Restoration of natural herbaceous vegetation and spatial variability of forest structure by gradual removal of Scots pine from former plantations

1.   Introduction

Over the last few centuries, coniferous plantations have replaced the natural forest ecosystems in large parts of Central Europe (Kint et al., 2006; Zerbe, 2002). The planting of Scots pine (Pinus sylvestris L.) in the lowlands of Europe followed extensive clearcutting. Fast growth and low habitat requirements, as well as a wide demand for pine wood in the timber and building industries, have been the main reasons for using this species for reforestation since the 18th century. At present, this Euro-Siberian species is the most widely distributed in the world, covering 28 million hectares (Durrant et al., 2016). Monospecific and even-aged Scots pine forests predominate the lowland landscapes of Poland, Germany, and other European countries, covering more than half of the total forested area (Aleksandrowicz-Trzcińska et al., 2017; Förster et al., 2021).

The natural occurrence of Scots pine is mainly limited to nutrient-poor sites within its natural range; however, due to the widespread plantations, it now dominates almost all site types, including the nutrient-rich sites of former broadleaved forests (Dobrowolska, 2006). The conversion of secondary forests dominated by Scots pine and pure stands of this species has become one of the main economic and ecological goals of forestry in Europe over the last few decades (Kint, 2005; Pach et al., 2018). The conversion of pure pine stands into mixed and broadleaved forests increases ecosystem stability and resistance to biotic and abiotic threats (Jactel et al., 2017). Furthermore, converting pure stands to mixed stands usually increases forest productivity, and the resulting income for forest managers (Bielak et al., 2014; Pretzsch et al., 2015). However, the conversion of pure and even-aged forest stands is often motivated by their influence on biodiversity, as spatially diverse, multilayered, and multispecies forests support higher levels of biodiversity than their simplified counterparts (Felton et al., 2010).

Management activities aimed at improving the naturalness and ecological quality of forest ecosystems may involve several strategies. Operations that provide space for a new generation of trees and promote a new cohort support the development of uneven-aged and multilayered forest stands (Kint et al., 2009; O'Hara, 2001). Regular spatial patterns in the distribution of trees that are typical of forest plantations and result from the regular spacing of planted trees can be transformed into random or clumped patterns via systematic thinning or aggregated retention harvesting (Crecente-Campo et al., 2009; Franklin et al., 2018; Nuutinen et al., 2021). In Scots pine monocultures planted on rich substrates, where temperate broadleaved forests have occurred in the past, the gradual removal of Scots pine trees and the promotion of natural regeneration of deciduous tree species accelerated the recovery of species composition typical of natural forests (Budde et al., 2011). Numerous studies have confirmed the positive effects of stand conversion practices in coniferous monocultures on various components and indicators of biodiversity, including herbaceous vegetation of the forest floor (Atkinson et al., 2015; Lust et al., 1998; Zerbe, 2002), tree diversity (Kint et al., 2009; Zerbe and Kreyer, 2007), lichens and beetles (Laarmann et al., 2013), mosses (Maciejewski and Zubel, 2009), forest stand structure (Crecente-Campo et al., 2009; Lust et al., 1998), and the number of large habitat trees (Vrska et al., 2017).

Forest stand conversion is typically undertaken in managed forests where timber production is the most important function. However, the experience of stand conversion gained in production forests may also be useful in protected areas where biodiversity conservation is a priority. In this study, we present the results of research conducted in Roztocze National Park (SE Poland) based on paired plots (also referred to as twin plots). In addition to natural forests of outstanding biological value, approximately one-third of the total forest area in the park is covered by former Scots pine plantations, which constitute a legacy of former forest management, and were included within the borders of the national park during its establishment and enlargements. These former pine plantations undergo natural succession processes and reveal trajectories towards natural ecosystems (Maciejewski, 2011; Maciejewski and Zubel, 2009). However, this process is very slow, and the national park services strive to accelerate the naturalization of former plantations by gradually removing pine trees. This study aimed to test whether these methods were effective in regenerating natural herbaceous vegetation and shifting the spatial distribution of trees and stand structures to more natural patterns. Specifically, we asked the following questions.

1) Does the removal of Scots pine trees support the recovery of natural herbaceous vegetation?

2) Does gradual removal of planted trees result in more natural spatial patterns of trees?

3) Does forest conversion support the recovery of natural features in forest structure?

We expected that Scots pine thinning resulted in an increased abundance of habitat-specific species in the forest floor, lower regularity of spatial tree distribution, and higher variability of structural indices.

