Skip to main content


Title: Investigation of the Relationship between the LIFE Index and RIVPACS: Putting LIFE into RIVPACS
Author: R T Clarke
Author: P D Armitage
Author: D Hornby
Author: P Scarlett
Author: J Davy-Bowker
Author: Environment Agency
Document Type: Monograph
Annotation: Environment Agency Project ID:EAPRJOUT_1263, Representation ID: 423, Object ID: 2473
In the UK, there are competing demands for both surface and groundwater resources. Sustained or repeated periods of low flows and/or slow flows are expected to impact on the plant and animal communities within rivers. To assess the potential impact of flow-related stresses on lotic macroinvertebrate communities, Chris Extence and colleagues from Anglian Region of the Environment Agency developed the Lotic-invertebrate Index for Flow Evaluation (LIFE). Extence et al. (1999) showed that for several individual sites, temporal variation in LIFE could be correlated with recent and preceding flow conditions. RIVPACS (River InVertebrate Prediction And Classification System), developed by CEH, the Environment Agency and their predecessors, is the principal methodology currently used by the UK government environment agencies to assess the biological condition of UK rivers. RIVPACS assesses biological condition at a site by comparing the observed macroinvertebrate fauna with the fauna expected at the site if it is unstressed and unpolluted, as predicted from its environmental characteristics. Biological condition is estimated currently using two Ecological Quality Indices (EQI) represented by the ratio (O/E) or observed (O) to expected (E) values of the number of Biological Monitoring Working Party (BMWP) taxa present and the ASPT (Average Score Per Taxon), denoted by EQITAXA and EQIASPT respectively. LIFE is based on the same macroinvertebrate sampling procedures as RIVPACS. In this R and D project, an assessment was made of the potential to use the RIVPACS reference sites and methodology to standardise LIFE across all physical types of site, as a ratio of observed to expected LIFE, denoted LIFE O/E. LIFE O/E then provides a standardised estimate of the severity of the impacts of any flow-related stress on the macroinvertebrate fauna at a site. The Environment Agency intend to use expected LIFE calculated using RIVPACS and LIFE O/E to determine the macroinvertebrate component in the Environmental Weighting (EW) system being developed within their Resource Assessment and Management (RAM) Framework for abstraction licensing and water resource assessments for Catchment Abstraction Management Strategies (CAMS). CEH have derived a numerical algorithm to provide predictions of the expected LIFE for any river site based on its values for the standard RIVPACS environmental predictor variables. This algorithm is compatible with the derivation of expected ASPT, gives appropriate lower weighting to taxa with lower expected probabilities of occurrence and hence should be used in preference to the current LIFECALCULATOR method. It is recommended that this new algorithm is incorporated into an updated Windows version of the RIVPACS software system to provide automatic calculation of observed LIFE, expected LIFE and hence LIFE O/E for any macroinvertebrate sample and river site. All analyses were based on family level log abundance category data from single season samples. The relative merits of using the minimum or average values of single season LIFE O/E or combined season sample LIFE O/E for annual assessments of flow related stress at a site need further investigation. Natural sampling variability alone can cause lower minimum values. An agreed standard method is needed for combining abundance category data for R and D Technical Report W6-044/TR1 v historical samples (i.e. pre- 2002) to enable sites assessments for future samples to be compared with historical data to estimate changes and trends. Seventy percent of the total variation in LIFE across all the high quality RIVPACS reference sites was explained by differences between the biological groupings of sites formed in the development of RIVPACS; this explanatory power was as high as for ASPT. Amongst these high quality unstressed sites, observed LIFE was correlated with the physical characteristics of a site. LIFE was positively correlated with site altitude and slope and the percentage substratum cover of boulders and cobbles; it was negatively correlated with stream depth and in-stream alkalinity and the percentage cover of sand and fine silt or clay sediment. When based on its standard suite of environmental predictor variables, RIVPACS predictions of expected LIFE were very effective overall, with correlations between observed life and expected LIFE of 0.78 for the 614 RIVPACS reference sites. Expected LIFE can vary between 5.93 and 7.92. LIFE O/E was centred around unity for the RIVPACS reference sites, with a small standard deviation of 0.056, less than the equivalent standard deviation for EQIASPT. Observed and expected LIFE should be recorded to two decimal places and LIFE O/E to three decimal places. Variation in observed LIFE and LIFE O/E was assessed for over 6000 of the biological sites sampled in the 1995 General Quality Assessment (GQA) national survey. These sites covered a very wide range of types and biological quality of site, including some which had been impacted by varying degrees of flow-related stress. Although observed LIFE ranged from 4.60 to 9.45, 90% of GQA sites had values in the narrow range 5.91-7.85. A provisional six grade system for LIFE O/E was developed based on the frequency distributions of values of LIFE O/E for the high quality reference sites and the wide ranging GQA sites. The lower limits for the grades were set at 1.00, 0.97, 0.93, 0.88 and 0.83; the lower limit of 1.00 for the top grade was chosen to give compatibility with the GQA grading system based on EQIASPT. The LIFE and ASPT indices are naturally correlated to some extent; macroinvertebrate families which require fast flowing conditions tend to also be susceptible to organic pollution, and vice versa. However, amongst the GQA sites the correlation between LIFE O/E and EQIASPT is only 0.69; the correlation between LIFE O/E and EQITAXA is only 0.39. The LIFE and GQA grades for the GQA sites were cross-compared. The LIFE and BMWP scoring systems do not appear to be completely confounded; suggesting that it may be possible to use the biota to differentiate flow-related stress from organic dominated stress. However, the apparent lack of agreement in site assessments using the two scoring systems must be at least partly due to the effects of sampling variation on both sets of O/E ratios. This will be correlated variation as the O/E ratios for