Do Regional Differences in Forest Distribution Affect Residents’ Preferences for Forest Ecosystem Services?

Abstract

This study investigates how residents of Jeollabuk-do, South Korea perceive and emphasize forest ecosystem services, focusing on whether the distribution of forests between the eastern mountainous areas and the western lowlands influences their preferences. We applied the Choice Experiment (CE) method to gauge willingness to pay (WTP) for seven key forest ecosystem services and a tax-related attribute. Between 10 and 14 August 2023, we conducted an online survey with 400 participants (out of 4177 invited). Only 20% of respondents were aged 50 or older, despite this age group making up nearly half of the region’s population. On the surface, no significant statistical differences appeared between the two areas regarding overall preferences, perhaps unsurprising, given their shared administrative framework and cultural background. However, a closer look at marginal WTP values uncovered clear regional priorities: residents in the eastern region placed a higher value on erosion control (KRW 23,559–33,109), while those in the west assigned greater priority to biodiversity improvement (KRW 30,225–43,961). Although the sheer distribution of forests may not drastically reshape general preferences, the specific forest characteristics of each area still shape what people care about most. These insights underscore the significance of tailoring forest ecosystem management policies to fit local needs, such as prioritizing erosion control in hilly regions and enhancing biodiversity in flatter areas.

1. Introduction and Theoretical Framework

Currently, 31% of the Earth’s land surface is covered by forests, which are key contributors to the biodiversity preservation of flora and fauna and act as carbon sinks in response to climate change [1]. These forests supply essential resources to residents and deliver non-use advantages like enhanced air quality and biodiversity conservation to people regardless of proximity [2,3]. Across many countries, forests are widely acknowledged for their economic contributions, including job creation, and their non-economic roles, such as environmental protection and the preservation of cultural heritage. As a result, they serve as essential components of ecological welfare and sustainable development for human societies [4,5,6].

In the Republic of Korea, forests cover an area of 6,348,834 hectares, accounting for 63.2% of the national land area [7]. This forest coverage rate ranks fourth among OECD countries, following Finland, Japan, and Sweden [8]. Many people have a close relationship with forests and enjoy their various benefits. These benefits are ecosystem services [9,10]. Smith et al. [11] defined ecosystem services as the full range of ecological, social, and economic benefits people obtain from nature. Sannigrahi et al. [12] defined ecosystem services as contributions from ecosystems that directly or indirectly improve human welfare and survival. The types of ecosystem services are broadly categorized into four: provisioning services, which provide water, food, and medicinal resources; regulating services, which improve environmental quality through carbon storage, erosion control, and other means; cultural services, which impact tourism, education, and recreation; and supporting services, which form the basis for different services by supporting genetic diversity and species habitats [13].

While international acknowledgment of the value of forests has grown, forests in many regions remain threatened, underscoring the critical necessity for conservation strategies informed by research identifying priority ecosystem services. Although previous global and national studies have evaluated ecosystem services, Jeollabuk-do, South Korea, offers an exceptionally unique case for studying the relationship between forest distribution and local preferences. With its eastern mountainous, high-forest-cover area and western flat, low-forest-cover area, Jeollabuk-do embodies sharp regional contrasts rarely examined in comparable studies.

Preferences for ecosystem services vary among individuals [14] and differ based on the demanders’ socio-cultural characteristics and economic levels [15,16]. Preferences for ecosystem services also vary depending on the living environment [17,18,19]. Urban residents with less access to forests prefer regulating and cultural services and a high-quality natural environment. In contrast, rural residents, who closely relate to forests, tend to prefer provisioning services [19]. While it is essential to provide ecosystem services that meet the preferences of the demanders [19,20,21], the ecosystem services offered can vary based on environmental characteristics such as forest distribution [22]. However, little research has focused on preferred ecosystem services regarding local forest distribution, and forest management plans tailored to these preferences remain underdeveloped. This study addresses these gaps by examining residents’ preferences for different types of forest ecosystem services [23,24], expanding upon prior works on willingness to pay (WTP) for ecosystem services.

Affordance, or action potential, refers to the relationship between the environment and human behavior [25]. It was systematized by American ecological psychologists [26], who defined it as ‘everything that the environment offers and stimulates to humans’. Gibson [26] highlighted the significance of the interactions between personal, social, and physical environments in environmental stimuli, suggesting that the climate influences humans positively or negatively. However, affordance is subjective, meaning behavior varies depending on how individuals perceive ecological stimuli [27].

Heft [28] reinterpreted traditional affordance, asserting that environments providing specific affordances, like climbing a tree or swimming, actively guide and shape individual behavior. He also noted that if individuals’ sociocultural backgrounds differ, the resulting behaviors influenced by affordances will also vary. Andersson and McPhearson [29], Building on Chemero [30], defined affordance as ‘the relationship between the ability to perceive and act and the characteristics of the environment’, arguing that affordance is influenced not only by a single attribute of the environment but also by internal and external properties of the ecological context and persons’ sociocultural elements.

