Ecological Strategy for Restoring the Forest Ecosystem in the Taebaek Region of Baekdudaegan Mountains in Korea

Abstract

This study was conducted to classify the types of degraded land in the Taebaeksan region within the Baekdudaegan mountain range in Korea using image analysis and to assess the degree of degradation through a detailed survey of each type. Based on the findings, candidate tree species for restoration and strategies for land rehabilitation were proposed. The analysis identified 1589 degraded sites, classified into nine types of degradation. Using the Self-Organizing Map (SOM) method, these sites were further categorized into six environmental types. A survey was conducted on 25 degraded sites (encompassing 73 specific locations) in the Taebaeksan region, revealing that the sites fell into seven out of nine degradation categories: grassland, agricultural land, forest roads, ranches, cemeteries, logging sites, and abandoned quarries. The excluded categories were buildings and military facilities. A more focused survey was conducted on the 73 largest sites across the degradation types, with a reference ecosystem selected based on topographic analysis and succession processes. Diagnostic evaluations identified logging sites and severely degraded areas as the most urgent priorities for rehabilitation. Candidate tree species for restoration were proposed for each degradation type and environmental condition, based on ecological importance and reference ecosystem characteristics. This study provides a comprehensive assessment of degradation types following ecological restoration procedures and offers valuable baseline data for selecting restoration species and designing restoration strategies in conjunction with reference ecosystem studies.

1. Introduction

The Baekdudaegan Mountain Range forms the foundation of Korea’s territory, spanning 1400 km from Baekdusan to Jirisan, and serves as the country’s core ecological axis [1]. In South Korea, it extends approximately 684 km from Hyangnobong Peak to Cheonwangbong Peak in Jirisan and includes the 2013 ceasefire line. The Baekdudaegan Protected Area, covering a total of 701 km, is divided into a core area and a buffer zone, encompassing six provinces and 32 cities and counties in administrative terms, and is under systematic management [2,3].

Baekdudaegan is recognized for its high ecological conservation value, supporting biodiversity and acting as a key axis that integrates not only traditional, human geography, industrial, and cultural values but also the rich natural environment of the Korean Peninsula [1,3,4,5]. National efforts to preserve Baekdudaegan have led to the enactment of the Baekdudaegan Protection Act in December 2003, spearheaded by the Korea Forest Service, which designated the Baekdudaegan Protected Area in 2005. These efforts have continued with the goal of expanding the protected area [5,6].

Notably, with the adoption of the Kunming–Montreal Global Biodiversity Framework (GBF) at the 2022 Conference of the Parties on Biological Diversity and the expansion of the Baekdudaegan Protected Area in 2023, the protected area reached a total of 2776.5 km² by 2024 [7]. Additionally, under Article 4 of the Baekdudaegan Protection Act, a 10-year basic plan for its protection is established, proposing long-term conservation and management plans to lay the foundation for sustainable management [2,3,8].

Despite these legal protections, many areas within the Baekdudaegan have already suffered degradation due to development activities before its designation as a protected area. The region remains under threat from indiscriminate forest resource utilization, increased visitation, and the effects of climate change, posing significant challenges to systematic conservation and management [3,9,10,11].

Therefore, it is imperative to analyze the current status of degradation in the Baekdudaegan Protected Area through comprehensive systematic surveys and propose restoration and management plans based on diagnostic evaluations. However, previous studies have primarily focused on identifying forest degradation trends through field surveys and have only suggested fragmented restoration and management measures [1,3,5,12]. Furthermore, research specifically addressing the classification and analysis of degraded areas in the Taebaeksan region of the Baekdudaegan is lacking. This underscores the need for research to scientifically investigate the actual state of degradation in the Taebaeksan area and establish a foundation for restoration efforts in the Baekdudaegan Mountain Range.

Ecosystem restoration involves reestablishing biological, chemical, and physical linkages between natural processes and ecosystems [13]. Internationally, large-scale restoration projects have been implemented to address ecosystem degradation and climate change. For example, Brazil has integrated artificial planting with natural regeneration to accelerate rainforest recovery, a strategy considered crucial for tropical forest resilience [14]. Similarly, the European Union has launched the “EU Biodiversity Strategy for 2030”, emphasizing nature-based solutions (NbS) for sustainable restoration [15]. Additionally, the “Rewilding North America” initiative in the United States and Canada focuses on enhancing ecological connectivity and restoring complex ecosystems through large-scale habitat protection [16]. Effective restoration of degraded ecosystems is essential for biodiversity, ecosystem health, climate change mitigation, and overall human well-being [17]. Through extensive consultations, UN Decade partners have worked to achieve the goals of the UN and Rio Conventions by 2023, promoting sustainable restoration practices, addressing the root causes of ecosystem degradation, and advocating for policies that ensure fairness and enhanced recovery of biodiversity and ecosystem health. The restoration process prepared under these principles includes: (1) identifying and assessing the extent and scale of degradation, (2) formulating a restoration plan that considers financial, ecological, socio-economic, and cultural factors, (3) implementing the restoration project, (4) ensuring short- and long-term management based on specific requirements, and (5) monitoring and evaluating the restoration sites [17,18].

However, domestic restoration projects often lack comprehensive systematic surveys and diagnostic evaluations of degraded areas. They rely heavily on large-budget construction methods modeled on past green restoration projects, often applying uniform restoration strategies across areas with distinct environmental characteristics [3,19,20,21].

Therefore, this study investigates the current status of land degradation in the Taebaeksan region of the Baekdudaegan Mountain Range; analyzes the urgency of restoration needs, types of degradation, and environmental characteristics through diagnostic evaluation; and utilizes this information as a foundation for establishing restoration goals and strategies. Based on this approach, the following hypotheses are proposed.

This study hypothesizes that environmental variables, including climate, topography, hydrology, and soil, significantly influence the classification of degradation types and the selection of appropriate restoration species in the Taebaeksan region. Additionally, it is expected that nature-based solutions (NbS) will provide greater long-term ecological sustainability than conventional engineering-based restoration approaches by enhancing biodiversity and improving ecosystem resilience. Furthermore, the use of native species adapted to the region is anticipated to facilitate a more effective restoration process and contribute to long-term ecological stability. Lastly, the restored vegetation is expected not only to resemble the reference ecosystem in composition but also to perform ecological functions similar to those of the pre-disturbance ecosystem, ensuring sustainable restoration. To validate these hypotheses, this study employs analyses of species diversity, vegetation indices, and ecological similarity, providing a scientific foundation for sustainable forest restoration strategies in the Baekdudaegan Mountain Range.

2. Materials and Methods

The research process is illustrated in Figure 1 and consists of the following steps. First, remote sensing and ArcGIS 10.8.1 (Esri, Redlands, CA, USA) data were utilized to analyze degraded forest areas, followed by classification of degradation types using a Self-Organizing Map (SOM). Next, a vegetation survey and comparative analysis with reference ecosystems were conducted to assess ecological conditions, which served as the basis for selecting priority restoration sites and appropriate restoration species. Finally, monitoring and adaptive management were implemented to evaluate the effectiveness of restoration efforts and establish long-term restoration strategies.

2.1. Study Area

In this study, the boundary was established based on the Baekdudaegan Protection Area as defined by the Korea Forest Service’s 2nd Baekdudaegan Protection Basic Plan (2016–2025). The research was conducted in the Taebaeksan area (Gangneung–Taebaek: 157 km) as the study area, one of the five regions defined in the basic plan for Baekdudaegan protection.

The Taebaeksan area stretches from Guryongryeong in Hongcheon-gun, Gangwon-do, to Gitdaebaegibong in Yeongwol-gun, covering a total length of 157 km and a total area of 75,372 ha. Geographically, it is located between 128°33′ and 128°35′ east longitude and 37°48′ and 37°08′ north latitude [2].

The Taebaeksan National Park area, covering 70.01 km² (comprising the park resource conservation district, park resource environment district, and park village district), was excluded from the survey. It was determined that this area would fall under the direct management and research of the National Park Service, which oversees natural environment protection and conservation. Instead, this study concentrated on conducting surveys of degraded areas located within the core and buffer zones (Figure 2).

