for C. hystrix, thus producing small and medium-sized timber. Fifty years was set as the time horizon in the simulation to observe the effects of longer rotations for the silvicultural options. From the perspectives of local people and managers, extending the selected species’ growing periods by around 20 years is practically feasible and acceptable. The larger-sized timber resulting from the longer growing period is expected to increase the economic benefit, given the current wood-supply gap in China, as described above. On the other hand, extending the observations for 20 years is a conservative approach given the trade-off between the length of the projection period and its accuracy. This conservative approach reduces uncertainties associated with increasing disease and mortality risk as trees age (Rich et al. 2007).

Table 1. Details of the five management alternatives

Alternative	Pm-pure-st	Pm-pure-hd	Ch-pure	Mea		Mua
Species	Pm	Pm	Ch	Pm	Ch	Pm	Ch
Planting time (y)	0	0	0	0	0	0	25th
Planting density	3000	3300	2500	1250	1250	3000	750
Thinning timing (y)	12th/20th /30th/38th	12th/20th /30th/38th	10th/15th/20th/25th/ 30th/35th	10th/15th/20th/25th/30th		10th/18th/25th	None
Thinning intensity	40%	40%	30%	40%	30%	60%	0%

Ch = Castanopsis hystrix; hd = high density; Mea = even-aged mixture; Mua = uneven-aged mixture; Pm = Pinus massoniana; st = standard (normal) density.

The silvicultural treatments were programmed using the PICUS-specific scripting language. Based on the field investigation, thinning activities were programmed as thinning from below (Jäghagen and Lageson 1996). The required monthly weather data to drive the model were obtained from the China Meteorological Data Service Center (China Meteorological Administration 2017). The soil characteristics of the field trial sites were obtained from the laboratory analysis of soil samples collected in the field. Differences in soil conditions were scrutinised and found to be negligible and statistically not significant.

Timber production was calculated from harvested wood from thinnings and the presumed harvest at the end of the assessment period (i.e. liquidation value). Based on timber production, the timber value was assessed using a techno-economic assessment, as indicated below. Carbon sequestration was measured by the average carbon stock over the assessment period in living trees, including stems, branches, leaves and roots. The carbon storage in wood removed from the forest was not included in the analysis. Given great uncertainties in the carbon price, carbon values and the combined values of timber and carbon for each option were preliminarily explored under three carbon price scenarios (see discussion).

Techno-economic assessment

Techno-economic analysis is commonly used to compare the economic performance of various alternatives (Newman et al. 2009). Potential values of the wood extracted from thinnings and the presumed final cuts were considered in this study as the economic revenue. Net present value (NPV), equivalent annual annuity (EAA), internal rate of return (IRR), and the sensitivity of NPV to harvesting cost, timber price and discount rate were analysed as four dominant indicators for assessing timber economic performance (Cubbage et al. 2007; Cubbage et al. 2010; Wagner 2011; Keča et al. 2012).

NPV is a measurement of profit calculated by subtracting the present values of cash outflows, including initial establishment cost, from the present value of cash inflows over time (Newman et al. 2009). The derivation of NPV complies with the profit function theory specifically in the forestry field (Li et al. 2020). Thus, NPV is typically used as the major decision indicator and EAA is an equivalent indicator usually used to convert the total present value to an equivalent annual value. In this study, options with higher NPVs and EAAs was considered more profitable. The discount rate is usually assumed to be constant over the assessment period and takes the risk and time value of money into account (Tee et al. 2014). The average real interest rate in China from 2000 to 2018 was 2%, and forest management is usually considered a long-term investment with the risks of vandalism, wildfire, pests and other natural hazards (Yin and Newman 1996; Linden and Leppänen 2003; Zhang and Pearse 2011; Zu 2012; Pinheiro and de Ribeiro 2013; World Bank 2019). The risk rate of planted-forest management was set at 1%, following previous specific studies in China (Wang 2010; Kang 2014). Therefore, the base real discount rate was set at 3%. In the analysis, NPV and EAA were formulated as:

(1) $NPV=\sum _{t=0}^{t=n}\frac{R_t-C_t}{{\left(1+r\right)}^t}$(1)

(2) $EAA=\frac{NPV\cdot r}{1-{\left(1+r\right)}^{-n}}$(2)

where $R_t$ is the potential revenue of timber at year t; $C_t$ is the potential cost of silvicultural activities at year t; and $r$ is the discount rate.

