Herbaceous Layer Response to Overstory Vegetation Changes in Quercus mongolica Fisch. ex Ledeb. Forests in Korea

Abstract

The development of overstory vegetation can considerably influence the composition and dynamics of herbaceous layer vegetation. However, the type of ecological processes underlying these changes remain poorly understood. We aimed to analyze changes in herbaceous layer species composition in Quercus mongolica Fisch. ex Ledeb. forests in central South Korea and identify the ecological processes driving succession, using zeta diversity and species turnover. We also sought to address regional bias in existing long-term monitoring data. Permanent 1 ha survey plots were established according to International Long Term Ecological Research Network guidelines, divided into 100 subplots. Data on species composition, crown openness, transmitted light, and structural variables were collected through four surveys (2014, 2015, 2017, and 2020) between 2014 and 2020. Zeta diversity and turnover metrics were used to evaluate succession dynamics. Species richness, cover, and turnover in the herbaceous layer were significantly correlated with overstory structure and rock cover. Crown openness and transmitted light declined but did not correlate with species turnover. Zeta diversity shifted from a power function model (2014–2017) to an exponential model (2020), indicating a shift from deterministic to stochastic processes. Successional changes in herbaceous vegetation may indicate a potential shift in forest structure in Q. mongolica stands—from stable, deterministic patterns to more variable, stochastic processes—highlighting the need for long-term monitoring in dynamic forest ecosystems.

1. Introduction

In temperate forests, species diversity is largely determined by the herbaceous layer [1,2,3]. Accordingly, understanding the diversity and dynamics of herbaceous vegetation, along with its relationships to biotic factors (e.g., the basal area and density of woody plants) and abiotic factors (e.g., canopy openness, light availability, and soil exposure), forms the foundation for analyzing stand structure. This knowledge of species diversity plays a critical role in designing, applying, and revising policies for the sustainable use and conservation of natural resources [4]. The development of the herbaceous layer is strongly influenced by the overstory canopy and is generally affected by variables such as light intensity, nutrient availability, throughfall, and soil moisture [5]. Additionally, herbaceous vegetation provides ecological functions such as protecting seeds and seedlings that serve as food sources for animals and offering shelter within forest ecosystems [6]. Although the herbaceous layer comprises a small portion of the total biomass, it minimizes nutrient loss from soil due to rainfall and supplies essential nutrients that support the development of woody vegetation [7,8,9].

Building upon this ecological importance, growing international concern over climate change and biodiversity loss—underscored by major agreements such as the 1992 United Nations Framework Convention on Climate Change and the 2015 Paris Agreement—has emphasized the need for long-term ecological monitoring to understand and conserve forest ecosystems [10]. In response, global and national efforts have established long-term ecological research (LTER) programs to track temporal changes in biodiversity, ecosystem functions, and environmental drivers. The U.S. National Science Foundation’s LTER program, initiated in 1980, laid the foundation for building ecological time-series datasets essential for evaluating ecosystem structure and function [11,12,13]. In South Korea, the Korea Long Term Ecological Research (KLTER) network was established in 1997 to monitor forest biodiversity, stand dynamics, and environmental changes across diverse ecosystems, including forests, coastal zones, and islands [14].

In addition to spatial considerations, forest ecosystems are fundamentally shaped by temporal changes in species composition and vertical structure [15,16]. These forest or stand dynamics involve shifts in vegetation layers and fluctuations in abiotic factors such as light availability, nutrient status, and soil moisture. Understanding these gradual transitions requires long-term data that capture ecological processes such as competition, succession, and niche differentiation [17,18,19].

Consequently, many studies have explored how the herbaceous layer reflects vertical stratification and mediates overstory–understory interactions, particularly in forests dominated by Quercus species. These widely distributed canopy trees influence understory light environments and soil characteristics, thereby shaping herbaceous composition [20,21,22]. Comparable vertical linkages have also been observed in tropical ecosystems, where changes in canopy structure due to invasive species or natural disturbance affect floristic diversity and edaphic conditions [23,24].

Alongside structural patterns, increasing attention has been directed toward the temporal dynamics of herbaceous vegetation in response to forest gap formation, disturbance, and succession. For instance, Halpern and Spies [3] demonstrated the sensitivity of herbaceous species to post-disturbance canopy changes, while Mölder et al. [25] found that thinning practices significantly modified herb-layer diversity through microclimatic changes. These studies underscore the herbaceous layer’s role as a dynamic indicator of ecosystem change.

In the Korean context, herbaceous vegetation dynamics have been investigated primarily in Pinus-dominated forests, focusing on the relationship between understory structure and site conditions [26,27]. Additional studies have examined responses of herb-layer vegetation to canopy disturbances and gap formation [28,29], and the spatial distribution of herbaceous species in island regions has been analyzed using ordination techniques such as Detrended Correspondence Analysis [30]. However, due to the complex topography and high floristic heterogeneity of Korean forests, identifying clear successional trends and structural relationships is difficult over short temporal scales. In this regard, long-term ecological monitoring serves as an essential tool for assessing ecosystem change and biodiversity under both natural and anthropogenic influences [31,32,33].

Nevertheless, most KLTER sites are concentrated in the Gangwon region (temperate northern zone) and in southern warm-temperate forests, resulting in a spatial gap in the central region. To address this imbalance and contribute to a more comprehensive understanding of national forest ecosystems, this study focuses on the early-stage structural changes in the herbaceous layer of temperate Quercus mongolica Fisch. ex Ledeb. forests in central South Korea. Given the herbaceous layer’s responsiveness to overstory conditions and its role as an early indicator of ecological succession [34], our objective is to analyze temporal changes in understory composition and provide baseline data for future studies on forest dynamics and long-term monitoring.

2. Materials and Methods

2.1. Conditions and Environment for Permanent Survey Plot Designation

The permanent survey plots were established following the guidelines of the ILTER network, based on three key principles: ① ecological representativeness, ② environmental homogeneity and accessibility, and ③ securing infrastructure and legal stability [11,35].

Among these, ecological representativeness was the primary consideration. Plots were selected to reflect major regional ecosystems and nationally significant vegetation types. A preliminary survey conducted in 2013 identified representative communities within the Research Forest of Kyungpook National University (KNU). Based on this, stands with Q. mongolica in the overstory and Lindera obtusiloba Blume in the shrub layer were confirmed as representative, and permanent plots were selected accordingly [36]. These Q. mongolica-L. obtusiloba communities are characteristic of South Korean temperate forests, particularly in the central region [37].

To support a comprehensive understanding of national ecosystems, addressing the spatial bias in existing long-term monitoring sites is essential [33]. Most ILTER-designated sites in South Korea are concentrated in either the southern or northern regions [11]. This study aimed to provide baseline data from the underrepresented central region, thereby reducing geographic bias (Figure 1).