2.2. Classification of Degradation Types Using Self-Organizing Map (SOM)

To conduct a survey and diagnostic evaluation of degraded land by type in the Taebaeksan area, the causes of degradation and environmental variables were analyzed, and type classification was performed using orthoimages of the Taebaeksan area based on aerial photographs taken between 2017 and 2019. A field survey was conducted to examine and analyze the vegetation, current plant status, and the actual condition of the extracted degraded land and reference ecosystems (control zones).

A reference ecosystem is a healthy ecosystem composed of biological communities and abiotic components that serve as a model for ecological restoration. The selected reference ecosystem should exhibit minimal human disturbance, maintain connectivity with surrounding forests, and share key ecological characteristics with the degraded site, such as altitude, aspect, and environmental conditions. It should be identified in nearby areas with comparable ecological attributes [20,22,23].

Orthoimages, which are orthorectified images with a spatial resolution of 25 cm, were created from aerial photos taken biennially across the country by the National Geographic Information Institute of the Ministry of Land, Infrastructure, and Transport. These images were analyzed accounting for temporal and spatial variations in illumination conditions, including seasonal changes, solar altitude, and azimuth at the time of capture. Causes of land degradation were determined through visual analysis of the orthoimages. The degradation was categorized into buildings, roads, agricultural land, logging sites, hiking trails, and other types. In cases where degradation was difficult to identify due to extensive natural succession over a long period, the area was classified as grassland without specifying a specific cause.

Environmental variable analysis through orthoimages focused on selecting factors affecting the survival and growth of vegetation for restoration purposes. Environmental variables were analyzed for climate, topography, hydrology, soil, and vegetation. Prior to conducting the Self-Organizing Map (SOM) analysis, a Spearman correlation analysis was performed to explore the relationships among environmental factors. The key variables included climate factors (precipitation, temperature, frost occurrence days), topographic characteristics (elevation, slope), and indicators related to soil and vegetation distribution. The correlation analysis results were visualized using a heatmap, which revealed interactions among environmental factors. Degraded environment types were classified using a Self-Organizing Map (SOM), an unsupervised-neural-network-based clustering method that reduces high-dimensional data into a two-dimensional map while preserving topological relationships. The classification of degraded area types was conducted using the kohonen package in the R software version 4.4.2 (R Foundation for Statistical Computing, Vienna, Austria), with the SOM neural network weight vectors optimized through 20,000 training iterations.

Among the degraded areas identified through orthoimage analysis, excluding the Taebaeksan National Park area, a survey was conducted on a total of 73 degraded sites and 59 reference ecosystems located near the degraded areas and accessible over a certain area (

Table A1. List of Investigation Sites for Vegetation Survey in Mt. Teabaek Section.

Type	No.	Address	Lat	Long
Degraded sites	1	Gangwon-do Taebaek-si Sangsami-dong san65-2	37°15′39.70″	129°00′16.70″
	2	Gangwon-do Taebaek-si Hwajeon-dong 143-27	37°12′54.40″	128°57′52.60″
	3	Gangwon-do Pyeongchang-gun Daegwallyeong-myeon Hoenggye-ri san1-119	37°41′36.42″	128°45′31.02″
	4	Gangwon-do Gangneung-si Wangsan-myeon Mokgye-ri san460-1	37°35′16.30″	128°51′52.90″
	5	Gangwon-do Taebaek-si Hasami-dong 524-157	37°20′34.10″	129°00′28.10″
	6	Gangwon-do Taebaek-si Hwajeon-dong 143-27	37°12′49.80″	128°58′28.20″
	7	Gangwon-do Gangneung-si Okgye-myeon Sangye-ri san347	37°34′27.15″	128°55′32.90″
	8	Gangwon-do Pyeongchang-gun Daegwallyeong-myeon Hoenggye-ri san2-1	37°40′43.00″	128°44′54.00″
	9	Gangwon-do Gangneung-si Wangsan-myeon Daegi-ri san1-12	37°35′13.10″	128°49′38.08″
	10	Gangwon-do Gangneung-si Wangsan-myeon Wangsan-ri san2-10	37°38′10.80″	128°46′47.30″
	11	Gangwon-do Samcheok-si Hajang-myeon Beoncheon-ri san57-2	37°20′58.60″	129°00′26.90″
	12	Gangwon-do Gangneung-si Wangsan-myeon Mokgye-ri san460-1	37°34′42.60″	128°51′18.90″
	13	Gangwon-do Pyeongchang-gun Daegwallyeong-myeon Hoenggye-ri 14-42	37°41′19.60″	128°45′06.70″
	14	Gangwon-do Pyeongchang-gun Daegwallyeong-myeon Hoenggye-ri san1-134	37°42′17.80″	128°43′32.10″
	15	Gangwon-do Pyeongchang-gun Daegwallyeong-myeon Hoenggye-ri 704	37°43′36.30″	128°43′25.50″
	16	Gangwon-do Donghae-si Samhwa-dong 714	37°28′01.10″	129°01′49.60″
	17	Gangwon-do Gangneung-si Seongsan-myeon Eoheul-ri san1-29	37°42′24.00″	128°46′20.40″
	18	Gangwon-do Taebaek-si Hasami-dong san220-75	37°20′05.10″	128°59′58.90″
	19	Gangwon-do Gangneung-si Wangsan-myeon Daegi-ri san1-13	37°34′58.55″	128°48′56.14″
	20	Gangwon-do Gangneung-si Wangsan-myeon Songhyeon-ri san242-58	37°35′06.40″	128°50′31.00″
	21	Gangwon-do Gangneung-si Wangsan-myeon Daegi-ri san1-13	37°34′15.61″	128°49′17.02″
	22	Gangwon-do Gangneung-si Wangsan-myeon Songhyeon-ri 86-1	37°32′54.00″	128°51′25.06″
	23	Gangwon-do Gangneung-si Okgye-myeo Namyang-ri san287	37°32′35.00″	128°57′29.90″
	24	Gangwon-do Gangneung-si Yeongok-myeon Samsan-ri san1-1	37°50′55.92″	128°39′33.38″
	25	Gangwon-do Gangneung-si Wangsan-myeon Wangsan-ri san1	37°39′28.40″	128°45′14.30″
Reference sites	1	Gangwon-do Taebaek-si Sangsami-dong	37°15′28.10″	128°59′59.26″
	2	Gangwon-do Taebaek-si Hwajeon-dong	37°13′24.60″	128°57′22.50″
	3	Gangwon-do Pyeongchang-gun Daegwallyeong-myeon Hoenggye-ri	37°41′39.68″	128°45′18.79″
	4	Gangwon-do Gangneung-si Wangsan-myeon Mokgye-ri	37°35′27.50″	128°51′57.00″
	5	Gangwon-do Taebaek-si Hasami-dong	37°21′03.70″	129°00′25.60″
	6	Gangwon-do Taebaek-si Hwajeon-dong	37°13′24.60″	128°57′22.50″
	7	Gangwon-do Gangneung-si Okgye-myeon Sangye-ri	37°34′25.65″	128°54′58.65″
	8	Gangwon-do Pyeongchang-gun Daegwallyeong-myeon Hoenggye-ri	37°40′40.48″	128°45′05.20″
	9	Gangwon-do Gangneung-si Wangsan-myeon Daegi-ri	37°35′17.40″	128°49′36.30″
	10	Gangwon-do Gangneung-si Wangsan-myeon Wangsan-ri	37°38′10.80″	128°46′47.30″
	11	Gangwon-do Samcheok-si Hajang-myeon Beoncheon-ri	37°21′03.70″	129°00′25.60″
	12	Gangwon-do Gangneung-si Wangsan-myeon Mokgye-ri	37°35′27.50″	128°51′57.00″
	13	Gangwon-do Pyeongchang-gun Daegwallyeong-myeon Hoenggye-ri	37°41′18.80″	128°45′07.20″
	14	Gangwon-do Pyeongchang-gun Daegwallyeong-myeon Hoenggye-ri	37°42′31.00″	128°44′14.00″
	15	Gangwon-do Pyeongchang-gun Daegwallyeong-myeon Hoenggye-ri	37°43′39.00″	128°43′22.10″
	16	Gangwon-do Donghae-si Samhwa-dong	37°28′01.10″	129°01′49.60″
	17	Gangwon-do Gangneung-si Seongsan-myeon Eoheul-ri	37°42′22.70″	128°46′20.30″
	18	Gangwon-do Taebaek-si Hasami-dong	37°20′05.50″	128°59′58.00″
	19	Gangwon-do Gangneung-si Wangsan-myeon Daegi-ri	37°34′56.77″	128°48′55.25″
	20	Gangwon-do Gangneung-si Wangsan-myeon Songhyeon-ri	37°35′11.70″	128°50′22.90″
	21	Gangwon-do Gangneung-si Wangsan-myeon Daegi-ri	37°34′14.78″	128°49′17.36″
	22	Gangwon-do Gangneung-si Wangsan-myeon Songhyeon-ri	37°32′04.00″	128°51′23.59″
	23	Gangwon-do Gangneung-si Okgye-myeo Namyang-ri	37°32′53.60″	128°51′43.80″
	24	Gangwon-do Gangneung-si Yeongok-myeon Samsan-ri	37°50′51.95″	128°39′34.69″
	25	Gangwon-do Gangneung-si Wangsan-myeon Wangsan-ri	37°39′20.20″	128°44′43.00″

