WP 11: Comparing and linking assessment systems based on invertebrates

Objectives

To compare and intercalibrate the results of different 'invertebrate methods', particularly in terms of errors, precision, relation to reference conditions and possible class boundaries of grades of Ecological Status. To investigate the cost-effectiveness of different approaches to the collection and processing of macroinvertebrate samples involving varying levels of resource intensity.

Methodology/work description

The information available to this Workpackage derives from the earlier Workpackages 3, 4, 6, 7, 8 and 9, together with an additional series of specially collected samples in space, time and intensity.

In combination the source Workpackages provide:

• high quality, consistently identified macroinvertebrate data from sites sampled and sorted in eleven Member States using standardised field and laboratory protocols for the most commonly used field techniques;
• databases of reliable quality on ecological river quality from other national and international projects;
• the metrics used by EU Member States for assessing the Ecological Status of their lotic waterbodies.

The methodology, databases and metrics will by evaluated by:

• quantification of the uncertainties associated with the most commonly used protocols for the collection of macroinvertebrate data;
• quantification of the sample processing and identification biases of each laboratory for several protocols;
• using a variety of taxonomic groups and morphological/environmental information in the applied metrics and related data evaluation techniques.

Sampling variation and variation of results: Replicate data acquired during the sampling workshops will be used to assess the comparative uncertainties associated with the prescriptive sampling and sample processing protocols of several field methods. Comparisons will be made of the following components of samples:

• mean numbers of taxa per sample per sampling protocol
• range, standard deviation and confidence intervals of mean numbers of taxa per sample
• between- and within-method variation in the functional composition of samples
• means and variance terms for the values of commonly used metrics and other calculation methods derived from replicate samples for each field protocol

In conjunction with data on sample processing biases (see below), these analyses will demonstrate the differences in the basic data submitted to a range of assessment protocols and how different levels of probability of assignment of sites to grades of Ecological Status may be derived from these assessment protocols.

Sample processing biases: Audit data acquired from Workpackage 9 will be used to assess the comparative biases associated with the prescriptive sample processing protocols of each of AQEM and RIVPACS. Comparisons will be made of the following components of the audit programme:

• The biases associated with each Member State. Are these a function of stream type, taxonomic difficulties particular to individual ecoregions or the comparative expertise and experience of individual consortium members within laboratories ?
• The biases associated with the separate processing protocols associated with each of the assessment programmes.
• The range, indicator values (if any) and community function of the taxa least successfully removed from samples and/or identified in each Member State.

Development of an error module: The variations, errors and biases associated with the sampling and processing of macroinvertebrate samples (as determined above) will be integrated into a single uncertainties model. This module will be developed as software packages for incorporation in ex-isting assessment systems. A similar system already exists within RIVPACS but the uncertainties determined from experimental programmes within the UK will be compared with those obtained from pan-European sampling. The functionality of the error module will be:

• Assessment of the statistical significance of observed temporal or spatial changes in the composition of macroinvertebrate samples (having taken all sources of methodological variation into account).
• Assessment of the probabilities that a given test site is truly in a particular grade of Ecological Status.
• Assessment of the probability that a site has truly changed grade over time. As a function of the uncertainties associated with biological monitoring, a site may appear to have changed its grade of Ecological Status over time without this change being significant.

Inter-calibration of 'invertebrate methods' and recommendations for class boundaries: The availability of data collected from common sites sampled concurrently by several field protocols will enable inter-calibration of procedures. Three tests will be used as part of the inter-calibration process:

• A comparison of the grades of Ecological Status attributed to test sites when samples collected using the AQEM and other national protocols are each used within the AQEM model.
• A comparison of the grades of Ecological Status attributed to test sites when samples collected using the AQEM and RIVPACS methodologies are each used within the RIVPACS model. The restricted geographic coverage of RIVPACS will only allow comparisons of samples taken within the United Kingdom.
• A comparison of the grades of Ecological Status attributed to test sites when samples collected at given sites using a certain protocol (e.g. AQEM) are assessed using the site evaluation systems associated with other standard national invertebrate protocols (e.g. RIVPACS, IBE, PERLA).
• Recommendations for class boundaries: From the results obtained with the above tests we will derive a matrix of possible class boundaries for the ecological state attributed to different methods and stressors.

Each test will incorporate the use of the error modules in order to compare an evaluation of the differences in the relative precision with which alternative procedures can be used to detect change. Comparisons of change will be based on evaluations of the spatial differences between sites of dif-ferent levels of presumed Ecological Status in the project sampling network.

These steps will be complemented by data on sample processing and cost effectiveness taken by a single partner (The Netherlands) as part of Workpackage 11. Additional samples will be taken in close collaboration with water authorities, following a stepwise procedure. The sample processing will particularly address questions of sampling (do we have to sample all habitats at a certain sampling surface area or can we use estimates of habitats and correct the numbers of organism from small samples to correct for the actual share of each?), sorting and processing of counts (what results are obtained when not all organism are sorted by hand, or only large animals are collected, or only subsamples are sorted?), and identification (how many individuals must be identified to get a valid picture of the species present and how many to get a valid picture of the abundances). But also aspects of seasonality (what season must be sampled and give most informative results?). Finally, attention will be given to the relationship between the sampling effectiveness and the results from metric calculations. Questions like "do metrics need this and/or that information to such detail?" should be answered.

The aim is, to produce a scientifically based, sound and especially cost effective methodology for comparison with the full standard protocols.

Overview