Analysis of macrobenthic communities in Flanders,Belgium, using a stepwise input variable selection procedure with artificial neural networks

The effect of environmental conditions on river macrobenthic communities was studied using a dataset consisting of 343 sediment samples from unnavigable watercourses in Flanders, Belgium. Artificial neural network models were used to analyse the relation among river characteristics and macrobenthic communities. The dataset included presence or absence of macroinvertebrate taxa and 12 physicochemical and hydromorphological variables for each sampling site. The abiotic variables served as input for the artificial neural networks to predict the macrobenthic community. The effects of the input variables on model performance were assessed in order to identify the most diagnostic river characteristics for macrobenthic community composition. This was done by consecutively eliminating the least important variables and, when beneficial for model performance, adding previously removed ones again. This stepwise input variable selection procedure was tested not only on a model predicting the entire macrobenthic community, but also on three models, each predicting an individual taxon. Additionally, during each step of the stepwise leave-one-out procedure, a sensitivity analysis was performed to determine the response of the predicted macroinvertebrate taxa to the input variables applied. This research illustrated that a combination of input variable selection with sensitivity analyses can contribute to the development of reliable and ecologically relevant ANN models. The river characteristics predicting presence or absence of the benthic macroinvertebrates best were the Julian day, conductivity, and dissolved oxygen content. These conditions reflect the importance of discharges of untreated wastewater that occurred during the period of investigation in nearly all Flemish rivers.     

An approach to multi-class imbalanced problem in ecology using machine learning

Ecologists collect their data manually by visiting multiple sampling sites. Since there can be multiple species in the multiple sampling sites, manually classifying them can be a daunting task. Much work in literature has focused mostly on statistical methods for classification of single species and very few studies on classification of multiple species. In addition to looking at multiple species, we noted that classification of multiple species result in multi-class imbalanced problem. This study proposes to use machine learning approach to classify multiple species in population ecology. In particular, bagging (random forests (RF) and bagging classification trees (bagCART)) and boosting (boosting classification trees (bootCART), gradient boosting machines (GBM) and adaptive boosting classification trees (AdaBoost)) classifiers were evaluated for their performances on imbalanced multiple fish species dataset. The recall and F1-score performance metrics were used to select the best classifier for the dataset. The bagging classifiers (RF and bagCART) achieved high performances on the imbalanced dataset while the boosting classifiers (bootCART, GBM and AdaBoost) achieved lower performances on the imbalanced dataset. We found that some machine learning classifiers were sensitive to imbalanced dataset hence they require data resampling to improve their performances. After resampling, the bagging classifiers (RF and bagCART) had high performances compared to boosting classifiers (bootCART, GBM and AdaBoost). The strong performances shown by bagging classifiers (RF and bagCART) suggest that they can be used for classifying multiple species in ecological studies.     

Modeling Dynamic Systems with Efficient Ensembles of Process-Based Models

Ensembles are a well established machine learning paradigm, leading to accurate and robust models, predominantly applied to predictive modeling tasks. Ensemble models comprise a finite set of diverse predictive models whose combined output is expected to yield an improved predictive performance as compared to an individual model. In this paper, we propose a new method for learning ensembles of process-based models of dynamic systems. The process-based modeling paradigm employs domain-specific knowledge to automatically learn models of dynamic systems from time-series observational data. Previous work has shown that ensembles based on sampling observational data (i.e., bagging and boosting), significantly improve predictive performance of process-based models. However, this improvement comes at the cost of a substantial increase of the computational time needed for learning. To address this problem, the paper proposes a method that aims at efficiently learning ensembles of process-based models, while maintaining their accurate long-term predictive performance. This is achieved by constructing ensembles with sampling domain-specific knowledge instead of sampling data. We apply the proposed method to and evaluate its performance on a set of problems of automated predictive modeling in three lake ecosystems using a library of process-based knowledge for modeling population dynamics. The experimental results identify the optimal design decisions regarding the learning algorithm. The results also show that the proposed ensembles yield significantly more accurate predictions of population dynamics as compared to individual process-based models. Finally, while their predictive performance is comparable to the one of ensembles obtained with the state-of-the-art methods of bagging and boosting, they are substantially more efficient.     

