Kohonen Artificial Neural Networks Research Articles

Feeding Patterns of Fish in Relation to the Trophic Status of Reservoirs: A Case Study of Rutilus rutilus (Linnaeus, 1758) in Five Fishing Waters in Serbia

The roach, Rutilus rutilus (Linnaeus, 1758), is one of the most common fish species in mesotrophic and eutrophic lakes throughout Europe. In the Serbian reservoirs selected for this study, this species accounts for the majority of juvenile fish biomass. The aim of this study was to investigate the diet composition of juvenile roach to assess their niche based on resource availability in five Serbian reservoirs with different trophic statuses. A modified Costello graph and Kohonen artificial neural network (i.e., a self-organizing map, SOM) were employed to examine the feeding habits of 142 specimens of roach caught in five reservoirs. Our results show that juvenile roach use zooplankton, benthic macroinvertebrates, algae and detritus in their diet. In addition, five neuron clusters (A, B, C, D and E) were isolated in the SOM output network. The SOM identifies specimens that share similar feeding patterns and categorizes them onto the same or adjacent neurons, determined by dominant prey. In terms of the number of specimens, cluster B was the most numerous, and the predominant prey of these specimens were Daphnia sp., Bosmina sp. and calanoid and cyclopoid copepods. The cluster with the lowest number of specimens is cluster C, and the specimens in it benefited from Chironomidae and Insecta. Due to the different trophic statuses of the reservoirs selected for this study, knowledge of fish feeding habits is essential for the formulation of effective conservation and management strategies for both the species and the reservoirs.

Assessing the level of digital maturity in the Three Seas Initiative countries

The dynamic development of the global economy, associated with the implementation of the Industry 4.0 idea, significantly affects the policies of individual countries and regions. Changes connected with this process include the digitalization of not only economy but practically all areas of life. In order to keep pace with this development and thus meet the growing competition on the global market, it is necessary to quickly adapt to the changes in question. This applies not only to all kinds of enterprises, but also to individual countries and regions. The following article assesses the level of digital maturity in a group of the EU countries that established the so-called Three Seas Initiative in 2015. These countries, like the EU as a whole, are taking a number of actions to modernize and adapt their businesses and economy to the global level. Based on a selected set of 20 determinants, characterizing the most important 9 technologies related to the idea of Industry 4.0, a comprehensive study was conducted to assess the level of digital maturity among these countries in terms of Industry 4.0 technologies implemented by enterprises and to determine their similarity in this regard. In order to achieve these objectives, a multi-stage research methodology was developed. In the first stage, statistical analyses of the adopted determinants were carried out (descriptive statistics, the Gini index, the Lorenz curves). Then, by means of the CRITIC and Grey Relational Analysis, the level of digital maturity in the use of Industry 4.0 digital technologies among these countries and their enterprises was assessed. Based on the determined values of the digital maturity index and its standard deviation, the digital maturity level classes of these countries were created. Their similarities, on the other hand, were identified using the Kohonen artificial neural networks. In addition, the Tau Kendall rank correlation coefficients were calculated between the economic parameters of studied countries and their digital maturity indicators. This comprehensive approach allowed for a reliable assessment of the digital maturity of studied countries, taking into account the most important digital technologies. Based on the results, the highest level of digital maturity in terms of the use of Industry 4.0 technologies in enterprises was found for the Czech Republic, Lithuania and Estonia, and the lowest level for Hungary, Romania and Bulgaria. The results should also be used when developing a common economic policy for countries in similar groups and for the whole Three Seas Initiative. The research carried out fills the research gap in assessing the digital maturity of a large group of EU countries located in its central and eastern parts and presents a new original research methodology in this area, also in terms of the links between the level of digitalization of enterprises and the level of economic development of countries.