2.   Methods

2.1   Study area

The study was carried out in the Roztocze National Park (RPN), located in SE Poland (Fig. 1). It was established in 1974 to protect natural forest resources that were previously protected in several nature reserves scattered across the Central Roztocze Highlands. The area of the national park is 84.82 ​km². The climate of Central Roztocze is classified as warm-summer humid continental, according to the Köppen-Geiger classification (Peel et al., 2007). The average duration of the growing season (days with a daily temperature above 5 ​℃) is approximately 210 days a year. The mean annual temperature is 7.2 ℃–7.4 ​℃ and the mean annual precipitation ranges between 650 and 750 ​mm (Kaszewski, 2008). The topography of the Roztocze region is characterized by a range of hills that run in the NW-SE direction. The hills are composed of silicate-carbonate marine deposits from the Upper Cretaceous period, mainly opokas and gaises. The terrain is complicated by numerous valleys filled with post-glacial deposits, including fluvial and aeolian sands. The differences between the tops of the hills and valley bottoms reach 100 ​m (Maruszczak, 1998). Of the 14 soil types identified within the national park, rendzinas, podzols, and rusty soils predominate (Koba and Miśta, 2015).

Figure  1.  Top-left panel (a): Localization of Roztocze National Park (grey circle) in Central-European context. Main panel (b): Localities of study plot pairs in Roztocze National Park (black rectangles).

DownLoad: Full-Size Img PowerPoint

The forests are comprised of 32 native tree species (80% of all native tree species that grow in Poland). Some of these species are characteristic of the mountains, e.g., European beech (Fagus sylvatica L.), silver fir (Abies alba Mill.) and Norway spruce (Picea abies (L.) H.Karst), whereas the others are typical of lowlands, such as Scots pine (Pinus sylvestris L.), European hornbeam (Carpinus betulus L.), small-leaved and large-leaved lindens (Tilia cordata Mill., and T. platyphyllos Scop.), together creating a mosaic of coexisting forest stands (Izdebski et al., 1992; Szwagrzyk et al., 2012). Submountain rich beech forests and oak-hornbeam forests are the most common types of broadleaved woodlands, whereas secondary Scots pine forests and natural silver fir forests are the most common representatives of coniferous woodlands.

2.2   Study objects

We sampled pairs of 0.5 ​ha (50 ​m ​× ​100 ​m) study plots in three locations in Roztocze National Park (Fig. 1). Each pair consisted of a plot subjected to stand conversion activities over the last few decades, and a control plot excluded from management for nearly half a century. Control plots (nos. 5, 6, and 8) were established between 1971 and 1973 as part of a project on the productivity of native forest communities in the Roztocze Highlands (Izdebski et al., 1976, 1977). Since then, these plots have been excluded from any kind of human intervention and resampled several times to track spontaneous changes in vegetation and forest dynamics (Maciejewski, 2011; Pielech et al., 2022). The plots subjected to stand conversion (nos. 5A, 6A, and 8A, hereinafter referred to as converted plots) were established in 2007 as part of a project on the dynamics of native tree species in Roztocze National Park. These plots were placed next to the control plots under the same habitat conditions to form pairs for future comparison. They have been subject to low-intensity management for several decades, and these practices mainly included the removal of Scots pine aimed at the conversion to natural forests.

2.3   Management history

In 2021, we summarized all available information regarding recent and historical management activities, including management plans (since 1934), forestry maps (since 1946), documents related to the implementation of management plans (before the establishment of the RPN), and protection plans. Based on these documents, as well as the knowledge of the national park staff, we have detailed knowledge of the management history in our study plots. Owing to previous measurements of permanent plots, we have precise data on the intensity of stand conversion activities. Table 1 summarizes the information on our study objects.

Table  1.  Characteristics of permanent plots used in this study. Intensity of Scots pine removal refers to total number (N) and total basal area (BA, m²) of Scots pine trees removed since 2007.

Plot ID	Type	Plantation established	Plot established	Soil type	Potential vegetation type	Treatment intensity (N/BA)
5	Control	1903	1971	Calcaric Leptic Cambisol	Rich beech forests	–
5A	Managed	1903	2007	Calcaric Leptic Cambisol	Rich beech forests	50/6.63
6	Control	1917	1971	Albic Podzol	Mixed forest	–
6A	Managed	1917	2007	Albic Podzol	Mixed forest	23/1.52
8	Control	1912	1973	Albic Podzol	Acidophilous beech/oak forest	–
8A	Managed	1912	2007	Albic Podzol	Acidophilous beech/oak forest	46/4.07

 | Show Table

DownLoad: CSV

The management treatments used for restoration purposes in plots 5A, 6A, and 8A stemmed from common forestry management practices. In the first step, the individuals of the target tree species were selected. The target species were defined as those that were typical components of restored communities. In the second step, Scots pine trees that negatively affected the growth and regeneration of the target species were gradually removed. These two steps resemble the selection thinning approach commonly used in Central European forestry. However, as opposed to selection thinning, which aims to increase the production capacity and timber quality, the treatment applied in RPN is aimed at restoring the natural species composition and structure of forest communities. Our treatment plots differed in terms of the intensity of cutting the Scots pine trees (Table 1).

