Explanatory notes on ecosystem level responses

Monitoring using macroinvertebrate community structure

Macroinvertebrate communities have been sampled from a number of sites in Magela Creek at the end of significant wet season flows, each year from 1988 to the present. The design and methodology have been gradually refined over this period (changes are described in the 2003–04 Supervising Scientist Annual Report (pp 32-34, section 2.2.3). The design is now a balanced one comprising upstream and downstream sites at two ‘exposed’ streams (Gulungul and Magela Creeks) and two control streams (Burdulba and Nourlangie Creeks).

Samples are collected from each site at the end of each wet season (between April and May). For each sampling occasion and for each pair of sites for a particular stream, dissimilarity indices are calculated. These indices are a measure of the extent to which macroinvertebrate communities of the two sites differ from one another. A value of zero percent indicates macroinvertebrate communities are identical in structure while a value of 100 percent indicates totally dissimilar communities, sharing no common taxa.

In the 2007–08 Supervising Scientist Annual Report (pp 30-32, section 2.2.3), improvements to the presentation and statistical analysis of macroinvertebrate data were described. By deriving dissimilarity values for each of the five possible randomly-paired upstream and downstream replicates, powerful analyses are available that can be used to test whether or not macroinvertebrate community structure has altered significantly at the exposed sites for the recent wet season of interest. For this multi-factor ANOVA, only data gathered since 1998 have been used. (Data gathered prior to this time were based upon different and less rigorous sampling and sample processing methods, and/or absence of sampling in three of the four streams.)

Disturbed sites may be associated with significantly higher paired-site dissimilarity values compared with undisturbed sites. Compilation of the full macroinvertebrate dataset from 1988 to 2008, and data from the paired sites in the two ‘exposed’ streams, Magela and Gulungul Creeks, for 2009, have been completed with results shown in Figure 1. This figure plots the paired-site dissimilarity values using family-level (log-transformed) data, for the two ‘exposed’ streams and the two ‘control’ streams.

Inferences that may be drawn from the time series data for Magela Creek shown in Figure 1 are weakened because there are no baseline (pre-1980) data upon which to assess whether or not significant changes have occurred as a consequence of mining. However, a four-factor ANOVA based upon replicate, paired-site dissimilarity values and using the factors Before/After (BA; fixed), Control/Impact (CI; fixed), Year (nested within BA; fixed) and Site (nested within CI; random) showed no significant difference in the change in dissimilarity values from earlier years (back to 1998) to those from 2008, between the control and exposed streams (ie the BA x CI interaction is not significant). While the Year x Site (BA CI) interaction is significant in the same analysis (p = 0.014), this simply indicates that dissimilarity values for the different streams – regardless of their status (Before, After, Control, Impact) – show differences through time. The dissimilarity plots shown in Figure 1 corroborate these results, showing reasonable constancy in the mean dissimilarity values for each stream across all years.

Paired upstream-downstream dissimilarity values (using the Bray-Curtis measure) calculated for community structure of macroinvertebrate families.

Figure 1  Paired upstream-downstream dissimilarity values (using the Bray-Curtis measure) calculated for community structure of macroinvertebrate families in several streams in the vicinity of the Ranger mine for the period 1988 to 2009. The dashed vertical lines delineate periods for which a different sampling and/or sample processing method was used. Dashed horizontal lines indicate mean dissimilarity across years. Dissimilarity values represent means (± standard error) of the 5 possible (randomly-selected) pairwise comparisons of upstream-downstream replicate samples within each stream.

Dissimilarity indices such as those used in Figure 1 may also be ‘mapped’ using multivariate ordination techniques to depict the relationship of the community sampled at any one site and sampling occasion with all other possible samples. Samples close to one another in the ordination indicate a similar community structure. Figure 2 depicts the ordination derived using the pooled (average) within-site macroinvertebrate data (unlike the replicate data used to construct the dissimilarity plot from Figure 1). The data points displayed are (i) for each of the sites sampled in Magela and Gulungul Creeks downstream of Ranger for each year of study (to 2009), (ii) Magela and Gulungul Creek upstream (control) sites for 2009, and (iii) all other control sites sampled up to 2008 (Magela and Gulungul upstream sites, all sites in Burdulba and Nourlangie). Because the data-points associated with the two exposed sites (on Magela and Gulungul Creeks) are generally interspersed amongst the points representing the control sites, this indicates that these exposed sites have macroinvertebrate communities that are similar to those occurring at control sites. This was verified using ANOSIM testing (ANalysis Of SIMilarity, effectively an analogue of the univariate ANOVA), a statistical approach used to determine if exposed sites (Magela and Gulungul downstream) are significantly different from control sites in multivariate space. ANOSIM conducted on pooled (within-site) data from all years to 2009 showed no significant separation of exposed and control sites for the respective comparisons (P>0.05).

Collectively, these graphical and statistical results provide good evidence that changes to water quality downstream of Ranger as a consequence of mining between 1994 and 2009 have not adversely affected macroinvertebrate communities.

Ordination plot of macroinvertebrate community structure data from sites sampled in several streams in the vicinity of Ranger mine for the period 1988 to 2009. Data from Magela and Gulungul Creeks for 2009 are indicated by the enlarged symbols.

Figure 2  Ordination plot of macroinvertebrate community structure data from sites sampled in several streams in the vicinity of Ranger mine for the period 1988 to 2009. Data from Magela and Gulungul Creeks for 2009 are indicated by the enlarged symbols.

Further information about ecosystem monitoring at Ranger Mine can be found in the Supervising Scientist Annual Reports.