IRR is the return rate that makes the NPV for all cash flows of a given option equal to zero. Generally, options with IRRs greater than the pre-set minimum acceptable return rate (MARR) would be financially acceptable (Newman et al. 2009). Simply, the discount rate mentioned above, considering both the interest rate and risk, was used as MARR. IRR was formulated as below in the context of this study:

(3) $\sum _{t=0}^{t=n}\frac{R_t-C_t}{{\left(1+IRR\right)}^t}=0$(3)

Sensitivity analysis in the context of techno-economics can be used to determine the effect of the uncertainty of involved factors on economic feasibility (Newman et al. 2009). Over a 50-year time horizon, socioeconomic factors are likely to vary significantly with changes in policies, markets and technologies. Specifically, timber price is usually affected by the relationship between supply and demand. A rise in labour or transport costs could lead to an increase in harvesting cost, and improvements in forest management measures and mechanisation could lead to a decrease. The discount rate varies closely with socioeconomic circumstances. For the sensitivity analysis, the range of harvesting cost was set at ($-$50%, $+$50%) of the base cost and the range of timber price was set at ($-$50%, $+$50%) of the base price. Profit-making private forest enterprises may prefer a higher discount rate. Thus, discount rates from 1% to 8% were used to analyse the effect on NPVs. In addition, the sensitivity of the silvicultural regimes matters. Given the current lack of data on variations in related parameters, no quantitative analysis was undertaken, but a qualitative discussion was conducted. The quantitative sensitivity index (SI) was formulated as below:

(4) $\mathrm{S}\mathrm{I}=\frac{\mathrm{\Delta }\mathrm{N}\mathrm{P}\mathrm{V}/\mathrm{N}\mathrm{P}\mathrm{V}}{\left|\mathrm{\Delta }X/X\right|}$(4)

where $X$ is the fluctuating variable and $\mathrm{\Delta }X$ is change in the variable.

Socioeconomic data of related costs and prices were obtained from two sources: published studies and literature, including the China Statistical Yearbook, the China Forestry Statistical Yearbook, research papers and reports (Cai et al. 2007; Zhang 2014; National Bureau of Statistics of China 2019; State Forestry Administration of China 2019; Zhang et al. 2019); and social interviews with experts working on the forest product trade and employees at local forest farms. The original currency unit was Chinese yuan converted to US dollars at an exchange rate of 1:7 of US dollars to Chinese yuan. The ratio of merchantable timber volume to overall stocking volume was obtained from the local survey referring to DBH.

Results

Temporal dynamics of the quantity of stocking volume and carbon sequestration

To further describe growth trends over time, we divided the 50-year assessment period into three stages: an early stage, from years 1 to 18; a middle stage, from years 19 to 35; and a late stage, from years 36 to 50.

In the early and middle stages, Pm-pure showed the highest stocking volume, Ch-pure had the lowest and Mea was in between, while Mua had a relatively large fluctuation (Figure 2). The rankings of Pm-pure, Ch-pure and Mea were maintained as before in the late stage, but Mua showed a rapid increase up to a maximum value of around 340 m³ ha⁻¹ at the end of the simulation period.

Figure 2. The development of stocking volume over time for all silvicultural options

The average carbon stock for Pm-pure-st, Pm-pure-hd, Ch-pure, Mea and Mua over the period was 56.4 tonne ha⁻¹, 55.9 tonne ha⁻¹, 51.7 tonne ha⁻¹, 56.8 tonne ha⁻¹ and 59.0 tonne ha⁻¹, respectively. In the early stage, Ch-pure showed the highest carbon storage and Pm-pure the lowest. In the middle stage, Ch-pure had less carbon storage than Pm-pure, and Mua had the lowest carbon storage. At the late stage (>year 36), Mua stored the most carbon, at 122 tonne ha⁻¹, and Ch-pure had the least carbon stock, at 79 tonne ha⁻¹ (Figure 3).

Figure 3. The development of carbon storage for all silvicultural options

Potential economic value of timber

Cash flows

Cash flow is usually used as a straightforward way to show economic inflows and outflows. The revenues and costs for each option were calculated by year and explicitly shown as cash flows (Figure 4). The quantities and intervals of economic payback varied among the options, but no apparent conclusion could be drawn on cash flow. Nevertheless, this exercise laid the foundation for calculating NPV, EAA and IRR (in the next section).