Fires, vegetation, and human—The history of critical transitions during the last 1000 years in Northeastern Mongolia

Fires are natural phenomena that impact human behaviors, vegetation, and landscape functions. However, the long-term history of fire, especially in the permafrost marginal zone of Central Asia (Mongolia), is poorly understood. This paper presents the results of radiocarbon and short-lived radionuclides (210Pb and 137Cs) dating, pollen, geochemical, charcoal, and statistical analyses (Kohonen's artificial neural network) of sediment core obtained from Northern Mongolia (the Khentii Mountains region). Therefore, we present the first high-resolution fire history from Northern Mongolia covering the last 1000 years, based on a multiproxy analysis of peat archive data. The results revealed that most of the fires in the region were likely initiated by natural factors, which were probably related to heatwaves causing prolonged droughts. We have demonstrated the link between enhanced fires and "dzud", a local climatic phenomenon. The number of livestock, which has been increasing for several decades, and the observed climatic changes are superimposed to cause "dzud", a deadly combination of droughts and snowy winter, which affects fire intensity. We observed that the study area has a sensitive ecosystem that reacts quickly to climate change. In terms of changes in the vegetation, the reconstruction reflected climate variations during the last millennium, the degradation of permafrost and occurrence of fires. However, more sites with good chronologies are needed to thoroughly understand the spatial relationships between changing climate, permafrost degradation, and vegetation change, which ultimately affect the nomadic societies in the region of Central and Northern Mongolia.

Diet seasonality and food overlap of Perca fluviatilis (Actinopterygii: Perciformes: Percidae) and Rutilus rutilus (Actinopterygii: Cypriniformes: Cyprinidae) juveniles: A case study on Bovan Reservoir, Serbia

European perch, Perca fluviatilis Linnaeus, 1758 and roach, Rutilus rutilus (Linnaeus, 1758) are the most common species present in mesotrophic and eutrophic lakes throughout Europe. Their biomass, especially in juvenile stages, contributes the most to the fish production of these ecosystems. In Bovan Reservoir, these two species constitute the bulk of the juvenile fish biomass. This study aimed to investigate the feeding composition of these two species in order to evaluate their niche overlap due to the availability of resources during different seasons. Traditional diet analysis indices and Kohonen artificial neural network (i.e., a self-organizing map, SOM) were used to investigate the diet of 158 individuals of both species and evaluate their food niche overlap. The indicator value (IndVal) was applied to identify indicator food categories based on which the contents of their alimentary tracts were grouped first into neurons and then into clusters on the SOM. Our results showed that juvenile fish used zooplankton and benthic prey in their diet. Roach often fed on nonanimal prey, while perch of age 0+ used fishes in the diet. Additionally, four clusters of neurons were isolated on the SOM output network. The distribution of perch and roach alimentary tracts in neurons indicated no high degree of competition between them. While diet analyses indices show which food category is generally important in specimensʼ diet, the SOM recognizes those specimens and arranges them together into the same or adjacent neurons based on dominant prey. Understanding fish feeding habits is critical for the development of conservation and management plans. Since Bovan is a eutrophic reservoir, our knowledge of fish feeding habits needs to be considered for stocking strategies in the future.

Renewable energy consumption in economic sectors in the EU-27. The impact on economics, environment and conventional energy sources. A 20-year perspective

The main reason for the ongoing process of energy transition in the global economy is the need to protect the environment, especially the climate. Increased public awareness about the destruction of the Earth's ecosystem and technological development have resulted in increased importance of renewable energy sources, which are increasingly replacing conventional sources. In addition to the environmental aspect, renewable energy also gives a chance for greater energy independence and the ability to produce energy by countries without conventional energy resources. These factors cause individual countries and their groups to take more and more decisive actions to develop renewable energy. Particularly ambitious plans in this area are presented by the European Union, which is the undisputed leader in introducing pro-environmental legislation. Therefore, the following article presents the results of research on the degree of renewable energy use in selected sectors of the economy and households in the European Union countries between 2000 and 2019. The impact of the use of renewable energy sources based energy on economic indicators and on the consumption of energy from conventional sources was also determined. Due to the different levels of economic development and wealth of countries, the study also refers to the groups of “old” and “new” European Union countries. The study showed the concentration and dynamics of change in the consumption of renewable energy in the studied sector and households for individual countries and their groups. The Kohonen artificial neural networks were used to analyze similarities between the European Union countries in terms of the use of energy from renewable energy sources in the sectors studied. In addition, correlations were established between economic and environmental parameters for the whole economy, industry and agriculture/forestry/fishing sectors and the consumption of energy from renewable energy sources, as well as the impact of this energy consumption on the consumption of conventional energy. The results obtained showed a significant increase in the consumption of energy from renewable energy sources in the European Union countries in the studied period, as well as significant differences in the use of this energy among the studied countries, their groups and sectors. Also, it was found that renewable energy sources based energy consumption has a positive impact on economic growth, the reduction of greenhouse gas emissions and the reduction of conventional energy consumption in practically all European Union countries. The results obtained indicate that this impact is greater in the “old” than in the “new” European Union countries.

