Next Article in Journal
Increase in Grain Production Potential of China Under 2030 Well-Facilitated Farmland Construction Goal
Previous Article in Journal
A Model for Future Development Scenario Planning to Address Population Change and Sea Level Rise
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Land
Volume 14
Issue 8
10.3390/land14081537
land-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

The Relationship of Forest Fragmentation to Scots Pine Forest Mortality

by
Debebe Dana Feleha
1,2,
Pawel Netzel
1,* and
Jakub Talaga
1
1
Faculty of Forestry, Department of Forest Resources Management, University of Agriculture in Krakow, Aleja Mickiewicza 21, 31-120 Krakow, Poland
2
Department of Natural Resource Management, College of Agriculture, Wolaita Sodo University, Wolaita Sodo P.O. Box 138, Ethiopia
*
Author to whom correspondence should be addressed.
Land 2025, 14(8), 1537; https://doi.org/10.3390/land14081537
Submission received: 31 May 2025 / Revised: 23 July 2025 / Accepted: 25 July 2025 / Published: 27 July 2025
(This article belongs to the Section Land Planning and Landscape Architecture)
Browse Figures
Review Reports Versions Notes

Abstract

Forest mortality (FM) is influenced by several independent factors, including forest fragmentation (FF) at different spatial scales and multi-scales, site conditions, and stand characteristics. The aim of this study was to investigate the relationship and effect of FF at various spatial scales on the probability of Scots pine FM. The presented study also analyzed the relationship of the multi-scale fragmentation index effect on forest dieback. The relationship between multiple stressors emphasizes the distinct role of FF in influencing pine FM probability. Data on forest cover, deadwood volume of Scots pine forest, and environmental variables were obtained from the Forest Information System for Europe, the Polish National Forest Inventory, and existing databases, respectively. A generalized additive model approach was used to develop models. The results showed that, at small (50–600 m), large (800–3000 m), and multi spatial scales, the FF effect on Scots pine FM probabilities was statistically significant. There is a partial effect of multi-scale fragmentation on the probability of Scots pine FM, given a holistic view of the fragmentation effect that captures both small and large-scale effects. The study concludes that to calculate FF for a particular area, analyzing different scales and capturing multi-scale level fragmentation indices is crucial to studying the cumulative effect of fragmentation on the probability of Scots pine FM.

1. Introduction

Temperate forest ecosystems occupy 25% of the total forest cover of Earth’s surface [1]. They offer various ecosystem benefits at local and international levels, such as preserving biodiversity, storing carbon, and regulating water for cities, farming, and hydroelectricity [2]. However, human activities such as agricultural and urban expansion, infrastructure, and other developmental activities have caused significant alterations in temperate forests [3,4]. As a result, temperate forests are characterized by a mosaic of forests with agricultural land and are 1.5 times more fragmented than tropical forests [5]. Moreover, they are affected by long-term forest fragmentation (FF) effects—loss of biodiversity, loss of ecological services, and increased edge effects [6]—which caused forest mortality (FM) [7,8]. The present study has concentrated on the Scots pine, which is the most geographically widespread species of tree on Earth [9]. It is imperative to acknowledge the ecological and economic significance of this tree species in Europe, which is of paramount importance [10]. In Poland, Scots pine constitutes the predominant forest-forming tree species, occupying over 60% of the total forest area [11].
Forest mortality (FM) is the death of trees or woody vegetation within a forest ecosystem [12,13]. It is a natural process that occurs as a result of various factors, including biotic and a biotic factors [14,15]. Moreover, FM occurs when the tree’s production of carbohydrates through photosynthesis is exceeded by its respiration. As a result, the tree dries out and ceases to photosynthesize [16]. FM can occur at the individual tree level, affecting individual trees or small groups of trees, or at large scales, affecting entire stands of forests. It can have significant ecological, environmental, and socio-economic impacts influencing forest structure, composition, biodiversity, carbon cycling, and ecosystem services [17].
Forest fragmentation (FF) is the process of dividing large and continuous forests into smaller and more isolated patches [18,19] with a creating variety of edges and edge zones [19,20]. An edge is the boundary or interface between a forest fragment and another habitat type, while an edge zone is the transition zone between forested and non-forested areas within a given edge distance [21]. The formation of an edge zone begins when a new forest edge is created, which can happen due to deforestation, logging, or other forms of disturbance [22]. This leads to the exposure of the previously interior forest to new environmental conditions in the edge zone [22,23].
The edge zone is distinguished and affected by the following edge effects: (1) The zone exhibits higher temperatures due to minimum tree cover, a lower vegetation density, and a greater degree of direct sunlight exposure compared to the interior zone [24]. (2) A higher canopy gap and a change in environmental conditions in the edge zone lead to overheating, desiccation, and increased competition from light-demanding species, which can stress shade-tolerant trees and lower relative humidity [25,26,27]. (3) Drought stress in the edge zone is a consequence of a complex interaction between increased solar radiation exposure, altered microclimate, and competition for water resources. This stress can impact the structure, composition, and services provided by the forest [28]. (4) Wind speed, turbulence, and wind throw are high at the edge zone because the edge zone has reduced shelter. The sudden change in surface characteristics and the irregular shape and roughness of trees can weaken their structure, leading to breakage or uprooting during storms [29]. (5) This zone is particularly susceptible to the introduction of invasive species, pests, and disease due to the disturbances it experiences, including logging, road construction, fire, and clearing.
These disturbances can have a detrimental impact on the forest ecosystem, rendering it more susceptible to the invasion of invasive species and the colonization by pests and diseases [30]. As a result, this boundary area is highly affected by edge effects that cause the accelerated mortality of trees [31,32].
The NFI is a database designed to assess and monitor the condition of forest resources throughout the country [33]. It plays a key role in forest management, protection, and policy-making. The NFI’s main tasks are to collect data; ensure sustainable forest management, protect the environment; make data available to the public; researchers, and other interested parties; develop a systematic sampling design based on a grid pattern; and maintain an up-to-date forest inventory database.
A significant amount of research has been conducted on the effect of forest edges [34,35,36,37]. However, only a limited number of studies have examined the influence of edge effects on the probability of dieback of tree resources in fragmented temperate forests [38]. The majority of research has centered on evaluating the impact of the edge effect on both the structural and species diversity of the forest, primarily as a result of variations in microclimate, management techniques, and the composition of the tree community compared to the interior of the forest stand [34].
FF in Poland has been studied at different scales [39,40,41]. Research has demonstrated that FF has significantly impacted biodiversity and ecosystem functioning in Poland [42,43]. However, there are no studies that look at the connection between the degree of FF at different scales and the multi-scale effect on the probability of FM in Scots pine.
The Oleszyce Forest Inspectorate (OFI) is characterized by complex topography and a wide range of fragmentation statuses. It is also dominated by Scots pine forests. We selected this area to study the relationship between the degree of FF at various spatial scales, the combinations of these scales, and the probability of Scots pine mortality, as expressed by the volume of dead wood removed during sanitation cuts.
The aim of this study is to answer the following research questions: (1) How do fragmentation index values at different spatial scales and multi-scales influence Scots pine FM probability change when key drivers such as site conditions and stand characteristics are removed? (2) Do all fragmentation scales significantly explain patterns of Scots pine FM, and which specific scales are most effective in predicting mortality? (3) What relationships exist between different scales and multi-scale fragmentation indices with Scots pine FM probability? (4) Does the influence of the multi-scale fragmentation index on the probability of Scots pine mortality differ from the influence of each scale? Which method is more advanced for studying the effects of FF on FM?
Answering these questions is crucial to improving our understanding of the complex relationships between FF degrees at all scales and the probability of Scots pine mortality.
Most studies focus on a single scale, but our multi-scale approach tests how fragmentation scales shape ecological outcomes. Understanding FM probabilities in pine ecosystems requires a multi-scale approach because single-scale assessments fail to capture the complex interactions between fine- and broad-scale fragmentation effects [44]. By quantifying fragmentation indices at multiple scales (50–3000 m) and analyzing their combined influence, our approach improves ecological practicality compared to the traditional single-scale method [45,46]. This framework also reveals important mechanistic insights that cannot be obtained through single-scale studies, such as how interactions between fine-scale edge effects (50–600 m), landscape-scale fragmentation (800–3000 m), and all-scale fragmentation (50–3000 m) amplify or mitigate mortality risk. These findings enable targeted management strategies, such as prioritizing reforestation in highly fragmented zones [47], while advancing our understanding of forest resilience under environmental change. By integrating these interactions across multiple scales, our method provides a more ecologically advanced representation of fragmentation effects and helps identify critical, scale-specific conservation strategies. This advancement is particularly valuable for managing Scots pine forests under increasing anthropogenic and climatic pressures because traditional single-scale assessments may overlook key drivers of tree mortality.

