The Distribution, Recruitment and Movement of in Far Western

Author Kerezsy, Adam

Published 2010

Thesis Type Thesis (PhD Doctorate)

School Griffith School of Environment

DOI https://doi.org/10.25904/1912/2533

Copyright Statement The author owns the copyright in this thesis, unless stated otherwise.

Downloaded from http://hdl.handle.net/10072/365885

Griffith Research Online https://research-repository.griffith.edu.au The distribution, recruitment and movement of fish in far

Adam Kerezsy

Bachelor of Arts Bachelor of Applied Science - Environmental Science (Honours)

Griffith School of Environment Science, Environment, Engineering and Technology Griffith University, Nathan, Queensland,

Submitted in fulfilment of the requirements of the degree of Doctor of Philosophy

January 2010

ii Abstract

The systems of far western Queensland, such as the Bulloo in the Bulloo- Bancannia Basin and the Cooper, Diamantina and Georgina in the Eyre Basin, are catchments that have been subjected to minimal anthropogenic disturbance. These are characterised by extreme flow variability by virtue of their geographical locations in the semi-arid and arid areas of . In general, the fish communities from these areas are poorly known, especially those from the most remote catchments such as the Georgina and its tributaries. The first aim of this study was to more-accurately establish the distributional range of fish throughout far western Queensland through extended spatial and temporal sampling of all major river systems from the Bulloo west to the border.

Fish sampling commenced in September 2006 and continued in December 2006, January 2007, April 2007, August 2007 and November 2007 before concluding in March and April 2008. Extra data from sampling undertaken in November 2008 and as recently as October 2009 has occasionally been included in the thesis chapters where it is relevant to the analysed results or the interpretation and discussion sections. In general, a minimum of three sites was always sampled on each sampling occasion and in each catchment – from east to west the Bulloo, Kyabra, Barcoo, Thomson, Cooper, Diamantina, Georgina and Mulligan. Major flooding occurred in the Georgina and Mulligan catchments in January and February 2007 and in the Barcoo, Thomson and Cooper catchments in January and February 2008. In contrast, neither the Diamantina nor Kyabra catchments experienced overbank flooding for the duration of the study. Consequently, comparison of fish communities between catchments with vastly different hydrological histories became possible using the assembled dataset. The results from the Mulligan catchment are particularly interesting as they provide the first fish records from this watershed and are amongst the few documented instances of fish species migrating long distances in an ephemeral Australian desert river. Seven species were demonstrated to colonise the Mulligan following flooding, and range extensions were also established for golden goby, Glossogobius aureus , in the Diamantina catchment, and the translocated sleepy cod, Oxyeleotris lineolatus , in the Thomson catchment.

iii Biota must possess adaptations or life-cycle traits that enable them to survive in the environments in which they live. Fish from arid-zone rivers that frequently experience total drying and/or extended drying periods live at the realistic limit of freshwater habitation. In the second part of this study, the recruitment patterns of fish species from the Bulloo-Bancannia and basins in Queensland were investigated using length-frequency analysis of samples from all sampling occasions. In particular, this data was analysed to evaluate whether recruitment is contingent upon flow events or season, or whether fish in isolated arid-zone river systems evince more opportunistic recruitment strategies in order to survive in river systems where channel and flood flows are rare and unpredictable. The results demonstrate that the majority of extant fish species are capable of maintaining their populations in isolated waterholes in the absence of flow events.

In river systems where flow is dynamic and stochastic, fish habitat is similarly changeable, for pools and reaches may become inundated only sporadically. Nevertheless, such temporary habitats may be colonised by vagile fish species provided connection with source habitats is established periodically. The current study documents the colonisation of a remote desert river – the Mulligan – by fish from its parent river – the Georgina – following connection during flooding. Additionally, movement patterns are analysed at a smaller scale within the Thomson and Barcoo catchments by examining fish usage of temporarily inundated channels following smaller connection flows.

Concepts detailing river function with reference to hydro-ecological models predicated on regular flood flows (Flood Pulse Concept) or connected river channels (River Continuum Concept) are not easily applied to rivers in the Australian arid-zone due to the stochasticity of flow events and the frequency and duration of dry spells. The current study considers the results drawn from the distribution, recruitment and movement studies in relation to existing models describing riverine function, and concludes that a source-sink population model may be useful for describing the dynamics of native fish in central Australia. Under this model, fish recruitment generally occurs along a constant timeframe irrespective of antecedent flows and flooding, whereas movement is highly opportunistic and likely to occur – for some if not all species – whenever migration pathways become available.

iv Given the current lack of information regarding isolated arid-zone rivers in Australia and the unique opportunities they present for both conservation and future study, the final part of this thesis examines the current status of freshwater fish assemblages and species in far western Queensland and the challenges and opportunities that currently exist in both research and management. Although the catchments considered during the study all demonstrate a robust native fish fauna, this is likely to be a result of their isolation and the existence of natural flow regimes. Maintaining these flow regimes, as well as attempting to prevent the spread of threatening processes such as alien and translocated species, should be prioritised by relevant management, policy and research agencies and institutions.

v vi Declaration

This work has not previously been submitted for a degree or diploma at any university. To the best of my knowledge and belief, the thesis contains no material previously published or written by another person except where due reference is made in the thesis itself.

Adam Kerezsy January 2010

All photographs in this thesis are the work of the author except where otherwise indicated.

vii viii Table of Contents

Abstract...... iii Declaration...... vii Table of Contents ...... ix List of Tables ...... xii List of Figures ...... xv List of Appendices ...... xxi Acknowledgements ...... xxii 1. Introduction...... 1 1.1 Ecological adaptations in arid environments ...... 2 1.2 Freshwater ecology in arid environments ...... 4 1.3 The current study...... 8 1.4 Aims of the current study...... 11 1.5 Structure of the thesis ...... 12 2. Study Area ...... 13 2.1 Location ...... 13 2.2 Climate...... 15 2.3 Hydrology ...... 16 2.4 Topography and geomorphology ...... 21 2.5 Vegetation ...... 22 3. Fish of the Lake Eyre and Bulloo-Bancannia Basins...... 24 3.1 Existing freshwater ecological studies...... 24 3.2 Existing knowledge and knowledge gaps in relation to fish in far western Queensland...... 25 3.2.1 ...... 25 3.2.2 Plotosidae ...... 26 3.2.3 ...... 29 3.2.4 Melanotaeniidae...... 30 3.2.5 Pseudomugilidae...... 31 3.2.6 Ambassidae...... 31 3.2.7 Percichthyidae...... 33 3.2.8 Terapontidae...... 35 3.2.9 ...... 38 3.2.10 Eleotridae...... 39 4. Variability of the study sites: Hydrology and water quality...... 43 4.1 Hydrology of the catchments preceding and during the study...... 45 4.1.1 Sites in the Mulligan catchment ...... 47 4.1.2 Sites in the Georgina catchment...... 52 4.1.3 Sites in the Diamantina catchment...... 58 4.1.4 Sites in the greater Cooper catchment...... 64 4.1.4.1 Sites in the Thomson catchment ...... 64

ix 4.1.4.2 Sites in the Barcoo catchment ...... 72 4.1.4.3 Sites in the catchment ...... 80 4.1.4.4 Sites in the Kyabra catchment ...... 84 4.1.5 Sites in the Bulloo catchment...... 86 4.2 Water quality at the sampling sites, 2006 – 2008...... 89 4.2.1 Methods...... 89 4.2.2 Results...... 90 4.2.3 Discussion ...... 97 4.3 Summary of the sampled catchments 2006 - 2008: aridity, season, hydrology and water quality...... 99 5. The distribution of fish in the Queensland Lake Eyre and Bulloo-Bancannia basins ...... 102 5.1 Introduction ...... 102 5.2 Methods...... 111 5.2.1 Field methods ...... 111 5.2.2 Data analysis...... 112 5.3 Results...... 115 5.3.1 Fish species presence/absence...... 115 5.3.2 Fish species abundance in the Mulligan catchment...... 125 5.3.3 Fish species abundance in the Georgina catchment...... 127 5.3.4 Fish species abundance in the Diamantina catchment...... 129 5.3.5 Fish species abundance in the Thomson catchment ...... 131 5.3.6 Fish species abundance in the Barcoo catchment...... 134 5.3.7 Fish species abundance in the Cooper catchment...... 135 5.3.8 Fish species abundance in the Kyabra catchment...... 136 5.3.9 Fish species abundance in the Bulloo catchment ...... 137 5.4 Discussion ...... 138 6. Fish recruitment in the Queensland Lake Eyre and Bulloo-Bancannia basins...... 153 6.1 Introduction ...... 153 6.2 Methods...... 159 6.2.1 Data analysis...... 160 6.3 Results...... 162 6.3.1 Bony bream ( Nematolosa erebi )...... 162 6.3.2 Cooper Creek catfish ( Neosiluroides cooperensis ) ...... 165 6.3.3 Hyrtl’s Tandan ( Neosiluris hyrtlii )...... 166 6.3.4. Silver tandan ( Porochilus argenteus )...... 170 6.3.5. Australian smelt ( Retropinna semoni )...... 175 6.3.6. Desert rainbowfish ( Melanotaenia splendida tatei ) ...... 178 6.3.7. Glassfish or Northwest Ambassis ( Ambassis sp.)...... 182 6.3.8. Yellowbelly ( Macquaria sp.) ...... 185 6.3.9. Banded grunter ( Amniataba percoides ) ...... 189 6.3.10. Welch’s grunter ( Bidyanus welchi )...... 190 6.3.11. Spangled perch ( Leiopotherapon unicolor )...... 191 6.3.12. Barcoo grunter ( Scortum barcoo )...... 197 6.3.13 Golden goby ( Glossogobius aureus )...... 197

x 6.3.14 Carp gudgeon ( Hypseleotris sp.) ...... 198 6.4 Discussion ...... 202 7. Movement, colonisation and extirpation of fish in four catchments in far western Queensland...... 220 7.1 Introduction ...... 220 7.2 Methods...... 226 7.2.1 Data analysis...... 226 7.3 Results...... 229 7.3.1 Fish communities at permanent and ephemeral sites in the Georgina and Mulligan catchments, 2006 - 2008 ...... 229 7.3.2 Fish assemblages in recently-filled ephemeral sites and permanent sites in the Thomson and Barcoo catchments...... 230 7.3.3 The influence of antecedent hydrology on fish communities in ephemeral waterholes ...... 232 7.3.4 Location of permanent waterholes in the Georgina and Mulligan catchments...... 234 7.3.5 distances following flooding in the Georgina and Mulligan catchments in January and February 2007...... 236 7.3.6 Fish species presence/absence at distances from sites in the Georgina and Mulligan catchments following major flooding...... 237 7.3.7 Fish communities in isolated waterholes in the Georgina/Mulligan catchments: the influence of migration distance on assemblage structure ...... 239 7.3.8 Fish species colonisation in the Mulligan, Georgina, Thomson and Barcoo catchments...... 242 7.3.9 Waterhole drying in the Georgina and Mulligan catchments...... 251 7.3.10 Localised extinctions at waterhole scale in the Georgina and Mulligan catchments...... 252 7.4 Discussion ...... 255 8. Fish populations in the rivers of far western Queensland: discussion and conclusions...... 266 8.1 Summary of data from the current study ...... 266 8.1.1 Fish distribution patterns...... 267 8.1.2 Fish recruitment...... 271 8.1.3 Fish movement...... 273 8.2. Concepts of fish ecology in Australian arid-zone rivers...... 275 8.3. Fitting the conceptual model: examples from the Queensland ...... 285 8.4. Fish communities within the Queensland Lake Eyre and Bulloo-Bancannia basins: current status, threats, management implications and recommendations ..289 Appendices...... 297 References...... 409

xi List of Tables

Table 3.1 Summary table of fish species present in the major catchments of the Queensland Lake Eyre and Bulloo-Bancannia basins...... 41 Table 4.1 Hydrological history of Vergemont Creek for the period December 2006 – March 2008...... 66 Table 4.2 Hydrological history of Native Waterhole for the period December 2006 – March 2008...... 67 Table 4.3 Hydrological history of Waterloo for the period September 2006 – November 2008...... 70 Table 4.4 Hydrological history of the Thomson Main Channel waterholes for the period September 2006 – March 2008...... 71 Table 4.5 Hydrological history of Coolagh for the period June 2006 – March 2008. .74 Table 4.6 Hydrological history of Coolagh 2 for the period June 2006 – March 2008 ...... 75 Table 4.7 Hydrological history of Coolagh 3 for the period June 2006 – March 2008 ...... 78 Table 4.8 Hydrological history of Isisford for the period December 2006 – March 2008...... 79 Table 4.9 Hydrological history of Murken for the period September 2006 – March 2008...... 81 Table 4.10 Hydrological history of Currareva for the period September 2006 – March 2008...... 82 Table 4.11 Hydrological history of Shed for the period September 2006 – March 2008...... 83 Table 4.12 Hydrological history of all Kyabra Creek waterholes for the period September 2006 – March 2008...... 85 Table 4.13 Hydrological history of all waterholes for the period April 2007 – March 2008...... 88 Table 4.14 Tukey’s post hoc test results for pairwise comparisons of surface water temperature between sampling periods ...... 90 Table 4.15 Tukey’s post hoc test results for pairwise comparisons of surface dissolved oxygen results between catchments and sampling periods ...... 92 Table 4.16 Tukey’s post hoc test results for pairwise comparisons of turbidity between catchments (n.s. = not significant)...... 96 Table 4.17 Fish sampling occasions and antecedent flows in the Lake Eyre and Bulloo-Bancannia basins between September 2006 and March/April 2008, with notes on salinity levels...... 100 Table 5.1 Fish species presence/absence in sampled catchments of the Queensland Lake Eyre and Bulloo-Bancannia basins from 2006 - 2008...... 116 Table 5.2 Summary of One-Way ANOSIM results comparing fish presence/absence throughout the Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008...... 118 Table 5.3 SIMPER analysis comparing fish presence/absence by catchment in the Queensland Lake Eyre and Bulloo-Bancannia Basins...... 119 Table 5.4 SIMPER analysis comparing fish species presence/absence in relation to season (or sampling time) in the Queensland Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008...... 122

xii Table 5.5 SIMPER analysis comparing fish species presence/absence in relation to antecedent hydrology in the Queensland Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008...... 124 Table 5.6 SIMPER analysis comparing fish species presence/absence in relation to waterhole type in the Queensland Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008 ...... 125 Table 5.7 Fish species recorded at Lake Mary and Walkaba/Jimberella waterholes in the Georgina catchment in November 2008...... 129 Table 5.8 Fish species recorded at Rocky Crossing, Spring Creek and Conn waterhole in the Diamantina catchment in November 2008...... 131 Table 5.9 Fish species recorded at Lake Dunn, Waterloo and Thomson Main Channel waterholes in the Thomson catchment in November 2008 ...... 133 Table 6.1. Size categories (standard length SL in millimetres) used for subsequent analysis of length frequency distributions of sampled fish species in the Lake Eyre and Bulloo-Bancannia basins, September 2006 – March/April 2008...... 160 Table 6.2 Summary of One-Way ANOSIM results comparing the size structure of bony bream populations throughout the Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008 ...... 163 Table 6.3 Summary of One-Way ANOSIM results comparing the size structure of Hyrtl’s tandan populations throughout the Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008...... 168 Table 6.4 SIMPER analysis comparing size frequency of Hyrtl’s tandan in relation to antecedent flow in the Queensland Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008 ...... 168 Table 6.5 Summary of One-Way ANOSIM results comparing the size structure of silver tandan populations throughout the Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008...... 172 Table 6.6 SIMPER analysis comparing size frequency of silver tandan in relation to season (or sampling time) in the Queensland Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008...... 172 Table 6.7 Summary of One-Way ANOSIM results comparing the size structure of Australian smelt populations throughout the greater Cooper Creek catchment from September 2006 – March/April 2008...... 176 Table 6.8 SIMPER analysis comparing size frequency of Australian smelt in relation to antecedent flow in the Cooper Creek catchment September 2006 – March/April 2008...... 177 Table 6.9 SIMPER analysis comparing size frequency of Australian smelt in relation to sampling time in the Cooper Creek catchment ...... 177 Table 6.10 Summary of One-Way ANOSIM results comparing the size structure of desert rainbowfish populations ...... 181 Table 6.11 Summary of One-Way ANOSIM results comparing the size structure of glassfish populations ...... 185 Table 6.12 SIMPER analysis comparing size frequency of glassfish populations in relation to antecedent flow...... 185 Table 6.13 Summary of One-Way ANOSIM results comparing the size structure of yellowbelly populations...... 187 Table 6.14 Summary of One-Way ANOSIM results comparing the size structure of banded grunter populations...... 190 Table 6.15 Summary of One-Way ANOSIM results comparing the size structure of spangled perch populations...... 193

xiii Table 6.16 SIMPER analysis comparing size frequency of spangled perch in relation to season (or sampling time) in the Queensland Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008. (Early summer = September – December, late summer = January – April, winter = May – August. Insignificant differences not included in this table.) ...... 193 Table 6.17 SIMPER analysis comparing size frequency of spangled perch populations in relation to antecedent flow...... 194 Table 6.18 Summary of One-Way ANOSIM results comparing the size structure of carp gudgeon populations...... 200 Table 7.1 SIMPER analysis comparing fish assemblages in the Georgina and Mulligan catchments at all permanent and ephemeral sites during the study ...... 230 Table 7.2 SIMPER analysis comparing fish assemblages at sites in the Thomson and Barcoo catchments categorised by presence in permanent and ephemeral waterholes ...... 232 Table 7.3 SIMPER analysis comparing fish assemblages at ephemeral sites filled from major flooding with those filled from smaller flows...... 234 Table 7.4 Migration distances* to previously dry waterholes in the Georgina and Mulligan catchments in January/February 2007...... 237 Table 7.5 Fish species sampled at distances from confluence sites with main channel waterholes in the Georgina and Mulligan catchments, April 2007...... 238 Table 7.6 Summary of One-Way ANOSIM results comparing fish assemblages at distances from the main channel of the in the Georgina and Mulligan catchments ...... 241 Table 7.7 SIMPER analysis comparing fish assemblages at sites in the Georgina and Mulligan catchments categorised by distance from the main channel of the Georgina ...... 242 Table 7.8 Summary table of vagility for 10 fish species from the Georgina/Mulligan and/or Barcoo/Thomson catchments during the study...... 255 Table 8.1 Fish species presence in the studied catchments 2006 – 2008...... 269 Table 8.2 Summary of no-flow and flow dependent recruitment for fish species from the study area...... 273

xiv List of Figures

Figure 1.1 A working model describing fish ecology in the Australian arid-zone...... 10 Figure 2.1 Australian drainage divisions...... 14 Figure 2.2 Mean monthly maximum (above) and minimum temperatures (below) from four weather stations located in the study area...... 16 Figure 2.3 Hydrograph of the Georgina River at Roxborough Downs from 1967 – 2008...... 17 Figure 2.4. Mean annual run-off (mean annual flow divided by catchment area) ± co- efficient of variation in annual flows in the Georgina, Diamantina, Cooper and Bulloo catchments (a: above), and mean monthly flow (mean daily flow for each month averaged across years when data is available) for the same catchments (b: below)....18 Figure 2.5. Mean number of zero flow days (top line) and mean duration of high flows (bottom line) in the Georgina, Diamantina, Cooper and Bulloo catchments ...... 21 Figure 3.1 Native fish species from the major catchments of the Queensland Lake Eyre and Bulloo-Bancannia basins...... 42 Figure 4.1 Site map showing the location of all sites sampled during the study in the Mulligan, Georgina, Diamantina, Thomson, Barcoo, Cooper, Kyabra and Bulloo catchments between September 2006 and March/April 2008 ...... 44 Figure 4.2 The location of sampling sites in November 2008 showing Lake Mary and Walkaba/Jimberella waterhole in the Georgina catchment, Conn waterhole, Mayne River and Spring Creek in the Diamantina catchment and Lake Dunn in the Thomson catchment...... 45 Figure 4.3 Hydrology of the studied catchments between January 2005 and June 2009...... 46 Figure 4.4 Pulchera Waterhole in April 2007 (left), August 2007 (middle) and November 2007 (right)...... 50 Figure 4.5 S Bend Gorge in April 2007 (left) and shortly before it dried in August 2007 (right)...... 50 Figure 4.6 Dune in April 2007 (left) and August 2007 (right)...... 50 Figure 4.7 Kunnamuka Swamp in April 2007 (left) and August 2007 (right)...... 51 Figure 4.9 Daily discharge (ML) of the Georgina River at Roxborough Downs from October 2006 – May 2008 ...... 52 Figure 4.10 Daily heights for the Georgina River at Roxborough Downs during the period of flooding in early 2007 ...... 53 Figure 4.11 The Main Channel site on the Georgina River, showing the connected channel in April 2007 (left), and waterhole recession following a drying period in August (middle) and November (right)...... 56 Figure 4.12 Lake Idamea showing the difference between full capacity...... 56 Figure 4.13 Lower Lake at full capacity following the recession of floodwaters in April 2007 (left), and as it dried down in August 2007 (middle) and November 2007 (right)...... 56 Figure 4.14 Parapituri showing dry season recession between August 2007 (left) and November 2007 (right)...... 57 Figure 4.15 Lake Mary, the only waterhole close to the northern Lake Eyre Basin divide that retained water in the Georgina catchment in November 2008...... 57 Figure 4.16 Walkaba waterhole in the Georgina catchment, November 2008...... 57 Figure 4.18 Daily discharge of the at Diamantina from October 2006 – May 2008. (Source: Queensland Department of Environment and Resource Management)...... 58

xv Figure 4.17 An example of the landscape surrounding sampling sites in the Diamantina catchment showing eroded escarpments stretching to the north-east from the Brighton Downs/Diamantina Lakes border...... 60 Figure 4.19 Lake Billyer during a drying period between April 2007 (left) and August 2007 (middle),...... 60 Figure 4.20 2 Mile waterhole showing water level recession between April (left) and August (right) 2007 ...... 60 Figure 4.21 Hunter’s Gorge, showing the difference in water level created by a small flow in early November 2007 (right) and the level of the waterhole in August 2007 (left)...... 60 Figure 4.22 Warracoota in August 2007 (left) and November 2007 (right) demonstrating the rise in water levels created by a small flow in early November.....61 Figure 4.23 Lake Constance in August (left) and November (right) 2007...... 61 Figure 4.24 Rocky Crossing on the Mayne River, Diamantina catchment, in November 2008 (left)...... 61 Figure 4.25 The rockhole on Spring Creek, showing lancewood and river red gum riparian vegetation (right)...... 61 Figure 4.26 Conn Waterhole, the highest – or most northerly - permanent water on the Diamantina River, in November 2008...... 61 Figure 4.27 Daily discharge of the Thomson River at Longreach from January 2005 – May 2008...... 65 Figure 4.28 The Vergemont Creek site at its lowest level during the study in December 2006 following an extended period of no-inflows...... 68 Figure 4.29 Native Waterhole following an extended period of no-inflows in December 2006...... 68 Figure 4.30 The sampling site at Waterloo in August 2007 following a recent flow that filled the waterhole...... 68 Figure 4.31 Sites sampled in the Thomson River Main Channel...... 69 Figure 4.32 Lake Dunn, in the upper Thomson catchment, in November 2008...... 69 Figure 4.33 Daily discharge of the at Retreat from January 2005 – May 2008...... 73 Figure 4.34 Coolagh waterhole on the Barcoo River in June 2006 showing overhanging riparian vegetation...... 76 Figure 4.35 Coolagh 2 following drying in December 2006 (left) and immediately after re-filling in November 2007 (right)...... 76 Figure 4.36 Coolagh 3 waterhole during a drying phase...... 76 Figure 4.37 The sampling site at Isisford in November 2007 ...... 76 Figure 4.38 Waterholes sampled in the Cooper catchment (l to r): Murken waterhole (December 2006), Currareva (December 2006) and Shed (September 2006)...... 77 Figure 4.39 Springfield waterhole in Kyabra Creek in September 2006 following an extended period of zero inflows...... 77 Figure 4.40 Springfield South receding during a drying period from September 2006 (left), to December 2006 (centre) and January 2007 (right)...... 77 Figure 4.41 The sampling site at One Mile waterhole on Kyabra Creek during sampling in November 2007, and showing the effects of a small within-channel flow between evening (left) and the following morning (right)...... 77 Figure 4.42 Daily discharge of the Bulloo River at Quilpie from January 2005 – May 2008...... 86 Figure 4.43 The Main Channel site on the Bulloo River (left) and the Shearing Shed site (right) (August 2006) ...... 87

xvi Figure 4.44 Mean surface water temperature (± standard error) pooled from all sites sampled from September 2006 to November 2008...... 91 Figure 4.45 Mean (± standard error) water temperature at the surface (grey bars) and at a depth of 2 metres (black bars) at all sites from December 2006 to March 2008 ..91 Figure 4.46 Mean surface dissolved oxygen (% saturation) ± standard error in permanent main channel waterholes, ephemeral main channel waterholes and ephemeral waterholes/lakes pooled from all sites sampled across the study area, 2006 - 2008...... 93 Figure 4.47 Mean surface dissolved oxygen (% saturation) ± standard error by catchment in the Lake Eyre and Bulloo-Bancannia basins, 2006 – 2008...... 93 Figure 4.48 Mean surface dissolved oxygen (% saturation) ± standard error from September 2006 to November 2008 at all sites in the Lake Eyre and Bulloo-Bancannia basins, 2006 – 2008...... 94 Figure 4.49 Mean (± standard error) dissolved oxygen at the surface (grey bars) and at a depth of 2 metres (black bars) at all sites from December 2006 to March 2008...... 94 Figure 4.50 Mean pH ± standard error by catchment in the Lake Eyre and Bulloo- Bancannia basins, 2006 – 2008...... 95 Figure 4.51 Mean turbidity (Secchi depth in cm) ± standard error by catchment, 2006 - 2008...... 96 Figure 4.52 Mean conductivity (µS/cm) by catchment, 2006 – 2008...... 97 Figure 5.1 Two Dimensional NMS ordination plot of fish communities transformed for presence/absence across four sampling periods in seven catchments ...... 117 Figure 5.2 Mean (±S.E.) number of species sampled at all sites in the Queensland Lake Eyre and Bulloo-Bancannia basins in late summer, winter and early summer.122 Figure 5.3 Mean (±S.E.) number of species sampled in the Bulloo (grey bars), Kyabra (pink bars), Cooper (red bars), Thomson (orange bars), Barcoo (purple bars), Diamantina (blue bars), Georgina (green bars) and Mulligan catchments (yellow bars) in April 2007, August 2007, November 2007 and March/April 2008...... 123 Figure 5.4 Proportional abundance of sampled fish species between April 2007 and March/April 2008 in the Mulligan catchment ...... 126 Figure 5.5 Sites in the Mulligan catchment, showing the location of Dune Pond and Kunnamuka Swamp and their isolation from the main channel of the Mulligan to the east...... 127 Figure 5.6 Proportional abundance of sampled fish species between April 2007 and March/April 2008 in the Georgina catchment ...... 128 Figure 5.7 Proportional abundance of sampled fish species between April 2007 and March/April 2008 in the Diamantina catchment...... 130 Figure 5.8 Proportional abundance of sampled fish species between September 2006 and March/April 2008 in the Thomson catchment ...... 132 Figure 5.9 A sleepy cod sampled from Waterloo Waterhole in the Thomson catchment in November 2008...... 133 Figure 5.10 Proportional abundance of sampled fish species between September 2006 and March/April 2008 in the Barcoo catchment...... 134 Figure 5.12 Proportional abundance of sampled fish species between September 2006 and March/April 2008 in the Kyabra catchment...... 136 Figure 5.13 Proportional abundance of sampled fish species between September 2006 and March/April 2008 in the Bulloo catchment...... 137 Figure 5.14 Desert rainbowfish sampled during the current study included vividly coloured specimens from the Mulligan catchment (left) and less-colourful specimens from all other catchments (right)...... 142

xvii Figure 6.1 Total percentage frequency of size classes of bony bream summed from all sampled catchments in April 2007 (light bars), August 2007 (dark bars), November 2007 (clear bars) and March/April 2008 (hatched bars)...... 162 Figure 6.2 Two Dimensional NMS ordination plot of log 10(x + 1) total catch data for bony bream in four size classes (50, 100, 150 and >150mm SL), across four sampling periods in eight catchments...... 163 Figure 6.3 Bony bream size structure from Pulchera waterhole in the Mulligan catchment in November 2007 following a drying period since February...... 164 Figure 6.4 Total percentage frequency of size classes of Cooper Creek catfish from all sampled greater Cooper catchments (Thomson/Barcoo/Cooper) summed from September 2006 to March/April 2008...... 165 Figure 6.5 Total percentage frequency of size classes of Hyrtl’s tandan summed from all sampled catchments...... 166 Figure 6.6 Two Dimensional NMS ordination plot of log 10 (x + 1) total catch data for Hyrtl’s tandan in four size classes (100, 150, 200 and >200mm SL), across four sampling periods in seven catchments ...... 167 Figure 6.7 Length-frequency distributions of Hyrtl’s tandan in the Georgina (a) and Cooper (b) catchments in April 2007 and March/April 2008...... 169 Figure 6.8 Total percentage frequency of size classes of silver tandan summed from all sampled catchments...... 170 Figure 6.9 Two Dimensional NMS ordination plot of log 10 (x + 1) total catch data for silver tandan in four size classes (100, 150, 200 and >200mm SL), across four sampling periods in eight catchments ...... 171 Figure 6.10 Length frequency distributions of silver tandan through time in Kyabra Creek (a: top) and the Thomson River (b: bottom)...... 173 Figure 6.11 Hydrograph of the Georgina River at Roxborough Downs (a), silver tandan sampled in the Georgina catchment from April 2007 to March/April 2008 (b), and silver tandan sampled in the Mulligan catchment from April 2007 to November 2007 (c)...... 174 Figure 6.12 Two Dimensional NMS ordination plot of log 10 (x + 1) total catch data for Australian smelt in three size classes (20, 40, >40mm SL), across seven sampling periods in three catchments...... 175 Figure 6.13 Seasonal distribution of Australian smelt ≤20mm (SL) from combined sites in the greater Cooper catchment...... 178 Figure 6.14 Total percentage frequency of size classes of all desert rainbowfish collected from all sampled catchments...... 179 Figure 6.15 A comparison of the number of desert rainbowfish sampled during March/April 2008...... 179 Figure 6.16 Two Dimensional NMS ordination plot of log 10 (x + 1) total catch data for desert rainbowfish in four size classes (20, 40, 60 and >60mm SL), across four sampling periods in eight catchments ...... 180 Figure 6.17 Length frequency distribution graphs of desert rainbowfish populations in the Georgina catchment...... 181 Figure 6.18 Total percentage frequency of size classes of glassfish summed from all sampled catchments...... 182 Figure 6.19 Percentage frequency of size classes of glassfish in relation to antecedent hydrology...... 183 Figure 6.20 Two Dimensional NMS ordination plot of log 10 (x + 1) total catch data for glassfish in three size classes...... 184

xviii Figure 6.21 Total percentage frequency of size classes of yellowbelly summed from all sampled catchments...... 186 Figure 6.22 Two Dimensional NMS ordination plot of log 10 (x + 1) total catch data for yellowbelly in four size classes...... 186 Figure 6.23 Yellowbelly length frequency distributions and antecedent hydrology .188 Figure 6.24 Percentage frequency of size classes of banded grunter from the Georgina catchment...... 189 Figure 6.25 Total percentage frequency of size classes of Welch’s grunter collected from all sampled catchments ...... 190 Figure 6.26 Total percentage frequency of size classes of spangled perch summed from all sampled catchments ...... 191 Figure 6.27 Two Dimensional NMS ordination plot of log 10 (x + 1) total catch data for spangled perch...... 192 Figure 6.29 Length frequency of spangled perch populations...... 195 Figure 6.30 The size structure of spangled perch populations in drying waterholes in the Mulligan catchment ...... 196 Figure 6.31 Total percentage frequency of size classes of Barcoo grunter populations ...... 197 Figure 6.32 Total percentage frequency of size classes of golden goby populations from sites in the Georgina catchment...... 198 Figure 6.33 Total percentage frequency of size classes of carp gudgeon populations from all sampled catchments ...... 199 Figure 6.34 Two Dimensional NMS ordination plot of log10(x + 1) total catch data for carp gudgeon in three size classes ...... 199 Figure 6.35 Carp gudgeon length frequency in the Kyabra catchment...... 201 Figure 6.36 A conceptual model of the recruitment behaviour of fish species in the Lake Eyre and Bulloo-Bancannia basins...... 219 Figure 7.1 Two Dimensional NMS ordination plot of log 10 (x + 1) CPUE data for fish communities in the Georgina and Mulligan catchments ...... 229 Figure 7.2 Two Dimensional NMS ordination plot of log 10 (x + 1) CPUE data for fish assemblages at sites in the Thomson and Barcoo catchments...... 231 Figure 7.3 Two Dimensional NMS ordination plot of log 10 (x + 1) CPUE data for fish assemblages at sites experiencing major flooding (overbank flows) in the 3 months prior to sampling (circles) and bank-full flows or less (stars) during the sampling period...... 233 Figure 7.4 The location of 8 permanent waterholes in the Georgina catchment in November 2008...... 235 Figure 7.5 Ephemeral waterholes in the Pituri Creek/Georgina catchment and the Mulligan catchment...... 236 Figure 7.6 A typical sand dune in the eastern Simpson Desert close to Kunnamuka Swamp ...... 239 Figure 7.8 Size frequency histograms of populations of bony bream pooled from sites >100 kilometres from the main channel (left) and <10 kilometres from the main channel following flooding in the Georgina and Mulligan catchments ...... 243 Figure 7.9 Size frequency histograms of populations of bony bream pooled from ephemeral and permanent sites in the Thomson and Barcoo catchments ...... 243 Figure 7.10 Size frequency histograms of populations of Hyrtl’s tandan...... 244 Figure 7.11 Size frequency histograms of populations of silver tandan pooled from sites >100 kilometres from the main channel (left) and <10 kilometres from the main channel following flooding in the Georgina and Mulligan catchments ...... 245

xix Figure 7.12 Size frequency histograms of populations of silver tandan pooled from ephemeral (left) and permanent (right) sites in the Thomson and Barcoo catchments ...... 246 Figure 7.13 Size frequency histograms of populations of desert rainbowfish ...... 247 Figure 7.14 Size frequency histograms of populations of glassfish...... 248 Figure 7.15 Size frequency histograms of populations of yellowbelly...... 249 Figure 7.16 Size frequency histograms of populations of spangled perch pooled from sites >100 kilometres from the main channel (left) and <10 kilometres from the main channel following flooding in the Georgina and Mulligan catchments ...... 250 Figure 7.17 Size frequency histograms of populations of spangled perch pooled from ephemeral and permanent sites in the Thomson and Barcoo catchments ...... 250 Figure 7.19 Fish populations sampled at S Bend Gorge in April 2007 (light bars) and two weeks prior to complete drying in August 2007 (dark bars)...... 252 Figure 7.20 Fish populations sampled at Lower Lake in April 2007 (light bars) and three weeks prior to complete drying in August 2007 (dark bars)...... 253 Figure 7.21 Fish populations sampled at Pulchera waterhole in April 2007 (light bars), August 2007 (dark bars) and prior to drying in November 2007 (clear bars)...... 254 Figure 8.1 A working model for fish ecology in the Australian arid-zone ...... 283 Figure 8.2 The relationship between migration capability or vagility and recruitment strategies for fish species from the rivers of far western Queensland...... 284 Figure 8.3 A modified source/sink model for fish communities in Australian arid zone rivers...... 285 Figure 8.4 Case studies demonstrating the source/sink concept by reference to the temporal changes in the fish assemblage at a source waterhole (Waterloo; top) and a sink waterhole (Pulchera; bottom) ...... 287

xx List of Appendices

Appendix 1. Techniques for sampling larval and juvenile fish in waterholes of the Queensland Lake Eyre Basin: A preliminary field study, June 2006...... 297 Appendix 2. Water quality data September 2006 – November 2008, Lake Eyre and Bulloo-Bancannia Basins...... 325 Appendix 3. Length frequency histograms – all fish species, all sites, Queensland Lake Eyre and Bulloo-Bancannia basins, 2006 - 2008...... 328 Appendix 4. Field identification of catfish in the Queensland Lake Eyre and Bulloo Bancannia Basins...... 396 Appendix 5. Hardyhead ( Craterocephalus spp.) in the Queensland Lake Eyre Basin ...... 402

xxi Acknowledgements

Researching and writing a PhD is a reasonably large undertaking, and the immensity of the task is probably only realised by those who have similarly immersed themselves in a single topic for longer than three occasionally lonely years. Without help, they just don’t happen. So here’s the list – apologies for any omissions.