Gl: grassland, Al: agricultural land, Fr: forest roads, Ra: ranches, Ct: cemeteries, Ls: logging sites, Aq: abandoned quarries.

).

2.3. Vegetation Survey and Analysis

The vegetation survey was conducted using plots measuring 25 m² (5 m × 5 m). To assess the proportion of degraded areas by type, the average relative importance value was calculated by summing the relative frequency and relative cover of tree species in the survey area [24]. Additionally, species abundance was analyzed by quantifying both the number of species within a specific area and the number of individuals within a community.

The species diversity index (H′), a measure that simultaneously considers species richness (the number of species present) and species evenness (the degree of homogeneity in species abundance), was analyzed. Furthermore, the evenness index (J′) and dominance index (D) were calculated to assess community structure [25,26,27] (

Table 1. List of formulas for determining plant diversity.

Diversity Indices	Formula
Relative Importance Value	(RF + RC)/2
Species Diversity Index (H′)	$-\sum \left(ni/N \ast log ni/N\right)$
Evenness Index (J′)	$H^{\prime }/logS$
Species Dominance Index (D)	$1-\sum ni\left(ni-1\right)/N\left(N-1\right)$

RF: relative frequency, RC: relative cover of tree species, N: total number of plants, ni: number of specific plants, S: number of species.

).

To determine the similarity in species composition across different types of degraded areas, the Multi-Response Permutation Procedures (MRPP) test, a non-parametric method for testing differences between groups, was performed. The distance scale for this analysis was calculated using Sorensen’s method [28,29].

Field surveys were carried out at each type of degraded area and corresponding reference ecosystem survey site to record all identified vascular plant species. Unidentified plants were classified and identified using herbarium references [30,31,32].

The life form of a plant, which represents the morphological and functional adaptations a plant has developed over time in response to its growth environment, was analyzed. Life forms were categorized into dormant forms, reproductive forms (types of underground organs and dispersal organs), and growth forms. These were further classified and organized by biological type [32,33,34].

Based on the surveyed flora, the naturalization rate, an indicator of human disturbance, was calculated as follows [35]:

$$\begin{gathered} Naturalization rate \left(\%\right)=(Number of naturalized plant species / \\ Total number of plant species)\times 100 \end{gathered}$$

Reference ecosystems were identified within a 5 km radius of each degraded area, taking into account the environmental conditions of the Taebaeksan area in the Baekdudaegan Protected Area, such as elevation, slope, and orientation. Forest stands were selected based on the succession process, with a minimum of three reference sites chosen per degraded area.

Diagnostic evaluations were conducted to quantitatively assess a total of eight factors: the area of the target site, land ownership, forest restoration type, stratigraphic structure, degree of disturbance, community similarity index, secondary degradation, and number of usable species. These factors were evaluated based on the urgency of restoration needs.

Additionally, for each type of degradation, restoration species were proposed by identifying species with a frequency of occurrence (critical value) of 3 or higher in the reference ecosystem, categorized according to their life forms (tree, shrub, herbaceous, fern, etc.).

3. Results

3.1. Classification of Degraded Area Types

Analysis of the Baekdudaegan Taebaeksan area, which encompasses a total area of 75,372 ha, revealed 1589 degraded sites through orthoimage analysis. These sites covered a total area of 1765.7 ha, representing approximately 2.3% of the total area.

Analysis by cause of degradation revealed the following distribution: roads (619 sites), buildings (302 sites), agricultural lands (230 sites), ranches (134 sites), cemeteries (71 sites), grasslands (70 sites), logging areas (69 sites), abandoned quarries (5 sites), military facilities (2 sites), and others (87 sites).

In terms of area coverage, the proportions were distributed as follows: ranches and roads each accounted for 33.8%, abandoned quarries for 14.8%, logging lands for 14.5%, agricultural lands for 8.7%, grasslands for 5.7%, buildings for 0.6%, cemeteries for 0.1%, military facilities for 0.2%, and others for 1.6%.

Despite their relatively small number, abandoned quarries and logging sites occupied large areas due to extensive resource extraction activities. Conversely, roads, ranches, and agricultural lands in mountainous regions exhibited a high frequency of occurrence and substantial total area. This pattern likely results from sustained human access and significant development pressure, particularly in areas where forest resources are easily accessible (

Table 2. The status of degraded sites by damage type in the Mt. Teabaek section.

	Type	Road	Construction	Agricultural Land	Ranches	Cemeteries	Grassland	Logging Sites	Abandoned Quarries	Other
Status
Number of patches		619	302	230	134	71	70	69	5	89
Area of patches (ha)		353.1	10.6	153.6	596.8	1.8	100.6	263.1	261.3	31.8
Ratio of total degraded sites		20	0.6	8.7	33.8	0.1	5.7	14.5	14.8	1.8

).

Using the Self-Organizing Map (SOM) analysis, the 1589 degraded sites within the Baekdudaegan Taebaeksan area were classified into six environmental types based on their topographical and ecological characteristics (

Table 3. Degraded site information and environment variable values.

Environment Variable	Type 1	Type 2	Type 3	Type 4	Type 5	Type 6
Number of degraded sites	99	1137	95	68	119	71
Mean area of degraded patches (ha)	284.2	1358.3	26.1	10.8	76.5	9.7
Average elevation (m)	1036	828	800	426	557	752
Average slope (°)	11.7	16.3	8.9	10.3	14.3	15.1
Wetlands	0.29	0.02	0.12	0.31	1.00	0.07
Annual mean precipitation (mm)	1657	1485	1388	1397	1422	1471
Number of heavy rain days	2.9	2.4	2.1	2.2	2.2	2.3
Annual mean temperature (°C)	5.1	7.1	7.5	8.0	7.6	7.4
Coldest monthly average temperature (°C)	−10.0	−7.3	−7.1	−6.4	−6.8	−7.0
Number of frost days	178.7	158.9	155.6	153.0	154.4	155.2

Type 1: highland; Type 2: medium altitude; Type 3: nutrient-poor soil type; Type 4: deep soil profile type; Type 5: wetland; Type 6: mixed forest.

). The SOM analysis incorporated multiple environmental variables including elevation, slope, aspect, soil characteristics, and vegetation cover to generate distinct environmental clusters. This classification approach enabled the identification of patterns and relationships among degraded sites, providing a foundation for developing targeted restoration strategies for each environmental type. The resulting six environmental types exhibited distinct characteristics in terms of their physical environment and degradation patterns, suggesting the need for type-specific restoration approaches.