Figure 4. Cash flows for all silvicultural options

NPV, EAA and IRR

The NPV, EAA and IRR were calculated for each option (Table 2). For all options, IRR was higher than the MARR, indicating that all options were economically feasible, with a ranking of Mea $>$ Mua $>$ Ch-pure $>$ Pm-pure-hd $>$ Pm-pure-st. The ranking in terms of NPV and EAA was Mea $>$ Ch-pure $>$ Mua $>$ Pm-pure-hd $>$ Pm-pure-st, indicating that Mea was the most profitable option, followed by Ch-pure, and Pm-pure-st was the least profitable.

Table 2. Net present value (NPV), equivalent annual annuity (EAA) and internal rate of return (IRR) for all alternatives

	Pm-pure-st	Pm-pure-hd	Ch-pure	Mea	Mua
NPV ($ ha^−1)	12 707.43	13 037.29	14 720.57	15 242.43	13 755.00
EAA ($ ha^−1)	475.40	487.74	550.72	570.24	514.59
IRR	11.92%	12.11%	12.32%	13.56%	13.34%

Ch = Castanopsis hystrix; hd = high density; Mea = even-aged mixture; Mua = uneven-aged mixture; Pm = Pinus massoniana; st = standard (normal) density.

In general, all options were economically acceptable. Mea was the most favourable: it returned the highest profit, with an NPV of $15 242.43 ha⁻¹, and had the strongest profitability capacity, with an IRR of 13.56%. Pm-pure was the least favourable for both these metrics, and Ch-pure and Mua were between Mea and Pm-pure for both (Table 2).

Sensitivity analysis

The SIs are shown as slopes in Figure 5. They decreased linearly with an increase in harvesting cost, while sensitivity to a change in harvesting cost varied between options (Figure 5a). Pm-pure was the most sensitive, Mua was the second-most sensitive, and Ch-pure was the least sensitive. The average SI of the harvesting cost was 0.27. NPV increased linearly with an increase in timber price, with the magnitude of the increase varying among the options (Figure 5b). Mua was the most sensitive to an increase in timber price, Pm-pure was the second-most sensitive, and Ch-pure was the least sensitive. The average SI of the timber price was 1.41.

Figure 5. Sensitivity analysis by option – NPV with: (a) changing harvesting cost; (b) changing timber price; and (c) changing discount rate

NPV decreased exponentially with an increasing discount rate (Figure 5c). Pm-pure and Mua were the most sensitive options and Ch-pure and Mea were the least sensitive. For the high-expected-return scenario (6% discount rate), NPVs for Pm-pure-st, Pm-pure-hd, Ch-pure, Mea and Mua were $4091 ha⁻¹, $4254 ha⁻¹, $5096 ha⁻¹, $5540 ha⁻¹ and $4655 ha⁻¹, respectively. Their relative ranking was the same as for the baseline scenario (3% discount rate).

Overall, in terms of the silvicultural options, Pm-pure and Mua were the most sensitive to changes in timber price, harvesting cost and discount rate, while Ch-pure was the least sensitive to changes in all three aspects. This indicates that the economic performance of the Ch-pure option is most stable, while Pm-pure and Mua would be risky options given the inherent uncertainties of a changing world.

Comparison of the options

Based on the above results, all five silvicultural options were compared in terms of the economic benefit of timber and carbon sequestration in physical units (Figure 6). Here, EAA from the techno-economic assessment was used as the indicator for the economic benefit of timber, while the average of carbon stock was employed as the indicator for carbon sequestration. In general, Mea showed the highest economic benefit, while Mua performed best in terms of carbon sequestration. Specifically, in terms of the two P. massoniana monoculture options, Pm-pure-hd showed higher economic benefit but less carbon sequestration than Pm-pure-st. In terms of the monocultures of both species, Ch-pure showed higher economic benefit but less carbon sequestration than Pm-pure. In terms of the mixed stands versus the monocultures, Mea performed significantly better in economic benefit than the monocultures of both species (Pm-pure and Ch-pure). With respect to carbon sequestration, Mea performed significantly better than Ch-pure and slightly better than Pm-pure. Mua showed a significant advantage over others on carbon sequestration; it performed slightly better than Pm-pure on economic benefit but worse than Ch-pure and Mea.

Figure 6. Economic benefit and carbon sequestration of all silvicultural options

Note: Circles stand for the normalised sensitivity index (SI): red for timber price; blue for harvesting cost; and green for discount rate

Discussion

This study assessed the effects of five species-composition-oriented silvicultural options on timber value and carbon sequestration using an integrated assessment framework for two representative species in subtropical China. The silvicultural options, which focus on fast-growing species and valuable wood production, are of considerable interest in the region, and the assessment of silvicultural options could bring insights of value to various stakeholders. To enhance the robustness of the conclusions, we discuss below the reasons behind the results, the effects of variation of the silvicultural regime, and a ‘Pareto improvement’ comparison among the five options.