Given the interaction between human behavior and cognitive processes with the environment, much research has been conducted on the affordance of natural environments. Laaksoharju [31] analyzed the affordance of trees for children aged 7–12 in Finland, reporting that the affordances provided by trees varied with the children’s growth. Sharma-Brymer et al. [32] noted that the affordances of school forests offer children opportunities for exploration and play, helping them identify local communities and cultures. Additionally, Guardini [33] reported that local forests’ affordances of environmental elements like trees and rocks vary based on individuals’ preferences and abilities. Vilar et al. [34] observed that the affordances provided by natural environments differ from those offered by urban environments. Thus, considering the affordances people perceive from their forest environments, it is necessary to design forest governance approaches that reflect regional affordance differences.

Perceptions and preferences for ecosystem services vary according to specific regions, regional characteristics, and individual values and experiences [35,36]. Personal preferences can be estimated through willingness to pay (WTP) and willingness to accept (WTA), and expressing these in economic terms is an essential method for demonstrating the significance of ecosystem services in policy [37].

Methods for evaluating the value of ecosystem services include stated preference, revealed preference, monetary valuation, benefit–cost analysis, cost-based, and value transfer techniques [38]. Among these, revealed preference is advantageous for estimating the value of cultural and supporting services, which is challenging to quantify because it presents hypothetical scenarios to respondents through surveys [39,40]. Revealed preference methods are divided into the Contingent Valuation Method (CVM) and the Choice Experiment (CE). Although the CVM has the considerable benefit of assessing non-market values [41], it can only estimate WTP for one attribute of the environmental good. In contrast, the CE has the advantage of estimating WTP for various characteristics of the ecological good [42]. Jo et al. [43] used CE to assess preferences for types of urban forests in Seoul and reported that urban residents preferred biodiversity enhancement as an ecosystem service. They also emphasized that policymakers and forest managers should consider citizens’ preferences in urban forest management to enhance residents’ satisfaction and ecological welfare. Local forests have a close relationship with residents and positively impact them [44,45]. Therefore, the management and planning of local forests should be based on identifying residents’ preferences, as these are closely related to their lives.

This research seeks to determine preferences for managing ecosystem services according to the forest distribution ratio. This will furnish crucial data for formulating management plans and policies for ecosystem services based on the forest distribution ratio, filling key gaps in regional and international forest ecosystem service research.

2. Materials and Methods

2.1. Study Site

The study area for this research is Jeollabuk-do, South Korea, focusing on residents living in this region. Jeollabuk-do accounts for 8.04% of South Korea’s total area and features a slightly elongated east–west orientation. The eastern region is mountainous, while the western region is primarily flat plains. Due to the topographical influences, there is a relatively significant climate difference between the eastern and western parts compared to the north and south [46]. The eastern part of Jeollabuk-do consists mainly of forests, forming a mountainous area, whereas the western part is predominantly flat, creating a plains area. The forest distribution status in Jeollabuk-do is as follows (Figure 1).

According to Korea Forest Services [47], the forest stock in the eastern region is significantly higher than in the western region, with the eastern region having many broad-leaved forests and the western region having many coniferous forests. This indicates a clear difference in forest types by region. These differences in forest stock and forest types lead to variations in the ecosystem services provided. However, Jeollabuk-do encounters difficulties in executing forest policies due to the relatively reduced scale of forest administration compared to other administrative regions [48].

Analyzing preferences for ecosystem services according to forest distribution in Jeollabuk-do, where the characteristics of the western and eastern regions are distinct, can optimize management to provide the forest ecosystem services preferred by residents, thereby increasing efficiency.

2.2. Choice Experiment

The Choice Experiment (CE) used to analyze preferences is a type of stated preference method that evaluates the economic value of environmental goods. Preference methods include the Contingent Valuation Method (CVM) and the CE. While the CVM estimates the value of a single attribute based on hypothetical scenarios, the CE presents alternatives combined with various attributes and levels, allowing for comparing preferences for each attribute [49,50].

2.2.1. Choice Experiment Theoretical Model

U_ni represents the utility of respondent n when choosing alternative i, and V_ni is defined as shown in Equation (1) [51].

$${\mathit{U}}_{\mathit{n}\mathit{i}}={\mathit{V}}_{\mathit{n}\mathit{i}}+{\mathit{e}}_{\mathit{n}\mathit{i}}$$

(1)

Eni denotes the unobservable utility component while V_ni comprises the linear sum of m attribute vectors X_i, as shown in Equation (2). The characteristic theory of demand best captures the nature of the choice experiment method, which assumes that goods are composed of multiple attributes [52].

$${\mathit{V}}_{\mathit{n}\mathit{i}}=\sum _{\mathit{k}=1}^{\mathit{m}}{\mathit{\beta }}_{\mathit{k}}{\mathit{X}}_{\mathit{i}}$$

(2)

The probability that the nth respondent chooses alternative i over j is given by Equation (3). This indicates that the utility from choosing i is greater than the utility from choosing j.

$${\mathit{P}}_{\mathit{n}\mathit{i}}=\mathbf{P}\mathbf{r}\left[\right({\mathit{V}}_{\mathit{n}\mathit{j}}+{\mathit{e}}_{\mathit{n}\mathit{j}})>{(\mathit{V}}_{\mathit{n}}+{\mathit{e}}_{\mathit{n}\mathit{i}})]$$