To ensure environmental homogeneity, we established survey plots with uniform slope direction. Areas with evident natural or anthropogenic disturbances (e.g., landslides) were excluded. Only undisturbed sites were chosen, and a minimum buffer of 50 m from forest roads was maintained to minimize edge effects. Accessibility was also considered to facilitate continued fieldwork.

The study area is located within the KNU Research Forest, under the jurisdiction of the College of Agriculture and Life Sciences at KNU, which provides legal protection and formal forest management. The study site was designated as an educational and experimental forest in 1961. The forest covers approximately 635 hectares, consisting of 262 ha of coniferous forest, 216 ha of broadleaved forest, and 74 ha of mixed forest, with the remaining area comprising non-stocked forest land. A 35 ha zone has also been designated as a restricted management area for ecological research. The forest has been managed primarily for academic and research purposes. The surrounding vegetation is representative of temperate forest ecosystems in Korea, predominantly featuring Quercus mongolica [38].

The geographic coordinates of the sites are N 36°10′42.3″–N 36°12′17.8″, E 128°59′45.9″–E 129°1′21.5″. The site is situated at Mt. Myeonbongsan (1121 m) in Cheongsong-gun, Gyeongsangbuk-do, with Mt. Bohyeonsan (1124 m) nearby. The area is characterized by high elevation, steep terrain, and several valleys [36].

The climate at the survey site is continental, with marked temperature differences between summer and winter, and rainfall concentrated in the summer months, from June to late August [39]. Meteorological data from 2010 to 2020 (Figure 2) show a mean annual rainfall of 967 mm. Annual rainfall was lower in 2015 and 2017, at 707.3 mm and 783.9 mm, respectively, with particularly severe spring droughts. In contrast, rainfall was higher in 2017 (1008.0 mm), 2018 (1139.7 mm), and 2020 (1285.8 mm), especially in July and August.

Mean temperatures followed a regular annual pattern, with maximum values in July or August and minimum values in December. These patterns indicate that the site exhibits a typical continental climate with four distinct seasons. The summers of 2018 and 2020 had particularly high mean temperatures, reaching 25.7 °C in July 2018 and 25.8 °C in August 2020.

Table 1. Topographic characteristics of the permanent survey plots.

Index	Plot 1	Plot 2	Plot 3	Plot 4
Aspect	N	N	NE	N
Rock coverage (%)	35.0	20.2	45.9	21.3
Altitude (m)	543	579	636	770
Soil type	Brown forest soil
Soil texture	Sandy loam
Soil humidity	Slightly dry
Landform	Middle hill

shows the topographic characteristics of the permanent survey plots. The slope direction was north to northeast, rock exposure was 30.6%, and the mean elevation was 632 m. The soil type was brown forest soil with a sandy loam texture, and the soil environment was slightly dry [40]. The plots were located on the midslope.

2.2. Permanent Survey Plot Placement and Survey Methods

To facilitate comparison with other long-term ecological monitoring studies, the survey area was initially planned as a 1 ha (100 × 100 m) subplot, in accordance with the standard area recommended by the National Institute of Forest Science [14]. However, due to limitations such as topographic conditions, the distribution of representative plant communities, and accessibility for long-term monitoring, a single 1 ha plot was deemed unsuitable. Instead, we positioned four separate 50 × 50 m plots that better represented the vegetation patterns than a single 1 ha plot. Each plot was further subdivided into twenty-five 10 × 10 m subplots, yielding a total of 100 subplots.

After marking the central region, compass surveying was conducted, tags were placed at 10 m intervals, and plot boundaries were delineated with a rope. The permanent plots were established in June 2014, and tree surveys were conducted four times over 7 years (2014, 2015, 2017, and 2020) between July and September.

The tree layer was defined as individuals with a diameter at breast height (DBH) of at least 2 cm and a height of at least 3 m. The shrub layer included individuals with a DBH of at least 2 cm and a height of 1.5–3 m. For each tree, species were identified, DBH was measured, and an aluminum tag with a unique ID was attached. DBH measurements were taken annually at the same point, approximately 3 cm below the tag.

Figure 3 shows a layout of the survey plots for herbaceous vegetation and crown photography. Understory vegetation was assessed using the Braun-Blanquet method [41], which is widely used in Asian vegetation surveys and effective for evaluating species composition [42]. In each subplot, five 2.5 × 2.5 m herbaceous layer plots were established. Species presence and mean coverage from these were used as vegetation data for the entire subplot. Species identification relied on research reports from the Korea National Arboretum and a fern atlas [43,44,45,46,47,48], with nomenclature based on the Knowledge System of National Species in Korea [49].

To ascertain the effects of overstory crown openness and transmitted light on the herbaceous layer, crown photographs were taken using a Canon EOS 6D camera (Canon Inc., Tokyo, Japan) equipped with a Canon EF 8–15 mm fisheye lens. Photographs were consistently taken from a fixed point at the center of each subplot, at a height of 1.2 m and facing north. A total of 100 photographs were taken per year. To minimize lighting bias, images were captured between 05:00 and 10:00 or between 15:00 and 18:00, or on overcast days [50,51,52]. Leaf litter depth around each photography point was also measured annually.

2.3. Analysis

To ascertain changes in stand structure within the tree and shrub layers, the basal area and tree density were analyzed. Basal area was calculated by determining the cross-sectional area of each tree using its diameter at DBH, and the values were converted to m²/ha. Crown openness and transmitted light were quantified using fisheye lens photographs, which were analyzed with Gap Light Analyzer ver. 2.0 (Simon Fraser University, Burnaby, BC, Canada; Figure 4).

To monitor changes in overall herbaceous layer coverage, vegetation cover surveys were conducted, and the species diversity, evenness, and richness were analyzed. Species richness was defined as the total number of plant species present within each survey plot, while species diversity and evenness were calculated using the Shannon Index [53].

To evaluate species dominance and changes in composition within the herbaceous layer, importance values (IVs) were calculated [54]. Importance values were defined as the average of the relative coverage and relative frequency. Relative coverage was estimated based on the median of the Braun-Blanquet cover class for each species [55] (

Table 2. Braun-Blanquet cover scale and median values.

Braun-Blanquet Scale	Range of Cover (%)	Median Value
5	75–100	87.5
4	50–75	62.5
3	25–50	37.5
2	12.5–25	18.75
1	<12.5: numerous individuals	9.375
+	<5: few individuals	4.69
r	<1: few and unique individuals	1.01

).