Use of different approaches to model presence/absence of Salmo marmoratus in Piedmont (Northwestern Italy)

In Piedmont (Italy) the environmental changes due to human impact have had profound effects on rivers and their inhabitants. Thus, it is necessary to develop practical tools providing accurate ecological assessments of river and species conditions. We focus our attention on Salmo marmoratus, an endangered salmonid which is characteristic of the Po river system in Italy. In order to contribute to the management of the species, four different approaches were used to assess its presence: discriminant function analysis, logistic regression, decision tree models and artificial neural networks. Either all the 20 environmental variables measured in the field or the 7 coming from feature selection were used to classify sites as positive or negative for S. marmoratus. The performances of the different models were compared. Discriminant function analysis, logistic regression, and decision tree models (unpruned and pruned) had relatively high percentages of correctly classified instances. Although neither tree-pruning technique improved the reliability of the models significantly, they did reduce the tree complexity and hence increased the clarity of the models. The artificial neural network (ANN) approach, especially the model built with the 7 inputs coming from feature selection, showed better performance than all the others. The relative contribution of each independent variable to this model was determined by using the sensitivity analysis technique. Our findings proved that the ANNs were more effective than the other classification techniques. Moreover, ANNs achieved their high potentials when they were applied in models used to make decisions regarding river and conservation management.     

The Application of Predictive Modelling for Determining Bio-Environmental Factors Affecting the Distribution of Blackflies (Diptera: Simuliidae) in the Gilgel Gibe Watershed in Southwest Ethiopia

Blackflies are important macroinvertebrate groups from a public health as well as ecological point of view. Determining the biological and environmental factors favouring or inhibiting the existence of blackflies could facilitate biomonitoring of rivers as well as control of disease vectors. The combined use of different predictive modelling techniques is known to improve identification of presence/absence and abundance of taxa in a given habitat. This approach enables better identification of the suitable habitat conditions or environmental constraints of a given taxon. Simuliidae larvae are important biological indicators as they are abundant in tropical aquatic ecosystems. Some of the blackfly groups are also important disease vectors in poor tropical countries. Our investigations aim to establish a combination of models able to identify the environmental factors and macroinvertebrate organisms that are favourable or inhibiting blackfly larvae existence in aquatic ecosystems. The models developed using macroinvertebrate predictors showed better performance than those based on environmental predictors. The identified environmental and macroinvertebrate parameters can be used to determine the distribution of blackflies, which in turn can help control river blindness in endemic tropical places. Through a combination of modelling techniques, a reliable method has been developed that explains environmental and biological relationships with the target organism, and, thus, can serve as a decision support tool for ecological management strategies.     

Comparison of the Abiotic Preferences of Macroinvertebrates in Tropical River Basins

We assessed and compared abiotic preferences of aquatic macroinvertebrates in three river basins located in Ecuador, Ethiopia and Vietnam. Upon using logistic regression models we analyzed the relationship between the probability of occurrence of five macroinvertebrate families, ranging from pollution tolerant to pollution sensitive, (Chironomidae, Baetidae, Hydroptilidae, Libellulidae and Leptophlebiidae) and physical-chemical water quality conditions. Within the investigated physical-chemical ranges, nine out of twenty-five interaction effects were significant. Our analyses suggested river basin dependent associations between the macroinvertebrate families and the corresponding physical-chemical conditions. It was found that pollution tolerant families showed no clear abiotic preference and occurred at most sampling locations, i.e. Chironomidae were present in 91%, 84% and 93% of the samples taken in Ecuador, Ethiopia and Vietnam. Pollution sensitive families were strongly associated with dissolved oxygen and stream velocity, e.g. Leptophlebiidae were only present in 48%, 2% and 18% of the samples in Ecuador, Ethiopia and Vietnam. Despite some limitations in the study design, we concluded that associations between macroinvertebrates and abiotic conditions can be river basin-specific and hence are not automatically transferable across river basins in the tropics.     