(3)

In Equation (4), V represents the contribution of attribute i to the utility, and the probability part is assumed based on the Gumbel distribution or Type I extreme value distribution, estimating the parameter values accordingly [53].

$$P_{ni}={\displaystyle \frac{e^{V_{ni}}}{\sum _j{e}^{V_{nj}}}}$$

(4)

Beyond these general formulas, the conditional logit, mixed logit, and latent class models integrate these probabilities differently. The conditional logit assumes homogeneous preferences and calculates choice probabilities using Equation (4). The mixed logit extends this by allowing β coefficients to vary randomly across respondents, capturing unobserved preference heterogeneity. The latent class model groups respondents into classes with distinct preference structures, estimating class-specific β values and the probability of class membership. Although this study primarily used the conditional logit, it tested the independence of irrelevant alternatives (IIA) assumption using the Hausman test, which confirmed that the IIA assumption was satisfied (

Table 1. Comparison of Conditional Logit Model Estimates (Distribution Ratios) for IIA Testing.

Region	Eastern Region		Western Region
	Model 1 (3 Alternatives)	Model 2 (2 Alternatives)	Model 1 (3 Alternatives)	Model 2 (2 Alternatives)
Water provision (Reference: <25% Broadleaf)
Moderate	0.0239	0.0366	−0.0193	−0.0568
Rich	0.3290	0.3551	−0.0111	−0.0540
Timber provision (Reference: <25% Conifer)
Moderate	−0.0110	−0.0348	0.0573	0.0371
Rich	0.0368	0.0222	0.1606	0.1665
NTFP provision (Reference: <25% Short-term Income or Fruit Trees)
Moderate	0.0513	0.0413	0.0805	0.0432
Rich	−0.0245	−0.0223	−0.0164	−0.0468
Erosion Control (Reference: 33% Understory Cover)
Moderate	0.4747	0.4823	0.1089	0.1578
Rich	0.6670	0.6872	−0.0101	0.0216
Carbon Storage (Reference: Low Canopy Density)
Moderate	0.3222	0.3003	−0.0273	−0.0669
Rich	0.5576	0.5361	−0.1039	−0.1211
Recreation (Reference: Hiking)
Moderate	0.2596	0.2748	0.0210	0.0036
Rich	0.3869	0.3865	0.0535	0.0626
Biodiversity (Reference: Poor Species Diversity and Richness)
Moderate	0.3694	0.3898	0.0463	0.0506
Rich	0.5740	0.5877	0.0750	0.0794
Tax	−0.0201	−0.0194	0.0010	0.0014

Note: Coefficient estimates are reported. No significance tests were conducted; therefore, significance stars are not applied.

).

2.2.2. Sampling and Survey Design

The survey targeted residents of Jeollabuk-do. According to Statistics Korea [54], the total population of Jeollabuk-do as of 2022 was 1,774,248, accounting for 3.43% of South Korea’s total population. Considering a sample error of 95% and a confidence level of 4–5%, a sample size of 385–601 was deemed appropriate. We collected 400 valid responses using stratified sampling to approximate the demographic structure of the population, including gender, age, and urban/rural distribution. A designated research agency conducted the survey. However, older adults (aged 50 and above) were underrepresented in our sample (20% vs. 48.8% in the population), introducing potential age bias. Moreover, only seven respondents reported working in agriculture or forestry, potentially limiting our ability to analyze preferences by occupation.

The survey was conducted online between 10 and 14 August 2023, immediately after the monsoon season, which may have temporarily increased public attention to erosion control. This limitation will be addressed in future studies. While the online method facilitated broad participation, it may have introduced hypothetical bias.

The choice experiment included eight attributes: seven representing forest ecosystem services and one representing a tax attribute for estimating marginal willingness to pay. These attributes were derived from previous studies, such as the National Institute of Forest Science [55]. A literature review determined crucial forest ecosystem service attributes in South Korea, including water provision, timber provision, NTFP provision (NTFP) supply, soil conservation, carbon sequestration, natural disaster mitigation, recreational provision, and biodiversity improvement. Similarly, Jo et al. [43] examined preferences for urban forest ecosystem services and identified comparable attributes, including water provision, timber provision, NTFP supply, erosion control, carbon storage, water quality improvement, air quality improvement, recreation, nature education, and biodiversity improvement. The levels of each attribute were determined based on previous empirical findings.

