Next Article in Journal
Water and Vegetation as a Source of UAV Forest Road Cross-Section Survey Error
Previous Article in Journal
Drought Driving Factors as Revealed by Geographic Detector Model and Random Forest in Yunnan, China
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Forests
Volume 16
Issue 3
10.3390/f16030506
forests-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

Assessing Ash (Fraxinus excelsior L.) Dieback Dynamics in the Białowieża Forest, Poland, Using Bi-Temporal High-Resolution Remote Sensing Data

by
Agnieszka Kamińska
1,
Maciej Lisiewicz
1,*,
Bartłomiej Kraszewski
1,
Miłosz Tkaczyk
2,
Krzysztof Stereńczak
1 and
Emilia Wysocka-Fijorek
1
1
Department of Geomatics, Forest Research Institute, Braci Leśnej 3 Street, Sękocin Stary, 05-090 Raszyn, Poland
2
Forest Protection Department, Forest Research Institute, Braci Leśnej 3 Street, Sękocin Stary, 05-090 Raszyn, Poland
*
Author to whom correspondence should be addressed.
Forests 2025, 16(3), 506; https://doi.org/10.3390/f16030506
Submission received: 5 February 2025 / Revised: 7 March 2025 / Accepted: 11 March 2025 / Published: 13 March 2025
(This article belongs to the Section Forest Inventory, Modeling and Remote Sensing)
Browse Figures
Versions Notes

Abstract

:
The ash dieback epidemic, caused by the fungus Hymenoscyphus fraxineus, has been a significant issue in Europe for over 20 years, severely affecting ash tree populations (Fraxinus excelsior L.). In the Białowieża Forest, ash trees now represent less than 1% of the species composition, with a sharp decline observed over the past several decades. This study aims to map the dynamics of ash mortality in the Białowieża Forest and assess the influence of habitat and stand factors on the severity of mortality. We utilized bi-temporal high-resolution remote sensing data from 2015 to 2019 to track the decline of ash trees and to identify factors affecting mortality. The analysis employed a combination of Boosted Regression Trees (BRTs) and hotspot analyses. Our results show that between 2015 and 2019, 29% of the living ash trees in the canopy layer of the forest died. The findings indicate that ash mortality was most pronounced in stands with a high proportion of ash trees, particularly where dead deciduous trees were already present. Intensive dieback of other deciduous trees was also noted in these stands. This study provides valuable insights into the factors influencing ash mortality dynamics and demonstrates the potential of remote sensing for large-scale monitoring of tree health. The results have important implications for forest management and conservation, offering baseline data that can inform strategies to manage ash dieback and guide targeted interventions in affected forest areas.

1. Introduction

Tree dieback has become a critical environmental challenge, potentially leading to the mortality of individual trees or entire forest ecosystems [1]. Several factors contribute to tree dieback, including pests, diseases, climate change, pollution, and deforestation [2]. These factors can weaken trees, making them more susceptible to further damage and ultimately causing their demise. Tree dieback has severe ecological consequences, as it affects biodiversity, disrupts ecosystems, and reduces the availability of essential resources such as oxygen and habitats for wildlife [3].
The emergence of ash dieback is seen as a global challenge since many countries and regions are facing diseases affecting their native tree species due to changing climate. The invasive nature of pathogens like Hymenoscyphus fraxineus (initially linked to East Asia [4]) showcases how such diseases can adapt and spread globally, impacting ecosystems and biodiversity. In the context of Europe, the decline of European ash (Fraxinus excelsior L.) has been particularly alarming since the first symptoms of ash dieback were documented in the early 1990s [5,6,7,8]. The problem was first noted in northeastern Poland and, over subsequent years, spread to adjacent regions [9,10,11]. Today, this decline is observed across most of Europe [12,13,14]. The primary culprit is the mentioned fungus Hymenoscyphus fraxineus (formerly known as Chalara fraxinea), which causes leaf wilt and defoliation, crown thinning, darkening of wood, necrosis, tumor-like growths on trunks and branches, and ultimately tree death [4,15]. Other pathogens, for instance, the fungus Rigidoporus ulmarius [16], may further weaken these trees. Pathogens of the genus Phytophthora, which are responsible for the death of root systems, also contribute to the dieback of ash trees [17]. This poses a particular threat as ash trees develop best in moist habitats, where the threat from these pathogens is particularly high.
Regardless of the region, the widespread decline of ash trees has resulted in the loss of over 90% of Fraxinus excelsior populations across its entire natural range [18,19,20]. The vulnerability of ash to this disease is increasing due to the intensifying disturbances in soil water relations, which are linked to water shortages caused by climate change—namely, rising temperatures and decreasing precipitation [2,4,15]. These conditions further hinder ash recovery efforts, making regeneration difficult and, in some cases, impossible [21]. Although certain ash populations may show resistance to infections by Hymenoscyphus fraxineus [22], offering a potential avenue for recovery, the observed regeneration has not produced the anticipated outcomes [21].
Regionally, the Białowieża Forest, one of the last and largest remaining parts of the primeval forest that once covered much of Europe, has witnessed ash mortality rates as high as 99% over the last twenty-six years [18]. A study by Cholewińska et al. [18] shows that the highest ash mortality is observed in younger stands. The habitat in which ash trees grow is also important. Fraxinus excelsior is a species with high soil requirements that needs fertile, moist sites with a neutral pH value [23]. In the Białowieża Forest, significant differences in ash mortality have been observed across various habitats. Ash trees thrive in their typical habitats, such as alder carr and mixed deciduous forests, but are less resilient in less humid deciduous forests [18].
In light of these alarming mortality rates and their variation across different habitats, there is a growing need for comprehensive and efficient monitoring approaches. Remote sensing techniques have become increasingly prevalent in a broad spectrum of forest management and ecological applications, ranging from monitoring land-cover changes to mapping individual trees [24,25]. By combining high-resolution remote sensing data with robust statistical methods, it is possible to accurately identify tree species along with their health status (alive or dead) [26,27] and to pinpoint factors that influence the spatial dynamics of tree mortality over specific time intervals [28,29,30]. However, studies focusing on ash dieback within the context of remote sensing (particularly for forest dieback detection, mapping, and monitoring) remain limited compared to research on other forest species, even when using commonly available satellite data [31]. Fraxinus excelsior is prevalent in highly mixed and heterogeneous forest environments, where individual species are typically separated using crown segmentation procedures and pre-classified by species [32].
To effectively minimize damage during outbreaks, reliable models capable of predicting areas at risk of tree mortality are necessary. Such insights are crucial for preserving optimal forest health. Accurate mortality modeling should take into account stand structure, environmental factors, and variations in topography [33]. According to the authors’ knowledge, there are no available publications that examine the role of habitat and stand factors in determining the severity of ash mortality using remote sensing data.
The objectives of this study were as follows:
  • To map the dynamics of ash mortality using bi-temporal high-resolution remote sensing data;
  • To assess the influence of habitat and stand factors on the severity of ash mortality.
This study focused on understanding the spatial dynamics and factors of ash mortality in the Białowieża Forest in Poland, by leveraging bi-temporal high-resolution remote sensing data from 2015 and 2019. We used data on individual trees classified by species and health status according to the method described by Lisiewicz et al. [34], in combination with information from CIR images acquired in 2019, to investigate the spatial distribution of ash mortality over a four-year period using a 1 ha grid cell framework. Our analysis employed two approaches: spatial hotspot analysis using global and local Moran’s coefficients, and a machine learning technique, Boosted Regression Trees, which have proven effective in our earlier studies [28,29].