To analyze shifts in species composition and spatial distribution in the herbaceous layer across the permanent survey plots, non-metric multidimensional scaling (NMDS) was performed. Based on Euclidean distance, two axes were selected according to the highest R² values, and species composition was visualized in a two-dimensional space [56]. NMDS is a suitable method for analyzing non-parametric data in ecosystems with a high likelihood of discontinuity [56]. Explanatory variables included in the NMDS analysis were the basal area and density of the seven dominant tree species, total basal area, total tree density, evenness, species richness, species diversity, turnover rate, crown openness, transmitted light, understory vegetation cover, rock exposure, and leaf litter depth. The Sørensen distance was used in a multi-response permutation procedure (MRPP) to test for interannual heterogeneity in species composition in NMDS space [56].

To track temporal dynamics in understory species composition, specifically, the entry and exit of species, we analyzed turnover rates and zeta diversity. The turnover rate was calculated using the formula defined by Hallet et al. [57]:

This was defined, between two time points, as

$${\displaystyle \frac{\mathit{Number} \mathit{of} \mathit{species} \mathit{gained}+\mathit{Number} \mathit{of} \mathit{species} \mathit{lost}}{\mathit{Total} \mathit{number} \mathit{of} \mathit{species} \mathit{appearing} \mathit{in} \mathit{both} \mathit{time} \mathit{points}}}\times 100$$

This calculation was used to derive the mean annual turnover at the community level. Additionally, we applied the same principle to assess species-level dynamics, by quantifying how frequently each species was gained or lost across the 100 subplots between 2014 and 2020. From these values, we computed the Turnover Intensity, defined as the mean of the immigration and emigration rates. This index reflects the overall magnitude of species-level compositional change.

Zeta diversity was used to measure species turnover and compositional similarity across multiple survey plots. It represents the number of shared species across n plots (zeta order) and is visualized as a decreasing curve. As the zeta diversity reflects the number of shared species, it is inherently influenced by plot-level species richness [58]. For this analysis, 250 Monte Carlo simulations were performed. To interpret species turnover patterns, both exponential and power function regression models were fitted to the zeta diversity data, and model fit was assessed using Akaike’s Information Criterion (AIC) [59]. AIC evaluates the balance between model accuracy and complexity, with smaller AIC values indicating better model performance [60].

The test result for model fit can be either positive or negative, but lower values indicate a more reliable model [60]. Annual ecosystem changes can be interpreted based on the significance of the exponential and power function models [58]. The exponential model supports the hypothesis of stochastic processes, characterized by rapid shifts in species composition due to random appearance and loss. In contrast, the power function model supports deterministic processes, where species composition changes in a more structured and stable manner due to species turnover, niche processes, and environmental filtering [60]. The stochastic process hypothesis suggests that species composition is shaped by random habit formation and incidental factors, aligning with neutral theory [60]. Conversely, the niche process hypothesis posits that heterogeneity in species distribution is primarily driven by environmental variables and interspecific interactions [61,62].

For the Pearson correlation analysis, we used 100 subplot units (10 × 10 m each) as replicates. Thirteen explanatory variables were measured in each subplot, including the basal area and density of the tree and shrub layers, leaf litter depth, rock exposure, species diversity, evenness, species richness, vegetation cover, species turnover, transmitted light, and crown openness.

The R packages Hmisc, vegan, and zetadiv in R Studio ver. 4.3.1. (R Foundation for Statistical Computing, Vienna, Austria) were used to perform one-way analysis of variance (ANOVA) for annual comparisons, Pearson correlation analysis, and zeta diversity analysis [63]. PC-ORD ver. 7.0 (MjM software, Corvallis, OR, USA) was used for the MRPP test and NMDS analysis.

3. Results

3.1. Changes in the Tree and Shrub Layers

Table 3. Changes in basal area in the tree and shrub layers. T: tree layer, S: shrub layer (units: m²/ha).

Contents	2014		2015		2017		2020
	T	S	T	S	T	S	T	S
Pinus densiflora Siebold & Zucc.	14.277	0.126	14.415	0.126	14.251	–	14.251	–
Quercus mongolica Fisch. ex Ledeb.	13.670	0.274	13.713	0.324	13.665	0.428	13.711	0.435
Betula schmidtii Regel	1.810	0.159	1.676	0.354	1.639	0.407	1.554	0.431
Fraxinus sieboldiana Blume	0.604	0.070	0.686	0.085	0.712	0.086	0.795	0.091
Rhus tricocarpa Miq.	0.315	0.093	0.334	0.106	0.348	0.114	0.349	0.170
Acer pseudosieboldianum (Pax) Kom.	0.125	0.122	0.131	0.137	0.154	0.125	0.177	0.141
Rhododendron schlippenbachii Maxim.	0.128	0.062	0.148	0.076	0.159	0.083	0.188	0.094
Lindera obtusiloba Blume	0.106	0.053	0.099	0.060	0.105	0.068	0.105	0.088
Rhododendron mucronulatum Turcz.	0.267	0.051	0.067	0.064	0.077	0.054	0.074	0.057
Sorbus alnifolia (Siebold & Zucc.) K.Koch	0.049	0.083	0.052	0.103	0.062	0.119	0.076	0.141
Prunus serrulata var. pubescens (Makino) Nakai	0.145	0.086	0.151	0.088	0.108	0.086	0.110	0.086
Styrax obassia Siebold & Zucc.	0.032	0.064	0.034	0.080	0.041	0.091	0.066	0.098
Magnolia sieboldii K.Koch	–	0.102	–	0.123	–	0.141	–	0.157
Betula davurica Pall.	0.265	–	0.268	–	0.251	–	0.286	–
Ilex macropoda Miq.	0.211	0.057	0.215	–	0.209	–	0.219	–
Quercus variabilis Blume	0.179	–	0.181	–	0.177	–	0.190	–
Fraxinus rhynchophylla Hance	0.082	0.109	0.085	0.127	0.092	0.119	0.095	0.138
Tilia amurensis Rupr.	0.042	0.152	0.043	0.166	0.044	0.173	0.059	0.188
Acer pictum subsp. mono (Maxim.) Ohashi	0.007	0.031	0.004	0.035	0.006	0.045	0.015	0.045
Carpinus laxiflora (Siebold & Zucc.) Blume	0.008	0.082	0.010	0.097	0.010	0.100	0.012	0.015
Ulmus davidiana var. japonica Rehder) Nakai	0.008	–	0.008	–	0.008	–	0.013	–
Philadelphus tenuifolius Rupr. & Maxim.	–	0.044	–	0.059	–	0.060	–	0.063
Corylus sieboldiana Blume	–	0.052	–	0.083	–	0.102	–	0.105
Symplocos chinensis f. pilosa (Nakai) Ohwi	–	0.066	–	0.071	–	0.073	–	0.492
Maackia amurensis Rupr.	0.004	0.075	0.004	0.080	0.005	0.080	0.005	–
Quercus serrata Thunb.	0.003	0.337	0.003	0.342	0.004	0.363	0.004	0.374
Pinus koraiensis Siebold & Zucc.	–	0.152	–	0.188	–	0.188	–	0.188
Rhus sylvestris Siebold & Zucc.	–	0.080	–	0.085	–	0.107	–	0.119
Weigela subsessilis (Nakai) L.H.Bailey	–	0.038	–	0.045	–	0.057	–	0.067
Total	32.335	2.623	32.327	3.104	32.114	3.269	32.354	3.783
Total (all layers)	34.958		35.431		35.383		36.137