Comparison of predictions of developmental toxicity for compounds of solvent data set

ABSTRACT We have considered a series of 235 compounds technically classified as solvents. Chemically, they belong to different classes. Their potential developmental toxicity was evaluated using two models available on platform VEGA HUB; model CAESAR and the Developmental/ Reproductive Toxicity library (PG) model. Models provide beside the prediction of developmental toxicity additional information on similar compounds from models’ training sets. In the report, first, we compare the predictions. Second, the sets of similar compounds have been used to implement the clustering scheme. The Kohonen artificial neural network method has been applied as a clustering method. The clusters obtained have been discussed for both models.

Application of the Kohonen neural network for monitoring tissue oxygen supply under hypoxic conditions

The article presents the results of application of the Kohonen artificial neural network (KANN) in assessing the oxygen status of human tissues, as well as for studying the adaptive-compensatory response of the body to functional load. In the experiment, the registered digital oxygen images of the tissue of 31 subjects were distributed into three classes using the KANN. Each group is characterised by different resistance of the organism to hypoxia. The research results have shown the effectiveness of using an artificial neural network structure and the possibility of its implementation for recognition of the functional state of a person under conditions of metabolic hypoxia; it seems relevant and has theoretical and practical significance in the framework of ecological physiology.

Kohonen Artificial Neural Network and Multivariate Analysis in the Identification of Proteome Changes during Early and Long Aging of Bovine Longissimus dorsi Muscle Using SWATH Mass Spectrometry.

To study proteomic changes involved in tenderization of Longissimus dorsi, Charolais heifers and bulls muscles were sampled after early and long aging (12 or 26 days). Sensory evaluation and instrumental tenderness measurement were performed. Proteins were analyzed by gel-free proteomics. By pattern recognition (principal component analysis and Kohonen’s self-organizing maps) and classification (partial least squares-discriminant analysis) tools, 58 and 86 dysregulated proteins were detected after 12 and 26 days of aging, respectively. Tenderness was positively correlated mainly with metabolic enzymes (PYGM, PGAM2, TPI1, PGK1, and PFKM) and negatively with keratins. Downregulation in hemoglobin subunits and carbonic anhydrase 3 levels was relevant after 12 days of aging, while mimecan and collagen chains levels were reduced after 26 days of aging. Bioinformatics indicated that aging involves a prevalence of metabolic pathways after late and long periods. These findings provide a deeper understanding of changes involved in aging of beef and indicate a powerful method for future proteomics studies.

Descriptive screening and lexicon development of non‐aged artisanal cachaça sensorial profile using principal component analysis and Kohonen artificial neural networks

AbstractCachaça is a distilled spirit made from sugarcane, exclusively produced in Brazil, and appreciated worldwide. This paper seeks to evaluate the sensory characteristics of 24 nonaged artisanal cachaça samples from Salinas (Minas Gerais, Brazil) through descriptive analysis, as well as chemometrically treat the obtained data based on principal components analysis (PCA) and Kohonen's neural network. The attributes (23) were divided between aroma (11) and flavor (12). PCA does not show good differentiation of nonaged cachaça samples. On the other hand, by using Kohonen's neural network it was possible to group samples according to their aroma and flavor characteristics in 9 and 10 distinct groups, respectively. A reduced number of descriptors could be used to describe the flavor of cachaça samples (alcohol, acidic, sweet, bitter, citric, tar, and burning), as significant correlations (R > 0.70, p < .05) exist among them with fruity, bagasse, fermented sugarcane juice, and astringent descriptors. This diminution on descriptors numbers could be able to reduce the workload of the judging panel with no losses to the sample' sensory characterization. The use of Kohonen's network chemometric treatment for treat sensory data showed to be a better alternative that PCA approach in this study.Practical applicationsThe growth in production and the appreciation of cachaça in the domestic and foreign markets have directed the production of the drink in Brazil with a focus on its quality and added value, seeking to obtain international recognition and increase exports. In this sense, the lexicon development and sensory characterization of samples from Salinas‐MG is very important to compare with commercial and artisanal samples from other regions in future studies. This work created a sensory profile for samples from this region, which despite its national and international popularity, does not exist yet. Finally, this is the first study that describes a lexicon development and sensory characterization of cachaças from Salinas‐MG and describes data with artificial neural networks, a potential approach for further studies.