Relying as the study did on multiple sampling across a vast area of isolated Australia, I firstly acknowledge the land owners and managers of the many stations, reserves and national parks upon which I camped and worked; the Richardsons from Leopardwood Park on the Bulloo, the Taylors from Coolagh on the Barcoo, Bob and Bronwen Morrish from Springfield on Kyabra Creek, the Kidds and Smiths from the Cooper close to , the Fergusons from Durham Downs (a little further down the Cooper), the Emmotts from Noonbah on the Thomson, the Youngs from Brighton Downs on the Diamantina, the Kingstons, John Clemments, Ian Andreassen and Ronelle Frazer from Diamantina Lakes National Park, and in the Georgina the McGlincheys from Badalia , the Bryants from Roxborough Downs , the Fennells, formerly of Linda Downs , the Millers from Rocklands and the Bryces from Glenormiston . For facilitating access to all NAPCo properties I thank Delphine Puxtey. In the Mulligan, thanks are due to Scott and Sajida Morrison, former managers of Ethabuka, and to the many Bush Heritage Australia staff with whom I’ve liaised and become acquainted with since 2006.

Without the support of a large group of unpaid volunteers, I would never have been able to cover the kilometres necessary to assemble the database. Leanne Faulks, Spider Tyack, Joel Huey and Ryan Woods accompanied me on the earliest sampling trips. In December 2006, January 2007, November 2007 and March 2008 Tim Smith came along for the ride, and in January 2007 and March 2008 his family (Leanne, Sarah and Josh) and mine (Al, Saffi and Emma) also put up with smelly nets and dirty cars. Mick Brigden came along on the first big trip in April 2007, and in the far west we were also joined by Vanessa Bailey, Alun Hoggett, Joan Powling and Don Cook. During this trip we spent quality time with Angus Emmott and Bill Wilkes at Ethabuka. We didn’t know at the time that it would be Bill’s last trip to the desert, but

xxii every time I return to that part of the country I think of him. Angus jumped in the passenger seat again in August 2007 (though not – it must be added – into the admittedly cold water). I was extremely fortunate to be asked by Jenny Silcock to accompany her on the ever-so-slightly storm-affected November 2008 excursion, which allowed me to collect extra samples throughout the Georgina and further demonstrate the presence of golden goby in the Diamantina. Jen’s help with everything from mapping to accommodation, all things botanical and even occasional fish-wenchery is duly acknowledged.

Across and outside the country, many of the Australian freshwater fish geeks have provided support, advice and general amusement. The short list of slightly-eccentric- yet-almost-always enthusiastic fishos includes Harry Balcombe, Jeff Johnson, Helen Larson, Mark Kennard, Brendan Ebner, Linters and JK and their respective mobs from Canberra and , Damien and Aaron up at JCU, Michael Hammer down in Adelaide, the ex-pat Peter Unmack over elsewhere, Rollsy, and Morgs and Beatts from over the other side of the desert. In the government, many people have kindly bent the rules where necessary or provided timely advice. These charitable types include Vanessa Bailey from what used to be called EPA, Jaye Lobegeiger from what used to be called NRW and Gary Muhling, Michael Hutchison and Butch from what used to be called DPI&F. Darren Shepherd from Toohey Forest Environmental Ed Centre became particularly talented at offering me a couple of days work here and there when it was most needed. At Griffith Uni I owe Harry a huge debt of gratitude for helping to keep me heading in roughly the right direction, to Deslie Smith for helping to keep all the financial things heading in the right direction, to Lacey Shaw for deftly managing to alter her phone voice from ‘Centre for Riverine Landscapes’ to ‘Australian Rivers Institute’ almost overnight, to Fran Sheldon for slinging me a bit of work and to Ben Stewart-Koster and Patrick Laceby for being able to quote much of Gran Torino verbatim and in a similar manner to Clint Eastwood.

In early 2008 I took some leave from the PhD, and through Jenny became acquainted with the slightly-intense-but-altogether-fascinating world of Russell Fairfax and Rod Fensham, and through Rod, ultimately a real job with Bush Heritage Australia. So thanks to Jen, Rod, Russ and the Hairy Todgers (Kate, Max, Murray, Paul etc..) for widening the net yet again in 2009. An acknowledgements section would not be

xxiii complete without mentioning Dr Martin Denny – he of the impossibly large swag and Atomic coffee machine – and Richard ‘Dr Duck’ Kingsford – he of the eternal passion and handy cynicism.

I thank my supervisors Professor Angela Arthington and Professor Stuart Bunn for assistance with the theoretical direction of this thesis, for their intellectual guidance and for allowing me to work on a PhD topic that broke almost all the rules regarding size of study area and riskiness of experimental design. Special thanks are due to Angela for providing timely encouragement, thoughtful editing and useful guidance that has resulted in a much-improved and comprehensive piece of work. My candidature was supported by a Griffith University Postgraduate Research Award and a top-up scholarship from the eWater CRC. In addition I thank the Australian Rivers Institute and Griffith University for financial, infrastructure and equipment support over the last four years.

I owe a particularly huge debt of gratitude to my family – Al, Saffi and Emma. In 2006 I dragged them from our little farm in NSW to a big city, and then spent most of the next four years back out on other peoples’ farms in outback Queensland. Their patience is unbelievable. I also thank my parents, Jen ő and Geraldine, my sister Julie and my brother-in-law Gav who provided us – me most particularly – with extra support when the wheels fell off.

Last, I thank two inspirational people who have played pivotal roles in the direction my life has taken since 2003 - Robyn Watts, who successfully dragged me through Honours, and Angus Emmott, who has occasionally pushed and pulled me along through the work in western Queensland.

This thesis is about real things in the real world – most often fish in muddy waterholes. It has been deliberately written using simple language where possible, so that anybody who may be interested – pastoralists, land managers, scientists, community people – can hopefully read and understand it. I sincerely hope that it is a useful body of work that can be used to inform the management of Australia’s arid rivers and ecosystems.

xxiv

xxv 1. Introduction

Arid to semi-arid regions comprise 47% of the earth’s land surface, yet the ecology of these areas, and particularly of arid-zone rivers, remains poorly known (Kingsford and Thompson 2006). In general, concepts describing the functioning of large rivers have been developed with reference to temperate and tropical areas (Vannote et al. 1980; Junk et al. 1989), and consequently, such concepts have been demonstrated to have deficiencies when applied to arid systems that represent the extreme margin of freshwater habitation (Puckridge 1999).

Australia is the driest inhabited continent and is characterised by large areas devoid of both surface water and large river channels. Nevertheless, many of the arid-zone rivers that do exist are unique as they support ecological communities that have not been subjected to human-induced changes such as regulation for water abstraction (Kingsford et al. 2006b). These free-flowing rivers therefore represent an ideal study area within which to investigate ecological concepts and processes in arid systems (Balcombe and Arthington 2009), despite the fact that – ironically – they flow for such comparatively short periods of time and exist as isolated waterholes for the remainder. Although Australian arid-zone rivers such as the Cooper and Diamantina have been demonstrated to possess some of the most variable flow regimes on earth (Puckridge et al. 1998), these rivers support robust populations of native flora (Capon 2003), invertebrates (Marshall et al. 2006), birds (Kingsford et al. 2006a), turtles (Georges et al. 2006) and fish (Puckridge 1999; Costelloe et al. 2004; Arthington et al. 2005; Balcombe et al. 2007; Balcombe and Arthington 2009).

Using data from eight catchments collected across a range of flow conditions from extended dry periods through to episodes of major flooding, the studies presented in this dissertation explore the distribution and biogeography of fish species across catchments, their broad recruitment patterns and their ability to migrate to distant and formerly isolated areas when unpredictable flow events occur. The findings from the individual chapters are then considered together and with reference to existing concepts and models of riverine function and population biology in order to conceptualise the ecology of fish communities in these highly variable landscapes.

1 The final section of the thesis presents research recommendations and possible management implications for river systems in far western Queensland based upon these studies and existing literature.

1.1 Ecological adaptations in arid environments

Desert ecosystems, or areas that receive <500mm of rain annually (Kingsford and Thompson 2006) are prone to severe climatic unpredictability. Prolonged dry periods, occasional rainfall and extremely high temperature variation generally characterise such systems, and for biota to survive and reproduce they must exhibit physiological, behavioural and/or life history traits that enable them to persist under variable conditions. These traits can be extremely varied, ranging from migratory behaviour enabling species to move to areas containing more resources to hibernation and aestivation in order to avoid the harshest seasonal conditions. In general, aridity itself, or the amount and frequency of rainfall and subsequent run-off, is the key determinant influencing biological diversity and population abundance in such systems (Stromberg 2007).

Resistance, or the ability to cope with prolonged dry periods, and resilience, the ability to recover from prolonged dry periods, have been identified as crucial to the persistence of species in ecosystems subject to temporal disturbance (Lake 2000; Lake 2003; Bond et al . 2008). For species to survive in arid environments, they must possess certain traits enabling them to avoid localised extirpation such that reproduction, dispersal and colonisation can occur when conditions are favourable (Fausch and Bramblett 1991). Resilience traits include the possession of wide environmental tolerances, broad or adaptable food preferences (Medeiros 2004; Balcombe et al . 2005), the ability to maintain small ‘source’ populations in favourable habitats (Dunning et al . 1992), opportunistic usage of renewed flows (Scheurer and Fausch 2003) and rapid population recovery under more favourable conditions (Fagan et al . 2002; Robertson 2000; Balcombe et al. 2007). For some aquatic species, such as the brassy minnow, Hybognathus hankinsoni , population maintenance and increased population opportunism appear to be tightly interwoven, with larvae and juveniles dispersing to potentially new areas and breeding adults more inclined to inhabit

2 ‘safer’ areas (Scheurer and Fausch 2003). For species demonstrating resistance traits, such as brine shrimp ( Artemia spp.), persistence is linked not with maintaining mature populations but through a reproductive strategy involving the deposition of drought tolerant eggs. Similarly, plants in the Cooper Creek floodplain of western Queensland maintain their populations by persisting in a dormant state as soil seed banks, such that when rain or flow occurs, germination of a certain proportion can occur, but with a reserve amount of seed retained to allow for future germinations (Capon and Brock 2006). These examples demonstrate that the strategies an organism must possess in order to survive in arid environments range from the ability to withstand sub-optimal conditions within generations to the possession of a reproductive strategy facilitating persistence between generations (Lake 2003). Additionally, in order for an or plant to exploit a particular survival strategy, the risks associated with that strategy must be outweighed by the advantages afforded (Kozakiewics and Szacki 1995). In the case of the hairy-footed dunnart, Sminthopsis youngsoni , a central Australian marsupial that has been demonstrated to travel long distances in a radio-telemetry study, the benefits, such as increased access to resources, are more conducive to the long and short-term survival of the species and individual than the risks, such as energetic demand (Haythornthwaite and Dickman 2006). Conversely, for the desert tortoise, Gopherus agassizii , of California, site fidelity is always strong, but activity and movement have been demonstrated to occur far less in drier years (Freilich et al . 2000).

Arid zone biota fall into two broad categories: highly vagile ecological generalists with broad environmental tolerances that have extensive distributions in a variety of habitats, and locally endemic species with a limited distributional range. For species in the first category, it is worth noting that ‘desert’ populations of widespread species often exhibit different life cycles or aspects of life cycles than their relatives in temperate or coastal areas. As an example, the Western pond turtle, Clemmys marmorata , has an extended distribution in the western United States, but populations in Oregon hibernate for up to eight months in terrestrial environments, whereas populations in southern California hibernate for only two months. In contrast, populations of this turtle from the Mojave Desert do not appear to practice terrestrial hibernation at all, and only leave the water for nesting purposes (Lovich and Meyer 2002). Species falling into the second category – local endemics – can frequently have

3 extremely restricted ranges, and this generally makes them particularly vulnerable to threatening processes like habitat alteration and invasion or predation by alien species (Fairfax et al . 2007; Read 2003). Well-known examples of the extinction of local endemics resulting from contact with ‘outside’ communities include the extermination of the flightless bird fauna (such as dodos, Raphus cucllatus , and moas, Dinornis spp.) from islands of the Pacific following contact with humans. Localised endemic populations in arid environments, such as the fish species living in Australian Great Artesian Basin springs (Wager and Unmack 2000; Fairfax et al. 2007; Kodric-Brown et al. 2007) and lizard communities in northern (Read 2003) live within similar habitat ‘islands’ and are consequently similarly prone to extirpation from habitat degradation or introduced predators.

1.2 Freshwater ecology in arid environments

Considerable effort has been directed towards the development of concepts aimed at describing aquatic ecosystem processes (Vannote et al . 1980; Junk et al . 1989; Thorp and Delong 1994), and the presence or absence of flow, flow variability, timing, magnitude and intensity are considered to be the foundations upon which riverine processes are based (Walker et al. 1995; Bunn and Arthington 2002; Bunn et al. 2006). The River Continuum Concept (RCC: Vannote et al. 1980) is focused on longitudinal changes in organic matter and biota from the headwaters to the river mouth in temperate European rivers with highly predictable hydrology. The Flood Pulse Concept (FPC: Junk et al. 1989), developed primarily to describe tropical systems, considered the exchange of nutrients and biota between the main river channel and temporarily-inundated , and found that a predictable ‘flood- pulse’ underpinned ecosystem processes in such systems. The FPC suggests there is likely to be high species diversity in regular flood-pulse environments and that present species are likely to evince flood-cued life-cycle adaptations (Junk et al . 1989). In the Riverine Productivity Model (RPM: Thorp and Delong 1994 and 2002), the emphasis shifted to localised primary production as the main driver of riverine food webs. Working in a central Australian river, Puckridge (1999) found that theories developed in temperate and tropical areas, such as the RCC, FPC and RPM were not directly applicable to the Australian arid-zone. This led to his development of the ‘Flow Pulse

4 Model’ (1999). Specifically, Puckridge’s model re-defined the floodplain to include all periodically inundated areas of a river including channel sections, and expanded the definition of flow to include any hydrological occurrence from localised run-off events to major flooding. Puckridge’s work therefore provides a potentially more relevant model for arid-zone rivers, especially given that the rivers do not persist as permanent channels but as isolated waterholes for the majority of the time, and that in these highly variable environments, within-channel flows, however small, may be as ecologically significant as infrequent overbank flows (floods). Australian studies investigating fish recruitment have also demonstrated that the application of concepts based on a generally predictable, annual flood (Junk et al . 1989) may not necessarily apply in Australian dryland rivers. In particular, the prediction that native fish require flood events to breed has been questioned, as the recruitment of many native fish species in the Murray-Darling Basin has been demonstrated to occur independently of high flow events (Humphries et al. 1999; Meredith et al . 2002; Mallen-Cooper and Stuart 2003). These studies, and others from more arid areas (Puckridge 1999; Balcombe and Arthington 2009), have therefore highlighted the importance of low or no flow periods for the recruitment of native fish species and the benefits to larval and juvenile survival afforded by a variable hydrograph. Despite the existence of these studies relating to Australian , there remains an identified need for further work to elucidate the behaviour of organisms in rivers with highly variable and erratic flow regimes, particularly in river systems that are not subject to flow regulation (McMahon and Finlayson 2003; Propst et al. 2008). As a consequence, and against the additional backdrop of impending climate change, recent Australian studies and reviews are increasingly becoming focused on themes that are linked with Australia’s exceptional ‘dryness’, such as the impacts and effects of extended drought (Bond et al. 2008) and the importance of refuge waterholes in arid areas (Sheldon et al . in press). Nevertheless, although maintenance of natural flow variability has been identified as crucial to species’ persistence in arid aquatic environments (Eby et al. 2003; Arthington et al. 2005; Sheldon et al. in press), current knowledge gaps pertaining to the biology and ecology of particular species are an obstacle to effective management of arid-zone riverine systems. The current study attempts to address this knowledge gap for fish communities living at the limit of conditions for survival and recruitment in central Australia.

5 Investigation of arid-zone aquatic ecology at adequate spatial and temporal scales is problematic, predominantly because of the unpredictable nature of flow events and subsequent biological responses. This situation does not occur to the same degree in temperate and tropical systems, where rainfall and flooding occur far more regularly, and generally on an annual basis. In arid Australian rivers that demonstrate the most variable hydrology on Earth (Puckridge et al. 1998), community composition is often shaped by the presence or absence of flow, since flow variability drives the presence, connection and disconnection of riverine habitat (Sheldon and Thoms 2006). In practical terms, this means that survey-based population research conducted in systems with highly variable flow histories must meet two criteria. First, surveys must be conducted at a large enough spatial scale to encapsulate a variety of possible local and/or regional flow conditions and events, from zero flows to major floods. Second, surveys must be conducted on an adequate temporal scale to encounter as many sequential flow events and seasonal changes as possible. Although research aimed at investigating the community ecology and life history strategies of aquatic biota is increasingly employing a multi-scale approach (Sheldon and Walker 1998; Labbe and Fausch 2000; Scheurer and Fausch 2003; Arthington et al. 2005; Marshall et al. 2006), the need to target research at sufficiently large spatial scales (catchment or multiple catchments) without compromising detail at smaller scales (reach) remains a challenge (Fausch et al. 2002). Additionally, the difficulty of incorporating the variability of arid-zone rivers into assessments of their ecological condition has been recognised (Sheldon 2005), and is similarly challenging without access to long-term temporal data-sets. One solution to these difficulties is to substitute ‘space’ for ‘time’. Thus, although the temporal timescale of the current study is comparatively short (2 years) and therefore potentially constrained in demonstrating the effects of flow variability (Eby et al. 2003), this has been ameliorated to some degree by increasing the spatial scale of the study to include eight semi-arid to arid-zone catchments in far western Queensland. The study therefore captures the dynamics of fish assemblages across a wide range of flow conditions from areas where rainfall and discharge patterns are both spatially and temporally variable.

In addition to tracking population trends, the study of fish assemblages at broad spatial scales has the potential to contribute to biogeographic knowledge. The presence of geographical barriers at a range of spatial scales has shaped the

6 community composition of aquatic assemblages over millenia (Rahel 2007). Studies of more recent biogeographical changes such as the introduction of alien species have demonstrated that community composition is impacted considerably by the introduction of species new to an area (Howe et al. 1997; Ivantsoff and Aarn 1999; Canonico et al . 2005; Roberts et al. 1995; Stuart and Jones 2006; Rayner et al . 2009), and recent phylogeographic studies suggest that the identification of species boundaries and times of divergence may provide valuable historical information that will contribute positively to ecological studies in freshwater systems (Musyl and Keenan 1992; Huey et al. 2006; Hammer et al. 2007; Faulks 2009).

With particular regard to arid-zone riverine systems, biogeographic research has the potential to provide data associated with historical migration pathways and speciation that will greatly assist research efforts aimed at elucidating the community composition of extremely remote populations and their relationships to populations in areas of greater connectivity. Arid-zone river systems with highly variable flow regimes, unpredictable patterns of connectivity and few alien species or other human- induced perturbations therefore offer unique opportunities to study localised colonisation and extinction events of aquatic biota. Such systems exist in central Australia.

Studying the reproductive behaviour of the biota of freshwater systems in remote areas is comparatively difficult because distance, cost and local weather events are likely to impact upon the frequency and success of data collection activities. Despite these constraints, data collected from isolated river systems may present several advantages when considering the recruitment of aquatic biota. In the Australian arid zone, where fish may frequently complete their entire life-cycles within a single, isolated waterhole during prolonged dry periods, and where flooding is rare and obvious, regular sampling, length measurement and cohort analysis (Puckridge et al. 2000; Nunn et al. 2002; Allen et al. 2005) have the potential to yield breeding and recruitment information that cannot be as easily obtained in flowing or connected systems. Consequently, traditional approaches aimed at determining the reproductive behaviour of fish, such as experimental breeding and rearing studies under artificial conditions (Lake 1967; Llewellyn 1973), or destructive techniques such as otolith examination (Pritchard 2004; Brown and Wooden 2007) and the calculation of the

7 gonadosomatic index (GSI: Pusey et al. 2004), can be replaced to a degree by the observation of cohorts and changes in population and community structure through time. This is especially advantageous in isolated arid-zone systems where destructive sampling of specimens is undesirable, on-site analysis is impossible and transportation of preserved samples for laboratory analysis is problematic due to the biomass of material involved.

Despite the recent increase in studies aimed at examining the biota and ecological processes of the Australian arid zone (Puckridge et al . 1998; Puckridge et al . 2000; Bunn et al . 2003; Costelloe et al . 2004; Arthington et al . 2005; Balcombe and Arthington 2009; Sheldon et al . in press), the spatial extent of field-based studies in far western Queensland has been extremely limited. Within the Queensland Lake Eyre Basin, research-based sampling of freshwater fish has only occurred at a limited number of sites in the Cooper and Diamantina catchments (Costelloe et al. 2004; Arthington et al. 2005; Balcombe and Arthington 2009). A significant part of the present study is therefore devoted to increasing basic biogeographic knowledge of the rivers of far western Queensland and the search for new information on species distributions in desert environments. By including poorly-known river systems such as the Bulloo, Barcoo, Georgina and Mulligan catchments in this study it is possible to directly address some of these knowledge gaps.

1.3 The current study

The current study is focused on fish communities of the Queensland Lake Eyre and Bulloo-Bancannia basins in central Australia and aims to supplement preceding work that has been undertaken primarily in the Cooper catchment (Puckridge 1999; Arthington et al. 2005; Balcombe et al. 2007; Balcombe and Arthington 2009) and in the Diamantina and more remote Neales catchment (Costelloe et al. 2004).

The thesis and component studies are underlain by the following working model regarding fish ecology in the Australian arid zone (Figure 1.1).

8 During extended dry periods, all fish species are capable of completing their life cycles in physically separated waterholes (Figure 1.1: Dry). As dry conditions intensify, the populations in shallower waterholes may become extirpated (Figure 1.1: Late dry). During dry periods, processes and responses such as recruitment and extirpation occur in isolation at the smallest scale (waterhole: Figure 1.1). When small flows occur, fish may take advantage of temporal migration pathways as either larvae, juveniles or adults, or may continue to complete their life cycles within the general area of the original waterhole (Figure 1.1: Connection flow). During these flow- connection periods, various processes and responses, such as recruitment and colonisation, occur at a larger spatial scale (reach: Figure 1.1), as migration pathways are open for comparatively short time periods and transfer of individuals to areas within a connected reach is possible. When floods occur, migration pathways may develop in multiple directions, including within the confines of pre-existing channels and across previously dry floodplains. During floods, fish can migrate long distances and colonise new and previously isolated areas as either adults or juveniles, or can remain in the general area of the original waterhole and continue to complete their life-cycles in situ (Figure 1.1: Flood.). Depending upon the size of the flood, processes and responses such as recruitment and migration potentially operate at large to very large spatial scales, and can occur throughout entire catchments if major flooding links all previously isolated waterholes. Although late dry (or prolonged dry) periods always follow dry periods, connection flows, and to a lesser extent large floods, occur unpredictably. Thus, although a temperate or tropical river may be expected to follow the dry/late dry /connection flow /flood sequence, these hydrological variations may occur out of sequence in arid-zone rivers (Figure 1.1).

9 Waterhole andprocesses Waterhole scale responses

Dry

Flood Waterholes disconnected Life-cycles completed Migration across floodplain within-waterhole and along migration pathways Catchment/sub-catchment scale Catchment/sub-catchment responses and processes

Late Dry Connection flow Waterholes disconnected Reach scaleprocesses Reach andresponses Life-cycles completed Migration of adults/juveniles within-waterhole along migration pathways Local extirpations

Figure 1.1 A working model describing fish ecology in the Australian arid-zone, depicting dry and late dry season habitats on the right, and habitats connected by flow on the left. Grey shapes indicate wet waterholes and clear shapes indicate dry waterholes. Grey arrows indicate fish movement/migration. Black arrows represent possible hydrologic sequences in rivers, with dashed arrows representing the potential for sequence variation in arid-zone rivers.

10 1.4 Aims of the current study

The current study considers fish species and communities across extended spatial and temporal scales within the Queensland Lake Eyre and Bulloo-Bancannia basins to investigate the following: a) The distribution of fish species and populations with reference, where appropriate, to biogeography and hydrology. b) The recruitment patterns of individual fish species, particularly in relation to the presence or absence of flow, and to test the hypothesis that arid-zone fish species are opportunistic and able to breed independently of flow events during prolonged dry periods. c) The migration, colonisation and extirpation/extinction patterns of fish species within four catchments of the study area, and to test the hypothesis that arid zone fish species will colonise recently-wetted habitats following inundation. d) The applicability of existing conceptual models for describing ecological processes in arid zone aquatic environments, consideration of alternative models that may be relevant to the specific area under study, and implications of the study for management of the rivers of far western Queensland.

11 1.5 Structure of the thesis

The study area includes the Mulligan, Georgina, Diamantina, Thomson, Barcoo, Cooper, Kyabra and Bulloo catchments. General introductions to the study area and fish species present are given in Chapters 2 and 3, respectively. Fish family names, scientific names, common names and species’ authors are given in Chapter 3, and thereafter common names are used when referring to species. Chapter 4 provides a detailed description of the study sites and hydrological changes through time. Chapter 5 presents data relating to fish species distributions and Chapter 6 investigates recruitment patterns. Chapter 7 considers fish colonisation in four of the eight catchments. In Chapter 8, the concluding chapter, a general summary of the results is presented, followed by a synthesis of the disparate studies and results into an appropriate conceptual model describing patterns of fish recruitment and movement in Australian arid-zone rivers. Last, research recommendations and the possible management implications of this work in far western Queensland are presented.

12 2. Study Area

2.1 Location

The study was conducted in catchments of the Queensland Lake Eyre Basin and also in the Bulloo-Bancannia Basin, which is situated immediately to the east of the Lake Eyre Basin in Queensland.

The Lake Eyre Basin covers 1.2 million square kilometres and occupies parts of Queensland, the Northern Territory, and South Australia (Figure 2.1). The Bulloo-Bancannia Basin is far smaller, covering only 100 000 square kilometres and an elongated area of Queensland and New South Wales (McMahon et al 2005; Midgley et al . 1991) (Figure 2.1). The endorheic rivers and waterbodies within both the Lake Eyre and Bulloo-Bancannia Basins drain to the south-west and consequently into the desert rather than the sea. In particularly wet seasons, run-off and drainage patterns can result in sporadic filling events in Lake Eyre, however in most years this does not occur (Knighton and Nansen 1994). Running as they do through semi-arid and arid country, rivers within the Lake Eyre and Bulloo-Bancannia basins are substantially different from coastal catchments and rivers that have their headwaters in high rainfall areas. Rather than persisting as discrete and constantly wet channels, these rivers are better described as sporadic channels and waterholes situated within a variable floodplain; where a waterhole exists one year, it may not the next. The waterholes and rivers of far western Queensland occupy a unique and iconic niche in Australian folklore and exploration (Madigan 1946; Murgatroyd 2002; Silcock 2009), and this iconic status is largely due to their crucial role in helping to establish the pastoral industry in an extremely dry climate (Durack 1959; Bowen 1987). Well-known pastoral empires such as those established by the Duracks and Costellos in the 1860s and later Sir all included large tracts of land in the Queensland Lake Eyre and Bulloo-Bancannia basins (Durack 1959; Bowen 1987).

13

Figure 2.1 Australian drainage divisions. Note the Lake Eyre and Bulloo-Bancannia divisions in the centre-right of the diagram (and continent). Source: National Land and Water Resources Audit, Natural Heritage Trust.

From east to west, the catchments included in the study are the Bulloo, Cooper, Diamantina, Georgina and Mulligan. The Cooper Creek catchment was investigated at sub-catchment scale by surveying sites in the Barcoo and Thomson rivers, Kyabra Creek and Cooper Creek itself. Cooper Creek forms approximately 30 kilometres north-east of the township of Windorah at the confluence of the Thomson and the Barcoo rivers (see Chapter 4, Figure 4.1 for site map).

In general, the rivers studied have their headwaters in northern Queensland and are therefore predominantly fed by run-off from summer rainfall. Exceptions include the Bulloo and Barcoo rivers, as neither have headwaters as far north as the Thomson, Diamantina and Georgina catchments.

14 2.2 Climate

The rivers of the Queensland Lake Eyre and Bulloo-Bancannia basins occur in the Australian semi-arid to arid zone. Rainfall decreases along north-south and east-west gradients within the study area. Consequently, districts in the north and east of the study area, such as the Thomson and Bulloo rivers, experience some of the highest mean annual rainfall (in excess of 600mm per year) whereas an area in the extreme west, such as the , experiences the lowest mean annual rainfall (approximately 120mm per year). In general, rain is less likely to occur from May to October than from November to April across the study area, with all weather stations frequently recording nil monthly rainfall readings in these months (Bureau of Meteorology 2009). In contrast, summer rainfall associated with the northern monsoon is more likely, but is often highly variable and occasionally extremely heavy. As an example of this variability, rainfall on the 21st of January 2007 across the catchment ranged from zero at Blackall and Isisford in the Barcoo catchment to 169mm at Bedourie in the Georgina catchment (Bureau of Meteorology 2009).

Temperature in the Queensland Lake Eyre and Bulloo-Bancannia basins is far more predictable than rainfall (Figure 2.2), but also exhibits a slight increase along an east- west gradient as the landscape changes from semi-arid to arid and, ultimately, to the sand ridges of the Simpson Desert. Average monthly maximum and minimum temperatures at Blackall, in the eastern part of the study area, range from 30.18 to 15.49 °C, whereas similar figures for Boulia, in the western section of the study area, are 31.63 to 16.88 °C (Figure 2.2). Despite these subtle differences, the maximum daily temperatures on record (44.6 °C for Blackall and 48.3 °C for Boulia) provide examples of the increases in temperature that can occur from the eastern to the western areas of the Queensland Lake Eyre Basin. In general, the highest temperatures always occur in January and the lowest in July (Figure 2.2), with extreme minimum daily temperatures occasionally dropping below zero (-2.0 °C for Blackall; - 5.5 °C for Boulia). As expected in the arid and semi-arid zones, evaporation exceeds precipitation in the Lake Eyre and Bulloo-Bancannia basins in western Queensland. Monthly evaporation follows a similar trend to temperature,

15 ranging from about 60mm in the winter months to 200mm in December, January and February (all quoted figures: Bureau of Meteorology 2009).

40

35

30

25

20 Celsius ° 15

10

5

0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Figure 2.2 Mean monthly maximum (above) and minimum temperatures (below) from four weather stations located in the study area. Green = Boulia, Georgina catchment, blue = Winton, Diamantina catchment, red = Windorah, Cooper catchment, grey = Adavale, Bulloo catchment. Source: Bureau of Meterology.

2.3 Hydrology

Obtaining accurate discharge data for the rivers and streams of western Queensland is difficult for a number of reasons. Flows, when they occur, are often of a short duration, and may only occur in a certain section of a river with a total length of up to 1000 kilometres (Diamantina River) or 1500 kilometres (Cooper Creek). As there are a limited number of stream gauging stations spread throughout the Lake Eyre Basin (and comparatively few in Queensland), and as many of these stations have either been de-commissioned or operated only sporadically, many flows – especially smaller flows – remain unrecorded (Figure 2.3). This situation is best illustrated in the

16 Mulligan River in the far west of the Queensland Lake Eyre Basin, and in Kyabra Creek in the far east; no gauging station exists on either waterway.

Georgina River at Roxborough Downs 1967 - 2008

350000

300000

250000

200000 ML/day 150000

100000

50000 No data: 29/09/88 – 05/06/05

0 1974 1977 2007

Figure 2.3 Hydrograph of the Georgina River at Roxborough Downs from 1967 – 2008. Note the prolonged period (1988 – 2005) when the gauging station was de- commissioned, and the variability of flows in years when data is available.

Despite the lack of hydrological data available for the rivers of far western Queensland, records from operational gauging stations permit some general patterns to be inferred. Mean annual run-off increases from west to east, with the Georgina catchment far drier than the Diamantina, the Diamantina far drier than the Cooper and the Cooper far drier than the Bulloo (Figure 2.4a). Nevertheless, mean monthly flow data indicates that the late summer period from January to March is the most likely time when flows will occur in all catchments, with the exception of the Bulloo, where monthly flows commonly occur over an extended timeframe from November to May (Figure 2.4b).

17 a)

35

30

25

20

15

10

Megalitres 1000 x per year (ML) 5

0 Georgina Diamantina Cooper Bulloo

b) 7

6 Georgina River Diamantina River Cooper Creek 5 Bulloo River

4

3

2 Megalitres 1000 per x day (ML) 1

0 Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec

Figure 2.4. Mean annual run-off (mean annual flow divided by catchment area) ± co- efficient of variation in annual flows in the Georgina, Diamantina, Cooper and Bulloo catchments (a: above), and mean monthly flow (mean daily flow for each month averaged across years when data is available) for the same catchments (b: below). Data derived from the following gauging stations: Roxborough Downs (Georgina), (Diamantina), Currareva (Cooper) and Autumnvale (Bulloo). Data courtesy of Mark Kennard, Australian Rivers Institute, Griffith University.

18 The geography of the Queensland Lake Eyre Basin, and the complex series of channels that comprise the river valleys and floodplains, combine to create variable discharge patterns depending on the amount of rainfall and where this rainfall occurs. Despite the fact that rainfall and run-off are generally highest in the north-eastern areas, discharge – especially from widespread flooding – is often likely to be higher further downstream as rivers, channels and creeks converge. Using the annual figures available for the Cooper Creek catchment, maximum figures for the Thomson River at Longreach range up to 10 800 000 ML/year (all figures quoted from McMahon et al. 2005). For the Alice River at Barcaldine and the Barcoo River at Blackall – both stations that can be considered higher upstream in their respective sub-catchments than Longreach in the Thomson – the figures are far lower at 455 000 ML/year and 387 000 ML/year respectively. Below the confluence of the Thomson and Barcoo rivers, the maximum figure recorded for Cooper Creek at Currareva near Windorah is 235 000 000 ML/year, thus demonstrating that, as floodwaters converge, discharge at downstream sites is often likely to increase. In contrast, minimum flows at Longreach (19 100 ML/year) and Blackall (7 100 ML/year) indicate that, for the years from which data is available, small flows occur on an annual basis in upstream areas, whereas at Currareva, (0 ML/year), flow is dependent on a sufficient level of discharge from either the Thomson or Barcoo Rivers.

Across the Queensland Lake Eyre Basin, mean annual streamflows indicate that areas in the Cooper and Diamantina catchments, such as Currareva (3 150 000 ML/year) and Diamantina Lakes (2 870 000 ML/year) have a higher discharge than those in the Georgina catchment, like Roxborough Downs (1 090 000 ML/year), however there are two important caveats. Firstly, the available streamflow information is patchy, and the number of years for which records exist range from below 8 (at Glengyle on Eyre Creek and at Retreat on the Barcoo River) up to 42 for Cooper Creek at Currareva. Secondly, and as discussed above, the channel/floodplain nature of the topography of western Queensland means that discharge in different areas of a catchment will vary depending on the amount of country floodwaters or flows can ‘colonise’ as they move downstream and due to extra inflows from tributaries and channels. As examples, the mean annual discharge of the Diamantina is higher at Diamantina Lakes (2 870 000 ML/year), a site upstream of Birdsville where the discharge is 1 300 000 ML/year. In contrast, in the Cooper catchment, mean annual discharge at Stonehenge (2 700 000

19 ML/year), a site downstream of Longreach, is higher than at Longreach (1 280 000 ML/year), presumably due to run-off from a tributary -Vergemont Creek - merging with the Thomson River between the two gauging stations. Further downstream, mean annual discharge of the Cooper is far higher at Currareva (3 150 000 ML/year) than it is much further downstream at Innamincka (1 430 000 ML/year), again due to the complex system of channels and floodplains that effectively ‘steal’ the water as it flows downstream (all figures quoted from McMahon et al. 2005).

Considering monthly and daily streamflow records gives a far more accurate picture of the variability of flows in far western Queensland. At all gauging stations, monthly and daily minimum flows are zero, indicating again that, for the majority of the time, the rivers exist as a series of disconnected waterholes. Maximum monthly discharge, like annual discharge, follows a general pattern of increasing from headwaters (such as Blackall at 325 000 ML/month) to middle sections (Currarreva at 15 900 000 ML/month), and then subsequent decrease in lower areas (Innamincka at 9 190 000 ML/month). Daily discharge, though influenced to a greater degree by local rainfall and run-off events, nevertheless exhibits the same pattern, since maximum daily discharges are likely to have occurred during a flood month and in a flood year. Daily discharge data demonstrates that westerly catchments, such as the Georgina and Diamantina, are likely to experience more prolonged yet more sporadic flooding events than the Cooper and Bulloo (Figure 2.5), despite the fact that the length of dry periods increases along an east-west gradient (Figure 2.5). Existing flow records generally indicate that the rivers in the study area experience decreasing flow predictability from east to west (Figure 2.5).

20 350

300

250 Mean number of zero flow days 200 Days 150

100 Mean duration of high flow spells above 25 th 50 percentile on flow duration curve.

0 Georgina Diamantina Cooper Bulloo

Figure 2.5. Mean number of zero flow days (top line) and mean duration of high flows (bottom line) from west to east in the Georgina, Diamantina, Cooper and Bulloo catchments. Data derived from the following gauging stations: Roxborough Downs (Georgina), Birdsville (Diamantina), Currareva (Cooper) and Autumnvale (Bulloo). Data courtesy of Mark Kennard, Australian Rivers Institute, Griffith University.