presents the changes in basal area within the permanent survey plots. The total basal area increased from 34.958 m²/ha in 2014 to 36.137 m²/ha in 2020, a rise of 1.179 m²/ha over 7 years. While the tree layer showed only a slight increase of 0.019 m²/ha, the shrub layer accounted for most of the change, increasing by 1.160 m²/ha. Compared to 2014, the basal area declined until 2017 and then increased in 2020.

In the tree layer, Pinus densiflora Siebold & Zucc., Betula schmidtii REGEL., and Rhododendron mucronulatum Tucz. var. mucronulatum showed decreasing trends, whereas Q. mongolica, Rhus trichocarpa (Miq.), and Fraxinus sieboldiana Blume showed increases. In the shrub layer, most species showed increasing trends, with Q. mongolica in particular displaying a notable rise in basal area.

In previous studies of Quercus-dominated stands in South Korea, a Q. mongolica stand on Mt. Gyebangsan increased by 1.59 m²/ha (from 40.53 m²/ha in 2007 to 42.12 m²/ha in 2012); a Q. serrata Thunb. ex. Murray stand in the Gwangneung Research Forest decreased slightly by 0.21 m²/ha (from 32.81 m²/ha in 2012 to 32.60 m²/ha in 2020); a Q. serrata stand on Mt. Geumsan, Namhae, increased by 1.22 m²/ha (from 31.40 m²/ha in 2007 to 32.62 m²/ha in 2011); and a long-term ecological monitoring site on Mt. Jeombongsan showed an increase of 2.70 m²/ha (from 40.00 m²/ha in 2012 to 42.70 m²/ha in 2020). Compared to these sites, the basal area in the present study was lower than those at Mt. Gyebangsan and Mt. Jeombongsan [14,64].

For the Q. mongolica stand in the current study, the basal area increased at an average rate of 0.197 m²/ha per year. Compared with other long-term ecological plots, except the Q. serrata stand at the Gwangneung Research Forest, woody plant growth in this study was relatively slower.

Table 4. Changes in tree density in the tree and shrub layers (units: stems/ha).

Contents	2014		2015		2017		2020
	T	S	T	S	T	S	T	S
Pinus densiflora Siebold & Zucc.	161	1	161	1	146	–	145	–
Quercus mongolica Fisch. ex Ledeb.	579	19	578	21	551	28	553	28
Betula schmidtii Regel	135	11	135	11	131	10	127	10
Fraxinus sieboldiana Blume	45	561	41	568	40	559	40	564
Rhus tricocarpa Miq.	66	137	64	133	63	127	59	127
Acer pseudosieboldianum (Pax) Kom.	29	52	29	54	29	50	28	63
Rhododendron schlippenbachii Maxim.	3	238	1	231	1	224	1	240
Lindera obtusiloba Blume	1	123	1	128	1	128	1	136
Rhododendron mucronulatum Turcz.	2	123	1	120	1	108	1	121
Sorbus alnifolia (Siebold & Zucc.) K.Koch	3	26	3	22	3	22	3	18
Prunus serrulata var. pubescens (Makino) Nakai	16	28	16	26	16	25	16	23
Styrax obassia Siebold & Zucc.	6	15	4	16	4	14	4	14
Magnolia sieboldii K.Koch	–	18	–	19	–	19	–	17
Betula davurica Pall.	8	–	8	–	7	–	7	–
Ilex macropoda Miq.	3	1	3	–	3	–	3	–
Quercus variabilis Blume	6	–	6	–	5	–	5	–
Fraxinus rhynchophylla Hance	3	2	3	2	3	1	3	1
Tilia amurensis Rupr.	1	1	1	1	1	1	1	1
Acer pictum subsp. mono (Maxim.) Ohashi	2	1	2	1	2	1	2	3
Carpinus laxiflora (Siebold & Zucc.) Blume	1	8	1	8	1	8	1	8
Ulmus davidiana var. japonica Rehder) Nakai	1	–	1	–	1	–	1	–
Philadelphus tenuifolius Rupr. & Maxim.	–	8	–	9	–	9	–	9
Corylus sieboldiana Blume	–	6	–	8	–	8	–	8
Symplocos chinensis f. pilosa (Nakai) Ohwi	–	6	–	6	–	6	–	6
Maackia amurensis Rupr.	1	1	1	1	1	1	1	–
Quercus serrata Thunb.	2	2	2	2	2	2	2	5
Pinus koraiensis Siebold & Zucc.	–	1	–	1	–	1	–	1
Rhus sylvestris Siebold & Zucc.	–	1	–	1	–	1	–	1
Weigela subsessilis (Nakai) L.H.Bailey	–	1	–	1	–	1	–	3
Total	1074	1391	1062	1391	1012	1354	1004	1407
Total (all layers)	2465		2453		2366		2397

presents the changes in stem density in the tree and shrub layers within the permanent survey plots. The overall tree density decreased by 54 stems/ha, from 2465 stems/ha in 2014 to 2411 stems/ha in 2020. In the tree layer, P. densiflora decreased by 16 stems/ha (from 161 to 145 stems/ha), Q. mongolica by 26 stems/ha (from 579 stems/ha to 553 stems/ha), and B. schmidtii by 8 stems/ha (from 135 stems/ha to 127 stems/ha).

In the shrub layer, the density declined in 2017 but rebounded by 2020, with Q. mongolica and L. obtusiloba increasing by 9 and 13 stems/ha, respectively. Overall, the change in tree density over the 7-year period was minimal.

At a Q. mongolica stand on Mt. Jeombongsan, the tree density decreased slightly from 915 trees in 2012 to 901 trees in 2020. Notably, Acer pseudosieboldianum (Pax) Kom. density increased significantly, while Q. mongolica declined [48]. This pattern is characteristic of mature stand structure unaffected by disturbance, with increasing dominance of shade-tolerant species and a decline in shade-intolerant ones [65]. In our study, both species showed decreasing trends in the tree layer, with a greater decline observed for Q. mongolica, consistent with previous findings.