A Spearman correlation analysis was conducted to examine the relationships among key environmental factors associated with forest degradation and restoration potential. The analyzed variables included climatic factors (precipitation, temperature, frost occurrence days), topographic factors (elevation, slope), soil characteristics (drainage suitability, soil depth), and vegetation types (coniferous forest, broadleaf forest, mixed forest). The correlations among these factors were analyzed in the context of forest degradation and restoration.

The analysis results revealed a negative correlation between elevation and the area of degraded patches. This indicates that areas at lower elevations tend to have larger degraded patches. A strong positive correlation was detected between temperature and frost occurrence days, with lower-temperature regions exhibiting higher numbers of frost occurrence days (Figure 3).

Precipitation and the number of rainy days demonstrated a positive correlation. This indicates that regions with higher annual precipitation experience more rainy days. A negative correlation was detected between wetland proportion and soil drainage suitability. These findings suggest that areas with higher wetland proportions tend to exhibit lower drainage suitability.

The highland mountain area type (Type 1) encompasses 99 degraded sites with an average elevation of 1036 m, distributed around Daegwallyeong Samyang Ranch, Odaesan Mountain, and Cheongoksan Mountain. The total degraded area for this type is 284.2 ha, representing 16.1% of the total degraded area in the Taebaeksan region. Analysis of natural forests within a 100 m radius of the degraded sites revealed that other broad-leaved forests comprised 32.4% and oak forests 13.1%. Based on these vegetation characteristics, broad-leaved trees are considered suitable as the primary restoration species.

The mid-altitude mountain type (Type 2), situated at an average elevation of 828 m, includes 1137 degraded sites. This type exhibits average climate, soil, and vegetation characteristics representative of the overall Taebaeksan region. Analysis of natural forests within a 100 m radius of the degraded sites showed oak forests accounting for 21.8% and pine forests 12.7%. Restoration efforts incorporating these dominant natural forest species are considered ecologically appropriate.

The nutrient-poor soil type (Type 3), located at an average elevation of 800 m, has only 18.0% of soil suitable for forestry—significantly lower than the average of other types (35.2%). Analysis of natural forests within a 100 m radius of the degraded sites indicated that pine forests (11.6%) had a relatively higher distribution than broad-leaved forests. Given these infertile soil conditions and existing vegetation patterns, Pinus densiflora, known for its high survival rate in nutrient-poor environments, is recommended as the primary restoration species.

The streamside forest type (Type 5), found at an average elevation of 650 m, occurs within 50 m of rivers and streams. This type represents a small-scale category, comprising only 0.6% of the total degraded land area. Given the risk of stream flooding and inundation, restoration plans should prioritize tree species with high moisture tolerance.

The deep soil profile type (Type 4) and mixed forest type (Type 6) present limitations in determining suitable restoration species through image analysis alone. For these types, the selection of appropriate restoration species will require additional field surveys and reference ecosystem analyses (Figure 4).

3.2. Survey of Current Conditions by Damage Type

3.2.1. Status of Degraded Sites by Damage Type

Based on the image analysis of the Baekdudaegan Mountain Range, the 25 largest degraded sites by area were identified and categorized into seven distinct types: grasslands, agricultural lands, forest roads, ranches, cemeteries, logging sites, and abandoned quarries. For each category, we quantitatively assessed key environmental factors, including the number of survey plots, elevation, slope, and degree of rock exposure.

Abandoned quarries represented the category with the highest number of degraded site survey plots, with 73 sites examined. Each survey plot was established with dimensions of 2 m × 2 m (4 m²). Among the categories, grasslands were observed at the highest mean elevation, averaging 977.6 m above sea level (

Table 4. Topographical factors of degraded sites by damage type in the Mt. Teabaek section.

	Type	Gl	Al	Fr	Ra	Ct	Ls	Aq	Total
Investigating Contents
Number of investigated sites		9	12	12	9	9	8	14	73
Average altitude (m)		977.6	773.4	789.4	901.1	901.1	599.6	823.9	823.7
Average slope (°)		7.6	1.2	1.8	2.8	2.8	4.0	6.8	3.9
Average rock exposure (%)		0	0.8	0	0	0	2.5	0	0.5

Gl: grassland, Al: agricultural land, Fr: forest roads, Ra: ranches, Ct: cemeteries, Ls: logging sites, Aq: abandoned quarries.

).

3.2.2. Importance Values at Degraded Sites by Damage Type

Analysis of the importance values of degraded sites in the Taebaeksan area revealed no tree layer in any degradation type, while the subtree layer was observed only in abandoned quarries. The dominant species composition was primarily artificial, consisting of I (27.1), Pinus densiflora (20.6), Robinia pseudoacacia (16.3), and Prunus serrulata Lindl var. pubescens (12.0).

In the shrub layer, logging sites showed six taxa in the following order: Larix kaempferi (55.7), Pinus densiflora (19.3), Aralia elata (5.7), Quercus serrata (7.5), Quercus mongolica (6.1), and Corylus sieboldiana var. mandshurica (5.7). Abandoned quarries exhibited seven taxa: Pinus densiflora (27.9), Amorpha fruticosa (38.0), Fraxinus rhynchophylla (12.6), Lespedeza cyrtobotrya (8.8), Prunus serrulata Lindl var. pubescens (4.9), Acer pictum Thunb var. mono (3.9), and Rhus chinensis (3.9).

The herb layer comprised a total of 198 taxa, with dominant species including Artemisia indica (8.2), Setaria viridis (5.5), Rubus crataegifolius (3.9), Erigeron annuus (3.5), and Chrysanthemum seticuspe (2.7). Among the top 30 species, eight were identified as naturalized plants: Erigeron annuus, Symphyotrichum pilosum, Galinsoga quadriradiata, Rumex acetosella, Taraxacum officinale, Erigeron canadensis, Oenothera biennis, and Trifolium pratense. This suggests the need for naturalized species management in forest restoration efforts (Table A1).

3.2.3. The Species Diversity Index at Degraded Sites by Damage Type

The species diversity index (H′) increases with the complexity and stability of an ecosystem [36]. In this study, the species diversity of degraded sites was recorded as follows: grasslands 1.437, agricultural lands 1.115, forest roads 1.376, ranches 1.574, cemeteries 1.562, logging sites 1.880, and abandoned quarries 1.602, with logging sites showing the highest species diversity. The reference ecosystems showed higher species diversity values: grasslands 1.860, agricultural lands 2.095, forest roads 1.921, ranches 1.915, cemeteries 1.881, logging sites 2.172, and abandoned quarries 1.889, with logging site reference ecosystems displaying the highest diversity. All reference ecosystems exhibited higher diversity indices than their corresponding degraded sites, indicating more stable conditions.

The evenness index (J′) approaches 1 when taxa are evenly distributed [37]. The mean species evenness was 0.648 (range: 0.585–0.738) for degraded sites and 0.657 (range: 0.621–0.718) for reference ecosystems, with reference ecosystems showing higher evenness values. Logging sites consistently showed the highest evenness values across all categories.

The dominance index (D) is known to indicate single-species dominance when above 0.9, dominance by one or two species when between 0.3 and 0.7, and multiple-species distribution when between 0.1 and 0.3 [36,38]. In this study, the mean species dominance was 0.686 (range: 0.552–0.808) for degraded sites and 0.791 (range: 0.772–0.834) for reference ecosystems, with reference ecosystems showing higher mean dominance. Based on these values, all survey sites were determined to be dominated by one or two taxa (Figure 5).

3.2.4. Comparison Community Similarity Analysis Between Degraded Sites and Reference Sites by Damage Type

Multi-Response Permutation Procedures (MRPP) analysis was performed to evaluate community similarity between degraded sites and their corresponding reference ecosystems across all degradation types. The analysis revealed similarity indices of 0.302 for grasslands, 0.412 for agricultural lands, 0.386 for forest roads, 0.400 for ranches, 0.234 for cemeteries, 0.232 for logging sites, and 0.315 for abandoned quarries. Among these, agricultural lands exhibited the highest similarity between degraded sites and reference ecosystems, while logging sites showed the lowest similarity.