Studying the Level of Sustainable Energy Development of the European Union Countries and Their Similarity Based on the Economic and Demographic Potential

The concept of sustainable economic development takes into account economic, social and environmental aspects and strives to achieve balance between them. One of the basic areas where it is required to revalue the current views on sustainable development is energy. The growing public awareness of environmental protection forces changes in this industry. Despite the global nature of this problem, its solution is perceived differently in various regions of the world. The unquestionable leader in introducing the idea of sustainable development economy is the European Union, where the energy sector is of key importance for the effectiveness of this process. In order to assess the sustainable energy development of the European Union countries, studies were conducted based on 13 selected indicators characterizing this sector in terms of energy, economy and environment. In order to assess the specificity of the European Union countries, these indicators were additionally compared to the gross domestic product value and the number of inhabitants of individual countries. For these cases, multi-criteria analyses were carried out using the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) method. It allowed the authors to rank the European Union countries in terms of their adaptation to a sustainable energy economy. Based on the determined values of indicators versus the gross domestic product and the number of inhabitants of the countries in question, these countries were also divided into similar groups with the use of the Kohonen artificial neural networks. These groups can pursue a common energy policy in the field of sustainable development. The aim of the research was to present a new approach to the assessment of sustainable energy development of the European Union countries. The extensive ratio analysis (13 indicators of the sustainable energy development), including the economic and demographic potential of individual countries, and the use of modern tools made it possible to acquire new knowledge in the field of sustainable energy development in the European Union countries. The results should be utilized for more effective sustainable energy development of the European Union countries.

Decision Routing Problems in A Wireless Sensor Network Based on A Neural Mechanism

This article proposes a solution for the routing problem in wireless sensor networks (WSN) based on a neural mechanism. The basic concepts of wireless sensor networks, artificial neural networks (ANNs), and WSN routing protocols are presented. The Kohonen ANN was selected to solve the problem of routing in wireless sensor networks based on a neural mechanism. A radio visibility matrix is proposed, which is a mathematical description of the connectivity of network nodes and the radio visibility of each node with respect to all other network nodes. Based on the Kohonen ANN trained by the constructive method, a method for WSN neural network clustering was developed. Two software-modeling environments are presented that were created to confirm the theory with respect to the logic of the developed methods for WSN clustering, their correction and the verification of their adequacy. Numerical results of modeling the solution of the routing problem in a wireless sensor network based on a neural mechanism by neural network clustering, the WSN matrix clustering method and the energy distance neural clustering protocol (EDNCP) are presented. It was found that the developed EDNCP protocol was 29% more efficient than known analogs.

Neural network classification of substorm geomagnetic activity caused by solar wind magnetic clouds

A Kohonen artificial neural network (ANN) was used to classify patterns of causal relationships between the level of geomagnetic activity in the auroral zone and plasma and magnetic field parameters in the body of an interplanetary magnetic cloud (IMO). Terrestrial and satellite observations during 33rd interplanetary magnetic clouds recorded from 1998 to 2012 are examined in detail. Experiments with the ANN during its fast training show that substorm discrimination by their intensity by three classes plus a “collector” for collecting atypical events is optimal for the study. An analysis of the classification result studies showed that each selected class of substorms corresponds to a specific set of perturbations of the plasma parameters and the magnetic field of the IMO. Using the integral characteristics of the plasma and the IMF components as input parameters of the ANN allowed us to detect the levels of the expected intensity of the AL index with an accuracy of up to 70%. The created ANNs can be used to restore the AL index both during periods of isolated magnetospheric substorms and during periods of a series of continuous successive substorms, one after another.

A Blended Artificial Intelligence Approach for Spectral Classification of Stars in Massive Astronomical Surveys.

This paper analyzes and compares the sensitivity and suitability of several artificial intelligence techniques applied to the Morgan–Keenan (MK) system for the classification of stars. The MK system is based on a sequence of spectral prototypes that allows classifying stars according to their effective temperature and luminosity through the study of their optical stellar spectra. Here, we include the method description and the results achieved by the different intelligent models developed thus far in our ongoing stellar classification project: fuzzy knowledge-based systems, backpropagation, radial basis function (RBF) and Kohonen artificial neural networks. Since one of today’s major challenges in this area of astrophysics is the exploitation of large terrestrial and space databases, we propose a final hybrid system that integrates the best intelligent techniques, automatically collects the most important spectral features, and determines the spectral type and luminosity level of the stars according to the MK standard system. This hybrid approach truly emulates the behavior of human experts in this area, resulting in higher success rates than any of the individual implemented techniques. In the final classification system, the most suitable methods are selected for each individual spectrum, which implies a remarkable contribution to the automatic classification process.