Conversely, in the shrub layer, both Q. mongolica and A. pseudosieboldianum showed increasing population densities, suggesting within-stratum competition typical of secondary succession. These trends indicate a less mature stand structure compared to previously studied long-term plots of similar vegetation.

3.2. Changes in Crown Openness and Transmitted Light

Figure 5 shows the changes in crown openness and transmitted light. Crown openness was 9.21% in 2014, increased to 11.11% in 2015, then declined to 6.42% in 2017 and 6.40% in 2020. These annual changes were statistically significant (p < 0.001). Transmitted light followed a similar downward trend, decreasing from 3.19 µmol·m⁻²·d⁻¹ in 2014 and 3.18 µmol·m⁻²·d⁻¹ in 2015 to 2.44 µmol·m⁻²·d⁻¹ in 2017 and 2.26 µmol·m⁻²·d⁻¹ in 2020, consistent with the observed reduction in crown openness.

In Q. mongolica stands, fluctuations in crown openness and transmitted light are common. These stands are largely established through vegetative regeneration, where parts of the tree other than the main branches die sequentially. This leads to spatial competition, and new individuals emerge once the competition subsides [36,66,67]. Consequently, the crown undergoes cycles of opening and closing, resulting in repeated variations in both crown openness and light transmission.

3.3. Changes in Species Composition and Biodiversity Among Understory Vegetation

Figure 6 shows the understory plant cover at each survey time point. From 2014 to 2020, plant cover was recorded as 36.45 ± 0.55% (2014), 44.10 ± 0.64% (2016), 39.15 ± 0.64% (2017), and 55.55 ± 1.21% (2020). Understory plant cover was highest in 2020, and the differences between years were statistically significant (p < 0.001).

Table 5. Changes in herbaceous layer species diversity (one-way ANOVA, Tukey’s post-hoc test; a, b show post-hoc differences between the mean values).

Contents	2014	2015	2017	2020	F	p
H’	2.5057 ± 0.0241 ab	2.5274 ± 0.0270 a	2.4151 ± 0.0312 b	2.5114 ± 0.0323 ab	3.083	0.027
SR	19.08 ± 0.47	19.37 ± 0.51	18.96 ± 0.66	20.24 ± 0.62	1.019	0.384
Even.	0.8579 ± 0.0046 a	0.8609 ± 0.0041 a	0.8345 ± 0.0053 b	0.8455 ± 0.0054 ab	6.142	<0.001

H’, species diversity; SR, species richness; Even., evenness.

presents changes in species diversity in the herbaceous layer of the permanent survey plots. Over 7 years, the species diversity increased slightly from 2.5057 ± 0.0241 in 2014 to 2.5114 ± 0.0323 in 2020 (an increase of approximately 0.0057). Species richness increased from 19.08 ± 0.47 in 2014 to 20.24 ± 0.62 in 2020 (an increase of approximately 1.16), while evenness decreased from 0.8579 ± 0.0046 to 0.8455 ± 0.0054 (a decrease of approximately 0.0124). Annual differences in species diversity and evenness were statistically significant (p < 0.05), whereas differences in species richness were not significant (p = 0.384).

Table 6. Annual changes in importance values (IVs) of representative herbaceous species in the permanent survey plots. Species were selected based on their high IV rankings. A total of 139 taxa were recorded, but only dominant species with relatively high IVs are presented here.

Species	Year
	2014	2015	2017	2020
Fraxinus sieboldiana Blume	10.85	12.25	15.92	13.05
Vaccinium hirtum var. koreanum (Nakai) Kitam.	8.87	9.20	9.15	9.28
Lindera obtusiloba Blume	7.77	8.63	9.75	9.39
Rhododendron schlippenbachii Maxim.	8.20	8.17	7.74	9.81
Carex okamotoi Ohwi	8.03	8.01	7.63	6.41
Rhododendron mucronulatum Turcz.	7.09	7.13	6.48	4.68
Acer pseudosieboldianum (Pax) Kom.	4.95	4.52	4.27	3.27
Rhus tricocarpa Miq.	4.66	4.51	4.02	3.63
Ainsliaea acerifolia Sch.Bip.	4.08	3.13	3.95	5.70
Styrax obassia Siebold & Zucc.	3.29	3.56	3.55	3.65
Hosta capitata (Koidz.) Nakai	3.00	3.55	2.62	2.99
Tripterygium regelii Sprague & Takeda	0.99	1.05	1.68	2.91
Quercus mongolica Fisch. ex Ledeb.	1.64	2.41	2.55	2.18
Polygonatum odoratum var. pluriflorum (Miq.) Ohwi	3.05	2.26	1.37	1.57
Calamagrostis arundinacea (L.) Roth	2.57	2.13	1.58	1.88
Omitted (species)	20.95 (82)	19.49 (77)	17.74 (85)	19.59 (73)

presents the annual changes in IVs for individual species in the herbaceous layer of the permanent survey plots. In 2020, the species with the highest IVs, in descending order, were F. sieboldiana (13.05) > R. schlippenbachii (9.81) > L. obtusiloba (9.39) > Vaccinium hirtum var. koreanum (Nakai) Kitam. (9.28). Notably, F. sieboldiana showed the greatest increase in IV over the study period, rising by 2.20 between 2014 and 2020. Both L. obtusiloba and R. schlippenbachii, which are mostly found in Quercus spp. forests in South Korea, also showed increases of 1.62 and 1.61, respectively.

In contrast, A. pseudosieboldianum, which showed increased density in the shrub layer, showed a gradual decline in IV, while Q. mongolica showed a slight upward trend. As previously discussed, the tree density increased in the shrub layer, suggesting that stratification may have developed following the vertical expansion of A. pseudosieboldianum from the herbaceous layer.

Carex okamotoi, which is native to South Korea [68], showed a gradual decrease in IV, decreasing from 8.03 in 2014 to 6.41 in 2020. While F. sieboldiana maintained the highest IV overall, year-to-year fluctuations were observed among the top-ranking species. For instance, although the IV of L. obtusiloba slightly decreased from 9.75 in 2017 to 9.39 in 2020, it still represented an increase relative to 2014.

Using NMDS analysis, changes in species composition in the herbaceous layer were visualized in two-dimensional space (Figure 7). The stress value of the NMDS ordination was 17.445, indicating a good fit with real data [56]. Axes 1 and 2 had explanatory powers of 0.652 and 0.211, respectively, yielding a combined explanatory power of 0.863.

Between 2014 and 2017, the spatial distribution of species composition shifted incrementally, with directional changes observed at each time point. By 2020, a marked increase in the magnitude of compositional change was detected, along with greater variability in directional trends.