Species compositional similarity was higher between degraded sites or between degraded sites and their reference ecosystems than among reference ecosystems themselves. The low similarity between logging sites and their reference ecosystems, despite high species diversity, can be attributed to post-degradation interventions such as artificial reforestation and forest restoration, which resulted in vegetation communities significantly different from the original ecosystem composition (

Table 5. The similarity index among degraded sites and reference sites.

		Degraded Sites							Reference Sites
		Gl	Al	Fr	Ra	Ct	Ls	Aq	Gl	Al	Fr	Ra	Ct	Ls	Aq
Degraded sites	Gl
	Al	0.174
	Fr	0.165	0.213
	Ra	0.200	0.361	0.288
	Ct	0.065	0.340	0.354	0.344
	Ls	0.215	0.369	0.220	0.171	0.203
	Aq	0.274	0.450	0.308	0.430	0.333	0.345
Reference sites	Gl	0.000	0.068	0.230	0.217	0.240	0.060	0.082
	Al	0.085	0.319	0.197	0.255	0.389	0.136	0.222	0.185
	Fr	0.000	0.073	0.275	0.118	0.203	0.067	0.089	0.255	0.179
	Ra	0.036	0.100	0.111	0.211	0.250	0.000	0.120	0.246	0.262	0.160
	Ct	0.000	0.169	0.250	0.227	0.575	0.069	0.205	0.311	0.405	0.292	0.358
	Ls	0.069	0.355	0.180	0.315	0.325	0.275	0.303	0.247	0.442	0.195	0.326	0.333
	Aq	0.026	0.127	0.196	0.154	0.400	0.022	0.131	0.278	0.347	0.250	0.366	0.438	0.211

Gl: grassland, Al: agricultural land, Fr: forest roads, Ra: ranches, Ct: cemeteries, Ls: logging sites, Aq: abandoned quarries.

).

3.2.5. Status of Taxonomic Groups by Type

A total of 202 taxa were identified across all degraded sites, comprising 65 families, 144 genera, 179 species, 9 subspecies, and 14 varieties. The distribution of taxa across degradation types was as follows: 76 taxa in grasslands, 109 in agricultural lands, 66 in forest roads, 58 in ranches, 73 in cemeteries, 94 in logging sites, and 84 in abandoned quarries. The Taebaeksan area, known for its diverse flora [10], revealed several rare and endemic plant species designated by the Korea Forest Service, despite being degraded sites.

Three rare plant taxa were identified within the degraded sites: Actaea bifida, Salvia chanryoenica, and Lysimachia pentapetala. Additionally, eight endemic plant taxa were observed, including Aster koraiensis and Angelica amurensis (

Table 6. List of rare, endemic, invasive alien, and biological indicator species by type of damage.

Type		Rare Plant	Endemic Plants	Invasive Alien Species	Climate-Sensitive Biological Indicator Species
Degraded sites	Gl	-	Aster koraiensis, Angelica purpuraefolia	Ambrosia artemisiifolia	-
	Al	-	-	Ambrosia artemisiifolia, Rumex acetosella	-
	Fr	-	-	-	Lamium amplexicaule
	Ra		Aster koraiensis	Symphyotrichum pilosum, Rumex acetosella	-
	Ct	Actaea bifida (VU), Salvia chanryoenica (LC)	Actaea bifida, Salvia chanryoenica, Lonicera subsessilis	Humulus scandens	-
	Ls	-	Cirsium setidens, Vicia chosenensis, Lonicera subsessilis, Salix koriyanagi	Symphyotrichum pilosum, Humulus scandens	Equisetum hyemale
	Aq	Lysimachia pentapetala (VU)	Salix koriyanagi	Symphyotrichum pilosum	-
Reference sites	Gl	Aristolochia manshuriensis (LC) Viola albida (LC)	Angelica purpuraefolia, Lonicera subsessilis	-	-
	Al		Angelica purpuraefolia	-	Ribes mandshuricum, Equisetum hyemale
	Fr	-	Angelica purpuraefolia	-	-
	Ra	Scopolia parviflora (LC)	Scopolia parviflora	-	Equisetum hyemale
	Ct	-	-	-	Ribes mandshuricum
	Ls	-	Weigela subsessilis	-	Equisetum hyemale
	Aq	Viola albida (LC), Viola diamantiaca(LC)	Angelica purpuraefolia	-	-

Gl: grassland, Al: agricultural land, Fr: forest roads, Ra: ranches, Ct: cemeteries, Ls: logging sites, Aq: abandoned quarries.

).

Agricultural lands showed the highest number of taxa, likely due to their extensive soil exposure from annual tilling, which facilitates the germination of soil seed banks and species introduction from surrounding areas. Conversely, ranches exhibited the lowest number of taxa, presumably due to their significantly lower soil exposure, which limits the germination of soil seed banks and dispersed seeds, except for ornamental grasses and flowers intentionally planted for ranch operations.

Reference ecosystems contained a total of 195 taxa, comprising 33 families, 135 genera, 173 species, 5 subspecies, and 17 varieties—fewer than the degraded sites. This difference can be attributed to the environmental conditions of degraded sites, which lack stable stratification and receive relatively high light penetration, leading to the establishment of numerous pioneer species including ferns, annual herbs, and perennial herbs following disturbance.

A survey of plants in degraded sites and reference ecosystems by degradation type revealed that among degraded sites, logging sites showed the highest number of species (170), followed by agricultural land (128) and forest roads (114). In contrast, within reference ecosystems, agricultural land exhibited the highest number of species (109), followed by logging sites (94) and grassland (92) (Figure 6).

Analysis of naturalization rates across different degradation types showed that grassland had 11.8%, agricultural land 16.4%, forest roads 6.1%, ranches 20.6%, cemeteries 8.1%, logging sites 7.1%, and abandoned quarries 11.4%, with an overall rate of 10.4% (Figure 7). The higher naturalization rates in ranches and agricultural land can be attributed to regular management practices, such as tilling, which facilitate the introduction of naturalized plants and germination of soil seed banks more readily compared to other degradation types.

3.2.6. Life Form Analysis

Plant life form, which categorizes the morphological and functional adaptations that plants have developed over time in response to their environment [39], was analyzed for both degraded sites and reference ecosystems.

In degraded sites, the distribution of life forms was as follows: trees 10.4%, subtrees 2.0%, shrubs 11.9%, vines 6.45%, annual herbs 15.8%, perennial herbs 50.0%, and ferns 3.5%. In reference ecosystems, the distribution was trees 14.4%, subtrees 5.6%, shrubs 17.4%, vines 6.7%, annual herbs 2.1%, perennial herbs 46.7%, and ferns 7.2%.

The life form composition in degraded sites showed a lower proportion of tree species and a higher proportion of annual herbs compared to reference ecosystems. This composition indicates that the degraded sites are in early successional stages, likely due to frequent artificial or natural disturbances. It is expected that these sites will require significant time to achieve vegetation stability, and succession should be guided toward natural vegetation similar to the reference ecosystems [3].

3.2.7. Diagnostic Evaluation by Damage Type

A diagnostic evaluation was conducted on the 25 selected sites using eight factors: site area, land ownership, forest restoration type, stratification structure, degree of disturbance, community similarity, secondary damage, and number of available species. Each factor was scored for assessment. Based on the diagnostic evaluation, which prioritized public lands and sites with larger areas and greater vegetation disturbance and damage, Site 23, an abandoned quarry type, was identified as requiring the most urgent restoration. Among the bottom 10 sites, logging site types were the most prevalent.

The average evaluation scores by degradation type were grassland 57.0, agricultural land 50.3, forest roads 43.0, ranches 39.0, cemeteries 54.7, logging sites 35.6, and abandoned quarries 40.7. These results indicate that logging sites should be prioritized for restoration efforts (

Table 7. Diagnostic evaluation by damage type.

