Robin A. Matthews

DESIGN AND ANALYSIS OF MULTISPECIES TOXICITY TESTS FOR PESTICIDE REGISTRATION

The community conditioning hypothesis describes ecological structures as historical, nonequilibrial, and by definition complex. Indeed, the historical nature of eco- logical structures is seen as the primary difference between single-species toxicity tests and multispecies test systems. Given the complex properties of ecological structures, mul- tispecies toxicity tests need to be designed accordingly with appropriate data analysis tools. Care must be taken to ensure that each replicate shares an identical history, or divergence will rapidly occur. Attempting to realize homogeneity by linear cross inoculation or waiting for an equilibrium state to occur assumes properties that ecological structures do not have. Data analysis must also incorporate the dynamic and hyperdimensional nature of ecological structures. Univariate analysis of individual variables denies the fundamental character of ecological structures as complex systems. A variety of methods, such as correspondence analysis, nonmetric multidimensional scaling, and nonmetric clustering and association analysis, are available to search for patterns and to test their relationships to experimental treatments. Visualization techniques including Space-Time Worms and redundancy analysis are also critical in attempting to understand the dynamic nature of these structures. Reliance upon the traditional analysis methods, such as ANOVA and the estimation of LOECs (lowest observable effects concentrations) or NOECs (no observable effects concentrations), com- parable to those of single-species toxicity tests, is to be blind to the unique and complex nature of multispecies toxicity tests. Fundamental design criteria for multispecies toxicity tests, data analysis, and interpretation are presented.

Multivariate analysis of the impacts of the turbine fuel JP-4 in a microcosm toxicity test with implications for the evaluation of ecosystem dynamics and risk assessment

Turbine fuels are often the only aviation fuel available in most of the world. Turbine fuels consist of numerous constituents with varying water solubilities, volatilities and toxicities. This study investigates the toxicity of the water soluble fraction (WSF) of JP-4 using the Standard Aquatic Microcosm (SAM). Multivariate analysis of the complex data, including the relatively new method of nonmetric clustering, was used and compared to more traditional analyses. Particular emphasis is placed on ecosystem dynamics in multivariate space.The WSF is prepared by vigorously mixing the fuel and the SAM microcosm media in a separatory funnel. The water phase, which contains the water-soluble fraction of JP-4 is then collected. The SAM experiment was conducted using concentrations of 0.0, 1.5 and 15% WSF. The WSF is added on day 7 of the experiments by removing 450 ml from each microcosm including the controls, then adding the appropriate amount of toxicant solution and finally bringing the final volume to 3 L with microcosm media. Analysis of the WSF was performed by purge and trap gas chromatography. The organic constituents of the WSF were not recoverable from the water column within several days of the addition of the toxicant. However, the impact of the WSF on the microcosm was apparent. In the highest initial concentration treatment group an algal bloom ensued, generated by the apparent toxicity of the WSF of JP-4 to the daphnids. As the daphnid populations recovered the algal populations decreased to control values. Multivariate methods clearly demonstrated this initial impact along with an additional oscillation seperating the four treatment groups in the latter segment of the experiment. Apparent recovery may be an artifact of the projections used to describe the multivariate data. The variables that were most important in distinguishing the four groups shifted during the course of the 63 day experiment. Even this simple microcosm exhibited a variety of dynamics, with implications for biomonitoring schemes and ecological risk assessments.

CyanoHABIT: A Novel Game to Identify Harmful Freshwater Algae

CyanoHABIT (Cyanobacterial Harmful Algal Bloom Identifying Technology) is a proposed learning technology designed to give the general public the ability to self-teach identification of potentially harmful bloom-forming algae (HABs) in freshwater lakes. The primary users will be adults who are interested in helping government agencies distinguish potentially toxic algae blooms from other non-toxic blooms. Toxic algae in freshwater lakes present a serious threat to public safety, and while many algal blooms are not toxic, confirming toxicity can be time consuming and expensive. Many states have only one agency that is able to monitor, sample, and test water for HABs, and some states have no resources for this task [1]. Fortunately, relatively few freshwater algae are capable of forming toxins, and distinguishing between benign algae blooms and potentially toxic ones is a task that can be learned in a short time by most adults. With a better educated public, the time and resources of the professional public services can be concentrated on the cases where they are needed most. We developed a gamified trainer for use on smartphones and personal computers to teach this skill to the general public. Our focus was on education, enabling the user to learn the distinguishing features of toxic algae, and not provide a flip-book of pictures of algae. Preliminary testing indicates that the software is enjoyable to use, and that the users do acquire a valuable skill from its use.

Nonmetric clustering: new approaches for ecological data

Ecological studies and multispecies ecotoxicological tests are based on the examination of a variety of physical, chemical and biological data with the intent of finding patterns in their changing relationships over time. The data sets resulting from such studies are often noisy, incomplete, and difficult to envision. We have developed machine learning and visualization software to aid in the analysis, modelling, and understanding of such systems. The software is based on nonmetric conceptual clustering, which attempts to analyze the data into clusters that are strongly associated with several measured parameters. Our analysis and visualization tools not only confirmed suspected ecological patterns, but revealed aspects of the data that were unnoticed by ecologists using conventional statistical techniques.<<ETX>>