Studies suggest broadleaf forests have higher soil moisture retention and runoff control for water provision than conifers [56,57,58]. Based on FAO [59], forests with less than 25% broadleaf trees are classified as ‘poor’, 25%–75% as ‘moderate’, and more than 75% as ‘rich’. Coniferous trees are generally more commercially valuable for timber provision than broadleaf trees [60,61]. Thus, forests with less than 25% conifers are considered ‘poor’, 25%–75% as ‘moderate’, and over 75% as ‘rich’ [59]. For NTFP supply, the proportion of areas cultivated with fruit trees or short-term income forest products indicates supply levels [43,62]. Following FAO [59], less than 25% is ‘poor’, 25%–75% is ‘moderate’, and over 75% is ‘rich’. For soil erosion control, previous studies highlight the significance of vegetation cover in reducing erosion [63,64,65]. Accordingly, forests with 33% understory cover are rated as ‘poor’, 66% as ‘moderate’, and 99% as ‘rich’ [63]. For carbon storage, canopy density is positively associated with carbon sequestration potential [66,67,68,69]. According to Korea Forest Service standards, forests with 40% or less canopy density are ‘poor’, 41%–70% ‘moderate’, and above 71% ‘rich’. For recreational services, the diversity of available mountain activities was used as a proxy [18]. Forests offering only hiking are considered ‘poor’, those with hiking, camping, and climbing are ‘moderate’, and those offering additional activities such as MTB and paragliding are ‘rich’. For biodiversity, species diversity and richness were used as classification criteria [50,70,71,72,73], with levels categorized as ‘poor’, ‘moderate’, and ‘rich’. A tax attribute was adopted to show willingness to pay instead of entrance fees or donations to reflect better the public good nature of ecosystem services [74,75]. Referring to Jo et al., Moeltner et al. [43,76], tax levels were set at KRW 10,000 (‘poor’), KRW 20,000 (‘moderate’), and KRW 40,000 (‘rich’) per household per year.

Although 6561 (3⁸) combinations are possible with the eight attributes and three levels each, an orthogonal design was used to reduce complexity and multicollinearity [77,78]. This design yielded 27 alternatives randomly paired to create 351 (₂₇C₂) combinations. Each respondent was presented with four sets of choices. Each set included two attribute-based alternatives and a third ‘neither’ option to increase the likelihood of representing valid preferences [79]. Graphical illustrations and explanations of the attribute levels were provided to aid respondents’ understanding.

3. Results

3.1. Characteristics of Respondents

This study investigated regional residents’ preferences for forest ecosystem services and willingness to pay (WTP). The research and data collection were conducted following approval from the Institutional Review Board (IRB) (WKIRB-202307-SB-057). Data collection took place from 10 and 14 August 2023. The survey was emailed to residents aged 19 and older who had lived in Jeollabuk-do for more than one year. A total of 4177 emails were sent, and 790 residents accessed the survey link. Excluding those who did not meet the criteria, exceeded the sample size, or abandoned the survey, 490 responses were initially obtained, and after excluding 90 insincerely answered responses, 400 valid data sets were retained (

Table 2. Survey Response Status and Valid Sample Selection Process.

Status				Case Number (Person)	Proportion (%)
Survey link successfully sent	Received the survey link unchecked			3387	81.1
	Received survey link, checked	Out	Not targeted subject	83	2.0
			Exceeded the targeted subject	154	3.7
			Incomplete response	63	1.4
		Response completed	Untrustworthy data	90	2.2
			Completed response	400	9.6
Total				4177	100.0

Note: Coefficient estimates are reported. No significance tests were conducted; therefore, significance stars were not applied.

).

The socioeconomic characteristics of the sample were as follows. When compared to the population census and the social survey report of Jeollabuk-do, the respondents’ characteristics, except for gender, did not perfectly reflect the residents of Jeollabuk-do. This discrepancy is a limitation of online surveys [80], as people aged 50 and above, who have lower internet access and usage rates due to aging, find it challenging to participate in online surveys, resulting in a lower proportion of respondents over 50. To address this limitation in future research, methods such as post-stratification weighting or conducting additional in-person surveys in small, rural communities should be considered, as these groups may have limited access to online surveys (

Table 3. Socioeconomic Characteristics of Respondents Compared to Jeollabuk-do Population.

Category			Sample Size (%)	Proposition of Jeollabuk-do’s Total Population, as of 2020 (%)
Variables		Code
Age	20s	2	26.5	15.7
	30s	3	26.8	11.2
	40s	4	26.8	10.0
	50s and above	5	20.0	48.8
Sex	Male	1	50.0	49.8
	Female	2	50.0	51.1
Marriage	Single	0	52.0	12.0
	Married	1	48.0	88.0
Number of children	None	0	15.5	-
	1	1	13.3
	2	2	16.8
	3 above	3	2.5
Education	Less than a middle school graduate	1	0.8	48.1
	High school graduate	2	12.3	29.2
	Attended or graduated from a university	3	73.0	21.0
	Graduate school student or a graduate degree holder	4	14.0	1.7
Monthly household income	Less than KRW 1,000,000	0	6.3	21.7
	KRW 1,000,000 to less than KRW 2,000,000	1	5.5	18.3
	KRW 2,000,000 to less than KRW 3,000,000	2	21.0	20.2
	KRW 3,000,000 to less than KRW 4,000,000	3	16.8	14.7
	KRW 4,000,000 to less than KRW 5,000,000	4	14.3	9.5
	KRW 5,000,000 to less than KRW 6,000,000	5	14.3	6.4
	KRW 6,000,000 to less than KRW 7,000,000	6	8.0	3.5
	KRW 7,000,000 to less than KRW 8,000,000	7	6.3	1.8
	KRW 8,000,000 or more	8	7.8	3.9
Number of forest visits in the past year	None	1	24.8	-
	More than once	2	75.3
Purpose of visit	Forestry activities	1	0.3
	Relaxation/walking	2	46.5
	Nature experience/education	3	1.0
	Physical activity	4	27.0
	Others	5	0.5

Source: adapted from Jo et al. [43].