BagBoosting for tumor classification with gene expression data

MOTIVATION: Microarray experiments are expected to contribute significantly to the progress in cancer treatment by enabling a precise and early diagnosis. They create a need for class prediction tools, which can deal with a large number of highly correlated input variables, perform feature selection and provide class probability estimates that serve as a quantification of the predictive uncertainty. A very promising solution is to combine the two ensemble schemes bagging and boosting to a novel algorithm called BagBoosting. RESULTS: When bagging is used as a module in boosting, the resulting classifier consistently improves the predictive performance and the probability estimates of both bagging and boosting on real and simulated gene expression data. This quasi-guaranteed improvement can be obtained by simply making a bigger computing effort. The advantageous predictive potential is also confirmed by comparing BagBoosting to several established class prediction tools for microarray data. AVAILABILITY: Software for the modified boosting algorithms, for benchmark studies and for the simulation of microarray data are available as an R package under GNU public license at http://stat.ethz.ch/~dettling/bagboost.html.     

Effect of alien riparian vegetation and its removal on a highly endemic river macroinvertebrate community

Invasive alien trees along river banks can reduce indigenous biodiversity, while their removal can restore it. We assessed here family- and species-level responses of river benthic macroinvertebrate assemblages to three riparian vegetation types (natural, alien trees, cleared of alien trees) in the Cape Floristic Region biodiversity hotspot. High species beta diversity of this highly endemic fauna meant that between-river, as well as seasonal effects, dominated assemblage patterns. SASS5, a qualitative, rapid bioassessment technique, based on the sensitivity of the families present, was used as a measure of river health and, indirectly, of water quality. SASS indicated a decline in water quality conditions after alien clearing, a likely response to the greater insolation and apparent erosion of cleared banks, resulting in elevated temperatures and suspended solids and lowered oxygen levels. Overall, cleared and natural sites were more similar to each other than to alien sites, suggesting some post-clearing recovery. However, many sensitive, endemic taxa survived in alien-invaded sites, and more than in the natural sites. These endemic species made use of shady, cool, high-oxygen levels under the alien tree canopy. However, endemics declined in overall abundance in sites cleared of aliens, being replaced by more tolerant, widespread taxa. Clearance of the alien trees opened up the rivers to sunny conditions, which had a major impact on community composition. Vegetation types, oxygen levels and river width were important environmental variables affecting these macroinvertebrate responses. Re-establishment of invertebrate biodiversity matched that of indigenous vegetation, with the most sensitive endemic taxa only recovering after establishment of bushy indigenous and shade-producing fynbos. Therefore, for biodiversity conservation objectives to be achieved, it is essential that indigenous plants are maintained and encouraged during and after clearing to ensure the recovery of endemic and sensitive taxa.     

Macroinvertebrate indicators of lake acidification: analysis of monitoring data from UK, Norway and Sweden

Although the acid sensitivity of many invertebrate species in lakes is well known, methods for assessment of lake acidification based on macroinvertebrate samples are less developed than for rivers. This article analyses a number of existing metrics developed for assessment of river acidification, and evaluates their performance for assessment of lake acidification. Moreover, new species-based indicators of lake acidification were developed and tested. The selected dataset contains 668 samples on littoral macroinvertebrates from 427 lakes with almost 60% of the samples from Sweden and the rest from UK and Norway. Flexible, non-parametric regression models were used for explorative analyses of the pressure–response relationships. The metrics have been assessed according to their response to pH, the degree of non-linearity of the response and the influence of humic compounds. Acid-sensitive metrics often showed a threshold in response to pH between 5.8 and 6.5. Highly acid-tolerant metrics were typically dominant across the whole pH range. Humic level had a positive effect for most acid-sensitive metrics. Generally, most metrics showed a more non-linear response pattern for the humic lakes than for clear lakes. The significant relationship between these macroinvertebrate metrics and acidification shows that there is a potential for developing further the assessment systems for ecological quality of lakes based on these metrics, although the metrics explained a low % of the variation (<30%). In order to improve the predictive power of the biotic metrics across the acidified part of Europe, further harmonization and standardisation of sampling effort and taxa identification are needed.     