Site No.	Type	Size	Land Owner	Restoration Type	Layers	Disturbance	Similarity	Secondary Damage	Availability	Total
4	Grassland	20	20	10	2	2	8	6	4	72
17	Cemeteries	20	20	5	2	4	4	8	2	65
6	Agricultural land	5	20	10	4	2	8	4	4	57
2	Grassland	20	5	5	2	6	6	8	2	54
18	Cemeteries	20	5	0	4	4	4	8	6	51
5	Agricultural land	0	20	10	2	2	8	4	4	50
25	Abandoned quarries	15	5	0	6	6	6	8	4	50
16	Cemeteries	20	5	5	2	2	4	8	2	48
7	Agricultural land	10	5	10	2	2	8	6	4	47
8	Agricultural land	10	5	10	2	4	6	8	2	47
13	Ranches	5	10	10	2	2	8	6	4	47
10	Forest roads	10	5	10	2	2	6	8	2	45
24	Abandoned quarries	10	5	0	6	6	6	8	4	45
3	Grassland	15	5	5	2	2	6	8	2	45
11	Forest roads	10	5	10	2	2	6	8	2	45
12	Forest roads	5	5	10	2	2	8	6	4	42
14	Ranches	0	10	10	2	2	8	6	4	42
9	Forest roads	5	5	10	2	2	6	8	2	40
20	Logging sites	10	5	0	2	4	4	8	6	39
22	Logging sites	5	5	0	4	6	4	8	6	38
1	Logging sites	0	10	5	2	2	6	8	4	37
19	Logging sites	0	5	0	4	6	4	8	6	33
21	Logging sites	0	5	0	4	4	4	8	6	31
15	Ranches	0	5	5	2	2	4	8	2	28
23	Abandoned quarries	0	5	0	2	2	6	6	6	27

).

4. Discussion

Generally, forest restoration efforts have often involved species that did not align with environmental conditions [40], and restoration objectives were frequently not tailored to the specific site conditions and project requirements [41]. Particularly, to mitigate secondary damage caused by natural disasters, restoration measures have primarily focused on engineering approaches such as terrain stabilization, soil erosion prevention, and landslide mitigation. While these methods may provide short-term benefits, they may not sufficiently ensure sustainability from a long-term ecological restoration perspective [3,19,20].

The research involved field surveys to classify degradation types in the Baekdudaegan region. Based on this classification, appropriate restoration directions and species were proposed. Subsequently, the methodology included degradation type classification, importance value analysis for species selection, similarity analysis with reference ecosystems, plant life form analysis, and naturalization rate assessment [17]. Utilizing these methods, restoration species were identified based on environmental conditions. The results indicate that 91 species were selected, comprising 22 tree species, 11 subtree species, 20 shrub species, 4 annual herbaceous species, 20 perennial herbaceous species, and 14 pteridophyte species. Among them, Fraxinus rhynchophylla, Quercus mongolica, and Acer pseudosieboldianum were found to be applicable to all degradation types (

Table A2. Important Value of Plant Species by Type of Degraded Site.

Layer	Species	Gl	Al	Fr	Ra	Ct	Ls	Aq	Total
Subtree layer	Betula pendula	-	-					27.1	27.1
	Pinus densiflora	-	-	-	-	-	-	20.6	20.6
	Robinia pseudoacacia	-	-	-	-	-	-	16.3	16.3
	Prunus serrulata var. pubescens	-	-	-	-	-	-	12.0	12.0
	Morus bombycis	-	-	-	-	-	-	6.5	6.5
	Cornus controversa	-	-	-	-	-	-	6.5	6.5
	Fraxinus rhynchophylla	-	-	-	-	-	-	5.6	5.6
	Pinus koraiensis	-	-	-	-	-	-	5.6	5.6
	Total	0.0	0.0	0.0	0.0	0.0	0.0	100.0	100.0
Shrub layer	Larix kaempferi	-	-	-	-	-	55.7	-	31.2
	Pinus densiflora	-	-	-	-	-	19.3	27.9	20.6
	Amorpha fruticosa	-	-	-	-	-	-	38.0	13.5
	Aralia elata	-	100.0	-	-	-	5.7	-	8.3
	Fraxinus rhynchophylla	-	-	-	-	-	-	12.6	6.1
	Lespedeza cyrtobotrya	-	-	-	-	-	-	8.8	4.2
	Quercus serrata	-	-	-	-	-	7.5	-	4.0
	Quercus mongolica	-	-	-	-	-	6.1	-	3.2
	Corylus sieboldiana var. mandshurica	-	-	-	-	-	5.7	-	2.9
	Prunus serrulata var. pubescens	-	-	-	-	-	-	4.9	2.2
	Acer pictum var. mono	-	-	-	-	-	-	3.9	1.9
	Rhus chinensis	-	-	-	-	-	-	3.9	1.9
	Total	0.0	100.0	0.0	0.0	0.0	100.0	100.0	100.0
Herb layer	Artemisia indica	11.8	4.8	11.2	9.9	0.8	5.7	18.6	8.2
	Setaria viridis	3.9	4.8	4.1	19.3	3.7	0.4	2.3	5.5
	Rubus crataegifolius	2.7	-	3.5	2.3	2.5	9.7	0.6	3.9
	Erigeron annuus	2.7	6.7	6.4	3.1	2.3	0.7	4.9	3.5
	Chrysanthemum boreale	5.2	1.9	5.6	0.6	-	1.3	6.2	2.7
	Carex humilis var. nana	1.8	-	4.8	-	7.1	1.5	1.7	2.3
	Symphyotrichum pilosum	3.9	-	-	4.2	-	1.8	8.1	2.2
	Lespedeza bicolor	-	-	3.0	-	4.3	3.6	2.3	2.1
	Artemisia montana	2.4	1.9	3.9	0.8	-	3.0	0.6	2.0
	Plantago asiatica	0.9	1.6	7.0	1.1	-	0.9	-	1.8
	Carex lanceolata	3.9	-	1.1	1.2	1.5	2.8	1.1	1.8
	Miscanthus sinensis	4.9	-	1.1	-	8.0	-	-	1.7
	Galinsoga quadriradiata	-	10.6	-	1.1	-	-	-	1.6
	Carex planiculmis	-	-	4.6	-	0.8	1.9	2.7	1.5
	Artemisia japonica	3.0	-	0.6	0.8	-	0.3	7.9	1.4
	Persicaria nepalensis	-	6.3	0.6	3.1	-	-	-	1.4
	Rumex acetosella	-	4.9	-	4.4	-	-	-	1.3
	Commelina communis	-	5.6	0.4	-	-	1.9	-	1.3
	Kummerowia striata	0.9	-	5.6	2.0	-	-	-	1.3
	Potentilla fragarioides	-	0.5	0.4	-	3.8	2.5	-	1.2
	Taraxacum officinale	2.2	1.6	-	4.8	-	0.3	-	1.2
	Digitaria ciliaris	-	2.2	-	5.6	-	-	-	1.1
	Aster koraiensis	3.1	-	-	2.9	2.6	-	-	1.1
	Conyza canadensis	3.3	1.3	1.8	0.6	-	-	2.1	1.1
	Zoysia japonica	-	-	-	2.9	4.9	-	-	1.0
	Oenothera biennis	1.3	0.8	1.7	-	0.5	0.3	3.1	1.0
	Trifolium pratense	1.5	-	-	5.5	-	-	-	1.0
	Spiraea fritschiana	-	-	-	-	-	4.1	-	1.0
	Sigesbeckia orientalis subsp. Pubescens	-	4.0	-	2.7	-	-	-	0.9
	Sorbaria sorbifolia var. stellipila	-	0.5	-	-	-	3.4	-	0.9
	Others(168 species)	40.8	40.0	32.5	21.1	57.0	53.9	37.7	41.4
	Total	100.0	100.0	100.0	100.0	100.0	100.0	100.0	100.0

Gl: grassland, Al: agricultural land, Fr: forest roads, Ra: ranches, Ct: cemeteries, Ls: logging sites, Aq: abandoned quarries.