2. Materials and Methods

2.1. Study Area

Located on the border of Poland and Belarus (coordinates: 52°45′29″ N, 23°46′8″ E), the Białowieża Forest (BF) is one of the largest and last remaining examples of ancient European lowland forests. Renowned for its rich biodiversity, the BF boasts a diverse array of tree species. The Polish part of the BF spans approximately 62,000 hectares, which includes 10,500 hectares within the Białowieża National Park (BNP), 12,000 hectares designated as nature reserves, and 39,500 hectares classified as managed forests. Over 35% of this area is protected through the BNP and nature reserves, while the remainder consists of managed forests. These managed parts fall under the administration of three forest districts: Białowieża, Browsk, and Hajnówka (Figure 1). The terrain is predominantly flat, with only minor variations in relative altitude (131.6–195.6 m a.s.l.). The BF experiences a continental climate influenced by Atlantic air masses. The mean annual temperature is 6.8 °C, and the average annual precipitation is about 633 mm. Soils are largely sandy or loamy, with higher fertility observed in riverine zones.
Renowned for its exceptional biodiversity, the BF harbors a wide array of tree species. According to 2019 field measurements, the forest is dominated by common hornbeam (Carpinus betulus), Norway spruce (Picea abies), black alder (Alnus glutinosa), Scots pine (Pinus sylvestris), and birch (Betula spp.). The BF showcases a diverse composition, featuring towering conifers (e.g., Picea abies, Pinus sylvestris) and broadleaved giants such as oaks, ash, maple, and lime. The Białowieża Forest’s biodiversity is linked to its complex structure, which includes both undisturbed and managed areas. The forest also plays a crucial socio-economic role, supporting local communities through sustainable forestry, tourism, and agriculture. Forestry has been a traditional practice in the region for centuries. The Białowieża National Park and nearby reserves boost the local economy by attracting tourists, researchers, and nature enthusiasts. The forest is also a key site for scientific research on forest ecosystems, climate change, and biodiversity conservation.

2.2. Remote Sensing Datasets

2.2.1. Airborne Laser Scanning and CIR Imagery

In this study, data were sourced from two airborne laser scanning (ALS) datasets, one from the leaf-on season and another from the leaf-off season, along with a leaf-on color infrared (CIR) dataset from 2015. In 2019, one ALS dataset and a color infrared dataset were acquired during the leaf-on season. The choice of datasets from 2015 and 2019 allowed us to capture the seasonal decline of ash, which in turn enabled us to conduct a detailed analysis of dieback dynamics. In 2015, the ALS dataset was collected using the Riegl LMS-Q680i system (Riegl LMS GmbH, Horn, Austria). Data collection was conducted at an altitude of 500 m above ground level (AGL): for the leaf-on season, measurements were taken from 2 to 5 July 2015; for the leaf-off season, data were gathered on 25 November, 27 November, and 6–7 December 2015. A total of 135 flight lines were executed with a 40% overlap to guarantee complete coverage of the study area. The average point cloud densities were 11 points per square meter. The ALS dataset from 2019 was obtained using the Riegl VQ-780i system (Riegl LMS GmbH, Horn, Austria). Data collection took place on 3–6 and 23 August 2019, at an altitude of 685 m AGL. This dataset featured an average point cloud density of 19 points per square meter, with 88 individual flight lines performed and a 20% overlap. Both datasets were used to produce canopy height models with resolution of 0.5 m, on the basis of which individual tree detection was performed.
Between 2 and 5 July 2015, CIR aerial images were captured alongside the ALS datasets using an UltraCam Eagle camera (Vexcel GmbH, Graz, Austria) at an altitude of 3040 m AGL. These images had a GSD of 0.20 m and featured 90% forward overlap and 40% side overlap. In 2019, CIR aerial images were acquired for over 90% of the study area on 17 August using the same UltraCam Eagle camera, but from an altitude of 3420 m AGL, with the same GSD and 80% forward and 70% side overlaps. For the remaining area, images were taken on 11 September 2019, using a DMC III camera (Leica, Wetzlar, Germany) from an altitude of 4880 m AGL, maintaining the same parameters as the UltraCam Eagle. RGB values extracted from the CIR images were then attributed to both point clouds.

2.2.2. RS Data Processing

For the 2015 dataset, individual tree detection was carried out using the CHM-based method described in Stereńczak et al. [35]. Each tree was subsequently classified by species and health status (living or dead) based on ALS structural and intensity features, as well as the spectral characteristics from each band of the CIR aerial images [27,34]. A total of 11 living tree species were classified, including alder, ash, aspen, birch, hornbeam, lime, maple, oak, pine, spruce, and other deciduous trees, along with four dead categories: dead deciduous, dead pine, dead spruce, and snag. High values of overall accuracy (OA) were achieved in the classification process, with results of 82%. A total of 1972 reference trees were used. For validation, repeated 5-fold cross-validation (100 iterations) was employed, and averaged classification metrics (overall accuracy, user’s accuracy, producer’s accuracy, F1-score) were derived. The F1-score of ash was equal to 0.46; however, UA reached a value of 0.63. Using a dataset from 2019, a Random Forest [36] method was employed to reclassify tree species from a 2015 classification by distinguishing between living and dead trees, yielding an OA of 98.6%. To account for the natural dynamics in the Białowieża Forest stands, where some dead trees had fallen or been removed in managed areas, a method was developed to identify fallen or felled trees. This approach employed Classification Trees [37], identifying the median height difference between polygons derived from the 2015 and 2019 CHMs, specifically calculated for pixels one pixel away from the centroid, as the most influential variable. In the model, applying a threshold of 5.5 m for the median height difference between the two CHMs contributed to achieving high overall accuracy values of 97.0%. Trees identified as fallen during the data acquisition period were considered dead.