Spatio-temporal changes of benthic macroinvertebrates in a cold Arkansas tailwater

This paper investigates spatial, seasonal and long-term changes in benthic macroinvertebrates in riffles of a cold tailwater. Cold tailwaters initially disrupt previously existing macroinvertebrate assemblages, but little is known about the long-term biological effects of a stable cold thermal regime. Assemblages at an upstream and downstream site of the Little Red River, Arkansas were investigated almost 30 years apart (1971 and 1999). Based upon published literature demonstrating the stability of benthic assemblages within unaltered environments, we predicted that the assemblages would be similar for each variable investigated. The benthic macroinvertebrate assemblages can be characterized as low diversity, with a total of 17 taxa identified. Isopods and Diptera comprised ~80% of all individuals. Other than chironomids, insects and particularly EPT taxa were poorly represented. Recent macroinvertebrate densities were significantly greater compared to the historical study period for the downstream site. Assemblage comparisons revealed moderate differences between study periods. Macroinvertebrate density was significantly greater upstream than downstream in the 1971 study period, yet taxa richness was significantly greater downstream for both study periods. Faunal composition was significantly different for upstream and downstream sites. Seasonal differences in numerical standing crop were identified for the 1971 upstream and 1999 downstream data sets. Low to moderate levels of seasonal, spatial and historical variation among benthic macroinvertebrate assemblages were attributed to environmental (temperature and flow) stability. The lack of aquatic insects other than chironomids over a 30-year period is indicative of the extreme constraints placed upon insect development within this cold regulated river.     

Selection of indicative taxa for river habitats: a case study on benthic macroinvertebrates using indicator species analysis and the random forest methods

The aim of the study was to evaluate the exclusivity and/or preference of macroinvertebrate taxa for river habitats. Indicator species analysis and random forests methods were applied to the data set of macroinvertebrate samples taken from 58 sampling points. Samples were classified according to habitat types defined by the position in a river channel and local hydraulic characteristics. 86 macroinvertebrate taxa were included in the analyses. High indicative values for habitats (importance value ≥50 and/or indicator value ≥40) were identified for 26 taxa. The results of both methods can be considered similar. Merged habitats of channel margin (margin of main channel and side arms) were mainly defined by “negative” indicator taxa (correct classification of given samples was caused by non-occurrence and low abundances of certain taxa in this habitat). In general, there was only a small group of taxa preferring these habitats. Taxa were not fully habitat specific because they mostly occurred in two or three habitat types. This could be the result of autecological plasticity of individual taxa and the connectivity among habitats. According to the experience from this case study, it can be concluded that both random forests and IndVal methods are suitable for the detection of indicative species, and random forests method has some additional advantages.     

The use of benthic macroinvertebrate families for river quality assessment in Lithuania

The aim of this study was to evaluate the influence of environmental factors on the distribution of macroinvertebrate taxa in different sized lowland Lithuanian rivers. A secondary aim was to assess ecological river quality and to determine the most suitable biotic index. A final aim was to determine the most appropriate macroinvertebrate families for river quality assessment in Lithuania. Species composition and quantitative characteristics of benthic macroinvertebrate communities have been investigated using standard kick-sampling method by a standard hand net in 24 different river sites in spring. Physical and chemical environmental parameters were measured in the same study site as the macroinvertebrate sampling. A total of 186 taxa representing 66 families or higher taxonomic ranks of benthic macroinvertebrates have been identified. Water temperature and current velocity influenced the highest number of ivestigated families. Seven of the most tolerant and eleven of the most sensitive macroinvertebrate taxa for hydrochemical parameters related with organic pollution were determined. The DSFI method was founded to be the best index for assessment of ecological status for Lithuanian rivers until more accurate estimation method will be created.     