).

Restoration species were proposed based on both degradation and environmental types. The analysis incorporated 16 variables including climate, topography, hydrology, soil, and vegetation. Analysis of importance values and life forms by classified degradation types yielded 90 taxa: 20 tree taxa, 6 subtree taxa, 19 shrub taxa, 36 perennial herb taxa, 1 annual herb taxon, and 8 pteridophyte taxa. Among these, Quercus mongolica and Acer pseudosieboldianum were identified as highly suitable for all environmental types in the Taebaeksan area of Baekdudaegan, while 34 taxa were proposed for the nutrient-poor soil type and deep soil profile type (

Table A3. Proposal for Restoration Species by Damage.

Type	Scientific Name	Gl	Al	Fr	Ra	Ct	Ls	Aq
T1	Quercus aliena					O
	Betula costata							O
	Acer pictum var. mono		O	O	O		O	O
	Quercus variabilis	O	O	O		O		O
	Carpinus cordata	O	O					O
	Ulmus davidiana var. japonica	O	O	O			O
	Zelkova serrata		O		O	O
	Maackia amurensis		O		O		O
	Fraxinus mandshurica						O
	Betula davurica			O		O	O	O
	Fraxinus rhynchophylla	O	O	O	O	O	O	O
	Betula schmidtii	O		O		O		O
	Morus bombycis	O	O	O	O		O	O
	Lindera obtusiloba	O	O	O		O	O	O
	Pinus densiflora	O	O	O		O	O	O
	Fraxinus sieboldiana	O	O			O	O
	Quercus mongolica	O	O	O	O	O	O	O
	Pinus koraiensis					O		O
	Quercus serrata		O	O	O	O	O	O
	Cornus controversa	O	O	O				O
	Tilia amurensis	O	O	O	O		O	O
	Magnolia sieboldii	O		O	O	O	O	O
T2	Toxicodendron trichocarpum			O		O	O	O
	Syringa reticulata	O	O
	Euonymus macropterus	O
	Acer pseudosieboldianum	O	O	O	O	O	O	O
	Acer mandshuricum							O
	Rhus chinensis		O					O
	Acer tataricum subsp. ginnala				O
	Styrax obassis	O	O	O		O		O
	Euonymus oxyphyllus	O					O
	Acer barbinerve		O
	Euonymus sachalinensis							O
S	Lonicera maackii				O
	Stephanandra incisa	O	O	O	O	O	O	O
	Symplocos sawafutagi	O	O		O	O	O	O
	Corylus sieboldiana var. mandshurica	O	O		O		O	O
	Deutzia glabrata							O
	Clematis urticifolia		O	O			O
	Weigela florida	O	O	O				O
	Rubus crataegifolius	O	O	O	O	O	O	O
	Hydrangea macrophylla subsp. serrata			O
	Lespedeza bicolor	O	O	O		O	O
	Callicarpa japonica	O	O	O		O	O	O
	Lespedeza maximowiczii	O	O	O	O	O	O	O
	Sasa borealis	O	O	O	O		O
	Rhododendron mucronulatum	O	O	O		O		O
	Lespedeza cyrtobotrya	O	O	O			O	O
	Spiraea fritschiana	O					O
	Rhododendron schlippenbachii	O	O	O		O	O	O
	Zanthoxylum piperitum		O			O		O
	Syringa pubescens subsp. patula		O		O
	Aria alnifolia				O	O
H	Galium spurium						O
	Melampyrum roseum	O		O			O
	Impatiens textorii				O		O
	Crepidiastrum denticulatum	O	O	O		O	O	O
	Carex humilis var. nana	O	O	O	O	O	O	O
	Galium maximowiczii	O	O	O		O		O
	Carex lanceolata	O	O	O	O	O	O	O
	Artemisia stolonifera	O	O	O	O	O	O	O
	Astilbe chinensis	O	O		O		O	O
	Ainsliaea acerifolia	O	O		O	O	O	O
	Carex siderosticta	O	O	O	O	O	O	O
	Arisaema amurense		O				O	O
	Artemisia keiskeana	O		O		O	O
	Angelica decursiva	O	O			O		O
	Meehania urticifolia	O	O		O		O	O
	Isodon inflexus		O	O	O		O	O
	Calamagrostis arundinacea	O	O	O	O	O	O	O
	Isodon excisus	O	O	O	O	O	O	O
	Diarrhena japonica	O	O	O		O	O	O
	Oplismenus undulatifolius		O			O		O
	Aster scaber	O	O	O	O	O	O	O
	Pseudostellaria palibiniana	O	O	O	O		O
	Spodiopogon sibiricus	O	O	O	O	O	O	O
	Phryma leptostachya var. oblongifolia	O	O	O		O	O	O
	Athyrium niponicum	O	O	O	O	O	O
	Pteridium aquilinum var. latiusculum				O
	Botrychium ternatum	O			O
	Dryopteris crassirhizoma		O		O		O
	Osmunda cinnamomea	O	O				O
	Polystichum craspedosorum		O
	Athyrium yokoscense	O	O	O	O		O	O
	Equisetum hyemale		O		O		O
	Arachniodes borealis		O
	Deparia conilii						O
	Athyrium monomachii		O				O
	Matteuccia struthiopteris				O
	Dryopteris expansa					O
	Dennstaedtia wilfordii		O

Gl: grassland, Al: agricultural land, Fr: forest roads, Ra: ranches, Ct: cemeteries, Ls: logging sites, Aq: abandoned quarries.

).

For hypothesis validation, SOM analysis, floristic analysis, vegetation indices, and species diversity indices were employed. The consistency between predicted restoration species for specific environmental conditions and actual survey results demonstrated a high degree of agreement. These findings suggest that the proposed methodology provides a scientifically grounded basis for restoration planning. However, given that the analysis was conducted at a specific time and location, further long-term monitoring is necessary to evaluate the survival rate and community dynamics of restored vegetation. Additionally, when applying this approach to other ecosystems, adjustments should be made based on regional environmental conditions and climate variations.

The research outcomes extend beyond a specific region, having relevance to international sustainable forest restoration policies. The International Union for Conservation of Nature (IUCN) and the United Nations (UN) Decade on Ecosystem Restoration emphasize ecological approaches to sustainable forest restoration. The species selection and restoration methods derived from this study have potential applications in countries with similar ecosystems. Furthermore, forest restoration plays a crucial role in achieving carbon neutrality [40]. The proposed restoration species and climate-adaptive restoration strategies can serve as carbon sinks. The importance of forest restoration has been highlighted in international climate strategies such as the United Nations Framework Convention on Climate Change (UNFCCC) and the REDD+ program [42]. Notably, key species such as Quercus mongolica and Acer pseudosieboldianum are expected to have a significant impact on long-term carbon sequestration, strengthening the alignment of this research with global carbon neutrality initiatives.

The proposed restoration strategy aligns with the principles of Nature-Based Solutions (NbS), rather than being solely reliant on engineering approaches [43]. Global organizations such as the European Union (EU) and the World Bank have recently prioritized NbS as a key strategy for climate change mitigation and disaster prevention. The study contributes to these global trends by presenting practical applications of NbS. The proposed restoration species and techniques serve as representative examples of NbS and should be further validated through comparisons with existing research on NbS effectiveness. Collaborative efforts with countries possessing similar ecological conditions could facilitate the broader implementation of these methods.

Additionally, the findings contribute to international forest restoration networks, offering opportunities for broader cooperation. Given the ecological significance of the Baekdudaegan region as a key component of the Northeast Asian ecological corridor, trilateral collaboration between South Korea, China, and Japan should be explored. Moreover, efforts should be made to establish transnational cooperation strategies to support ecosystem-based restoration policies. Among the 94 native plant species designated by the Korea Forest Service for forest restoration, 31 species identified in this study were found to be consistent with official restoration species lists. This suggests that the study’s findings can directly contribute to policy development and ecological restoration practices [44]. A comparison between the species proposed in this study and those officially announced revealed 31 matching taxa. Therefore, the species proposed in this study are considered applicable to forest restoration policy development and ecological restoration practices (

Table A4. Proposal for Restoration Species by Environmental Type.

