What drives the ecological quality of surface waters? A review of 11 predictive modeling tools.

What policy is needed to ensure that good-quality water is available for both people's needs and the environment? The EU Water Framework Directive (WFD), which came into force in 2000, established a framework for the assessment, management, protection and improvement of the status of water bodies across the European Union. However, recent reviews show that the ecological status of the majority of surface waters in the EU does not meet the requirement of good status. Thus, it is an important question what measures water management authorities should take to improve the ecological status of their water bodies. To find concrete answers, several institutes in the Netherlands cooperated to develop a software tool, the WFD Explorer, to assist water managers in selecting efficient measures. This article deals with the development of prediction tools that allow one to calculate the effect of restoration and mitigation measures on the biological quality, expressed in terms of Ecological Quality Ratios (EQRs). To find the ideal modeling tool we give a review of 11 predictive models: 10 models from the field of Machine Learning and, additionally, the Multiple Regression model. We present our results in terms of a 'prediction-interpretation competition'. All these models were tested in a multiple-stressor setting: the values of 15 stressors (or steering factors) are available to predict the EQR values of four biological quality elements (phytoplankton, other aquatic flora, benthic invertebrates and fish). Analyses are based on 29 data sets from various water clusters (streams, ditches, lakes, channels). All 11 models were ranked by their predictive performance and their level of model transparency. Our review shows a trade-off between these two aspects. Models that have the best EQR prediction performance show non-transparent model structures. These are Random Forest and Boosting. However, models with low prediction accuracies show transparent response relationships between EQRs on the one hand and individual steering factors on the other hand. These models are Multiple Regression, Regression Trees and Product Unit Neural Networks. To acknowledge both aspects of model quality - predictive power and transparency - we recommend that models from both groups are implemented in the WFD Explorer software.

Intraguild predation behaviour of ladybirds in semi-field experiments explains invasion success of Harmonia axyridis.

Harmonia axyridis has been introduced as a biological control agent in Europe and the USA. Since its introduction, it has established and spread, and it is now regarded as an invasive alien species. It has been suggested that intraguild predation is especially important for the invasion success of H. axyridis. The aim of this study was to compare the intraguild predation behaviour of three ladybird species (Coccinella septempunctata, Adalia bipunctata, and H. axyridis). Predation behaviour was investigated in semi-field experiments on small lime trees (Tilia platyphyllos). Two fourth-instar larvae placed on a tree rarely made contact during 3-hour observations. When placed together on a single leaf in 23%-43% of the observations at least one contact was made. Of those contacts 0%-27% resulted in an attack. Harmonia axyridis attacked mostly heterospecifics, while A. bipunctata and C. septempunctata attacked heterospecifics as often as conspecifics. In comparison with A. bipunctata and C. septempunctata, H. axyridis was the most successful intraguild predator as it won 86% and 44% of heterospecific battles against A. bipunctata and C. septempunctata respectively, whilst A. bipunctata won none of the heterospecific battles and C. septempunctata won only the heterospecific battles against A. bipunctata. Coccinella septempunctata dropped from a leaf earlier and more often than the other two species but was in some cases able to return to the tree, especially under cloudy conditions. The frequency with which a species dropped did not depend on the species the larva was paired with. The results of these semi-field experiments confirm that H. axyridis is a strong intraguild predator as a consequence of its aggressiveness and good defence against predation from heterospecific species. The fact that H. axyridis is such a strong intraguild predator helps to explain its successful establishment as invasive alien species in Europe and the USA.

Towards a renewed research agenda in ecotoxicology.

New concerns about biodiversity, ecosystem services and human health triggered several new regulations increasing the need for sound ecotoxicological risk assessment. The PEER network aims to share its view on the research issues that this challenges. PEER scientists call for an improved biologically relevant exposure assessment. They promote comprehensive effect assessment at several biological levels. Biological traits should be used for Environmental risk assessment (ERA) as promising tools to better understand relationships between structure and functioning of ecosystems. The use of modern high throughput methods could also enhance the amount of data for a better risk assessment. Improved models coping with multiple stressors or biological levels are necessary to answer for a more scientifically based risk assessment. Those methods must be embedded within life cycle analysis or economical models for efficient regulations. Joint research programmes involving humanities with ecological sciences should be developed for a sound risk management.