2.4 Topography and geomorphology

Amongst the oldest geological sequences in the Queensland Lake Eyre Basin are the eroded mesas and gibber plains of the Cretaceous Winton Formation (Wopfner 1963). These structures are particularly noticeable from Diamantina Lakes north to Winton in the Diamantina catchment. Alluvium and dune sands accumulated between the higher strata in the Quaternary (Senior and Mabbutt 1979), and the anastomosing and braided channels of the Cooper, Diamantina and Georgina rivers meander through the lowest sections of their respective valleys (Knighton and Nansen 1994). High rainfall conditions were replaced with increasing aridity from approximately 780 million years ago in central Australia, and consequently there was a substantial reduction in

21 flow regimes in the area (Bowler 1990; Nansen et al. 2008). Aridity became pronounced from approximately 0.9 – 1.6 million years ago (Chen and Barton 1991) and extreme from 500 000 years ago (Gardner et al. 1997).

Only 30% of the Lake Eyre Basin is >250m above sea level (ASL) and the lowest point in Lake Eyre itself is 15 metres below sea level (Maroulis et al. 2007). The headwaters of the rivers studied rarely reach 300 metres ASL, and throughout the study area these rivers exhibit only slight decreases in elevation (Maroulis et al. 2007). As examples, Blackall on the Barcoo River is located at approximately 270m ASL, and Longreach on the Thomson River is at 180m ASL. Downstream of the junction of the Barcoo and Thomson rivers, Windorah is at 125m ASL, however the difference in elevation is mitigated by the long distances the rivers cover, as even in a straight line the distances between Windorah and Blackall and Windorah and Longreach are approximately 250 and 200 kilometres respectively. Consequently, the topography of both the Queensland Lake Eyre and Bulloo-Bancannia basins is predominantly flat. Sand dunes are present in the landscape from as far east as Kyabra Creek, and become longer and closer together towards the Birdsville area and the Simpson Desert, where they run in a parallel north-north-west to south-south-east orientation (Shephard 1992).

2.5 Vegetation

Aquatic macrophytes are rarely present in the waterholes of the Lake Eyre and Bulloo-Bancannia basins, predominantly due to the turbidity of the waterways and the negative effect this has on plant photosynthesis. As a consequence, aquatic plants (such as Nardoo spp.) are generally confined to waterholes where turbidity is low, such as certain spring-fed areas in the Georgina catchment. Upland plant communities generally grade from mulga, Acacia aneura ,-dominated areas in the east (such as the Bulloo) to gidgee, Acacia cambagei and Acacia georginae , communities in the west. Riparian communities surrounding the deeper and more permanent waterholes are characterised by a common suite of species including coolibah, Eucalyptus coolabah , river red gum, Eucalyptus camaldulensis, belalie, Acacia stenophylla , creek wilga, Eremophila bignoniflora , bauhinia, Lysiphyllum gilvum , tea trees, Melaleuca

22 trichostachya , and Melaleuca leucadendron (in the Georgina River only), lignum, Muehlenbeckia cunninghamii , rat’s tail couch, Sporobolus mitchellii , Warrego grass, Paspalidium jubiflorum and sedges, Cyperus spp. (Jenny Silcock, Queensland Department of Environment and Resource Management, personal communication). Introduced plants, such as Parkinsonia aculeate , prickly acacia, Acacia nilotica and Noogura burr, Xanthium pungens , are frequently common in riparian zones of the greater Cooper, Diamantina and Georgina catchments.

23

3. Fish of the Lake Eyre and Bulloo-Bancannia Basins

3.1 Existing freshwater ecological studies

Both the Lake Eyre and Bulloo-Bancannia basins remain comparatively poorly studied when compared with the more heavily-populated coastal drainages and the Murray-Darling, a large inland basin in south-eastern Australia (Allen et al. 2002; Pusey et al . 2004). The first surveys of fish in the Lake Eyre Basin were conducted in South Australia in the late 1970s (Glover and Sim 1978; Glover 1979; Glover 1982). Fish surveys at sub-Basin scale in Queensland did not occur until 1995 (Long and Humphrey) and 2001 (Bailey and Long), and a general text regarding fish in the Lake Eyre Basin was published by the in 2000 (Wager and Unmack). More recently, multi-disciplinary surveys and studies have occurred primarily in the Queensland Cooper Creek catchment (Dryland Refugium Project: Arthington et al. 2005) and across a wider spatial scale incorporating sites in both Queensland and South Australia (ARIDFLO: Costelloe et al. 2004). The existence of these research programs has led to an increase in published literature pertaining to aquatic systems and processes in the Lake Eyre Basin including studies relating to primary production (Bunn et al. 2003), macroinvertebrates (Sheldon and Thoms 2006), and the relationships between flow and fish (Puckridge et al. 1998; Puckridge 1999; Arthington et al. 2005; Balcombe et al. 2007; Balcombe and Arthington 2009). Additionally, genetic studies have frequently targeted or incorporated species from the Lake Eyre Basin and include work on invertebrates (Hughes and Hillyer 2003; Hughes et al. 2004) and fish (Huey et al. 2006; Pritchard 2004; Hammer et al . 2007). Scientific literature directed towards fish ecology in the Bulloo-Bancannia Basin is limited to a study by Midgley et al. (1991). As a consequence, the Bulloo-Bancannia system is frequently grouped with the Lake Eyre Basin and information is appropriated regarding species’ life histories and environmental requirements (Wager and Unmack 2000). It should be emphasised that existing knowledge of Australian arid-zone fish ecology has been derived from a comparatively small geographical area. In Queensland, the Cooper catchment below Windorah remains unstudied, as do

24 the smaller rivers in both the upper Thomson and Barcoo catchments. Studies in the Diamantina are similarly limited to the upper-mid reaches of the system in the vicinity of Diamantina Lakes (Costelloe et al. 2004). The current study is the first to survey the Mulligan River in any capacity and the first to undertake temporal sampling in the Georgina and Bulloo systems.

3.2 Existing knowledge and knowledge gaps in relation to fish in far western Queensland

Although some fish species in rivers of the Lake Eyre Basin experience population booms following flood events (Puckridge et al . 2000, Arthington et al . 2005; Balcombe and Arthington 2009), there remains a lack of data relating to finer-scale distribution of species, recruitment patterns of most species and the roles played by recruitment, migration, colonisation and mortality in structuring fish assemblages through time. In the following sections the native fish species known to occur in the Queensland Lake Eyre Basin are discussed by family with reference to existing knowledge and current knowledge gaps.

3.2.1 Clupeidae

The Clupeidae, or herrings, are a widespread family with a global distribution and a comparatively ancient lineage traceable to the Cretaceous (Allen et al . 2002). In the Lake Eyre and Bulloo-Bancannia basins the only Clupeid fish is the bony bream, Nematolosa erebi (Figure 3.1a; Table 3.1). Bony bream are one of the most widely distributed freshwater fish in Australia and, as detritivores and algivores, their presence is thought to play an important role in the functioning of aquatic ecosystems (Pusey et al. 2004). Bony bream are common throughout the central Australian region, where they have been observed to suffer large population declines when waterholes experience extended dry periods in Cooper Creek (Arthington et al. 2005). Bony bream appear to spawn across a range of habitat conditions in the Murray- Darling Basin, including exposed environments such as Lake Alexandrina, South

25 Australia (Puckridge and Walker 1990) and receding still-water environments such as Lake Cargelligo, New South Wales (Kerezsy 2005). It seems likely that bony bream may utilise a ‘hit or miss’ reproductive strategy, given their wide distribution and ability to withstand, and recover from, population decline. In Cooper Creek, constant recruitment (including population increases following dry periods) has been demonstrated for this species in the Windorah area of south-western Queensland (Balcombe and Arthington 2009), and the existence of a reproductive strategy that is non-reliant on flow (but nevertheless advantaged by elevated or serial flows) is supported by evidence from South Australia (Puckridge et al. 2000). Bony bream appear to be amongst the first species to colonise newly-inundated areas (Puckridge et al . 2000), and have been recorded in recently-inundated floodplain habitats as larvae, juveniles and adults in the Windorah area of Cooper Creek (Balcombe et al. 2007). Although bony bream appear to be a highly vagile and opportunistic species in Cooper Creek, the behaviour of this species in the more westerly (and hence more arid) catchments – particularly the Georgina – is unknown. Given that bony bream populations are likely to play a major role in underpinning food webs in areas where they occur, the current study aims to monitor bony bream in these arid catchments and examine the recruitment and colonisation strategies of this species in Australian desert systems.

3.2.2 Plotosidae

Eel-tailed Plotosid catfish are represented in the Lake Eyre Basin by four species, including the widespread and common Hyrtl’s tandan, Neosiluris hyrtlii (Figure 3.1c; Table 3.1), the locally abundant silver tandan, Porochilus argenteus (Figure 3.1d; Table 3.1), the comparatively rare and endemic Cooper Creek catfish, Neosiluroides cooperensis (Figure 3.1b; Table 3.1), and the Dalhousie catfish, Neosiluris gloveri , which is endemic to the Dalhousie Springs complex in South Australia. The existence of a fifth species inhabiting the Bulloo-Bancannia drainage has also been suggested (Wager and Unmack 2000; Table 3.1). The of freshwater catfish in Australia remains uncertain, although the number of potential species suggested in earlier literature (up to 18; Merrick and Schmida 1984) is probably over-estimated. As an example of this uncertainty, recent genetic work suggests that silver tandan may be

26 more closely related to Hyrtl’s tandan than previous research has indicated and that this species may be better placed in the genus Neosiluris (Chris Hardy, CSIRO, personal communication).

Hyrtl’s tandan and silver tandan have a widespread distribution throughout the Lake Eyre and Bulloo-Bancannia basins whereas Cooper Creek catfish are restricted to waterways within the Cooper Creek catchment (Allen et al . 2002). Although both silver tandan and Hyrtl’s tandan form aggregations or shoals, sampling of Cooper Creek catfish to-date has indicated that this species does not exhibit such behaviour (Arthington et al . 2005). It seems likely that Cooper Creek catfish is a more solitary animal, perhaps more similar in terms of life history to Tandanus tandanus from the Murray-Darling Basin than to either Hyrtl’s tandan or silver tandan. Genetically, Cooper Creek catfish are descended from a far more ancient lineage originating from marine and Tandanus ancestors (Chris Hardy, CSIRO, personal communication), which may account for these observed behavioural differences.

Both Hyrtl’s tandan and silver tandan have been sampled in large numbers in Cooper Creek and have been observed to suffer large population declines when waterholes experience extended dry periods (Arthington et al. 2005). In contrast, Cooper Creek catfish have always been sampled in small numbers if at all (Costelloe et al. 2004; Arthington et al. 2005). This evidence, though limited, suggests that different reproductive strategies may be employed by these species, despite the fact that the spawning and reproductive behaviour of all present catfish species have not been studied in detail. Available evidence from waterholes in the Windorah area of south- western Queensland suggests that both Hyrtl’s tandan and silver tandan are most likely to spawn during the warmer summer months (Balcombe and Arthington 2009). Nevertheless, slight differences appear to exist between these species, with silver tandan possessing a comparatively flexible recruitment strategy closely aligned with season, and Hyrtl’s tandan possibly requiring flows or floods in order to reproduce (Balcombe and Arthington 2009). In the case of Hyrtl’s tandan, similar preferential reproduction associated with elevated summer flows has been demonstrated in the morphologically similar in the northern Murray-Darling Basin (Balcombe et al . 2006). Although the biology and ecology of Cooper Creek catfish remains relatively unknown, a laboratory spawning has been reported (Unmack

27 1996). The Cooper Creek catfish appears to produce relatively few eggs (approximately 1000) of a very large size (3-4mm: Unmack 1996). Given their low fecundity, it seems unlikely that Cooper Creek catfish would spawn in unfavourable conditions and more likely that Cooper Creek catfish spawn far more infrequently and selectively, perhaps in response to currently unknown triggers. Cooper Creek catfish grow to a large size (up to 600mm; Allen et al . 2002), suggesting that this species may be long-lived, thus allowing such a strategy to ensure successful recruitment albeit at infrequent intervals. During sampling in temporary floodplain habitats in Cooper Creek in the vicinity of Windorah in south-western Queensland, no Cooper Creek catfish were sampled, however larval, juvenile and adult silver tandan and larval and adult Hyrtl’s tandan were present on the floodplain (Balcombe et al. 2007).

Detailed studies of the spawning and reproductive behaviour of Plotosid catfish in Australia is currently limited to the work completed on Tandanus tandanus in the Murray-Darling Basin (Davis 1977a, 1977b and 1977c). Orr and Milward (1984) studied the localised breeding migration and subsequent spawning of Hyrtl’s tandan and a related species, Neosiluris ater , in the near Townsville, however the larvae and juveniles of each species were difficult to identify. Unlike Tandanus tandanus , the most common Plotosid in the Murray-Darling Basin, the available evidence suggests that Hyrtl’s tandan does not build a ‘nest’, but rather scatters fertilised eggs over the substrate (Orr and Milward 1984). Given the demonstrated shoaling behaviour of silver tandan, it seems likely that this species may also be an egg-scatterer.

The large fluctuations observed in populations of both silver tandan and Hyrtl’s tandan suggest that periods of ‘boom’ and ‘bust’ in both of these species may be linked with flow variability, however their recruitment and movement capabilities require investigation at wider spatial scales before more accurate predictions can be made. Last, the possibility that other species of Plotosid catfish exist in either the Bulloo-Bancannia or Lake Eyre basins has been suggested (Wager and Unmack 2000), and gaining relevant field data can only be achieved by sampling across both areas.

28 3.2.3 Retropinnidae

Australian smelt, Retropinna semoni , are generally found throughout south-eastern Australia, and the Cooper Creek catchment in the Lake Eyre Basin represents their most north-westerly distribution (Figure 3.1e; Table 3.1). Recent phylogeographical evidence indicates that Australian smelt in the Lake Eyre Basin may be a separate species (Hammer et al. 2007), suggesting colonisation of Cooper Creek is more likely to have occurred from the south rather than the east. The apparent absence of Australian smelt from the Bulloo-Bancannia drainage, which is situated immediately to the east of Cooper Creek, and their absence from both the Diamantina and Georgina catchments, lends further evidence to this notion (Jeff Johnson, Queensland Museum, personal communication), and also indicates that colonisation of the western Lake Eyre Basin drainages (through Lake Eyre and Goyder’s Lagoon) has either not occurred or was unsuccessful. Australian smelt is a pelagic, schooling species that reaches sexual maturity within the first nine months (Pusey et al. 2004). Sampling completed in waterholes in South Australia indicates that the abundance of larval Australian smelt increases following flooding in Cooper Creek (Puckridge et al. 2000), and both larval and juvenile Australian smelt have been detected in temporary floodplain habitats in the vicinity of Windorah in south-western Queensland (Balcombe et al. 2007). Although the data was not collected in central Australia, Australian smelt spawning is generally linked to water temperature (15°C: Allen et al. 2002) and has been demonstrated to be concentrated in late winter and spring (Milton and Arthington 1985). Spawning activity has also been linked to particular stages in the lunar cycle (Ebner 2004). Given that studies suggest that low-flow conditions are conducive to successful recruitment of Australian smelt in the coastal rivers of south- eastern Queensland and in the Murray-Darling Basin (Pusey et al. 2004), it seems likely that this species may exhibit a similar seasonal reproductive strategy in Cooper Creek due to the necessity of population maintenance in an unpredictable environment.

29 3.2.4 Melanotaeniidae

The family Melanotaeniidae is represented in the Lake Eyre and Bulloo-Bancannia basins by a single sub-species, Melanotaenia splendida tatei , known as the desert rainbowfish (Figure 3.1f; Table 3.1). Taxonomic revision and the ‘discovery’ of localised species, sub-species and colour variants across northern Australia and New Guinea have characterised this family of small and colourful fishes (Allen et al. 2002), and their popularity as hardy aquarium species has led to captive breeding of many species (Leggett and Merrick 1987). Despite this, the general ecology of desert rainbowfish in the wild is poorly known. Results from Cooper Creek in South Australia suggest that the desert rainbowfish demonstrates increased larval abundance following flow events (Puckridge et al . 2000), and this is supported by a study of the closely-related Eastern rainbowfish ( M. s. splendida ) which indicates that although this species is reproductively active throughout the year in the Townsville area, spawning activity increases during the wet season (Beumer 1979). As M. s. splendida (and, presumably, M. s. tatei) reach sexual maturity within their first year and rarely live for more than two years (Pusey et. al . 2004), it seems reasonable to assume that in the unpredictable hydrological regimes experienced in central Australia that desert rainbowfish would exhibit a flexible reproductive strategy that would not be reliant on flows or floods. Courtship and subsequent spawning of this species in the wild have been reported from Lake Dunn in the Lake Eyre Basin as occurring in late autumn (Wager and Unmack 2000), an observation that appears to lend evidence to the assertion that desert rainbowfish may be highly opportunistic spawners. Desert rainbowfish have a wide distributional range in inland Australia encompassing the entire Lake Eyre Basin, the Bulloo-Bancannia Basin and the north-western rivers of the Murray-Darling Basin (Wager and Unmack 2000; Lintermans 2007). The range of desert rainbowfish in the waterways of central Australia, combined with the demonstrated presence of juvenile and adult desert rainbowfish in temporary floodwaters in the vicinity of Windorah in south-western Queensland (Balcombe et al. 2007), indicate that this species may exhibit well-developed abilities to persist in drying waterholes and to re-colonise formerly dry areas upon inundation.

30 3.2.5 Pseudomugilidae

A single species of Pseudomugilid fish - the red-finned blue-eye, Scaturiginichthys vermeilipinnis - has an extremely restricted distribution limited to the Edgbaston springs complex near Aramac (Allen et al. 2002; Table 3.1) in the Queensland Lake Eyre Basin. The presence of a Pseudomugilid species in the area is curious as all other members of this family are coastal species. Temporal sampling of the Edgbaston spring complex since 1989 indicates that red-finned blue-eye are declining and that this may be due to the presence of alien gambusia, Gambusia holbrooki (Fairfax et al. 2007). The red-finned blue-eye is listed as endangered under both state and national legislation (Nature Conservation Act 1992; Environment Protection and Biodiversity Conservation Act 1999) and as critically enadangered by the International Union for the Conservation of Nature (IUCN 2010). In March 2009, a fish survey of Edgbaston springs found only four populations of red-finned blue-eye and a maximum total population of approximately 3000 individuals (Kerezsy 2009). This species can therefore be considered one of the rarest and smallest freshwater fish in Queensland (and Australia), as adults are fully grown at 30mm (Allen et al . 2002). Red-finned blue-eye are capable of withstanding diurnal temperature fluctuations in excess of 20° celsius (Fairfax et al. 2007) and inhabit springs that are generally <5cm deep (personal observation). Although the isolation of spring complexes such as Edgbaston have necessitated the evolution of specialised adaptations such as small body size and tolerance of extreme temperatures, it seems equally likely that some of these traits – particularly small body size – may also explain the inability of this species to successfully colonise more connected areas of the Lake Eyre Basin such as the main channels of the Cooper, Diamantina and Georgina drainages.

3.2.6 Ambassidae

Members of the Ambassidae are small perch-like fishes usually referred to as glassfish due to their largely transparent bodies. In common with the Melanotaeniidae and Eleotridae families, the Ambassidae have been subject to on-going taxonomic revision and resulting changes and alterations to common and scientific names. Hence, within the Lake Eyre Basin, the species formerly (Puckridge 1999) referred to

31 as Mueller’s glassfish, Ambassis muelleri , is currently known as Northwest ambassis (Ambassis sp.) and is awaiting description as either a single or multiple species (Allen et al. 2002; Figure 3.1g). Ambassid fishes have not been recorded by either Midgley et al . (1991) or Wager and Unmack (2000) as occurring in the Bulloo-Bancannia Basin, however the Northwest ambassis is listed as occurring in the Bulloo River in a more generic and recent text (Allen et al. 2002). Confirmation of the presence of Ambassid fishes from the Bulloo-Bancannia Basin is therefore important in the first instance (Table 3.1). If Ambassid fishes definitely occur in the Bulloo-Bancannia a further priority will exist in relation to the determination of the possible species, given that current records suggest that Ambassis agassizii occurs at least as far west as the north-western edge of the Murray-Darling Basin (Balcombe et al . 2006) and Ambassis sp. occurs westwards from the Cooper system in the eastern Lake Eyre Basin (Arthington et al . 2005). 1 Specific information relating to the recruitment and life history of glassfish in central Australia is lacking, however data relating to the closely-related Ambassis agassizii , from eastern coastal drainages and the Murray- Darling Basin suggest a young age at sexual maturity (<12 months), short life-span (3 – 4 years) and a peak in spawning activity in spring and summer (Milton and Arthington 1985). It seems likely that Ambassis sp. in the Lake Eyre and Bulloo- Bancannia basins would possess adaptable reproductive strategies not reliant on flow or flood events in order to maintain their populations, however very little is known of the life history of this undescribed species in far western Queensland. Sampling events conducted in Queensland reaches of Cooper Creek reveal that Ambassis sp. occur sporadically but occasionally in very large numbers following flow events, but also that these populations suffer massive declines as waterholes dry and recede (Arthington et al. 2005). Determining the degree to which floods or elevated flows influence glassfish abundance, reproduction and dispersal would be beneficial, particularly as the related species A. agassizii is considered to be declining in the Murray-Darling where naturally-occurring floods and flows have generally been replaced or altered by river regulation (Lintermans 2007).

1 The term ‘glassfish’ is used throughout this thesis to refer to the undescribed Northwest Ambassis, (Ambassis sp.)

32 3.2.7 Percichthyidae

The yellowbelly or Lake Eyre golden perch, Macquaria sp. (Figure 1.3h; Table 3.1) occurs throughout central Australia, although it has also been suggested that the yellowbelly that inhabits the Bulloo-Bancannia Basin may be a sub-species (Wager and Unmack 2000). It should be noted that the potential number of species or sub- species of yellowbelly/golden perch that exist in Australian waterways is the subject of on-going conjecture and research (Musyl and Keenan 1992; Faulks 2009), and also that the various common names for this species (golden perch, yellowbelly, callop) can be similarly confusing. Golden perch (yellowbelly) from central Australia are - at the very least - extremely similar to the described golden perch, Maquaria ambigua , which is widespread throughout the Murray-Darling Basin. Studies aimed at determining the taxonomic status of the various golden perch populations (including a population in the coastal-draining Fitzroy catchment in ) have reached alternate conclusions regarding their evolutionary history, with Musyl and Keenan (1992) suggesting an inland origin (followed by radiation) and Faulks (2009) a coastal origin in the Fitzroy. It would appear that the most likely explanation for yellowbelly evolution and range expansion is the latter, with coastal populations crossing the Great Dividing Range during the Pleistocene and occupying the three inland drainages – the Murray-Darling, Bulloo-Bancannia and Lake Eyre – when wetter conditions provided more widespread hydrological connectivity. Faulks (2009) presents evidence to suggest that the resolution of the golden perch/yellowbelly speciation conundrum could be achieved by declaring the Fitzroy River golden perch a new species and labelling the Lake Eyre and Bulloo-Bancannia populations as an Evolutionary Significant Unit (ESU) derived from M. ambigua in the Murray-Darling. Faulks (2009) also found genetic separation between yellowbelly in the eastern Lake Eyre Basin rivers (the Barcoo, Thomson, Cooper and also the Bulloo), and those further west – an unsurprising result given the aridity of central Australia and the isolation this imposes on wholly aquatic biota.

Although studies of the Lake Eyre yellowbelly are currently comparatively rare (Pritchard 2004), the Murray-Darling species has been studied in more detail, and much of this information has been appropriated for all yellowbelly in the literature (Allen et al. 2002). Yellowbelly (all) are piscivorous ambush predators, and occupy

33 similar niches at the top of their respective food chains. In the case of fish from the Murray-Darling, this niche is shared with other large Percichthyid fishes such as , Maccullochella peelii peelii , trout cod, Maccullochella macquariensis, Macquarie perch, Maquaria australasica and, in some areas, alien species such as redfin, Perca fluviatilis , and trout, Salmo trutta and Onchorynchus mykiss . In contrast, the Lake Eyre yellowbelly is the dominant large piscivore in the rivers in which it lives. In the Murray-Darling, yellowbelly were harvested commercially until 2003 (Reid et al. 1997), remain a target species for recreational fishing and are emerging as an important aquaculture species (Bruce Sambell, Ausyfish, personal communication). With the exception of commercial fishing, the same attributes are common to yellowbelly in central Australia, and with regard to aquaculture, the Lake Eyre variant is considered to be superior in many respects to the Murray-Darling variety (Sambell, personal communication). Available evidence suggests that Murray- Darling yellowbelly may undertake long migrations associated with recruitment (Reynolds 1983) or dispersal (Mallen-Cooper et al . 1995). Historically, a strong linkage has been made between flood-flows and yellowbelly reproduction in the Murray-Darling Basin (Lake 1967, Mackay 1973; Cadwallader 1978 and 1979). More recently, Mallen-Cooper and Stuart (2003) demonstrated that within-channel flows (as opposed to floods) resulted in stronger year classes than flood flows for this species. Evidence from waterholes in the Windorah area in south-western Queensland suggests that yellowbelly recruit over an extended timeframe in Cooper Creek (Balcombe and Arthington 2009). Further support for this hypothesis is provided by an otolith study by Pritchard (2004) suggesting that although golden perch recruitment may be higher following flood or flow events, this species breeds constantly within the Lake Eyre Basin. Yellowbelly have a wide distribution in the Lake Eyre Basin (Wager and Unmack 2000), and are large fish with high energy requirements. Together with their demonstrated longevity (Pritchard 2004), these factors combine to indicate that, despite the unpredictability of the Lake Eyre and Bulloo-Bancannia aquatic environments, yellowbelly are consistently able to maintain their populations. These factors, when combined, indicate that this species may be able to recruit successfully in no-flow environments in far western Queensland. Although yellowbelly have been observed in temporary floodplain habitats as larvae, juveniles and adults (Balcombe et al. 2007), the migratory abilities of this species are unstudied

34 in central Australia. In general these fish are assumed to possess similar recruitment- related movement patterns to their more southerly-distributed cousins.

3.2.8 Terapontidae

Terapontid fishes are common across northern Australia, where they often exhibit localised endemism within a limited range of river systems (Allen et al. 2002). In southern Australia the family is currently represented only by the silver perch, Bidyanus bidyanus , in the Murray-Darling Basin. Fossil Terapontids are some of the oldest fish fossils known from Australia (Unmack 2001), suggesting this family has a long history in Australian waterways. Colloquially, Terapontid fishes are often referred to as ‘grunters’ due to the sound they make when removed from the water. The trophic guilds occupied by Terapontids are unusually varied, and they range from primarily carnivorous fishes, such as sooty grunter, Hephaestus fuliginosus , to more omnivorous species such as silver perch. Within the Lake Eyre Basin, Terapontid fishes include spangled perch, Leiopotherapon unicolor (Figure 3.1k; Table 3.1), Welch’s grunter, Bidyanus welchi (Figure 3.1j; Table 3.1 ), Barcoo grunter, Scortum barcoo (Figure 3.1 l: Table 3.1 ), and banded or barred grunter, Amniataba percoides (Figure 3.1i; Table 3.1). Translocations of Murray-Darling silver perch are also suggested to have occurred in both southern and northern areas of the Lake Eyre Basin by Wager and Unmack (2000). In the Bulloo-Bancannia, a similar suite of species has been recorded with the exception of banded grunter (Midgley et al. 1991).

Spangled perch, like bony bream and Hyrtl’s tandan, have an extremely wide distribution throughout Australian drainage basins (Allen et al. 2002). Spangled perch reproduction has been studied under laboratory conditions using fish from Tottenham in the New South Wales Murray-Darling Basin which were then transported to Sydney (Llewellyn 1973). Although rising water levels have historically been associated with the recruitment behaviour of this species, it is currently considered that rising temperature is more likely to be a recruitment cue (Pusey et al . 2004). In far western Queensland, spangled perch have been demonstrated to evince large population declines in receding waterholes of Cooper Creek (Arthington et al. 2005) and an extended spawning season with a peak linked to summer flow elevation has

35 been suggested for this species (Balcombe et al. 2007). Like other occasionally abundant species in the Lake Eyre Basin, such as silver tandan and bony bream, it seems likely that spawning itself may occur regularly and without reliance on flow.

Spangled perch are generally regarded as an extremely hardy species (Allen et al. 2002), and their ability to survive in central Australia is likely to be due to broad dietary and environmental tolerances. Frequently reported anecdotal accounts of spangled perch surviving in vehicle wheel ruts and swimming across recently-flooded paddocks (Wager and Unmack 2000) warrant further investigation of the colonisation and migratory abilities of this species, particularly as enhanced mobility and tolerance to depressed levels of dissolved oxygen would create competitive advantages for an arid-zone fish. Studies of spangled perch in more tropical systems indicate that migration associated with breeding or accessing floodplain habitats plays an important role in the life cycle of this species (Pusey et al . 2004), and it seems likely that these behavioural traits would be replicated in the rivers of far western Queensland.

Welch’s grunter is endemic to the Lake Eyre and Bulloo-Bancannia basins, and morphologically closely resembles both silver perch and Barcoo grunter. The larvae and juveniles of all three species are reported to be extremely difficult – if not impossible – to tell apart (Costelloe et al. 2004; Bruce Sambell, Ausyfish, personal communication). Despite growing to an edible size, Welch’s grunter remain comparatively under-studied, both in terms of general biology and reproductive behaviour, although it is suggested that this species spawns following summer flow events (Merrick and Schmida 1984; Allen et al. 2002). Interest in Welch’s grunter as a potential aquaculture species has recently occurred, no doubt associated with the success of raising both silver perch and Barcoo grunter under culture conditions. Given the current lack of biological and ecological information in existence regarding this species, it is difficult to make predictions relating to habitat preferences, migration and breeding behaviour until more research is completed in the Lake Eyre and Bulloo-Bancannia basins.

Barcoo grunter is endemic to the Lake Eyre and Bulloo-Bancannia basins and closely resembles both Welch’s grunter and silver perch. Barcoo grunter have become a reasonably popular aquaculture species as they can be induced to spawn artificially

36 and they achieve high growth rates under culture conditions (personal observation). In aquaculture, Barcoo grunter are usually referred to as ‘Jade Perch’ as their common name is considered to be inferior in marketing terms. Despite the growth of the ‘Jade Perch’ industry, little is known of either the basic biology or reproductive behaviour of Barcoo grunter in their natural environment, although they are presumed to spawn in response to flood events (Allen et al . 2002). As with Welch’s grunter, research directed towards the life cycle and environmental requirements of this species is required in order to make predictions regarding habitat preferences, movement and breeding behaviour.

Given the limited distribution of both Welch’s and Barcoo grunter, and the fact that both species are likely to be used in Australian freshwater aquaculture, a need exists to prioritise the study of these in their natural environment. Recent aquaculture spawning experiments suggest that Welch’s grunter may interbreed with silver perch (Bruce Sambel, Ausyfish, personal communication). As translocated silver perch may already be present in the Lake Eyre Basin (Wager and Unmack 2000), this situation does not bode well for maintenance of the genetic integrity of this species in its natural habitat.

Banded grunter has a limited distribution in the Queensland Lake Eyre Basin and this species has only been recorded from the Georgina catchment (Bailey and Long 2001). Although a single record exists for this species at the Queensland Museum from the Diamantina catchment (Jeff Johnson, personal communication), the collection location – Lake Muncoonie on Eyre Creek - is associated with the Georgina system and is north of Goyder’s Lagoon, the recognised junction of the Diamantina and Georgina catchments in South Australia. It seems likely that banded grunter are comparatively recent biogeographical additions to the Lake Eyre Basin fish fauna (or the last remnants of an ancient population), as this species is also widely recorded from coastal drainages from the Ashburton River in Western Australia across to the in Queensland (Allen et al. 2002). The wide distribution of banded grunter suggests that it is almost certainly an ecological generalist and therefore capable of withstanding a similarly wide range of environmental conditions, however no detailed research has occurred to date regarding the behaviour of this species in the Lake Eyre Basin. Evidence from both the region (Bishop et al. 2001)

37 and the (Pusey et al. 2004) suggests that banded grunter are opportunistic spawners unlikely to rely on flooding. These findings therefore suggest that this Terapontid species, like spangled perch, but perhaps unlike Welch’s or Barcoo grunter, may possess a flexible life history enabling it to adapt, or succeed, in highly variable systems such as the Georgina.

3.2.9 Gobiidae

The Gobiidae is a very large family of marine and freshwater benthic fishes, and three species of Gobiid fish occur in the Queensland Lake Eyre Basin. Elizabeth springs gobies, Chlamydogobius micropterus (Table 3.1), and Edgbaston springs gobies, Chlamydogobius squamigenus (Table 3.1), as their common names indicate, are restricted in their ranges to Elizabeth and Edgbaston springs, respectively. These endemic species are similar (and related) to the desert goby, Chlamydogobius eremius , from northern South Australia. Both localised goby species attain a comparatively small size (3 – 5cm) and exhibit highly specialised survival adaptations including the ability to gulp air (thus facilitating survival in waters with low dissolved oxygen), the ability to withstand extreme temperature fluctuations (5 – 41 ° celsius) and the ability to tolerate high salinity and pH levels (Allen et al . 2002). Both species, along with the red-finned blue-eye, are listed under state and national endangered species legislation (NCA 1992; EPBC 1999) and by the International Union for the Conservation of Nature as critically endangered (IUCN 2010). The third species of goby recorded from the Queensland Lake Eyre Basin is the golden goby, Glossogobius aureus (Figure 3.1m; Table 3.1), which grows far larger (up to 160mm; Pusey et al. 2004). 2 Like banded grunter, golden goby appear to be a predominantly tropical species with an isolated population existing in the Georgina catchment. Also like banded grunter, the colonisation of this western river in the Queensland Lake Eyre Basin does not appear to have facilitated further colonisation of the more easterly catchments despite periodic connection events. In the first instance, knowledge gaps relating to the geographical distribution of this species need to be

2 Early surveys and studies have mostly described medium-sized gobies in the Lake Eyre Basin as Glossogobius giurius (Merrick and Schmida 1984), however it is currently generally accepted that all specimens are G. aureus (Helen Larson, Museum and Art Gallery of the Northern Territory, personal communication).

38 addressed, especially given that populations in the appear to evince a marine larval stage which would not be possible in inland rivers (Larson and Martin 1990). Additionally, survival adaptations in the unpredictable arid zone Georgina catchment are likely to be related to the ability to survive and recruit in drying waterholes, whereas the environmental conditions prevalent in the ocean- draining rivers of northern Australia are essentially opposite (such as waterholes annually replenished from comparatively predictable monsoonal rainfall). Golden gobies can be considered amongst the most poorly known species in the Queensland Lake Eyre Basin as a result of their limited distribution and the paucity of data that exists relating to the Georgina catchment (Bailey and Long 2001). Addressing this knowledge gap can only be achieved through more thorough surveys and targeted projects in far western Queensland.

3.2.10 Eleotridae

Gudgeons of the Eleotridae family are generally common and diverse in most Australian drainage basins, although this family is comparatively under-represented in central Australia. Endemic purple-spotted gudgeons ( Mogurnda spp.) occur in springs and waterways in the Northern Territory and South Australia, and there is a possibility that isolated populations may exist elsewhere, most particularly in the Bulloo and Barcoo catchments (Wager and Unmack 2000; Table 3.1) 3. Nevertheless, within the Queensland Lake Eyre and Bulloo-Bancannia basins the only common Eleotrids are small gudgeons of the Hypseleotris complex that are restricted to the Cooper and Bulloo catchments. Confusion surrounds the taxonomy of the Hypseleotris gudgeons, especially the differences between three (or four) species, all of which appear to inhabit the Murray-Darling Basin. Hypseleotris klunzingeri , commonly known as the western carp gudgeon, has been described by Ogilby in 1898 (Merrick and Schmida 1984; McDowall 1996; Allen et al. 2002). Hypseleotris sp. 4 (Midgley’s Carp Gudgeon) and sp. 5 (Lake’s Carp Gudgeon)(Larson and Hoese 1996), which have also been referred to as sp. 1 and sp. 2 (Merrick and Schmida 1984; Allen et al . 2002 ) have not been described. In addition, a further species, the

3 Purple-spotted gudgeons ( Mogurnda spp.) have been reported from museum records in the Bulloo catchment (Midgley et al. 1991) and from personal observation in the Barcoo catchment (Unmack, personal communication) For the purposes of this study the status of this species complex in the studied catchments is considered to be unknown and hence they have been omitted from this review.

39 Murray-Darling carp gudgeon ( Hypseleotris sp. 3) is mentioned by both Unmack (2000) and Allen et al . (2002), and hybridization between some members of the Hypseleotris complex is suggested by Bertozzi et al . (2000), Thacker and Unmack (2005), Unmack (personal communication) and Hammer (personal communication). Given this level of confusion, it is difficult to draw any conclusions as to the species (or multiple species, or hybrid species) which may inhabit far western Queensland, and solving the taxonomic problem of the Hypseleotris gudgeon complex remains an on-going challenge to those working in freshwater fish research in Australia. Consequently, it is most common for these species (or hybrids) to be treated as a species complex or group (Lintermans 2007).