Type	Scientific Name	Type 1	Type 2	Type 3	Type 4	Type 5	Type 6
T1	Quercus aliena						O
	Acer pictum var. mono				O	O
	Quercus variabilis		O		O		O
	Prunus padus			O
	Carpinus cordata				O
	Ulmus davidiana var. japonica			O		O
	Zelkova serrata			O	O		O
	Maackia amurensis	O		O
	Fraxinus rhynchophylla	O	O	O	O	O
	Prunus serrulata f. spontanea				O
	Acer triflorum				O
	Morus bombycis				O
	Carpinus laxiflora				O		O
	Pinus densiflora		O		O		O
	Fraxinus sieboldiana						O
	Quercus mongolica	O	O	O	O	O	O
	Quercus serrata		O	O	O		O
	Cornus controversa				O
	Tilia amurensis	O			O	O
	Magnolia sieboldii		O		O
T2	Toxicodendron trichocarpum						O
	Acer pseudosieboldianum	O	O	O	O	O	O
	Rhus chinensis				O
	Acer tataricum subsp. ginnala				O
	Styrax obassis				O	O	O
	Acer barbinerve			O
S	Stephanandra incisa	O		O	O
	Ribes mandshuricum			O
	Symplocos sawafutagi	O		O
	Sambucus williamsii			O
	Corylus sieboldiana var. mandshurica	O		O
	Clematis urticifolia					O
	Rubus crataegifolius	O	O		O	O	O
	Lindera obtusiloba		O		O	O	O
	Lespedeza bicolor				O
	Lonicera praeflorens						O
	Callicarpa japonica				O		O
	Lespedeza maximowiczii	O	O			O	O
	Sasa borealis	O	O	O		O
	Rhododendron mucronulatum		O				O
	Lespedeza cyrtobotrya		O
	Rhododendron schlippenbachii		O				O
	Zanthoxylum piperitum				O		O
	Syringa pubescens subsp. Patula			O
	Aria alnifolia	O
H	Carex humilis var. nana		O	O	O		O
	Galium maximowiczii						O
	Dendranthema zawadskii var. latiloba						O
	Carex lanceolata	O	O			O
	Aster ageratoides					O
	Achnatherum pekinense					O
	Viola albida var. chaerophylloides			O			O
	Artemisia stolonifera	O	O		O	O
	Ainsliaea acerifolia		O	O
	Carex siderosticta	O	O		O	O
	Polygonatum odoratum var. pluriflorum	O
	Arisaema amurense			O
	Eupatorium japonicum						O
	Artemisia keiskeana						O
	Adenophora remotiflora			O
	Viola selkirkii	O
	Impatiens textorii				O
	Meehania urticifolia				O
	Thalictrum tuberiferum			O
	Isodon inflexus	O				O
	Atractylodes ovata	O
	Saussurea grandifolia					O
	Calamagrostis arundinacea	O	O
	Isodon excisus	O	O	O	O
	Liparis kumokiri			O
	Boehmeria japonica						O
	Viola japonica			O
	Diarrhena japonica		O		O	O
	Syneilesis palmata						O
	Convallaria keiskei			O
	Oplismenus undulatifolius				O		O
	Aster scaber					O	O
	Pseudostellaria palibiniana	O		O	O
	Spodiopogon sibiricus	O	O				O
	Viola albida			O
	Phryma leptostachya var. oblongifolia					O	O
	Crepidiastrum denticulatum					O
	Athyrium niponicum	O
	Pteridium aquilinum var. latiusculum	O
	Dryopteris crassirhizoma	O		O
	Osmunda cinnamomea			O
	Athyrium yokoscense	O		O
	Equisetum hyemale			O		O
	Arachniodes borealis			O
	Athyrium monomachii			O

Type 1: highland; Type 2: medium altitude; Type 3: poor soil; Type 4: deep soil; Type 5: wetland; Type 6: mixed forest.

and

Table A5. Proposal for Restoration Species by Damage Type Selected by the Notice.

Type	Scientific Name	Gl	Al	Fr	Ra	Ct	Ls	Aq
T1	Betula costata							O
	Acer pictum var. mono		O	O	O		O	O
	Quercus variabilis	O	O	O		O		O
	Zelkova serrata		O		O	O
	Fraxinus rhynchophylla	O	O	O	O	O	O	O
	Betula schmidtii	O		O		O		O
	Lindera obtusiloba	O	O	O		O	O	O
	Pinus densiflora	O	O	O		O	O	O
	Quercus mongolica	O	O	O	O	O	O	O
	Pinus koraiensis					O		O
	Quercus serrata		O	O	O	O	O	O
	Cornus controversa	O	O	O				O
	Tilia amurensis	O	O	O	O		O	O
	Magnolia sieboldii	O		O	O	O	O	O
T2	Toxicodendron trichocarpum			O		O	O	O
	Acer pseudosieboldianum	O	O	O	O	O	O	O
	Rhus chinensis		O					O
	Styrax obassis	O	O	O		O		O
S	Stephanandra incisa	O	O	O	O	O	O	O
	Symplocos sawafutagi	O	O		O	O	O	O
	Lespedeza bicolor	O	O	O		O	O
	Lespedeza maximowiczii	O	O	O	O	O	O	O
	Rhododendron mucronulatum	O	O	O		O		O
	Lespedeza cyrtobotrya	O	O	O			O	O
	Aria alnifolia				O	O
H	Carex humilis var. nana	O	O	O	O	O	O	O
	Ainsliaea acerifolia	O	O		O	O	O	O
	Isodon inflexus		O	O	O		O	O

Gl: grassland, Al: agricultural land, Fr: forest roads, Ra: ranches, Ct: cemeteries, Ls: logging sites, Aq: abandoned quarries.

).

Enhancing collaboration on sustainable forest restoration at an international level will be crucial, and the proposed restoration strategies and species selections could serve as important references for global ecological restoration initiatives.

5. Conclusions

This study systematically classified the types of degradation in the Baekdudaegan Protected Area, particularly in the Taebaeksan region, and proposed a restoration strategy based on environmental characteristics. Through field surveys, importance analysis, similarity analysis with reference ecosystems, and plant lifeform evaluations, a total of 91 suitable restoration species were identified. Among them, Fraxinus rhynchophylla, Quercus mongolica, and Acer pseudosieboldianum were found to be applicable to all types of degradation. These findings provide a scientific foundation for ecological restoration and suggest practical applications that can contribute to sustainable forest management and restoration planning.

This study goes beyond simply selecting restoration species by incorporating data-based restoration planning, thereby enhancing the scientific validity of restoration strategies. In particular, Self-Organizing Map (SOM) analysis was used to verify the suitability of the selected restoration species. Furthermore, 31 of the restoration species proposed in this study were found to be included in the list of 94 native plant species designated by the Korea Forest Service.

However, as this study was conducted at a specific time and location, further research is needed to evaluate the long-term survival rates and community dynamics of the restored vegetation. Additionally, biological interactions and ecological succession processes were not fully considered, nor were socioeconomic factors. Future research should incorporate continuous monitoring, remote sensing, and AI-based vegetation analysis to further refine restoration strategies.

This study presents a scientifically rigorous and practically applicable ecological restoration strategy for the Baekdudaegan Protected Area, aiming to enhance the objectivity and precision of restoration planning. Future research should focus on tracking long-term vegetation changes influenced by climate change and assessing the ecological stability of restored plant communities. Strengthening collaboration with local communities will also be crucial to ensuring a balanced approach that integrates both ecological and social considerations. With further research and continuous technological advancements, the restoration strategies proposed in this study can be further refined and contribute to improving the effectiveness of ecological restoration efforts.