).

3.2. Results of Estimation

The survey was conducted twice to examine residents’ forest ecosystem service preferences in the eastern and western regions. Each respondent was presented with three alternatives per set, repeated four times, resulting in 1600 data points from eastern region residents and 1600 from western region residents, totaling 3200 data points. The analysis was performed using R and analyszed using conditional logistic regression [81]. The study revealed an absence of notable differences in forest ecosystem service preferences between the eastern and western regions. A log-likelihood ratio test confirmed no significant differences between the groups, implying that residents’ preferences do not vary based on forest distribution ratios. Additionally, hypothesis testing across attributes showed no significant differences among them.

Despite these non-significant differences, investigating possible reasons why the hypothesis was not supported is essential. Possible explanations include collective cultural values, limited direct interaction with forests, or other confounding factors, such as the high proportion of urban residents (77.88% of Jeollabuk-do’s total population [48]), which may homogenize regional preferences. Additionally, the relatively small number of respondents engaged in agriculture or forestry (only seven individuals) may limit the capture of preference heterogeneity based on economic activities [19].

The survey design used 27 orthogonally designed profiles paired into four choice sets per respondent. Considering the cognitive burden imposed by multi-attribute choice tasks, a prior pilot test was conducted to assess the clarity and feasibility of the survey design. The pilot involved pre-testing with a small sample of respondents, which helped refine the number of tasks and attribute explanations to minimize respondent fatigue. Additionally, comprehension checks were included in the primary survey to ensure that lay participants could comprehend intricate attribute levels, such as ‘understory vegetation cover of 33%, 66%, 99%’ for erosion control and ‘low/medium/high’ canopy densities for carbon storage (

Table 4. Conditional Logistic Regression Results on Forest Ecosystem Service Preferences of Eastern and Western Jeollabuk-do Residents.

Ecosystem Service (Attributes)	Attributes and Levels	Eastern (E)	Western (W)	Hypothesis Testing (βE = βw) χ2 (1)
Alternative specific constant (ASC)	-	1.614 *	1.377 *	-
Water provision (Standard level: less than 25% of deciduous trees)	25% to less than 75% of deciduous trees	0.024	0.214 *	2.001
	More than 75% of deciduous trees	0.329	0.339 *	0.006
Timber provision (Standard level: less than 25% of coniferous trees)	25% to less than 75% of coniferous trees	−0.011	−0.053	0.100
	More than 75% of coniferous trees	0.037	0.125	0.449
NTFP provision (Standard level: less than 25% of forest products and fruit tree planting)	25% to less than 75% of Forest products and fruit trees planting	0.051	−0.140	2.066
	More than 75% of Forest products and fruit trees planting	−0.024	−0.088	0.223
Erosion control (Standard level: area covered by the forest 33%)	The area covered by the forest is 66%	0.475 ***	0.217 ***	2.256
	The area covered by the forest is 99%	0.667 ***	0.510 ***	1.385
Carbon storage (Standard level: low canopy density)	Medium canopy density	0.322 ***	0.310 **	0.008
	High canopy density	0.558 ***	0.521 ***	0.073
Recreation (Standard level: trekking)	Trekking, camping, and climbing	0.260 ***	0.235 *	0.035
	Trekking, camping, climbing, MTB, paragliding, etc.	0.387 ***	0.246 *	1.101
Biodiversity improvement (Standard level: poor)	Average	0.369 ***	0.550 ***	1.764
	Rich	0.574 ***	0.799 ***	0.095
WTP for forest ecosystem service (Tax) (KRW/household/year)	-	−0.020 ***	−0.018 ***	-

***: p < 0.001, **: p < 0.01, *: p < 0.05.

).

The study did not find statistically significant differences in preferences, contrasting with prior research indicating that preferences for forest ecosystem services vary according to the natural context in which individuals reside [82,83]. However, Williams et al. [84] found that while there could be differences in preferences for forest ecosystem services between urban and rural areas, these differences do not apply to all residents. Furthermore, Lapointe et al. [19] suggested that differences in economic activities between urban and rural residents could lead to varying perceptions of ecosystem services. In this study, only seven respondents were engaged in agriculture or forestry, which is a tiny proportion. This could explain the lack of statistically significant differences between the preferences of eastern and western region residents.

Nonetheless, observable trends emerged in preference rankings between the two regions. Eastern region residents prioritized erosion control, biodiversity improvement, and carbon storage, whereas western region residents prioritized biodiversity improvement, carbon storage, and erosion control. These findings align with previous research suggesting a global shift toward prioritizing regulating and supporting services over provisioning services [1,43,85]. However, the statistical robustness of these trends is limited, and future studies should validate WTP findings against actual behavioural data to reduce hypothetical bias.