Carbon sequestration, timber production and the value of the five options

The five options fluctuated over time in terms of both stocking volume and carbon sequestration. This can mostly be explained by species’ characteristics and species-oriented silvicultural treatments. Pm-pure had a higher stocking volume than Ch-pure and Mea at the early and middle stages of the assessment period. This is a similar result to Qin et al. (2011), which can be explained by the faster growth of P. massoniana compared with C. hystrix (Qin et al. 2011). Mua showed the highest value at the late stage, which can be explained by the increased tree density and space utilisation efficiency resulting from the multilayer forest structure formed from planting C. hystrix under the canopy. Mua had a relatively large fluctuation over the period, which could be ascribed to the high-intensity thinnings over time. Ch-pure showed higher carbon sequestration than Pm-pure and Mea in the early stage. This result is similar to those of He et al. (2013) and Lu et al. (2014), which can be explained by the higher carbon storage in individual broadleaved trees compared with coniferous individuals (He et al. 2013; Cook et al. 2014; Lu et al. 2014). In the middle stage, multiple high-intensity thinnings in Ch-pure significantly changed its carbon sequestration volume. The high carbon sequestration of Mua at the late stage can be ascribed to its high plant density.

All options are economically acceptable, indicating the economic feasibility of social capital investment in planted-forest development in the region and providing support for China’s national forest policies, which encourage investment in forest development (State Forestry Administration of China and the Ministry of Finance of China 2019). Mea showed higher profitability than the monocultures, which is consistent with the study by Salek and Sloup (2012). The results could facilitate the development of mixed-species stands with higher profitability. The absolute NPVs of the five options may be higher than investment returns in other countries. This is because the market price for small-diameter wood is higher in China due to the huge demand arising from rapid economic development over recent decades. The profitability ranking of Pm-pure, Ch-pure and Mea in this study was consistent with a previous study using a 30-year time horizon (Zhang et al. 2019). In terms of absolute values, the IRRs of Pm-pure, Ch-pure and Mea in this study were generally consistent with and fall within the range obtained in the study by Zhang et al. (2019). The IRR of Pm-pure was slightly higher than the base IRR in that study, where it was 9–10%. The higher price of the large-dimension P. massoniana timber resulting from longer production cycles is the main reason for the increased IRR. However, the IRRs of Ch-pure and Mea are lower than in the study by Zhang et al. (2019), where the base IRR was 17% and 21%, respectively. The major reason is that higher prices were used for C. hystrix timber in that study (Cai et al. 2007; Zhang 2014).

Discount rate plays an important role in absolute economic performance. It is affected by multiple factors related to risk and time preferences and the duration of a project, and it may vary by time and country (Sauter and Mußhoff 2018). In China, a nominal discount rate of 6–8% is usually used in forestry departments (Market Supervision Administration of Guangdong Province of China 2019; Zhang et al. 2019). Deducting the average inflation rate in China of around 3%, the range of the real discount rate is 3–5% (National Bureau of Statistics 2019). Given the relatively long period in this study, a real discount rate of 3% as the base case is reasonable in China; this rate was also used in Su’s (2015) study. Taking the inflation rate into account, this rate is equivalent to a nominal discount rate of around 6%. This is close to 5%, which was the nominal discount rate used by Salek and Sloup (2012) in Viet Nam and by Knoke et al. (2014) in the tropical Andes in Ecuador, and the 8% used by Cubbage et al. (2007), Cubbage et al. (2010), Cubbage et al. (2014) and Cubbage et al. (2020) in their global timber investment research. It should be noted that profit-making entities may prefer a higher discount rate in NPV calculations.

‘Pareto improvement’ comparison among the five options

Of the two P. massoniana monoculture options, Pm-pure-hd showed higher economic benefit but less carbon sequestration compared with Pm-pure-st. In the current market in China, timber is priced based on diameter class rather than precise diameter. From the simulation output, the mean diameters for Pm-pure-st and Pm-pure-hd fell into two adjacent price levels, which resulted in a substantial difference in timber price. Thus, although dense planting might negatively affect diameter, the total economic benefit might be higher if the diameters remain at the same price level by properly adjusting planting density. Because China’s current domestic timber market is a seller’s market, and there is almost no domestic supply of large-dimension timber in the local market (e.g. with diameter >40 cm), the price of large-dimension timber with DBH 40 cm, 50 cm or even 60 cm is not distinguished from middle-dimension timber such as DBH 30 cm. Thus, there has been limited economic incentive to grow long-term large-dimension timber in practice due to the time preference, which in turn limits the local timber market.

