• About this site|
• Site index|
• Contact us|
• Jobs

A User Guide for the Macroinvertebrate Community Index

Acknowledgements

Part 1: Background to Biotic Indices

Part 2: Guidelines for using the MCI, QMCI and SQMCI

1 Applying MCI indices in different freshwater environments

2 Taxonomy

3 Monitoring and reporting

4 Design of monitoring programmes

5 Detecting trends

6 Reporting the MCI

References

Appendix 1: Calculating the MCI: Excel Macros and User-defined Functions

Appendix 2: Use of MCI, QMCI and SQMCI Throughout New Zealand

3 Monitoring and reporting

Biological systems are complex and unstable in space and time. As a result, biologists often feel compelled to study all of their components, but one need not sample everything. For monitoring it is more important to focus on biological attributes that respond reliably to human activities, are minimally affected by natural variation, are cost effective to measure and can be presented in a way that conveys useful information to water managers or to the general public. Macroinvertebrate biomonitoring using biotic indices has a long history of meeting these objectives (Karr and Chu 1999).

This section looks at different types of bioassessments and biomonitoring programmes and provides some guidance on how the MCI and its variant indices should be used within these programmes.

3.1 Compliance monitoring and environmental impact assessments

3.1.1 Compliance monitoring

Compliance monitoring is routine monitoring to ensure that an activity that is allowed under a resource consent is not having any significant adverse effects.

We recognise that regional councils have the responsibility for determining the scope of consent monitoring on a case-by-case basis. However, we have seen examples of monitoring conditions that seem to us to be overly burdensome for consent holders. There really is no need to monitor everything. Put simply, if a stream or river has healthy macroinvertebrate communities then it is almost certain that other ecosystem components will be in good shape too. There is often no need to monitor water quality, bed sediments, periphyton biomass or fish populations to obtain reasonable assurance that a consented activity is not having significant adverse effects. If macroinvertebrate biomonitoring does reveal disturbing trends, or if a problem is observed when sampling for macroinvertebrates, then that is when additional investigations should be undertaken.

There is some justification for more frequent or more intensive monitoring programmes for new consents, but if no adverse effects have been detected by the first review (after perhaps five years), then we believe that monitoring requirements should be reduced. In our view, annual macroinvertebrate sampling represents the minimum desirable level for compliance monitoring.

The first step in choosing which MCI variant to use is to ensure that the chosen index is sensitive to the type of impact or disturbance that the consented activity is expected to have (or might have if something untoward happens - bearing in mind that under the Resource Management Act 1991 (RMA) impacts from consented activities are not supposed to have any significant adverse effects). One decision that will have a significant effect on costs is whether or not quantitative data are required.

The design of compliance monitoring programmes is discussed further in Section 4.

3.1.2 Assessments of environmental effects (AEEs)

Assessments of environmental effects (AEE) are undertaken when a new activity that is expected to have effects on the environment is proposed. The AEE will form part of the application for consent to undertake this activity under the provisions of the RMA. An AEE is also likely to be required when the term of an existing consent expires and permission is required for the activity to continue.

Biotic indices are a measure of stream health and can be used as part of an AEE to show the effect of an activity, provided the activity is capable of causing the kinds of changes to macroinvertebrate communities that biotic indices can detect. For example, it is entirely appropriate to use the MCI (or one of its variants) to assess an activity that has the potential to cause nutrient enrichment or sedimentation. However, if there was a proposed discharge containing chemicals that had no effect on macroinvertebrate communities (but did have other adverse effects on the environment), then the MCI would not be an appropriate tool to use. For example, Hickey and Clements (1998) found that the QMCI did not detect the impacts of heavy metal pollution in streams on the Coromandel Peninsula. They noted that this was because the QMCI had "incorrect tolerance scores for some taxa to heavy metals." This is not really a valid criticism of the QMCI, which was developed to detect organic pollution and nutrient enrichment, but rather a warning about using indices for assessing impacts (e.g. metal toxicity) for which they were not designed.

Although research is still in progress, there is evidence to suggest that the MCI has limitations when assessing the effects of extremely low flows. As flows reduce (whether during natural droughts or as the result of abstraction), macroinvertebrate communities in many stony streams change from being dominated by mayflies, stoneflies, and caddisflies, to being dominated by chironomids, worms, snails and hydroptilid caddisflies. This occurs because periphyton on the stone surfaces changes from a thin diatom film to thick algal mats or even filamentous algae. These changes are reflected in a sharp decrease in the MCI. However, once the entire riverbed is covered with thick periphyton, the MCI stabilizes (perhaps around 80) even though flow, wetted perimeter, and space for aquatic communities continue to decrease. It follows that MCI values alone should not be used to support arguments that extreme abstractions do not have significant adverse effects on aquatic communities.

3.2 State of the environment monitoring and reporting

Effective management of water resources (or environmental quality) requires knowing about changes that occur in the environment and having an understanding of the underlying cause(s) of any changes that might be predicted or observed. It is also desirable to be able to distinguish anthropogenic (human-caused) changes from natural ones. This kind of information is gathered mainly by State of the Environment (SoE) monitoring and reporting, which in New Zealand usually is undertaken by regional and unitary councils.