2.4   Vegetation sampling and measurements

Field work was conducted in the summer of 2021 and organized into two campaigns dedicated to (1) sampling herbaceous vegetation of the forest floor and (2) measuring the structure of the forest stand. In addition, for broadleaved forest plots (5-5A), we conducted additional sampling in the spring of 2021 to ensure that we recorded all spring geophytes typical of this forest ecosystem.

2.4.1   Herbaceous vegetation

To analyze changes in the species composition and the abundance of plant species of the forest floor, we conducted floristic surveys recording all vascular plants within 0.5 ​m² circular plots (364 circular sample plots in total; hereinafter, to distinguish between large rectangular 0.5 ​ha study plots and small circular 0.5 ​m² plots, we refer to the latter as to ‘circular plots’) and estimated percentage cover of each identified species. Plant cover was estimated visually with 1% accuracy for low cover values and to the nearest 5% for cover values greater than 20%. For study plots designated as pairs 5-5A, we sampled 70 circular plots each, whereas for pairs 6-6A and 8-8A, we sampled 56 circular plots each. Differences in the number of circular plots sampled stem from the original study design proposed in the early 1970s, when permanent plots were established (Izdebski et al., 1976, 1977). The authors of this methodology proposed 70 circular plots in deciduous forest sites and 56 circular plots in coniferous and mixed forest sites because of the spatial variability of vegetation on the forest floors in these habitats. To maintain consistency in long-term sampling and allow for the analysis of vegetation trajectories, we followed the initial sampling scheme (Fig. 2).

Figure  2.  Distribution of 0.5 ​m² circular plots (red dots) established to sample herbaceous vegetation within 0.5 ​ha study plot pairs (5-5A, 6-6A, and 8-8A).

DownLoad: Full-Size Img PowerPoint

2.4.2   Forest structure measurements

Forest structure measurements were conducted using the Field-Map technology (Institute of Forest Ecosystem Research – IFER, Jílové u Prahy, Czech Republic, https://www.ifer.cz/) to map the spatial distribution of trees and measure the structural attributes of a forest stand. We recorded the position, species identity, diameter at breast height (DBH), and viability of each tree with a DBH equal to or greater than 7 ​cm. DBH was calculated as the mean of two perpendicular measurements of the tree trunk. In addition, we used dendrometric tape to measure the perimeter of the tree trunks. Perimeter measurements were used to control for possible errors because we searched our dataset for major discrepancies between the perimeter calculated from the DBH and that measured with a tape.

2.5   Data analyses

2.5.1   Herbaceous vegetation

For each circular sample plot, we analyzed the richness and vascular plant diversity of forest floor species by computing several indices, including species richness (number of species), Shannon and Simpson diversity indices, and the Shannon evenness index, i.e. Shannon diversity index divided by the logarithm of the species number. Subsequently, we compared the average values of these indices between the managed and control plots, assessing the statistical significance of the observed differences using the Wilcoxon test (since the data did not follow a normal distribution). To explore differences in the species composition of herbaceous vegetation between the managed and control plots, we performed Non-metric Multidimensional Scaling (NMDS) ordination. In addition, we tested these differences for statistical significance with permutational analysis of variance (PERMANOVA) using the vegan::adonis2 function with 999 permutations (Oksanen et al., 2019).

For detailed comparisons of herbaceous plant species in the community, we only considered plant species that reached a minimum frequency of 10% across all circular sample plots within particular plot pairs. In this way, we prioritized species with a substantial presence within the ecosystem studied, ensuring that our analyses would encompass the most ecologically relevant species in the forest floor and disregard rare species. We compared plant species coverage between plots subjected to the treatment and that of the control group. Given the non-normal distribution of the data, we used the Wilcoxon test.