2. Research Material and Methods

2.1. Study Site Description

A research study was conducted at the Oleszyce Forest Inspectorate (OFI) (see Figure 1). The OFI includes forests and agricultural land [48]. The OFI is located in southeastern Poland, in the northeastern part of the Subcarpathian Voivodeship (N 54.0788°, E 13.4787°, at a maximum altitude of 272 m and minimum altitude of 195 m above sea level. The total area of the territory is 33,754 hectares. Of this area, 11,805 hectares are covered by the inspectorate’s forests. There are 20,179 forest stands in the OFI, among these 11,660 stands are occupied by Pine. The inspectorate’s maximum mean temperature is 18.7 °C, and minimum mean temperature is 7.8 °C during the vegetation season (May–August). The maximum mean precipitation is 841 mm, and minimum mean precipitation is 519 mm. The Forest Inspectorate’s soil is predominantly brown (44%), followed by rusty (20.3%), fawn (9.3%), and soil-gravel (6.7%). The pine stands are on average 66 years old and 22.3 m tall, have a diameter of 30 cm, and have a volume of 230.54 m3 (Table 1). Moreover, the species composition of forest stands is dominated by Scots pine, which makes up 54.9% of the total forest stands in the Forest Inspectorate, followed by oak (11.90%), alder (10.9%), Beech (9.7%), birch (6.3%), and others (6.3%). The most common forest habitat in the OFI is mixed coniferous forest, which comprises 36.3% of the total area. It is followed by fresh mixed forest (11.3%) and fresh forest (10.6%).

2.2. Research Methods

The research methodology was carried out in four stages. First, forest data sources were identified. We have selected data from a time period that corresponds to a dry period in Poland. Information on forest fragmentation was calculated. Then, data on deadwood volume from sanitary cutting and stand characteristics of Scots pine forests were obtained. Additionally, environmental data, including climate, topography, soil types, and geological types, was gathered. Second, data preprocessing was performed. Third, the data were analyzed using the generalized additive model (GAM). Fourth, the partial dependence graphs showing the relationship and effect of the fragmentation index on the probability of Scots pine FM were explained and interpreted.
The stages of the methods used are shown in the flowchart in Figure 2. Yellow indicates the data sources. Green indicates the data collection and preprocessing stage. Blue indicates the modeling stage. The gray box denotes the analysis of the results.

2.3. Data Sources and Data Collection

In this study, data were collected from three sources: (1) The forest cover data was obtained from the Forest Information System for Europe [50]. The essential data contain forestland cover with a resolution of 10 m registered in 2018. The spatial reference system of the data is the European LAEA (EPSG: 3035) projection. We used forest masks taken from Poland’s land cover. The forest mask was limited to forest under national management. (2) Sanitation cutting data: It is the harvesting of trees for the removal of insects or disease from a forest stand [51]. In order to protect the forests, sanitary cutting is mandatory in the forests managed by the State Forests in Poland. It is carried out on an ongoing basis depending on the occurrence of tree mortality [52]. In Polish State Forests, the collection of mortality data is conducted by individual forest districts as a component of their forest management operations, subsequently compiling it within the National Forest Information System. This system offers insights into the number of sanitary cuts executed within each stand that had not been previously designated for silvicultural treatments or final cuts. The dead wood volume of each tree removed in sanitary cuts is carefully measured. The data encompass information pertaining to the causes of mortality, which are categorized into two distinct groups: (i) bark beetle/drought and (ii) wind. For the purposes of this research, data pertaining to Scots pine mortality resulting from drought and bark beetles was selected. Data was collected on the basis of information contained at National Forest Inventory (NFI) for the OFI. The data on the total volume of trees harvested as part of sanitary cutting operations in each forest stand in each year of the 2015–2022 periods were the only measure of the pine mortality in the research. (3) The environmental variables data were collected from existing sources. For information on soil types, data was obtained from the soil map [53], and for geology types, the data source is the geological maps of the National Geological Institute. Soil types are classified based on main soil types as gleyearth, mads, fawn, and rusty. Likewise, geology types are classified based on the origin of the geology of the area: clay, lesses, Eolic sands, sands, gravels, and limestone. Information on the altitude and slope of the area are obtained from a numerical terrain model developed from the LiDAR scan of Poland [54]. We obtained climate data from the Polish Institute of Meteorology and Water Management [49,52] for the years 2015 to 2022.

2.4. Data Preprocessing

Data preprocessing is performed to avoid data’s impact on the performance of the analysis and the model. The forest fragmentation (FF) data was calculated as values for each forest stand for the centriod of the spatial raster layer and extracted in numeric form for each different scale and multi-scale FF. Checking and filling or removing the missing values in the data set was performed. To fill in missing values, we used the average of the five nearest adjacent values when there were a few missing values in a record. If the number of missing values in a record was large, we removed the entire record. We calculated the climatic water balance (CWB) to describe water and climate conditions. The solution proposed by [55,56] was adopted as the method for calculating potential evapotranspiration. The Standardized Climatic Water Balance (SCWB) has also been calculated [57]. The SCWB was calculated for each 100 m × 100 m grid cell. The CWB and SCWB were spatially averaged for the individual forest stands.
We calculated interquartile range (IQR) and used it to remove outliers. The outlier detection method identifies outliers as points that fall below the first quartile minus 1.5(IQR) or above the third quartile plus 1.5(IQR), as described in [58].
After completing all the steps, the missing values and outliers were identified and removed or replaced. Then, the continuous explanatory variables, such as elevation, rainfall, and temperature, were normalized to a range between zero and one. The categorical response and explanatory variables were encoded. The categorical response variable means that FM is encoded as class 0 for forest stands without deadwood volume of Scots pine removed during sanitation cutting and as class 1 for forest stands with deadwood volume of Scots pine removed during sanitation cutting. The categorical explanatory variable is encoded as follows: Soil fertility level was encoded as class 0 for forests grown off of farmland and class 1 for forests grown in farmland. Soil types were encoded as follows: class 0 = gley earth; class 1 = mads; class 2 = fawn; class 3 = rusty. Geology types were encoded as follows: class 0 = clay; class 1 = lesses; class 2 = eolian sands; class 3 = sands and gravels; class 4 = limestone.
Finally, we found that the value of the response variable is strongly imbalanced. This dataset contains 955 records in class 1 (forest stands with deadwood volume of Scots pine) and 8573 records in class 0 (forest stands without deadwood volume of Scots pine). We implemented a combination of bootstrap sampling and random oversampling machine learning methods to balance the given minority class dataset for further analysis by sampling with replacement. This was carried out to improve the representation of the data and performance of models in scenarios where the minority class is of greater interest.