Ecological vulnerability in wildlife: application of a species-ranking method to food chains and habitats.

Nature development in The Netherlands is often planned on contaminated soils or sediments. This contamination may present a risk for wildlife species desired at those nature development sites and must be assessed by specific risk assessment methods. In a previous study, we developed a method to predict ecological vulnerability in wildlife species by using autecological data and expert judgment; in the current study, this method is further extended to assess ecological vulnerability of food chains and terrestrial and aquatic habitats typical for The Netherlands. The method is applied to six chemicals: Cd, Cu, Zn, dichlorodiphenyltrichloroethane, chlorpyrifos, and ivermectin. The results indicate that species in different food chains differ in vulnerability, with earthworm-based food chains the most vulnerable. Within and between food chains, vulnerability varied with habitat, particularly at low trophic levels. The concept of habitat vulnerability was applied to a case study of four different habitat types in floodplains contaminated with cadmium and zinc along the river Dommel, The Netherlands. The alder floodplain forest habitat contained the most vulnerable species. The differences among habitats were significant for Cd. We further conclude that the method has good potential for application in mapping of habitat vulnerability.

Estimation of in situ sediment-to-water fluxes of polycyclic aromatic hydrocarbons, polychlorobiphenyls and polybrominated diphenylethers.

Sediment-water fluxes of hydrophobic organic chemicals (HOC) may affect the quality of surface waters. Here, we present an approach to derive such fluxes from (a) in situ HOC concentration gradients measured with passive samplers and (b) mass transfer coefficients measured with a novel flux method using Empore disks. For eight undisturbed sediments, this method identified whether the sediment acted as a source or as a sink for HOCs. The analysis also identified which type of transport resistance governed sediment water exchange. For seven inland locations, exchange was limited by benthic boundary layer transport, showing no dependencies on sediment or chemical properties other than concentration. For one river mouth location, exchange was limited by slow in-bed intraparticle diffusion. A biphasic dual compartment radial diffusion model adequately described the data for this location. Fast desorption was interpreted as molecular diffusion retarded by microscale dual domain sorption to amorphous as well as black carbon (BC). Slow desorption was invariant with LogK(ow) and consistent with intraorganic matter diffusion through BC particles. Finally, it is discussed how these findings can be translated into a general framework for flux based exposure assessment.

Wildlife vulnerability and risk maps for combined pollutants.

Ecological risk and vulnerability maps can be used to improve the analysis of pollutant risks and communication to stakeholders. Often, such maps are made for one pollutant at the time. We used the results of wildlife vulnerability analysis, a novel trait-based risk assessment approach, to map overall vulnerability of habitats in Denmark to various metals and one insecticide. These maps were combined with maps of estimated soil concentrations for the same compounds divided by their Maximum Permissible Concentrations. This combination yielded relative risk maps that can be used to assess where the highest risk conditions to wildlife from these individual pollutants in Denmark occur (hot spot identification). In order to show how cumulative risk maps can be made, the maps of the individual pollutants were combined assuming different mechanisms of joint toxicity: no addition, concentration addition, antagonism and synergism. The study demonstrated that with an accurate set of geographical and ecological data one can use the results of vulnerability analysis to make relevant ecological risk maps that show hot spot areas for risks of single or cumulative risks from soil pollutants.

Contrasting sensitivities to toxicants of the freshwater amphipods Gammarus pulex and G. fossarum.

Amphipods are an important component of freshwater ecosystems. They are very often used in ecotoxicology, particularly the freshwater amphipod Gammarus pulex. However, there is scarce information on the sensitivity to toxicants of other species within the genus Gammarus. The present study aims to: (1) to compare sensitivities to ivermectin and cadmium between two species of freshwater amphipods (G. pulex and G. fossarum); (2) to compare sensitivities to these toxicants between juveniles and adults within each species; and (3) to assess whether the sensitivity to toxicants of these co-generic species is related with the wideness of their natural distribution area. Eight independent short-term bioassays (96 h) were conducted to assess sensitivity for ivermectin and cadmium for juvenile and adult life stages for each species. The LC50 (mortality) and EC50 (mortality plus immobility) were calculated to 48 and 96 h of continuous exposure. Our results showed that G. pulex was less tolerant to ivermectin than G. fossarum, the reverse being true for cadmium. In general, juveniles of both species were less tolerant to cadmium than adults. In the case of ivermectin, only for G. fossarum EC50 values were different between life stages. These results suggest that the risk assessment of toxicants to freshwater amphipods should include bioassays with the most sensitive species and life stage.