2.5.2   Spatial patterns and structural indices

Ripley's L function was used to examine the spatial arrangement of the measured trees. We were interested in whether the regular distributions of trees typical of forest plantations changed to random or clumped patterns as a result of stand conversion activities. Ripley's L function is a well-established method that allows for the analysis of spatial patterns in tree distribution; however, it also considers a range of spatial scales (Ripley, 1976, 1977). Often, trees show a regular pattern under small spatial scales (e.g., a few meters), whereas under larger spatial scales, they are distributed in a random or clustered manner (Dixon, 2002). For each plot, we calculated the transformed L-function for Ripley's K function using the spatstat::Lest function (Baddeley et al., 2015). The Besag's transformation of the original K function to its L derivative linearize the expected values and stabilize the variance (Ward and Ferrandino, 1999). Ripley's function counts the number of trees within a circular plot of radius centered on each tree in the plot. The calculations were repeated with increasing radius. In addition, we ran Monte Carlo simulations with 999 iterations to calculate the 95% simulation envelope and assess the significance of the dispersion or clustering of trees in the plots.

To examine the effect of Scots pine removal on the structural characteristics of the forest, we calculated a set of structural indices, including the basal area (BA), density of living trees, mean and maximum DBH, and quadratic mean diameter (QMD). We used these metrics to conduct a pairwise comparison between treatments, specifically comparing plots that underwent conversion with those of the control group.

All statistical analyses were performed using R version 4.1.3 (R Core Team, 2022) and additional packages for specific data analysis as well as for data wrangling and graphical presentation of the outcomes, including spatstat (Baddeley et al., 2015), maptools (Bivand and Lewin-Koh, 2021), ggplot2 (Wickham, 2016), ggpurb (Kassambara, 2020), dplyr (Wickham et al., 2020a), tidyr (Wickham and Henry, 2020b), and vegan (Oksanen et al., 2019). Basic spatial analyses and maps presenting the distribution of trees in the study plots were performed in ArcGIS Pro 2.9.2 (ESRI Inc., Redlands, https://www.esri.com/en-us/arcgis/products/arcgis-pro/).

3.   Results

3.1   Diversity, composition and abundance of herbaceous species

Species composition varied among the plot pairs studied, primarily because each pair represented a different forest type (Table 1). In total, we recorded 60 herbaceous species in the six study plots. The species composition of forest floor vegetation differed between converted and control plots, and these differences were the strongest for pair 5-5A. In the NMDS ordination space, the convex hulls for both groups overlapped only partially, suggesting substantial segregation. These differences were less profound in plots 6-6A and 8-8A (Fig. 3). However, PERMANOVA confirmed that only the plot pairs 5-5A and 6-6A differed significantly. In contrast, the differences in species composition in plot pair 8-8A were statistically insignificant (Table 2). Detailed comparisons were performed for 20 frequent species with a minimum frequency of 10% in all circular sample plots for particular plot pairs. We observed significant differences in species abundances in two out of the three pairs of examined plots, i.e. 5-5A and 8-8A.

Figure  3.  Non-metric multidimensional scaling (NMDS) for three pairs of plots compared. Left-hand plots present distribution of converted and control plots in ordinal spaces along first and second NMDS axes. Background polygons (convex hulls) represent portions of ordinal space occupied by each group. Right-hand panels show distribution of plant species in ordinal spaces with position of plots in the background.

DownLoad: Full-Size Img PowerPoint

Table  2.  Results of permutational analysis of variance (PERMANOVA) testing differences in species composition between converted and control plots.

Plot pair	df	Sum of squares	R2	F	p
5-5A	1	5.206	0.174	28.872	0.001
6-6A	1	0.603	0.036	3.772	0.024
8-8A	1	0.426	0.019	1.817	0.103

 | Show Table

DownLoad: CSV

In the case of pair 5-5A, the treatment plot had a higher abundance of species naturally associated with mesic broadleaved forests, i.e., typical to this specific forest habitat (Table 1; Fig. 4). These species included Athyrium filix-femina (L.) Roth, Circaea alpina L., Galium odoratum Scop., Impatiens noli-tangere L., Veronica montana L., and Viola reichenbachiana Jord. ex Boreau. Light-demanding species, such as Geranium robertianum L., Rubus hirtus Waldst & Kit. agg., R. idaeus L., and Urtica dioica L, also increased in presence in plots subjected to cutting. Species that prefer acidic or moderately acidic conditions, such as Oxalis acetosella L. and Maianthemum bifolium (L.) F. W. Schmidt exhibited lower abundance in the treatment plot than in the control plot.

Figure  4.  Comparison of herbaceous vegetation between plots 5 and 5A. Vertical line represents median, while points indicate outliers. Statistical significance: ns, p ​> ​0.05 (not significant); **, p ​≤ ​0.01; ***p ​≤ ​0.001; ****p ​≤ ​0.0001.

DownLoad: Full-Size Img PowerPoint

For the pair of plots 6-6A, many circular plots had no or a very small number of species; therefore, only three species reached the 10% threshold (see subsection 2.5.1. Herbaceous vegetation), and were included in further analyses. No significant differences in the abundance were observed (Fig. 5).