3.3. Forest Ecosystem Services of Eastern Regions

The preferences of eastern region residents for forest ecosystem services showed statistically significant changes in all attributes except for water provision (‘poor’ to ‘moderate’), timber provision, and NTFP provision, influencing respondents’ choices. The average marginal WTP amounts varied with changes in ecosystem service attributes, with erosion control, biodiversity improvement, carbon storage, recreation, water provision, NTFP supply, and timber provision in that order. Among all ecosystem service attributes, erosion control had the highest marginal WTP, with households willing to pay KRW 23,559 per year for an improvement from ‘poor’ to ‘moderate’ and KRW 33,109 from ‘poor’ to ‘rich’. For biodiversity improvement and carbon storage, the marginal WTP amounts were KRW 28,492 and KRW 27,678, respectively, for an improvement from ‘poor’ to ‘moderate’, and KRW 18,337 and KRW 15,991 for an improvement from ‘poor’ to ‘rich’ (

Table 5. Conditional Logistic Regression and Marginal Willingness to Pay (MWTP) for Forest Ecosystem Service Attributes among Eastern Jeollabuk-do Residents.

Ecosystem Service (Attributes)	Levels	Coefficient	p > z	MWTP (Unit: KRW)
				Mean	95% CI
Alternative specific constant (ASC)	-	1.614 *	0.000	-	-
Water provision (Standard level: less than 25% of deciduous trees)	25% to less than 75% of deciduous trees	0.024	0.781	1185	−7332	9844
	More than 75% of deciduous trees	0.329	0.000	16,329	8020	27,410
Timber provision (Standard level: less than 25% of coniferous trees)	25% to less than 75% of coniferous trees	−0.011	0.895	−0.550	−9122	8011
	More than 75% of coniferous trees	0.037	0.666	1829	−6647	10,654
NTFP provision (Standard level: less than 25% of forest products and fruit tree planting)	25% to less than 75% of Forest products and fruit trees planting	0.051	0.553	2545	−5862	11,851
	More than 75% of Forest products and fruit trees planting	−0.024	0.779	−1216	−10,006	7752
Erosion control (Standard level: area covered by the forest 33%)	The area covered by the forest is 66%	0.475 ***	0.000	23,559	14,217	36,638
	The area covered by the forest is 99%	0.667 ***	0.000	33,109	22,894	48,510
Carbon storage (Standard level: low canopy density)	Medium canopy density	0.322 ***	0.000	15,991	7366	27,217
	High canopy density	0.558 ***	0.000	27,678	18,294	41,565
Recreation (Standard level: trekking)	Trekking, camping, and climbing	0.260 ***	0.003	12,884	4321	23,585
	Trekking, camping, climbing, MTB, paragliding, etc.	0.387 ***	0.000	19,202	10,468	31,211
Biodiversity improvement (Standard level: poor)	Average	0.369 ***	0.000	18,337	9565	30,345
	Rich	0.574 ***	0.000	28,492	18,902	42,470
WTP for forest ecosystem service (Tax) (KRW/household/year)	-	−0.020 ***	0.000	-	-	-

Pseudo R² = 0.3086, ***: p < 0.001, *: p < 0.05.

).

While most attributes showed higher marginal WTP amounts with improved levels, some levels of timber provision and NTFP supply, as well as taxes, had negative values, indicating a negative influence on respondents’ choices. Specifically, timber provision showed a negative WTP for an improvement from ‘poor’ to ‘moderate’, and NTFP supply showed a negative WTP for an improvement from ‘poor’ to ‘rich’, suggesting respondents were resistant to paying higher taxes for provisioning services.

While most attributes showed higher marginal WTP amounts with improved levels, some levels of timber provision and NTFP supply, as well as taxes, had negative values, indicating a negative influence on respondents’ choices. Negative WTP values can economically indicate that respondents perceive the service or its associated costs as a burden or a reduction in their welfare (disutility). For example, respondents may view increased timber or NTFP supply as offering minimal personal benefit while increasing public costs (e.g., taxes), lowering their utility. Alternatively, negative WTP estimates may reflect estimation noise rather than valid negative preferences, especially when confidence intervals are wide or include zero, suggesting model fit limitations or insufficient statistical strength.

The trends in respondents’ preferences, such as favouring erosion control, biodiversity improvement, and carbon storage over timber and NTFP supply, can be attributed to changing perceptions of forests [86,87]. While previous studies reported that rural residents emphasize provisioning services [88,89], recent research by Khaing and Youn [86] and Weinbrenner et al. [87] suggests that a shift from utilization-focused to conservation-focused forest management has led to a preference for regulating services over provisioning services. Additionally, the eastern region of Jeollabuk-do has seen less development and consists mainly of mountainous areas. The survey was conducted immediately after the summer monsoon season, which likely heightened the preference for erosion control [90]. This supports previous findings that residents who have experienced natural disasters tend to be more aware of climate change risks [91]. Thiemann et al. [92] reported that perceptions of ecosystem services can change with climate change awareness.

