One of the principle challenges that face applied ecologists is to isolate effects of interest (signal) from natural spatial or temporal variability (noise). Although a plethora of biological metrics and analysis techniques exist and are commonly used in environmental assessment programs, often surprisingly little is known of their inherent errors or weaknesses. The main objective of this Workpackage 12 is to compare and constrast the discriminatory power of selected indicator (organism) groups and metrics to detect change with known error. Further, we will develop a strategy or guidelines for determining which indicator group or groups are best for assessing the ecological effects of selected human-induced stressors.
The data that will be further evaluated in this Workpackage, are:
With these three data sources, a number of evaluations will be performed, partly in conjunction with Workpackage 11. The goals and methods of these evaluations are:
12.1 Which group is best suited for determining the ecological effects of selected stressors in a particular geographic region ?
Determining what group or groups of indicators are best suited for assessing the ecological effects of a known stressor requires knowledge of a number of indicator-inherent properties. Two factors, in particular, strongly affect our ability to detect ecological change:
Ideally, the indicator metric should reflect changes that are only associated with a single stressor and dose-response relationships should be derived. The expected change of selected indicator groups and metrics will be studied using data from this project and the previous EU-funded AQEM project. In brief, the response of single indicator groups and metrics will be assessed by determining organism-response relationships along known pollution gradients.
Organism group/metric response to known stress will be evaluated in a number of ways.
12.2 How are the different organism groups/methods affected by type I and II errors ?
The suitability of organism groups and metrics for the detection of ecological stressor is largely dependent on how they are affected by type I (false positive) and II (false negative) errors. For detecting the susceptibility to type I and II errors, the following steps will be performed:
12.3 Which organism group can be used on which spatial scale?
12.4 Which organism group is suited for early and late warnings ?
According to the conceptual model, organism groups or metrics vary in their suitability to detect stress. This conjecture will be tested using the results from 12.1, 12.2 and 12.3. We hypothesise that organism groups that have a high false positive frequency, due to natural (seasonal) variability, may be better suited as early-warning indicators of change. Conversely, organism groups or metrics that have low b error may also have slow response time. Hence, these may be classified as late-warning, but more certain indicators of change if/when change occurs. Using a suite of "complementary" organism groups, such as a combination of an early-warning metric (e.g. periphyton) with a slower, but statistically more reliable metric (e.g. macrophytes) may result in more optimal or sensitive monitor-ing.
The results of 12.1, 12.2 and 12.3 will be used to determine if a complementary, organism-group approach gives a better estimate of ecological change. Recommendations will be developed to assist in selecting single or combinations of complementary organism groups to be statistical confident that if/when ecological change occurs that it will be detected. Information on how organism groups and metrics respond to selected stressors, their expected response or impact (magnitude and direction of change) and the errors associated with estimates of change will be taken into account used to ascertain the applicability of metrics for monitoring ecological change.
12.5 How can information derived from different taxonomic groups and habitat surveys be inter-calibrated in order to provide an integrated assessment of the Ecological Status of sites?
A broad suite of approaches will be investigated in order to determine the most appropriate for cross calibrating and integrating information obtained from different taxonomic and hydromor-phological sources. No a priori conclusions about the most appropriate approach(es) have been made and different approaches and combinations of data sources may be appropriate in different circumstances.
Correlation and regression provide a simple, initial approach to the calibration of methods. Within each of the different types of samples collected, some components (taxa, statistics or other metrics derived from the biological sample) are likely to respond to the same stress (albeit more or less strongly), whilst other components (taxa or metrics derived from the data) may not respond at all. The responses may be positive or negative. Although we will test whole data metrics, it is more likely that particular components of the data will provide the best information for scaling and therefore calibrating the different sample types.
A more complex approach is provided by pattern recognition techniques. These can be used to identify metrics with common pattern across all samples and to associate these patterns to intensities of individual stresses. Metrics relating to different sample types will be most useful for calibrating the different sample types. Regression statistics provide means of calibration. Fuzzy procedures would then provide a suitable approach for combining metrics from different sample types. A useful addition would be to determine whether these relationships differ between geographical regions and/or stream types.
Existing knowledge of the autecology of each taxon will provide information about the significance of the results of these analyses.
12.6 What recommendations can be made for the use of groups for various purposes ?
The effect of stressors, scale and time on metrics and organism groups and the susceptibility to errors will lead to final recommendations concerning which method(s) is/are the best to use in which situation. This will be the base for Workpackage 15.
12.7 What recommendations can be made for the definition and delineation of class boundaries ?
As specified for invertebrate-based methods in Workpackage 11 the data will be used to derive a matrix of possible class boundaries of grades of 'Ecological Status' associated with different methods and stressors. This will be performed for each organism group and combinations of groups and will be restricted to those stream types sampled in Workpackages 7 and 8.
 |
Overview |  |