The environmental variables most closely associated with shifts in species composition included the tree density, species diversity, species richness, and herbaceous layer plant cover. Among species, L. obtusiloba and F. sieboldiana had particularly strong influences on composition, likely due to their status as dominant shrubs in the permanent survey plots. Their relatively even distribution and increased inter-individual variation likely contributed to their strong compositional impact.

To test the heterogeneity of species composition plotted in 2D space, an MRPP test was performed (

Table 7. MRPP test results in the herbaceous layer, by year (T: T statistic, A: similarity index).

Compared Year			T	A	p
2014	vs.	2015	−12.846	0.038	<0.001
2014	vs.	2017	−6.603	0.018
2014	vs.	2020	−62.919	0.136
2015	vs.	2017	−5.361	0.016
2015	vs.	2020	−26.801	0.065
2017	vs.	2020	−44.340	0.100

). The MRPP T statistic reflects the mean distance between plots based on species composition; more negative values indicate greater heterogeneity and clearer separation between groups. The A value is a similarity index that compares the observed within-group agreement to that expected by chance, with values closer to one indicating higher within-group similarity, zero indicating randomness, and negative values indicating less similarity than expected by chance [56].

The T statistic and A were lowest in 2015 and 2017, indicating high heterogeneity and species composition, similar to that of randomly grouped data. In contrast, the heterogeneity between groups was greatest in 2014 and 2020, and the similarity index was closer to one, indicating high within-year similarity and distinct differences in species composition between those years. Notably, the comparison between 2017 and 2020 showed the second-highest heterogeneity, suggesting that significant shifts in herbaceous layer composition began to emerge after 2017. Thus, the seventh year marks the point at which clear changes in species composition became evident.

3.4. Species Turnover for Vegetation in the Herbaceous Layer

To assess species turnover and the underlying ecological processes in the herbaceous layer, zeta diversity decline curves were analyzed (Figure 8). To interpret the changes in zeta diversity, both the diversity values and slope were examined until convergence to zero. The zeta order value, corresponding to shared habitats, showed a rapidly decreasing number of shared species, at 49 in 2014, 55 in 2015, 52 in 2017, and 39 in 2020. These trends were also reflected in the slope of the zeta diversity curve. Up to a zeta order of 1–24, the slope became shallower over time, while at zeta orders of 25 or above, a rapid decline in the zeta diversity was observed in 2020.

The graph of decreasing zeta diversity supports a power function model from 2014 to 2017 and an exponential model in 2020. Thus, after around 7 years, the hypothesis regarding the drivers of ecosystem change shifted. In other words, changes in species composition in the herbaceous layer initially supported deterministic processes, but over time shifted toward stochastic processes.

Figure 9 shows the results for species turnover in the permanent survey plots by year. The overall species turnover from 2014 to 2020 was 48.01 ± 0.98%. Turnover between 2014 and 2015 was 18.91 ± 0.80, between 2015 and 2017 was 42.55 ± 0.98, and between 2017 and 2020 was 43.89 ± 1.03, indicating that turnover increased with longer intervals between surveys. The differences in turnover between years were statistically significant (p < 0.001).

Table 8. Turnover rates of plant species in the herbaceous layer in the permanent survey plots. Img. is the immigration rate, Emg. is the emigration rate, units: %. Turnover Intensity represents the average of the immigration and emigration rates, serving as a summary indicator of the temporal dynamism of each species.

Species	Img.	Emg.	Turnover Intensity
Quercus mongolica Fisch. ex Ledeb.	5.86	4.41	5.14
Rhus tricocarpa Miq.	4.02	5.55	4.79
Prunus serrulata var. pubescens (Makino) Nakai	4.48	4.55	4.52
Sorbus alnifolia (Siebold & Zucc.) K.Koch	2.76	4.55	3.66
Lindera obtusiloba Blume	4.02	2.99	3.51
Atractylodes ovata (Thunb.) DC.	4.83	1.14	2.98
Styrax obassia Siebold & Zucc.	2.99	2.84	2.92
Acer pseudosieboldianum (Pax) Kom.	2.07	3.41	2.74
Rhododendron mucronulatum Turcz.	2.18	3.13	2.66
Vaccinium hirtum var. koreanum (Nakai) Kitam.	2.30	2.84	2.57
Disporum smilacinum A. Gray	3.91	1.00	2.45
Calamagrostis arundinacea (L.) Roth	1.61	3.27	2.44
Rhododendron schlippenbachii Maxim.	1.49	3.13	2.31
Symplocos chinensis f. pilosa (Nakai) Ohwi	3.91	0.71	2.31
Tripterygium regelii Sprague & Takeda	2.30	1.71	2.00
Fraxinus sieboldiana Blume	1.61	2.13	1.87
Omitted (species)	49.66 (82)	52.64 (76)	–

shows the species turnover by plant species over 7 years in the permanent survey plots. A total of 98 species immigrated, while 92 species emigrated. The frequency ratio of immigrating species was highest, in descending order, for Q. mongolica (5.86), Atractylodes ovata (Thunb.) DC. (4.83), Prunus serrulata var. pubescens (Makino) Nakai (4.48), and Disporum smilacinum A. Gray and Symplocos chinensis for. pilosa (Nakai) Ohwi (both 3.91). In contrast, the species with the highest emigration frequencies were R. tricocarpa (5.55), and P. serrulata var. pubescens and Sorbus alnifolia (Siebold & Zucc.) K. Koch (both 4.55), followed by Q. mongolica (4.41), A. pseudosieboldianum (3.41), and Calamagrostis arundinacea (L.) Roth (3.27).

Plant species with high immigration ratios in the permanent plots included Q. mongolica, P. serrulata var. pubescens, L. btusiloba, A. ovata, Styrax obassia Siebold & Zucc., D. smilacinum, S. chinensis for. pilosa, and Tripterygium regelii Sprague & Takeda Species, and those with high emigration ratios included R. tricocarpa, P. serrulata var. pubescens, S. alnifolia, A. pseudosieboldianum, R. mucronulatum, V. hirtum var. koreanum, C. arundinacea, and Fraxinus sieboldiana. Notably, most of the dynamic turnover in frequency ratio was observed in canopy tree species, with Q. mongolica, R. tricocarpa, and P. serrulata var. pubescens exhibiting both frequent immigration and emigration events.

3.5. Correlations Between Biotic and Abiotic Factors over 7 Years

Correlations between various environmental variables were analyzed over 7 years, including species diversity and turnover in the herbaceous layer (Figure 10). Species diversity and species richness of the understory vegetation showed negative correlations with tree density and basal area in the shrub layer, and with tree density alone in the tree layer (p < 0.05). Abiotic factors were negatively correlated with rock cover (p < 0.01).

Herbaceous layer cover showed negative correlations with tree density and basal area in the shrub layer, and with rock cover (p < 0.01), while showing positive correlations with species diversity and richness (p < 0.01).