LONG-TERM RISK CLASS MIGRATIONS OF NON-BANKRUPT AND BANKRUPT ENTERPRISES

This paper investigates how the process of going bankrupt can be recognized much earlier by enterprises than by traditional forecasting models. The presented studies focus on the assessment of credit risk classes and on determination of the differences in risk class migrations between non-bankrupt enterprises and future insolvent firms. For this purpose, the author has developed a model of a Kohonen artificial neural network to determine six different classes of risk. Long-term analysis horizon of 15 years before the enterprises went bankrupt was conducted. This long forecasting horizon allows one to identify, visualize and compare the intensity and pattern of changes in risk classes during the 15-year trajectory of development between two separate groups of companies (150 bankrupt and 150 non-bankrupt firms). The effectiveness of the forecast of the developed model was compared to three popular statistical models that predict the financial failure of companies. These studies represent one of the first attempts in the literature to identify the long-term behavioral pattern differences between future “good” and “bad” enterprises from the perspective of risk class migrations.

Determination of plant communities based on bryophytes: The combined use of Kohonen artificial neural network and indicator species analysis

Abstract Mosses and liverworts are good bioindicators, not only of environmental pollution and the substrates on which they develop, but also of the plant communities in which they grow. The present study classifies bryological releves and identifies epigeic species significantly associated with three forest phytocoenoses important for Central Poland: Tilio-Carpinetum with Abies alba, Abietetum polonicum and managed forests; importantly, the study takes an approach using a Kohonen artificial neural network (i.e. self-organising map, SOM), its first use in bryological studies, together with Indicator Species Analysis based on IndVal index. The number of species significantly associated with natural forests was found to be 2–3 times higher than in managed forests, which reflected the differences in the quality of forest habitats. Additionally, the approach not only confirmed that some taxa play a diagnostic role (e.g.: Pleurozium schreberi, Dicranum polysetum, Dicranum scoparium, Thuidium tamariscinum, Atrichum undulatum, Eurhynchium angustirete) but also indicated new species which may have diagnostic potential for the studied communities. Our findings demonstrate that the epigeic bryophyte layer can play an important role in separating the studied phytocoenoses, and thus in the syntaxonomy of forest communities.

Neural-network analysis of socio-medical data to identify predictors of undiagnosed hepatitis C virus infections in Germany (DETECT)

BackgroundChronic hepatitis C virus (HCV)-infection is a slowly debilitating and potentially fatal disease with a high estimated number of undiagnosed cases. Given the major advances in the treatment, detection of unreported infections is a consequential step for eliminating hepatitis C on a population basis. The prevalence of chronic hepatitis C is, however, low in most countries making mass screening neither cost effective nor practicable.MethodsWe used a Kohonen artificial neural network (ANN) to analyze socio-medical data of 1.8 million insurants for predictors of undiagnosed HCV infections. The data had to be anonymized due to ethical requirements. The network was trained with variables obtained from a subgroup of 2544 patients with confirmed hepatitis C-virus (HCV) infections excluding variables directly linked to the diagnosis of HCV. All analyses were performed using the data mining solution “RayQ”. Training results were visualized three-dimensionally and the distributions and characteristics of the clusters were explored within the map.ResultsAll 2544 patients with confirmed chronic HCV diagnoses were localized in a clearly defined cluster within the Kohonen self-organizing map. An additional 2217 patients who had not been diagnosed with hepatitis C co-localized to the same cluster, indicating socio-medical similarities and a potentially elevated risk of infection. Several factors including, age, diagnosis codes and drug prescriptions acted only in conjunction as predictors of an elevated HCV risk.ConclusionsThis ANN approach may allow for a more efficient risk adapted HCV-screening. However, further validation of the prediction model is required.

