Supervising Scientist Division

3 Environmental research and monitoring (Surface water monitoring)

Supervising Scientist Annual Report 2004–2005

Supervising Scientist, Darwin, 2005
ISBN 0 642 24395 6
ISSN 0 158-4030

3.6 Changes in billabong fish communities sampled in the stream monitoring program

Since mining activities commenced at Ranger in 1979, changes unrelated to mining have occurred in stream catchments that, if not well understood, have the potential to confound conclusions drawn about the environmental impact of mining. These changes may be associated with natural climatic events or phases, or may be due directly or indirectly to invasive species removed from, introduced to, or which have increased their range throughout, the region. Hence an important ongoing task is to gain a sound understanding of the dynamics and factors affecting populations and communities of the key biota in streams adjacent to ARR mine sites. In this way, changes in metrics used to summarise responses of these monitoring organisms can be correctly attributed to mining or non-mining-related causes. Two such investigative studies are reported here, focusing in fish communities sampled in deeper channel and shallow lowland billabongs.

3.6.1 Channel billabongs

In the Supervising Scientist annual report for 2003–04, and section 2.2.3 (this report) , a significant decline over time was noted in the paired-site dissimilarity measure used as the basis for detecting changes in the fish communities of Mudginberri Billabong on Magela Creek, downstream of Ranger mine, relative to Sandy Billabong on Nourlangie Creek, a control site.

Large discrepancies in the abundances of numerically-dominant fish species between billabongs are particularly influential in inflating dissimilarity measures. In the 2003–04 annual report, the decline in the dissimilarity measure was attributed to the particularly high abundances of chequered rainbowfish ( Melanotaenia splendida inornata) and to a lesser extent glassfish ( Ambassis spp) in Mudginberri Billabong in the early years of the study, relative to Sandy Billabong. This result has subsequently been confirmed. Thus, in Table 3.3, the influence of these numerically-dominant fish species in channel billabongs upon correlation and regression results for the dissimilarity versus time relationship is shown.

Table 3.3 Influence of numerically-dominant fish species
in channel billabongs upon correlation and regression
results for the paired-site dissimilarity value, and time
  Correlation ( ρ ) Regression parameters
R2 P
All taxa -0.45 0.20 0.0003
Glassfish species removed -0.36 0.13 0.005
Chequered rainbowfish removed -0.31 0.07 0.02
Both taxa removed -0.17 0.03 0.20

Removal of each species, particularly chequered rainbowfish, reduces the significance of the decline in the dissimilarity measure. Removal of both fish species results in a non-significant result (P>0.05).

The relative abundances of chequered rainbowfish and glassfish in both Mudginberri and Sandy billabongs are plotted as a time series in Figure 3.10. (Sampling in Mudginberri Billabong commenced in 1989 though sampling in Sandy did not commence until 1994.) In 1996, the visual-observation sampling method was conducted relatively early in the wet-dry recessional flow period in Mudginberri Billabong, and the high counts of both fish species observed in this year reflect fish migrating upstream through the billabong.

Figure 3.10 Relative abundances of chequered rainbowfish and glassfish in channel billabongs over time

Figure 3.10 Relative abundances of chequered rainbowfish and glassfish in channel billabongs over time

A feature of the abundance plots of Figure 3.10 is the decline in rainbowfish in Mudginberri Billabong since 1989. Omitting anomalous 1996 data (see above), this decline is shown to be significant in regression analysis (R2 = 0.547, p = 0.001). The decline in this fish species in Mudginberri, relative to Sandy, is the main reason for the corresponding decline in the paired-site dissimilarity measure (Table 3.3). For this reason, possible causes of this decline were examined. Potential correlates of rainbowfish abundance were sought from water quality (natural and related to wet season wastewater discharges from Ranger mine) and quantity (stream discharge) variables.

In addition, rainbowfish (and glassfish) observe very significant migrations in Magela Creek after wet season spawning and recruitment on the (downstream) floodplain (ARRRI Annual Research Summary 1987–88). Thus, the abundances of these fish in Mudginberri Billabong at the time of annual sampling reflect, to a large extent, breeding and recruitment success on the Magela floodplain. (This migration phenomenon appears to be less pronounced in Nourlangie Creek, eriss unpublished data, possibly explaining to some extent the smaller magnitude and/or variability in numbers of the same two fish species in Sandy Billabong.) Therefore, correlates that could explain changes to floodplain conditions over time, as these affect rainbowfish breeding and recruitment success, were also sought.

Key environmental correlates of rainbowfish abundance and decline in Mudginberri Billabong over time are shown in Figure 3.11. The results may be summarised under the following three categories.

Water quality associated with Ranger mine wastewater discharges

The dominant contaminants associated with Ranger mine wastewater discharges to Magela Creek are uranium, magnesium and sulfate. Uranium (U) data derived from Magela Creek prior to 2000 are unreliable, due to contamination and instrumentation problems (ie poor detection limits). Therefore, magnesium (Mg) data were analysed as a reasonably reliable, surrogate measure of mine wastewater contaminant concentrations in Magela Creek. Because Mg concentrations are naturally and inversely correlated with stream discharge in Magela Creek, the net input of Mg from Ranger was derived, this being the difference in median wet season concentration between downstream (compliance site) and upstream (control) locations. The plot of net wet season Mg concentration and corresponding rainbowfish abundance in Mudginberri Billabong for that wet season is provided in Figure 3.11. No significant relationship is observed. This is not surprising: concentrations of U and Mg in Magela Creek arising from mine wastewater discharges are at least two orders of magnitude lower than those known to adversely affect larval fishes, including, in the case of U, chequered rainbowfish (eg Supervising Scientist Annual Reports 2003–04, section 3.4.1 & 2004–05, section 3.1).