a site are all calculated from the same sample(s). Further research is needed urgently to assess the influence of sampling variation on the observed relationship between LIFE O/E and EQIASPT and thus the extent to which they can be used to identify different forms of stress. R and D Technical Report W6-044/TR1 vi The sensitivity of RIVPACS predictions of expected LIFE to flow related characteristics at a site was assessed by simulating alterations to stream width, depth, discharge category and substratum composition. Within a site type, realistic changes led to relatively small changes, usually less than 0.3, in expected LIFE. This suggest that RIVPACS predictions of expected LIFE are robust and mostly vary with the major physical types of site. Ideally, the RIVPACS predictions of the atargeta or expected LIFE, should not involve variables whose values when measured in the field may have already been altered by the flow-related stresses whose effects LIFE O/E is being used to detect. Using new predictions not involving the RIVPACS variables based on substratum particle size composition, stream width and depth, the change in expected LIFE is less than 0.10 for over 70% of sites and the change in LIFE O/E is less than 0.02 for 80% of sites. However, omitting these variables, especially mean substratum particle size, lead to significant increases and hence over-predictions of expected LIFE for large and/or slowflowing lowland river sites (notably in RIVPACS site groups 33-35), which then underestimated LIFE O/E for this type of site. This problem needs resolving. Further research is needed to assess the potential for improving predictions without these flow-related variables using temporally-invariant GIS-derived variables such as upstream catchment or river corridor geological composition. An ecological or environmental index is of little value without some knowledge of its susceptibility to sampling variation and other estimation errors. Sampling variation in observed LIFE was assessed using the replicated sampling study sites involved in quantifying sampling variation of ASPT and number of BMWP taxa, as used in the uncertainty assessment of EQIs in RIVPACS III+. Sampling variation in LIFE was found to be small relative to differences between physical types of site. There was no evidence that sampling differences between operators affected LIFE. The sampling standard deviation of LIFE decreased with the number of LIFE-scoring families present at a site; a predictive equation has been derived. It is recommended that this relationship is used in any future assessment of uncertainty in values of LIFE O/E. The RIVPACS reference sites were selected because, at the time of sampling, they were considered to be of high biological quality and not subject to any form of environmental stress, whether from toxic or organic pollution or flow-related problems. The current study included the first quantitative assessment of the flow conditions in the year of sampling each reference site relative to the flows in other years at the same site. Reference sites were carefully linked to the most appropriate national flow gauging station using the CEH national river network GIS (Geographic Information System) derived from the CEH-corrected Ordnance Survey 1:50000 blue-line river data. For most types of reference site there was no relationship between autumn sample LIFE O/E and the relative mean summer (June-August) flow in the immediately preceding summer. Three lowland stream reference sites of the same biological type were identified as having low LIFE O/E and sampled in years of relatively low summer flows. It is recommended that these three sites are not involved in RIVPACS predictions of expected LIFE. Removing these three sites, which are all from RIVPACS site group 33, may also reduce the problem, R and D Technical Report W6-044/TR1 vii discussed above, of over-predicting expected LIFE for large lowland river sites in RIVPACS site groups 33-35 when flow-related variables are excluded from the predictions. A large subset (c. 2000) of the biological GQA sites sampled in the 1995 national survey were linked, using the GIS, to suitable gauging stations of similar Strahler stream order within 10km which had complete summer flow data in 1995 and in at least four other years. One important factor influencing the ability to detect relationships between LIFE and flows was that river flows were less, often much less, than average in all regions of England and Wales in 1995. The general correlations between autumn sample LIFE O/E and relative summer flows in the preceding summer were statistically significant, but weak, both overall and for sites within each biological type. Correlations were strongest for intermediate size non-lowland streams occurring mainly in northern and south-west England and Wales, which include flashy rivers where the macroinvertebrates are more likely to be dependent on recent flows. However, the vast majority of the GQA sites with very low values of LIFE O/E (i.e. less than 0.8) had mean summer flows in 1995 which were ranked amongst the lowest 20% of all years with flow data available. These GQA sites are likely to have been suffering from flow related stress in 1995. In contrast, a large proportion of GQA sites with relatively low flows had relatively high values of LIFE O/E in autumn 1995. The autumn 1995 macroinvertebrate fauna at many of these sites may be dependent on flow conditions over longer or earlier periods than just the preceding summer. In this study, the only flow variable considered was relative mean summer flow and this was correlated with autumn sample LIFE O/E across all GQA sites. The correlations were less than those found by Extence et al (1999) within individual sites between observed LIFE and the best of a large range of flow variables measured over a period of years. More research is needed on developing relationships between LIFE O/E and flow parameters whose time period and form vary with the type of site. Autumn 2000 was a period of very high flows in many regions, which contrast with the generally low flows in 1995. It may be useful to compare differences in LIFE O/E with differences in flows between the two years amongst those sites with matched flow data that were surveyed in both the 1995 and 2000 GQA surveys. R and D Technical Report W6-044/TR1 viii CONTENTS Page EXECUTIVE SUMMARY 1.
Publisher: Environment Agency
Subject Keywords: Low flows; Biological monitoring; RIVPACS; Macroinvertebrates; Life; Slow flows; Ecological stress; Catchment Abstraction Management Strategies; Resource assessment and management (ram) framework
Extent: 182
Total file downloads: 9

Download PDF    Display PDF in separate tab