Hypseleotris gudgeons in southern Australia have been demonstrated to prefer low- flow environments for spawning (King et al. 2003), and the presence of gravid females and a large variation in sizes at seasonally different sampling periods suggests that this species (or species complex) may have an extended spawning period or may spawn year round (Kerezsy 2005). Spawning in response to flooding has also been reported (Milton and Arthington 1984) and work completed in the South Australian section of Cooper Creek suggests that Hypseleotris gudgeons demonstrate increased recruitment activity following flood events (Puckridge 1999). During sampling undertaken on the Cooper Creek floodplain in the vicinity of Windorah in south- western Queensland, both juvenile and adult carp gudgeons have been detected (Balcombe et al. 2007), indicating that this species may opportunistically colonise ephemeral areas following inundation. In common with Australian smelt, rainbowfish and glassfish, Hypseleotris gudgeons appear to be short-lived and reach maturity quickly (Pusey et al. 2004). It is therefore likely that this species may utilise highly adaptive reproduction strategies within the Lake Eyre and Bulloo-Bancannia basins. Within the southern and western Murray-Darling Basin, Hypseleotris gudgeons are often sampled in large numbers during fish sampling operations in regulated rivers (Budd 2005; Kerezsy 2005). In contrast, the number of Hypseleotris sp. sampled in Cooper Creek has been comparatively small during all relevant studies (Arthington et al . 2005; Balcombe et al. 2007; Balcombe and Arthington 2009). This disparity introduces the concept of the recruitment success of such species in regulated versus unregulated systems, and possibly suggests that the modifications made to river systems in the Murray-Darling Basin may have inadvertently resulted in conditions

40 which favour population expansion of this species. Alternatively, it is equally possible that populations of carp gudgeons in far western Queensland – where ecosystems and food chains are unaffected by large numbers of introduced species – are regulated by naturally occurring populations of piscivorous species. The distribution of carp gudgeons raises a biogeographical question relating to their occurrence in the Lake Eyre and Bulloo-Bancannia basins and the rivers of the north-western Murray-Darling Basin in Queensland. This suggests that carp gudgeons (Figure 3.1n; Table 3.1) most probably colonised the Lake Eyre Basin from the east, yet have been unable to reach the Diamantina and Georgina catchments.

Table 3.1 Summary table of fish species present in the major catchments of the Queensland Lake Eyre and Bulloo-Bancannia basins. Dots indicate presence, question marks indicate possible presence requiring confirmation (based on existing literature).

Family Species Common Distribution by catchment name Georgina Diamantina Cooper Bulloo

Clupeidae Nematolosa erebi (Günther 1868) Bony bream ● ● ● ● Plotosidae Neosiluroides cooperensis Cooper Creek ● Allen & Feinberg 1998 catfish Neosiluris hyrtlii Steindachner 1867 Hyrtl’s tandan ● ● ● ● Neosiluris sp. Bulloo false- ? spined catfish Porochilus argenteus Zietz 1896 Silver tandan ● ● ● ● Retropinnidae Retropinna semoni (Weber 1895) Australian smelt ● Melanotaeniidae Melanotaenia splendida tatei Desert ● ● ● ● (Peters 1866) rainbowfish Pseudomugilidae Scaturiginichthys vermeilipinnis Red-finned blue- ● Ivantsoff, Unmack, Saeed & eye Crowley 1991 Ambassidae Ambassis sp. Northwest ● ● ● ●/? Ambassis or Glassfish Percichthyidae Macquaria sp. Yellowbelly ● ● ● ●

Terapontidae Amniataba percoides Banded or Barred ● (Günther 1864) grunter Bidyanus welchi Welch’s grunter ● ● ● ● McCulloch & Waite 1917 Leiopotherapon unicolor Spangled perch ● ● ● ● (Günther 1859) Scortum barcoo Barcoo grunter ● ● ● ● McCulloch & Waite 1917 Gobiidae Chlamydogobius micropterus Elizabeth Springs ● Larson 1995 goby Chlamydogobius squamigenus Edgbaston goby ● Larson 1995 Glossogobius aureus Golden goby ● Akihito & Meguro 1975 Eleotridae Hypseleotris spp. Carp gudgeon ● ●

Mogurnda spp. Purple-spotted ? ? gudgeon

41

a) Bony bream b) Cooper Creek catfish c) Hyrtl’s tandan

d) Silver tandan e) Australian smelt f) Desert rainbowfish

g) Glassfish h) Yellowbelly i) Banded grunter

j) Welch’s grunter (A. Emmott) k) Spangled perch l) Barcoo grunter

m) Golden goby (Emma Kerezsy) n) Carp gudgeon

Figure 3.1 Native fish species from the major catchments of the Queensland Lake Eyre and Bulloo-Bancannia basins.

42 4. Variability of the study sites: Hydrology and water quality

This study was carried out in eight catchments and on nine occasions in the Queensland Lake Eyre and Bulloo-Bancannia basins, commencing in May 2006 and concluding in November 2008.

In May 2006 trials of sampling equipment were undertaken in the Barcoo River in the vicinity of Coolagh waterhole west of Blackall (Figure 4.1; Appendix 1). In September 2006, December 2006 and January 2007 sampling was carried out in the greater Cooper catchment, with a minimum of three sites located in the Kyabra Creek, Thomson River, Barcoo River and Cooper Creek sub-catchments (Figure 4.1).4 In April, August and November 2007, and finally in March/April 2008, sampling was carried out in a total of eight catchments including the Mulligan River, Georgina River, Diamantina River, Thomson River, Cooper Creek, Barcoo River, Kyabra Creek and the Bulloo River (Figure 4.1).

In November 2008 sampling was conducted at extra sites within the Queensland Lake Eyre Basin in an effort to extend distributional data for cryptic or invasive species. Consequently, sites at an upper reach in each respective catchment (Georgina, Diamantina and Thomson/Cooper) were selected (Figure 4.2).

4 Throughout the document the term ‘greater Cooper catchment’ is used ro refer to the Thomson, Barcoo, Kyabra and Cooper sub-catchments collectively.

43 Lake Idamea Georgina Main Channel Lower Lake

S-Bend Parapituri Ocean Bore Kunnamuka Swamp Lake Billyer Thomson Main Channel Hunter’s Gorge Dune Pond Warracoota Native Pulchera Lake Constance Coolagh (all) Vergemont Waterloo Isisford

Currarreva Shed Murken Bulloo Main Channel Bulloo House Springfield Bulloo Shed Springfield South One Mile

Figure 4.1 Site map showing the location of all sites sampled during the study in the Mulligan, Georgina, Diamantina, Thomson, Barcoo, Cooper, Kyabra and Bulloo catchments between September 2006 and March/April 2008. (Map courtesy Vanessa Bailey, Queensland Environmetal Protection Authority/Queensland Department of Environment and Resource Management).

44 Lake Mary

Lake Dunn

Walkaba/Jimberella waterhole

Conn waterhole

Rocky crossing, Mayne River

Spring creek, Goneaway

kilometres

100 200 300

Figure 4.2 The location of sampling sites in November 2008 showing Lake Mary and Walkaba/Jimberella waterhole in the Georgina catchment, Conn waterhole, Mayne River and Spring Creek in the Diamantina catchment and Lake Dunn in the Thomson catchment. (Map courtesy Jenny Silcock, Queensland Environmental Protection Agency/Queensland Department of Environment and Resource Management).

4.1 Hydrology of the catchments preceding and during the study

Major flooding occurred in the Bulloo catchment in summer 2007 and 2008 (Figure 4.3), in the greater Cooper catchment in summer 2006 and 2008 (Figure 4.3) and in the Georgina catchment in summer 2007 (Figure 4.3). Although flows occurred in the Diamantina catchment in summer 2006 and summer 2007, this catchment did not experience major flooding (Figure 4.3). All flood events occurred during the period from January to April in all catchments (Figure 4.3)

45

200 Bulloo catchment 150

100

50

200 Cooper catchment 150

100

50

200 Diamantina catchment 150

Megalitres1000 x (ML) 100

50

200 Georgina

150 catchment no data January to 100 April 2005 due to de-commissioned 50 gauging station Jan– 08 Apr May –May Jun 09 Sep– 05 Dec Jan –06 Apr – May Aug06 Sep– 06 Dec Jan –07 Apr – May Aug07 Sep– 07 Dec Jan –05 Apr –May Aug 05

Figure 4.3 Hydrology of the studied catchments between January 2005 and June 2009. The data has been taken from the following gauging stations: Quilpie (Bulloo catchment), Longreach (Cooper catchment), Diamantina Lakes (Diamantina catchment) and Roxborough Downs (Georgina catchment). Source: Queensland Department of Environment and Resource Management.

46 4.1.1 Sites in the Mulligan catchment

The Mulligan River rises approximately 10 kilometres east of the Northern Territory border and meanders in a south and south-easterly direction for approximately 300 kilometres before it meets Eyre Creek south-west of the township of Bedourie (Figure 4.1). The Mulligan River is an ephemeral desert stream and forms the north-eastern boundary of the Simpson Desert. Although the headwaters of the Mulligan River are situated in a gorge, for the majority of its length the Mulligan is a sandy river bed consistent with its location in the eastern dunefields of the Simpson Desert. The Mulligan River was dry in November 2006 (Scott Morrison, Ethabuka Station, personal communication), however major flooding in far western Queeensland occurred in January and February 2007 (see Section 4.1.2). The Mulligan River was in flood for approximately four weeks following the rainfall in January and February 2007 (Scott Morrison, Ethabuka Station, personal communication). Following the recession of floodwaters the Mulligan had become a series of disconnected waterholes by March 2007.

In April 2007 five sites (Pulchera Waterhole, Dune Pond, Kunnamuka Swamp, Ocean Bore, S Bend Gorge) were sampled in the Mulligan catchment (Figure 4.1). No rainfall or inflows occurred in the catchment between April and August 2007, and consequently only two sites (Pulchera Waterhole and S Bend Gorge) contained sufficient water to enable sampling to be undertaken in August. By November 2007, the Mulligan was dry with the exception of one site (Pulchera Waterhole), and by March/April 2008 this site had also dried.

Pulchera Waterhole

Pulchera Waterhole is considered the largest waterhole on the Mulligan River (Angus Emmott, Chair, Lake Eyre Basin Community Advisory Council, personal communication) and is situated approximately halfway between the headwaters and Eyre Creek (Figure 4.1). During the current study, Pulchera Waterhole was approximately 5 kilometres long, 300 metres wide and 2.5 metres deep in April 2007 following the flood event of early 2007. By November 2007, Pulchera had receded to approximately 2 kilometres long, 100 metres wide and up to 0.7 metres deep (Figure

47 4.4). Pulchera waterhole occupies a swale between two low-lying dunes in the eastern Simpson Desert dunefield. Consequently, the substrate of Pulchera Waterhole consists of sand overlain by a layer of silt up to 0.2 metre deep. The north-western shoreline of Pulchera waterhole is characterised by a gypsum outcrop approximately 300 metres long. Fringing vegetation comprised gidgee trees, spinifex and associated desert vegetation growing comparatively close (<5 metres) to the water’s edge in April 2007 but separated by an increasingly wide river bank as the water level receded (Figure 4.4).

S Bend Gorge

S Bend Gorge is situated in the headwaters of the Mulligan River approximately 20 kilometres east of the Queensland/Northern Territory border. At S Bend Gorge the river channel is approximately 10 metres wide and is surrounded on both sides by eucalypts and steep rocky hillsides and gullies (Figure 4.5). The substrate of the Mulligan River at S Bend Gorge comprises sand overlain by scattered rocks from the surrounding gorge. Following the recession of floodwaters resulting from summer rain, S Bend Gorge was 1.2 metres deep in April 2007. By August 2007 S Bend Gorge had become a single pool approximately 40 metres long and 5 metres wide, and was 0.4 metres deep at its deepest point (Figure 4.5). Two weeks after the August samples were taken, S Bend Gorge had dried, thus preventing any further sampling of this site (Len Rule, Craven’s Peak Station, personal communication).

Dune Pond

Dune Pond is a depression at the base of a sand dune in the eastern Simpson Desert. Dune Pond does not lie on a watercourse and is situated approximately 30 kilometres north-west of Pulchera Waterhole (Figure 4.1). Dune Pond filled during the flooding event in January 2007 and was approximately 40 metres long, 10 metres wide and 0.1 metre deep in April 2007 (Figure 4.6). Dune pond dried up within a week of the April sample being taken (Scott Morrison, Ethabuka Station, personal communication), thus preventing further sampling of the site.

48 Kunnamuka Swamp

Kunnamuka Swamp is an ephemeral wetland located between two Simpson Desert sand dunes. Kunnamuka Swamp is approximately 35 kilometres south-west of S Bend Gorge and consequently geographically isolated from the main channel of the Mulligan River (Figure 4.1). Kunnamuka Swamp filled during flooding in January 2007 when it was approximately 5 kilometres long, up to 300 metres wide and up to 0.6 metres deep (Len Rule, Craven’s Peak station, personal communication). Kunnamuka Swamp dried between the April and August 2007 sampling events, thus preventing further sampling of the site (Figure 4.7). Essentially a drowned swale, Kunnamuka Swamp was characterised by inundated gidgee trees and a sand substrate.

Ocean Bore

Ocean Bore is a groundwater-fed reservoir on Craven’s Peak station located approximately 30 kilometres south-west of S Bend Gorge (Figure 4.1). Ocean Bore is circular, with a diameter of approximately 15 metres and is up to 3 metres deep (Figure 4.8). Ocean Bore was sampled in April 2007 only.

49

Figure 4.4 Pulchera Waterhole in April 2007 (left), August 2007 (middle) and November 2007 (right).

Figure 4.5 S Bend Gorge in April 2007 (left) and shortly before it dried in August 2007 (right).

Figure 4.6 Dune pond in April 2007 (left) and August 2007 (right).

50

Figure 4.7 Kunnamuka Swamp in April 2007 (left) and August 2007 (right).

Figure 4.8 Ocean Bore in April 2007.

51 4.1.2 Sites in the Georgina catchment.

The Georgina River rises north of in the border area of Queensland and the Northern Territory (Figure 4.2). The Georgina catchment includes tributaries such as Eyre Creek, King Creek, Pituri Creek and the , and merges with the Diamantina River at Goyder’s Lagoon in South Australia. Unlike the Mulligan River, the Georgina is the westernmost catchment in the Queensland Lake Eyre Basin to exhibit an anastomosing channel system underlain by deep cracking clays. These channels are surrounded by slightly higher rock and gibber plains. The Georgina River held extremely low water levels in November 2006 (Stephen Bryce, Glenormiston Station, personal communication), with water limited to deep waterholes situated within the main channel. Major flooding in far western Queensland occurred in January and February 2007 (Figure 4.9), and the Georgina River was in flood for approximately four weeks (Figure 4.10). Following the recession of floodwaters the Georgina had become a series of disconnected waterholes by March 2007.

250000

200000

Flood peak 22 Jan – 2 Feb, 2007

150000 ML/day

100000

50000

0 April 2007 August 2007 November 2007 March/April 2008

Figure 4.9 Daily discharge (ML) of the Georgina River at Roxborough Downs from October 2006 – May 2008. Note the complete absence of flow in the Georgina during summer 2007/2008 (Source: Queensland Department of Environment and Resource Management).

52 9

8

7 MajoMajor 6 r 5 ModerateModerate 4 MinorMinor 3 Daily height (metres) Daily 2

1

0 1 10 20 30 1 10 20 January/February 2007

Figure 4.10 Daily heights for the Georgina River at Roxborough Downs during the period of flooding in early 2007 (Source: Queensland Department of Environment and Resource Management; for flood heights: Australian Bureau of Meteorology).

Georgina Main Channel

The Georgina Main Channel site is located approximately 110 kilometres west of the township of Boulia (Figure 4.1). The site is situated on Glenormiston Station which is owned by the North Australian Pastoral Company. The Georgina Main Channel site comprises a deep section of river that is approximately 50 metres long, 15 metres wide and up to 3 metres deep for the majority of each year (Stephen Bryce, Glenormiston Station, personal communication). The site is surrounded by steep river banks and a canopy of ti-tree and river red gum (Figure 4.11). In April 2007, following flooding in the Georgina catchment in January and February, the site was connected to other areas in the main channel of the Georgina River (Figures 4.1 and 4.11). Three months later in August 2007 the site had receded to its average dimensions following a very slight within-channel connection flow in June/July (Stephen Bryce, Glenormiston Station, personal communication). By November 2007, following a period of zero flow, the waterhole had receded to be approximately 30 metres long, 8 metres wide and up to 2.5 metres deep (Figure 4.11). The Georgina Main Channel site still held a similar level of water in March/April 2008 despite a dry summer. This suggests that, like many waterholes in the Georgina catchment, this waterbody is maintained by sub-surface springs as well as surface run-off (Stephen

53 Bryce, Glenormiston Station, and Shane McGlinchey, Badalia Station, personal communications).

Lake Idamea

Lake Idamea is a shallow (up to 2 metres deep) lake located on Pituri Creek approximately 5 kilometres west of the main channel of the Georgina River on Glenormiston Station (Figure 4.1). Lake Idamea filled following flooding in the Georgina catchment in January 2007, and in April 2007 it was approximately 2 kilometres long and 500 metres wide (Figure 4.12). Pituri Creek did not receive the small mid-winter flow experienced by the main channel of the Georgina River in 2007, and consequently the shoreline of Lake Idamea had receded by 15 metres in August 2007 and a further 20 metres by November 2007 (Figure 4.12). By March/April 2008, and following a dry summer, Lake Idamea had receded to a small ‘sump’ approximately 200 metres long by 75 metres wide and up to 0.4 metre deep (Figure 4.12). The shoreline of Lake Idamea is surrounded by coolibah woodlands, however for the majority of the time the lake is separated from riparian vegetation, and it frequently dries completely, the most recent dry period being in 2004 (Stephen Bryce, Glenormiston Station, personal communication).

Lower Lake

Lower Lake is a shallow (up to 2 metres deep) lake located on Pituri Creek approximately 10 kilometres west of the main channel of the Georgina River and 5 kilometres downstream of Lake Idamea on Glenormiston Station (Figure 4.1). Lower Lake was dry prior to flooding in the Georgina catchment in January and February 2007 (Stephen Bryce, Glenormiston Station, personal communication), and in April 2007 it was approximately 1.6 kilometres long and 150 metres wide (Figure 4.13). Pituri Creek did not receive the small mid-winter flow experienced by the main channel of the Georgina River in 2007, and consequently the shoreline of Lower Lake had receded by 45 metres in August 2007 (Figure 4.13). Lower Lake had dried completely by November 2007 (Figure 4.13).

54 Parapituri (also known as Paravituari)

Parapituri is a long (10 kilometre) section of the main channel of the Georgina River located along an east-west axis approximately 50 kilometres west of the township of Boulia (Figure 4.1). The southern shore of Parapituri is on Herbert Downs, a North Australian Pastoral Company property, while the northern shore is bordered by the smaller family-owned properties Wirrilyerna and Badalia. Parapituri is a comparatively deep (up to 5 metres) section of channel with a maximum width of approximately 150 metres. Parapituri is characterised by steep banks and ti-tree, lignum and coolibah riparian communities. Parapituri had dried to an average maximum depth of 1.5 metres by November 2007 following a period of zero flow since June/July 2007 (Shane McGlinchey, Badalia Station, personal communication)(Figure 4.14). Parapituri still held a similar level of water in March/April 2008 despite a very dry summer. This suggests that, like many waterholes in the Georgina catchment, this waterbody is maintained by sub-surface springs as well as surface run-off (Stephen Bryce, Glenormiston Station, and Shane McGlinchey, Badalia Station, personal communications).

Lake Mary

Lake Mary was the northernmost waterhole in the Georgina catchment that held water in November 2008 (Gavin Miller, Rocklands, personal communication) and is situated approximately 10 kilometres north of the township of Camooweal and 10 kilometres east of the Northern Territory/Queensland border on Rocklands, a property owned by the Western Grazing Company (Figure 4.2). Lake Mary was approximately 500 metres long, 70 metres wide and up to 1 metre deep in November 2008 (Figure 4.15). Like many waterholes in the Georgina catchment, Lake Mary is underlain by limestone.

55

Figure 4.11 The Main Channel site on the Georgina River, showing the connected channel in April 2007 (left), and waterhole recession following a drying period in August (middle) and November (right).

Figure 4.12 Lake Idamea showing the difference between full capacity following flood inundation in April 2007 (left), the subsequent period of no-flows to November 2007 (middle), and the Lake at its lowest point during the current study in March/April 2008 (right).

Figure 4.13 Lower Lake at full capacity following the recession of floodwaters in April 2007 (left), and as it dried down in August 2007 (middle) and November 2007 (right).

56

Figure 4.14 Parapituri showing dry season recession between August 2007 (left) and November 2007 (right).

Figure 4.15 Lake Mary, the only waterhole close to the northern Lake Eyre Basin divide that retained water in the Georgina catchment in November 2008.

Figure 4.16 Walkaba waterhole in the Georgina catchment, November 2008.

Walkaba/Jimberella

Walkaba waterhole on the Jimberella stock route is approximately 200 kilometres south of Lake Mary in the Georgina catchment (Figure 4.2). In November 2008, there were no water-filled waterholes between Lake Mary and Walkaba within the Queensland Lake Eyre Basin (personal observation). Walkaba is the longest permanent waterhole in the Georgina catchment (Pat Fennell, former owner, Linda Downs, personal communication), and in November 2008 was approximately 100 metres wide, greater than 3 metres deep and greater than 10 kilometres long (Figure 4.16). Walkaba, like many waterholes in the Georgina catchment, is spring-fed (Pat Fennell, personal communication), and is underlain by limestone.

57 4.1.3 Sites in the Diamantina catchment

The Diamantina River rises to the north of the township of Winton where it is fed by a conglomeration of smaller creeks (Figure 4.1). Like the Georgina River, the Diamantina is best described as a series of channels composed of deep cracking clays within far drier, rocky uplands. The Diamantina catchment is characterised by eroded escarpments for much of its length, particularly the central area where the majority of the described sites are situated (Figure 4.17). The Diamantina did not experience major flooding during the study period, and flows were restricted to minor flooding in summer 2006/2007 and connection flows in July 2007 and summer 2007/2008 (Figure 4.18). Sites utilised in the Diamantina catchment during the current study include Lake Billyer and 2 Mile on the Australian Agriculture company-owned Brighton Downs station and Hunter’s Gorge, Warracoota and Lake Constance located within Diamantina Lakes National Park (Figure 4.1). Permission to conduct sampling within Diamantina Lakes was granted by Queensland National Parks and Wildlife Service (Permit No: WITK04571107). Additional sites sampled in the Diamantina catchment in November 2008 included Rocky Crossing on the Mayne River, Spring Creek on the Tonkoro/Goneaway border and Conn waterhole, the northernmost permanent waterhole in the Diamantina catchment (Jenny Silcock, Queensland Environmental Protection Authority, personal communication) (Figure 4.2).

250000

200000

150000 ML/day

100000

50000

0 Jan/Feb 2007 July 2007 Dec 2007 – Jan 2008

Figure 4.18 Daily discharge of the Diamantina River at Diamantina Lakes from October 2006 – May 2008. (Source: Queensland Department of Environment and Resource Management).

58 Lake Billyer

Lake Billyer is a shallow ephemeral lake on an outer eastern channel of the Diamantina floodplain on Brighton Downs station (Figure 4.1). Lake Billyer typically fills from run-off from the ironstone and mesa country located on the adjacent property Mount Windsor. Lake Billyer was approximately 3 kilometres long and 100 metres wide in April 2007, with an average maximum depth of 0.8 metres (Figure 4.19). The shoreline of Lake Billyer receded 30 metres between April and August 2007, by which time the depth of the water was 0.4 metres (Figure 4.19). By October 2007 Lake Billyer had dried (Linda Young, Brighton Downs Station, personal communication), however storm-fed rainfall over Mount Windsor station resulted in the Lake filling again during a two week period in the first half of November 2007, partially inundating the narrow strip of coolibah and lignum vegetation that surrounds the Lake. By March/April 2008, the water level in Lake Billyer had receded back to the same level as August 2007.

2 Mile

2 Mile is a narrow (up to 10 metres wide) waterhole on the eastern side of the main Diamantina channel complex located to the south of the Brighton Downs homestead (Figure 4.1). In April 2007 2 Mile was approximately 350 metres long and 0.9 metres deep, and by August 2007 2 Mile had receded to be only 50 metres long, 3 metres wide and less than 0.3m deep overlying a 0.4m layer of silt (Figure 4.20). Although a canopy of coolibah and river red gum branches is present at 2 Mile, the waterhole lacks any riparian understorey (Figure 4.20).

59

Figure 4.17 An example of the landscape surrounding sampling sites in the Diamantina catchment showing eroded escarpments stretching to the north-east from the Brighton Downs/Diamantina Lakes border.

Figure 4.19 Lake Billyer during a drying period between April 2007 (left) and August 2007 (middle), and following re-filling from storm rainfall in November 2007 (right).

Figure 4.20 2 Mile waterhole showing water level recession between April (left) and August (right) 2007. 2 Mile dried within 2 weeks of the August samples being taken (Bob Young, Brighton Downs Station, personal communication).

Figure 4.21 Hunter’s Gorge, showing the difference in water level created by a small flow in early November 2007 (right) and the level of the waterhole in August 2007 (left).

60

Figure 4.22 Warracoota in August 2007 (left) and November 2007 (right) demonstrating the rise in water levels created by a small flow in early November.

Figure 4.23 Lake Constance in August (left) and November (right) 2007.

Figure 4.24 Rocky Crossing on the Mayne River, Diamantina catchment, in November 2008 (left).

Figure 4.25 The rockhole on Spring Creek, showing lancewood and river red gum riparian vegetation (right).

Figure 4.26 Conn Waterhole, the highest – or most northerly - permanent water on the Diamantina River, in November 2008.

61 Hunter’s Gorge

Hunter’s Gorge is a deep (>3 metres) waterhole located between two eroded escarpments in Diamantina Lakes National Park (Figure 4.1). Hunter’s Gorge was approximately 2 kilometres long and 150 metres wide in August 2007 (Figure 4.21). The dimensions of Hunter’s Gorge had increased slightly in November 2007 due to a small flow which resulted from storm run-off in sections of the Diamantina catchment during the first two weeks of November 2007 (Ian Andreassen, Ranger, Diamantina Lakes National Park, personal communication) (Figure 4.21). The water level of Hunter’s Gorge continued to fluctuate slightly due to sporadic flows throughout summer 2007/2008 (John Clemments, Ranger, Diamantina Lakes National Park). The shoreline of Hunter’s Gorge is characterised by gently sloping banks and thickets of lignum, the majority of which were locally flooded by the connection flows in summer 2007/2008.

Warracoota

Warracoota is a deep (>3 metres) waterhole located on an outer western channel in the Diamantina catchment (Figure 4.1). Warracoota was up to 2 kilometres long and 50 metres wide in August 2007, and rose approximately 0.5 metre following run-off from storm activity in parts of the Diamantina catchment in early November 2007. The water level in Warracoota had stabilised to August 2007 levels by March/April 2008 following sporadic connection flows in summer 2007/2008. Warracoota is a steep- sided waterhole with lignum, coolibah and river red gum communities persisting to the water’s edge and forming a riparian canopy along the river banks (Figure 4.22).

Lake Constance

Lake Constance is a shallow (<1 metre) lake located on the western side of the main Diamantina channel complex in Diamantina Lakes National Park (Figure 4.1). Lake Constance was approximately 3 kilometres long and up to 200 metres wide in August 2007 and these levels did not alter in November 2007. Although the Lake dried down between August and November, localised storms and resulting run-off re-filled it two weeks prior to the November 2007 samples being taken. By March/April 2008 the

62 water level in Lake Constance had receded to the August 2007 level following small sporadic summer flows (Figure 4.23).

Rocky Crossing, Mayne River

The Mayne River rises in the eastern Diamantina catchment and enters the Diamantina channel complex during periods of high flow in the Mt Windsor/Brighton Downs area (Figures 4.1 and 4.2). Situated close to the headwaters of the Mayne River, Rocky Crossing (on the Winton/Jundah road) is a deep waterhole (>3m) that was approximately 2 kilometres long and 100 metres wide in November 2008. As the name suggests, Rocky Crossing is underlain by broken rock and cobble and is fringed by river red gum (Figure 4.24).

Spring Creek, Goneaway National Park

Spring Creek is a narrow creekline fed by a permanent rock-hole on the Tonkoro/Goneaway National Park border (Figures 4.2 and 4.25). Spring Creek was approximately 4 kilometres long in November 2008, but never wider than 10 metres. Although Spring Creek was generally <1 metre deep, the section in the vicinity of the permanent rockhole was >3m deep (Figure 4.25). Spring Creek can be considered part of the Mayne River sub-catchment and therefore part of the Diamantina catchment, however it should be noted that Spring Creek is dry for the majority of the time and only fills following localised rainfall.

Conn waterhole

Conn waterhole is the northernmost permanent waterhole in the Diamantina catchment (Silcock 2009) and is situated west of Winton at the confluence of the Diamantina River with Wokingham Creek and the Western River (Figure 4.2). Conn waterhole was sampled in November 2008 when it was approximately 3 kilometres long, 100 metres wide and up to 2.5 metres deep (Figure 4.26).

63 4.1.4 Sites in the greater Cooper catchment

The greater Cooper catchment includes the Barcoo and Thomson Rivers as well as sites in Kyabra Creek and Cooper Creek itself downstream of the confluence of the Barcoo and Thomson rivers (Figure 4.1). As the Cooper catchment provides a useful area in which to conduct within-catchment (or reach) comparisons, the Cooper sub- catchments were treated separately in the current survey.

4.1.4.1 Sites in the Thomson catchment

The Thomson River rises above the town of Longreach in western Queensland and is also fed by smaller creeks (such as Vergemont Creek) as it follows a southerly course through districts such as Stonehenge and Jundah (Figure 4.1). Waterways in the Thomson catchment generally comprise a series of channels underlain by cracking clay situated within a floodplain and surrounded by higher country. Sites in the Thomson River utilised in the current study include Vergemont and Native waterholes (both in the Vergemont Creek channel system) and Waterloo and Thomson Main Channel waterholes in the Thomson channel system (Figure 4.1). Smaller waterholes and alternative sites were also used in the Thomson Main Channel area when road and weather conditions prevented access to the Main Channel itself, and to study fish colonisation of small recently-inundated waterholes. A site in the upper Thomson and close to the north-east Lake Eyre Basin divide – Lake Dunn – was sampled in November 2008 only (Figure 4.2).

Sites in the Thomson River experienced a small amount of flooding in summer 2006/2007, and occasional within-channel connection events throughout 2007. Major flooding occurred in the Thomson River from January 18 – January 21, 2008 (Bureau of Meteorology 2008), however the period of elevated flows and floodplain inundation lasted from approximately December 2007 to April 2008 (personal observation; Figure 4.27).

64 250000 Flood peak 18 Jan – 21 Jan 08

200000

150000 ML/day 100000

50000

0 Jan/Feb 2006 Summer 06/07 Summer 07/08

Figure 4.27 Daily discharge of the Thomson River at Longreach from January 2005 – May 2008. (Source: Queensland Department of Environment and Resource Management).

Vergemont Creek

Vergemont Creek is a comparatively large waterhole in the Vergemont system situated approximately 10 kilometres west of the homestead on Noonbah station, which is itself approximately 150 kilometres south-west of Longreach (Figure 4.1). The Vergemont Creek site was never less than 300 metres long, 15 metres wide and 1.5 metres deep during the current study (Figure 4.28), and the hydrological history of the Vergemont Creek site is summarised in Table 4.1 for the duration of the study.

65 Table 4.1 Hydrological history of Vergemont Creek for the period December 2006 – March 2008. Date Antecedent flows Change in depth Connectivity Dec 2006 No flows in the N/A. Maximum depth 1.5 Isolated preceding 6 months metres Jan 2007 Small flow in mid- Slight rise following flow Connected to other waterholes December event. Maximum depth 1.8 in the same channel briefly metres. during mid-December but isolated again by January. April 2007 Small flows in Maximum depth 2.2 Connected to other waterholes January. metres. in the same channel during January but isolated again shortly afterwards. August 2007 Small flow in July. Maximum depth 2.5 Connected to other waterholes metres. in the same channel during July but isolated again shortly afterwards. November Small flow during the Maximum depth 3 metres. Remains connected to 3 2007 first week of smaller upstream waterholes November (prior to on the same channel when sampling). sampled. No connection to other channels (laterally situated on the floodplain). March/April Moderate flooding in Maximum depth >3 Connected to other Vergemont 2008 January. metres. Creek waterholes and the Thomson River itself during January flooding. Isolated by March.

Native Waterhole

Native is a small waterhole in the Vergemont system situated approximately 3 kilometres west of the homestead on Noonbah station, which is itself approximately 150 kilometres south-west of Longreach (Figure 4.1). Native is situated on a different channel of the Vergemont system from the Vergemont Creek site. The Native site was never less than 20 metres long, 8 metres wide and 1.1 metres deep during the current

66 study (Figure 4.29), and the hydrological history of the site is summarised in Table 4.2 for the duration of the study.

Table 4.2 Hydrological history of Native Waterhole for the period December 2006 – March 2008. Date Antecedent flows Change in depth Connectivity Dec 2006 No flows in the N/A. Maximum depth 1.5 Isolated preceding 6 months metres Jan 2007 No flows Maximum depth 1.1 Isolated metres. April 2007 Small flows in Maximum depth 1.6 Connected to other waterholes January. metres. in the same channel during January but isolated again shortly afterwards. August 2007 No flows. Maximum depth 1.3 Isolated metres. November No flows Maximum depth 1.1 Isolated 2007 metres. March/April Moderate flooding in Maximum depth 2.5 Connected to other Vergemont 2008 January. metres. Creek waterholes and the Thomson River itself during January flooding. Isolated by March.

67

Figure 4.28 The Vergemont Creek site at its lowest level during the study in December 2006 following an extended period of no-inflows.

Figure 4.29 Native Waterhole following an extended period of no-inflows in December 2006.

Figure 4.30 The sampling site at Waterloo in August 2007 following a recent flow that filled the waterhole.

68

Figure 4.31 Sites sampled in the Thomson River Main Channel.

Top: The Main Channel site in September 2006 (left) and dry in December 2006 (right). Middle: Three alternative sites sampled in August 2007 (l to r) Thomson 1 st Channel, Thomson Small and Thomson Tiny. Thomson 1 st Channel was also sampled in November 2008. Bottom: Thomson Backflow, the only site that could be accessed in November 2007 due to wet conditions.

Figure 4.32 Lake Dunn, in the upper Thomson catchment, in November 2008.

69 Waterloo

Waterloo is a comparatively large waterhole in the main channel system of the Thomson River (Figure 4.30). Waterloo is situated approximately 5 kilometres south of the bridge crossing the Thomson River at Lochern National Park and is therefore approximately 30 kilometres north-east of the township of Stonehenge (Figure 4.1). During the current study Waterloo remained longer than 1 kilometre, wider than 30 metres and its maximum depth at the sampling site never fell below 2 metres. The hydrological history of Waterloo is given in Table 4.3 for the duration of the study.

Table 4.3 Hydrological history of Waterloo for the period September 2006 – November 2008. Date Antecedent flows Change in depth Connectivity Sep 2006 No flows N/A. Maximum depth 3 Isolated. metres. Dec 2006 No flows Maximum depth 2.5 Isolated metres. Jan 2007 No flows Maximum depth 2 metres. Isolated April 2007 Flows in January. Maximum depth 3 metres. Connected to other waterholes in the Thomson system in January. Isolated by April. August 2007 Flows in July. Maximum depth 3.3 Connected to other waterholes metres. during flow event in July but isolated again by August. November Flows in November Maximum depth 3.5 Connected to, and part of, the 2007 metres. Thomson River during sampling. March/April Major flooding in Maximum depth >3.5 Connected to, and part of, the 2008 January metres. Thomson River during sampling. November Connection flows in No change in depth Isolated from the main 2008 October channel of the Thomson for approximately 2 weeks prior to sampling.

70 Thomson River Main Channel .

The Thomson River Main Channel site is situated approximately 3 kilometres south of the bridge across the Thomson River at Lochern National Park and immediately south of the Lochern National Park boundary (Figure 4.1). Access to the Thomson River Main Channel site is through a series of outer channels (Figure 4.31). Consequently, although the Main Channel site was sampled effectively in September 2006, April 2007 and March/April 2008, alternative sites on the outer channels were sampled in August and November 2007 and in November 2008. In December 2006 the site was dry, and no sample was taken from the site in January 2007 due to high flows. The hydrological history of the Thomson River Main Site and alternative sites is given in Table 4.4. Table 4.4 Hydrological history of the Thomson Main Channel waterholes for the period September 2006 – March 2008. Date Antecedent flows Change in depth Connectivity Sep 2006 No flows N/A. Maximum depth 0.8 Isolated. metres. Dec 2006 No flows Waterhole dry. N/A Jan 2007 High flows prevent Maximum depth 2 metres. Connected to, and part of the sampling. Thomson River. April 2007 Flows in January. Maximum depth 2.5 Connected to other waterholes metres. in the Thomson system in January. Isolated by April. August 2007 Flows in July prevent access to original site. Three smaller sites sampled, all recently- filled as a result of the July flows include (length x width x depth): Thomson 1 st channel: 500m x 20m x 1.2m Thomson Small: 20m x 4m x 0.4m Thomson Tiny: 4m x 2m x 0.1m November Flows in November prevent access to original site. Recently-inundated Thomson 2007 Backflow site sampled adjacent to the main flow. This site is connected to the Main Channel, is 6 metres wide and maximum depth 0.8 metres. March/April Major flooding in Maximum depth 2.5 Part of the main connected 2008 January. metres. channel of the Thomson from (approx.) December – April. November Flows in October prevent access to original site. Recently-inundated Thomson 1 st 2008 channel site sampled. This site is connected to the Main Channel, is 10 metres wide and has a maximum depth of 1.2 metres.