Figure  5.  Comparison of herbaceous vegetation between plots 6 and 6A. Vertical line represents median, while the points indicate outliers. Abbreviation ‘ns’ denotes results lacking statistical significance.

DownLoad: Full-Size Img PowerPoint

On the contrary, for pair 8-8A, treatment plots had a significantly higher abundances of acidophilic species characteristic of this specific habitat type (Table 1; Fig. 6), and these species included Luzula pilosa (L.) Willd., Melampyrum pratense L., and Trientalis europaea L.

Figure  6.  Comparison of herbaceous vegetation between plots 8 and 8A. Vertical line represents median, while the points indicate outliers. Statistical significance: ns, p ​> ​0.05 (not significant); **, p ​≤ ​0.01; ***p ​≤ ​0.001; ****p ​≤ ​0.0001.

DownLoad: Full-Size Img PowerPoint

When compared to the control, plots subjected to the removal of pine trees had increased species richness and diversity of the forest floor vegetation in the studied communities in two out of the three study plots (Fig. 7). Notably, the species richness was significantly higher in the converted plots 5A and 8A than in their corresponding control plots. Moreover, the values of the Shannon and Simpson diversity indices, along with the evenness index, were also significantly higher. In contrast, such positive changes were not observed in plot 6A, which was characterized by the lowest cutting intensity.

Figure  7.  Comparison of species richness and diversity indices between plots subjected to stand conversion and control plots. Statistical significance: ns, p ​> ​0.05 (not significant); * p ​≤ ​0.05; ** p ​≤ ​0.01; *** p ​≤ ​0.001; **** p ​≤ ​0.0001.

DownLoad: Full-Size Img PowerPoint

3.2   Structural forest characteristics

In total, 1,977 trees of nine species (excluding individuals with DBH<7 ​cm) were recorded in the six study plots. Besides Scots pine, the dominance of which was an effect of the former plantation, other dominant tree species were European beech and European hornbeam in plots 5-5A; silver fir, Norway spruce and European beech in plots 6-6A; and European beech and sessile oak (Quercus petraea (Matt.) Liebl.) in plots 8-8A (Fig. 8).

Figure  8.  Pairs of study plots presenting tree distributions.

DownLoad: Full-Size Img PowerPoint

3.2.1   Spatial distribution of trees

Ripley's L function revealed differences between the control plots and the plots subjected to stand conversion; however, there was no common pattern, and each pair showed different changes (Fig. 9). A comparison of plots 6 and 6A revealed that conversion resulted in a remarkable increase in the L-function values, showing a tendency from a random to clustered tree distribution pattern. However, the deviation from randomness is not statistically significant. When comparing plots 8 and 8A, the shape of the L function was similar for both plots, except for the distances between 5 and 15 ​m. Our analyses showed an increase in the L function values; however, in this case, the deviation from randomness was not significant. Comparison of plots 5-5A showed a significant regular pattern at small scales (up to 5 ​m), but a random pattern in converted stands.

Figure  9.  Comparison of Ripley's L function between plots subjected to Scots pine removal and the control plots. The first and second columns show the shapes of the L functions and simulation envelopes, allowing for assessments of the significance of deviation from randomness. The third column presents the overlapped L functions for both converted and control plots.

DownLoad: Full-Size Img PowerPoint

3.2.2   Structural indices

We detected differences in the structural characteristics between the converted and control plots. The total basal area and the density of live trees were lower in the converted plots because of the selective removal of the pine trees (Table 3). In contrast, the mean DBH and QMD were higher in the converted stands, and the maximum diameters were higher in the converted plots in two of the analyzed pairs of plots. The vigorous growth of European beech was the main factor responsible for the increase in the mean diameter in plot 5A when compared to the control plot 5. In plot 8A, the high contribution of oaks, both sessile and pedunculate (Quercus robur L.), was a driving factor for the higher mean DBH when compared to the control plot. The differences in mean diameters between plots 6 and 6A for the analyzed tree species were subtle, apparently because of the low intensity of cutting.

Table  3.  Comparison of structural characteristics between plots subjected to Scots pine removal (5A, 6A, 8A) and the control plots (5, 6, 8).