Species turnover in the herbaceous layer correlated with various ecological factors, showing negative correlations with tree density and basal area in the shrub layer, as well as significant negative correlations with leaf litter depth and rock cover. However, turnover did not show statistically significant correlations with transmitted light or crown openness (p > 0.05).

Overall, both biodiversity and turnover in the herbaceous layer were influenced by the species composition of the shrub layer. These results suggest that sharp changes in shrub layer population densities exert dynamic ecological pressure on short-term species composition in the understory vegetation.

4. Discussion

4.1. Relationships of Herbaceous Layer Species Composition with the Tree and Shrub Layers

We aimed to monitor short-term changes over 7 years in a temperate forest in South Korea by establishing permanent survey plots. Through this, we investigated changes in the tree and shrub layers, crown openness, and herbaceous layer species composition, and examined which ecological factors influenced the herbaceous species composition.

An analysis of change in the basal area and tree density within the 1 ha permanent survey plots revealed minimal changes in the overall species composition of the tree layer, which occupies a higher layer than the herbaceous layer. Pearson correlation analysis revealed no correlations between the basal area in the tree layer and either species diversity or plant cover in the herbaceous layer. These findings suggest that the stand structure of this temperate forest is still undergoing ecological succession. The basal area in the upper-story vegetation serves as an indirect indicator of canopy cover, and in less mature stands, the correlation between tree canopy cover and herbaceous layer species composition tends to be weak [69]. Thus, the forest in our study appeared to be in a successional phase.

One notable observation was the significant decline in P. densiflora, a major coniferous species in the tree layer. This trend reflects the ongoing successional process in which coniferous trees gradually give way to broadleaf species, resulting in the transformation of a mixed conifer–broadleaf forest into a deciduous broadleaf forest. Such a transition is characteristic of temperate forests undergoing secondary succession, which are often initially dominated by one or two key species before evolving into broadleaf-dominated systems [70].

Frequent disturbances in the tree layer, which lead to the formation of forest gaps, play a key role in species composition and diversity until canopy stabilization occurs [71]. These disturbances are a critical driver of secondary succession. The changes observed in the tree and shrub layers in our permanent plots align with previous research findings on species composition dynamics in secondary forests [71]. In early stages, species such as Q. mongolica and P. densiflora were dominant until the crown stabilized. However, minor disturbances arising from interspecific competition promoted gap formation, allowing for the expansion of the shrub layer [71].

With the emergence of forest gaps caused by natural mortality in the tree layer, a noticeable increase was observed in the basal area and density for shrub species, such as Rhododendron schlippenbachii, Rhododendron mucronulatum, L. obtusiloba, and Fraxinus sieboldiana, rather than in tree species. The formation of forest gaps has been shown to promote increased shrub layer density [72], and these structural changes have significant effects on the development of the herbaceous and understory layers [72,73,74].

Additionally, our results indicated a gradual decrease in transmitted light over the study period. This is likely due to the expansion of shrub layer vegetation, as photographs were consistently taken at a height of 1.2 m from the ground. The increased cover of the shrub layer over time contributed to a reduction in light penetration to the forest floor.

4.2. Biodiversity and Species Turnover in the Herbaceous Layer

Species diversity in the herbaceous layer showed statistically significant differences depending on the survey year, but no clear increasing or decreasing trend was observed. In contrast, evenness showed a declining trend over time. This pattern is commonly observed in temperate coniferous forests, where herbaceous layer diversity can vary irregularly due to spatiotemporal factors [75,76].

Evenness reflects how plant species are distributed across ecological niches within a community and indicates the extent of resource use in a habitat [77,78]. Evenness increases when species are present in relatively equal proportions and decreases when specific species become dominant [78]. Low evenness suggests that resources are concentrated among a few species, which may lead to increased invasion by external species and the loss of native species [78,79]. In this study, the gradual decline in evenness suggests that prolonged dominance by certain species led to uneven species turnover. In other words, this decrease, combined with ecological niche monopolization, was associated with abrupt shifts in species composition and increased exosystem instability, rather than gradual succession [80]. Thus, evenness serves as a key ecological indicator for assessing the extent and direction of species turnover.

This trend in declining evenness was also supported by the analysis of zeta diversity and species turnover. When changes in zeta diversity were analyzed, the decline in 2014, 2015, and 2017 was best modeled by a power function, whereas in 2020, a clear shift towards a decreasing exponential function was observed. The power function decline observed in 2014–2017 reflects niche-based structural species turnover, i.e., stable changes in species composition influenced by environmental factors. A power function decline typically arises when species composition across regions shows a stable structure due to niche-based influences such as environmental gradients [81,82]. This indicates that the understory vegetation maintained relative structural stability during that period.

In contrast, the exponential decline observed in 2020 suggests sudden, stochastic species between regions [83], pointing to ecological instability or disruption in the forest ecosystem. An exponential model is indicative of unpredictable, random turnover processes becoming dominant [81,82]. This shift reflects a transition from deterministic, niche-based filtering during succession to a community structure aligning more closely with neutral theory, driven by stochastic processes [60,84,85].

This shift corresponds temporally with the spread of dominant shrub layer species—Lindera obtusiloba, Rhododendron schlippenbachii, and Fraxinus sieboldiana—after 2017. The increased occupancy of these species likely affected resource availability in the herbaceous layer, supporting the notion of an indirect influence of overstory vegetation changes on herbaceous layer dynamics [25,86,87].

Moreover, the permanent survey plots exhibited a high degree of species composition homogeneity, with the same dominant species found even in environmentally diverse habitats. This widespread dominance contributes to the loss of rare species and regional specificity and ultimately drives the exponential decline in zeta diversity.

Despite potential contradictions among different theories of species composition change, real ecosystems are shaped by the complex interactions of numerous heterogeneous factors [83]. Therefore, detailed and continuous long-term monitoring will be essential for understanding these dynamics within the permanent survey plots studied.

Although species turnover in the herbaceous layer of the permanent survey plots increased over time, the zeta diversity graph showed a shallow decreasing curve in all survey years up to a zeta order of 24. However, beyond a zeta order of 24, the magnitude of decline was greatest in 2020. This suggests that, over the 7 years, species composition changed during each survey, but the extent of these changes was distributed throughout the plots. The 2020 turnover, which fits an exponential function, appears to have been caused by random or uniform environmental pressures, resulting in the same species undergoing turnover across all survey plots.