71 Lake Dunn

Lake Dunn is situated in the north-eastern section of the Lake Eyre Basin in the area known as the desert uplands (Figure 4.2). Lake Mary fills from run-off from the Great Dividing Range. Outflow from Lake Dunn feeds Reedy Creek, Cornish Creek and ultimately the Thomson River (Figure 4.2). Lake Dunn is a large circular lake with a diameter of approximately 5 kilometres and an average maximum depth of 1.5 metres in November 2008 (Jenny Silcock, Queensland Department of Environment and Resource Management, personal communication) (Figure 4.32).

4.1.4.2 Sites in the Barcoo catchment

The Barcoo River rises above the town of Blackall in western Queensland and is also fed by smaller waterways (such as the Alice River) as it follows a westerly and then south-westerly course through districts such as Isisford and Retreat, where it meets the Thomson and becomes Cooper Creek (Figure 4.1). Waterways in the Barcoo catchment generally comprise a series of channels underlain by cracking clay situated within a floodplain and surrounded by higher country. Sites in the Barcoo River used in the current study include three waterholes in the Coolagh complex and the Isisford Weir (Figure 4.1). In December 2006 and August 2007 extra sites were sampled in the Barcoo catchment including 8 Mile, to the north-east of Isisford and Oma, to the south of Isisford. Results from these waterholes have been omitted from the study and consequently their hydrological history is not discussed here.

The Barcoo catchment experienced major flooding in January 2008 and a period of extended floodplain inundation and elevated or fluctuating water levels from November 2007 – March 2008 (Figure 4.33). Minor flooding and/or within-channel connecting flows occurred in the Barcoo catchment in April 2006 and January 2007 (Figure 4.33).

72 250000

Flood peak Jan 24 – Jan 28 2008

200000

150000 ML/day 100000

50000

0 April 06 Jan 07 Dec 07 – Feb 08

Figure 4.33 Daily discharge of the Barcoo River at Retreat from January 2005 – May 2008. (Source: Queensland Department of Environment and Resource Management).

Coolagh

Coolagh is a large waterhole located approximately 120 kilometres north-west of Blackall at the confluence of the Barcoo and Alice rivers (Figure 4.1). Coolagh was inaccessible in January 2007 due to a minor flood and associated road conditions. A sample could not be taken from Coolagh in December 2006 due to thunderstorm activity. The hydrological history of Coolagh is given in Table 4.5. Coolagh is a comparatively deep waterhole and was never less than 2.5 metres deep at the sampling site. The waterhole is generally 2.5 kilometres long and up to 300 metres wide, and is characterised by steep banks and overhanging ti-tree, coolibah and red gum (Figure 4.34).

73 Table 4.5 Hydrological history of Coolagh for the period June 2006 – March 2008. Date Antecedent flows Change in depth (at Connectivity sampling site) June 2006 Flows in April N/A. Maximum depth 3.5 Connected to other waterholes metres. in the Barcoo main channel in April and May. Isolated in June. Sep 2006 No flows Maximum depth 2.5 Isolated metres. Dec 2006 No sample due to thunderstorms. Jan 2007 No sample due to minor flooding and inaccessibility. April 2007 Minor flood flows in Maximum depth 3.5 Connected to other waterholes January. metres. in the Barcoo system in December, January and February. Isolated by April. August 2007 No flows. Maximum depth 3 metres. Isolated. November Flows in November Maximum depth 3.5 Connected to, and part of, the 2007 metres. Barcoo River during sampling. March/April Major flooding in Maximum depth >3.5 Connected to, and part of, the 2008 January; elevated metres. Barcoo River during flows November – sampling. March.

Coolagh 2

Coolagh 2 is a small ephemeral waterhole located approximately 200 metres north of the Coolagh waterhole but within the same floodplain (Figure 4.1). Coolagh 2 was inaccessible in January 2007 due to a minor flood and associated road conditions. The hydrological history of Coolagh 2 is given in Table 4.6. Coolagh 2 is a shallow waterhole and was never more than 1.5 metres deep at the sampling site. The waterhole is generally 300 metres long and up to 20 metres wide when full (Figure 4.35).

74 Table 4.6 Hydrological history of Coolagh 2 for the period June 2006 – March 2008 Date Antecedent flows Change in depth Connectivity June 2006 Flows in April N/A. Maximum depth 1 Connected to other waterholes metres. in the Barcoo main channel in April and May. Isolated in June. Sep 2006 No flows Maximum depth 0.6 Isolated metres. Dec 2006 No flows Dry Isolated Jan 2007 No sample due to minor flooding and inaccessibility. April 2007 Minor flood flows in Maximum depth 0.8 Connected to other waterholes January. metres. in the Barcoo system in December, January and February. Isolated by April. August 2007 No flows. Maximum depth 0.5 Isolated. metres. November Flows in November Maximum depth 1.2 Isolated following connection 2007 metres. event in the first week of November. March/April Major flooding in Maximum depth 1.5 Isolated following flood 2008 January; elevated metres. recession in early March 2008. flows November – March

75

Figure 4.34 Coolagh waterhole on the Barcoo River in June 2006 showing overhanging riparian vegetation.

Figure 4.35 Coolagh 2 following drying in December 2006 (left) and immediately after re-filling in November 2007 (right).

Figure 4.36 Coolagh 3 waterhole during a drying phase in April 2007 (left), dry in August 2007 (centre) and re-filled following summer flooding in March 2008 (right).

Figure 4.37 The sampling site at Isisford in November 2007 immediately after a flow peak and subsequent recession.

76

Figure 4.38 Waterholes sampled in the Cooper catchment (l to r): Murken waterhole (December 2006), Currareva (December 2006) and Shed (September 2006).

Figure 4.39 Springfield waterhole in Kyabra Creek in September 2006 following an extended period of zero inflows.

Figure 4.40 Springfield South receding during a drying period from September 2006 (left), to December 2006 (centre) and January 2007 (right).

Figure 4.41 The sampling site at One Mile waterhole on Kyabra Creek during sampling in November 2007, and showing the effects of a small within-channel flow between evening (left) and the following morning (right). The Creek rose 65cm overnight.

77 Coolagh 3

Coolagh 3 is a small waterhole located approximately 100 metres north of the Coolagh waterhole but within the same floodplain (Figure 4.1). Coolagh 3 was inaccessible in January 2007 due to a minor flood and associated road conditions. The hydrological history of Coolagh 3 is given in Table 4.7. Coolagh 3 is a shallow waterhole and was never more than 0.8 metres deep at the sampling site. The waterhole is generally 100 metres long and up to 6 metres wide (Figure 4.36).

Table 4.7 Hydrological history of Coolagh 3 for the period June 2006 – March 2008 Date Antecedent flows Change in depth Connectivity June 2006 Flows in April N/A. Maximum depth 0.6 Connected to other waterholes metres. in the Barcoo main channel in April and May. Isolated in June. Sep 2006 No flows Dry Isolated Dec 2006 No flows Dry Isolated Jan 2007 No sample due to minor flooding and inaccessibility. April 2007 Minor flood flows in Maximum depth 0.3 Connected to other waterholes January. metres. in the Barcoo system in December, January and February. Isolated by April. August 2007 No flows. Dry Isolated. November Flows in November Maximum depth 0.5 Isolated following connection 2007 metres. event in the first week of November. March/April Major flooding in Maximum depth 0.8 Isolated following flood 2008 January; elevated metres. recession in early March 2008. flows November – March

78 Isisford

Isisford is the town waterhole located immediately east of the township of the same name (Figure 4.1). Isisford was the only Barcoo waterhole sampled during the flooding in January 2007. The hydrological history of Isisford is given in Table 4.8. Isisford is a deep waterhole and was always deeper than 2 metres at the sampling site (Figure 4.37). The waterhole is generally 2 kilometres long and up to 200 metres wide. At its southern terminus Isisford is bordered by a concrete weir and is therefore one of the few sites in the Queensland Lake Eyre Basin that could be considered regulated.

Table 4.8 Hydrological history of Isisford for the period December 2006 – March 2008 Date Antecedent flows Change in depth Connectivity Dec 2006 No flows N/A. Maximum depth 2 Isolated metres. Jan 2007 Minor flooding in Maximum depth 4 metres Barcoo River connected and December and at sampling site. flowing during January 2007. January. Floodplain site used (depth = 1.2 metres). April 2007 Minor flood flows in Maximum depth 3 metres. Connected to other waterholes January. in the Barcoo system in December, January and February. Isolated by April. August 2007 No flows. Maximum depth 2.5 Isolated. metres. November Flows in November Maximum depth 3 metres. Barcoo River connected and 2007 flow receding during sampling in November 2007. March/April Major flooding in Maximum depth >3 Isolated following flood 2008 January; elevated metres. recession in March. flows November – March

79 4.1.4.3 Sites in the Cooper Creek catchment

Cooper Creek forms at the confluence of the Thomson and Barcoo rivers to the north- east of the township of Windorah, and then meanders as a series of channels within a clay floodplain up to 80 kilometres wide to the Queensland/South Australian border at Nappa Merie and thence to Lake Eyre. Sites in Cooper Creek surveyed during the current study include Murken and Currareva waterholes (both in the ‘western’ channel system, and supplied with water predominantly by flows from the Thomson River) and Shed waterhole on the eastern edge of the channels and supplied predominantly by Barcoo River flows (Figure 4.1). Although extra sites were sampled at a more southerly location on Durham Downs station in September 2006, these sites were not re-sampled and have not been included in the study. Hence, they are not discussed here.

Major flooding occurred in Cooper Creek in summer 2007/2008 as a result of upstream flooding in both the Thomson and Barcoo rivers. Smaller within-channel connection events occurred in Cooper Creek sites (particularly Murken and Currareva) during summer 2006/2007, again as a result of upstream hydrological changes in the two ‘feeder’ rivers.

Murken

Murken is a large waterhole located south-east of Windorah (Figures 4.1 and 4.38). Murken was always in excess of 2 kilometres long and 200 metres wide during the current study, and the depth of the sampling site was never less than 1.5 metres. The hydrological history of Murken for the study period is given in Table 4.9. During the November 2007 sampling period Cooper Creek rose 2.5 metres overnight, thus preventing the retrieval of sampling equipment. This equipment was later retrieved during sampling in March/April 2008.

80

Table 4.9 Hydrological history of Murken for the period September 2006 – March 2008. Date Antecedent flows Change in depth Connectivity Sep 2006 No flows N/A. Maximum depth 2 Isolated metres. Dec 2006 No flows Maximum depth 1.7 Isolated metres. Jan 2007 Within-channel flows Maximum depth 4 metres. Connected to other waterholes in late December and in the Windorah area (eg: early January. Currareva). April 2007 Flows in January. Maximum depth 2 metres. Connected to other waterholes in the Cooper system in December, January and February. Isolated by April. August 2007 Small flow in July. Maximum depth 2.5 Connected to other waterholes metres. in the Windorah area (eg: Currareva) during the July flow. Isolated by August. November High flows in Maximum depth 2 metres Connection to other 2007 November during followed by an overnight waterholes in the Cooper sampling. rise of 2.5 metres. Creek, Barcoo and Thomson catchments. March/April Major flooding in Maximum depth 2.5 Connection to all other 2008 January; elevated metres. waterholes Nov – March. flows November – March.

Currareva

Currareva is a large waterhole located 10 kilometres east of Windorah (Figures 4.1 and 4.38). The sampling site chosen at Currareva is on a small tributary which enters the main Currareva waterhole at the eastern end of the main bridge crossing (Figure 4.1). The Currareva site was highly variable in terms of depth and width due to the occasionally large flows which exited Currareva during high flow or flood periods (Table 4.10). Changes in dimensions of the site (length x width in metres) are

81 included in the hydrological history of this site due to this variability. In January 2007 and November 2007 high flows prevented sampling of the Currareva site. In both instances an alternative site approximately 500 metres downstream was used (Table 4.10).

Table 4.10 Hydrological history of Currareva for the period September 2006 – March 2008. Date Antecedent flows Change in depth Dimensions Connectivity Sep 2006 No flows N/A. Maximum 50 metres x 15 Isolated depth 3 metres. metres Dec 2006 No flows Maximum depth 15 metres x 5 Isolated 2.5 metres. metres. Jan 2007 Within-channel flows in late December and early January create a deep (8 metres +) swiftly flowing site in January, unsuitable for setting passive fishing equipment. An alternative site was utilized further downstream on Currareva with the following dimensions: 20m x 6m x 2m. April 2007 Flows in January. Maximum depth 3 50 metres x 15 Connected to other metres. metres. waterholes in the Cooper system in December, January and February. Isolated by April. August Small flow in Maximum depth 4 > 1 km x 25 Connected to other 2007 July. metres. Small flow metres. waterholes in the during sampling. Windorah area (eg: Murken) during July and August. November Within-channel flows in November create a deep (10 metres +) swiftly flowing site in 2007 unsuitable for setting passive fishing equipment. An alternative site was utilized further downstream on Currareva with the following dimensions: 20m x 6m x 2m. The depth of this site rose overnight due to in-flows by 1.5 metres, with the dimensions increasing to 50 metres x 20 metres in the same period. March/April Major flooding in Maximum depth > 1 km x 25 Connection to all other 2008 January; elevated 3.5 metres. Small metres. waterholes Nov – flows November flow during March. – March. sampling.

82 Shed

Shed is a large waterhole located on an outer eastern channel of the Cooper floodplain approximately 25 kilometres east of Windorah (Figures 4.1 and 4.38). The hydrological history of Shed waterhole (Table 4.11) is different from both Murken and Currareva as flows in the Barcoo (as opposed to the Thomson) River are generally required to inundate this site. Shed is generally > 2 kilometres long and up to 300 metres wide, and at the sampling site Shed was never shallower than 2 metres deep during the study.

Table 4.11 Hydrological history of Shed for the period September 2006 – March 2008. Date Antecedent flows Change in depth Connectivity Sep 2006 No flows N/A. Maximum depth 3 Isolated metres. Dec 2006 No flows Maximum depth 2.7 Isolated metres. Jan 2007 No flows Maximum depth 2 metres. Isolated April 2007 Flows in January. Maximum depth 3 metres. Connected to other waterholes in the eastern Cooper system in January and February. Isolated by April. August 2007 Small local flow in Maximum depth 3.2 Connected to other waterholes July. metres. in the immediate area briefly. Isolated by August. November No flows. Maximum depth 2.5 Isolated. 2007 metres. March/April Major flooding in Maximum depth >3 Isolated following flood 2008 January; elevated metres. recession in March. flows November – Connected to upstream and March downstream waterholes Nov – Mar.

83 4.1.4.4 Sites in the Kyabra catchment

Kyabra Creek is situated to the east of the main Cooper channels at Windorah and rises to the west of the township of Adavale (Figure 4.1). Unlike the Thomson and Barcoo rivers or Cooper Creek below the confluence of these rivers, Kyabra Creek did not experience major flooding in January 2008 or at any other time during the study (Bob Morrish, Springfield Station, personal communication). Flow gauging stations are absent from Kyabra Creek and the following information has therefore been collated through personal observation and the records of Bob Morrish, the owner of Springfield.

Springfield

Springfield is a large waterhole up to 5 kilometres long and 300 metres wide (Figure 4.39). During the current study the depth at the sampling site fluctuated by up to 1 metre depending upon the timing of antecedent within-channel flow events. The hydrological history of Springfield waterhole is given in Table 4.12.

Springfield South

Springfield South is a small ephemeral waterhole located immediately south of Springfield waterhole (Figures 4.1 and 4.40; Section 1.4.4.1). The hydrological history of Springfield South is summarised in Table 4.12.

One Mile

One Mile is situated a small distance south (upstream) of both Springfield and Springfield South waterholes (Figures 4.1 and 4.41). One Mile waterhole was never less than 500 metres long, 70 metres wide and 1 metre deep during the current study. The hydrological history of One Mile and all Kyabra Creek waterholes is summarised in Table 4.12.

84 Table 4.12 Hydrological history of all Kyabra Creek waterholes for the period September 2006 – March 2008. Date Antecedent flows Change in depth Connectivity Sep 2006 No flows N/A. All waterholes isolated Springfield - maximum depth 3 metres. Springfield South – 1.2 metres One Mile – N/A Dec 2006 No flows Springfield - maximum depth 2.7 All waterholes isolated metres. Springfield South – 0.7 metres. One Mile – 1 metre Jan 2007 Within-channel flow Springfield - maximum depth 3.3 Two week connection mid-December. metres. event (all waterholes) Springfield South – 1.2 metres. immediately following One Mile – 1.4 metres flow. April 2007 Within-channel flow Springfield - maximum depth 3.5 Three week connection January and February. metres. event (all waterholes) Springfield South – 1.6 metres. immediately following One Mile – 1.7 metres. flow. August 2007 No flows Springfield – maximum depth 3 Isolated metres. Springfield South – 1 metre. One Mile – 1.2 metres. November Connection flow Springfield – overnight rise to 3.2 Re-connected during 2007 during sampling. metres sampling. Springfield South – overnight rise to 1.2 metres. One Mile – overnight rise to 1.85 metres. March/April Connection flows Springfield – maximum depth 3.2 Isolated following flow 2008 occur in early summer metres. recession in January (November and Springfield South – 1.2 metres. 2008. December). One Mile – 1.3 metres.

85 4.1.5 Sites in the Bulloo catchment

The Bulloo River rises north of Adavale in southern central Queensland. The Bulloo catchment is a unique single-river endorheic catchment in Australia, and in wet years the river channel floods out into an area known as the Bulloo overflow south-west of . Rather than exhibiting an anastomosing channel system underlain by deep cracking clays (such as the greater Cooper, Diamantina and Georgina systems), the Bulloo generally has more defined channels (Figure 4.43). The Bulloo River experienced dry seasons in 2005 and 2006 (Figure 4.42; Vin Richardson, Leopardwood Park Station, personal communication), with water limited to deep waterholes situated within the main channel. Major flooding occurred in the Bulloo catchment in January 2007 and again in January 2008 (Figure 4.42). Following the recession of floodwaters the Bulloo had become a series of disconnected waterholes by March 2008.

250000

200000 Flood peak 21 -22 Jan 2008

150000 Flood peak 25 – 26 Jan 2007 ML/day 100000

50000

0 Jan 05 Jan 06 Jan 07 Jan 08

Figure 4.42 Daily discharge of the Bulloo River at Quilpie from January 2005 – May 2008. (Source: Queensland Department of Environment and Resource Management).

86 Bulloo Main Channel

The Bulloo Main Channel site is located approximately 30 kilometres upstream of the township of Adavale (Figure 4.1). The Bulloo Main Channel site comprises a deep section of river that is rarely dry and that is approximately 200 metres long, 15 metres wide and up to 2.5 metres deep for the majority of each year (Vin Richardson, Leopardwood Park station, personal communication). The site is surrounded by steep river banks and a canopy of river red gum and ghost gum (Figure 4.43). The hydrological history of the Bulloo Main Channel site (and other Bulloo sites) is given in Table 4.13.

Figure 4.43 The Main Channel site on the Bulloo River (left) and the Shearing Shed site (right) (August 2006).

87 Table 4.13 Hydrological history of all Bulloo River waterholes for the period April 2007 – March 2008. Date Antecedent flows Change in depth Connectivity April 2007 Major flood in January Main channel - maximum All waterholes connected depth 2.5 metres. during summer flows; Shearing Shed – 2.2 disconnected for 4 weeks at metres. time of sampling. House 2.2 metres. August 2007 No flows Main channel – maximum Isolated depth 2.2 metres. Shearing Shed – 1.9 metres. House – 1.9 metres. November Minor flooding 2 Main channel – maximum Connected during early 2007 weeks prior to depth 2.5 metres summer flows (November). sampling. Shearing Shed – 2.4 Disconnected during metres. sampling. House – 2.4 metres. March/April Major flooding in Main channel– maximum Isolated following flow 2008 January; elevated depth 2.5 metres. recession in February 2008. flows November – Shearing Shed – 2.5 March metres. One Mile – 2.5 metres.

Shearing Shed and House sites in the Bulloo catchment

Shearing Shed and House sites on the Bulloo River are large waterholes located approximately 3 kilometres from one another and 20 kilometres downstream from the Bulloo Main Channel site (Figures 4.1 and 4.43). Both sites are approximately 1 kilometre long, 100 metres wide and 2.5 metres deep at their deepest points when full. Both Shearing Shed and House sites in the Bulloo catchment are located on outer channels of the Bulloo, and consequently the banks are neither as steep nor as heavily timbered as the Main Channel site (Figure 4.43). The hydrological history of all Bulloo sites is given in Table 4.13.

88 4.2 Water quality at the sampling sites, 2006 – 2008

4.2.1 Methods

Water quality parameters including temperature, dissolved oxygen, pH, turbidity and salinity were measured at each site and on each sampling occasion between the hours of 3 and 5 pm when fish sampling equipment was deployed (see Chapters 5, 6 and 7). Measured water quality parameters were found to be unreliable in January 2007 due to faulty equipment and have therefore been omitted from the data and analyses (Appendix 2). Temperature, dissolved oxygen, pH and conductivity were tested using handheld TPS meters that were calibrated prior to each sampling trip, and turbidity was measured using a Secchi disc. With the exception of turbidity, all water quality parameters were measured at a depth of ten centimetres. In waterholes with a depth >2m, temperature and dissolved oxygen readings were also taken at 2 metres.

Water quality results were analysed using one-way analysis of variance (ANOVA) in SPSS Version 11.5 to test for differences in water quality parameters between fixed factors such as sampling times (September 2006, December 2006, April 2007, August 2007, November 2007, March/April 2008, November 2008), catchments (Mulligan, Georgina, Diamantina, Thomson, Barcoo, Cooper, Kyabra and Bulloo) and waterhole type. Waterhole type was defined as follows: Permanent within-channel – waterholes located within main channel or channels of a river valley that did not go dry between September 2006 and November 2008. Ephemeral within-channel – waterholes located within main channel or channels of a river valley that dried at least once between September 2006 and November 2008. Ephemeral floodplain – waterholes distant from main channel complexes that filled following flooding between September 2006 and November 2008.

Levene’s Test was used to satisfy assumptions of homogeneity, and statistically significant results were accepted at P≤0.05. In instances where water quality parameters were found to be significantly different with regard to either sampling time, catchment or waterhole type, pairwise results were further analysed using Tukey’s post hoc tests.

89 4.2.2 Results

Surface water temperature ranged from 12°C at the River site in the Bulloo catchment in August 2007 to 34.9°C at Springfield in the Kyabra catchment in December 2006. Mean water temperatures were significantly variable through time (df: 6; F = 64.248; p <0.001), with temperatures in August and September lower than all other sampling periods and temperatures in December generally higher (Table 4.14, Figure 4.44; Appendix 2). Water temperature at depth (2 metres) was lower than surface water temperature on all sampling occasions except August 2007 (Figure 4.45).

Table 4.14 Tukey’s post hoc test results for pairwise comparisons of surface water temperature between sampling periods. Non-significant results have been omitted from this table. Pairwise comparison p value September 2006 vs December 2006 <0.001 September 2006 vs August 2007 <0.001 September 2006 vs November 2007 <0.001 September 2006 vs November 2008 <0.001 December 2006 vs April 2007 <0.001 December 2006 vs August 2007 <0.001 December 2006 vs March/April 2008 <0.001 April 2007 vs August 2007 <0.001 April 2007 vs November 2007 0.036 April 2007 vs November 2008 0.014 August 2007 vs November 2007 <0.001 August 2007 vs March/April 2008 <0.001 August 2007 vs November 2008 <0.001

90 35

30

25

20

15

10

5 Water temperature (°C). Mean ± Mean Water (°C). temperature S.E.

0 Sep 06 Dec 06 Apr 07 Aug 07 Nov 07 Mar 08 Nov 08

Sampling event

Figure 4.44 Mean surface water temperature (± standard error) pooled from all sites sampled from September 2006 to November 2008.

35

30

25

20

15

10 Mean temperature (°C) ± (°C) S.E. temperature Mean 5

0 December 06 April 07 August 07 November 07 March 08 Sampling time

Figure 4.45 Mean (± standard error) water temperature at the surface (grey bars) and at a depth of 2 metres (black bars) at all sites from December 2006 to March 2008.

91 Dissolved oxygen within the top 10cm of water (‘surface’) varied through time (df: 6, F = 4.423, p < 0.001), by catchment (df: 7, F = 5.646, p < 0.001), and also depending upon waterhole type (df: 2, F = 12.091, p < 0.001). Dissolved oxygen ranged from 17.4% saturation in the Thomson Main Channel site in August 2007 to 162.2% saturation at Lake Idamea in the Georgina catchment during March/April 2008 (Appendix 2). Dissolved oxygen was always higher at ephemeral flooplain sites than in permanent or ephemeral main channel sites ( p = <0.001; Figure 4.46). In general, dissolved oxygen was significantly higher in the Georgina than in the other catchments (Table 4.15, Figure 4.47) and higher in April 2007 than at other sampling times (Table 4.15; Figure 4.48). Dissolved oxygen was consistently lower at depth (2 metres) than at the surface on all sampling occasions (Figure 4.49).

Table 4.15 Tukey’s post hoc test results for pairwise comparisons of surface dissolved oxygen results between catchments and sampling periods. Non-significant results have been omitted from this table. Fixed factor Pairwise comparison p value Catchment Georgina vs Bulloo <0.001 Georgina vs Kyabra 0.003 Georgina vs Thomson 0.002 Georgina vs Barcoo 0.016 Georgina vs Diamantina 0.035 Bulloo vs Cooper 0.029 Bulloo vs Mulligan 0.001 Sampling time April vs September 0.001 April vs November 2007 0.002

92 100

90

80

70

60

50

40

30

20

10

Mean dissolved oxygen (% saturation) ± saturation) (% oxygen dissolved Mean S.E. 0 Permanent main channel Ephemeral main channel Ephemeral floodplain

Waterhole type

Figure 4.46 Mean surface dissolved oxygen (% saturation) ± standard error in permanent main channel waterholes, ephemeral main channel waterholes and ephemeral floodplain waterholes/lakes pooled from all sites sampled across the study area, 2006 - 2008.

120

100

80

60

40

20 Mean dissolved oxygen (% saturation) ± saturation) (% oxygen dissolved Mean S.E. 0 Bulloo Kyabra Cooper Thomson Barcoo Diamantina Georgina Mulligan

Catchment

Figure 4.47 Mean surface dissolved oxygen (% saturation) ± standard error by catchment in the Lake Eyre and Bulloo-Bancannia basins, 2006 – 2008.

93 100

90

80

70

60

50

40

30

20

Mean dissolved oxygen (% saturation) ± saturation) (% oxygen dissolved Mean S.E. 10

0 Sep 06 Dec 06 Apr 07 Aug 07 Nov 07 Mar 08 Nov 08

Sampling times

Figure 4.48 Mean surface dissolved oxygen (% saturation) ± standard error from September 2006 to November 2008 at all sites in the Lake Eyre and Bulloo-Bancannia basins, 2006 – 2008.

90

80

70

60

50

40

30

20

10 Mean dissolved oxygen (% saturation) ± saturation) (% oxygen dissolved Mean S.E.

0 December 06 April 07 August 07 November 07 March 08

Sampling time

Figure 4.49 Mean (± standard error) dissolved oxygen at the surface (grey bars) and at a depth of 2 metres (black bars) at all sites from December 2006 to March 2008.

94 pH ranged from 5.4 at Native Waterhole in the Thomson catchment in August 2007 to 9 at a number of sites (Springfield South and One Mile in the Kyabra catchment in December 2006, Waterloo and Vergemont in the Thomson catchment in August 2007 and Lake Constance in the Diamantina catchment in March/April 2008). pH differences at catchment scale were minimal (df: 7, F = 1.403, p = 0.209), with most readings between 7 and 8 (Figure 4.50; Appendix 2).

8

7.5

7

pH. Mean ± Mean pH. S.E. 6.5

6 Bulloo Kyabra Cooper Thomson Barcoo Diamantina Georgina Mulligan

Catchment

Figure 4.50 Mean pH ± standard error by catchment in the Lake Eyre and Bulloo- Bancannia basins, 2006 – 2008.

Turbidity was significantly different between catchments (df: 7, F = 14.682, p<0.001), and water in the Georgina catchment was always far clearer than in any of the other catchments (Figure 4.51; Table 4.16). The most turbid water (Secchi depth 2cm) was recorded in the Diamantina catchment at Hunter’s Gorge and Lake Constance in November 2007 and at Lake Billyer in March/April 2008. The clearest water (Secchi depth 85cm) was recorded at Parapituri Waterhole in the Georgina catchment in November 2007 (Appendix 2).

95 45

40

35

30

25

20

15

Turbidity (cm). Mean ± Mean (cm). Turbidity S.E. 10

5

0 Bulloo Kyabra Cooper Thomson Barcoo Diamantina Georgina Mulligan

Catchment

Figure 4.51 Mean turbidity (Secchi depth in cm) ± standard error by catchment, 2006 - 2008.

Table 4.16 Tukey’s post hoc test results for pairwise comparisons of turbidity between catchments (n.s. = not significant). All pairwise comparisons not recorded in the table were non-significant. Pairwise comparison p value Georgina vs Bulloo, Kyabra, Cooper, Thomson, Barcoo, Diamantina <0.001 Georgina vs Mulligan n.s. Mulligan vs Bulloo n.s. Mulligan vs Kyabra 0.005 Mulligan vs Cooper <0.001 Mulligan vs Thomson 0.030 Mulligan vs Barcoo n.s. Mulligan vs Diamantina <0.001

Salinity was variable between catchments (df: 7, F = 2.673, p = 0.013) and was higher in the Georgina than in the other catchments (Figure 4.52). Within the Georgina catchment, salinity was also variable (Figure 4.52). Parapituri Waterhole consistently

96 returned the highest salinity levels in the Georgina catchment, and conductivity increased in this waterhole from 1020 µS/cm in April 2007 to 9180 µS/cm in August and 24000 µS/cm in November 2007 as the waterhole receded (Figure 4.52; Appendix 2).

4500

4000

3500

3000 2575.79 2500

2000

1500 1318.53

1000

500 Electrical conductivity (µS/cm). Mean (µS/cm). Electrical±Mean conductivity S.E. 177.59 139.61 149.43 204.58 208.39 47.08 0 Bulloo Kyabra Cooper Thomson Barcoo Diamantina Georgina Mulligan

Catchment

Figure 4.52 Mean conductivity (µS/cm) by catchment, 2006 – 2008.

4.2.3 Discussion

Annual surface water temperature in the Queensland Lake Eyre and Bulloo- Bancannia basins generally ranges from <15°C in winter to >30°C in summer. This suggests that all aquatic organisms that are extant in the study area generally have a high tolerance of annual temperature fluctuations. Thermal stratification of the water column occurs in all seasons except winter, an unsurprising result given the location of the sites in the arid and semi-arid zones and their exposure to solar radiation.

Dissolved oxygen was highest in formerly dry sites that filled following floodplain inundation. These results can be explained by the sudden influx of rainfall-derived

97 run-off that accumulated at such sites following flood events. High dissolved oxygen concentrations in April 2007 are similarly the result of a large amount of floodwater remaining at recently-inundated sites. Comparatively high dissolved oxygen concentrations in the Georgina catchment almost certainly occur due to the geomorphology of this catchment, as unlike all others in the study area, permanent waterholes in the Georgina catchment are underlain and consistently replenished by subterranean aquifers (Shane McGlinchey, Badalia Station; Stephen Bryce, Glenormiston Station; John Bryant, Roxborough Downs Station; Pat Fennell, formerly Linda Downs Station, personal communications).

Both turbidity and conductivity were noticeably different in the Georgina and Mulligan catchments than in the more easterly sites for the duration of the study. In both catchments, certain sites (most obviously Parapituri in the Georgina but also Pulchera in the Mulligan) exhibited a pattern of increasing salinity as they entered a drying phase following flooding in January and February 2007. It seems likely that dilution of saline waterholes occurred following flooding and that conductivity then began increasing as the waterholes receded. In the case of Parapituri – the most saline waterhole – this increase was particularly noticeable, as conductivity increased from 1020 µS/cm to 24 000 µS/cm in only seven months (April – November 2007). These results therefore indicate that aquatic biota in the Georgina catchment have comparatively wide tolerances to salinity fluctuations. In comparison, all catchments east of the Georgina recorded comparatively low conductivity for the duration of the study as few waterholes registered a figure higher than 300 µS/cm (Appendix 2). Similar increases in electrical conductivity have been noted in the Cooper catchment during prolonged dry periods (Sheldon and Fellows in press ). The most turbid water was consistently sampled in the Diamantina catchment, where visibility within the water column was frequently limited to 2 or 3 centimetres. In the neighbouring Georgina catchment, and also in the Mulligan, turbidity readings of up to 75cm were recorded, and this again is most likely attributable to the increased salinity in these catchments (Appendix 2).

Waterholes in all studied catchments can be characterised by high temperature variation in relation to season, high dissolved oxygen variation in relation to flooding and rainfall and relatively constant neutral-tending-alkaline pH values. All sites in the

98 Bulloo, Cooper, Thomson, Barcoo and Diamantina catchments have low electrical conductivity and are highly turbid, whereas sites in the Georgina and Mulligan catchments often exhibit low turbidity and high conductivity. It should be noted that, though many sites in the Georgina catchment are spring-fed, the salinity of source aquifers is variable, and consequently some waterholes are far more saline than others (Shane McGlinchey, Badalia Station; Stephen Bryce, Glenormiston Station; John Bryant, Roxborough Downs Station; Pat Fennell, formerly Linda Downs Station; Jenny Silcock, Queensland, Department of Environment and Resource Management personal communications).

4.3 Summary of the sampled catchments 2006 - 2008: aridity, season, hydrology and water quality.

The catchments studied range from the Bulloo in the east, where comparatively high flows are likely to occur between November and May, to the Georgina in the west, where flows are most likely to occur only between January and March (see Chapter 2). Additionally, the number of zero-flow days per year is historically far higher in the Georgina, and increases along a west-east gradient to the Bulloo (see Chapter 2). The four main catchments (Bulloo, Cooper, Diamantina and Georgina) therefore evince increasing aridity from east to west, and represent a suitable spatial scale within which the main hypotheses of this study relating to the distribution, recruitment and movement of fish can be tested (see Chapters 1, 5, 6 and 7).

Mean daily temperatures across all studied catchments are above 25°C between the months of September and April (Chapter 2), however flow is most likely to occur in all catchments from January to April (Chapter 2). This suggests that dividing the calendar year into three (as opposed to four) seasons may be most appropriate in order to address the aims of the study, with the ‘seasons’ best described as late summer (January to April, when flows/floods are most likely), winter (May to August, when temperatures are low but flow is highly unlikely to occur) and early summer (September to December, when temperatures are high but flow is unlikely). Accordingly, sampling during the current study was undertaken on three occasions

99 during late summer (January 2006, April 2007 and March/April 2008), on three occasions during early summer (September 2006, December 2006 and November 2007) and once during winter (August 2007) (Table 4.17). The time constraints associated with completing the sampling program and associated analysis and written work precluded further sampling during winter periods.

Table 4.17 Fish sampling occasions and antecedent flows in the Lake Eyre and Bulloo-Bancannia basins between September 2006 and March/April 2008, with notes on salinity levels.

Flow in the Season preceding 3 Late summer Winter Early summer months (Jan – Apr) (May – Aug) (Sep – Dec) Overbank to major Mulligan (April 07) flooding Georgina (April 07) Thomson (March 08) Barcoo (March 08) Cooper (March 08) Bulloo (April 07) Bulloo (March 08) Within-channel Diamantina (April 07) Thomson (August 07) Diamantina (Nov 07) connection flows Diamantina (March 08) Cooper (August 07) Thomson (Nov 07) Thomson (April 07) Barcoo (September 06) Barcoo (April 07) Barcoo (Nov 07) Cooper (January 07) Cooper (Nov 07) Cooper (April 07) Kyabra (Nov 07) Kyabra (January 07) Bulloo (Nov 07) Kyabra (April 07) Kyabra (March 08) No flows Georgina (March 08) Mulligan (August 07)* Mulligan (Nov 07)* Thomson (January 07) Georgina (August 07)* Georgina (Nov 07)* Kyabra (December 06) Diamantina (Aug 07) Thomson (Sept 06) Barcoo (August 07) Thomson (Dec 06) Kyabra (August 07) Cooper (Sept 06) Bulloo (August 07) Cooper (Dec 06) Kyabra (Sep 06) * Salinity increases in these catchments as water levels drop during a prolonged drying phase.