	Basal area (BA, m2·ha−1)						Live tree density (N·ha−1)						Mean DBH (cm)						Maximum DBH (cm)						Quadratic mean diameter (QMD)
	Study plots						Study plots						Study plots						Study plots						Study plots
	5	5A	6	6A	8	8A	5	5A	6	6A	8	8A	5	5A	6	6A	8	8A	5	5A	6	6A	8	8A	5	5A	6	6A	8	8A
All species	58.4	35.5	45.9	42.9	51.3	34.4	442	188	1206	924	728	466	38.9	46.9	18.6	20.6	26.9	27.5	103.1	90.4	51.8	53.0	59.6	61.7	41.0	49.0	22.0	24.3	30.0	30.7
Scots pine (Pinus sylvestris)	35.7	12.2	35.6	35.2	44.3	29.2	232	76	348	322	386	242	43.5	44.9	35.6	36.9	37.5	38.6	65.1	59.1	51.8	53.0	59.6	61.7	44.3	45.3	36.1	37.3	38.2	39.2
Norway spruce (Picea abies)	–	–	1.4	0.4	0.4	<0.1	–	–	120	34	42	2	–	–	11.3	11.1	10.8	13.5	–	–	35.8	19.2	29.2	13.5	–	–	12.2	11.5	11.7	13.5
Silver fir (Abies alba)	4.6	1.6	6.8	5.6	0.2	<0.1	38	16	618	448	8	2	35.4	31.0	11.3	12.1	18.3	16.4	68.1	75.1	28.1	25.0	24.5	16.4	39.3	35.6	11.9	12.7	19.2	16.4
European beech (Fagus sylvatica)	9.1	20.7	2.1	1.7	5.5	3.2	42	86	116	120	238	164	47.1	53.1	13.5	11.5	15.9	14.5	103.1	90.4	36.0	44.1	34.8	35.6	52.5	55.4	15.1	13.3	17.2	15.7
European hornbeam (Carpinus betulus)	8.9	0.9	–	–	–	–	130	10	–	–	–	–	28.9	34.0	–	–	–	–	49.5	40.9	–	–	–	–	29.6	34.3	–	–	–	–
Sessile oak (Quercus petraea)	–	–	–	–	0.7	1.1	–	–	4.0	–	–	8	–	–	8.0	–	–	33.9	–	–	8.6	–	–	44.3	–	–	8.0	–	–	35.6
Pedunculate oak (Quercus robur)	–	–	–	–	–	–	–	–	–	–	46	44	–	–	–	–	12.7	15.9	–	–	–	–	26.1	36.0	–	–	–	–	13.6	18.1
Northern red oak (Quercus rubra)	–	–	–	–	0.1	–	–	–	–	–	6	–	–	–	–	–	12.7	–	–	–	–	–	15.7	–	–	–	–	–	13.2	–
Silver birch (Betula pendula)	–	–	–	–	0.1	<0.1	–	–	–	–	2	4	–	–	–	–	26.0	9.5	–	–	–	–	26.0	10.9	–	–	–	–	26.0	9.6

 | Show Table

DownLoad: CSV

4.   Discussion

4.1   Effect of Scots pine removal on herbaceous vegetation

The results of our study shed light on the intricate relationship between forest conversion through the removal of planted Scots pine trees and the naturalization of herbaceous vegetation, highlighting the significant influence of initial forest type on these dynamics. Our findings underscore the positive impact of silvicultural practices on beech forests, which are renowned for their diverse ecosystem services (Augustynczik and Yousefpour, 2021). Beech forests play a pivotal role in Europe's forest habitats, and the preservation and rejuvenation of their tree communities should be a central focus of forest management approaches aimed at balancing biodiversity conservation and timber production.