2.5. Model Development

In order to analyze the effect of the FF relationship on the probability of Scots pine FM, we used generalized additive model (GAM). Among the many methods for building models, Ref. [59] showed the effectiveness of GAM. GAM models are flexible and powerful statistical modeling techniques for analyzing complex relationships between response and multiple explanatory variables. They involve modeling a relationship between a response variable and each of its explanatory variables individually, using a smoothing and tensor function, combining these in an additive manner. In this way, non-linear relationships between variables can be captured, making GAM models suitable for use with complex and non-linear structures. GAM uses a combination of both linear and nonlinear functions to describe the relationship between the dependent and independent variables. The strength of GAM is its ability to deal with highly non-linear and non-monotonic relationships between the response and the set of explanatory variables. In addition, GAMs can handle both continuous and categorical explanatory variables, as well as interactions between variables. The GAM model can help develop ecological models that better explain underlying data, improving our understanding of ecological systems [60].
A crucial step in the application of the GAM model is the selection of the appropriate level of smoothing for a predictor. This is best achieved by specifying the level of smoothing using the concept of effective degrees of freedom. GAM can model interactions between predictors by including tensor product smooth terms. These terms capture the joint effects of multiple predictors on the response variable [61]. Moreover, GAM can model the combined effects on the response variable simultaneously [62].
GAM model maintains a high degree of interpretability. This is achieved by representing the effects of predictors as smoothing functions.
The detection of potential forest dieback in stands involves classifying them into two categories. The generalized additive model is used as a classifier, and its response is thresholded to obtain class 0 (no forest mortality, or FM) or class 1 (FM). The model’s response before thresholding is related to the probability of FM occurrence.
The variance inflation factor (VIF) was used to detect multicollinearity among multiple continuous explanatory variables. High collinearity between the predictors leads to high VIF values [63]. We started with all the predictors, and we calculated the VIF values for each. Then, we iteratively removed the variable with the highest VIF value until all the remaining predictors had VIF values of around 5.0.
In the model-building approach, we used the select=TRUE function in mgcv to enable double penalty shrinkage. We diagnosed the model summary, looking into p-values, and again checked the term significance evaluation. Effective degrees of freedom (EDFs) > 1 represent non-linear effects, and an EDF near 0 indicates that the term has been effectively removed. p-values < 0.05 indicate significance. After completing these steps, the GAM variable selection balances statistical significance, model parsimony, and domain relevance. To know the importance variables involved in the model, we calculated the chi-square value of each variable from the model. Initially, we developed a model for site conditions and stand characteristics that did not include forest fragmentation indices. During this stage, we removed the following categorical explanatory variables: forest grown on farmland versus off farmland, soil types, and geological types. We also removed the following redundant continuous variables: height, diameter, and age. These variables were found to be insignificant to the model. The final base model describing the probability of Scots pine forest mortality is built on: site index, altitude, area, stand volume, carbon–nitrogen ratio, and interactions: SCWB1,2,3, AMR1,2,3 and AMT1,2,3 All predictor variables’ full names are stated in the description of Table 1.
The FF index was calculated to gather information about forest fragmentation. This index is based on an analysis of forest patterns, with similarity calculated using the Jensen–Shannon divergence approach [41]. The FF index ranges from 0 to 1, taking a value close to 0 for areas sparsely covered by trees. A value of 1 indicates areas that are continuously covered by forests. The FF index was calculated for OFI at different scales (diameter of neighborhood): 50 m, 100 m, 200 m, 400 m, 600 m, 800 m, 1000 m, 1200 m, 1400 m, 1600 m, 1800 m, 2000 m, 2500 m, and 3000 m. We added the FF index to the model as the sixth term. Fourteen models were developed in this way. After analyzing each scale independently, we calculated multiscale FF indices using combinations of the 50–600 m, 800–3000 m, and 50–3000 m scales.

3. Results

Each model included the following independent variables representing twelve site conditions: altitude; SCWB1, SCWB2, and SCWB3; AMR1, AMR2, and AMR3; AMT1, AMT2, and AMT3; carbon-to-nitrogen ratio (C/N); area; and two stand characteristics: site index (SI) and stand volume. To account for the impact of FF on FM, the model included the FF index. This enabled a detailed evaluation of the contribution of each scale of fragmentation to Scots pine mortality. Furthermore, the multiscale FF index was analyzed as a fragmentation representation in the model. This allows us to study the impact of FF on FM without focusing on only one scale.
A total of eighteen classification models were developed during modeling. The initial model (base model) incorporated site conditions and stand characteristics. In that model, all predictors were statistically significant, with an average p-value of 2 · 10 − 16 . The base model explained the effects of site conditions and stand characteristics. After adding the FF index, seventeen additional models were developed, including the impact of FF. The FF index was statistically significant at different p-values (Table 2). We visualized the partial effects of the fragmentation index at different fragmentation scales (Figures 4–6). Since the base model explained the variability of each forest stand’s environment, analyzing the partial effects of the FF index allowed us to determine the relationship between FF and FM without the impact of environmental conditions.
Additionally, the significance of the various independent variables included in the model was analyzed. Figure 3 shows the significance of these variables. Variables from the basic model (Figure 3A) and fragmentation indicators (Figure 3B) are presented separately. Note the distinctive impact of the multi-scale forest fragmentation factor, which combines scales ranging from 50 to 3000 m. Its impact is more than double that of FF ratios for the other scales.
Analyses of the base model and the base model with a fragmentation influence demonstrated consistent, moderately predictive performance across all evaluation metrics (Table 2). The base model had an area under the curve (AUC) of 0.818. Adding fragmentation scales improved model performance across all metrics evaluated. The AUC increased slightly, peaking at 0.835 for the 200–1000 m scale and stabilizing at 0.818–0.835 for the other scales. Model accuracy ranged from 0.734 to 0.758, showing slight improvement with the inclusion of fragmentation scales compared to the baseline model (accuracy = 0.734). Similarly, Kappa values, which assess the agreement between predicted and observed classifications, increased modestly, ranging from 0.467 to 0.517 compared to the baseline model (Kappa = 0.467).

3.1. The Relationship Between FF and the Probability of FM at Different Spatial Scales

To determine which spatial scale most strongly drives Scots pine mortality, we analyzed the effects of fragmentation separately at each scale. The small-scale effect matters more than the large-scale effect when it comes to the probability of Scots pine mortality. Each fragmentation scale (from 50 m to 3000 m) has a different effect on the probability of FM (see Figure 4, Figure 5 and Figure 6). FF at scales from 50 m to 600 m has a linear relationship with FM, whereas FF at scales from 800 m to 3000 m has a nonlinear relationship.
We observed a negative relationship between fragmentation and the probability of pine mortality at spatial scales of 50, 100, 200, 400, and 600 m in higher fragmentation ranges (0.985–1, 0.993–1, 0.972–1, 0.686–0.886, and 0.85–1, respectively). A negative relationship was seen when increased fragmentation correlated with decreased mortality, resulting in a positive effect ranging from −0.0151 to −0.0157 at 50 m, from −0.0415 to −0.0439 at 100 m, from −0.084 to −0.0007 at 200 m, from −0.0011 to −0.092 at 400 m, and from −0.041 to −0.253 at 600 m. Conversely, a positive relationship was observed at intermediate fragmentation ranges (0.2997–1.0000, 0.0170–0.9210, 0.5984–0.9645, 0.4086–0.7751, and 0.3890–0.8210, respectively). A positive relationship occurred when fragmentation and tree mortality increased. The relationship became neutral at fragmentation index values of 1 (at 50 and 100 m), 0.9658 (at 200 m), 0.775 (at 400 m), and 0.821 (at 600 m) (see Figure 4). In this case, changes in fragmentation do not lead to measurable increases or decreases in mortality, which highlights the lack of a relationship between fragmentation and mortality.
We observed a negative relationship between forest fragmentation (FF) and probability at spatial scales of 800, 1000, 1200, 1400, 1600, 1800, 2000, 2500, and 3000 m in the higher fragmentation ranges (0.643–0.813, 0.76–0.951, 0.792–1.000, 0.793–1.000, 0.567–0.854, 0.621–0.874, 0.497–0.789, 0.579–0.782, and 0.513–0.884, respectively). Specifically, increased fragmentation was correlated with decreased mortality. Conversely, a positive relationship occurred within intermediate fragmentation ranges (0.61–0.442, 0.74–0.47, 0.792–1, 0.793–1, 0.567–0.854, 0.621–0.874, 0.497–0.789, 0.506–0.379, and 0.424–0.513, respectively). However, the effect remained negative, with coefficients including 0.0.172 (800 m), 0.67 (1000 m), 0.875 (1200 m), 0.736 (1400 m), 1.57 (1600 m), 1.37 (1800 m), 1.857 (2000 m), 1.87 (2500 m), and 1.571 (3000 m). The relationship became neutral at fragmentation index values of 0.601 and 0.82 (800 m), 0.75 and 0.94 (1000 m), 0.792 and 0.91 (1200 m), 0.79 and 0.532 (1400 m), 0.573 and 0.843 (1600 m), 0.876 and 0.612 (1800 m), 0.793 and 0.53 (2000 m), 0.803 and 0.523 (2500 m), and 0.799 and 0.531 (3000 m) (Figure 5 and Figure 6). This study examined the relationship between fragmentation and mortality, not the reverse.
Note that when classifying by the trend in the FF–FM relation, all analyzed scales can be divided into three groups, as illustrated in Figure 4, Figure 5 and Figure 6.