3.4. Forest Ecosystem Services of Western Region

The preferences of western region residents for forest ecosystem services were statistically significant for all attribute levels except for timber provision and NTFP provision, which influenced respondents’ choices. The marginal willingness to pay (WTP) amount for changes in ecosystem service attributes was ranked as follows: biodiversity improvement, carbon storage, erosion control, water provision, recreation, timber provision, and NTFP supply. Among all ecosystem service attributes, biodiversity improvement had the highest marginal WTP, with households willing to pay KRW 30,225 per year for an improvement from ‘poor’ to ‘moderate’ and KRW 43,961 for an improvement from ‘poor’ to ‘rich’. For carbon storage and erosion control, the marginal WTP amounts were KRW 17,058 and KRW 14,910, respectively, for an improvement from ‘poor’ to ‘moderate’, and KRW 28,661 and KRW 28,029 for an improvement from ‘poor’ to ‘rich’ (

Table 6. Conditional Logistic Regression and Marginal Willingness to Pay (MWTP) for Forest Ecosystem Service Attributes among Western Jeollabuk-do Residents.

Ecosystem Service (Attributes)	Levels	Coefficient	p > z	MWTP (Unit: KRW)
				Mean	95% CI
Alternative specific constant (ASC)	-	1.377 ***	0.000	-	-	-
Water provision (Standard level: less than 25% of deciduous trees)	25% to less than 75% of deciduous trees	0.214 *	0.039	11,770	697	26,620
	More than 75% of deciduous trees	0.339 **	0.001	18,644	7360	35,776
Timber provision (Standard level: less than 25% of coniferous trees)	25% to less than 75% of coniferous trees	−0.053	0.603	−2905	−14,661	8343
	More than 75% of coniferous trees	0.125	0.213	6877	−4072	19,635
NTFP provision (Standard level: less than 25% of forest products and fruit tree planting)	25% to less than 75% of Forest products and fruit trees planting	−0.140	0.167	−7721	−20,290	3480
	More than 75% of Forest products and fruit trees planting	−0.088	0.391	−4847	−17,972	6222
Erosion control (Standard level: area covered by the forest 33%)	The area covered by the forest is 66%	0.217 ***	0.009	14,910	3432	30,781
	The area covered by the forest is 99%	0.510 ***	0.000	28,029	15,770	47,745
Carbon storage (Standard level: low canopy density)	Medium canopy density	0.310 **	0.003	17,058	5497	33,718
	High canopy density	0.521 ***	0.000	28,661	16,069	49,722
Recreation (Standard level: trekking)	Trekking, camping, and climbing	0.235 *	0.020	12,901	1823	26,452
	Trekking, camping, climbing, MTB, paragliding, etc.	0.246 *	0.017	13,533	2429	28,059
Biodiversity improvement (Standard level: poor)	Average	0.550 ***	0.000	30,225	17,445	51,707
	Rich	0.799 ***	0.000	43,961	29,160	71,818
WTP for forest ecosystem service (Tax) (KRW/household/year)	-	−0.018 ***	0.000	-	-	-

Pseudo R² = 0.2779, ***: p < 0.001, **: p < 0.01, *: p < 0.05.

).

Like the eastern region, western region residents did not prefer timber and NTFP supply. However, the most preferred ecosystem services differed between the regions. While eastern region residents preferred erosion control the most, western region residents preferred biodiversity improvement. According to McDonald [93], urban residents preferred conserving the natural environment more than rural residents. Furthermore, Ma et al. [94] and Yang et al. [95] found that urban residents are more aware of the health impacts of ecosystem destruction and environmental pollution. In contrast, rural residents are less aware of these impacts. The western region of Jeollabuk-do consists mainly of flat land and has undergone more development than the mountainous eastern region. Biodiversity improvement is a highly preferred ecosystem service regardless of country or region [96,97]. However, western region residents showed a stronger preference for Biodiversity improvement than eastern region residents, corresponding with prior research indicating that urban dwellers prefer natural environment conservation more [93].

Although most attributes showed higher marginal WTP with improved levels, the negative WTP estimates observed for some levels of timber provision, NTFP supply, and taxes necessitate meticulous interpretation. Negative WTP can economically or behaviorally indicate perceived disutility, meaning respondents view the proposed change as decreasing their welfare, often due to added costs without perceived personal benefit. For example, increased timber or NTFP supply might be seen as commercial or extraction-focused, which does not align with conservation-oriented preferences. However, it is also possible that some negative estimates arise from estimation noise or statistical uncertainty, especially when confidence intervals are wide or include zero. Therefore, negative WTP signals caution or resistance, but it should be contextualized within the estimates’ ecological meaning and statistical precision.

4. Discussion

This study was based on the hypothesis that there would be differences in ecosystem service preferences according to the distribution ratio of forests in Jeollabuk-do. The study surveyed residents aged 19 and older who had lived in Jeollabuk-do for over a year. Using a choice experiment analyzed through the conditional logit model, it investigated their marginal willingness to pay (WTP) to improve forest ecosystem services. Previous studies have often estimated preferences for a single type of forest ecosystem service [98,99] or focused on urban populations [43,85]. However, this research is enhanced by examining preferences that consider the distribution ratio of forests and the ecosystem services provided by regional forests.