100 Hydrology within the sampled catchments during the study included periods of no flow for up to three months in all sampled catchments, periods following within- channel flows in all catchments except the Georgina and Mulligan and periods following major flooding in the Georgina/Mulligan, greater Cooper and Bulloo catchments (Table 4.17). The sampling events therefore represent an adequate spread of antecedent flow conditions within and between seasons from which to draw conclusions regarding the main hypotheses relating to the distribution, recruitment and movement patterns of fish in these rivers (see Chapter 1). The timeframe of the current study was necessarily brief due to the constraints of post-graduate study and the requirement that doctoral work is completed within a timeframe as close as possible to three years. Despite this comparatively short timeframe, samples were procured following major flooding and periods of within-channel flows and no flows during different seasons from several catchments (Table 4.17). Although samples were not available from any catchments in either winter or early summer following major flooding in the preceding three months (Table 4.17), the historical hydrology of all catchments indicates that sampling over a longer time period would deliver similar results, as major flooding almost always occurs in central Australia in late summer (specifically January to March) (see Chapter 2).

Water quality parameters such as dissolved oxygen, pH and temperature displayed little variation across the studied catchments. Spatially, the western rivers (the Georgina and Mulligan) were always less turbid than the rivers in the Diamantina, greater Cooper and Bulloo catchments (this chapter). Sampling along an extended timeframe demonstrated that salinity is likely to increase in the Georgina catchment (and to a lesser extent the Mulligan) as waterholes enter a drying phase (Figure 4.17).

101 5. The distribution of fish in the Queensland Lake Eyre and Bulloo-Bancannia basins

5.1 Introduction

The distribution of living things on earth is regulated by processes operating at a multitude of spatial and temporal scales ranging from prehistoric climatic conditions and changes (Nanson et al. 2008) to local colonisation and extinction events (Fairfax et al. 2007). During the modern era, these processes have been further impacted by human-induced perturbations to natural systems such as the domestication of plants and animals, hunting pressure and the translocation of species around the globe (Dudgeon et al. 2006). Introductions of biota and habitat degradation affect ecosystems world-wide and continue to change the species composition of communities within them (Box et al. 2008; Humphries and Winemiller 2009). Consequently, maintenance of species richness is an important theme in conservation biology (Amarasinghe and Welcomme 2002; Dudgeon et al. 2006).

Accurately mapping the distribution of species is a fundamental step in the process of detailing the drivers of recruitment and persistence and identifying potential threats to populations such that management and conservation strategies can be developed. Unfortunately, in the case of many recently-extinct and/or endangered species, this fundamental step was either never taken or was taken too late (Quammen 1996). There can be little doubt, for example, that efforts to conserve the populations of extinct mammals such as Steller’s sea cow, Ectopistes migratorius, and the quagga, Equus quagga quagga , might have been commenced if the dwindling distributions of such species had been known. In south-eastern Australia, the river systems of the Murray-Darling Basin are a good example of a region where large-scale surveying began after the introduction of alien fish species and the regulation of river flows (Llewellyn 1983; Harris and Gehrke 1997; Humphries and Winemiller 2009), thus possibly obscuring endemic species’ original ranges. Additionally, species with a fragmented distributional range are often susceptible to higher extinction risk (Fagan et al. 2002; Hughes et al . 1999), and therefore establishing the range of these species

102 is fundamental to the development of recovery plans and extinction amelioration strategies (Knight and Arthington 2008). As examples, fishes from the Sonoran Desert in Mexico and the United States with the most fragmented historic distributions have been demonstrated to be the most likely to suffer local extirpations (Fagan et al. 2002), and highly range-limited Australian fish species, such as those from the spring complex at Edgbaston in western Queensland, have similarly suffered local extinctions (Fairfax et al. 2007).

Establishing the current geographical range of fish species is difficult due to factors such as their high vagility, long migration pathways and the occasionally massive geographical areas that species may inhabit. This is most obvious in the case of oceanic and coastal species, where established ranges frequently cross continental borders and species new to science continue to be discovered. Investigating the distribution of fish species with a life history completed in freshwater is slightly less problematic, but is still contingent upon the completion of adequate surveys at appropriate spatial and temporal scales (Labbe and Fausch 2000; Knight and Arthington 2008). The possession – or at least existence – of historical records enables the impacts of anthropogenic changes to be assessed, yet in arid Australia – both the remote Lake Eyre and the comparatively populated Murray-Darling basins - these records are recent and patchy, generally dating only from the 1970s (Glover and Sim 1978; Glover 1979; Glover 1982; Llewellyn 1983). In contrast, from examination of long-term datasets, habitat alteration, such as flow reduction, has been demonstrated to contribute to species losses from arid-zone streams in Kansas (Cross and Moss 1987). Consequently, current species distribution guides that rely on a combination of scientific surveys and anecdotal records will certainly require revision as more work in remote and/or inaccessible areas is completed (Allen et al. 2002; Pusey et al. 2004), and filling the existing knowledge gaps regarding the distribution of Australian freshwater fish should be a priority for management agencies nationwide. This situation is especially acute in far western Queensland, the state where the headwaters of all the major Lake Eyre Basin rivers (the Cooper, Diamantina and Georgina), as well as the Bulloo, are located.

Conducting fish survey work in outback Australia can be both difficult and expensive because of the ephemeral nature of waterholes, the large distances involved and the

103 frequently high daytime temperatures. Consequently, central Australia generally remains poorly studied with regard to fish distributions and aquatic communities (Box et al. 2008). Although fish species lists have been generated for the major catchments these are frequently based on single-occasion sampling (Bailey and Long 2001) and have not been conducted in wholly ephemeral catchments. There are, of course, sound reasons for this deficiency that are principally related to the expense of conducting fish surveys in ephemeral arid environments: surveys conducted in comparatively large, relatively permanent waterholes are more likely to yield records.

In Queensland, the Cooper Creek catchment in the general vicinity of the township of Windorah has been studied more intensively than either the Diamantina or Georgina catchments and data from this area has contributed to published scientific papers associated with the Dryland Refugium project (Arthington et al. 2005; Balcombe et al. 2007; Balcombe and Arthington 2009). During the ARIDFLO surveys (Costelloe et al. 2004), sites in the Queensland Diamantina and Cooper catchments were studied, as well as sites from the Cooper and Diamantina in South Australia and the in the south-western section of the Lake Eyre Basin. A 1995 Queensland Department of Primary Industries survey included sites in both the Thomson River (Cooper catchment) and Diamantina catchment (Long and Humphery 1995), however the only survey at Queensland sub-Basin scale (Cooper, Diamantina and Georgina catchments) is a report by the Queensland Department of Natural Resources and Mines (Bailey and Long 2001). A basin-wide identification text has also been published by the Queensland Department of Primary Industries (Wager and Unmack 2000). Nevertheless, the necessarily restricted spatial scale of all existing work completed in the Queensland Lake Eyre Basin suggests that knowledge of fish distribution within these arid catchments should be considered fragmentary, and this is most notable in the Georgina catchment. This paucity of data can be considered extreme compared with the extensive data sets that have been compiled over many years for fish communities in arid North America (Eby et al. 2003).

The results of existing surveys and studies confirm the presence of nine native species throughout the Queensland Lake Eyre Basin, a further three confined to the greater Cooper Creek catchment and at least one confined to the Georgina catchment (see

104 Chapter 3). Alien fish species present in the Queensland Lake Eyre Basin include gambusia, Gambusia holbrooki and goldfish, Carassius auratus , with the distribution of both species restricted to the Cooper Creek catchment. Translocated fish in the Queensland Lake Eyre Basin are known to include Murray cod, Maccullochella peelii peelii and Murray-Darling golden perch, Macquaria ambigua (Bailey and Long 2001) and may also include silver perch , Bidyanus bidyanus (Wager and Unmack 2000). The distribution of stocked Murray-Darling golden perch in the Lake Eyre Basin is unlikely to be revealed due to the morphological similarity of this species to the endemic sub-species. Murray cod, originally stocked in the late 1990s, are occasionally caught by recreational anglers close to the original release point in the Thomson River (Vanessa Bailey, Queensland Department of Environment and Resource Management, personal communication). Silver perch are morphologically similar to Welch’s grunter, and superficially similar to Barcoo grunter, but their distribution and/or presence in the Lake Eyre Basin is currently unknown. 5

The fish community of the Bulloo-Bancannia Basin, situated immediately east of the Lake Eyre Basin in western Queensland, is known only from surveys carried out in 1986 and 1989 and reported by Midgley et al. (1991). This report confirms the presence of six species in the Basin, including bony bream, an unidentified catfish, (Neosiluris sp.), a gudgeon, ( Hypseleotris sp.), desert rainbowfish, golden perch, spangled perch and Welch’s grunter. In addition, museum records include a single specimen of purple-spotted gudgeon, Mogurnda adspersa (Queensland Museum) and the alien mosquitofish, Gambusia affinis (Australian Museum) as also occurring in the Bulloo-Bancannia Basin (Midgley et al. 1991).

Though the Bulloo-Bancannia Basin is not technically part of the Lake Eyre Basin, recent studies suggest a strong genetic affinity between the two areas that is identifiable in aquatic biota ranging from crayfish (Hughes and Hillyer 2003) to mussels (Hughes et al. 2004) and fish (Faulks 2009). Indeed, Wager and Unmack (2000) include records from the Bulloo-Bancannia Basin in a government text relating to fish of the Lake Eyre Basin. The species mentioned in this publication as occurring within the Bulloo-Bancannia catchment include those mentioned above, with certain

5 Wager and Unmack (2000) suggest that silver perch have been translocated to two areas the Lake Eyre Basin..

105 qualifications. Wager and Unmack (2000) list three species of catfish, including Hyrtl’s tandan, silver tandan and a third undescribed species, the false-spined catfish Neosiluris sp., known from only two specimens. They list the golden perch (yellowbelly) occurring in the Bulloo as Macquaria sp. rather than Macquaria ambigua , and indicate that this species – or sub-species – is more likely to be related to the Lake Eyre golden perch than to M. ambigua from the Murray-Darling Basin. A recent study generally dispels the delineation between Bulloo and Lake Eyre yellowbelly, considering them part of the same Evolutionary Significant Unit (Faulks 2009). Wager and Unmack (2000) record two possible species of carp gudgeons from the Bulloo-Bancannia – the western carp gudgeon, Hypseleotris klunzingeri, and the undescribed Midgley’s carp gudgeon, Hypseleotris sp., and in addition to Welch’s grunter, also record Barcoo grunter from the Bulloo-Bancannia Basin.

In general, past studies suggest that the aquatic fauna of the Bulloo-Bancannia is likely to be a subset of the species from the nearby Thomson/Barcoo/Cooper catchment in the adjacent Lake Eyre Basin, with notable omissions being Australian smelt and Cooper Creek catfish. The extant Bulloo-Bancannia fish populations are therefore aligned not with the north-western rivers of the Murray-Darling Basin (to the east) but with the catchments of western Queensland within the Lake Eyre Basin. Currently, the Bulloo catchment is managed in conjunction with the north-western Murray-Darling rivers (the Warrego and Paroo) rather than with the Lake Eyre Basin rivers. Comprehensive surveying of fish communities throughout the Bulloo- Bancannia Basin is needed, particularly as this catchment is geographically – and, presumably, biogeographically – intermediate between the Murray-Darling and Lake Eyre Basins.

Fish survey work in the Lake Eyre and Bulloo-Bancannia basins currently presents a generally cohesive if fragmentary audit of these areas, and details a core group of at least seven species with a range extending across the rivers of both basins. Nevertheless, anomalies in the current data exist, and point to a discrepancy in many instances between the scientific and grey literature. In the Bulloo catchment, records of false-spined catfish, Neosiluris sp., are restricted to Wager and Unmack (2000), and only a single record of purple-spotted gudgeon, Mogurnda adspersa , exists (Jeff Johnson, Queensland Museum, personal communication). Additionally, the presence

106 of Northwest ambassis (glassfish) is recorded in this catchment by Allen et al. (2002) but not by Wager and Unmack (2000) or Midgley et al. (1991). Further west, golden goby, Glossogobius aureus , are reported to be present in the Georgina catchment by Wager and Unmack (2000) and are also mentioned in Pusey et al. (2004), however the only published government report that included the Georgina catchment did not record this species (Bailey and Long 2001). These records indicate that increased sampling effort is required throughout the Lake Eyre and Bulloo-Bancannia basins in order to gain a more accurate understanding of existing fish distributions. This is especially important in Queensland, the Australian state where populations of locally endemic species such as Cooper Creek catfish and golden goby are most likely to be extant and/or abundant due to the number of permanent waterholes present within all three of the major Lake Eyre Basin catchments (Silcock 2009).

It is important to note that with the exception of the surveys undertaken as part of the Dryland Refugium project (Arthington et al. 2005; Balcombe et al. 2007; Balcombe and Arthington 2009), the ARIDFLO project (Costelloe et al. 2004) and the notable body of work produced in South Australia by Puckridge (1999), the majority of studies of fish communities in the Lake Eyre Basin have been single-occasion surveys with no long-term spatial and temporal dimension. Additionally, existing surveys have concentrated on large waterholes on the larger rivers, such as the greater Cooper (predominantly Thomson/Cooper) and Diamantina drainages and – to a much lesser extent - the more remote Georgina drainage (Bailey and Long 2001). The comparatively recent surveys of the mostly ephemeral Neales River by ARIDFLO (Costelloe et al. 2004) and Kyabra Creek by the Dryland Refugium team (Arthington et al. 2005; Balcombe and Arthington 2009) represent the first inclusions of small, hydrologically-isolated catchments in comparative studies with a temporal dimension.

In addition to being the only Lake Eyre Basin studies incorporating an extended temporal scale, the work reported by Puckridge et al . (2000), Costelloe et al. (2004), Arthington et al. (2005), Balcombe et al. (2007) and Balcombe and Arthington (2009) are also unique due to their consideration of the influence of flow regime and recent flow history on fish distributions and patterns of assemblage structure. Given that Lake Eyre Basin rivers have been demonstrated to possess some of the most variable flow regimes on earth (Puckridge et al . 1998), and that Australia’s indigenous arid-

107 zone fauna are adapted to surviving in rivers characterised by low flow conditions for the majority of the time (McMahon and Finlayson 2003; Arthington et al . 2005), these studies are valuable in that their sampling designs seek to consider the ecological effects of space, time and flow on fish communities. Nevertheless – and again due to the difficulties and expense associated with conducting research in outback Australia – studies that have incorporated Queensland catchments are characterised by design omissions. ARIDFLO, conducted at Basin scale, considered sites in the Cooper and Diamantina at both up and downstream locations (as well as the remote Neales River in South Australia), but did not consider the Georgina, the third largest catchment in the Lake Eyre Basin. Although the Dryland Refugia project was primarily concerned with comparing variation between and within catchments at reach scale in the Cooper Creek, Warrego and Border-Rivers catchments, this study did not include sites in the Barcoo River, despite the Barcoo’s secondary role (behind the Thomson) of providing flow and floodwaters to the downstream Cooper, both in Queensland and into South Australia.

The work reported by both ARIDFLO (Costelloe et al. 2004) and Dryland Refugium (Arthington et al. 2005; Balcombe et al. 2007; Balcombe and Arthington 2009) build on the prior studies of Puckridge et al. (2000) and demonstrate that fish distribution and abundance in the Lake Eyre Basin are influenced by both antecedent flow history and seasonal factors such as temperature. Floods, or a series of high flows over subsequent years, result in greater catch-per-unit-effort, population peaks and range extensions, whereas samples taken in winter frequently display reduced abundances and suggest fish inactivity during cold periods.

The current study has retained the temporal scale approach instigated by Puckridge (1999), ARIDFLO (Costelloe et al. 2004) and the Dryland Refugium project (Arthington et al. 2005; Balcombe et al. 2007; Balcombe and Arthington 2009), and aims to supplement the preceding research in the Lake Eyre and Bulloo-Bancannia basins by investigating the distribution of fish at catchment and sub-Basin (Queensland only) scale by incorporating the following design elements: 1. Inclusion of the Georgina River catchment as well as the Diamantina and Cooper drainages.

108 2. Inclusion of sites spread throughout the greater Cooper catchment, including both the Barcoo and Thomson rivers as well as the Cooper itself below the confluence of the major feeder rivers. Sites in Kyabra Creek, a small catchment situated to the east of the main Cooper channels near Windorah and first surveyed by the Dryland Refugia project, have also been retained in this study. 3. Inclusion of the Bulloo River, situated in the endorheic Bulloo-Bancannia Basin immediately east of the Lake Eyre Basin in Queensland. 4. Inclusion of the Mulligan River, an ephemeral desert catchment situated within the eastern dunefields of the Simpson Desert.

Based on these design elements and the incorporation of extended spatial and temporal sampling, this section of the study addresses the following hypotheses:

1. Fish communities will exhibit spatial differences due to catchment barriers. Specifically: a) The fish communities in the Bulloo catchment will be different from all others as it is an endorheic catchment separated from the Lake Eyre Basin rivers. b) The fish communities in the greater Cooper catchment (Thomson, Barcoo, Cooper and Kyabra sub-catchments) will be similar to each other, but different from those of all other catchments. c) The fish communities in the Diamantina and Georgina catchments will be similar to each other as the rivers join sporadically at Goyder’s Lagoon, but different from the communities in the Bulloo and greater Cooper catchments. d) The fish communities in the ephemeral Mulligan catchment are likely to be the same, or a sub-set, of those in the Georgina, the Mulligan’s parent catchment. 2. Fish communities will exhibit temporal variability associated with sampling time/season, and species richness will be lower in winter. 3. Fish presence/absence will exhibit variability depending upon antecedent hydrology, with species richness increasing following flooding.

109 4. Fish presence/absence patterns will exhibit variability between permanent and ephemeral habitats (waterholes).

110 5.2 Methods

Studies relating to presence/absence of fish species were conducted in waterholes of the Bulloo-Bancannia and Lake Eyre basins as described in Chapter 4. A minimum of three waterholes were sampled in each catchment during each sampling period with the exception of the Mulligan catchment, where waterholes evaporated quickly following a filling event in January 2007. Consequently, three sites were sampled in the Mulligan catchment in April 2007, two in August 2007 and one in November 2007. The Mulligan River had dried completely by March/April 2008. In November 2008, samples were taken opportunistically at northern sites in the Thomson, Diamantina and Georgina catchments with the express purpose of gaining extra distributional information. Consequently, these results have been omitted from all temporal analyses as they were only sampled once. Late in 2009, as this thesis was being prepared, opportunistic fish sampling also occurred in the Mulligan catchment. These results have not been presented and are mentioned only if they are particularly relevant to the main dataset (see Appendix 5).

5.2.1 Field methods

Fish populations were sampled in all waterholes and on all sampling occasions using a combination of three methods; large fyke nets, small fyke nets (also called glass eel nets) and a manually-dragged larval trawl net. Large double-winged fyke nets with a 13mm stretched mesh and 8 metre wings were set parallel to the bank with their openings facing in opposite directions upstream and downstream from a central post. Cod-ends were secured above the water surface in order to allow air-breathing vertebrates to survive if they became entrapped. Small fyke nets with a stretched mesh of 2mm and a wing width of 3 metres were set in an identical manner. All fyke nets were set in the afternoon (as close as possible to 2 pm) and retrieved the following morning (as close as possible to 9 am). At each site a larval trawl net was also dragged through the water for 5 minutes. A more detailed discussion of small fyke net and larval trawl sampling techniques is provided in Appendix 1. Following the clearing of fyke nets and larval trawl nets all fish were held in shaded water-filled

111 buckets prior to processing. Fish species were identified using a combination of published literature relating to fishes of arid Australian rivers (Wager and Unmack 2000; Allen et al. 2002), however species identification of carp gudgeons, Hypseleotris spp., was not attempted and their records were pooled under this genus. All sampled fish were measured from the tip of the snout to the caudal peduncle in order to obtain a standard length (SL) measurement for each individual specimen. Following identification and measurement native species were returned to the water alive and alien species were euthanased using a dilute oil of cloves solution followed by refrigeration. All sampling and euthanasia were carried out under General Fisheries Permit (No: PRM03315D) issued by the Queensland Department of Primary Industries and under a Griffith University Ethics Agreement (AES/09/06/AEC). Permission to sample at all sites was sought and obtained from station owners and the Queensland Environmental Protection Agency (now Queensland Department of Environment and Resource Management).

5.2.2 Data analysis

Data relating to each fish species at each site on each sampling occasion was combined for the three sampling methods used in order to calculate catch-per-unit- effort (CPUE). All fyke net samples were standardised to a 19 hour set time (as per Arthington et al. 2005) and larval trawls were standardised to 5 minutes at each site on each sampling occasion. Subsequent analysis of fish communities was performed on samples taken in April 2007, August 2007, November 2007 and March/April 2008 for fish communities in the Bulloo, Diamantina and Georgina systems. In the Mulligan system, all waterholes were dry prior to March/April 2008 and consequently analysis could be performed only on samples from April, August and November 2007. In most greater Cooper catchments (Thomson, Cooper and Kyabra catchments), data was available over a longer timeframe and included samples taken in September 2006, December 2006, January 2007, April 2007, August 2007, November 2007 and March/April 2008. Due to difficulties associated with storm conditions and road access, sites in the Barcoo catchment were not sampled in either December 2006 or January 2007.

112 Bray-Curtis similarity matrices (Bray and Curtis 1957) were constructed using CPUE totals transformed for presence/absence using PRIMER-E Version 5. Ordination analyses (Clarke 1993) were performed using hybrid non-metric multi-dimensional scaling in PRIMER-E Version 5 in order to identify obvious patterns of similarity among fish communities in catchments and in relation to factors such as season (or sampling time), antecedent flow and waterhole type. One-way analysis of similarities (ANOSIM) was then used to test for the influence of catchment, flow, season and waterhole type on fish communities using the same Bray-Curtis matrices. Results from ANOSIM calculate a test statistic ‘R’ identifying the observed differences between treatments compared with the differences among replicates within treatments (Clarke and Warwick 1994).

To test for the influence of season, sampling times were categorised as either early summer (September to December), late summer (January to April) or winter (May to August). To test for the influence of flow, antecedent hydrology (the 3 months prior to sampling) was categorised as major flooding, minor to moderate flooding, within- bank connection flows or no flows. Flooding was defined as per the definitions of the Bureau of Meteorology (2008), however the following explanations clarify these classifications for the purposes of the study. 1. Major flooding: Overbank flooding causing inundation of all previously dry floodplain areas and tributaries. 2. Minor to moderate flooding: Overbank flooding causing inundation of some (but not all) previously dry floodplain areas and tributaries. 3. Within-bank connection flow: A flow that occurs within a river channel, linking waterholes and filling some previously dry areas such as backwaters and anabranches. 4. No flows: Zero flow in the preceeding three months To test for the influence of waterhole type, waterholes were categorised as either permanent within-channel (waterholes situated in the main channel of rivers that did not dry during the study), ephemeral within channel (waterholes situated in the main channel of rivers that dried at least once during the study) or ephemeral lakes (floodplain waterholes that filled and dried at least once within the study period). In instances where ANOSIM revealed significant pairwise differences between fish communities explained by catchment, season, flow or waterhole type, SIMPER analysis in PRIMER-E Version 5 was used to calculate the average dissimilarity

113 between paired samples and allocate the contribution each species made to this dissimilarity (Clarke and Warwick 1994).

114 5.3 Results

5.3.1 Fish species presence/absence

Seventeen native and alien fish species were sampled in eight catchments in far western Queensland from September 2006 to November 2008 (Table 5.1). Bony bream, silver tandan, desert rainbowfish and spangled perch were the most widely distributed species and occurred in all catchments (Table 5.1). Hyrtl’s tandan, glassfish, yellowbelly and Barcoo grunter were recorded from seven of the eight sampled catchments, and Welch’s grunter were recorded from six (Table 5.1). The remaining species had more limited distributions, with Cooper Creek catfish collected from the Thomson, Barcoo and Cooper catchments, Australian smelt sampled from all greater Cooper catchments (Thomson/Barcoo/Cooper/Kyabra), banded grunter sampled from the Georgina and Mulligan catchments, golden goby sampled from the Georgina and Diamantina catchments and carp gudgeons sampled from the Bulloo and greater Cooper catchments (Table 5.1). A single sleepy cod was sampled in the Thomson catchment (Table 5.1). Alien fish species were confined to the greater Cooper catchment during the study, with goldfish recorded from the Thomson and Barcoo catchments and gambusia recorded from the Cooper downstream of the Thomson/Barcoo confluence (Table 5.1). The lowest number of species was recorded from the Mulligan catchment (7) and the highest from the Thomson catchment (14) (Table 5.1). Fish were not present at two sites in the Mulligan catchment, including Kunnamuka Swamp, an ephemeral lake that filled following flooding in early 2007, and Ocean Bore (see Chapter 4).

115 Table 5.1 Fish species presence/absence in sampled catchments of the Queensland Lake Eyre and Bulloo-Bancannia basins from 2006 - 2008. Percentages represent the frequency of species presence in relation to the total number of sites sampled in each catchment (eg: 100% = present at every site, smaller percentages = present at fewer sites). Empty areas indicate the species was absent from the catchment.

Species Catchment (total number of sites sampled 2006 – 2008) Mulligan Georgina Diamantina Thomson Barcoo Cooper Kyabra Bulloo (6) (12) (14) (27) (19) (21) (24) (12) Nematolosa erebi 83.3 83.3 78.6 100 89.5 100 100 100 Bony bream Neosiluroides cooperensis 44.4 26.3 33.3 Cooper Creek catfish Neosiluris hyrtlii 75 42.9 62.9 47.4 80.9 58.3 66.6 Hyrtl’s tandan Porochilus argenteus 83.3 50 71.4 74 57.9 85.7 91.6 91.6 Silver tandan Retropinna semoni 70.8 33.3 26.3 33.3 Australian smelt Melanotaenia splendida tatei 50 91.6 35.7 51.9 42.1 23.8 75 58.3 Desert rainbowfish Ambassis sp. 83.3 100 22.2 36.8 9.5 37.5 83.3 Glassfish Macquaria sp. 41.6 100 100 100 95.2 95.8 75 Yellowbelly Amniataba percoides 16.6 83.3 Banded or Barred grunter Bidyanus welchi 8.3 7.1 22.2 10.5 28.6 25 Welch’s grunter Leiopotherapon unicolor 100 50 42.9 77.8 42.1 42.9 87.5 100 Spangled perch Scortum barcoo 33.3 16.6 21.4 29.6 26.3 47.6 8.3 Barcoo grunter Glossogobius aureus 75 7.1 Golden goby Hypseleotris sp. 59.3 47.4 19 83.3 58.3 Carp gudgeon Oxyeleotris lineolatus 3.7 Sleepy cod Carassius auratus 7.4 10.5 Goldfish Gambusia holbrooki 14.3 Gambusia or Mosquitofish TOTAL NUMBER OF 7 11 9 14 13 13 11 8 SPECIES

116 Multivariate analysis of fish presence/absence throughout the Queensland Lake Eyre and Bulloo-Bancannia basins indicate that sites in catchments such as the Mulligan (yellow symbols), Georgina (green symbols) and to a lesser extent the Bulloo (grey symbols) form distinct groups and support different fish communities from sites in the Diamantina and greater Cooper catchments (Figure 5.1). Sites in the Diamantina and greater Cooper catchments are less differentiated and occupy the central portion of the ordination space (Figure 5.1). Although there is no clear grouping of sites sampled at particular times, winter samples (August 2007) often present as outliers in the ordination (Figure 5.1).

Stress 0.2

Figure 5.1 Two Dimensional NMS ordination plot of fish communities transformed for presence/absence across four sampling periods in seven catchments. Triangles = April 2007, squares = August 2007, diamonds = November 2007 and circles = March/April 2008; Grey = Bulloo catchment, pink = Kyabra sub-catchment, red = Cooper sub-catchment, orange = Thomson sub-catchment, purple = Barcoo sub- catchment, blue = Diamantina catchment, green = Georgina catchment and yellow = Mulligan sub-catchment.

117 Analysis of Similarities (ANOSIM) indicates that the species composition of fish communities in most catchments in the Queensland Lake Eyre and Bulloo-Bancannia basins are significantly different from one another (Table 5.2). Exceptions are the sub- catchments within the greater Cooper system: Kyabra Creek versus the Thomson River, Cooper Creek versus the Thomson and Barcoo Rivers and the Thomson River versus the Barcoo River (all non-significant; Table 5.2). It should be noted that these four sub-catchments are periodically hydrologically connected. Differences in fish species presence/absence can also be explained by sampling season (or time), antecedent flow regime and waterhole type (Table 5.2).

Table 5.2 Summary of One-Way ANOSIM results comparing fish presence/absence throughout the Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008. All transformations presence/absence. Factor Global R P Significant pairwise tests Catchment 0.314 0.001 Bulloo River vs Kyabra Creek (0.029) Bulloo River vs Cooper Creek (0.001) Bulloo River vs Thomson River (0.004) Bulloo River vs Barcoo River (0.005) Bulloo River vs Mulligan River (0.001) Bulloo River vs Georgina River (0.001) Bulloo River vs Diamantina River (0.002) Kyabra Creek vs Cooper Creek (0.001) Kyabra Creek vs Barcoo River (0.011) Kyabra Creek vs Mulligan River (0.001) Kyabra Creek vs Georgina River (0.001) Kyabra Creek vs Diamantina River (0.001) Cooper Creek vs Barcoo River (0.032) Cooper Creek vs Mulligan River (0.001) Cooper Creek vs Georgina River (0.001) Cooper Creek vs Diamantina River (0.001) Thomson River vs Mulligan River (0.001) Thomson River vs Georgina River (0.001) Thomson River vs Diamantina River (0.002) Barcoo River vs Georgina River (0.001) Barcoo River vs Diamantina River (0.024) Mulligan River vs Diamantina River (0.008) Georgina River vs Diamantina River (0.001) Season 0.064 0.035 Late summer vs winter (0.038) Antecedent 0.096 0.001 Within-channel flow vs major flood (0.004) flow Within-channel flow vs no flow (0.001) Major flood vs minor/moderate flood (0.001) Waterhole 0.136 0.009 Permanent channel vs ephemeral lake (0.003) type Ephemeral channel vs ephemeral lake (0.047)

118 Table 5.3 SIMPER analysis comparing fish presence/absence by catchment in the Queensland Lake Eyre and Bulloo-Bancannia Basins from April 2007 – March/April 2008. Abbreviations: Bu = Bulloo, Ky = Kyabra, Co = Cooper, Th = Thomson, Ba = Barcoo, Di = Diamantina, Ge = Georgina, Mu = Mulligan.

Species Average abundance per sample % contribution to observed differences (>5%) Bu Ky Co Th Ba Di Ge Mu BuKy BuCo BuBa BuDi BuGe BuMu KyCo KyBa KyMu KyGe KyDi Bony bream 15 39.5 54.08 17.08 38.15 12.6 267.93 61.2 ------6.78 Cooper Creek ------6.2 6.27 - - - 7.01 6.01 - - - catfish Hyrtl’s tandan 91.25 - 16.75 214.3 10 9.6 41.79 - 15.21 8.38 11.31 12.01 8.73 18.18 9.68 10.34 11.84 7.56 12.01 Silver tandan 10.83 7.25 22.83 20.23 1.15 59 - 57.0 5.79 6.31 11.92 7.9 9.81 - 6.97 11.53 - 8.32 7.9 Australian smelt - - 54.5 7.38 - - - - 11.91 16.4 4.67 8.85 - - 12 8.75 8.28 5.88 8.85 Desert rainbowfish 5.5 8 - - - - 184.36 102.8 13.41 10.29 12.25 13.84 7.42 10.59 12.92 13.22 5.36 5.06 13.84 Glassfish - - - - 3.46 - 108.79 44.2 16.57 9.36 10.74 - 8.55 13.92 - 8.13 15.13 12.14 - Yellowbelly 6.33 26.83 24.25 15.85 17.69 16.8 - - 7.55 - 5.06 - 10.1 22.22 - - 21.93 9.19 - Banded grunter ------205.14 - - - - - 13.98 - - - - 11.82 - Welch’s grunter ------9.55 6.11 - 7.72 - - 8.7 6.64 6.64 5.11 7.72 Spangled perch 41.25 7 15.83 6.31 10.54 - 41.29 40.4 - 10.03 12.52 14.13 9.78 - 10.81 11.4 - 8.09 14.13 Barcoo grunter - - 2.92 ------8.92 - 5.52 - 5.42 10.05 5.21 5.2 - 5.52 Golden goby ------4.64 - - - - - 10.88 - - - - 9.23 - Carp gudgeon 7.5 5.83 - 3.46 0.85 - 12.49 11.45 12.30 17.19 13.13 20.85 13 11.89 15.92 11.26 17.14 (continued)

119 Bu Ky Co Th Ba Di Ge Mu CoBa CoMu CoGe CoDi ThMu ThGe ThDi BaGe BaDi MuDi GeDi Bony bream 15 39.5 54.08 17.08 38.15 12.6 267.93 61.2 - - - 6.02 - - 6.47 5.29 9.48 6.27 6.78 Cooper Creek ------8.42 5.42 - 6.92 5.48 - 7.18 - 7.31 - - catfish Hyrtl’s tandan 91.25 - 16.75 214.3 10 9.6 41.79 - 9.92 9.05 6.35 10.61 9.45 6.9 11.5 7.91 11.26 8.17 8.95 Silver tandan 10.83 7.25 22.83 20.23 1.15 59 - 57.0 10.09 5.17 6.82 8.97 8.48 7.07 11.3 7.5 13.32 6.44 9.21 Australian smelt - - 54.5 7.38 - - - - 14.36 14.57 12.37 19.31 6.57 5.58 8.83 - 5.39 - - Desert rainbowfish 5.5 8 - - - - 184.36 102.8 6.28 12.55 9.34 8.4 13.44 10 9.03 10.81 9.99 14.38 10.73 Glassfish - - - - 3.46 - 108.79 44.2 7.58 10.98 10.63 3.82 11.2 10.69 - 9.68 8.88 16.01 15.2 Yellowbelly 6.33 26.83 24.25 15.85 17.69 16.8 - - - 13.7 7.61 - 16.8 8.49 - 9.31 - 21.7 10.61 Banded grunter ------205.14 - - - 10.2 - - 10.85 - 11.77 - - 13.32 Welch’s grunter ------6.82 5.34 - 7.28 - - 5.44 - - - - Spangled perch 41.25 7 15.83 6.31 10.54 - 41.29 40.4 9.28 8.99 6.59 9.81 5.79 7.25 12.15 7.6 11.46 13.99 8.57 Barcoo grunter - - 2.92 - - - - - 9.47 7.8 6.64 10.08 5.72 - 7.47 - 7.55 6.39 - Golden goby ------4.64 - - - 7.92 - - 8.43 - 9.07 - - 10.02 Carp gudgeon 7.5 5.83 - 3.46 0.85 - 7.3 - - - 9.57 8.11 12.93 - 7.89 - -

120 SIMPER analysis of fish species presence/absence within the Queensland Lake Eyre and Bulloo-Bancannia basins indicates that species with limited distributional ranges generally contribute strongly to the differences in species composition among catchments (Table 5.3). In the Georgina catchment, the presence of banded grunter and golden goby always separated this catchment from all others (Table 5.3), and in the Bulloo and all greater Cooper catchments the presence of carp gudgeon always separated these catchments from the Diamantina, Georgina and Mulligan (Table 5.3). Bony bream, which were present at all sites and in almost all samples, made a very low or nil contribution to the separation of catchments (Table 5.3). In contrast, species present in all (or most) catchments but in widely varying average abundances per sample typically made substantial contributions to the separation of catchments (Table 5.3). Examples include the high average abundance of desert rainbowfish in samples from the Georgina and Mulligan catchments compared with all others and the variable average abundance of Hyrtl’s tandan in all catchments except the Mulligan, where this species was not recorded (Table 5.3).

The mean number of species sampled at all sites was lowest in winter and highest in late summer (Figure 5.2). Analysis of the seasonal variation of fish species presence/absence within the Queensland Lake Eyre and Bulloo-Bancannia basins using SIMPER indicates that a number of species, most notably Hyrtl’s tandan, desert rainbowfish, spangled perch and glassfish, are likely to be absent or comparatively rare in samples taken during winter but are present in late summer (Table 5.4).

121 7

6

5

4

3

Number of species of Number 2

1

0 Late summer Winter Early summer

Figure 5.2 Mean (±S.E.) number of species sampled at all sites in the Queensland Lake Eyre and Bulloo-Bancannia basins in late summer, winter and early summer.