3.2. The Relationship Between FF and the Probability of FM at Different Spatial Multi-Scale Indices

The partial dependence analysis reveals that the FF index’s relationship with FM at the multi-scale level exhibits different tendencies: positive, negative, and neutral (Figure 7). Specifically, a positive relationship was observed between fragmentation indices ranging from 0.3 to 0.96 for a scale composition of 50–600 m, from 0.607 to 0.123 and from 0.832 to 0.889 for a scale composition of 800–3000 m, and from 0.287 to 0.738 for all scale compositions. This indicates that fragmentation within this range increases the probability of FM. In this case, an increase in fragmentation and forest mortality is observed. Conversely, a negative relationship was found between fragmentation indices of 0.96 and 1.00 for scale compositions of 50–600 m, 0.608–0.832 for scale compositions of 800–3000 m, and 0.730–0.930 for all scale compositions, showing that fragmentation in this range positively affects pine forest mortality. This means that as fragmentation increases, forest mortality decreases. However, at a fragmentation index of 0.889 for a scale composition of 50–600 m, 0.608–0.821 for a scale composition of 800–3000 m, and 0.731–0.931 for all scale compositions, there is no significant influence of FF on FM. This implies that when forests become highly fragmented, the likelihood of forest mortality neither increases nor decreases. The highest probabilities of forest mortality occur at fragmentation indices of 0.5 (for scales of 50–600 m and 50–3000 m) and 0.25 (for scales of 800–3000 m).

3.3. The Spatial Presentation of the Relationship Between FF and the Probability of FM at Multi-Scales

Spatial maps of tree mortality probability (Figure 8B) and the partial effects of fragmentation (Figure 8A) at the multiscale level (50–3000 m), which is most signifiicant to the model, are essential to forest conservation efforts. These maps visually demonstrate how fragmentation increases mortality risk and triggers harmful ecological feedback loops that threaten ecosystem integrity. These maps instantly reveal geographic mortality hotspots, prompting investigation into their suggested potential relationships through visual patterns. Furthermore, these maps offer a powerful assessment of the geographic plausibility of underlying statistical models. The wspatial display of these results is crucial for research, policy, and practice because maps significantly enhance comprehension by revealing high-risk zones and fragmentation severity and providing essential landscape context. This enables targeted mitigation, helps policymakers prioritize funding, and makes complex data accessible to non-experts, fostering dialogue and collaboration to develop solutions. By pinpointing areas that require specific resilience strategies, spatial maps transform model outputs into actionable intelligence, driving concrete policies such as ecological connectivity laws, infrastructure guidelines that avoid biodiversity hotspots, and national conservation agendas. For forest managers, these maps are indispensable. They communicate complex risk information to policymakers, community leaders, and the public far more effectively than raw models do. This clarity raises critical awareness and serves as a potent catalyst for action.

4. Discussion

Our study presents a novel methodological framework for analyzing how fragmentation drives Scots pine mortality across spatial scales. This framework disentangles the effects of FF from other factors, such as site conditions and stand characteristics. Our three-step methodology—(1) isolating the partial effects of fragmentation, (2) identifying scale-specific thresholds ranging from 50 to 3000 m, and (3) assessing the usability of the multiscale FF index—provides a replicable approach to quantifying fragmentation effects. Rigorously validated models (Table 2) confirm the framework’s reliability, balancing moderate predictive accuracy with ecological interpretability. These results challenge conventional single-scale analyses by demonstrating that mortality patterns result from the interaction of local processes (e.g., edge effects) and broader, landscape-scale disturbances.
This framework utilizes both discrete and composite spatial scales. This approach surpasses the limitations of single-scale analyses by explicitly quantifying the synergistic modulation of mortality risk by cross-scale interactions between fine-scale edge effects (50–600 m), landscape-scale fragmentation (800–3000 m), and integrated all-scale fragmentation (50–3000 m). These critical mechanistic insights, which are unobtainable via single-scale investigations, elucidate the complex ecological drivers of FM. Furthermore, the framework improves analytical precision by reducing scale-related errors and uncertainties, resulting in optimized model outputs. These robust, scale-integrated findings directly inform and advance evidence-based forest management and conservation planning.
Our results show that the base model confirmed that site conditions and that stand characteristics were the primary drivers of mortality in Scots pine forest, consistent with previous studies [45,64]. However, after removing the effect of environment by creating the base model, the relation between FF and FM could be shown across spatial scales (50–3000 m). FF significantly influenced mortality probability and improved model performance (Table 2).
Analyses of the partial dependence of FF on FM reveal that the impact of fragmentation on Scots pine mortality varies across spatial scales. Depending on the scale range considered, the impact can be positive, negative, or neutral (Figure 4, Figure 5 and Figure 6). This finding is supported by research indicating that different ecological processes dominate at different fragmentation scales [65]. The influence of fine-to-intermediate-scale fragmentation (50–600 m) on tree mortality probability highlights the critical role of local environmental conditions, including edge effects, microclimate shifts, and heightened disturbance susceptibility ([8,10]; Figure 4). Broad-scale fragmentation (800–3000 m) disrupts landscape connectivity and resource availability, reducing forest resilience to drought and pests ([66,67]; see Figure 5 and Figure 6). The discrepancy between studies emphasizing edge effects [22] and those emphasizing landscape-scale forest loss [65] may reflect ecosystem-specific thresholds. In contrast, some studies report scale-invariant fragmentation effects [68], possibly due to analyses conducted on homogeneous landscapes.
Our study advances the field of forest fragmentation ecology by systematically analyzing fragmentation scale indices at defined spatial extents (50–600 m and 800–3000 m) and integrating them into composite, multi-scale indices. Multi-scale fragmentation indices are essential for understanding the complex impacts of fragmentation on forest mortality. Different scales reveal different mechanisms: local scales highlight edge effects and microhabitat changes, and landscape scales reveal connectivity and population dynamics. Using multi-scale approaches improves the accuracy of mortality risk assessments and supports better forest management and conservation planning [41]. This approach quantifies the synergistic effects on pine mortality. Recent research shows that the effects of fragmentation on mortality depend strongly on scale. At local scales, fragmentation often increases mortality due to higher exposure to edge effects and reduced habitat quality. At larger scales, however, the effects can be more complex and sometimes positive for certain species due to geometric effects [41,69,70].
The positive relationship between forest fragmentation and mortality, particularly at high fragmentation indices (see Figure 4, Figure 5 and Figure 6), underscores the necessity of targeted conservation efforts. Fragmented forests exhibit reduced ecological resilience and increased vulnerability to environmental stressors [71,72], a pattern consistent with our findings. Disrupting ecosystem stability directly increases mortality risk through loss of forest connectivity, altered microclimates, and enhanced edge effects [73]. The effect of forest fragmentation from 800 m to 3000 m has a nonlinear relationship with pine mortality probability; fragmentation effects vary in magnitude and direction. We observed a linear relationship between the 50 to 600 m scale of fragmentation and pine mortality, indicating a direct and consistent effect. These results are in line with previous studies showing that fragmentation effects are linear on a fine scale [74,75].
The multi-scale fragmentation effect demonstrates how mortality patterns arise from the interaction between edge effects and fragment configuration, such as size, shape, and isolation. We observed scale-dependent effects of FF on the mortality probability of Scots pine forests. To advance our understanding, we grouped relatively similar scales and developed a composite scale (multi-scale). This approach helps us understand the cumulative effects of the indices on mortality across these similar scales. This multi-scale analysis is important for determining the average effect of these scales and for providing recommendations for policy direction based on the study. To our knowledge, this approach improves upon the traditional single-scale approach and provides valuable insights into the study of multi-scale effects. This perspective enables researchers to identify the thresholds at which the effects of fragmentation are most pronounced, offering valuable insights for targeted conservation and management interventions. By incorporating fragmentation metrics at relevant scales, managers can better predict mortality patterns and prioritize areas for restoration or protection [72]. These interventions can be modified to fit the specific scale of forest fragmentation and assist decision-makers in allocating resources efficiently to areas where fragmentation-induced mortality is most severe. Practical applications of multi-scale fragmentation research on Scots pine forests are very important. They identify critical thresholds of mortality risks. Using this information, managers can prioritize the restoration of landscape connectivity in areas crossing these thresholds. It provides an early warning and helps with climate-resilient reforestation by designing tree-planting schemes using scale-dependent thresholds. Our findings provide the spatial proposal for effective Scots pine conservation in the face of climate change, demonstrating that fragmentation policies must be as multi-scalar as the ecological processes they aim to protect. This approach could be extended to other foundation species, such as oak and beech, creating a new standard for scale-explicit forest management. This study bridges the gap between analyses at scales ranging from 50 m to 3000 m, minimizing scale-dependent biases [46], and providing a robust assessment of fragmentation effects [66,76].
It is critically important to display models’ results of mortality probability (Figure 8B) and fragmentation effects (Figure 8A) through spatial maps for research, practical applications, and policy development. These maps significantly enhance comprehension by instantly revealing spatial patterns, such as high-risk zones and areas of severe fragmentation. They provide an essential, contextual understanding of the landscape. Furthermore, maps enable targeted action by guiding precise mitigation efforts and allowing policymakers to prioritize funding for high-impact areas. Their visual nature also makes complex data accessible to non-experts, fostering crucial dialogue between researchers, NGOs, and local communities to co-design effective solutions. By identifying regions that require specific resilience strategies, spatial maps transform abstract model outputs into actionable intelligence. This intelligence enables concrete policy outcomes, such as laws mandating ecological connectivity, infrastructure guidelines that avoid biodiversity hotspots, and national conservation agendas.
A composite fragmentation metric that synthesizes multiple scales captures the average effect on Scots pine forest mortality probability and provides a holistic understanding of ecological impacts. Adopting a multi-scale perspective advances the study of fragmentation effects beyond the limitations of traditional single-scale research and provides a framework for conservation and sustainable forest management.
While this study provides insights into Scots pine mortality driven by dry periods, several limitations must be acknowledged. First, our focus on dry-period data means the potential influence of wet periods on mortality remains unassessed. Additionally, wind destruction events and the inherent methodological considerations of chronological data analyses may contribute to mortality patterns to some degree. These limitations concerning data scope, disturbance factors, and analytical methods represent important avenues for future research.
Forest fragmentation poses serious threats to forest ecosystems and climate stability. Addressing these issues requires a multifaceted approach integrating ecological restoration, sustainable land use, and community engagement. Implementing targeted strategies can mitigate the negative effects of fragmentation, restore degraded forests, and increase ecosystem resilience for future generations.