2.3. Statistical Analyses

The study area was segmented into 100 m grid cells [28,29,38], and changes in ash mortality over four years were analyzed using spatial modeling. Ash mortality was determined by counting the number of trees that died within each raster cell during the period from 2015 to 2019. The analysis examined response variables in relation to focal cell predictors, which included forest stand characteristics, site conditions, and individual tree-related factors (Table 1). Stand-related attributes were obtained from the Polish State Forests database (SILP) and the forest digital map to describe specific habitat and stand features for each cell. In total, this study included 33,488 cells containing ash trees, which were analyzed to assess ash mortality dynamics.
For this analysis, we resampled the 2015 DEM (originally at a 0.5 m resolution) to 100 m resolution by averaging the cell values. This resampling was conducted to eliminate artifacts and noise associated with small topographic features exhibiting significant height variations. The resulting DEM served as an input layer for calculating topographic characteristics, including the Topographic Position Index (TPI). The TPI is determined by computing the difference between the actual elevation of each cell and the average elevation within a specified surrounding area, in this case, a 5000 m radius. The TPI raster was then categorized into three classes: −1, 0, and 1, representing areas that are lower, flat, or higher relative to the mean elevation of the surrounding landscape, respectively. Slope, representing the maximum rate of elevation change between a cell and its neighboring cells, and aspect, which indicates the downslope direction of this maximum change, were calculated for eight directional categories (N, N-E, E, S-E, S, S-W, W, N-W). All mentioned topographic characteristics were computed using R [39].
To examine selected habitat and stand attributes, we utilized stand-related characteristics from the Polish State Forests database (SILP) along with data from the forest digital map. These variables delineate forest stands (alternatively referred to as subcompartments), which serve as the primary administrative units used by the Polish State Forest Administration. The forest stand characteristics were mapped onto rectangular grid cells (1 ha each) covering the entire study area, with each grid cell being assigned the characteristics of the forest subcompartment that occupied the largest area within it. A complete list of predictors is provided in Table 1.

2.3.1. Hotspot Analysis

Spatial patterns were assessed using both global and local Moran’s I coefficients [40]. The global Moran’s I measures the degree of positive or negative spatial autocorrelation across the entire study area. At the local level, the local Moran’s I (LISA) values were calculated for each location to determine its similarity with neighboring areas and to evaluate statistical significance. Five distinct LISA categories were identified [41]:
  • Locations with no significant local autocorrelation (class 0);
  • High–high locations with high values and similar neighbors (class 1);
  • Low–low locations with low values and similar neighbors (class 2);
  • Low–high locations with low values and high-value neighbors (class 3);
  • High–low locations with high values and low-value neighbors (class 4).
The significance of the global and local Moran’s I coefficients was assessed using the “conditional permutation” method [40]. Spatial cluster analysis and the identification of spatial outliers were conducted using GeoDa (version 1.12; [42]). A significance level of α = 0.01 was applied to all statistical tests.

2.3.2. Boosted Regression Tree Analyses

Boosted Regression Trees (BRTs) [43] were used to identify key factors influencing ash dieback dynamics. The BRT models were trained using 70% of the dataset and tested on the remaining 30%. The model with the highest cross-validation (CV) correlation was selected as the best-performing model. Each predictor’s relative importance was calculated (summing to 100% across all variables) to indicate its contribution to the model. Partial dependence functions were used to visualize the fitted function, showing the effect of each variable on the response while accounting for the average influence of all other predictors [43].
The models were implemented in R version 4.4.0 [39] using the gbm package [44,45] with a Gaussian response distribution and a 10-fold cross-validation procedure. Each model was configured with a tree complexity of 5 (number of nodes), a learning rate of 0.05, and a bag fraction of 0.75 (the proportion of data sampled at each iteration). A flowchart illustrating the processing steps is shown in Figure 2.

3. Results

3.1. Ash Mortality

In 2015, a total of 145,069 living ash trees were identified in the canopy layer. Over the four-year period from 2015 to 2019 in the Białowieża Forest (BF), 41,565 ashes perished, accounting for 29% of the total population. The highest mortality rate was observed in the Białowieża Forest District at 32%, while the lowest rate was recorded in the Browsk Forest District at 24% (Table 2). Figure 3 presents maps illustrating the spatial distribution of ash tree mortality over the four-year period.

3.2. HotSpot Analysis

Ash mortality in the Białowieża Forest (BF) from 2015 to 2019 exhibited a clustering pattern. The global Moran’s I value was 0.42, indicating a significant positive spatial autocorrelation (pseudo p-value < 0.001). Hotspot analysis identified various LISA classes (0, 1, 3, and 4), which are illustrated in Figure 3.
Many hotspots were identified (2808 units). Most “high–high” cells were located in the middle part of Białowieża Forest area, most in BNP (1242 units), followed by Białowieża F.D. (845 units). In turn, the fewest hotspots were located in Hajnówka F.D. (581 units). Locations with high values and similar neighbors accounted for 4.5% of the BF and were situated in deciduous stands. In these areas, an average of five ash trees died during the analyzed 4-year period. In addition, the highest number and mortality of other deciduous trees, and conversely, the lowest number and mortality of conifers, were recorded in these locations. Additionally, among all classes, hotspots exhibited the highest average values for the number of living ash trees and dead deciduous trees (Table 3).
Cold clusters were not depicted, but spatial outliers were observed (class 3—448 units; class 4—911 units). The locations with high values but low-value neighbors (class 4) were spread over all study area (Browsk F.D.—340 units; Hajnówka F.D.—337 units; Białowieża F.D.—132 units; BNP—102 units). Spatial autocorrelation techniques allowed the depiction of single dead ash trees. Unlike hot cluster locations, the lowest number and mortality of other deciduous trees, and the highest number and mortality of conifers, were recorded in these locations. Additionally, these locations were characterized by the lowest average values of the number of alive ash and dead deciduous trees (Table 3).
Locations with low values but surrounded by high-value neighbors were situated near hot clusters (Browsk F.D.—26 units; Hajnówka F.D.—92 units; Białowieża F.D.—138 units; BNP—192 units), in deciduous stands with many living deciduous trees. In these areas, ash tree mortality between 2015 and 2019 was minimal, and the mortality of other tree species was also low.
An interaction plot was analyzed to examine the number of dead ash trees per 1-hectare cell from 2015 to 2019 across LISA classes identified through hotspot analysis and the ’dominant species’ category. Regardless of the dominant species, the high–high LISA classes showed the highest response values, followed by the low–high class. The lowest values were observed in the low–low class, where no ash mortality was recorded. The highest number of dead ash trees per 1 ha cell was observed in locations dominated by ash (mean = 6.29 ± 4.69), and the lowest in locations dominated by birch (mean = 3.16 ± 2.38). “Low–high” locations were not detected in ash, lime, and maple locations (Figure 4).
In summary, the hotspot analyses demonstrated that environmental and stand-related factors influenced the ash mortality. The species composition of the stands was a key factor affecting ash decline during the analyzed period. Mortality increased in areas with a high number of ash, particularly in the presence of already dead deciduous trees. Intensive dieback of other deciduous trees was also observed in these stands.

3.3. BRT Analyses

The results of the BRT analysis indicated that the number of living ash trees was the primary factor influencing the response, with a relative importance of nearly 80%. The second most important factor was mortality of other deciduous species (5%) (Table 4). Other variables had minimal impact (<3%). The BRT model explained 87% of the randomly selected input data and predicted 82% of the excluded data.
The fitted functions in the BRT model were visualized using partial dependence functions, which demonstrate the influence of a variable on the response while controlling for the average effects of other variables (Figure 5). These functions revealed that the response increased with the number of living ash trees. A positive correlation between ash mortality and other deciduous mortality was also observed. The intensity of ash dieback appeared similar across all forest districts and the BNP. As anticipated, topographic factors played a minor role in the analysis.

4. Discussion

The removal of ash from the species composition of forest stands is not unique to the Białowieża Forest. Large-scale dieback and widespread decline of ash trees in forest ecosystems were first observed in Poland and across Europe in the early 20th century [46]. This phenomenon has persisted and now affects many European countries [47]. The current state of affairs is undoubtedly influenced by recent extreme weather phenomena that weaken trees, which in turn makes them more susceptible to infection [48]. Ash health depends on various biotic and abiotic factors. Key biotic factors include habitat conditions and thus the stand’s species composition and the humidity and age of the ash trees [23]. The most critical abiotic factors include infections by pathogens: Hymenoscyphus fraxineus and those pertaining to the genus Phytophthora [6,17,47,49].
Over the past several decades, there has been a sharp decline in the number of ash trees in the Białowieża Forest. The available data do not allow for pinpointing the exact time when the initial symptoms of dieback, such as leaf wilting and partial necrosis, were first observed in this region. In his study, Paluch [50] noted a significant reduction in the number of ash trees in the research plots from 1997 to 2012, and in some of the study plots (about 20%), ash trees have entirely disappeared from the stand species composition.
Currently, there are very few ash trees in the forest, constituting less than 1% of the species composition of the Białowieża Forest [34]. Our findings indicate that the dieback process continues, with 29% of ash trees in the canopy layer dying during the study period (2015–2019). The intensity of ash dieback was similar across all forest districts and the BNP.
Our results confirm that both host and environmental factors affect the timing of ash mortality. The species composition of ash stands, which largely reflects habitat conditions, plays an important role in ash mortality. Previous studies [51] have found that ash mortality was lower in poor habitats and in stands with lower ash tree density. In our study, both the BRT and hotspot analyses indicated that the stand’s species composition was the primary factor influencing ash mortality during the four-year outbreak period (2015–2019) in the Białowieża Forest. Mortality increased in stands with a high number of ash trees, especially in the presence of already dead deciduous trees, many of which were likely dead ash trees. Intensive dieback of other deciduous trees was also observed in these stands.
These findings align with previous studies indicating that ash mortality is higher in fertile, humid habitats closer to the species’ ecological optimum [52,53]. Fluctuations in groundwater levels also have a significant impact on ash mortality [54]. It should also be noted that the Białowieża Forest suffered from extreme drought and heat during the period studied and before, which may have contributed to the increased mortality [55].
The analysis of the impact of age and height classes on ash mortality is primarily based on ground-based data and sampling part of the ash population [23,53,54]. Although ash dieback impacts trees of all ages, young trees and saplings are especially vulnerable [56,57,58]. Older trees affected by H. fraxineus can repeatedly regenerate substantial portions of their crowns. However, the presence of an additional pathogen can markedly speed up the progression of ash dieback, potentially resulting in tree mortality [10,56,59]. Comparative studies of disease progression in ash stands of different ages over several years [60] have shown that older ash trees show signs of serious weakening and increased mortality. In our study, the age of the trees had no effect on ash tree mortality.
In our study, remote sensing data with hotspot and BRT analysis has proven to be a valuable tool in assessing ash mortality on a large forest complex of unique significance. So far, various studies have utilized remote sensing technologies like UAV-derived imagery, WorldView-2 satellite imagery, and hyperspectral remote sensing to effectively detect, classify, and monitor the health of ash trees across large geographic areas. For instance, Flynn et al. [61] highlighted the potential of UAV-based greenness measurements and spatial patterning analysis for detecting ash dieback, demonstrating that post-peak greenness assessments allow for accurate identification of disease symptoms. Similarly, analysis of WorldView-2 imagery enabled the classification of tree species and quantification of ash mortality levels, achieving promising classification accuracies of 83% for tree species and 73% for varying degrees of ash damage [62]. Moreover, hyperspectral imagery has facilitated the mapping of ash dieback distribution over extensive forested landscapes [63]. In our study, we employed single-tree classification based on airborne laser scanning data to identify and assess ash trees individually, enabling a more detailed evaluation in the context of spatial mortality patterns across the forest.
However, the method we used to detect ash tree mortality has certain limitations. The high species diversity within the stands and the low proportion of ash trees resulted in moderate accuracy in recognizing and classifying individual trees, making the classification of ash less satisfactory than desired. This limitation was primarily due to the challenges of distinguishing ash trees from other deciduous species in mixed-species stands using remote sensing data.
Despite these limitations, this study has significant strengths. It provided a comprehensive analysis of the entire study area, offering valuable insights into the condition and spatial distribution of the entire ash population. By leveraging bi-temporal high-resolution remote sensing data, the methodology proved effective for assessing large-scale tree mortality dynamics. This approach holds substantial potential for application in other regions, enabling the creation of operational risk maps for tree mortality in extensive forested areas. Such maps could play a crucial role in guiding forest management and conservation strategies, offering a practical tool for identifying vulnerable areas and implementing targeted interventions.

5. Conclusions

This study investigated the dynamics of ash dieback in the Białowieża Forest using bi-temporal high-resolution remote sensing data. Although ash trees currently make up less than 1% of the species composition, the dieback of this species continues. Between 2015 and 2019, 29% of the living ash trees in the forest canopy died. The methods and results presented in this study enabled the mapping of ash mortality dynamics, as well as the evaluation of the influence of habitat and stand factors on the severity of ash mortality.
During the four-year outbreak period (2015–2019), ash mortality in the Białowieża Forest was strongly influenced by the species composition of the stands. Mortality was higher in stands with a greater number of ash trees, particularly when other deciduous trees, many of which were likely dead ash trees, were also present. These stands also exhibited significant dieback in other deciduous species.
Overall, our results provide baseline information for forest management and conservation by enabling the creation of operational risk maps and targeted intervention strategies. They also underscore the complex interplay between biotic and abiotic factors in ash dieback dynamics. While our approach proved effective for individual tree analysis, limitations remain, particularly in differentiating closely related species within heterogeneous stands, which suggests that future research should incorporate additional remote sensing data from different periods to improve classification accuracy. Moreover, future studies should extend this approach to other regions and consider a broader range of disturbance factors to further validate and enhance predictive models, ultimately contributing to more effective forest health monitoring and management practices.

Author Contributions

Conceptualization, A.K. and M.L.; Methodology, A.K.; Validation, A.K. and M.L.; Formal Analysis, A.K.; Data Curation, M.L.; Writing—Original Draft Preparation, A.K. and M.L.; Writing—Review & Editing, B.K., E.W.-F., M.T. and K.S.; Visualization, A.K., M.L. and B.K.; Project Administration, A.K.; Funding Acquisition, A.K. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the project titled ‘Comprehensive spatial analysis of the dieback of dominant tree species using multi−temporal ALS data and CIR imagery in the Białowieża Primeval Forest’ and carried out by the Forest Research Institute from 2021–2024 (internal No. 901510). Data were funded through the project ‘LIFE+ ForBioSensing PL Comprehensive monitoring of stand dynamics in Białowieża Forest supported with remote sensing techniques’, which is co-funded by the EU Life Plus program (contract number LIFE13 ENV/PL/000048) and The National Fund for Environmental Protection and Water Management in Poland (contract number 485/2014/WN10/OP−NM−LF/D).

Data Availability Statement

The data presented in this study are available on request. The data underlying this article are stored in the repository of the Forest Research Institute.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Anderegg, W.R.L.; Martinez-Vilalta, J.; Cailleret, M.; Camarero, J.J.; Ewers, B.E.; Galbraith, D.; Gessler, A.; Grote, R.; Huang, C.; Levick, S.R.; et al. When a Tree Dies in the Forest: Scaling Climate-Driven Tree Mortality to Ecosystem Water and Carbon Fluxes. Ecosystems 2016, 19, 1133–1147. [Google Scholar] [CrossRef]
  2. Allen, C.D.; Macalady, A.K.; Chenchouni, H.; Bachelet, D.; McDowell, N.; Vennetier, M.; Kitzberger, T.; Rigling, A.; Breshears, D.D.; Hogg, E.H.; 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]
  3. Anderegg, W.R.L.; Kane, J.M.; Anderegg, L.D.L. Consequences of Widespread Tree Mortality Triggered by Drought and Temperature Stress. Nat. Clim. Change 2013, 3, 30–36. [Google Scholar] [CrossRef]
  4. Gross, A.; Hosoya, T.; Queloz, V. Population Structure of the Invasive Forest Pathogen Hymenoscyphus pseudoalbidus. Mol. Ecol. 2014, 23, 2943–2960. [Google Scholar] [CrossRef]
  5. Carroll, D.; Boa, E. Ash Dieback: From Asia to Europe. Plant Pathol. 2024, 73, 741–759. [Google Scholar] [CrossRef]
  6. Timmermann, V.; Børja, I.; Hietala, A.M.; Kirisits, T.; Solheim, H. Ash Dieback: Pathogen Spread and Diurnal Patterns of Ascospore Dispersal, with Special Emphasis on Norway. EPPO Bull. 2011, 41, 14–20. [Google Scholar] [CrossRef]
  7. Kowalski, T. O Zamieraniu Jesionów. Trybunał Leśnika 2001, 4, 6–7. [Google Scholar]
  8. Stocki, J. Przyczyny Zamierania Drzew w i Drzewostanów Jesionowych w Polsce. Głos Lasu 2001, 4, 17–19. [Google Scholar]
  9. Pautasso, M.; Aas, G.; Queloz, V.; Holdenrieder, O. European Ash (Fraxinus excelsior) Dieback—A Conservation Biology Challenge. Biol. Conserv. 2013, 158, 37–49. [Google Scholar] [CrossRef]
  10. Bakys, R.; Vasaitis, R.; Barklund, P.; Thomsen, I.M.; Stenlid, J. Occurrence and Pathogenicity of Fungi in Necrotic and Non-Symptomatic Shoots of Declining Common Ash (Fraxinus excelsior) in Sweden. Eur. J. For. Res. 2009, 128, 51–60. [Google Scholar] [CrossRef]
  11. Przybył, K. Fungi Associated with Necrotic Apical Parts of Fraxinus excelsior Shoots. For. Pathol. 2002, 32, 387–394. [Google Scholar] [CrossRef]
  12. Solheim, H.; Hietala, A.M. Spread of Ash Dieback in Norway. Balt. For. 2017, 23, 144–149. [Google Scholar]
  13. Clark, J.; Webber, J. The Ash Resource and the Response to Ash Dieback in Great Britain. In Dieback of European Ash (Fraxinus spp.): Consequences and Guidelines for Sustainable Management; Vasaitis, R., Enderle, R., Eds.; Swedish University of Agricultural Sciences: Uppsala, Sweden, 2017; pp. 228–237. [Google Scholar]
  14. McCracken, A.R.; Douglas, G.C.; Ryan, C.; Destefanis, M.; Cooke, L.R. Ash Dieback on the Island of Ireland. In Dieback of European Ash (Fraxinus spp.): Consequences and Guidelines for Sustainable Management; Vasaitis, R., Enderle, R., Eds.; Swedish University of Agricultural Sciences: Uppsala, Sweden, 2017; pp. 125–139. [Google Scholar]
  15. Baral, H.-O.; Queloz, V.; Hosoya, T. Hymenoscyphus fraxineus, the Correct Scientific Name for the Fungus Causing Ash Dieback in Europe. IMA Fungus 2014, 5, 79–80. [Google Scholar] [CrossRef] [PubMed]
  16. Lücking, R.; Moncada, B.; Hawksworth, D.L. Gone with the Wind: Sequencing Its Type Species Supports Inclusion of Cryptolechia in Gyalecta (Ostropales: Gyalectaceae). Lichenologist 2019, 51, 287–299. [Google Scholar] [CrossRef]
  17. Tkaczyk, M.; Nowakowska, J.A.; Oszako, T. Phytophthora Species Isolated from Ash Stands in Białowieża Forest Nature Reserve. For. Pathol. 2016, 46, 660–662. [Google Scholar] [CrossRef]
  18. Cholewińska, O.; Keczyński, A.; Smerczyński, I.; Jaroszewicz, B. European Ash (Fraxinus excelsior L.) Dieback in a Core Area of Białowieża National Park. Natl. Park. Nat. Reserv. 2018, 37, 3–18. [Google Scholar]
  19. Matisone, I.; Matisons, R.; Laiviņš, M.; Gaitnieks, T. Statistics of Ash Dieback in Latvia. Silva Fenn. 2018, 52, 9901. [Google Scholar] [CrossRef]
  20. Skovsgaard, J.P.; Thomsen, I.M.; Skovgaard, I.M.; Martinussen, T. Associations among Symptoms of Dieback in Even-Aged Stands of Ash (Fraxinus excelsior L.). For. Pathol. 2010, 40, 7–18. [Google Scholar] [CrossRef]
  21. Enderle, R.; Metzler, B.; Riemer, U.; Kändler, G. Ash Dieback on Sample Points of the National Forest Inventory in South-Western Germany. Forests 2018, 9, 25. [Google Scholar] [CrossRef]
  22. Pušpure, I.; Matisons, R.; Laiviņš, M.; Gaitnieks, T.; Jansons, J. Natural Regeneration of Common Ash in Young Stands in Latvia. Balt. For. 2017, 23, 209–217. [Google Scholar]
  23. Dobrowolska, D.; Hein, S.; Oosterbaan, A.; Wagner, S.; Clark, J.; Skovsgaard, J.P. A Review of European Ash (Fraxinus excelsior L.): Implications for Silviculture. For. Int. J. For. Res. 2011, 84, 133–148. [Google Scholar] [CrossRef]
  24. Fassnacht, F.E.; Latifi, H.; Stereńczak, K.; Modzelewska, A.; Lefsky, M.; Waser, L.T.; Straub, C.; Ghosh, A. Review of Studies on Tree Species Classification from Remotely Sensed Data. Remote Sens. Environ. 2016, 186, 64–87. [Google Scholar] [CrossRef]
  25. Hansen, M.C.; Loveland, T.R. A Review of Large Area Monitoring of Land Cover Change Using Landsat Data. Remote Sens. Environ. 2012, 122, 66–74. [Google Scholar] [CrossRef]
  26. Kamińska, A.; Lisiewicz, M.; Stereńczak, K. Single Tree Classification Using Multi-Temporal ALS Data and CIR Imagery in Mixed Old-Growth Forest in Poland. Remote Sens. 2021, 13, 5101. [Google Scholar] [CrossRef]
  27. Kamińska, A.; Lisiewicz, M.; Stereńczak, K.; Kraszewski, B.; Sadkowski, R. Species-Related Single Dead Tree Detection Using Multi-Temporal ALS Data and CIR Imagery. Remote Sens. Environ. 2018, 219, 31–43. [Google Scholar] [CrossRef]
  28. Kamińska, A.; Lisiewicz, M.; Kraszewski, B.; Stereńczak, K. Habitat and Stand Factors Related to Spatial Dynamics of Norway Spruce Dieback Driven by Ips typographus (L.) in the Białowieża Forest District. For. Ecol. Manag. 2020, 476, 118432. [Google Scholar] [CrossRef]
  29. Kamińska, A.; Lisiewicz, M.; Kraszewski, B.; Stereńczak, K. Mass Outbreaks and Factors Related to the Spatial Dynamics of Spruce Bark Beetle (Ips typographus) Dieback Considering Diverse Management Regimes in the Białowieża Forest. For. Ecol. Manag. 2021, 498, 119530. [Google Scholar] [CrossRef]
  30. Grodzki, W.; Starzyk, J.R.; Kosibowicz, M. Impact of Selected Stand Characteristics on the Occurrence of the Bark Beetle Ips typographus (L.) in the Beskid Żywiecki Mountains. For. Res. Pap. 2014, 75, 159–169. [Google Scholar] [CrossRef]
  31. Gašparović, M.; Pilaš, I.; Klobučar, D.; Gašparović, I. Monitoring Ash Dieback in Europe—An Unrevealed Perspective for Remote Sensing? Remote Sens. 2023, 15, 1178. [Google Scholar] [CrossRef]
  32. Polk, S.L.; Chan, A.H.Y.; Cui, K.; Plemmons, R.J.; Coomes, D.A.; Murphy, J.M. Unsupervised Detection of ASH Dieback Disease (Hymenoscyphus fraxineus) Using Diffusion-Based Hyperspectral Image Clustering. In Proceedings of the IGARSS 2022—2022 IEEE International Geoscience and Remote Sensing Symposium, Kuala Lumpur, Malaysia, 17–22 July 2022; pp. 2287–2290. [Google Scholar]
  33. Sproull, G.J.; Adamus, M.; Szewczyk, J.; Kersten, G.; Szwagrzyk, J. Fine-Scale Spruce Mortality Dynamics Driven by Bark Beetle Disturbance in Babia Góra National Park, Poland. Eur. J. For. Res. 2016, 135, 507–517. [Google Scholar] [CrossRef]
  34. Lisiewicz, M.; Kamińska, A.; Kraszewski, B.; Kuberski, Ł.; Pilch, K.; Stereńczak, K. Comprehensive Mapping of Individual Living and Dead Tree Species Using Leaf-On and Leaf-Off ALS and CIR Data in a Complex Temperate Forest. For. Int. J. For. Res. 2025, cpaf007. [Google Scholar] [CrossRef]
  35. Stereńczak, K.; Kraszewski, B.; Mielcarek, M.; Piasecka, Ż.; Lisiewicz, M.; Heurich, M. Mapping Individual Trees with Airborne Laser Scanning Data in an European Lowland Forest Using a Self-Calibration Algorithm. Int. J. Appl. Earth Obs. Geoinf. 2020, 93, 102191. [Google Scholar] [CrossRef]
  36. Breiman, L. Random Forests. Mach. Learn. 2001, 45, 5–32. [Google Scholar] [CrossRef]
  37. Breiman, L.; Friedman, J.H.; Olshen, R.A.; Stone, C.J. Classification and Regression Trees, 1st ed.; Chapman and Hall/CRC: Boca Raton, FL, USA, 1984. [Google Scholar]
  38. Kärvemo, S.; Van Boeckel, T.P.; Gilbert, M.; Grégoire, J.C.; Schroeder, M. Large-Scale Risk Mapping of an Eruptive Bark Beetle—Importance of Forest Susceptibility and Beetle Pressure. For. Ecol. Manag. 2014, 318, 158–166. [Google Scholar] [CrossRef]
  39. R Development Core Team. R: A Language and Environment for Statistical Computing; R Foundation for Statistical Computing: Vienna, Austria, 2025; Available online: https://www.r-project.org/ (accessed on 27 January 2025).
  40. Anselin, L. Local Indicators of Spatial Association—LISA. Geogr. Anal. 1995, 27, 93–115. [Google Scholar] [CrossRef]
  41. Zhang, C.; Luo, L.; Xu, W.; Ledwith, V. Use of Local Moran’s I and GIS to Identify Pollution Hotspots of Pb in Urban Soils of Galway, Ireland. Sci. Total Environ. 2008, 398, 212–221. [Google Scholar] [CrossRef]
  42. Anselin, L.; Syabri, I.; Kho, Y. GeoDa: An Introduction to Spatial Data Analysis. Geogr. Anal. 2006, 38, 5–22. [Google Scholar] [CrossRef]
  43. Elith, J.; Leathwick, J.R.; Hastie, T. A Working Guide to Boosted Regression Trees. J. Anim. Ecol. 2008, 77, 802–813. [Google Scholar] [CrossRef]
  44. Elith, J.; Leathwich, J. Boosted Regression Trees for Ecological Modelling and Prediction. 2017. Available online: https://citeseerx.ist.psu.edu/document?repid=rep1&type=pdf&doi=3fa2b8826c881d732169995869b9d356c6996029 (accessed on 6 March 2025).
  45. Ridgeway, G. The Gbm Package. Generalized Boosted Regression Models. 2024. Available online: https://cran.r-project.org/web/packages/gbm/gbm.pdf (accessed on 6 March 2025).
  46. Gil, W.; Łukaszewicz, J.; Paluch, R.; Zachara, T. Jesion Wyniosły. Hodowla i Zagrożenia.; Gil, W., Ed.; PWRiL: Warsaw, Poland, 2010; ISBN 978-83-09-99019-2. [Google Scholar]
  47. Pacia, A.; Borowik, P.; Hsiang, T.; Marozau, A.; Matić, S.; Oszako, T. Ash Dieback in Forests and Rural Areas—History and Predictions. Forests 2023, 14, 2151. [Google Scholar] [CrossRef]
  48. Dahlsjö, C.A.L.; Malhi, Y. Unravelling a Hidden Synergy: How Pathogen-Climate Interactions Transform Habitat Hydrology and Affect Tree Growth. Sci. Total Environ. 2024, 954, 176325. [Google Scholar] [CrossRef]
  49. Jung, T.; Nechwatal, J. Phytophthora Gallica Sp. Nov., a New Species from Rhizosphere Soil of Declining Oak and Reed Stands in France and Germany. Mycol. Res. 2008, 112, 1195–1205. [Google Scholar] [CrossRef] [PubMed]
  50. Paluch, R. Rate and Direction of Changes in Tree Species Composition of Natural Stands in Selected Forest Associations in the Białowieża Forest. For. Res. Pap. 2015, 75, 385–406. [Google Scholar] [CrossRef]
  51. Skovsgaard, J.P.; Wilhelm, G.J.; Thomsen, I.M.; Metzler, B.; Kirisits, T.; Havrdová, L.; Enderle, R.; Dobrowolska, D.; Cleary, M.; Clark, J. Silvicultural Strategies for Fraxinus excelsior in Response to Dieback Caused by Hymenoscyphus fraxineus. For. Int. J. For. Res. 2017, 90, 455–472. [Google Scholar] [CrossRef]
  52. Bartelheimer, M.; Poschlod, P. Functional Characterizations of Ellenberg Indicator Values—A Review on Ecophysiological Determinants. Funct. Ecol. 2016, 30, 506–516. [Google Scholar] [CrossRef]
  53. Turczański, K.; Rutkowski, P.; Nowiński, M.; Zawieja, B. Health Status of European Ash (Fraxinus excelsior L.) in Relation to the Moisture of Selected Forest Sites. Sylwan 2020, 164, 133–141. [Google Scholar] [CrossRef]
  54. Turczański, K. Site Conditions for Decline of Ash Stands in Europe. Acta Sci. Pol. Silvarum Colendarum Ratio Et Ind. Lignaria 2020, 19, 67–73. [Google Scholar] [CrossRef]
  55. Boczoń, A.; Kowalska, A.; Ksepko, M.; Sokołowski, K. Climate Warming and Drought in the Bialowieza Forest from 1950–2015 and Their Impact on the Dieback of Norway Spruce Stands. Water 2018, 10, 1502. [Google Scholar] [CrossRef]
  56. Husson, C.; Caël, O.; Grandjean, J.P.; Nageleisen, L.M.; Marçais, B. Occurrence of Hymenoscyphus pseudoalbidus on Infected Ash Logs. Plant Pathol. 2012, 61, 889–895. [Google Scholar] [CrossRef]
  57. Grosdidier, M.; Scordia, T.; Ioos, R.; Marçais, B. Landscape Epidemiology of Ash Dieback. J. Ecol. 2020, 108, 1789–1799. [Google Scholar] [CrossRef]
  58. Kjær, E.D.; McKinney, L.V.; Nielsen, L.R.; Hansen, L.N.; Hansen, J.K. Adaptive Potential of Ash (Fraxinus excelsior) Populations against the Novel Emerging Pathogen Hymenoscyphus pseudoalbidus. Evol. Appl. 2012, 5, 219–228. [Google Scholar] [CrossRef]
  59. Lygis, V.; Vasiliauskas, R.; Larsson, K.H.; Stenlid, J. Wood-Inhabiting Fungi in Stems of Fraxinus excelsior in Declining Ash Stands of Northern Lithuania, with Particular Reference to Armillaria Cepistipes. Scand. J. For. Res. 2005, 20, 337–346. [Google Scholar] [CrossRef]
  60. Madsen, C.L.; Kosawang, C.; Thomsen, I.M.; Hansen, L.N.; Nielsen, L.R.; Kjær, E.D. Combined Progress in Symptoms Caused by Hymenoscyphus fraxineus and Armillaria Species, and Corresponding Mortality in Young and Old Ash Trees. For. Ecol. Manag. 2021, 491, 119177. [Google Scholar] [CrossRef]
  61. Flynn, W.R.M.; Grieve, S.W.D.; Henshaw, A.J.; Owen, H.J.F.; Buggs, R.J.A.; Metheringham, C.L.; Plumb, W.J.; Stocks, J.J.; Lines, E.R. UAV-Derived Greenness and within-Crown Spatial Patterning Can Detect Ash Dieback in Individual Trees. Ecol. Solut. Evid. 2024, 5, e12343. [Google Scholar] [CrossRef]
  62. Waser, L.T.; Küchler, M.; Jütte, K.; Stampfer, T. Evaluating the Potential of Worldview-2 Data to Classify Tree Species and Different Levels of Ash Mortality. Remote Sens. 2014, 6, 4515–4545. [Google Scholar] [CrossRef]
  63. Chan, A.H.Y.; Barnes, C.; Swinfield, T.; Coomes, D.A. Monitoring Ash Dieback (Hymenoscyphus fraxineus) in British Forests Using Hyperspectral Remote Sensing. Remote Sens. Ecol. Conserv. 2021, 7, 306–320. [Google Scholar] [CrossRef]
Forests 16 00506 g001
Figure 1. Map illustrating the location of the Białowieża Forest, including its subdivisions into forest districts and the national park. The distribution of European ash trees, as recorded in July 2015, was determined using the methodology described by Kamińska et al. [27] and Lisiewicz et al. [34].
Figure 1. Map illustrating the location of the Białowieża Forest, including its subdivisions into forest districts and the national park. The distribution of European ash trees, as recorded in July 2015, was determined using the methodology described by Kamińska et al. [27] and Lisiewicz et al. [34].
Forests 16 00506 g001
Forests 16 00506 g002
Figure 2. Data processing workflow.
Figure 2. Data processing workflow.
Forests 16 00506 g002
Forests 16 00506 g003
Figure 3. Hotspot analysis results for ash mortality in the Białowieża Forest from 2015 to 2019. The numbers in brackets represent the local indicators of spatial association (LISA) classes.
Figure 3. Hotspot analysis results for ash mortality in the Białowieża Forest from 2015 to 2019. The numbers in brackets represent the local indicators of spatial association (LISA) classes.
Forests 16 00506 g003
Forests 16 00506 g004
Figure 4. Interaction plot of the number of ash dead per 1 ha cell between 2015 for 2019 for LISA classes (1—“high–high”; 3—“low–high”; 4—“high–low”) identified by hotspot analysis and ‘dominant species’.
Figure 4. Interaction plot of the number of ash dead per 1 ha cell between 2015 for 2019 for LISA classes (1—“high–high”; 3—“low–high”; 4—“high–low”) identified by hotspot analysis and ‘dominant species’.
Forests 16 00506 g004
Forests 16 00506 g005
Figure 5. Partial regression plots showing the key predictors of ash mortality based on BRT models are presented. The fitted function (on the y-axis) illustrates the impact of each variable on dieback dynamics, accounting for the effects of all other predictors. Percentages represent the relative contribution of each variable to the BRT model. Tick marks at the top of each plot correspond to the 10th percentiles of the data.
Figure 5. Partial regression plots showing the key predictors of ash mortality based on BRT models are presented. The fitted function (on the y-axis) illustrates the impact of each variable on dieback dynamics, accounting for the effects of all other predictors. Percentages represent the relative contribution of each variable to the BRT model. Tick marks at the top of each plot correspond to the 10th percentiles of the data.
Forests 16 00506 g005
Table 1. Variables included in the analysis.
Table 1. Variables included in the analysis.
Variable TypeVariableDescription
treealive ashNumber of alive ash
other deciduous mortalityNumber of deciduous without ash dying between 2015 and 2019
dead deciduousNumber of dead deciduous
other alive deciduousNumber of alive deciduous without ash
coniferous treesNumber of coniferous
coniferous mortalityNumber of coniferous without ash dying between 2015 and 2019
avg heightMean height of trees
cover allPercent of a grid cell covered by tree crowns
stand and habitatdominant tree speciesDominant tree species
habitat typeBasic unit in the classification of forest habitats in Poland
dominant ageAge of the dominant tree species
topographicaspectThe compass direction
slopeRate of change of elevation
tpi 500Topographic Position Index calculated with a 500 m radius of influence.
areadistrictName of Forest District
reserve
Table 2. Results of detecting living and dead ash trees in the top forest layer of the Białowieża Forest, presented separately for different forest districts and the Białowieża National Park.
Table 2. Results of detecting living and dead ash trees in the top forest layer of the Białowieża Forest, presented separately for different forest districts and the Białowieża National Park.
Area TypeNumber of Detected
Alive Ash		Number of Detected
Dead Ash		Percentage of Dead
Ash
20152015–20192015–2019
Białowieża F.D.38,29212,08831.57%
Browsk F.D.25,828609223.59%
Hajnówka F.D.35,55210,27428.90%
Białowieża N.P.45,39713,11128.88%
Total145,06941,56528.65%
Table 3. Descriptive statistics (mean ± SD) for LISA classes. Identical letters (a, b, c, d) indicate homogeneous groups with no significant differences, as determined by Tukey’s HSD test (α = 0.01).
Table 3. Descriptive statistics (mean ± SD) for LISA classes. Identical letters (a, b, c, d) indicate homogeneous groups with no significant differences, as determined by Tukey’s HSD test (α = 0.01).
FactorWhole Area LISA Class
0134
number_of_dead_ash1.24 ± 2.070.92 ± 1.48
b		4.76 ± 3.82
d		0 ± 0
a		1.2 ± 0.57
c
number of alive ash4.33 ± 4.643.65 ± 3.44
b		12.24 ± 7.81
c		3.98 ± 3.34
b		2.25 ± 1.95
a
dead_deciduous9.86 ± 10.868.27 ± 9.19
b		21.67 ± 18.0
c		9.83 ± 10.12
b		6.20 ± 7.34
a
other alive deciduous206.3 ± 80.0205.6 ± 81.5
b		220.3 ± 50.5
c		233.0 ± 67.2
c		175.3 ± 95.4
a
other_deciduous_mortality14.15 ± 13.6913.67 ± 12.91
a, b		19.31 ± 18.58
c		11.28 ± 8.44
a		15.04 ± 18.29
b
coniferous trees81.77 ± 83.8085.06 ± 84.90
b		36.45 ± 28.95
a		47.90 ± 43.45
a		130.1 ± 113.2
c
coniferous_mortality26.67 ± 37.5927.85 ± 38.54
b		12.85 ± 17.12
a		15.51 ± 22.66
a		36.03 ± 46.28
c
Table 4. The relative contribution (%) of various variables to ash mortality, based on the general BRT analysis, is provided. The direction of each predictor’s association with ash mortality is indicated by [↑] for a positive relationship and [↓] for a negative relationship.
Table 4. The relative contribution (%) of various variables to ash mortality, based on the general BRT analysis, is provided. The direction of each predictor’s association with ash mortality is indicated by [↑] for a positive relationship and [↓] for a negative relationship.
Predictor TypeVariableRelative Contribution [%]
treealive ash79.44 [↑]
other deciduous mortality5.42 [↑]
dead deciduous2.66 [↑]
other alive deciduous<1
coniferous trees1.51
coniferous mortality<1
avg height1.59
cover all<1
stand and habitatdominant tree species2.28
habitat type<1
dominant age<1
topographicaspect1.26
slope1.31
tpi 500<1
areadistrict<1
reserve<1
training
data correlation		0.87
CV correlation0.82
standard error0.004
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