Table 5.4 SIMPER analysis comparing fish species presence/absence in relation to season (or sampling time) in the Queensland Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008. (Early summer = September – December, late summer = January – April, winter = May – August.) Species Average abundance per sample Percent contribution to observed differences (>5%). Early summer Late summer Winter Late summer vs winter Bony bream 100.74 76.81 13.08 5.14 Hyrtl’s tandan 29.57 91.45 - 11.54 Silver tandan 10.70 39.30 7.46 9.48 Australian smelt 11.48 - - 6.62 Desert rainbowfish - 47.60 20.58 10.43 Glassfish 17.13 25.91 - 9.42 Yellowbelly 14.52 17.64 9.92 6.07 Spangled perch 9.57 30.51 5.65 9.96 Carp gudgeon 1.7 - 2.54 8.89

122 Antecedent hydrology had an influence on the number of species present in fish samples taken throughout the Queensland Lake Eyre and Bulloo-Bancannia basins between April 2007 and March/April 2008. The highest numbers of species were recorded in the Georgina catchment in April 2007 following major flooding in January 2007, and in the Cooper catchment in March/April 2008 following flooding in January and February 2008 (Figure 5.3). SIMPER analysis indicates that Hyrtl’s tandan and spangled perch were most commonly sampled following major flooding and that both glassfish and desert rainbowfish were uncommon following minor to moderate flooding and within-channel flows (Table 5.5).

250000 250000

Flood peak Jan 24 – Jan 28 2008

200000 200000

Flood peak 22 Jan – 2 Feb, 2007

150000 150000 ML/day ML/day 100000 100000

50000 50000

0 April 2007 August 2007 November 2007 March/April 2008 0 April 06 Jan 07 Dec 07 – Feb 08

11

10

9

8

7

6

5

4 3

2

Mean number species of Mean number 1

0 April 07 August 07 November 07 March/April 08

Figure 5.3 Mean (±S.E.) number of species sampled in the Bulloo (grey bars), Kyabra (pink bars), Cooper (red bars), Thomson (orange bars), Barcoo (purple bars), Diamantina (blue bars), Georgina (green bars) and Mulligan catchments (yellow bars) in April 2007, August 2007, November 2007 and March/April 2008. Hydrographs of the Georgina River (top left) and Cooper Creek (top right) show major flooding in these catchments in January 2007 and January 2008 respectively, and arrows indicate the highest species richness in both catchments occurring after major flood events.

123 Table 5.5 SIMPER analysis comparing fish species presence/absence in relation to antecedent hydrology in the Queensland Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008. Abbreviations: NF: No flow, WCF: Within- channel connecting flow, MiMoF: Minor to moderate flooding, MaF: Major flooding. Species Average abundance per sample Percent contribution to observed differences (>5%). NF WCF MiMoF MaF WCF vs MaF MiMoF vs MaF WCF vs NF Bony bream 114.23 20.81 19.50 95.05 - - - Hyrtl’s tandan 19.50 8.34 45.71 177.65 9.53 10.25 9.20 Silver tandan 6.03 8.91 63.21 46.80 7.06 7.83 9.34 Australian smelt - 22.41 - - 9.59 - 9.58 Desert rainbowfish 82.77 1.84 - 37.10 10.51 12.56 9.88 Glassfish 45.77 - - 25.45 11.66 12.25 10.58 Yellowbelly 7.37 18.44 24.36 13.45 6.09 6.13 6.76 Spangled perch 8.13 3.03 10.71 65.50 10.15 9.88 9.57 Barcoo grunter - - - - 7.09 7.68 - Cooper Creek catfish - - - - 5.34 6.19 - Carp gudgeon - - - - 8.01 7.51 8.43

Species such as glassfish and bony bream contributed most to species presence/absence differences between ephemeral lakes and within-channel waterholes (both permanent and ephemeral) (Table 5.6). Both bony bream and glassfish exhibited far higher average abundance per sample in ephemeral lake habitats than either permanent or ephemeral within-channel habitats (Table 5.6). In contrast, Hyrtl’s tandan demonstrated higher average abundance per sample in permanent within- channel waterholes. Both silver tandan and spangled perch demonstrated higher average abundance per sample in all ephemeral waterholes (both within-channel and floodplain lakes), whereas desert rainbowfish demonstrated high average abundances only in ephemeral within-channel waterholes (Table 5.6).

124 Table 5.6 SIMPER analysis comparing fish species presence/absence in relation to waterhole type in the Queensland Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008. (Permanent within-channel waterholes = no drying within the study period, ephemeral within-channel waterholes = at least one drying event during the study period, ephemeral lakes = floodplain lakes that dried at least once during the study period.) Species Average abundance per sample Percent contribution to observed differences (>5%). Permanent Ephemeral Ephemeral Permanent within- Ephemeral within- within-channel within-channel lakes channel vs ephemeral channel vs ephemeral lakes lakes Bony bream 43.28 83.21 127.88 24.21 23.64 Hyrtl’s tandan 34.11 - - 10.24 7.85 Silver tandan 7.64 58.89 43.88 14.73 15.16 Desert rainbowfish - 38.79 - 9.86 12.9 Glassfish - - 80.88 12.52 11.03 Yellowbelly 14.67 18.74 10.63 11.45 9.79 Spangled perch 11.05 37.47 25.94 7.71 9.68

5.3.2 Fish species abundance in the Mulligan catchment

Sampled sites in the Mulligan River filled during a flood event in January 2007 and dried down by April 2007 (Dune Pond), September 2007 (S Bend Gorge) and March/April 2008 (Pulchera waterhole; see Chapter 4). Bony bream, spangled perch and glassfish were comparatively abundant on all sampling occasions, whereas the numbers of silver tandan and desert rainbowfish declined dramatically as the system dried towards the end of 2007 (Figure 5.4). Banded grunter and Barcoo grunter were present in the Mulligan system in very small numbers in April and August 2007 (Figure 5.4). A single spangled perch was sampled from Dune Pond in April 2007, however no fish were recorded at a similarly isolated ephemeral site in the Mulligan (Kunnamuka Swamp) or at Ocean Bore (Figure 5.5; see Chapter 4). Dune Pond was situated at the base of a sand dune approximately 20 kilometres from the nearest section of the Mulligan River (Figure 5.5). In the ephemeral Mulligan catchment, the

125 number of present species declined as waterholes evaporated, with only three species present in November 2007 compared with seven in April 2007 (Figure 5.4).

250000

200000

Flood peak 22 Jan – 2 Feb, 2007

150000 ML/day

100000

50000 All sites dry 0 April 2007 August 2007 November 2007 March/April 2008

Bony bream Silver tandan Spangled perch Banded grunter Barcoo grunter Desert rainbowfish Glassfish

April 2007 August 2007 November 2007 n=456 n=846 n=230

Figure 5.4 Proportional abundance of sampled fish species between April 2007 and March/April 2008 in the Mulligan catchment. As the Mulligan is an un-gauged river, a hydrograph from Roxborough Downs in the neighbouring Georgina catchment has been provided as an approximation of hydrological conditions in the Mulligan catchment and to demonstrate the major flooding of January/February 2007 that inundated the formerly dry catchment (Source: Queensland Department of Natural Resources and Water). All sites were dry in the Mulligan prior to January 2007 (Scott Morrison, Ethabuka Station, personal communication) and approximately 2 months after the November 2007 samples were taken (personal observation).

126 S Bend Gorge Ocean Bore Kunnamuka Swamp Dune Pond Pulchera

Figure 5.5 Sites in the Mulligan catchment, showing the location of Dune Pond and Kunnamuka Swamp and their isolation from the main channel of the Mulligan to the east.

5.3.3 Fish species abundance in the Georgina catchment

Sites in the Georgina catchment included some that retained water for the entire sampling period (such as Parapituri and the Main Channel site) and others that dried down during the study period (Lower Lake by November 2007 and Lake Idamea shortly after March/April 2008; Stephen Bryce, Glenormiston Station, personal communication). Fish samples taken in the Georgina were consistently high in number with the exception of the sample taken in August 2007 (Figure 5.6). The fish community in the Georgina catchment was characterised by large numbers of glassfish, rainbowfish and bony bream on all sampling occasions, although the sample of rainbowfish was comparatively low in August 2007 (Figure 5.6). Banded grunter

127 and golden goby populations were present on all sampling occasions in similar proportions in the Georgina catchment, whereas populations of Hyrtl’s tandan, silver tandan and yellowbelly were more variable through time (Figure 5.6). Although both Barcoo and Welch’s grunter were recorded in the Georgina catchment, both species were uncommon in the samples. Despite a prolonged period of drying following April 2007, neither the number of present species nor overall abundance declined dramatically in the Georgina catchment (as it did in the Mulligan).

250000

200000

Flood peak 22 Jan – 2 Feb, 2007

150000 ML/day

100000

50000

0 April 2007 August 2007 November 2007 March/April 2008

Bony bream Yellowbelly Hyrtl's tandan Silver tandan Spangled perch Barcoo grunter Welch's grunter Banded grunter Desert rainbowfish Glassfish Golden goby April 2007 August 2007 November 2007 March/April 2008 n=3282 n=206 n=2411 n=3417

Figure 5.6 Proportional abundance of sampled fish species between April 2007 and March/April 2008 in the Georgina catchment and a hydrograph from Roxborough Downs (Source: Queensland Department of Natural Resources and Water).

Opportunistic samples taken following a prolonged period of drying in November 2008 at Lake Mary, (the most northerly waterhole) and Walkaba/Jimberella, (the largest permanent waterhole) in the Georgina catchment demonstrated a similar species list to samples taken at Parapituri and the Georgina Main Channel site

128 between April 2007 and March/April 2008 (Table 5.7). At both sites, all expected species were recorded except Barcoo and Welch’s grunter (Table 5.7).

Table 5.7 Fish species recorded at Lake Mary and Walkaba/Jimberella waterholes in the Georgina catchment in November 2008. Species Lake Mary Walkaba/Jimberella Nematolosa erebi * * Bony bream Neosiluris hyrtlii * * Hyrtl’s tandan Porochilus argenteus * * Silver tandan Melanotaenia splendida tatei * * Desert rainbowfish Ambassis sp. * * Northwest Ambassis or Glassfish Macquaria sp. * * Lake Eyre golden perch or Yellowbelly Amniataba percoides * * Banded or Barred grunter Leiopotherapon unicolor * * Spangled perch Glossogobius aureus * * Golden goby TOTAL NUMBER OF SPECIES 9 9

5.3.4 Fish species abundance in the Diamantina catchment

The Diamantina catchment did not experience major flooding throughout the study period and all sites remained on a drying trajectory following minor flooding in early 2007 (see Chapter 4). The total numbers of fish recorded in samples from the Diamantina catchment were generally low with the exception of the April 2007 sample (Figure 5.7). Bony bream, yellowbelly and silver tandan were the most commonly sampled fish species in the Diamantina catchment between April 2007 and March/April 2008 (Figure 5.7), with the April 2007 sample dominated by silver tandan and the March/April 2008 sample dominated by yellowbelly (Figure 5.7). Though present during all sampling events, Hyrtl’s tandan was generally uncommon in the Diamantina catchment with the exception of November 2007 (Figure 5.7). Spangled perch, Barcoo grunter, Welch’s grunter and desert rainbowfish were present in the Diamantina samples on most sampling occasions but were never abundant, and a single golden goby was sampled from Lake Billyer in April 2007 (Figure 5.7).

129 Opportunistic sampling of three more sites in the Diamantina catchment in November 2008 revealed a similar pattern of low species diversity (Table 5.8). This was especially evident at Rocky Crossing/Mayne River and Spring Creek/Goneaway, as only two species of fish were recorded from each site (Table 5.8). The sample from Conn waterhole, the highest in the Diamantina in November 2008, included the three commonly-encountered Diamantina species (bony bream, yellowbelly and silver tandan) and a golden goby specimen – the second individual recorded from the Diamantina catchment (Table 5.8).

250000

200000

150000 ML/day

100000

50000

0 Jan/Feb 2007 July 2007 Dec 2007 – Jan 2008

Bony bream Yellowbelly Hyrtl's tandan Silver tandan Spangled perch Barcoo grunter Welch's grunter Desert rainbowfish Golden goby

April 2007 August 2007 November 2007 March/April 2008 n=986 n=111 n=207 n=224

Figure 5.7 Proportional abundance of sampled fish species between April 2007 and March/April 2008 in the Diamantina catchment and a hydrograph from Diamantina Lakes (Source: Queensland Department of Natural Resources and Water).

130 Table 5.8 Fish species recorded at Rocky Crossing, Spring Creek and Conn waterhole in the Diamantina catchment in November 2008. Species Rocky Spring Creek Conn Crossing W’hole Nematolosa erebi * Bony bream Neosiluris hyrtlii * Hyrtl’s tandan Porochilus argenteus * Silver tandan Melanotaenia splendida tatei * Desert rainbowfish Macquaria sp. * * Lake Eyre golden perch or Yellowbelly Leiopotherapon unicolour * Spangled perch Glossogobius aureus * Golden goby TOTAL NUMBER OF SPECIES 2 2 4

5.3.5 Fish species abundance in the Thomson catchment

The fish community in the Thomson catchment was characterised by a core group of species including bony bream, yellowbelly, Hyrtl’s tandan, silver tandan, Australian smelt and carp gudgeon (Figure 5.8). Australian smelt dominated the samples in September 2006, whereas bony bream was the most abundant species during the summer samples taken in December 2006 and January 2007 (Figure 5.8). Samples in both April and August 2007 included low overall totals, however yellowbelly were the most common species during both sampling events (Figure 5.8). Following flooding in the Thomson catchment from December 2007 to February 2008 a comparatively large total number of fish was sampled in March/April 2008 and this sample was dominated by Hyrtl’s tandan (Figure 5.8). Spangled perch, Barcoo grunter, Welch’s grunter, Cooper Creek catfish, desert rainbowfish and glassfish were always comparatively uncommon in samples from the Thomson catchment, and two goldfish were sampled during November 2007 (Figure 5.8).

Opportunistic sampling of Lake Dunn in the upper Thomson catchment occurred in November 2008, and repeat sampling of the Waterloo and Thomson Main Channel sites was also completed. Fish species sampled at Lake Dunn contained a similar fish

131 assemblage to the samples from Waterloo and the Thomson Main Channel (Table 5.9) and previous samples from the Thomson catchment (Figure 5.8), however neither Australian smelt nor carp gudgeon were present. During the November 2008 sampling, a sleepy cod, Oxyeleotris lineolatus , was sampled from Waterloo (Table 5.9; Figure 5.9). This is the first record for this species in the Lake Eyre Basin.

250000 Flood peak 18 Jan – 21 Jan 08

200000 Bony bream Yellowbelly

150000 Hyrtl's tandan Silver tandan Cooper Creek catfish ML/day 100000 Australian smelt Carp gudgeon Spangled perch 50000 Barcoo grunter Welch's grunter Desert rainbowfish 0 Glassfish Jan/Feb 2006 Summer 06/07 Summer 07/08 March/April 2008 Goldfish n=3172

September 2006 December 2006 January 2007 April 2007 August 2007 November 2007 n=393 n=496 n=510 n=117 n=84 n=307

Figure 5.8 Proportional abundance of sampled fish species between September 2006 and March/April 2008 in the Thomson catchment and a hydrograph from Longreach (Source: Queensland Department of Natural Resources and Water).

132 Table 5.9 Fish species recorded at Lake Dunn, Waterloo and Thomson Main Channel waterholes in the Thomson catchment in November 2008. Species Lake Dunn Waterloo Thomson Nematolosa erebi * * * Bony bream Neosiluris hyrtlii * * * Hyrtl’s tandan Porochilus argenteus * * * Silver tandan Retropinna semoni * * Australian smelt Melanotaenia splendida tatei * * Desert rainbowfish Ambassis sp. * Northwest Ambassis or Glassfish Macquaria sp. * * * Lake Eyre golden perch or Yellowbelly Scortum barcoo * * Barcoo grunter Leiopotherapon unicolor * * * Spangled perch Hypseleotris spp. * * Carp gudgeon Oxyeleotris lineolata * Sleepy cod Gambusia holbrooki * Gambusia TOTAL NUMBER OF SPECIES 8 8 10

Figure 5.9 A sleepy cod sampled from Waterloo Waterhole in the Thomson catchment in November 2008. This is the first record of this species from the Lake Eyre Basin.

133 5.3.6 Fish species abundance in the Barcoo catchment

Fish sampling in the Barcoo catchment was disrupted by storm activity and subsequent road closures and site access difficulties in both December 2006 and January 2007. The fish community in the Barcoo catchment was characterised by low numbers of Australian smelt during all sampling events except November 2007 (Figure 5.10). Bony bream and yellowbelly were the most consistently common species in samples from the Barcoo catchment, with populations of most other species (such as Hyrtl’s tandan, silver tandan, carp gudgeon and spangled perch) relatively variable (Figure 5.10). Although Barcoo grunter, Welch’s grunter and Cooper Creek catfish were present in the Barcoo River, these species were never common (Figure 5.10). In contrast, populations of both desert rainbowfish and glassfish were comparatively common in samples from the Barcoo catchment in August 2007 and March/April 2008, and the Barcoo therefore represents the only greater Cooper catchment included in the current study where these species were not consistently rare.

250000

Flood peak Jan 24 – Jan 28 2008

200000

150000 ML/day 100000

50000

0 March/April 2008 April 06 Jan 07 Dec 07 – Feb 08 n=556

Bony bream Yellowbelly Hyrtl's tandan Silver tandan Cooper Creek catfish Australian smelt Carp gudgeon Spangled perch Barcoo grunter Welch's grunter Desert rainbowfish Glassfish Goldfish September 2006 April 2007 August 2007 November 2007 n=121 n=354 n=126 n=104

Figure 5.10 Proportional abundance of sampled fish species between September 2006 and March/April 2008 in the Barcoo catchment and a hydrograph from Retreat (Source: Queensland Department of Natural Resources and Water).

134 5.3.7 Fish species abundance in the Cooper catchment

Australian smelt, bony bream and yellowbelly were consistently the most abundant species in samples from the Cooper catchment between September 2006 and November 2007 (Figure 5.11). Following major flooding in summer 2007/2008 the sample in March/April 2008 exhibited comparatively similar numbers of common species such as bony bream, yellowbelly, Hyrtl’s tandan, silver tandan and spangled perch (Figure 5.11). Sites in the Cooper catchment in March/April 2008 also exhibited the largest numbers of Cooper Creek catfish, Barcoo grunter and Welch’s grunter recorded during the study (Figure 5.11). Desert rainbowfish, glassfish and carp gudgeon were consistently rare in the Cooper catchment, and this was also the only sampled catchment where gambusia were present (Figure 5.11).

Bony bream 250000 Yellowbelly

Flood peak Jan 24 – Jan 28 2008 Hyrtl's tandan Silver tandan 200000 Cooper Creek catfish Australian smelt 150000 Carp gudgeon Spangled perch

ML/day Barcoo grunter 100000 Welch's grunter Desert rainbowfish

50000 Glassfish March/April 2008 Gambusia 0 April 06 Jan 07 Dec 07 – Feb 08 n=793

September 2006 December 2006 January 2007 April 2007 August 2007 November 2007 n=367 n=473 n=750 n=538 n=261 n=834

Figure 5.11 Proportional abundance of sampled fish species between September 2006 and March/April 2008 in the Cooper catchment and a hydrograph from Retreat (Source: Queensland Department of Natural Resources and Water).

135

5.3.8 Fish species abundance in the Kyabra catchment

Kyabra Creek is an un-gauged catchment within the greater Cooper catchment where no flooding occurred throughout the study period (Bob Morrish, Springfield, personal communication). The fish community in Kyabra Creek exhibited only slight variation between September 2006 and March/April 2008 (Figure 5.12). Bony bream and silver tandan were abundant during all sampling events, and yellowbelly and Hyrtl’s tandan were also common on all occasions with the exception of January 2007 and August 2007, respectively (Figure 5.12). Australian smelt, spangled perch, carp gudgeons and desert rainbowfish displayed variable abundance in Kyabra Creek between September 2006 and March/April 2008, with Australian smelt most common in September 2006, spangled perch most common in December 2006 and January 2007, carp gudgeons most common in September 2006 and desert rainbowfish most common in August 2007 (Figure 5.12). Barcoo grunter, Welch’s grunter and glassfish were generally uncommon in Kyabra Creek throughout the study, and Cooper Creek catfish were not recorded (Figure 5.12).

September 2006 December 2006 January 2007 April 2007 n=389 n=541 n=875 n=429

Bony bream Yellowbelly Hyrtl's tandan Silver tandan Australian smelt Carp gudgeon Spangled perch Barcoo grunter Welch's grunter Desert rainbowfish Glassfish August 2007 November 2007 March/April 2008 n=249 n=400 n=290

Figure 5.12 Proportional abundance of sampled fish species between September 2006 and March/April 2008 in the Kyabra catchment.

136 5.3.9 Fish species abundance in the Bulloo catchment

The Bulloo catchment experienced overbank flooding during summer 2006/2007 and summer 2007/2008. With the exception of a lower overall total in August 2007, fish communities in the Bulloo catchment displayed very little variation between April 2007 and March/April 2008, with species abundances almost identical during these two sampling events (Figure 5.13). Hyrtl’s tandan was the most abundant fish species in the Bulloo catchment during all sampling periods, with yellowbelly, spangled perch, silver tandan and desert rainbowfish present in consistently similar population proportions throughout the study (Figure 5.13). Glassfish and carp gudgeon were the only species in the Bulloo to demonstrate seasonal population variation, with glassfish rare in November 2007 and carp gudgeon common in August 2007.

250000

200000 Flood peak 21 -22 Jan 2008

150000 Flood peak 25 – 26 Jan 2007 ML/day 100000

50000

0 Jan 05 Jan 06 Jan 07 Jan 08

Bony bream Yellowbelly Hyrtl's tandan Silver tandan Carp gudgeon Spangled perch Desert rainbowfish Glassfish

April 2007 August 2007 November 2007 March/April 2008 n=806 n=221 n=722 n=473

Figure 5.13 Proportional abundance of sampled fish species between September 2006 and March/April 2008 in the Bulloo catchment and a hydrograph from Quilpie (Source: Queensland Department of Natural Resources and Water).

137 5.4 Discussion

The catchments considered during the present study demonstrate strong populations of native Australian fish species and noticeably small populations of alien species. On this basis alone, the river systems of the Queensland Lake Eyre and Bulloo-Bancannia basins can therefore be considered to be in good ecological condition when compared with those in the Murray-Darling – a similar-sized Australian semi-arid to arid zone system (Harris and Gehrke 1997; Balcombe et al. 2006; Rayner et al. 2009). The spatial distribution of certain fish species is related to catchment boundaries, suggesting, as previous genetic studies of both invertebrates and fish have demonstrated, that species dispersal has been confined within individual catchments for an extended time period (Hughes and Hillyer 2003; Hughes et al. 2004; Huey et al. 2006). The role played by catchment boundaries in structuring fish assemblages is especially evident for species such as banded grunter, Cooper Creek catfish and Australian smelt, as these species were only collected within geographically separated catchments within the area considered by the study.

The variation in fish species presence/absence between catchments in the Queensland Lake Eyre and Bulloo-Bancannia basins during the present study indicates that fish assemblages are broadly similar (as predicted) throughout the greater Cooper catchment(s) and similar between the Georgina and Mulligan catchments (Hypotheses 1b and 1d, Section 5.1). Hypothesis 1a, which predicted the Bulloo catchment would have a distinct species composition, is also supported, however hypothesis 1c is refuted by the results. Sites in the Diamantina catchment, rather than exhibiting similarities to the Georgina and Mulligan catchments, instead were more similar in terms of fish species presence/absence to sites in the greater Cooper and Bulloo catchments. It seems likely that a combination of the species composition and hydrological history of the Diamantina and Georgina catchments may account for these differences. In the Georgina, banded grunter, glassfish and golden goby were common and comparatively abundant in all or most samples, whereas in the Diamantina banded grunter and glassfish were absent and golden goby noticeably rare. Additionally, yellowbelly, though present at all sites on all sampling occasions in the Diamantina, were comparatively rare in the Georgina and absent in the Mulligan.

138 Results from the Georgina catchment indicate that the number of present fish species is likely to rise at individual localities following major flooding. In contrast, results from the Diamantina catchment provide evidence that prolonged dry periods are likely to result in a more depauperate fish fauna in Australian dryland systems. This may have implications for the study of the effect of extended droughts and possible climate change on fish communities in arid areas (Matthews and Marsh-Matthews 2003), as it indicates that prolonged aridity is likely to reduce species diversity and abundance in arid-zone river systems.

The predominantly low number of species and abundances detected in winter support the findings of Costelloe et al. (2004) and Balcombe and Arthington (2009), and provide support for hypothesis 2 (Section 5.1). The results indicate that the number of detectable species and their abundance are likely to be limited in waterholes of the Australian arid zone during winter (August), but that population peaks and increases in species richness of fish communities following flooding are also likely to occur during the warmer months. These findings are in agreement with similar results from the South Australian Lake Eyre Basin (Puckridge et al . 2000) and Cooper Creek in Queensland (Arthington et al. 2005) and therefore suggest support for hypothesis 3 (Section 5.1). Results demonstrating a preference for ephemeral habitats by certain fish species (see below) indicate that hypothesis 4 (Section 5.1) is also supported by data from this study, however it should be noted that the number of present species declined dramatically in the ephemeral Mulligan catchment as it dried throughout 2007, a result not replicated in the Mulligan’s parent river, the Georgina. The presence of permanent waterholes in the Georgina, and their absence in the Mulligan, accounts for this difference, and suggests that species and individuals likely to capitalise on the conditions afforded by ephemeral habitats originate from source populations in permanent refuges.

Widespread species

The most widely distributed fish species in the Queensland Lake Eyre and Bulloo- Bancannia basins during the study period were spangled perch, silver tandan, desert rainbowfish and bony bream. These species were detected in all catchments from the Bulloo River west to the Mulligan and on all sampling occasions. Their presence in

139 the ephemeral Mulligan catchment indicates that these species, as well as glassfish, are likely to undertake long migrations (up to 250 kilometres upstream) in order to colonise newly-inundated areas in far western Queensland (see Chapter 7). Bony bream and spangled perch are considered to be the two most widely distributed Australian freshwater species, and both occur in multiple Australian drainage basins (Allen et al. 2002; Pusey et al . 2004). Consequently, their extended distribution throughout far western Queensland, including the remote and ephemeral Mulligan system is not unexpected. Results from the current study indicate that bony bream appear to display a preference for recently-inundated floodplain habitats such as large shallow lakes, and this is in agreement with previous studies in areas of the western and southern Murray-Darling Basin, such as Lake Cargelligo (Kerezsy 2005) and Lake Alexandrina (Puckridge and Walker 1990), as well as studies in Cooper Creek near Windorah in Queensland (Balcombe et al. 2007) and the downstream Cooper at Coongie Lakes in South Australia (Puckridge 1999). The presence of a spangled perch in the extremely remote and rapidly-drying Dune Pond site at the base of a sand dune in the eastern Simpson Desert in April 2007 gives further indication of the hardiness and migration abilities of this species and supports previous observations of ‘overland’ migrations during extreme rainfall events (Pusey et al. 2004). The generally high average abundance of spangled perch in ephemeral habitats recorded by this study similarly demonstrates the vagility of this species, however the variable presence of spangled perch throughout the study area is curious. Despite being present at all sites during all sampling times in the Mulligan and Bulloo catchments, and present in the Kyabra catchment in 87.5% of samples, spangled perch distribution was patchy ( ≤50% of samples) in catchments such as the Diamantina, Barcoo and Cooper. This evidence strongly suggests that spangled perch preferentially inhabit smaller and more ephemeral waterways throughout the Lake Eyre Basin, possibly due to reduced competition from other large-bodied piscivores (such as yellowbelly) in such areas.

Neither silver tandan nor desert rainbowfish possess a multi-basin range within Australia, and both are endemic to catchments in the centre of the continent. These species can therefore be considered arid-zone ecological specialists within their respective families (the Plotosidae and Melanotaeniidae). Silver tandan were morphologically identical throughout their range between September 2006 and March/April 2008 with the exception of an individual sampled in Pulchera waterhole

140 in the Mulligan catchment in April 2007. This stunted specimen was almost certainly a mutant silver tandan and its presence was probably related to enhanced recruitment conditions associated with elevated flood levels (Appendix 4). The high average abundance of silver tandan recorded from all samples taken in ephemeral habitats is indicative of the migratory abilities of this species. Desert rainbowfish, though exhibiting similar migratory behaviour, appeared to favour ephemeral waterholes within main channel watercourses rather than floodplain environments. This species was also recorded at Spring Creek, a rock-hole in the Mayne River sub-catchment of the Diamantina, indicating that populations of desert rainbowfish are capable of persisting in small, highly isolated areas where permanent water exists. Although it has been suggested that desert rainbowfish from the Georgina catchment and further west are differently coloured than populations in the more easterly rivers (Wager and Unmack 2000), desert rainbowfish were identical in all catchments except the Mulligan during the current study. The samples of desert rainbowfish from S Bend Gorge in the Mulligan catchment were characterised by fish with vivid orange, purple and/or yellow longitudinal stripes along each flank (Figure 5.14). Local colour variation is common in Melanotaeniid fish, and species from adjacent catchments frequently exhibit different colour patterns (Helen Larson, Northern Territory Museum and Art Gallery, personal communication). Nevertheless, the presence of a local colour variant of desert rainbowfish from the Mulligan River is notable, primarily because the Mulligan is an ephemeral river. The results therefore suggest that a source population of vividly-coloured desert rainbowfish is likely to exist in the Mulligan catchment. 6 The comparative abundance of desert rainbowfish in the Georgina catchment is in contrast to the general rarity of this species in the Cooper and Diamantina. It seems likely that the less-turbid water in the western catchments may be advantageous for this species, which generally inhabits the upper levels of the water column and is known to feed at or near the surface (Wager and Unmack 2000; Pusey et al. 2004). At the very least, these results suggest that future research directed towards this species could be completed more easily in the Georgina than in the more easterly catchments.

6 Sampling undertaken in October 2009 in the Mulligan catchment located a free-flowing bore on the Ethabuka/Marion Downs border that is the most likely source population for vividly-coloured desert rainbowfish. It seems likely that fish from this population may colonise the upper Mulligan following flooding (see Appendix 5).

141

Figure 5.14 Desert rainbowfish sampled during the current study included vividly coloured specimens from the Mulligan catchment (left) and less-colourful specimens from all other catchments (right).

Yellowbelly and Hyrtl’s tandan

Yellowbelly and Hyrtl’s tandan were distributed in every catchment except the Mulligan system in the Simpson Desert, despite the fact that migration from the flooded Georgina into the flooded Mulligan was possible in January 2007. Although genetic differences exist between yellowbelly populations in eastern and western catchments of the Lake Eyre Basin (Faulks 2009), all yellowbelly have been treated as a single species during the current study. The great majority of Hyrtl’s tandan sampled during the current study were easily identifiable, however specimens from the Georgina River in April 2007 and November 2008 and also from Lake Dunn in the upper Thomson catchment in November 2008 displayed different head profiles and it is possible that local varieties may exist (Appendix 4). Nevertheless, Hyrtl’s tandan from all catchments have been grouped together as one taxon during this study.

Results from this study suggest that the Georgina system may represent the western distributional limit for both yellowbelly and Hyrtl’s tandan in Queensland. Alternatively, it is possible that sampling in April and August 2007 failed to detect these species in the Mulligan, or that sampling following another flood may yield different results. 7 Evidence from other systems during the study demonstrates that of the two species, yellowbelly is a highly mobile fish that undertakes migrations to newly inundated areas in the Thomson and Barcoo systems (see Chapter 7), but that the species is not as common in the Georgina, thus lending support to the notion that

7 Opportunistic sampling of sites in the Mulligan catchment in October 2009 (and following flooding and connection earlier in the year) similarly found no yellowbelly or Hyrtl’s tandan.

142 the distributional range of yellowbelly may have a western limit in Queensland. Despite being present at all or nearly all sites on all sampling occasions in the Diamantina (100%), Thomson (100%), Barcoo (100%), Cooper (95%), Kyabra (95%) and Bulloo (75%) catchments, yellowbelly were rare in the Georgina (41.6%), a surprising result given that major flooding occurred in this catchment and is reported to have a positive impact on the recruitment success of this species in the more southern areas of its range (Lake 1967; Cadwallader 1978 and 1979). The results therefore suggest that the turbid waterways of the Diamantina and greater Cooper catchments may provide more suitable habitat for this ambush predator than the comparatively clear and occasionally saline conditions present in the Georgina catchment.

Hyrtl’s tandan was locally common in many of the waterholes sampled across the entire study area but was often absent from small, shallow waterholes, suggesting that this species may preferentially avoid ephemeral areas. Nevertheless, the presence of juvenile Hyrtl’s tandan in ephemeral habitats such as Lower Lake and Lake Idamea in the Pituri Creek system of the Georgina catchment following flooding in January 2007 indicates that this species probably utilises ephemeral areas as nursery habitats (Chapter 6). It is possible that migration distance may influence the presence of Hyrtl’s tandan, given that the Pituri Creek waterholes are comparatively close (<10 kilometres) to the main channel of the Georgina but the Mulligan River sites are up to 250 kilometres distant (see Chapter 7; Davey and Kelly 2007; Larned et al. 2009). During the current study all catfish sampled in the Bulloo catchment were positively identified as either Neosiluris hyrtlii or Porochilus argenteus using published guides (Allen et al. 2002). No evidence was found of a third species of false-spined catfish occurring in the Bulloo River (see Wager and Unmack 2000).

Glassfish

Glassfish were present in all catchments except the Diamantina during the current study, although it should be noted that this species was common only in the Georgina, Mulligan and Bulloo systems. In almost all greater Cooper sites glassfish abundance was low throughout the sampling period and remained low following the major flooding of January/February 2008. The only exception was the sample taken at Lake

143 Dunn in the upper Thomson catchment in November 2008, where comparatively strong numbers (>100) of glassfish were recorded. This is in stark contrast to the large numbers of glassfish sampled in the western catchments (Mulligan and Georgina) and comparatively robust populations sampled in the Bulloo throughout 2007. Thus, although there was a detectable flood response for this species, this response was not uniform across all catchments and suggests that population booms of glassfish in far western Queensland are likely to be erratic (Puckridge 1999). Similar results for this species have been reported in previous studies conducted in the Queensland Lake Eyre Basin (Arthington et al. 2005). Glassfish were always most abundant in samples taken in floodplain lakes separated from the main channel during the current study but were absent from these areas in the Diamantina catchment where major flooding did not occur. Further studies in the Diamantina catchment across a wider geographic area or along a longer temporal timeframe (preferably including a major flood event) would be useful with regard to establishing the status of this species in the upper (Queensland) Diamantina catchment, as recent surveys (Costelloe et al. 2004) reveal that glassfish, though common in South Australian reaches of the Diamantina, are rare in Queensland. 8 Given that opportunistic sampling undertaken at Diamantina sites in November 2008 (Mayne River, Spring Creek and Conn waterhole) also failed to detect this species, it seems highly likely that glassfish are either absent or rare in the upper reaches of the Diamantina catchment in Queensland.

The taxonomy of Ambassid fishes is uncertain, and the fish sampled during the current study were identified as the undescribed species Ambassis sp. (Northwest Ambassis) (Allen et al. 2002). Morphological examination of glassfish from the Bulloo drainage suggests that this species has fin ray counts intermediate between Ambassis sp. and the olive perchlet, Ambassis agassizii , which is found in the Murray- Darling and eastern coastal catchments (Pusey et al. 2004; Jeff Johnson, Queensland Museum, personal communication). It is therefore recommended that research regarding the taxonomy of Ambassid fishes be undertaken using a combination of morphological and genetic methods, and that this research should include representatives of glassfish from all inland catchments including the Bulloo.

8 During sampling for the ARIDFLO project, a total of 2,263 glassfish were recorded from the South Australian Diamantina compared with only 19 in Queensland (Costelloe et al. 2004)

144 Barcoo and Welch’s grunter

Barcoo and Welch’s grunter were not sampled in large numbers during the current study, and these results are therefore similar to other recent spatial/temporal surveys in the Lake Eyre Basin (Costelloe et al. 2004; Arthington et al. 2005). The distribution of both species included all greater Cooper catchments, the Diamantina and the Georgina between September 2006 and March/April 2008. In addition, Barcoo grunter were sampled in the Mulligan catchment, and both species were not recorded from sites in the Bulloo catchment. Results from the current study therefore indicate that both Welch’s and Barcoo grunter are present in the Queensland Lake Eyre Basin and that antecedent flooding is likely to lead to an increase in populations and allow Barcoo grunter to colonise previously dry catchments.

The distributional status of morphologically similar Terapontid fish in Australian catchments has undergone several re-interpretations within the last 25 years (Merrick and Schmida 1984; Allen et al. 2002), and speculation continues regarding the possibility that a third species may exist in the Lake Eyre Basin (Costelloe et al. 2004). During the current study, no evidence was found of a possible third Terapontid and all specimens were identified as either Bidyanus welchi or Scortum barcoo based on published guides (Wager and Unmack 2000; Allen et al. 2002). It is possible that Terapontid fish may be naturally rare in the upper Bulloo River where sampling occurred and that this may account for their absence from all samples. Alternatively, given the paucity of prior studies in the Bulloo drainage and the fact that Welch’s grunter were sampled rarely during the only existing prior study (Midgely et al. 1991), it appears likely that both Welch’s and Barcoo grunter may be uncommon in this system, or that the generalisations reported in published guides may be slightly incorrect (both species common in the Bulloo: Merrick and Schmida 1984; Wager and Unmack 2000; Allen et al. 2002). Resolving the status of Terapontid fish distributions in the Bulloo catchment is therefore recommended and could be achieved by conducting fish sampling at a variety of sites throughout the entire length of the Bulloo River and over a temporal gradient spanning at least one large flow event.