Stream discharge and natural water quality

Since the commencement of this study in 1989, there has been a general increase in wet season rainfall and associated stream discharge in Magela Creek. Consequently, total wet season discharge in the creek is significantly and negatively correlated with rainbowfish numbers in Mudginberri Billabong (Figure 3.11). How higher discharge per se would result in lower fish numbers is not clear. However, greater discharge volumes result in greater dilution of wet season surface waters and their solute concentrations. Median wet season values of electrical conductivity (EC) of Magela Creek waters upstream of Ranger were used as a surrogate measure of solute concentrations. Not surprisingly, median wet season EC in Magela Creek is significantly and positively correlated with rainbowfish numbers (Figure 3.11).

Magela Creek surface waters are extremely soft and poorly-buffered, factors accentuated at high wet season flows in the creek. It is possible that early life stages of rainbowfish are stressed under these conditions (which include relatively high acidity) and lack essential minerals for growth and development. There is some experimental evidence to support this. The fry of both chequered rainbowfish (ARRRI Annual Research Summary 1987–88) and the congener, the black-banded rainbowfish used in the SSD’s creekside testing program (section 2.2.3 of this annual report), exhibit reduced survival when exposed to creek waters during high flow events.

Figure 3.11 Environmental correlates of rainbowfish abundance in Mudginberri Billabong, 1989–2005

Figure 3.11 Environmental correlates of rainbowfish abundance in Mudginberri Billabong, 1989–2005

Conditions on Magela floodplain

Changes to vegetation communities on Magela floodplain, the main breeding area and recruitment source for chequered rainbowfish in Magela Creek, may have adversely affected rainbowfish populations. In particular, a number of grass species are rapidly expanding in range and densities on the floodplain, due partly to removal of feral buffalo that once grazed on these grasses and acted as a form of control. A particularly aggressive species is the exotic para grass (Urochloa mutica). In the Supervising Scientist annual report for 2003–04 (section 3.3.1) an assessment of the risk this grass poses to the region’s floodplains was described. The ecological impact of para grass is regarded as very significant because it forms dense monocultures, out-competes native vegetation, alters hydrological regimes and thereby reduces habitat and food supply for native animals.

The rapid expansion of para grass on Magela floodplain matches the period of decline of chequered rainbowfish in Mudginberri Billabong (Figure 3.11). It is quite possible that the expansion of this and other exotic and native grasses (eg Hymenachne species) has reduced available spawning areas, food supply and migration pathways for chequered rainbowfishes on and through the floodplains.

A feature of floodplain hydrology that may also affect fish breeding success and recruitment is the period of drying of the floodplain prior to annual re-wetting. The so-termed ‘Flood-Pulse’ Theory predicts and shows that primary and secondary production of floodplains is dependent upon the degree of seasonal drying and inundation. In particular, chemical cycling and nutrient release are dependent upon, and enhanced by, sediment and soil drying prior to wet season inundation. For Magela Creek, a shorter previous-dry-season and reduced drying of the floodplain may reduce ensuing wet season production on the floodplain nursery zone. The relationship between length of previous dry season (ie from Magela Creek cease-to-flow to commencement of flow, downstream of Ranger) and Mudginberri rainbowfish numbers is, in fact, significant (Figure 3.11), lending some support to this hypothesis.

It is worth noting that the expansion of grasses on Magela floodplain will have led to greater soil water retention and thereby, and independently, accentuated any reduction in floodplain drying observed since this monitoring study commenced.


The decline in rainbowfish numbers in Mudginberri Billabong over the period 1989 to 2005 does not appear to be related to any change in water quality associated with mine wastewater discharges from Ranger. Over time with further monitoring and analysis, it may be possible to distinguish and identify natural stream water quality, discharge and/or floodplain habitat factors responsible for changes to fish populations in Magela Creek billabongs. These causal factors may then be modelled to account for variation in monitoring response variable(s).

3.6.2 Shallow lowland billabongs

It is now three years since cane toads (Bufo marinus) invaded the area of billabong sampling sites on Nourlangie Creek and two years since they appeared at Magela Creek billabong sites. Possible effects of cane toads on fish communities are examined in Figure 3.12 by comparing numbers of fish in the two catchments before and after cane toad arrival. The fish were grouped into three trophic guilds with different potential risk from the presence of toxic toad life stages: ‘Carnivores’ (spangled grunter, mouth almighty, sleepy cod); ‘Benthic omnivores’ (four species of eel-tailed catfish); and ‘Microphagic omnivores’ (glassfish, rainbowfish, hardyheads). There is no evidence of a decline in any of the trophic groups following toad arrival. Consequently, it is concluded that there has been no measurable impact from cane toads on billabong fish communities. This result contributes valuable information to national assessments of risk to biodiversity posed by this invasive species. It also indicates that this invasive species is unlikely to confound assessments of mining impact at Ranger and supports the continued use of this monitoring design for this role.

Figure 3.12 Temporal patterns of abundance of fish in shallow billabongs before and after the arrival of cane toads in two catchments, Nourlangie Creek and Magela Creek

Figure 3.12 Temporal patterns of abundance of fish in shallow billabongs before and after the arrival of cane toads in two catchments, Nourlangie Creek and Magela Creek