Specifically, SoE monitoring and reporting programmes aim to:

• obtain representative data for each of the resources or resource compartments
• detect the presence and direction of trends
• identify the effects of activities − particularly land-use change − on resource quality
• determine the effectiveness of management initiatives directed at enhancing degraded resources.

Long-term data sets, including biomonitoring data sets, are vital (Likens 1998). They can address scientific and environmental questions at a scale that is realistic and applicable to environmental management, and can document the responses to disturbance by natural and anthropogenic events and activities.

3.2.1 Which biotic index should be used?

This view probably is quite contentious, given that most regional councils have reported the SQMCI (from coded-abundance data) or QMCI (from fixed-counts) along with the MCI in their SoE reports. So why do we recommend using only the MCI?

Most regional councils take several weeks (on each monitoring occasion) to collect their SoE samples. Most have rules about when to sample in relation to the last significant flood. However, when sampling is spread over several weeks (and even up to a month or two), there will be differences in macroinvertebrate community composition relating to when the sample was collected. For example, consider two sites that are identical in condition or "health". If they are sampled on the same day, their MCI, SQMCI or QMCI values will be about the same. However, if these two sites are sampled 30 days apart, it is likely that the biotic indices will be different. If the river has been in recession the entire time, the site sampled second will probably have lower index values because there will be increasing development of periphyton-associated communities with increasing dominance by low-scoring taxa such as chironomids, worms, snails and hydroptilid caddisflies. If there has been a significant fresh [A fresh is a sudden increase in stream or river flow due to rainfall or snow/ice-melt.] between sampling these sites, the site sampled second could have a higher index value, because low-scoring algal-associated taxa are displaced by mayflies, stoneflies and caddisflies that are characteristic of comparatively clean stone surfaces. The difference between index values is a consequence of when the samples were collected rather than a measure of the health of the sites. This problem affects the MCI to a lesser extent than the SQMCI or QMCI, because the list of species present at a site is affected less when samples are collected than the densities or relative abundances of taxa. [Research on the effects of floods and droughts on biotic index values in stony streams is currently being undertaken by John Stark under NIWA's Water Allocation FRST programme. One aim is to determine correction factors that would enable biotic index values to be standardised to factor out the influence of floods, droughts, season or sampling time. Results are expected by June 2007.]

For SoE reporting to the public or to other laypeople, the KIS (Keep It Simple) principle should apply. SoE reports that consider taxa richness, MCI and SQMCI may be confusing. These three indices do not measure the same thing, so assessments based on them are not necessarily in complete agreement, often requiring an experienced biologist to explain the differences.

Taxa richness is not strictly a measure of stream condition or stream health. In fact, highest taxa richness often is associated with slightly enriched streams (e.g. those experiencing diffuse-source nutrient enrichment from farmland), rather than pristine streams in reference condition. Low taxa richness can be associated with quite "sterile" environments with extremely pure water, perhaps where torrential water velocities and lack of nutrients result in low productivity. Such places are naturally unproductive. In other words, there is no valid basis to assume that high taxa richness is good and low taxa richness is bad. Furthermore, estimates of taxa richness are highly dependent on sample size, which, in turn, can be influenced by sampling or processing effort (which can vary markedly with different personnel). For these reasons, use of taxa richness for SoE reporting is not recommended.

We do not recommend the SQMCI (or QMCI) for SoE reporting either, because community percentage composition (more so than taxonomic composition) can change during the sampling period (which can be several weeks) in response to small freshes or as a flow recession lengthens. Consequently, differences between sites arise as a result of when samples were collected, and are therefore an artefact of the sampling regime rather than a true measure of stream health. We also found substantially higher variances in SQMCI-sb scores compared to MCI-sb scores from replicate samples collected in soft-bottomed streams in the Auckland regions (Stark and Maxted 2004, 2007). The high variance impaired the ability of the index to discriminate between sites of different qualities.

As we have seen, the MCI is a well-proven and reliable index for assessing stream health. Reliable MCI values can be derived from samples collected according to the national protocols, including fixed counts as low as 100 (although a minimum fixed count of 200 is recommended) (Stark et al 2001; Duggan et al 2003). The MCI is not affected by changes in percentage composition, and, consequently, is affected much less than the SQMCI and QMCI by flow-related or seasonal factors. The MCI is essentially a scaled average score per taxon, and therefore (being an average) is relatively unaffected by sample size (unlike taxa richness) provided that samples are collected according to the standard protocols (Stark et al 2001). The MCI normally is highly correlated with the SQMCI or QMCI (Stark 1993, 1998), so using indices that tell much the same story seems somewhat superfluous, and could be confusing for laypeople. As a result, we recommend the MCI as the index of choice for SoE reporting.

In theory, the MCI can be affected by samples containing taxa that have drifted into the sampling area from upstream habitat or tributaries. However, this also affects other indices such as taxa richness and EPT richness, and we do not believe this is a significant problem for bioassessments, although the effect of this on stream health assessments has not been evaluated. We suspect it would be more of an issue when sampling is undertaken soon after a significant fresh. Drift is a major mechanism for re-colonisation of downstream reaches, and who is to say whether one or two higher-scoring taxa that may have drifted in from upstream really belong there or not. All macroinvertebrate communities have rare taxa present.