Ecological vulnerability in wildlife: an expert judgment and multicriteria analysis tool using ecological traits to assess relative impact of pollutants.

Nature development in The Netherlands often is planned on contaminated soils and sediments of former agricultural land and in floodplain areas; however, this contamination may present a risk to wildlife species desired at those nature development sites. Specific risk assessment methods are needed, because toxicological information is lacking for most wildlife species. The vulnerability of a species is a combination of its potential exposure, sensitivity to the type of pollutant, and recovery capacity. We developed a new method to predict ecological vulnerability in wildlife using autecological information. The analysis results in an ordinal ranking of vulnerable species. The method was applied to six representative contaminants: copper and zinc (essential metals, low to medium toxicity), cadmium (nonessential metal, high toxicity), DDT (persistent organic pesticide, high toxicity), chlorpyrifos (persistent organophosphate insecticide, high toxicity), and ivermectin (persistent veterinary pharmaceutical, low to medium toxicity). High vulnerability to the essential metals copper and zinc was correlated with soil and sediment habitat preference of a species and with r-strategy (opportunistic strategy suited for unstable environments). Increased vulnerability to the bioaccumulating substances cadmium and DDT was correlated with higher position of a species in the food web and with life span and K-strategy (equilibrium strategy suited for stable environments). Vulnerability to chlorpyrifos and ivermectin was high for species with a preference for soil habitats. The ecological vulnerability analysis has potential to further our abilities in risk assessment.

Development of a feeding behavioural bioassay using the freshwater amphipod Gammarus pulex and the Multispecies Freshwater Biomonitor.

The present study reports the development of a feeding behavioural bioassay using the Multispecies Freshwater Biomonitor (MFB). This device is based on the quadruple impedance conversion technique to record online different behaviours of animals. Animal movements in the water generate specific frequencies, and the MFB can estimate the percentage of time producing each frequency (from 0.5 to 8.5 Hz) by means of a stepwise discrete Fourier transformation. Two feeding behavioural bioassays were conducted in order to know the frequencies related to feeding behaviour of the freshwater amphipod Gammarus pulex. The first bioassay assessed the effects of food presence in the amphipod behaviour. The second bioassay assessed the effects of cadmium on the feeding activity (measured as leaf weight loss) and behaviour (swimming, ventilation, and feeding recorded through the MFB) of G. pulex in order to check the suitability of the developed method. The results of the first bioassay showed that the frequencies ranging from 3.0 to 4.5 Hz were highly correlated with feeding activity, especially 3.5 and 4.0 Hz. In the second bioassay, we found that cadmium reduced feeding and ventilation behaviours. Our study showed that the MFB can be used to record the feeding behaviour of G. pulex exposed to toxicants. The developed feeding behavioural bioassay allows an accurate and automatic assessment of several endpoints, including feeding, swimming and ventilation. However, the study of the complex behaviour of G. pulex using the MFB needs further research, since some behaviours seem to generate similar frequencies.

Avoidance of polycyclic aromatic hydrocarbon-contaminated sediments by the freshwater invertebrates Gammarus pulex and Asellus aquaticus.

Contamination of sediments is a serious problem in most industrialized areas. Sediments are often contaminated with trace metals and organic contaminants like polycyclic aromatic hydrocarbons (PAHs) and polychlorinated biphenyls (PCBs). Bioassays are often used to determine the effect of contaminants on biota. However, survival or growth may not be the most sensitive endpoints. Behavioral changes often occur at much lower concentrations. Our study aimed to assess the effect of PAHs on habitat choice of two common freshwater invertebrates, the amphipod Gammarus pulex and the isopod Asellus aquaticus. We spiked clean field sediment with a mixture of four PAHs, fluoranthene, pyrene, chrysene, and benzo[k]fluoranthene, to a total concentration of 30 mg PAH/kg dry weight. Both species were offered a choice between PAH-spiked sediments and clean sediments in laboratory experiments. Results show that both species avoid PAH-spiked sediment. Origin of the population, either from a clean reference site or from a polluted site, did not affect habitat choice of either species.