In terms of the monocultures of the two species P. massoniana and C. hystrix, Ch-pure showed higher economic benefit but less carbon sequestration than Pm-pure. This can be attributed to the significantly higher price of C. hystrix timber, especially for large-dimension good-quality timber. The price gap results from the huge gap between the supply of and demand for good-quality, large-dimension timber in China (Cai et al. 2007).

Comparing the mixed stands with monocultures, Mea performed significantly better in terms of economic benefit than the monocultures Pm-pure and Ch-pure. Mea performed significantly better than Ch-pure in carbon sequestration and slightly better than Pm-pure. This can be attributed to the fact that the greater thinning of P. massoniana in Mea in the early stage facilitated the DBH growth of C. hystrix in the later stage. In addition, the greater thinning of P. massoniana in the early stage provided substantial short-term economic benefits. Thus, Mea balanced carbon sequestration as well as short- and long-term economic benefits.

Mua had a significant advantage in carbon sequestration but performed only slightly better than Pm-pure in economic benefit and worse than Ch-pure and Mea. Essentially, Mua made good use of the species’ characteristics of growing well at high densities, leading to higher carbon sequestration. But the high-intensity thinning in the early stage reduced the number of P. massoniana stems, and C. hystrix did not respond sufficiently in terms of DBH growth over 25 years, resulting in a reduced economic benefit for this option. Prolonging the time horizon might improve the economic performance of Mua.

The above comparison is in line with the concept of Pareto improvement, in which alternative A is defined as a Pareto improvement for alternative B if the switch from B to A harms nobody and increases benefits from at least one aspect (Just et al. 2008). Thus, in this study, Mea is a Pareto improvement for Ch-pure and Pm-pure, while Mua is a Pareto improvement for only Pm-pure.

The effects of the variation of silvicultural regimes

Only the abovementioned most feasible and reliable silvicultural regimes based on preliminarily examined field trials were modelled and simulated with PICUS, and differences in the silvicultural regime may affect timber value and carbon sequestration. Factors such as the species used, the mix of species, and the frequency, timing and intensity of thinning will affect the results. Species selection and composition are key factors with substantial influence on the ensuing treatments. Given the important role of P. massoniana in producing merchantable wood in a relatively short time, a lower thinning frequency or lower thinning intensity could be considered to an appropriate degree in the Pm-pure forest. This would increase the number of stems, which is likely to increase economic revenue in the current Chinese timber market, where timber price is less differentiated by size for this assortment type. Castanopsis hystrix is grown for large-sized high-value timber, and a higher thinning frequency or intensity could be considered to an appropriate degree in the Ch-pure forest. This would promote DBH growth and increase the economic value of the timber, although the trade-off between DBH growth and the number of stems brings some uncertainty, especially given the changing timber price. Mixing high-value broadleaved species with fast-growing coniferous species offers a means for meeting demand for both large-quantity and high-quality wood supply. However, the specific mix of species is worthy of a long-term field trial. Castanopsis hystrix can be mixed successfully with P. massoniana for afforestation, while Michelia macclurie Dandy also has potential to mix with P. massoniana. However, Mytilaria laosensis Lecte is considered an unsuitable species for mixing with C. hystrix because it grows faster in the early stage, which could substantially oppress C. hystrix, with the potential to cause mortality. As to the ratio of species in Mea, a higher proportion of P. massoniana with higher thinning intensity could be considered. This would likely increase short-term revenue by leveraging the fast growth of P. massoniana, but it would also increase the risk of oppressing the growth of C. hystrix, and fewer stems of C. hystrix would mean less carbon storage later. Determining the optimal mixture ratio range and the corresponding thinning schedule is worth further research. The rotation period could be extended: this could increase economic opportunities for Ch-pure, Mea and Mua because the unit price will increase as DBH increases for large-sized, high-value timber.