5. Conclusions

The study demonstrates that forest fragmentation (FF) has spatially scale-dependent effects on the probability of mortality in Scots pine that are independent of site conditions and stand characteristics. By systematically isolating the effects of fragmentation (FF), the multi-scale framework reveals synergistic interactions that amplify ecological vulnerability and explain a substantial proportion of mortality variability. Our analysis emphasizes the importance of integrating multi-scale perspectives to fully understand the cumulative ecological consequences of fragmentation. It shows that small-scale edge effects and connectivity loss interact with broader, landscape-scale disturbances to influence mortality patterns. Our findings underscore the necessity of scale-aware conservation strategies that prioritize maintaining forest connectivity, establishing buffer zones, and mitigating edge effects at all spatial scales. This approach bridges local and landscape-scale fragmentation dynamics, enhancing the precision of conservation planning and policymaking and providing actionable insights to strengthen resilience and mitigate mortality risks in Scots pine ecosystems.

Author Contributions

Conceptualization, D.D.F. and P.N.; methodology, D.D.F. and P.N.; validation, D.D.F. and P.N.; formal analysis, D.D.F.; investigation, D.D.F. and P.N.; data curtain, D.D.F. and J.T.; writing—original draft preparation, D.D.F.; writing—review and editing, D.D.F., P.N. and J.T.; visualization, D.D.F., J.T. and P.N.; supervision, P.N.; project administration, P.N. All authors have read and agreed to the published version of the manuscript.

Funding

These studies were conducted as part of a bilateral project between the Czech Science Foundation (project number 25-14626L) and the Polish National Science Center (project number OPUS LAP 2023/51/I/NZ9/01813).

Data Availability Statement

Data are available upon request. For further inquiries, please get in touch with the corresponding author directly.

Conflicts of Interest

The authors report no financial or personal conflicts of interest that could be perceived as influencing the results or interpretation of this research.