Conversion practices had a beneficial effect on herbaceous vegetation in the studied forest communities. This was especially evident when comparing the species composition of the undergrowth in the beech forests, where both richness and diversity were significantly higher in the converted plots than in the control plots. It is important to note that our analyses are quantitative in nature, comparing the overall richness and diversity indices. Numerous studies have shown that thinning may result in a higher plant species richness (Götmark et al., 2005; Widenfalk and Weslien, 2009; Yang et al., 2023). However, an increase in the number of species and their diversity may not always be desirable when considering the naturalization of economically managed forest ecosystems, and such an increase may be due to the greater presence of ruderal or invasive species in response to canopy openings during conversion (Heinrichs and Schmidt, 2009). Therefore, to obtain a deeper understanding of the characteristics of the detected changes, we also analyzed differences in species and interpreted these changes in the context of consistency with habitat types. From this perspective, our analyses indicate that the quantitative changes in the studied areas are closely linked to the regeneration of typical woodland-specialist species, characteristic of habitats naturally occurring before their transformation into Scots pine plantations. In a habitat typical of beech forests, plot subjected to the gradual removal of planted pine trees (plot 5A) was characterized by the increased abundances of herbaceous plants typical of mesic broadleaved forests. Analogously, in the habitat of acidophilous broadleaved forests, increased abundances of species typical of this specific habitat was found in the converted plot (8A). A notable increase in the abundance of native species closely related to specific habitats within plots subjected to conversion demonstrates the potential for successful conversion (Zerbe, 2002). The observed changes in herbaceous vegetation offer promising implications for ecosystem services, with potential benefits, such as enhanced soil stability (Löbmann et al., 2020) and improved nutrient cycling (Hobbie, 1992). These transformations facilitate the rapid decomposition of aging plant material within the herbaceous layer. Increased organic matter and root turnover from higher natural herbaceous vegetation cover introduce various organic inputs, including leaf litter and root exudates, which stimulate microbial diversity and activity, which are key factors in decomposition rates (Keiser et al., 2013; Elliott et al., 2015). These processes, combined with rapid turnover of herbaceous vegetation, supply essential nutrients that support both naturalization and regeneration of target tree species (Rawlik et al., 2021). Furthermore, the resulting nutrient inputs positively influence tree foliage production, causing increased leaf mass among overstory species (Albaugh et al., 2012; Norris et al., 2013; Elliott et al., 2015). Understanding these broader ecosystem functions not only justifies the adoption of forest conversion practices, but also presents opportunities for their refinement, tailored to various ecological objectives.

4.2   Effect of Scots pine removal on spatial distribution of trees

This study showed that forest stand conversion activities that rely on the selective removal of planted trees and support the regeneration of target species can affect the spatial distribution of trees, accelerate the naturalization process of former plantations, and restore natural forest ecosystems. Positive changes were observed in all three paired plots. The results of Ripley's L function analyses revealed changes in the spatial patterns in an expected direction, that is, from regular to random and from random to clustered. However, these changes were more subtle and weaker than expected. We consider this to be mainly due to the methods and intensity of stand conversion operations applied, but also due to the relatively short period of restoration management.

In our study in Roztocze National Park, Scots pine removal was used to convert the structure of the stand and accelerate the naturalization of former plantations, which was very similar to selection thinning, a method widely used in typically managed forests. In this type of management activity, trees are removed more or less regularly to provide the remaining planted trees with an equal amount of living space and resources. Selection thinning may, therefore, preserve regularity in tree distribution and fail as a method aimed at increasing spatial heterogeneity in forests. Nuutinen et al. (2021) used the Clark-Evans aggregation index to show that this kind of thinning cannot alter the regularity of the tree distribution in forest plantations. Similarly, other authors also confirmed that typical management practices can maintain, or even increase, regularity in the spatial distribution of trees (Boncina et al., 2007; Kint et al., 2003; Li et al., 2021; Puettmann et al., 2014). Previous studies have shown that post-thinning spatial arrangement of trees may depend on which component of the forest ecosystem was removed. Removal of dominant and co-dominant species led to more clustered spatial patterns, whereas removal of intermediate and suppressed individuals resulted in more regular arrangements (Kuehne et al., 2018). In addition, spatial patterns may also depend on the thinning intensity. Moderate thinning intensities may be insufficient to change spatial patterns over time because of the low mortality rates and low levels of ingrowth. In contrast, heavy thinning resulted in clustered or random patterns (Acquah et al., 2023). Therefore, if the restoration of natural spatial patterns is a priority, decisions on thinning intensities and removed components of forest stands should be taken in advance.

4.3   Structural differences between plots subjected to stand conversion and those left for natural succession

Forest stand conversion based on pine tree removal resulted in several positive changes in the structural characteristics of the study plots. With lower tree density, the mean DBH and QMD were higher in the converted plots than in the control plots. Tree species that comprise natural forest communities in the studied sites were responsible for this change, including European beech in plot 5A and pedunculate and sessile oaks in plot 8A. This suggests that the applied selective cutting of Scots pine trees supported forest stand naturalization by improving conditions for tree species consistent with habitat types. However, in plot 6A, the mean tree diameters (both DBH and QMD) were only slightly higher than those in the control plot. This could be explained by a very low intensity of Scots pine removal in this plot when compared to the 5A and 8A plots (BA of the removed pine trees was 1.52 ​m² vs. 6.63 ​m² and 4.07 ​m², respectively; see Table 1). Such a low intensity of cutting did not improve the conditions for growth and competition for other tree species. Furthermore, when comparing BA, plots 6 and 6A did not differ in terms of the BA of living pine trees (35.6 ​m² vs. 35.2 ​m²), even though 23 pine trees were removed in 6A (see Table 1). In plot 6A, the intensity of cutting was similar to that of natural mortality in control plot 6. The low intensity of cutting was probably compensated for by the increased growth rate in the converted plot, and therefore, the Scots pine contribution expressed by BA remained similar. Previous studies have shown that lite thinning (<20% of the trees removed) promoted forest regeneration, heavy thinning (>35% of the trees) facilitated forest growth, and only thinning of moderate intensity (20%–35% of the trees) created a stable and heterogeneous spatial structure (Wang et al., 2024).