To summarize, the permanent survey plots exhibited high species turnover in 2020, which superficially indicates substantial shifts in species composition but in reality, it reflects the sudden dominance of a few species. Although evenness temporarily increased between 2017 and 2020, it showed an overall decreasing trend from 2014 to 2020. This pattern, along with the consistent number of shared species across many plots, resulted in a shallow decline in zeta diversity. However, when the number of shared habitats was high (zeta order ≥ 25), the zeta diversity declined sharply, indicating increased heterogeneity in species composition. These results suggest that the understory vegetation structure remains influenced by complex ecological factors and that a small number of dominant species are broadly distributed across the survey plots, suggesting an ongoing spatial homogenization [34].

From the high zeta order interval (≥24) onward, the zeta diversity remained constant (zeta plateau) across all survey years. This pattern serves as further evidence of spatial homogenization, resulting from the broad distribution of a few dominant species within the herbaceous layer. In other words, it indicates a decline in regionally specific species and the onset of homogenization within the ecosystem [82]. Notably, in 2020, the zeta plateau was relatively brief and repeated, which was interpreted as a vegetation structure simultaneously characterized by dominance of certain species and regional fluctuations in species composition [34]. Over time, this suggests a transitional structure in the understory vegetation, indicative of an unstable successional stage [34].

Species diversity in vegetation can change depending on which species become dominant, and the sequence in which broadly distributed species initially establish themselves [88]. When we examined changes in IV, the IVs for F. sieboldiana and L. obtusiloba increased more than those of other species over the 7-year period. These changes in the spatial dominance of species likely influenced the overall species composition of the herbaceous layer [89,90].

The primary contributors to species turnover were mainly woody plant saplings. These species, particularly at early growth stages, are highly sensitive to environmental stressors such as water scarcity and light competition, which supports the likelihood of stochastic survival or failure [91,92]. Rather than a scenario where species with similar growth traits coexist without competition under uniform conditions, the data suggest that species composition was rapidly restructured through stochastic losses. Therefore, our results suggest that both niche-based selection and stochastic processes operate at different times during the mid-successional stage.

In particular, Q. mongolica, a species that forms key communities within the study site, exhibited the highest rates of immigration and emigration. This reflects active population dynamics and highlights the species’ sensitivity to environmental fluctuations [93,94]. Species with high turnover rates are likely to respond acutely to climate change, changes in stand structure, and other external disturbances. As such, these species could serve as important indicators for climate change monitoring [95].

In most temperate forests, species diversity is primarily influenced by the herbaceous and shrub layers rather than the tree layer [3]. Accordingly, closely examining understory vegetation dynamics is essential, as they play a critical role in maintaining biodiversity. The herbaceous layer is particularly sensitive to environmental changes [67] and requires narrower habitat conditions than shrubs. These characteristics mean that herbaceous species diversity is more susceptible to disturbance [88], underscoring the need for ongoing monitoring to better understand and document these ecological changes.

Although our study focused on a single research forest in South Korea, similar trends of understory homogenization and increasing dominance by a few woody species have been observed in temperate forests worldwide. For instance, a 14-year study in eastern Canada reported compositional shifts and a decline in understory diversity in hardwood forests, despite an overall increase in herbaceous cover [96]. Long-term surveys across European temperate forests revealed that deterministic turnover driven by environmental gradients gradually transitioned into more homogenized understory communities [97]. In northern Wisconsin, resampling of remnant upland forests after approximately 50 years showed a reduction in understory diversity and compositional homogenization, even as total cover increased [98]. Additionally, studies in Canadian forests demonstrated that disturbance-based regeneration treatments led to persistent dominance by early-successional species, a decrease in beta diversity, and increasingly homogeneous understory communities over time [99]. These global observations are consistent with our findings, which suggest a temporal shift from deterministic (niche-based) assembly processes to more stochastic dynamics, as reflected in the zeta diversity pattern transitioning from a power-law to an exponential model.

5. Conclusions

In this study, we confirmed that the ecological processes in herbaceous layer vegetation involved a deterministic phase from 2014 to 2017, followed by stochastic changes in species composition in 2020. These results show that, in secondary forests of South Korea, frequent disturbances, spatial competition among species, and species turnover during forest gap recovery contribute to a complex successional process. Through correlation analysis with individual ecological factors over 7 years, we found that species composition in the shrub layer had an even greater influence on herbaceous layer diversity and turnover than previously recognized. Extending earlier research on the effects of overstory vegetation, we confirmed that the structurally complex shrub layer exerts a substantial short-term influence on the herbaceous layer.

Using long-term ecological monitoring data collected over 7 years from a temperate forest in South Korea, we analyzed short-term changes in herbaceous layer vegetation. While limited by the relatively brief monitoring period, the study provides baseline data for elucidating early ecosystem dynamics. Changes in herbaceous vegetation are not easily explained by changes in species composition alone but instead reflect complex interactions among ecological factors. Herbaceous species respond simultaneously to a variety of biotic and abiotic influences, including stand structure, light availability, soil conditions, climate, interspecific competition, animal disturbance, and microbial interactions—all of which are intricately connected. Therefore, vegetation dynamics in the herbaceous layer should be interpreted as the outcome of these interconnected processes rather than isolated factors.

To better understand these dynamics, integrative analyses that combine diverse ecological variables are essential. Comprehensive approaches should include environmental parameters such as soil moisture, nutrient status, and microclimatic conditions, as well as biotic components like animal and microbial community changes and their interactions with vegetation. Such multifaceted analyses are crucial for revealing not only structural changes in herbaceous vegetation but also shifts in ecosystem processes.

Future long-term monitoring efforts should aim to document both basic indices, such as species inventories and composition, and to quantitatively assess the complex relationships between biodiversity and ecosystem function. Increasing attention is being paid to long-term monitoring of factors such as climate change, land-use change, and species migration, all of which align closely with the goals of the Kunming–Montreal Global Biodiversity Framework, adopted in December 2022. This framework outlines 23 action targets for halting and reversing biodiversity loss by 2030, including the following: ① Target 1: spatial planning and ecosystem conservation; ② Target 3: expansion of protected areas and OECMs; and ③ Target 14: integration of biodiversity values into environmental assessments, policy, and planning across sectors.

Long-term ecological monitoring provides scientific data essential for shaping conservation strategies, assessing biodiversity trends, and identifying priority areas for protection. Furthermore, as the concept of Nature’s Contribution to People (NCP) gains prominence in ecosystem services research, long-term monitoring becomes increasingly valuable. It offers quantitative indicators of temporal changes in NCP and supports the assessment of shifts in both the supply and demand of ecosystem services.

This study highlights the importance of tracking herbaceous vegetation dynamics as a key to understanding broader ecosystem changes. It also emphasizes the need to sustain long-term ecological monitoring. Going forward, efforts should include not only standardized survey design but also the development of monitoring systems that expand both spatial and functional coverage. Convergent, interdisciplinary research will be essential to fully elucidate the complex interactions among ecological components.