References

  1. Pan, Y.; Birdsey, R.A.; Fang, J.; Houghton, R.; Kauppi, P.E.; Kurz, W.A.; Phillips, O.L.; Shvidenko, A.; Lewis, S.L.; Canadell, J.G.; et al. A large and persistent carbon sink in the world’s forests. Science 2011, 333, 988–993. [Google Scholar] [CrossRef] [PubMed]
  2. Buytaert, W.; Cuesta-Camacho, F.; Tobón, C. Potential impacts of climate change on the environmental services of humid tropical alpine regions. Glob. Ecol. Biogeogr. 2011, 20, 19–33. [Google Scholar] [CrossRef]
  3. Chen, S.; Woodcock, C.E.; Bullock, E.L.; Arévalo, P.; Torchinava, P.; Peng, S.; Olofsson, P. Monitoring temperate forest degradation on Google Earth Engine using Landsat time series analysis. Remote Sens. Environ. 2021, 265, 112648. [Google Scholar] [CrossRef]
  4. Galicia, L.; Zarco-Arista, A.E. Multiple ecosystem services, possible trade-offs and synergies in a temperate forest ecosystem in Mexico: A review. Int. J. Biodivers. Sci. Ecosyst. Serv. Manag. 2014, 10, 275–288. [Google Scholar] [CrossRef]
  5. Morreale, L.L.; Thompson, J.R.; Tang, X.; Reinmann, A.B.; Hutyra, L.R. Elevated growth and biomass along temperate forest edges. Nat. Commun. 2021, 12, 7181. [Google Scholar] [CrossRef]
  6. Estreguil, C.; Caudullo, G.; de Rigo, D.; San-Miguel-Ayanz, J. Forest Landscape in Europe: Pattern, Fragmentation and Connectivity; Joint Research Centre, Institute for Environment and Sustainability: Ispra, Italy, 2013. [Google Scholar] [CrossRef]
  7. Debebe, D.; Feleha, D.; Tymińska-Czabańska, L.; Netzel, P. Academic Editor: Mark E. Harmon Forest Fragmentation and Forest Mortality—An In-Depth Systematic Review. Forests 2025, 16, 565. [Google Scholar] [CrossRef]
  8. Laurance, W.F.; Dell, B.; Turton, S.M.; Lawes, M.J.; Hutley, L.B.; McCallum, H.; Dale, P.; Bird, M.; Hardy, G.; Prideaux, G.; et al. The 10 Australian ecosystems most vulnerable to tipping points. Biol. Conserv. 2011, 144, 1472–1480. [Google Scholar] [CrossRef]
  9. Eckenwalder, J.E. Conifers of the World: The Complete Reference; Timber Press: Portland, OR, USA, 2009. [Google Scholar]
  10. Buras, A.; Schunk, C.; Zeiträg, C.; Herrmann, C.; Kaiser, L.; Lemme, H.; Straub, C.; Taeger, S.; Gößwein, S.; Klemmt, H.J.; et al. Are Scots pine forest edges particularly prone to drought-induced mortality? Environ. Res. Lett. 2018, 13, 025001. [Google Scholar] [CrossRef]
  11. Pasichnyk, D.; Lesinski, J.; Socha, J. Variability in biometric traits of Scots pine needles in Poland. Sylwan 2022, 166, 681–697. [Google Scholar] [CrossRef]
  12. Ambrose, M.J. Chapter 5—Tree mortality. Gen. Tech. Rep. SRS-261 2021, 261, 105–123. [Google Scholar]
  13. Senf, C.; Buras, A.; Zang, C.S.; Rammig, A.; Seidl, R. Excess forest mortality is consistently linked to drought across Europe. Nat. Commun. 2020, 11, 6200. [Google Scholar] [CrossRef]
  14. King, A.J.; Melbourne, B.A.; Davies, K.F.; Nicholls, A.O.; Austin, M.P.; Tuff, K.T.; Evans, M.J.; Hardy, C.M.; Cunningham, S.A. Spatial and temporal variability of fragmentation effects in a long term, eucalypt forest fragmentation experiment. Landsc. Ecol. 2018, 33, 609–623. [Google Scholar] [CrossRef]
  15. Ma, J.; Li, J.; Wu, W.; Liu, J. Global forest fragmentation change from 2000 to 2020. Nat. Commun. 2023, 14, 3752. [Google Scholar] [CrossRef] [PubMed]
  16. Filip, G.M.; Schmitt, C.L.; Scott, D.W.; Fitzgerald, S.A. Understanding and Defining Mortality in Western Conifer Forests. West. J. Appl. For. 2007, 22, 105–115. [Google Scholar] [CrossRef]
  17. Harmon, M.E.; Bell, D.M. Mortality in Forested Ecosystems: Suggested Conceptual Advances. Forests 2020, 11, 572. [Google Scholar] [CrossRef]
  18. Laurance, W.F. Theory meets reality: How habitat fragmentation research has transcended island biogeographic theory. Biol. Conserv. 2008, 141, 1731–1744. [Google Scholar] [CrossRef]
  19. Vieilledent, G.; Grinand, C.; Rakotomalala, F.A.; Ranaivosoa, R.; Rakotoarijaona, J.R.; Allnutt, T.F.; Achard, F. Combining global tree cover loss data with historical national forest cover maps to look at six decades of deforestation and forest fragmentation in Madagascar. Biol. Conserv. 2018, 222, 189–197. [Google Scholar] [CrossRef]
  20. Fischer, R.; Taubert, F.; Müller, M.S.; Groeneveld, J.; Lehmann, S.; Wiegand, T.; Huth, A. Accelerated forest fragmentation leads to critical increase in tropical forest edge area. Sci. Adv. 2021, 7, eabg7012. [Google Scholar] [CrossRef]
  21. Ries, L.; Fletcher, R.J.; Battin, J.; Sisk, T.D. Ecological Responses to Habitat Edges: Mechanisms, Models, and Variability Explained. Annu. Rev. Ecol. Evol. Syst. 2004, 35, 491–522. [Google Scholar] [CrossRef]
  22. Harper, K.; Macdonald, S.; Burton, P.; Chen, J.; Brosofske, K.; Saunders, S.; Euskirchen, E.; Roberts, D.; Jaiteh, M.; Esseen, P.A. Edge Influence on Forest Structure and Composition in Fragmented Landscapes. Conserv. Biol. 2005, 19, 768–782. [Google Scholar] [CrossRef]
  23. Laurance, W.F.; Lovejoy, T.E.; Vasconcelos, H.L.; Bruna, E.M.; Didham, R.K.; Stouffer, P.C.; Gascon, C.; Bierregaard, R.O.; Laurance, S.G.; Sampaio, E. Ecosystem Decay of Amazonian Forest Fragments: A 22-Year Investigation. Conserv. Biol. 2002, 16, 605–618. [Google Scholar] [CrossRef]
  24. Reinmann, A.B.; Hutyra, L.R. Edge effects enhance carbon uptake and its vulnerability to climate change in temperate broadleaf forests. Proc. Natl. Acad. Sci. USA 2017, 114, 107–112. [Google Scholar] [CrossRef] [PubMed]
  25. Davies-Colley, R.; Payne, G.; Elswijk, M. Microclimate gradients across a forest edge. N. Z. J. Ecol. 2000, 24, 111–121. [Google Scholar]
  26. Matlack, G.R. Microenvironment variation within and among forest edge sites in the eastern United States. Biol. Conserv. 1993, 66, 185–194. [Google Scholar] [CrossRef]
  27. Schwartz, N.B.; Uriarte, M.; DeFries, R.; Bedka, K.M.; Fernandes, K.; Gutiérrez-Vélez, V.; Pinedo-Vasquez, M.A. Fragmentation increases wind disturbance impacts on forest structure and carbon stocks in a western Amazonian landscape. Ecol. Appl. 2017, 27, 1901–1915. [Google Scholar] [CrossRef]
  28. Meeussen, C.; Govaert, S.; Vanneste, T.; Haesen, S.; Van Meerbeek, K.; Bollmann, K.; Brunet, J.; Calders, K.; Cousins, S.A.O.; Diekmann, M.; et al. Drivers of carbon stocks in forest edges across Europe. Sci. Total Environ. 2021, 759, 143497. [Google Scholar] [CrossRef]
  29. Chen, J.; Franklin, J.F.; Spies, T.A. Vegetation Responses to Edge Environments in Old-Growth Douglas-Fir Forests. Ecol. Appl. A Publ. Ecol. Soc. Am. 1992, 2, 387–396. [Google Scholar] [CrossRef]
  30. Barbati, A.; Bastrup-Birk, A.; Baycheva-Merger, T.; Bonhomme, C.; Bozzano, M.; Bücking, W.; Camia, A.; Caudullo, G.; Cienciala, E.; Cimini, D.; et al. State of Europe’s Forests 2011. Status and Trends in Sustainable Forest Management in Europe. In Proceedings of the FOREST EUROPE Ministerial Conference on the Protection of Forests in Europe, Oslo, Norway, 14–16 June 2011. [Google Scholar]
  31. Brando, P.M.; Balch, J.K.; Nepstad, D.C.; Morton, D.C.; Putz, F.E.; Coe, M.T.; Silvério, D.; Macedo, M.N.; Davidson, E.A.; Nóbrega, C.C.; et al. Abrupt increases in Amazonian tree mortality due to drought–fire interactions. Proc. Natl. Acad. Sci. USA 2014, 111, 6347–6352. [Google Scholar] [CrossRef]
  32. Tabarelli, M.; Cardoso da Silva, J.M.; Gascon, C. Forest fragmentation, synergisms and the impoverishment of neotropical forests. Biodivers. Conserv. 2004, 13, 1419–1425. [Google Scholar] [CrossRef]
  33. Lasy Państwowe z Dodatnim Wynikiem Finansowym za 2024 Rok. 2024. Available online: https://lochow.warszawa.lasy.gov.pl/aktualnosci/-/asset_publisher/1M8a/content/lasy-panstwowe-z-dodatnim-wynikiem-finansowym-za-2024-rok (accessed on 13 May 2025).
  34. Meeussen, C.; Govaert, S.; Vanneste, T.; Calders, K.; Bollmann, K.; Brunet, J.; Cousins, S.A.O.; Diekmann, M.; Graae, B.J.; Hedwall, P.O.; et al. Structural variation of forest edges across Europe. For. Ecol. Manag. 2020, 462, 117929. [Google Scholar] [CrossRef]
  35. Ordway, E.M.; Asner, G.P. Carbon declines along tropical forest edges correspond to heterogeneous effects on canopy structure and function. Proc. Natl. Acad. Sci. USA 2020, 117, 7863–7870. [Google Scholar] [CrossRef]
  36. Pinto, S.R.; Mendes, G.; Santos, A.M.; Dantas, M.; Tabarelli, M.; Melo, F.P.L. Landscape Attributes Drive Complex Spatial Microclimate Configuration of Brazilian Atlantic Forest Fragments. Trop. Conserv. Sci. 2010, 3, 389–402. [Google Scholar] [CrossRef]