Comparison in the carbon market

The ecosystem service of carbon sequestration could be monetarily evaluated in the carbon market, with this value combined with the economic benefit obtained from timber production to enable further comparison between options. Given the instability and variance of current global carbon trading markets, three carbon price scenarios were considered: $5 tonne⁻¹, $25 tonne⁻¹, and $100 tonne⁻¹ (Chikumbo and Straka 2012). Table 3 shows carbon values and the combined values for these prices.

Table 3. The carbon values and the combined values by scenario

Scenario	$5 tonne^−1		$25 tonne^−1		$100 tonne^−1
	Carbon value($ ha^−1 y^−1)	Combined value($ ha^−1 y^−1)	Carbon value($ ha^−1 y^−1)	Combined value($ ha^−1 y^−1)	Carbon value($ ha^−1 y^−1)	Combined value($ ha^−1 y^−1)
Pm-pure-st	1035.67	1511.07	5178.37	5653.77	20 713.48	21 188.88
Pm-pure-hd	1025.21	1512.95	5126.07	5613.81	20 504.29	20 992.03
Ch-pure	948.70	1499.42	4743.48	5294.20	18 973.90	19 524.62
Mea	1043.01	1613.25	5215.07	5785.31	20 860.28	21 430.52
Mua	1082.10	1596.69	5410.50	5925.09	21 641.99	22 156.58

Ch = Castanopsis hystrix; hd = high density; Mea = even-aged mixture; Mua = uneven-aged mixture; Pm = Pinus massoniana; st = standard (normal) density.

For a carbon price of $5 tonne⁻¹, Mea had the highest combined value, while the Mua option had the highest combined value at carbon prices of $25 tonne⁻¹ and $100 tonne⁻¹. The assumed carbon prices would improve the economics of all silvicultural options, especially for the mixed forest. This indicates that the realisation of forest carbon trading in the market could be helpful for afforestation and reforestation.

Conclusion

All five silvicultural options meet the minimum profitability requirement, with varying advantages. The long-term profitability of the P. massoniana monoculture is in doubt due to its high sensitivity to increases in harvesting cost. Mea (even-aged mixture of P. massoniana and C. hystrix) is the least sensitive to changing markets. Mua (uneven-aged mixture of P. massoniana and C. hystrix) is expected to perform better economically if timber prices increase.

Among the five silvicultural options, Mua performed better than the other options for carbon sequestration. Mea showed higher combined benefits for timber production and carbon sequestration than pure stands of both species and had an additional advantage of balancing short- and long-term benefits over the forest growth period. A monoculture of C. hystrix achieved a better economic performance than P. massoniana stands. An appropriately higher density of P. massoniana could improve economic performance in the context of current timber pricing and markets in China.

This study employed an interdisciplinary analysis framework to assess stand-level silvicultural options regarding species composition involving the integration of preliminary field trials, simulations with a forest dynamics model, and techno-economic and social implication analysis. The integrated analysis framework could extend the current planted-forest management assessment approach in China, which would be helpful in supporting stakeholders in their strategies and decisions under the current forest management transition. Additionally, taking the two dominant but distinct species as cases, this study showed the positive performance of the preliminary field trials in general and provided useful insights for the future of multipurpose management of planted forests.

The study has several limitations. It emphasised the stand level due to limitations in spatial data availability, but it would be beneficial for ecosystem sustainability to further optimise forest management at the estate-/landscape level (Chikumbo et al. 2000). In addition, given the context of forest management in China, we emphasised the stand-level assessment of promising silvicultural options. Multi-objective optimisation could further advance stand-level silvicultural regimes by considering a wider plethora of options to provide broader insights into factors such as initial planting stock and thinning frequency, timing and intensity (Chikumbo 2012). Multi-objective optimisation was not included in this research, due largely to a lack of quantitative techniques in forest management in China, such as forest growth functions research (particularly for Mea and Mua). Thus, we only considered promising silvicultural options preliminarily examined in field trials because the field trials could provide evidential support for the simulations. Admittedly, this is a conservative strategy, and we plan to undertake further efforts to overcome existing limitations and to conduct multi-objective optimisation studies at the stand and estate levels as a next step.

Acknowledgements

We appreciate the ECTF and Prof. Wenfu Guo’s support for the field investigation and Prof. Lin Qin’s support in laboratory-analysis data. We also appreciate the comments of Prof. Runsheng Yin and anonymous reviewers, and the editors’ support, which helped improve the manuscript.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This study was partially supported by Chinese Academy of Forestry under Grant [CAFYBB2020MC002]; National Natural Science Foundation of China under Grant [31270681] and Ministry of Science and Technology of China under Grant [201404201].