In light of the preliminary hypothesis, the findings indicated no statistically significant differences in ecosystem service preferences between eastern and western regions. This finding aligns in context with Jo et al. [43], which reported no significant difference in ecosystem service preferences between urban neighborhood parks and mountainous parks in Seoul, and Jo et al. [16], which analyzed asymmetric but sometimes similar stakeholder preferences for local forest ecosystem services across South Korea. This suggests that rather than simple geographic or physical differences, collective cultural and social backgrounds, a high urbanization rate (77.88%) [48], and everyday policy experiences within Jeollabuk-do may have integratively shaped residents’ perceptions and preferences for ecosystem services. Thiemann et al. [92] also reported that sociocultural factors better explain perceptions of ecosystem services than environmental factors, providing important context for interpreting the present results.

Although no significant differences were detected, notable trends emerged in the priority of services by region. Eastern region residents prioritized erosion control (highest marginal WTP: KRW 23,559–33,109), while western region residents prioritized biodiversity improvement (highest marginal WTP: KRW 30,225–43,961). This pattern corresponds with findings from Faccioli et al. [100] and Ureta et al. [101], emphasizing that residents emphasize services depending on their physical and ecological context. For example, the mountainous eastern region experienced a sharp increase in landslides in 2023, 12.7 times more than in 2021, emphasizing the urgency of erosion control [90]. In contrast, under greater development pressure, the western lowland region likely placed more importance on biodiversity conservation. Moreover, Toledo-Gallegos et al. [102] highlighted that even when regional averages show no significant difference, spatial clustering of WTP (hotspots and coldspots) can reveal localized preference heterogeneity. Future research applying advanced spatial analysis methods such as Moran’s I or Ripley’s K could uncover hidden local-scale preference patterns within Jeollabuk-do that are not apparent in simple east–west comparisons.

However, it is important to acknowledge some limitations. The study relied heavily on observed WTP trends without statistically validating whether these represent actual preference differences. Moreover, the negative WTP values for provisioning services (timber, non-timber products) may suggest a general communal shift toward conservation. Still, they could also result from estimation noise or resistance to tax increases. Similar global findings, such as those reported in FAO [1], point to declining demand for extractive forest uses and favor regulating and supporting services, reinforcing the need to interpret negative WTP carefully.

Another limitation relates to the sample composition. The underrepresentation of respondents aged 50 and older (20% in the sample versus ~49% in the population) likely skewed the data toward younger, urban, digitally connected participants, reducing the visibility of preferences for provisioning services, which older or rural populations might emphasize. Addressing this age bias in future studies through post-stratification weighting or in-person sampling is essential. Furthermore, the survey timing, conducted immediately post-monsoon, likely amplified preferences for erosion control, but this confounding effect was not explicitly quantified.

Therefore, future research may benefit from following the examples from Jo et al. [43], Toledo-Gallegos et al. [102], and Ureta et al. [101], conducting subgroup analyses by demographic (age, gender), socioeconomic (income, occupation), and spatial dimensions to understand heterogeneity in preferences better. Incorporating advanced analyses that reflect spatial clustering and regional WTP patterns would enhance explanatory power. Furthermore, integrating insights from cultural ecosystem service valuation and regional planning literature can significantly strengthen the theoretical contributions.

5. Conclusions

This study began with the hypothesis that residents’ preferences for forest ecosystem services would vary according to the forest distribution ratios in their respective regions. The results indicated that Jeollabuk-do residents’ preferences for forest ecosystem services did not significantly differ between areas with high forest distribution ratios (eastern region) and those with low forest distribution ratios (western region). This suggests that the residents of Jeollabuk-do may not be distinctly aware of or influenced by the forest distribution ratios in their areas, or it could be a limitation of the study design itself. Based solely on the study’s findings, it can be concluded that the forest distribution ratio does not substantially affect residents’ perceptions of forest ecosystem service preferences.

However, when analyzing the relative importance of preferred ecosystem services, it was observed that the top priorities differed between the two regions. Eastern region residents prioritized erosion control (with marginal willingness-to-pay values ranging from KRW 23,559 to 33,109). In comparison, western region residents prioritized biodiversity enhancement (with marginal willingness-to-pay values ranging from KRW 30,225 to 43,961). These differences are likely influenced by the characteristics and development levels of their respective regions, as well as the timing of the survey. This indicates a tendency in the preferred management direction of forest ecosystem services based on the forest distribution ratio of residents’ areas.

Considering these results, prioritizing erosion control in the eastern region and biodiversity enhancement in the western region could help reduce residents’ damages from natural disasters and conserve the environment, thus meeting the demand for preferred ecosystem services according to regional preferences. Furthermore, region-specific forest management recommendations can be proposed. Managers should emphasize erosion control in the eastern region by improving understory vegetation cover, reforesting with native broadleaf species, and enhancing slope stability. In the western region, biodiversity can be improved by restoring fragmented habitats, planting diverse native species, and establishing ecological corridors. These tailored approaches to forest ecosystem service management reflect residents’ preferences, enhance their perceived ecological well-being, and contribute specifically to reducing forest-based natural disaster risks while strengthening long-term ecosystem resilience.