  37. Riutta, T.; Clack, H.; Crockatt, M.; Slade, E.M. Landscape-Scale Implications of the Edge Effect on Soil Fauna Activity in a Temperate Forest. Ecosystems 2016, 19, 534–544. [Google Scholar] [CrossRef]
  38. Budniak, P.; Zięba, S. Effects of Forest Fragmentation on the Volume of Wood Resources in Managed, Pine-Dominated Forests in Poland. Forests 2022, 13, 590. [Google Scholar] [CrossRef]
  39. Kozak, J.; Ziółkowska, E.; Vogt, P.; Dobosz, M.; Kaim, D.; Kolecka, N.; Ostafin, K. Forest-Cover Increase Does Not Trigger Forest-Fragmentation Decrease: Case Study from the Polish Carpathians. Sustainability 2018, 10, 1472. [Google Scholar] [CrossRef]
  40. Mazgajski, T.; Żmihorski, M.; Abramowicz, K. Forest Habitat Loss and Fragmentation in Central Poland during the Last 100 Years. Silva Fenn. 2010, 44, 715–723. [Google Scholar] [CrossRef]
  41. Netzel, P.; Tyminska, L.; Dana Feleha, D.; Socha, J. New approach to assess forest fragmentation based on multiscale similarity index. Ecol. Indic. 2024, 158, 111530. [Google Scholar] [CrossRef]
  42. Rutkowski, R.; Jagołkowska, P.; Zawadzka, D.; Bogdanowicz, W. Impacts of forest fragmentation and post-glacial colonization on the distribution of genetic diversity in the Polish population of the hazel grouse Terastes bonasia. Eur. J. Wildl. Res. 2016, 62, 293–306. [Google Scholar] [CrossRef]
  43. Ziółkowska, E.; Ostapowicz, K.; Radeloff, V.C.; Kuemmerle, T. Effects of different matrix representations and connectivity measures on habitat network assessments. Landsc. Ecol. 2014, 29, 1551–1570. [Google Scholar] [CrossRef]
  44. Wu, J. Effects of changing scale on landscape pattern analysis: Scaling relations. Landsc. Ecol. 2004, 19, 125–138. [Google Scholar] [CrossRef]
  45. Allen, C.D.; Macalady, A.; Chenchouni, H.; Bachelet, D.; Mcdowell, N.; Vennetier, M.; Kitzberger, T.; Rigling, A.; Breshears, D.; Hogg, E.; et al. A global overview of drought and heat-induced tree mortality reveals emerging climate change risks for forests. For. Ecol. Manag. 2010, 259, 660–684. [Google Scholar] [CrossRef]
  46. McGarigal, K.; Cushman, S.; Neel, M.; Ene, E. FRAGSTATS: Spatial Pattern Analysis Program for Categorical Maps; University of Massachusetts: Amherst, MA, USA, 2012. [Google Scholar]
  47. Millar, C.I.; Stephenson, N.L. Temperate forest health in an era of emerging megadisturbance. Science 2015, 349, 823–826. [Google Scholar] [CrossRef]
  48. Klisz, M.; Buttò, V.; Rossi, S.; Morin, H.; Jastrzębowski, S. Intra-annual stem size variations converge across marginal populations of European beech. Trees 2020, 34, 255–265. [Google Scholar] [CrossRef]
  49. Netzel, P. Standardized Climatic Water Balance for Poland (1951–2023). 2025. Available online: https://landtax.urk.edu.pl/SCWB (accessed on 29 June 2025).
  50. Forest Information System of Europe. 2024. Available online: https://forest.eea.europa.eu (accessed on 13 May 2025).
  51. Pietzsch, B.W.; Peter, F.; Berger, U. The Effect of Sanitation Felling on the Spread of the European Spruce Bark Beetle—An Individual-Based Modeling Approach. Front. For. Glob. Chang. 2021, 4, 704930. [Google Scholar] [CrossRef]
  52. Socha, J.; Hawryło, P.; Leszczyński, K.; Tymińska-Czabańska, L. Height growth and site index models for main forest forming tree species in Poland. Sylwan 2023, 167, 617–632. [Google Scholar] [CrossRef]
  53. Ballabio, C.; Lugato, E.; Fernández-Ugalde, O.; Orgiazzi, A.; Jones, A.; Borrelli, P.; Montanarella, L.; Panagos, P. Mapping LUCAS topsoil chemical properties at European scale using Gaussian process regression. Geoderma 2019, 355, 113912. [Google Scholar] [CrossRef]
  54. European Digital Elevation Model (EU-DEM). 2024. Available online: https://www.eea.europa.eu/en/datahub/datahubitem-view/d08852bc-7b5f-4835-a776-08362e2fbf4b (accessed on 7 April 2025).
  55. Pascual Ferrer, J.; Candela Lledó, L. Water balance on the Central Rift Valley. In Case Studies for Developing Globally Responsible Engineers; Global Dimension in Engineering Education: Cham, Switzerland, 2015; pp. 563–591. [Google Scholar]
  56. Wilm, H.G.; Thornthwaite, C.W.; Colman, E.A.; Cummings, N.W.; Croft, A.R.; Gisborne, H.T.; Harding, S.T.; Hendrickson, A.H.; Hoover, M.D.; Houk, I.E.; et al. Report of the Committee on Transpiration and Evaporation, 1943-44. Trans. Am. Geophys. Union 1944, 25, 683–693. [Google Scholar] [CrossRef]
  57. Łabędzki, L.; Bąk, B. Standardized climatic water balance as drought index. Acta Agrophysica 2004, 3, 117–124. [Google Scholar]
  58. Tukey, J.W. Exploratory Data Analysis; Addison-Wesley Pub. Co.: Reading, MA, USA, 1977. [Google Scholar]
  59. Aertsen, W.; Kint, V.; van Orshoven, J.; Özkan, K.; Muys, B. Comparison and ranking of different modelling techniques for prediction of site index in Mediterranean mountain forests. Ecol. Model. 2010, 221, 1119–1130. [Google Scholar] [CrossRef]
  60. Guisan, A.; Edwards, T.C.; Hastie, T. Generalized linear and generalized additive models in studies of species distributions: Setting the scene. Ecol. Model. 2002, 157, 89–100. [Google Scholar] [CrossRef]
  61. Baayen, R.H.; Linke, M. Generalized Additive Mixed Models. In A Practical Handbook of Corpus Linguistics; Springer International Publishing: Cham, Switzerland, 2020; pp. 563–591. [Google Scholar] [CrossRef]
  62. Wood, S.N. Generalized Additive Models: An Introduction with R, Second Edition, 2nd ed.; Chapman and Hall/CRC: New York, MY, USA, 2017. [Google Scholar] [CrossRef]
  63. James, G.; Witten, D.; Hastie, T.; Tibshirani, R. (Eds.) Statistical Learning. In An Introduction to Statistical Learning: With Applications in R; Springer: New York, NY, USA, 2021; pp. 15–57. [Google Scholar] [CrossRef]
  64. McDowell, N.G.; Allen, C.D.; Anderson-Teixeira, K.; Aukema, B.H.; Bond-Lamberty, B.; Chini, L.; Clark, J.S.; Dietze, M.; Grossiord, C.; Hanbury-Brown, A.; et al. Pervasive shifts in forest dynamics in a changing world. Science 2020, 368, eaaz9463. [Google Scholar] [CrossRef]
  65. Fahrig, L.; Arroyo-Rodríguez, V.; Bennett, J.R.; Boucher-Lalonde, V.; Cazetta, E.; Currie, D.J.; Eigenbrod, F.; Ford, A.T.; Harrison, S.P.; Jaeger, J.A.G.; et al. Is habitat fragmentation bad for biodiversity? Biol. Conserv. 2019, 230, 179–186. [Google Scholar] [CrossRef]
  66. Anderegg, W.R.L.; Hicke, J.A.; Fisher, R.A.; Allen, C.D.; Aukema, J.; Bentz, B.; Hood, S.; Lichstein, J.W.; Macalady, A.K.; McDowell, N.; et al. Tree mortality from drought, insects, and their interactions in a changing climate. New Phytol. 2015, 208, 674–683. [Google Scholar] [CrossRef]
  67. Didham, R.K.; Kapos, V.; Ewers, R.M. Rethinking the conceptual foundations of habitat fragmentation research. Oikos 2012, 121, 161–170. [Google Scholar] [CrossRef]
  68. Taubert, F.; Fischer, R.; Groeneveld, J.; Lehmann, S.; Müller, M.S.; Rödig, E.; Wiegand, T.; Huth, A. Global patterns of tropical forest fragmentation. Nature 2018, 554, 519–522. [Google Scholar] [CrossRef]
  69. Bogaert, J.; Barima, Y.S.S.; Mongo, L.I.W.; Bamba, I.; Mama, A.; Toyi, M.; Lafortezza, R. Forest Fragmentation: Causes, Ecological Impacts and Implications for Landscape Management. In Landscape Ecology in Forest Management and Conservation: Challenges and Solutions for Global Change; Li, C., Lafortezza, R., Chen, J., Eds.; Springer: Berlin/Heidelberg, Germany, 2011; pp. 273–296. [Google Scholar] [CrossRef]
  70. Gelber, S.; Blowes, S.A.; Chase, J.M.; Huth, A.; Schurr, F.M.; Tietjen, B.; Zeller, J.W.; May, F. Geometric and demographic effects explain contrasting fragmentation-biodiversity relationships across scales. bioRxiv 2024. [Google Scholar] [CrossRef]
  71. Haddad, N.M.; Brudvig, L.A.; Clobert, J.; Davies, K.F.; Gonzalez, A.; Holt, R.D.; Lovejoy, T.E.; Sexton, J.O.; Austin, M.P.; Collins, C.D.; et al. Habitat fragmentation and its lasting impact on Earth’s ecosystems. Sci. Adv. 2015, 1, e1500052. [Google Scholar] [CrossRef]
  72. Fahrig, L. Effects of Habitat Fragmentation on Biodiversity. Annu. Rev. Ecol. Evol. Syst. 2003, 34, 487–515. [Google Scholar] [CrossRef]
  73. Laurance, W.F.; Nascimento, H.E.M.; Laurance, S.G.; Andrade, A.C.; Fearnside, P.M.; Ribeiro, J.E.L.; Capretz, R.L.; Fearnside, P.M. Rain Forest Fragmentation and the Proliferation of Successional Trees. Ecology 2006, 87, 469–482. [Google Scholar] [CrossRef]