4.4   Implications for forest stand conversions for nature conservation purposes

Forest management activities in protected areas are sometimes controversial because non-intervention or “hands-off” approaches are often favored by the public (Landres, 2010). Such conservation strategies are the most reasonable options for primeval and close-to-primeval forests, where natural processes have uninterrupted biodiversity for centuries (Jaroszewicz et al., 2019). However, in seminatural ecosystems developing under intense human pressure and a lack of natural disturbance regimes, management activities may have beneficial effects on biodiversity resources in protected areas (Sebek et al., 2015).

In this study, we found several positive effects of forest stand conversion on the naturalization of former pine plantations. The changes that occurred on the forest floor were highly favorable, resulting in increased species richness and diversity, as well as the successful establishment of species naturally found in broadleaved forests. However, the differences in the spatial patterns of tree distribution and, to some extent, in structural indices, between converted and control plots were lower than expected. We explain these subtle differences by the fact that operations aimed at stand conversion were mainly based on conventional methods used in typically managed forests. As mentioned above, the removal of planted trees using methods following selection thinning can preserve or even increase the regular distribution of canopy trees. The aim of this method is to selectively cut trees to create equal spaces for the remaining trees. This treatment conserves the spatial homogeneity of both old and young trees. This is a good strategy for production forests; however, if nature conservation is a priority, other methods that improve the spatial heterogeneity of tree distributions should be applied. Several authors have shown that systematic thinning is more effective at breaking tree regularity than selective cutting (Nuutinen et al., 2021). Generally, stand conversion for conservation purposes should be based on increasing spatial heterogeneity by forming a mosaic of patches with dense canopies, patches with moderate densities, and gaps of various sizes (Duflot et al., 2022). In addition, all management strategies based on emulating natural disturbances, such as those based on the gap dynamic model, could accelerate forest naturalization (Angelstam, 1998; Aszalós et al., 2022). For small-scale interventions, simulations of tree cutting and analysis of potential results using geostatistical tools could also be beneficial.

In addition to the appropriate method, the intensity of planned operations is crucial when planning forest stand conversion for conservation purposes. Our study showed that the low-intensity Scots pine removal applied in Roztocze National Park was not always sufficient to change the structure of the site. In plot 6A, where the cutting intensity was the lowest, the contribution of Scots pine expressed as BA was similar to that of the control plot. Therefore, to successfully accelerate the naturalization of former plantations, a higher intensity of cutting activities and its spatial variability depending on existing natural regeneration patches is strongly recommended. Possible public concerns regarding intensive tree cutting in national parks could be mitigated by information campaigns that explain the purpose and character of active protection.

5.   Conclusions

The selective cutting of planted trees can support the naturalization of former Scots pine plantations in protected areas. However, the specific ecological requirements of the target species and habitats, as well as the selection of the proper method and intensity of the canopy tree cuttings, are of vital importance for achieving the desired conservation outcomes. Strategies based on typical management practices, such as selection thinning, although supporting the naturalization of herbaceous vegetation, are not always the best approach, as they may preserve or even increase the regular distribution of trees. Therefore, other methods that increase spatial heterogeneity could be potentially a better choice, including systematic thinning or the simulation of cutting scenarios and the resulting spatial patterns using geostatistical tools. In addition, low-intensity cutting activities may not be sufficient to support natural regeneration or change the size structure. Unfortunately, in our study system, we were unable to test the effects of a broad range of management strategies and thinning intensities, which could be tested experimentally in further long-term studies.

CRediT authorship contribution statement

Remigiusz Pielech: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Adrian Wysocki: Writing – review & editing, Writing – original draft, Visualization, Validation, Investigation, Formal analysis, Data curation. Kacper Foremnik: Writing – review & editing, Writing – original draft, Visualization, Validation, Investigation, Formal analysis, Data curation. Marek Malicki: Investigation. Bartłomiej Surmacz: Writing – review & editing, Writing – original draft, Visualization, Validation, Investigation, Formal analysis, Data curation. Jerzy Szwagrzyk: Methodology, Funding acquisition, Conceptualization. Zbigniew Maciejewski: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization.

Data availability

Data are available on request from the authors.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Remigiusz Pielech reports financial support was provided by Polish State Forests. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.