Effect of Seasonal Changes on Predictive Model Assessments of Streams Water Quality with Macroinvertebrates

In this study we investigate how seasonal variability in aquatic macroinvertebrate communities affects the performance of a predictive model developed to assess environmental quality. Macroinvertebrates were sampled from nine not visibly disturbed sites located in different streams of the Mondego catchment across a full year. Organisms were identified to the lowest practicable taxonomic level and their abundances recorded at three taxonomic levels (order, family and lowest level). The seasonal samples were examined with regard to seasonal variation using three predictive models at order, family and lowest taxonomic level. The models showed increasing effect of seasonal changes across taxonomic levels, from order to the lowest level. When using the current models samples should be taken in the same season as the reference sites were sampled. Furthermore, data from more reference sites should be added to the model in order to encompass sufficient natural variation and allow the use of the model in different seasons. (© 2006 WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim)     

The influences of climatic variation and vegetation on stream biota: lessons from the Big Dry in southeastern Australia

Aquatic biodiversity faces increasing threats from climate change, escalating exploitation of water and land use intensification. Loss of vegetation in catchments (= watersheds) has been identified as a substantial problem for many river basins, and there is an urgent need to better understand how climate change may interact with changes in catchment vegetation to influence the ecological condition of freshwater ecosystems. We used 20 years of biological monitoring data from Victoria, southeastern Australia, to explore the influences of catchment vegetation and climate on stream macroinvertebrate assemblages. Southeastern Australia experienced a severe drought from 1997 to 2009, with reductions of stream flows >50% in some areas. The prolonged drying substantially altered macroinvertebrate assemblages, with reduced prevalence of many flow‐dependent taxa and increased prevalence of taxa that are tolerant of low‐flow conditions and poor water quality. Stream condition, as assessed by several commonly used macroinvertebrate indices, was consistently better in reaches with extensive native tree cover in upstream catchments. Prolonged drought apparently caused similar absolute declines in macroinvertebrate condition indices regardless of vegetation cover, but streams with intact catchment and riparian vegetation started in better condition and remained so throughout the drought. The largest positive effects of catchment tree cover on both water quality and macroinvertebrate assemblages occurred above a threshold of ca. 60% areal tree cover in upstream catchments and in higher rainfall areas. Riparian tree cover also had positive effects on macroinvertebrate assemblages, especially in warmer catchments. Our results suggest that the benefits of extensive tree cover via improved water quality and in‐channel habitat persist during drought and show the potential for vegetation management to reduce negative impacts of climatic extremes for aquatic ecosystems.     

Longitudinal variation in the composition of the benthic macroinvertebrate fauna of a typical North coast Jamaican river

Benthic macroinvertebrate fauna plays a major role in river ecosystems, especially those of tropical islands. Since there is no information on the distribution of benthic invertebrates along a Jamaican river, we report here on the composition of the benthic fauna of the Buff Bay river, on the Northern coast of Jamaica. A total of 14 samples were collected from five sites, using kick nets and a Surber sampler, between May 1997 and October 1998. We also examined the applicability of the rhithron/potamon model, and some of the premises of the River Continuum Concept (RCC) in relation to the distribution of invertebrate taxa. The results showed a total of 38 taxa of identified invertebrates. A group of dominant taxa, composed mainly of immature stages of insects, occurred at all sites. Two notable characteristics of the river were the absence of a true potamonic fauna and the low representation of the shredder functional feeding group in the community We conclude that, while there was minor variation in the composition of the benthic macroinvertebrate fauna among the sites, this was a response to local conditions within the river system. The characteristics of the community did not conform to either of the models.     

Fish distribution predictions from different points of view: comparing associative neural networks,geostatistics and regression models

Accurate prediction of species distributions based on sampling and environmental data is essential for further scientific analysis, such as stock assessment, detection of abundance fluctuation due to climate change or overexploitation, and to underpin management and legislation processes. The evolution of computer science and statistics has allowed the development of sophisticated and well-established modelling techniques as well as a variety of promising innovative approaches for modelling species distribution. The appropriate selection of modelling approach is crucial to the quality of predictions about species distribution. In this study, modelling techniques based on different approaches are compared and evaluated in relation to their predictive performance, utilizing fish density acoustic data. Generalized additive models and mixed models amongst the regression models, associative neural networks (ANNs) and artificial neural networks ensemble amongst the artificial neural networks and ordinary kriging amongst the geostatistical techniques are applied and evaluated. A verification dataset is used for estimating the predictive performance of these models. A combination of outputs from the different models is applied for prediction optimization to exploit the ability of each model to explain certain aspects of variation in species acoustic density. Neural networks and especially ANNs appear to provide more accurate results in fitting the training dataset while generalized additive models appear more flexible in predicting the verification dataset. The efficiency of each technique in relation to certain sampling and output strategies is also discussed.     