3.2.2 SoE monitoring versus research

In our view, SoE monitoring is not the best way to undertake basic research into the relationships between biological communities, land use, or other human activities that may affect water quality. This is not to say that data from SoE monitoring cannot be used for this purpose (as was done in the Auckland region; see Stark and Maxted 2004). Research objectives can be achieved by undertaking additional investigations in association with SoE monitoring, but it is important to keep the fundamental aims of SoE monitoring in focus, and not compromise the integrity of the SoE monitoring programme.

3.2.3 Planning an SoE monitoring programme

The design of monitoring programmes is discussed in Section 4 in more detail, but it is worth making some preliminary comments here. SoE monitoring involves sampling at defined (fixed) locations, at predefined intervals, according to the parameters being measured. Sites should be representative of least-disturbed conditions (reference) and impacted areas, and of a range of common land uses within the region. Monitoring methods should be robust, standardised, and applied consistently over time.

SoE monitoring networks have to be designed with management objectives in mind, and we consider the following matters should be at the forefront of design planning.

• What are the site locations?
• Use both reference sites and impacted sites, and samples should be representative in space (e.g. across different land uses and different River Environment Classification classes, and spread throughout the region).
• How many sites?
• What indicators will be measured?
• What degree of change do you want to detect (and how will it be distinguished from natural variability)?
• How often will monitoring be undertaken? (Samples should be representative in time.)
• What information is required?
• How will the data be analysed?
• How many data do the statistical analyses require?
• How much replication is required?
• How will data be translated into information that water managers can use?
• How much will it cost?
• Is there a long-term commitment to funding?

3.3 Biodiversity monitoring

Whereas the MCI focuses on cost-effective stream health assessment, there is increasing interest these days on freshwater macroinvertebrate biodiversity. This is a different issue, and here MCI-level identifications fall a little short of the ideal.

How best to undertake surveys to assess biodiversity is a topic that warrants further investigation. A large number of samples are almost certain to be required to obtain a complete list of the macroinvertebrate taxa in a particular stream habitat. Stark (1993b: Figure 1) showed, for example, that a single hand-net sample collected from riffle habitat contained about 57% (range 34%−78%) of the taxa collected in 12 replicates combined, and even 12 samples were insufficient to collect all the taxa that were likely to be present. Furthermore, other taxa found in non-riffle habitats (e.g. runs, pools, under banks) would require additional sampling effort to detect them. In our view, routine sampling for biodiversity is likely to be prohibitively expensive.

So how can we continue to undertake cost-effective SoE monitoring and collect data suitable for assessing biodiversity at the same time? Perhaps the solution to the bioassessment-biodiversity issue is to take single hand-net samples at each SoE site according to protocols C1 (hard-bottomed) or C2 (soft-bottomed) (Stark et al 2001), identify all taxa to the species level (where possible), and record only presence/absence. These data would enable calculation of MCI and MCI-sb values that would be consistent with those calculated from data collected previously. Additional samples could be collected from other habitats and scanned for taxa that were not found in the sample collected from the habitats targeted for SoE monitoring using protocols C1 or C2. The additional effort previously put into relative abundance counts or abundance coding could be put into the specific identification of more samples and habitats. A reference collection should also be assembled.

It is unlikely that sufficient sampling could be undertaken on each SoE monitoring occasion to compile complete macroinvertebrate biodiversity inventories for each sampling site. However, over time, the list of taxa recorded from each site will increase. Alternatively, every few years (say five or ten) extra sampling effort focusing on biodiversity assessment could be added to the SoE programme.

Although collecting presence-absence data for SoE monitoring may seem like a backward step, we believe that the positives outweigh the negatives. The positives include:

1. Cost-effective biomonitoring is retained (or even improved).
2. More sites can be monitored within a given budget because sample processing costs are reduced.
3. Species-level data suitable for biodiversity assessments are collected.
4. These data are suitable for developing a species-level MCI using the rank correlation iterative method developed by Bruce Chessman (as used by Stark and Maxted 2004, 2007 for the MCI-sb).
5. Assessments for soft-bottomed streams would be improved and much more cost-effective − the MCI-sb performs much better than its semi-quantitative and quantitative variants in soft-bottomed streams, and obtaining coded abundances or counts of invertebrates from samples from soft-bottomed streams is very time consuming.
6. Data quality control would be simplified - checking the reference collection and/or checking identifications in vials and for missed taxa in the sample residue.
7. The confounding influences of floods, low flows, season and extended sampling periods are minimised because they affect taxa richness and MCI much less than relative abundances, densities, SQMCI or QMCI.

On the other hand, if only presence-absence data are collected:

1. There are some well-performing indices that cannot be calculated (e.g. %EPT abundance), but they are often strongly correlated with MCI anyway, and so may be redundant.
2. There will be a loss of information, because knowing which taxa are dominant is useful information for an ecologist but understanding the influences of the confounding effects of flow and season is required to make full use of this information.