  74. Riitters, K.H. Forest Fragmentation. In Forest Health Monitoring 2005 National Technical Report; Gen. Tech. Rep. SRS-104; U.S. Department of Agriculture’s Forest Service Southern Research Station: Asheville, NC, USA, 2007; Volume 104, pp. 9–15. [Google Scholar]
  75. Riitters, K.; Wickham, J.; O’Neill, R.; Jones, K.B.; Smith, E. Global-Scale Patterns of Forest Fragmentation. Conserv. Ecol. 2000, 4, 3. [Google Scholar] [CrossRef]
  76. Turner, M.G. Landscape Ecology in North America: Past, Present, and Future. Ecology 2005, 86, 1967–1974. [Google Scholar] [CrossRef]
Land 14 01537 g001
Figure 1. The area under study is demarcated by a thick black line, which corresponds to the border of the Oleszyce Forest Inspectorate (C). Forest stands managed by the State Forests are marked with a thin black line. The forested area is marked in green. Additionally, the Oleszyce Forest Inspectorate is marked in red on a map of Poland (B). The location of Poland in Europe is shown in green (A).
Figure 1. The area under study is demarcated by a thick black line, which corresponds to the border of the Oleszyce Forest Inspectorate (C). Forest stands managed by the State Forests are marked with a thin black line. The forested area is marked in green. Additionally, the Oleszyce Forest Inspectorate is marked in red on a map of Poland (B). The location of Poland in Europe is shown in green (A).
Land 14 01537 g001
Land 14 01537 g002
Figure 2. A flow chart describes the processing path of the research.
Figure 2. A flow chart describes the processing path of the research.
Land 14 01537 g002
Land 14 01537 g003
Figure 3. The importance of independent variables in the adopted models: (A) the effect of the variables in the basic model in descending order, (B) the effect of forest fragmentation at different scales and for multi-scale indicators ordered by scale.
Figure 3. The importance of independent variables in the adopted models: (A) the effect of the variables in the basic model in descending order, (B) the effect of forest fragmentation at different scales and for multi-scale indicators ordered by scale.
Land 14 01537 g003
Land 14 01537 g004
Figure 4. The partial effects of the FF index at different scales (50 m, 100 m, 200 m, 400 m and 600 m) on stand-level mortality probability of Scots pine forest.
Figure 4. The partial effects of the FF index at different scales (50 m, 100 m, 200 m, 400 m and 600 m) on stand-level mortality probability of Scots pine forest.
Land 14 01537 g004
Land 14 01537 g005
Figure 5. The partial effects of the FF index at different scales (800 m, 1000 m, 1200 m, and 1400 m) on the stand-level mortality probability of Scots pine forest.
Figure 5. The partial effects of the FF index at different scales (800 m, 1000 m, 1200 m, and 1400 m) on the stand-level mortality probability of Scots pine forest.
Land 14 01537 g005
Land 14 01537 g006
Figure 6. The partial effects of the FF index at different scales (1600 m, 1800 m, 2000 m, 2500 m, and 3000 m).
Figure 6. The partial effects of the FF index at different scales (1600 m, 1800 m, 2000 m, 2500 m, and 3000 m).
Land 14 01537 g006
Land 14 01537 g007
Figure 7. The partial effects of forest fragmentation index at multi-scales (50–600 m), (800–3000 m), and (50–3000 m) on stand level probability of Scots pine forest mortality.
Figure 7. The partial effects of forest fragmentation index at multi-scales (50–600 m), (800–3000 m), and (50–3000 m) on stand level probability of Scots pine forest mortality.
Land 14 01537 g007
Land 14 01537 g008
Figure 8. This is a spatial representation of the effect of multiscale (50–3000 m) FF on FM (A) and Scots pine FM probability (B). The gray color shows the total forested area of the Oleszyce Forest Inspectorate.
Figure 8. This is a spatial representation of the effect of multiscale (50–3000 m) FF on FM (A) and Scots pine FM probability (B). The gray color shows the total forested area of the Oleszyce Forest Inspectorate.
Land 14 01537 g008
Table 1. Basic characteristics of the Scots pine forest stand and its site conditions in the Oleszyce Forest Inspectorate. The data were taken from the NFI database, the digital terrain model, and meteorological data, which were interpolated for the entire area of Poland [49].
Table 1. Basic characteristics of the Scots pine forest stand and its site conditions in the Oleszyce Forest Inspectorate. The data were taken from the NFI database, the digital terrain model, and meteorological data, which were interpolated for the entire area of Poland [49].
CharacteristicsMinimumMaximumMeanInter Quartile Range
Area [ha]0.136.275.324.45
Mean Slope [degrees]0.135.071.610.83
Mean Altitude [m]195.29271.64236.2330.12
Age [year]21556658
Height [m]1362212
Diameter [cm]1603023
Volume [m3/ha]0.88638.94230.54188.83
Site index [m]3.9349.7632.355.09
Density [trees/ha]24,873.7565,255.4851,698.868588.03
AMT1 [°C]18.4521.9020.0120.84
AMT2 [°C]18.4521.0919.9120.74
AMT3 [°C]18.4521.0919.9520.74
AMR1 [mm]5.7352.9827.1737.38
AMR2 [mm]5.7350.1225.7034.52
AMR3 [mm]10.1450.1226.0434.48
SCWB1−0.951.04−0.090.65
SCWB2−0.951.690.040.90
SCWB3−0.951.69−0.060.62
Note: SCWB1—Standardized climate water balance one year before sanitation cutting; SCWB2—Standardized climate water balance two year before sanitation cutting; SCWB3—Standardized climate water balance three year before sanitation cutting; AMR1—Average vegetation period minimum precipitation one year before sanitation cutting; AMR2—Average vegetation period minimum precipitation two year before sanitation cutting; AMR3—Average vegetation period minimum precipitation three year before sanitation cutting; AMT1—Average vegetation period maximum temperature one year before sanitation cutting; AMT2—Average vegetation period maximum temperature two year before sanitation cutting; AMT3—Average vegetation period maximum temperature three year before sanitation cutting.
Table 2. Models’ performance evaluation metrics for the base model and the base model with different scales of forest fragmentation. p-values are thet average values of all predictors for the base model and the p-values for fragmentation scales are for each scale.
Table 2. Models’ performance evaluation metrics for the base model and the base model with different scales of forest fragmentation. p-values are thet average values of all predictors for the base model and the p-values for fragmentation scales are for each scale.
Model EvaluationBase ModelBase Model with Fragmentation at a Given Scale
Site and Stand 50 m Scale 100 m Scale 200 m Scale 400 m Scale 600 m Scale
AUC0.8180.8320.8340.8350.8350.835
Accuraccy0.7340.7540.7370.7450.7320.746
Kappa0.4670.5070.5060.4990.5050.492
p-value< 2 · 10 − 16 < 2 · 10 − 16 < 2 · 10 − 16 < 2 · 10 − 16 < 2 · 10 − 16 < 2 · 10 − 16
800 m scale1000 m scale1200 m scale1400 m scale1600 m scale1800 m scale
AUC0.8350.8350.8340.8340.8330.833
Accuraccy0.7530.7550.7580.7510.7550.752
Kappa0.5050.5090.5170.5030.5110.513
p-value< 2 · 10 − 16 < 2 · 10 − 16 6.18 · 10 − 3 5.94 · 10 − 1 2.19 · 10 − 5 < 2 · 10 − 16
2000 m scale2500 m scale3000 m scaleMulti 50–600 mMulti 800–3000 mMulti 50–3000 m
AUC0.8320.8320.8160.8190.8180.817
Accuraccy0.7430.7530.7570.7420.7440.744
Kappa0.4850.5110.4770.4840.4880.488
p-value< 2 · 10 − 16 < 2 · 10 − 16 < 2 · 10 − 16 4.86 · 10 − 6 < 2 · 10 − 16 < 2 · 10 − 16
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Feleha, D.D.; Netzel, P.; Talaga, J. The Relationship of Forest Fragmentation to Scots Pine Forest Mortality. Land 2025, 14, 1537. https://doi.org/10.3390/land14081537

AMA Style

Feleha DD, Netzel P, Talaga J. The Relationship of Forest Fragmentation to Scots Pine Forest Mortality. Land. 2025; 14(8):1537. https://doi.org/10.3390/land14081537

Chicago/Turabian Style

Feleha, Debebe Dana, Pawel Netzel, and Jakub Talaga. 2025. "The Relationship of Forest Fragmentation to Scots Pine Forest Mortality" Land 14, no. 8: 1537. https://doi.org/10.3390/land14081537

APA Style

Feleha, D. D., Netzel, P., & Talaga, J. (2025). The Relationship of Forest Fragmentation to Scots Pine Forest Mortality. Land, 14(8), 1537. https://doi.org/10.3390/land14081537

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop