Introduction

Forests are essential natural resources providing a variety of economic, sociocultural, and ecological benefits. However, they are under constant threat from various factors including pests and diseases, which have significant impacts on their health and productivity. The industrial wood production sector is particularly vulnerable to forest pests and diseases, as they can cause significant damage to timber resources and affect the economic viability of the industry. Sevinç1 reports that forest diseases are the most important harmful factor affecting the industrial wood production, while insect outbreaks are the third important one. Thus, effective control methods are essential to protect forests and ensure sustainable industrial wood production. This article will discuss the effectiveness of some of the control methods employed against forest pests and diseases affecting industrial wood production, including the mechanical, the chemical, the biotechnical, and the biological control methods. These solutions play a critical role in maintaining the health and productivity of forests, ensuring a sustainable supply of industrial wood and other forest products2,3,4,5,6,7.

Food and Agriculture Organization (FAO) of United Nations reports that the demand for wood products have increased significantly over the years due to population growth and urbanization. As a result, industrial wood production became a crucial part of the forest industry, providing the foundation for a wide range of products such as paper, furniture, and construction materials. Moreover, the area of certified forests worldwide increased by 1.85% (8.4 million ha) reaching a new all-time high of 463 million ha8.

According to the Research Agency of the Forestry Commission (RAFC) of The United Kingdom, a total of 3.9 billion m3 of wood was obtained from global forests in 2020, of which around one half (49%) was used as fuel wood and the remainder as industrial roundwood. Global production of wood products in 2020 reached a total of 473 million m3 of sawn wood, 368 million m3 of wood-based panels, and 401 million tons of paper and paperboard. In addition, Europe produced 25% of the sawn wood, 22% of the wood-based panels, and 21% of all paper and paperboard produced in the world in 2020. North and Central America and Europe together, on the other hand, accounted for around three fifths (58%) of all industrial roundwood production in 2020. Moreover, globally, production of industrial roundwood increased by 7% between 2015 and 2020, resulting from increases in all regions except North and Central America. In addition, nearly three quarters (74%) of fuel wood production in 2020 took place in Asia and Africa9. Industrial wood production in the world by years9 is given in Table 1 and visualized in Fig. 1.

Table 1 Industrial wood production (million m3) in the world by years.
Full size table
Fig. 1
figure 1

Worldwide industrial wood production with the temporal changes and the production amounts over the years.

Full size image

When Table 1 and Fig. 1 are examined, it appears that industrial wood is mostly produced in Europe. The second greatest producer region appears as North and Central America, while Asia is the third producer region. Moreover, except for Europe, all regions seem to have increased their industrial wood production in 2020.

In the literature, there are various studies concerning forest pests and diseases, such as MacLean10 who studied the impact of forest pests and fire on timber yield reporting that damaging agents kill trees and destroy the commercial value of timber. In addition, Neumann and Marks11 examined the status and management of insect pests and diseases in Victorian softwood plantations in Australia. They discussed the use of pesticides along with the cultural, biological, and genetic techniques, highlighting the threat posed by accidentally introduced insect pests and diseases. They also identified some actions that were considered useful in providing the necessary controls in the forestry and timber sectors. On the other hand, Nair12 investigated the insect pests and diseases in Indonesian forest, assessed the major threats, research efforts, and investigated the related literature. The study suggests that the disease problems are less significant than the pests in the natural forests. In addition, Cudmore et al.13 studied the economic impacts of the invasive forest pests in Canada. It is reported that invasive pests reportedly cause billions of dollars in damage to the Canadian economy, as well as causing yield loss and environmental damage. The study also emphasizes the need for effective measures to prevent the spread of invasive pests and to control the outbreaks. Similarly, Lovett et al.14 performed a review study and synthesized information on invasions of nonnative forest insects and diseases in the United States. They also examined the ecological and economic impacts, pathways of arrival, and distribution of these insects and diseases by suggesting policies for reducing future invasions. Nambiar et al.15, however, studied the impacts of the diseases on sustainable wood supply to the pulp and paper industry in Indonesia and pointed out the risks of relying on a single tree-species to obtain wood. In addition, Thu et al.16 studied the new and emerging insect pest and disease threats to forest plantations in Vietnam. They suggest that resistant tree species, improved nursery hygiene, silvicultural operations, and integrated pest management schemes are required to protect the forests. In another study, Venkatesh et al.17 examined the diseases and insect pests challenge to meet wood production demand of teak timber (Tectona grandis L.), which is a valuable tree species for timber. They stated that updating the struggles against biotic factors such as insect pests, fungal and bacterial infections that limit teak tree production would help to obtain better quality wood.

Research gap and motivation of the study

Industrial wood production constitutes a fundamental component of various sectors, including construction, furniture manufacturing, and the paper and pulp industries. However, this industry is significantly affected by forest pests and diseases, which lead to considerable reductions in the industrial wood quality and amount. These biological threats result in substantial economic losses, particularly in countries with a well-developed wood products industry such as Türkiye.

Ensuring the sustainability of industrial wood production requires the implementation of effective and environmentally responsible pest and disease control strategies. This study investigates the impact and effectiveness of the mechanical, the chemical, the biotechnical, and the biological control methods on the industrial wood production in Türkiye. A machine learning-based analytical framework is employed to evaluate these methods in terms of both their operational efficiency and ecological compatibility.

Moreover, the study aims to contribute to public awareness by emphasizing the importance of developing innovative and nature-friendly approaches to forest pest and disease management. Promoting strategies that combine high effectiveness with environmental sustainability is essential for the long-term resilience and productivity of the forestry sector.

Materials and methods

Artificial neural networks

Artificial neural network is a type of machine learning model simulating the structure and function of the human brain. Artificial neural networks are used to solve complex problems in various fields, such as medicine, marketing, finance, and economics. The architecture of artificial neural networks consists of layers of interconnected nodes, which are also called neurons that process and transmit information. The mechanism of artificial neural networks is to make estimations or classifications by learning from the given data set.

The main architecture of artificial neural networks consists of three types of interconnected layers involving neurons, which are an input layer, a hidden layer, and an output layer. The input layer receives the data and sends them to the hidden layer to be processed. The output layer, to which the processed data is forwarded, provides the final estimation or classification. The output of each neuron is calculated using a function called activation function, which can be a simple linear function or a more complex non-linear function.

To train artificial neural network models, a data set is first divided into two parts as a training set and a testing set. The training set is used to adjust the weights and biases of the neurons in the network to minimize the error between the predicted output and the actual output. The testing set, however, is used to evaluate the performance of the constructed network. For more detailed information about artificial neural networks18,19,20,21 may be referred.

Multilayer perceptron model

Multilayer perceptron is a type of artificial neural network model that is widely used in machine learning applications. The architecture of a multilayer perceptron, which is also a type of deep learning model, consists of an input layer, one or more hidden layers, and an output layer. The input layer receives the input data, which is then processed by the hidden layers. The output layer provides the final prediction or classification based on the input data18,20,22,23,24,25. A schematic representation of a multilayer perceptron model architecture, which consists of an input layer with two neurons, three hidden layers comprising three, two, and five neurons respectively, and an output layer with a single neuron is provided in Fig. 2.

Fig. 2
figure 2

Schematic representation of a multilayer perceptron model consisting of an input layer with two neurons, three hidden layers comprising three, two, and five neurons respectively, and an output layer with a single neuron.

Full size image

Each node in a multilayer perceptron model receives input from other nodes in the previous layer and produces an output that is transmitted to nodes in the next layer. The output of each node is calculated using an activation function, which can be a simple linear function or a more complex non-linear function. To train a multilayer perceptron model, a dataset is partitioned into two parts as training set and testing set. The training set is used to adjust the weights and biases of the neurons in the network to minimize the error between the predicted output and the actual output. The validation set is used to evaluate the performance of the network on unseen data. The process of training multilayer perceptron models involves the use of the backpropagation algorithm, which is a gradient-based optimization algorithm. The backpropagation algorithm calculates the gradient of the error function with respect to the weights and biases of the neurons in the network. The weights and biases are then adjusted using this gradient to minimize the error. Multilayer perceptron models have been used in a wide range of applications such as speech recognition, computer vision, natural language processing, and financial forecasting. One of the main advantages of multilayer perceptron models is their ability to learn complex patterns in data and make accurate predictions. Multilayer perceptron models are also highly adaptable and can be used in a wide range of applications. However, multilayer perceptron models have some limitations, such as their tendency to overfit the data, which can result in poor performance on another data set. Another limitation is their computational complexity, which can make them slow to train and difficult to deploy on low-power devices. Fernandes de Mello and Antonelli Ponti26, Velo et al.27, and Zhou et al.28 may be referred for more detailed information about multilayer perceptron model.

Study area

Türkiye is a country rich in natural resources, including forests. The forest sector in Türkiye is an important part of the Turkish economy, providing significant contributions to employment, trade, and sustainable development. In addition, Türkiye is also a significant producer of industrial wood products in the world. Most of the industrial wood production in Türkiye is from state forests, which are owned and controlled by General Directorate of Forestry (GDF) consisting of 30 Regional Directorates of Forestry (RDFs). Figure 3 presents the forest density in Türkiye as of 202029.

Fig. 3
figure 3

Forest density map of the Regional Directorates of Forestry in Türkiye as of 2020, illustrating spatial variations in forest coverage across different administrative regions.

Full size image

According to GDF, while forest assets are decreasing in the world, Türkiye is one of the countries increasing their forest areas over the decades. As of 2023, forest assets in Türkiye have increased by 3 million ha since 1973. In addition, the amount of fuel wood production in Türkiye was 4.6 million m3 and industrial wood production was 25.5 million m3 in 202230. Figure 4 shows the annual industrial wood production of Türkiye in 2013–202231.

Fig. 4
figure 4

Yearly industrial wood production in Türkiye from 2013 to 2022, highlighting the production trends and fluctuations in the industrial wood amounts over the specified period.

Full size image

Figure 4 suggests that except for a slight decrease in 2017, Türkiye steadily increased its industrial wood production until the end of 2021. Afterwards, however, the production has a slight decrease again in 2022. The amounts of industrial wood production in the RDFs of Türkiye as of 2022 are presented in Table 2.

Table 2 Industrial wood production amounts in the regional directorates of forestry in Türkiye as of 2022.
Full size table

The data presented in Table 2 are visually represented in Fig. 5 for clearer comparison and interpretation.

Fig. 5
figure 5

Industrial wood production across the Regional Directorates of Forestry in Türkiye as of 2022, illustrating the spatial distribution and regional variations in the production levels.

Full size image

Table 2 and Fig. 5 show that industrial wood is produced in the highest amounts in Antalya, Muğla, Bolu, Kastamonu, and Zonguldak directories, respectively. The least production levels, however, are observed in Hatay, Erzurum, Elazığ, and Şanlıurfa directories. Industrial wood production densities in the RDFs in Türkiye as of 2022 are displayed in Fig. 6.

Fig. 6
figure 6

Industrial wood production densities across the Regional Directorates of Forestry in Türkiye as of 2022, highlighting the spatial variation in production intensity among the administrative regions.

Full size image

Figure 6 suggests that while the most intense regions of industrial wood production are the northwest and southwest regions, the production amounts in the eastern regions are relatively less in Türkiye. However, for a sustainable and economical industrial wood production, it is necessary to apply effective control methods against forest pests.

Main methods of controlling forest pests and diseases in Türkiye

Forests are threatened by many factors as well as many forest pests. Various methods have been developed to combat these pests. The most commonly used methods in Türkiye can be grouped under the following subheadings.

Mechanical control methods

Mechanical control methods applied in forests refer to methods applied by machinery, tools, or hands to prevent or reduce damage caused by pests, diseases, or invasive species. Some of these methods can be summarized as crushing, collecting, blocking, and capturing with traps. In the crushing method, pests or their eggs, which are generally found in large numbers, are crushed and killed using hands or a wire brush. When the collection method is applied, especially the pests living together and their eggs or larvae are collected and destroyed by cutting the plants. When applying the blocking method, the goods are covered with nylon to prevent insects such as bark beetles that live in forest areas or in shelled goods in warehouses from returning to the forest. The lower parts of the nylons are covered with soil to prevent harmful insects from getting out. Thus, a large amount of damage can be prevented. On the other hand, it is also possible to catch and destroy insects and other pests that are difficult to collect and live in hidden places by ambushing them with various trapping methods32,33,34,35.

Chemical control methods

Chemical control methods for protecting forests, however, refer to use of chemical substances to manage or control forest ecosystems, particularly to prevent or reduce the damage caused by pests, diseases, or invasive species. The main element in the chemical control methods is chemical compounds. These compounds are called herbalists or specifically, pesticides. Pesticides can be grouped according to the organisms they target, such as acaricides, fungicides, herbicides and insecticides. They can also be named according to their production methods, such as chemical pesticides and biopesticides32,33,34,35.

Biotechnical control methods

Biotechnical control is the use of certain techniques to control the normal biological and physiological behaviors of harmful organisms rather than directly destroying them. The main biotechnical control products can be grouped as traps, attractants or repellents, and chemicals and hormones that disrupt insect development. The traps can be grouped into color and light, food, and sound traps. Traps can be used alone, for example color traps (e.g. yellow sticky traps), but can also be integrated with other traps or pheromones. Pheromones are generally used in three ways to combat pests, such as monitoring, preventing mating, and mass capture. Monitoring refers to tracking the emergence and population of an insect in a certain area. Preventing mating, on the other hand, means preventing males and females of a pest species from meeting in a specific area, thus preventing reproduction. Similarly, mass capture refers to capture and destruction of the maximum possible number of individuals of a pest for control purposes32,33,35,36.

Biological control methods

There are many creatures that live by feeding on organisms that harm forest plants and plant products. Therefore, these creatures are natural enemies of forest pests, because they destroy pests and reduce their populations. The biological control methods for protecting forests refer to the use of living organisms or biological agents to manage or control forest ecosystems, particularly to prevent or reduce damage caused by pests, diseases, or invasive species. These methods involve the use of natural predators, parasites, or pathogens to control the populations of target species that threaten the health or productivity of forests. Biological control is a control method developed by taking advantage of the balance mechanisms in nature and has almost no negative effects on nature32,33,34,35,36,37.

Data and variables

The dataset presented in Table 3 includes the industrial wood production and the extent of forest areas in Türkiye where the mechanical, the chemical, the biotechnical, and the biological control methods were applied between 2013 and 2021.

Table 3 The data set used in the study.
Full size table

The extent of forest areas in Türkiye where the mechanical, the chemical, the biotechnical, and the biological control methods were applied, as presented in Table 3, is graphically depicted in Fig. 7.

Fig. 7
figure 7

Extent of forest areas in Türkiye treated with the mechanical, the chemical, the biotechnical, and the biological control methods between 2013 and 2021.

Full size image

Table 3 and Fig. 7 suggest that the most frequently used methods throughout the years are the mechanical, the biological, and the biotechnical methods. It can also be observed that the chemical control methods were applied in lesser amounts than the other methods. Moreover, while the use of the chemical control methods, which can be harmful to the environment, declined over the years, the biological and the biotechnical control methods, which are comparatively more environmentally friendly, gained popularity.

Results

Development of the multilayer perceptron model

The multilayer perceptron model was developed using Weka38 software. A learning rate of 0.3 was applied with a momentum score of 0.3 and a validation threshold of 20. The number of the iterations was set to 51. Moreover, 70% of the data was used for training and 30% for testing. The best evaluation metrics were obtained for the model having two hidden layers containing two and seven neurons, respectively. The constructed multilayer perceptron model is presented in Fig. 8.

Fig. 8
figure 8

The constructed multilayer perceptron model.

Full size image

The estimation performance metrics of the constructed multilayer perceptron model are provided in Table 4.

Table 4 Model performance metrics.
Full size table

Table 4 demonstrates that the artificial neural network, constructed based on a multilayer perceptron deep learning architecture, achieves a successful accuracy rate, as indicated by the calculated correlation coefficient of approximately 82%.

Evaluation of the deep learning model against the conventional regression model

Deep learning models have consistently demonstrated superior performance compared to classical models39,40. To validate the estimation accuracy of the deep learning model developed in the previous section, it is beneficial to construct a conventional linear regression model using the same dataset for comparison. A multiple linear regression model was fitted to the data presented in Table 3, with industrial wood production as the dependent variable and the remaining variables as independent variables. The resulting analysis of variance (ANOVA) for this model is summarized in Table 5.

Table 5 Analysis of variance (ANOVA) table for the multiple linear regression model.
Full size table

As shown in Table 5, the multiple linear regression model, which has been a conventionally used model in the literature, was highly unsuccessful in estimating the industrial wood production, as all the coefficients, and consequently the overall model, were found to be statistically insignificant. In contrast, the constructed artificial neural network estimated the industrial wood production with a satisfactory correlation score of 82%. This outcome is unsurprising, given that artificial neural networks are widely recognized for their superiority over conventional methods, which are often constrained by numerous theoretical assumptions and limitations.

Evaluation of the effectiveness of the forest pests and diseases control methods

To evaluate the contribution of the variables in the created deep learning model to the estimation, an attribute selection process was carried out and the ranked attribute scores are given in Table 6 in a descending order. The attribute selection was performed using Weka38 software with the “classifier attribute evaluation” module, using four folds and a threshold level of 0.05.

Table 6 Effectiveness of the forest pests and diseases control methods.
Full size table

When Table 6 is examined, it appears that all methods have enhancer effects on the amounts of industrial wood. However, the most effective methods emerge as the chemical control methods with an attribute score of 0.518. The second most effective methods, on the other hand, emerge to be the mechanical control methods with an attribute score of 0.447. The third most effective methods, however, are the biological control methods, which have an attribute score of 0.423. Finally, the methods relatively less effective compared to the others are biotechnical control methods with 0.410 attribute score. Additionally, it is possible to comment that the effectiveness levels of the biological and biotechnical control methods seem to be close to each other. Figure 9 illustrates the effectiveness evaluation of the forest pests and diseases control methods, ranked in decreasing order from most to least effective.

Fig. 9
figure 9

Effectiveness of the forest pests and diseases control methods, ranked in descending order from most to least effective.

Full size image

Discussion

The most striking finding of the study is that, although the chemical control methods were applied in the smallest quantities, they had the greatest positive impact on industrial wood production. Nonetheless, views on the use of the chemical control methods vary. For example, Abbas et al.41 highlight their negative environmental effects, whereas Woreta42 notes that the chemical pest control methods, first introduced in the 1950s, have been improved over time to reduce their environmental harm.

In a similar vein, although the mechanical control methods were applied in smaller quantities than biological and biotechnical methods in most years in Türkiye, they proved to be the second most effective approach for reducing forest pests and diseases, thereby supporting industrial wood production. It was also observed that the application levels of biological control methods, identified as the third most effective, remained relatively stable over the years, except for 2013. While biotechnical methods have gained increasing popularity over time, they appeared to be the least effective among the four control methods examined.

The finding that the chemical control methods are the most effective in increasing the industrial wood production is consistent with the results of Van Den Meersschaut and Lust43, who found that all chemical control methods, with one exception, were significantly more effective in eliminating pests than mechanical and biological methods. In addition, Wilhoit44 and Tudi et al.45 report that, without the use of pesticides, global agricultural production would experience losses of 78% in fruits, 58% in vegetables, and 32% in cereals. Similarly, Stephenson et al.46 state that the use of the chemical control methods doubled the production of the eight principal crops worldwide between 1965 and 1990, resulting in a 30% increase in yield. Holmes and MacQuarrie47 also support the effectiveness of the chemical control methods, stating that over the past 90 years, experimental trials and chemical treatments have been applied against numerous insect pests in Canada. In many cases, the chemical methods have proven to be a viable pest management option. Similarly, Bulman et al.48 report that the chemical control methods are among the most effective approaches for managing tree diseases in nurseries across Europe. From a different perspective, however, Akyol and Sarikaya49 note that the chemical control methods are approximately 1.4 times more expensive than the biological methods in Türkiye.

The mechanical control methods, identified as the second most effective approach, are generally less harmful to the environment compared to the chemical control methods32,50. Pomp et al.51 report that while the chemical control is the most effective option for managing Japanese stilt grass [Microstegium vimineum (Trin.) A. Camus], the mechanical control has also demonstrated effectiveness as a secondary method. Yigit et al.52 attribute the relatively lower effectiveness of the mechanical methods to their more limited applicability compared to the chemical methods. Furthermore, Sarikaya et al.53 note that the mechanical control is not effective against certain pest species, particularly when the level of damage is severe.

The biological control methods, which rank as the third most effective solution in supporting the increase in industrial wood production, are considered the most environmentally friendly. However, despite their growing success, controlling pathogens in forest ecosystems through biological means remains challenging due to the complex interactions and specific characteristics of both hosts and pathogens37. Garnas et al.54 further note that although the biological control methods offer clear advantages, the development of effective strategies can be slow and often involves substantial initial costs. Additionally, identifying, developing, and testing suitable biocontrol agents is a complex and demanding process.

Although the biotechnical control methods have been applied increasingly more than the other three methods, they have proven to be the least effective overall. Yayla et al.55 suggest that this may be because the biotechnical control methods alone are insufficient when pest populations reach high levels.

The research results clearly indicate that, although the chemical control methods are applied in the smallest quantities, they have the greatest positive impact on industrial wood production. Beyond their evident effectiveness, these methods are preferred due to their ease of application, cost-effectiveness, large-scale availability, and rapid action compared to other approaches47,56. However, the significant environmental damage caused by the chemical methods has long been recognized. For example, Balla et al.57 report that pesticides persist in the soil for extended periods, leaving toxic residues and contaminating the environment. Furthermore, if pesticides enter the hydrological system, they can cause severe or even fatal effects on terrestrial and aquatic organisms. Similarly, Allison et al.58 note that even modern pesticides with narrow target spectra can harm many non-target insect species. In addition to environmental and ecological risks, pesticides may pose serious threats to human health, as they can act as potent carcinogens by disrupting the endocrine and immune systems. Evidence of this link is provided by da Silva et al.59, who demonstrate an association between pesticide exposure and the development of head and neck cancers.

Considering the negative effects of the chemical control methods, the question arises whether it is possible to completely abandon these harmful approaches or whether non-chemical methods, such as the biological control, provide a truly effective and safe alternative. There appear to be some reservations regarding this issue. For example, Prospero et al.37 conclude that, to prevent biological control agents from harming the environment by destroying also non-target species, these agents must be registered and licensed by the responsible authorities prior to their use in forests, like the regulations applied to chemical pesticides. Furthermore, they recommend integrating biological control methods with other control strategies, as biological methods alone remain insufficiently effective. Regarding mechanical control methods, it should be noted that these approaches primarily involve the use of simple devices and instruments deployed in forests. However, these devices, often installed over large areas with considerable time and effort, are vulnerable to extreme weather events such as heavy rain and hurricanes, which can render them ineffective60.

Overall, chemical methods are known to have negative environmental impacts. However, given their advantages, including high effectiveness and rapid action, completely abandoning their use in the short term appears unrealistic. Therefore, as an interim measure, efforts should focus on minimizing their environmental harm through laboratory studies and targeted research. In contrast, a long-term and sustainable solution involves the development and widespread adoption of environmentally friendly control methods.

Conclusions

In conclusion, this study examined the effectiveness of various forest pests and diseases control methods on the industrial wood production. The research utilized an artificial neural network model based on multilayer perceptron architecture, a deep learning approach. For comparison, a conventional multiple linear regression model was also developed, but it proved statistically insignificant and failed to estimate the industrial wood production accurately. In contrast, the artificial neural network model demonstrated a strong estimation performance, achieving a correlation coefficient of approximately 82%.

The findings reveal that all control methods positively influence industrial wood production to varying degrees. The chemical control methods emerged as the most effective in increasing production, followed by the mechanical control methods as the second most effective, and the biological control methods ranking third. The biotechnical methods were identified as the least effective among the methods examined.

Despite the clear benefits of the chemical control methods, their negative environmental impacts are well-documented. Given their high effectiveness and rapid action, completely discontinuing their use in the short term appears unrealistic. Therefore, a practical short-term strategy is to focus on reducing their environmental harm through laboratory research and controlled application. For a long-term and sustainable approach, it is essential to develop and promote environmentally friendly control methods. As a continuation of this study, the effectiveness of both existing and newly developed control methods for forest pests and diseases can be further evaluated by employing various machine learning approaches for comparative analysis.