Macroinvertebrate community dynamics in a temperate European Atlantic river. Do they conform to general ecological theory?

Spatial and temporal dynamics of macroinvertebrate communities have usually been linked to several environmental and anthropic factors. The aim of this study is to elucidate how important are these factors in structuring macroinvertebrate communities from temperate regions. Regarding the macroinvertebrate number of taxa, the Habitat Template Model, the Dynamic Equilibrium Hypothesis and the Intermediate Disturbance Hypothesis will be tested in order to know how important the diversity of instream elements and the hydrological disturbance frequency are in defining the macroinvertebrate taxonomic richness. Thus, the structure and composition of macroinvertebrate communities were analysed in nine sites of the Pas River basin, a temperate Atlantic basin in northern Spain, during winter, spring, summer and autumn 2005, together with water physicochemical and environmental characteristics. Macroinvertebrate abundance increased downstream and during summer, probably favoured by lower hydraulic stress and water organic enrichment. As predicts the Habitat Template Model, the macroinvertebrate number of taxa was related to habitat heterogeneity. However, no clear relationship amongst macroinvertebrate richness and water quality was found. The macroinvertebrate taxonomic richness did not correspond exactly with the Dynamic Equilibrium Hypothesis and the Intermediate Disturbance Hypothesis because it was relatively high in the absence of hydrological disturbances (summer). Thus, disturbance events may play a secondary role in determining the seasonal dynamic of the number of taxa. However, hydrological disturbances can be considered the most important factors explaining the seasonal pattern of macroinvertebrate abundance. On the other hand, spatial patterns of macroinvertebrate community structure and composition were mainly determined by resource availability, hydraulic conditions, habitat heterogeneity and human alterations, whilst hydrological predictability and resource availability might play a major role in determining seasonal dynamics.     

Description and evaluation of a sampling strategy for macroinvertebrate communities in Basque rivers (Spain)

This paper describes the results of a pilot study on the macroinvertebrate communities occurring in a fast-flowing river in Northern Spain. The objective was to devise a sampling strategy for use on a series of Basque rivers included in a site classification study. Two contrasted sites were chosen for the pilot-study on the River Lea (Bizkaia), at the headwaters and near the mouth. Taxon richness and distribution across the river was examined using two sampling techniques at each site. It was concluded that, ideally, timed sampling units should be collected at intervals across a river transect in order to acquire a representative list of taxa. In addition, adequate sampling in marginal areas was stressed because a number of taxa appeared to be exclusive to marginal areas. Sampling location and total sampling effort were also shown to be a major consideration when taxon lists were acquired for the application of routine biological surveillance techniques.     

长白山源头溪流底栖动物群落结构季节动态

源头溪流是河流生态系统物质循环和能量流动的重要区域,对底栖动物的生物多样性维持具有重要意义。目前,针对我国源头溪流底栖动物群落结构的研究薄弱,对长白山源头溪流底栖动物季节动态的研究尚未见报道。采用野外原位定量取样的方法,力图阐释长白山源头溪流底栖动物群落结构的季节动态特征及其主要环境驱动因子。研究结果表明:(1)共计采集底栖动物90个分类单元,隶属于3纲9目38科。其中,水生昆虫85属,占绝对优势。底栖动物群落结构的季节动态明显,底栖动物密度及多样性在夏、秋季显著高于冬季和春季,并在冬季达到最低值。(2)底栖动物功能摄食类群以收集者占优势,其次为撕食者、捕食者和刮食者,滤食者相对丰度最低。不同功能摄食类群的季节动态不一致,但密度和物种丰富度整体表现为秋季最高。(3)水温、凋落叶分布和流速是长白山源头溪流底栖动物群落结构季节动态的主要环境驱动因子。本研究可为长白山源头溪流后续相关研究及长白山松花江水系生态修复提供基础数据支持及参考。