Kamińska, A.; Lisiewicz, M.; Kraszewski, B.; Tkaczyk, M.; Stereńczak, K.; Wysocka-Fijorek, E. Assessing Ash (Fraxinus excelsior L.) Dieback Dynamics in the Białowieża Forest, Poland, Using Bi-Temporal High-Resolution Remote Sensing Data. Forests 2025, 16, 506. https://doi.org/10.3390/f16030506

AMA Style

Kamińska A, Lisiewicz M, Kraszewski B, Tkaczyk M, Stereńczak K, Wysocka-Fijorek E. Assessing Ash (Fraxinus excelsior L.) Dieback Dynamics in the Białowieża Forest, Poland, Using Bi-Temporal High-Resolution Remote Sensing Data. Forests. 2025; 16(3):506. https://doi.org/10.3390/f16030506

Chicago/Turabian Style

Kamińska, Agnieszka, Maciej Lisiewicz, Bartłomiej Kraszewski, Miłosz Tkaczyk, Krzysztof Stereńczak, and Emilia Wysocka-Fijorek. 2025. "Assessing Ash (Fraxinus excelsior L.) Dieback Dynamics in the Białowieża Forest, Poland, Using Bi-Temporal High-Resolution Remote Sensing Data" Forests 16, no. 3: 506. https://doi.org/10.3390/f16030506

APA Style

Kamińska, A., Lisiewicz, M., Kraszewski, B., Tkaczyk, M., Stereńczak, K., & Wysocka-Fijorek, E. (2025). Assessing Ash (Fraxinus excelsior L.) Dieback Dynamics in the Białowieża Forest, Poland, Using Bi-Temporal High-Resolution Remote Sensing Data. Forests, 16(3), 506. https://doi.org/10.3390/f16030506

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