Deceptive ecological status of urban streams and rivers—evidence from diatom indices

AbstractUrbanization directly affects the ecological status of streams and rivers: Their beds and drainage basins are often heavily transformed and their habitat conditions disturbed. The aim of the study was to evaluate the extent of human impact in streams in the city of Łódź, Central Poland (~700,000 inhabitants), by pattern recognition analysis of benthic diatom assemblage samples using a Kohonen artificial neural network, principal component analysis, and indicator species analysis. The study also determined the ecological status of the streams using five diatom indices: Multimetric Diatom Index, Specific Pollution Sensitivity Index, Generic Diatom Index, Biological Diatom Index, and Trophic Diatom Index. The samples demonstrated considerable variation in ecological status. Examples of poor and bad ecological status were observed, clearly resulting from strong human pressure; however, good status and moderate status were also recorded, both at natural sites and, unexpectedly, those which were severely transformed, whose beds were covered with concrete and which were located downstream from combined sewer overflows. Such sites demonstrated high water conductivity of up to 1440 μS/cm, but this was accompanied by other more favorable conditions: The water flowed from upstream stretches enclosed in downtown underground canals with a fast current, low water temperature, and high average concentration of dissolved oxygen (approximately 10 mg/dm3). These conditions had a mitigating effect, shaping the diatom assemblages to a specific form with the highest species richness and the highest number of indicator taxa observed in the study. In summary, the hydromorphological transformation of streambeds may result in seemingly good ecological status at heavily degraded sites. Therefore, conclusions based on diatom‐based biomonitoring of urban streams and rivers should be drawn with caution.

Direct Analysis in Real Time-Mass Spectrometry and Kohonen Artificial Neural Networks for Species Identification of Larva, Pupa and Adult Life Stages of Carrion Insects.

Species determination of the various life stages of flies (Order: Diptera) is challenging, particularly for the immature forms, because analogous life stages of different species are difficult to differentiate based on morphological features alone. It is demonstrated here that direct analysis in real time-high-resolution mass spectrometry (DART-HRMS) combined with supervised Kohonen Self-Organizing Maps (SOM) enables accomplishment of species-level identification of larva, pupa, and adult life stages of carrion flies. DART-HRMS data for each life stage were acquired from analysis of ethanol suspensions representing Calliphoridae, Phoridae, and Sarcophagidae families, without additional sample preparation. After preprocessing, the data were subjected to a combination of minimum Redundancy Maximal Relevance (mRMR) and Sparse Discriminant Analysis (SDA) methods to select the most significant variables for creating accurate SOM models. The resulting data were divided into training and validation sets and then analyzed by the SOM method to define the proper discrimination models. The 5-fold venetian blind cross-validation misclassification error was below 7% for all life stages, and the validation samples were correctly identified in all cases. The multiclass SOM model also revealed which chemical components were the most significant markers for each species, with several of these being amino acids. The results show that processing of DART-HRMS data using artificial neural networks (ANNs) based on the Kohonen SOM approach enables rapid discrimination and identification of fly species even for the immature life stages. The ANNs can be continuously expanded to include a larger number of species and can be used to screen DART-HRMS data from unknowns to rapidly determine species identity.

Kohonen artificial neural networks and the IndVal index as supplementary tools for the quantitative analysis of palaeoecological data

Kohonen artificial neural networks and the IndVal index as supplementary tools for the quantitative analysis of palaeoecological data

Indicator macroinvertebrate species in a temporary Mediterranean river: Recognition of patterns in binary assemblage data with a Kohonen artificial neural network

Current classifications used in bioassessment programs, as defined by the Water Framework Directive (WFD), do not sufficiently capture the variability present in temporary Mediterranean streams. This may result in inaccurate evaluation of the water quality biological metrics and difficulties in setting reference conditions. The aim of the study was to examine if aquatic invertebrate data of increased taxonomical resolution but expressed on a binary abundance (frequent/rare) scale and referring to good bioindicator species only suffice to indicate clear gradients in water courses with high natural variability such as intermittent Mediterranean streams. Invertebrate samples were collected from 74 sites in the Quarteira River basin, located in southern Portugal. Their classification with the use of a Kohonen artificial neural network (i.e., self-organising map, SOM) resulted in five categories. The variables that drove this categorization were primarily altitude, temperature and conductivity, but also type of substrate, riparian cover and percentage of riffles present. According to the indicator species analysis (ISA), almost all the studied taxa were significantly associated with certain SOM categories except for the category that included sites with disrupted flow regime. The SOM and ISA allowed us to effectively recognize biotic and abiotic patterns. Combined application of both methods may thus greatly enhance the effectiveness and precision of biological surveillance and establish reference sites for specific channel units in streams with high natural variability such as intermittent Mediterranean streams.