145 Banded grunter

Banded grunter was the species with the most westerly distribution during the current study, and was only detected in the Georgina and Mulligan catchments. This supports and enhances existing distributional records of this species in Queensland, and indicates that banded grunter are most likely to have entered the Lake Eyre Basin via the northern Basin divide and made their way into the Georgina system from the rivers of the Gulf of Carpentaria. A specimen held at the Queensland Museum, though recorded as being from the Diamantina catchment, was caught at Lake Muncoonie in the lower Mulligan/Eyre Creek area and hence the Georgina catchment (Jeff Johnson, Queensland Museum, personal communication). Given that surveys of South Australian reaches of the Diamantina River and the Neales River in the western Lake Eyre Basin include records of this species (Costelloe et al . 2004), it seems likely that banded grunter may be present in the upper (Queensland) Diamantina catchment. During the current study, the comparatively large numbers and densities of this species encountered in the Georgina catchment indicate that the sampling methods used are adequate for detection of banded grunter. Consequently, accurately establishing the range of this species in the upper Diamantina is contingent on sampling over a large geographical area rather than employing different sampling techniques. The apparent absence of this species from large, permanent waterholes in the Diamantina such as Hunter’s Gorge and Warracoota may be considered curious given its presence in the South Australian Diamantina and its abundance in the Georgina. Nevertheless, failure to catch banded grunter in November 2008 at sites close to the northern and eastern Diamantina catchment boundaries (Conn Waterhole, Mayne River and Spring Creek) tends to confirm that this species does not occur in the upper reaches of this catchment. It therefore seems most probable that populations of banded grunter have radiated southwards from the Georgina catchment (Costelloe et al . 2004) and that separate colonisation of the more easterly Lake Eyre Basin drainages has not occurred. Sampling of the Diamantina catchment between Diamantina Lakes National Park and Birdsville would be desirable in order to more accurately gauge the distribution of banded grunter in far western Queensland.

146 Golden goby

The current study is the first to document the presence of golden goby along a temporal timeframe in the Georgina catchment, and the first to document the presence of this species in the Diamantina catchment (at Lake Billyer in April 2007 and at Conn Waterhole in November 2008). The golden goby is a morphologically similar species to flathead goby, Glossogobius giurus, and consequently the identification of the Lake Eyre Basin species was confirmed by sending specimens to Helen Larson, Curator of Icthyology at the Northern Territory Museum and Art Gallery. Golden goby were sampled in all Georgina catchment sites between April 2007 and March/April 2008 including Lake Idamea and Lower Lake in Pituri Creek. Golden goby were also present at Lake Mary, the most upstream waterhole in the Georgina, and at Walkaba/Jimberella Waterhole where they were collected in November 2008. These results indicate that the golden goby is a widespread species in the Georgina catchment and that during flow events this species is likely to move into previously dry areas. The presence of golden goby in all Georgina sites indicates that, like banded grunter, populations of this species in the Lake Eyre Basin are probably derived from ancestral stocks originating in the Gulf of Carpentaria. Golden goby have a wide range in coastal catchments throughout the tropical western Pacific (Allen et al. 2002), and northern Australian populations access marine environments as larvae (Pusey et al. 2004). The presence of golden goby as far south (and as far inland) as Boulia in far western Queensland indicate that populations of this species in the Lake Eyre Basin have developed a recruitment strategy that is not reliant on access to the sea. Despite the fact that some waterholes in the Georgina catchment are comparatively saline (such as Parapituri), many are not (such as Lake Mary and Walkaba/Jimberella), and golden goby were collected from all areas with a wide range of salinities. It seems possible that inland populations of golden goby may therefore represent a specialised local variety capable of completing its life history within fresh water. Golden goby almost certainly have a southern distributional limit within the Lake Eyre Basin, as this species appears to be absent from waterholes in South Australia (Costelloe et al. 2004; Dale McNeil, SARDI, personal communication). At present, golden goby in the Diamantina catchment can be considered rare as only two specimens were caught during this current study and no golden goby have been recorded during prior survey work in this catchment

147 (Costelloe et al. 2004; Vanessa Bailey, Queensland Environmental Protection Agency, personal communication). Accurately establishing the distributional range of golden goby in far western Queensland can only be achieved through extensive sampling effort directed at the Georgina and Diamantina catchments and including their headwaters and tributaries. Additionally, establishing the southern distributional limit of golden goby in the Georgina and Diamantina catchments should also be completed given the apparent absence of this species from South Australia. This work should be prioritised by management agencies as golden goby can be considered a unique and uncommon species, especially in the upper Diamantina.

Cooper Creek catfish

Between September 2006 and March/April 2008, Cooper Creek catfish were only sampled from sites in the Thomson and Barcoo rivers and Cooper Creek, and no specimens were sampled in Kyabra Creek (also part of the greater Cooper catchment). Although Cooper Creek catfish were relatively uncommon in the samples, this pattern is also evident in other surveys conducted in the greater Cooper catchment (Costelloe et al. 2004; Arthington et al. 2005). The apparent absence – or at the very least rarity – of Cooper Creek catfish in Kyabra Creek suggests that the distribution of this species may be generally confined to major watercourses and that it therefore has a geographically limited range within the Thomson, Barcoo and Cooper catchments. It seems possible that smaller tributaries within the greater Cooper catchment (such as Kyabra Creek) may be unsuitable habitats for this species due to the increased likelihood of prolonged dry periods, and that Cooper Creek catfish may be a specialised species that is generally limited to permanent – or stable – habitats (Poff and Allan 1995). Ensuring the survival of Cooper Creek catfish is thus contingent upon protection of the catchments where this endemic species occurs and should be considered a priority by agencies charged with managing the Thomson/Barcoo/Cooper drainages.

148 Australian smelt

Australian smelt were common throughout the greater Cooper catchment during the current study (September 2006 to March/April 2008), with numbers peaking between the months of September and January. It appears likely that the increase in smelt detected during this early summer period may be related to breeding activity of this species (see Chapter 6). Despite being common in the greater Cooper catchment, and present in the north-western Murray-Darling Basin (Balcombe et al. 2006), Australian smelt were not detected in the Bulloo catchment during the current study, and existing records also indicate that this species is absent from this catchment (Midgley et al. 1991; Wager and Unmack 2000; Allen et al. 2002). The results from the current study therefore support the findings of a genetic study of Australian smelt indicating that the Lake Eyre population in the greater Cooper catchment is more likely to have southern rather than eastern biogeographic origins (Hammer et al. 2007). This is further supported by sampling in the upper Thomson (Lake Dunn) in November 2008, where no Australian smelt were found. Accurately establishing the distributional range of Australian smelt in the Thomson catchment is therefore contingent on investigation of presence/absence in Thomson River waterholes north of Stonehenge and including all upper catchments, such as Towerhill, Torren’s, Aramac and Cornish Creeks and the Landsborough River. The apparent absence of Australian smelt from the Diamantina and Georgina catchments similarly suggests that colonisation of these catchments has either not occurred or has occurred unsuccessfully, despite migration pathways existing during sporadic filling events of Lake Eyre (hypothetically permitting colonisation of the Diamantina/Warburton from the Cooper) and Goyder’s Lagoon (hypothetically permitting colonisation of the Georgina from the Diamantina).

Carp gudgeon

The current study has confirmed the westward distribution of carp gudgeon in the Thomson and Cooper catchments but no representatives of this species were detected in the more westerly Diamantina and Georgina catchments. It therefore appears most likely that carp gudgeon may have radiated east from the Paroo and Warrego systems in the Murray-Darling Basin and the Bulloo catchment to the greater Cooper, with further westerly movement limited by the Cooper/Diamantina catchment divide.

149 During the current study, no isolated populations of purple-spotted gudgeons (Mogurnda spp.) were detected in the Bulloo or Barcoo catchments (Wager and Unmack 2000). However, there is a possibility that populations exist in the western rivers, given the presence of these species at isolated locations throughout the Lake Eyre Basin ( M. thermophila at Dalhousie Springs, M. clivicola in the Flinders Ranges and M. larapintae in the ; Allen et al. 2002).

Translocated species

The presence of a single sleepy cod at Waterloo waterhole in the Thomson catchment in November 2008 is a surprising result, especially given the fact that this waterhole has received a considerable amount of research-based sampling effort, as it was surveyed throughout the current study and also during the Dryland Refugium project (Arthington et al. 2005). The sleepy cod is a hardy species with a natural distribution encompassing the north-east coast and Gulf of Carpentaria divisions (Pusey et al. 2004). Within the last 5 – 10 years sleepy cod has also become a relatively popular aquaculture species (Bruce Sambell, Ausyfish, personal communication) and numerous range extensions have been recorded throughout Queensland (Michael Hutchison, Queensland Department of Primary Industries and Fisheries, personal communication). It therefore seems most likely that the presence of a sleepy cod in the Thomson catchment is the result of unauthorised translocation rather than a natural range extension. Nevertheless, further sampling and monitoring of waterholes in the Thomson/Barcoo/Cooper is recommended in order to accurately gauge the distribution of this species in these rivers, and to track any range expansion of potentially translocated populations.

Absent species

Lake Eyre hardyhead and desert goby are present and occasionally abundant in the South Australian Lake Eyre Basin (Glover and Sim 1978; Glover 1979; Glover 1982; Wager and Unmack 2000), but were not recorded from any Queensland sites during the current study. Costelloe et al. (2004) suggest that the lack of saline waterholes upstream of the confluence of the Georgina and Diamantina rivers at Goyder’s Lagoon may be a factor limiting the dispersal of these species to the upper

150 Diamantina. However, waterholes studied in the Georgina catchment during the current study frequently exhibited elevated salinity (see Chapter 4) and the presence of these species is therefore likely in this catchment. Following the conclusion of field sampling for the current study, hardyhead specimens were sampled in the Mulligan catchment in October 2009 (see Appendix 4). Although it appears most likely that this species is the Lake Eyre hardyhead, Craterocephalus eyrseii , positive identification of specimens using genetic techniques has commenced (see Appendix 5).

Similarly, no records of species known from spring complexes in Queensland, such as red-finned blue-eye, Edgbaston goby and Elizabeth Springs goby, were detected at any of the sampling sites.

Alien species

The absence of alien fish species from all catchments except the greater Cooper during the current study indicates that neither goldfish nor gambusia are established in the Diamantina and Georgina systems despite the existence of occasional migration pathways during periods of high flow (via Lake Eyre and Goyder’s Lagoon). The apparent absence of alien fish species from the Bulloo River is also notable, particularly as populations of goldfish, gambusia and carp, Cyprinus carpio , are present in the north-western Murray-Darling (to the east) and populations of goldfish and gambusia are present in the greater Cooper catchment (to the west). Although goldfish appear to be range-limited to the Thomson and Barcoo catchments, gambusia are more widespread, with populations also present in the remote Neales River in South Australia (Costelloe et al. 2004) and at Nocundra on the Wilson River, the most south-easterly sub-catchment in the Queensland Lake Eyre Basin (personal observation). Populations of gambusia are also established in springs at Edgbaston, north-east of Aramac, where they represent a threat to one of Australia’s most critically endangered freshwater fish, the red-finned blue-eye (Fairfax et al . 2007, personal observation). Although the low numbers of alien species recorded during the current study and recent temporal surveys in the greater Cooper catchment (Costelloe et al. 2004; Arthington et al . 2005) are encouraging, these species should nevertheless be considered potentially damaging to the aquatic ecosystems of all rivers in far western Queensland and education programs detailing their identification and alien

151 status should be instigated by all relevant management authorities. Monitoring of sites throughout the Queensland Lake Eyre Basin is recommended in order to document population peaks of invasive species if they occur. It is recommended that areas where gambusia are currently abundant, such as Nocundra and the springs at Edgbaston, be included in such a monitoring program.

Summary

The results presented in this chapter show that the Bulloo catchment is differentiated from the Lake Eyre Basin catchments, that rivers within the greater Cooper catchment (Kyabra/Cooper/Thomson/Barcoo) share a generally similar species mix and that this situation is also true for the Georgina and Mulligan catchments. Contrary to expectations, the Diamantina catchment was found to have a fish fauna more similar to the greater Cooper drainages than to the Georgina catchment. This finding should be noted by natural resource managers in far western Queensland, as currently the Diamantina and Georgina catchments are managed together (Lake Eyre Basin Co- ordinating Group 2000). The data presented in this chapter and in Chapter 4 provide evidence that the Georgina is both physico-chemically and biologically different from the more easterly rivers of the Queensland Lake Eyre Basin. Temporal variability in fish presence/absence was most obvious during the winter collection period (August 2007), when many species known to be present in warmer months were not recorded. Fish abundance in samples (all species) also declined in winter. In catchments experiencing major flooding (the Georgina/Mulligan in early 2007 and the Thomson/Barcoo/Cooper in early 2008), species richness and the abundance of most species increased, a result that supports previous studies in both the Lake Eyre and Murray-Darling basins (Puckridge et al. 2000; King et al. 2003; Costelloe et al. 2004). Last, certain fish species exhibited spatial preference for either permanent (Hyrtl’s tandan) or ephemeral (bony bream, silver tandan, glassfish) habitats across the study area and through time.

152 6. Fish recruitment in the Queensland Lake Eyre and Bulloo-Bancannia basins

6.1 Introduction

Reproductive strategies of all animals and plants serve to ensure the survival of juveniles through to maturity such that the species persists across its full geographic range. Consequently, reproductive strategies are necessarily highly variable, and range from mass hatching of and seeds to parent-reliant strategies concentrated on successfully raising small numbers of offspring as evinced by mammals. The term ‘recruitment’ specifically refers to the addition of new members to a population (Pusey et al. 2004). Fish are generally highly fecund animals, producing a large number of eggs corresponding to declining numbers of larvae, juveniles and – ultimately – adults (Jobling 1995). The reproductive strategies of fish are also highly variable, and range from apparently random dispersal of eggs in the water column to nest-building and parental mouth-brooding. Variability in larval and juvenile recruitment is thus the main factor influencing population fluctuations of fishes (Caley et al. 1996), and mortality of larvae is the main factor influencing recruitment variability (Cushing 1975). Fish eggs, larvae and juveniles are subject to mortality risk through predation (Rice et al. 1997), cannibalism (Folkvard 1997) and environmental factors such as stochastic drying events. Consequently, the condition of rearing habitats and the ability of adults to reproduce at opportune times have also been identified as major factors contributing to reproductive success (Humphries 2005).

The influence of flow in riverine systems has been recognised as the predominant factor driving all freshwater ecosystem structure and processes (Walker et al. 1995; Bunn and Arthington 2002), and more specifically the life histories of fish species in particular environments (Winemiller 1989). Studies demonstrating the relationship between predictable summer flooding in tropical areas and fish spawning (Welcomme 1985) provide a relatively clear and direct link between an environmental

153 phenomenon and a biological response, and have informed theories relating to the intrinsic linkage of the two events such as the Flood Pulse Concept (Junk et al . 1989). Nevertheless, appropriating such theories across wider spatial and environmental scales is problematic due to the specific physical, hydrological and biological attributes that river systems may possess (Leigh and Sheldon 2008). Australian studies in the Murray-Darling Basin, for example, increasingly conclude that fish in wild habitats may spawn regularly and that cohort success rather than spawning itself may be influenced by the presence or absence of flow (Humphries et al. 2002, King et al. 2003). Furthermore, although certain species, such as Murray cod, Maccullochella peelii peelii , have been demonstrated to initiate spawning after a particular temperature threshold has been reached (Humphries 2005; Koehn 2006), other species appear to demonstrate more flexible recruitment in relation to factors such as temperature and flow in Australian dryland systems (Humphries et al. 2002; Balcombe and Arthington 2009). In areas at the geographical extreme of freshwater fish habitation, such as desert rivers, flow of any magnitude is an erratic and highly unpredictable event, and it can therefore be assumed that the reproductive strategies of fish are adapted to ensuring species survival in the absence of regular flows.

Studying the breeding cycles and recruitment of fish species in natural systems is difficult as it is contingent upon sampling being conducted at appropriate spatial and temporal scales. Historically, this situation has occasionally been circumvented by removing animals from the wild and conducting controlled experiments on captive populations (Lake 1967; Beumer 1979). Although these studies have frequently demonstrated that certain species exhibit a breeding response to either elevated flow or temperature (Lake 1967; Llewellyn 1973), their application to understanding recruitment processes in natural systems is questionable as environmental conditions are always far more diverse in the wild than under culture situations. This is especially so in arid-zone river systems, where seasonal changes in temperature and day length are relatively predictable but hydrological changes are extremely erratic (Puckridge et al . 1998).

The practice of using sectioned calcified body parts such as scales (Robillard and Marsden 1996), fins (Cass and Beamish 1983) and most frequently otoliths (ear bones) (Secor et al. 1995; Mallen-Cooper and Stuart 2003) that produce periodic

154 growth increments has become a common method for ageing fish (Campana 2001). These techniques have enabled researchers to calculate age in a similar manner to counting rings on tree trunks, and thus birthdates and ages (days, weeks, months or years) have been determined for many marine and freshwater species. Within the Lake Eyre Basin in central Australia, otolith age analysis has been applied to yellowbelly (Pritchard 2004), however the study was focused on adult (as opposed to juvenile) fish. Despite the popularity of ageing studies based on examination of otoliths and other structures, Campana (2001) highlights two major sources of error associated with these methods, specifically process error associated with the examined structures, and error associated with the subjectivity required and the variation of interpretation between laboratories. There are also practical considerations that limit the application of this technique. In order for otoliths (in particular) to be extracted, fish must be killed, making this technique unsuitable for rare species, those that are rarely collected or those that live in isolated areas where repeat sampling over time is envisaged. Additionally, establishing an adequate sample size – particularly for species that may exhibit protracted or serial spawning behaviour – is likely to be difficult. For example, unless all cohorts of a serially-spawning fish are sampled, inaccurate conclusions regarding the life history of the species involved are likely to be drawn. Additionally, for fish species with an extended geographical range, samples should ideally include representatives of all populations, a potentially very expensive and almost impossible task. These considerations, combined with the unsuitability of otolith-based ageing for certain species (Allen et al. 2005) and the problems associated with validating daily increments (Campana 2001; Peterson 2003), indicate that this technique is unlikely to deliver results that could enhance the present study, where the aim is primarily to investigate recruitment trends of multiple species across a wide geographic area. More specifically, the current study is primarily concerned with detecting the presence or absence of juvenile cohorts in specific habitats through time as an indicator of antecedent breeding behaviour.

Studies relating to the reproduction and recruitment of fish frequently employ calculation of the gonadosomatic index (GSI) of sampled specimens, where the weight of the gonad is expressed as a percentage of total body weight (Bishop et al. 2001; Pusey et al. 2004). A number of classifications of maturity stages exist, generally ranging from immature (commonly Stage 1) through to ripe or spent

155 (commonly stages 4 to 7), the implication being that a ripe fish is ready to – or about to – spawn, whereas juvenile and developing fishes are not sexually mature (Kesteven 1960; Nikolsky 1963; Pollard 1972; Beumer 1979; Pusey et al . 2004). Studies incorporating GSI generally relate maturity stage to time of year and occurrence of flows and other factors such as temperature and photoperiod when making assertions regarding the reproductive activity of fish (Pusey et al. 2004). The calculation of GSI has the potential to demonstrate spawning times for fish species that spawn regularly, but the success of this method may be compromised for species that employ multiple or protracted spawning strategies. During recent sampling of yellowbelly in the in the Australian Murray-Darling Basin, for example, fish that had already spawned were found to exhibit elevated GSI values, suggesting they were still in spawning condition (Clayton Sharpe, personal communication). Results such as these indicate that, like otolith examination, GSI calculation may deliver inaccurate results if applied across multiple species and catchments in an effort to detect recruitment trends. Additionally, and again in common with otolith extraction, the calculation of GSI necessitates the destruction of sampled specimens. In the context of this study, where comparatively few individuals may exist in isolated waterholes, removal of members of the population is not desirable, especially when repeated population sampling along a temporal gradient is planned. The current study, which aims to present an overview of the recruitment patterns of as many present fish species as possible across eight arid-zone catchments, has therefore focused on analysis of length-frequency distributions (Nunn et al. 2002; Balcombe et al. 2007; Balcombe and Arthington 2009) rather than employing destructive techniques. Despite the fact that there is often a large variation in fish size in relation to age (Mallen-Cooper and Stuart 2003), this technique is particularly well-suited to the study of fish communities in isolated habitats, where it is possible to detect and track the presence of cohorts in spatial and temporal samples.

Multiple sampling of riverine sites across a temporal gradient has the potential to demonstrate recruitment patterns of extant fish species, especially in the arid and semi-arid zones where waterholes exist as discrete and disconnected refuges for the majority of the time. The extent to which recruitment may be linked to either seasonal or hydrological factors can also be inferred, particularly if distinctive abiotic variation occurs and is measured (Puckridge et al . 2000). Although the growth rates of

156 poikilothermic animals such as fish have been demonstrated to slow during colder periods of the year, new recruits are detectable in samples as long as standardised sampling equipment is used (Nunn et al. 2002). Carrying out such work in geographically isolated areas such as the Queensland Lake Eyre Basin necessitates certain compromises with regard to sampling frequency and a concomitant reduction in the representation of temporal events and patterns. However, this is ameliorated to a degree by the nature of the isolated waterholes themselves: a waterbody with no inflows that is physically separated from all others is more likely to yield useful data relating to new recruits than a connected river reach that may deliver and/or receive new recruits from another location, and over time, patterns are likely to become obvious, especially if individual hydrological connection events occur and are noted.

The current body of knowledge pertaining to fish recruitment in the Queensland Lake Eyre Basin is small (Puckridge 1999 and Puckridge et al. 2000; Pritchard 2004; Arthington et al . 2005; Balcombe and Arthington 2009). Pritchard (2004) used otolith examination to demonstrate that golden perch recruitment was stronger in flood than non-flood years. Balcombe et al. (2007) concluded that access to, and usage of floodplain habitats in the Cooper Creek catchment was evident for almost all present species, but that this usage was likely to vary with life history stage, with some, but not all species occurring in floodplain habitats as larvae, juveniles and adults. Balcombe and Arthington (2009) demonstrated that bony bream and yellowbelly are likely to breed continuously in Cooper Creek, that silver tandan breed seasonally, and that flow is likely to be a pre-requisite for the recruitment success of Hyrtl’s tandan. Indeed, despite the fact that most studies indicate that population booms occur following flooding, recruitment of most species also appears to be successful, albeit reduced, in drier years (Costelloe et al . 2004; Arthington et al. 2005; Balcombe and Arthington 2009). To-date, sampling-based studies of recruitment patterns of fish in the Lake Eyre Basin have generally been confined to the Cooper Creek catchment and have not included consideration of more westerly catchments such as the Georgina.

In the neighbouring Bulloo-Bancannia Basin, no studies exist regarding the recruitment of fish, and this basin, along with the Western Plateau and areas of the Kimberley and Gulf of Carpentaria, can be regarded as one of the most under-studied freshwater systems in Australia. The current level of interest in detailing the life

157 histories of fish in un-regulated areas is high, particularly as river regulation and development may have obscured these natural patterns in many more developed areas of the world (Bunn and Arthington 2002; Kingsford 2006a; Humphries and Winemiller 2009). Results from studies in the Murray-Darling Basin suggesting that river regulation is likely to suppress or inhibit the recruitment potential of native species (Humphries et al. 2002) further encourage investigation of fish recruitment in unregulated systems such as the rivers of the Queensland Lake Eyre and Bulloo- Bancannia basins.

In light of the paucity of data relating to the recruitment of fish in Australian arid zone river systems and the identified need (Puckridge 1999) to investigate fish recruitment in un-regulated rivers, the current study addresses the following hypotheses within and between the Bulloo, greater Cooper, Diamantina and Georgina/Mulligan catchments in far western Queensland.

1. The recruitment of some or all present species of fish will occur at local scale (within waterhole) during no-flow periods in the rivers of far western Queensland. 2. Fish (some or all present species) recruitment will be enhanced by periods of flow and/or flooding in the rivers of far western Queensland. 3. There will be a seasonal recruitment response by fish species in the rivers of far western Queensland.

158 6.2 Methods

Studies relating to the length-frequency patterns of all fish species were conducted in waterholes of the Bulloo-Bancannia and Lake Eyre basins as described in Chapter 4. Field methods were identical to those described in Chapter 5. Three waterholes were sampled in each catchment during each sampling period with the exception of the Mulligan catchment, where waterholes evaporated quickly following a filling event in January 2007. Consequently, three waterholes were sampled in the Mulligan catchment in April 2007, two in August 2007 and one in November 2007. By March/April 2008 the Mulligan catchment was dry in its entirety.

Standard length measurement data for each fish species at each site on each sampling occasion were combined for the three sampling methods used (large fyke nets, small fyke nets and larval trawl nets – see Chapter 5). The collections of length data for most fish species were then divided into four length frequency categories based on the observed maximum length of each species and the assumptions of prior studies (see Gehrke et al. 1999; Puckridge et al . 2000; Arthington et al. 2005; Balcombe et al. 2006; Balcombe et al. 2007; Balcombe and Arthington 2009; Table 6.1). In the case of both anguilliform species (such as Cooper Creek catfish, silver tandan and Hyrtl’s tandan), and infrequently sampled large-bodied species (such as Barcoo and Welch’s grunter), the minimum size category was set at <100mm SL (Table 6.1). In the case of small-bodied species with a life-span unlikely to exceed 2 – 3 years (Pusey et al. 2004) such as Australian smelt, carp gudgeon and glassfish, three (as opposed to four) size categories were used (Table 6.1). In the majority of cases subsequent analysis was performed on samples taken in April 2007, August 2007, November 2007 and March/April 2008 in order to allow comparison of commonly-sampled species across all catchments. For species limited to the greater Cooper catchment (such as Cooper Creek catfish and Australian smelt), data from seven sampling occasions was used (September 2006, December 2006, January 2007, April 2007, August 2007, November 2007 and March/April 2008), and data from seven sampling occasions was also used to construct length-frequency histograms where these longer-term results better illustrated recruitment patterns. Complete length frequency data for all species from all sampling periods is included as Appendix 3.

159 Table 6.1. Size categories (standard length SL in millimetres) used for subsequent analysis of length frequency distributions of sampled fish species in the Lake Eyre and Bulloo-Bancannia basins, September 2006 – March/April 2008. Cooper Creek catfish Bony bream Banded Desert Australian smelt Carp Hyrtl’s tandan Yellowbelly grunter rainbowfish Glassfish gudgeons Silver tandan Spangled perch Welch’s grunter Golden goby Barcoo grunter 100 50 30 20 20 15 150 100 60 40 40 30 200 150 90 60 >40 >30 >200 >150 >90 >60

6.2.1 Data analysis

Bray-Curtis similarity matrices (Bray and Curtis 1957) using PRIMER-E Version 5 were constructed using totals for each size category where the total sample exceeded

200 individuals per species, and all data was log 10 (x+1) transformed (Clarke and Warwick 1994). In cases where total samples were lower than this figure (such as for Cooper Creek catfish, Welch’s grunter, Barcoo grunter and golden goby), statistical analysis was not attempted and results were presented as length-frequency histograms.

Ordination analyses (Clarke 1993) were performed using hybrid non-metric multi- dimensional scaling in PRIMER-E Version 5 in order to identify obvious patterns of similarity in length-frequency for each species. One-way analysis of similarities (ANOSIM) was then used to test for the influence of either flow or season on length- frequency distributions for each species using the same Bray-Curtis matrices. Results from ANOSIM calculate a test statistic ‘R’ identifying the observed differences between treatments compared with the differences among replicates within treatments (Clarke and Warwick 1994).

To test for the influence of season, sampling times were categorised as either early summer (September to December), late summer (January to April) or winter (May to

160 August) (as per Chapters 4 and 5 of this thesis; Balcombe and Arthington 2009). To test for the influence of flow, antecedent hydrology (the three months prior to sampling) was categorised as major flooding, minor to moderate flooding, within- bank connection flows or no flows. Flooding was defined as per the definitions of the Bureau of Meteorology (2008), however the following explanations clarify these classifications for the purposes of the study. 1. Major flooding: overbank flooding causing inundation of all previously dry floodplain areas and tributaries. 2. Minor to moderate flooding: overbank flooding causing inundation of some (but not all) previously dry floodplain areas and tributaries. 3. Within-bank connection flow: a flow that occurs within a river channel, linking waterholes and filling some previously dry areas. 4. No flows: zero flow in the preceeding three months.

In instances where ANOSIM revealed significant pairwise differences between the length-frequency structure of a fish species population explained by either season or flow, SIMPER analysis in PRIMER-E Version 5 was used to calculate the average dissimilarity between paired samples and allocate the contribution each size class made to this measure of dissimilarity (Clarke and Warwick 1994).

161 6.3 Results

6.3.1 Bony bream ( Nematolosa erebi )

Samples of bony bream were taken throughout the greater Cooper catchment from September 2006 to March/April 2008 and throughout the Lake Eyre and Bulloo- Bancannia basins in April, August and November 2007 and in March/April 2008. All size classes were present during all sampling occasions (Figure 6.1). Multivariate analysis of the different size classes of bony bream showed no clear clustering of sites (Figure 6.2), and Analysis of Similarities (ANOSIM) suggested that there were no differences in the size structure of bony bream across the Lake Eyre and Bulloo- Bancannia basins that could be explained by either flow or season (Table 6.2). Bony bream <50mm SL were most common in April and November 2007 across all catchments. Although the April 2007 samples were taken following major flooding in the Georgina and Mulligan catchments, samples from November 2007 were taken following drying periods in all catchments.

20 n = 6028 18

16 Apr-07

14 Aug-07

12 Nov-07

10 March and April 2008

8 % Frequency % 6

4

2

0 50 100 150

Standard length (mm)

Figure 6.1 Total percentage frequency of size classes of bony bream summed from all sampled catchments in April 2007 (light bars), August 2007 (dark bars), November 2007 (clear bars) and March/April 2008 (hatched bars).

162

Stress = 0.17

Figure 6.2 Two Dimensional NMS ordination plot of log 10 (x + 1) total catch data for bony bream in four size classes (50, 100, 150 and >150mm SL), across four sampling periods in eight catchments. Triangles = April 2007, squares = August 2007, diamonds = November 2007 and circles = March/April 2008; Grey = Bulloo catchment, pink = Kyabra sub-catchment, red = Cooper sub-catchment, orange = Thomson sub-catchment, purple = Barcoo sub-catchment, blue = Diamantina catchment, green = Georgina catchment and yellow = Mulligan sub-catchment.

Table 6.2 Summary of One-Way ANOSIM results comparing the size structure of bony bream populations throughout the Lake Eyre and Bulloo-Bancannia basins from

April 2007 – March/April 2008. All transformations log 10 (x + 1).

Factor Global R P Season 0.039 Not significant Flow 0.047 Not significant

163 There was strong evidence suggesting recent bony bream recruitment had occurred in no-flow areas, and this is best exemplified by the sample taken at Pulchera waterhole in the Mulligan catchment in November 2007 (Figure 6.3). Flow had ceased in the Mulligan catchment in February 2007 (Scott Morrison, Ethabuka station, personal communication), and the waterhole had been drying rapidly from April 2007 onwards (personal observation). During sampling in August 2007, only a single bony bream (160mm SL) was sampled.

140

120

100

80

60

40 Number of individuals

20

0 20 40 60 Standard length (mm)

Figure 6.3 Bony bream size structure from Pulchera waterhole in the Mulligan catchment in November 2007 following a drying period since February. No fish larger than 65mm SL were sampled (n = 173).

164 6.3.2 Cooper Creek catfish ( Neosiluroides cooperensis )

Samples of Cooper Creek catfish were taken at sites in the Thomson, Barcoo and Cooper catchments on all sampling occasions from September 2006 to March/April 2008 but this species was not collected in Kyabra Creek. Juvenile Cooper Creek catfish <100mm SL were only collected in April 2007 and March/April 2008, and fish between 100 and 150mm SL were also most common in samples taken during late summer (January to May; Figure 6.4).

40 n = 153 35 Sep-06 30 Dec-06 Jan-07 25 Apr-07 Aug-07 20 Nov-07 March/April 2008

% Frequency % 15

10

5

0 100 200 300

Standard length (mm)

Figure 6.4 Total percentage frequency of size classes of Cooper Creek catfish from all sampled greater Cooper catchments (Thomson/Barcoo/Cooper) summed from September 2006 to March/April 2008.

165 6.3.3 Hyrtl’s Tandan ( Neosiluris hyrtlii )

Samples of Hyrtl’s tandan were taken throughout the Lake Eyre and Bulloo- Bancannia basins in April, August and November 2007 and in March/April 2008, and from the greater Cooper catchment from September 2006 to March/April 2008. This species was absent from all samples taken in the Mulligan catchment. Although most size classes were sampled on all occasions, samples taken in August 2007 were extremely low in number and comprised only 1.3% of the total fish catch. In contrast, samples taken in March/April 2008 comprised 70% of the total catch (Figure 6.5).

50 n = 5026 45

40

35

30

25

20 % Frequency % 15

10

5

0 100 150 200 Standard length (mm)

Figure 6.5 Total percentage frequency of size classes of Hyrtl’s tandan summed from all sampled catchments in April 2007 (light bars), August 2007 (dark bars), November 2007 (clear bars) and March/April 2008 (hatched bars).

Multivariate analysis of the different size classes of the Hyrtl’s tandan populations indicates a general separation of flooded catchments such as the Bulloo and Georgina from the Diamantina and greater Cooper catchments (Kyabra, Cooper, Thomson and Barcoo) (Figure 6.6). Additionally, samples taken in March/April 2008 in both the Thomson and Cooper catchments overlapped with the area of the ordination occupied by the Bulloo and Georgina samples collected over a longer timeframe, and both of

166 these catchments similarly experienced flooding (Figure 6.6). Samples from the Georgina catchment taken during the same sampling period (March/April 2008) were associated with the more commonly-grouped low and no-flow samples from the greater Cooper and Diamantina sites collected earlier in the study (Figure 6.6).

Stress = 0.13

Figure 6.6 Two Dimensional NMS ordination plot of log 10 (x + 1) total catch data for Hyrtl’s tandan in four size classes (100, 150, 200 and >200mm SL), across four sampling periods in seven catchments. Triangles = April 2007, squares = August 2007, diamonds = November 2007 and circles = March/April 2008; Grey = Bulloo catchment, pink = Kyabra sub-catchment, red = Cooper sub-catchment, orange = Thomson sub-catchment, purple = Barcoo sub-catchment, blue = Diamantina catchment and green = Georgina catchment. The circled area in the bottom left indicates grouping of post-flood samples.

Analysis of Similarities (ANOSIM) suggested that no significant variation in the size structure of Hyrtl’s tandan populations could be explained by season, but the presence of major antecedent flooding as opposed to within-channel flows made a significant difference to the size structure of this species (Table 6.3). Populations of Hyrtl’s tandan in areas that had experienced major flooding were exclusively dominated by

167 the smaller size classes (fish under 100mm and between 100 - 150mm SL), whereas in areas where only within-channel flows had occurred small fish were consistently rare or absent (Table 6.4, Figure 6.7).

Table 6.3 Summary of One-Way ANOSIM results comparing the size structure of Hyrtl’s tandan populations throughout the Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008. All transformations log 10 (x + 1).

Factor Global R P Significant pairwise tests Season 0.049 n.s. Antecedent 0.16 0.001 Within-channel flow vs major flood: 0.001 flow Within-channel flow vs no flow: 0.032

Table 6.4 SIMPER analysis comparing size frequency of Hyrtl’s tandan in relation to antecedent flow in the Queensland Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008. Abbreviations: NF: No flow, WCF: Within-channel connecting flow, MiMoF: Minor to moderate flooding, MaF: Major flooding. SL (mm) Mean abundance per sample Percent contribution to observed differences (>5%). NF WCF MiMoF MaF WCF vs MaF WCF vs NF <100 4.23 - - 198.58 31.33 14.66 100-150 67.38 2.34 34.40 61.25 27.66 30.36 151 - 200 4.85 10.43 33.50 - 22.61 29.51 >200 1.38 5.35 15.80 - 18.40 25.47

168

250000 a)

200000

Flood peak 22 Jan – 2 Feb, 2007

150000 ML/day

100000

50000

0 April 2007 August 2007 November 2007 March/April 2008

80

70 (n = 108) (n = 462)

60

50

40

30

20 % Frequency %

10

0 100 150 200 100 150 200 Standard length (mm)

b) 250000 Flood peak 23 Jan – 28 Jan 2008

200000

150000 ML/day

100000

50000

0 April 2007 August 2007 November 2007 March/April 2008

90 80 (n=35) (n=120) 70

60

50

40

30

20 % Frequency % 10

0 100 150 200 100 150 200

Standard length (mm)

Figure 6.7 Length-frequency distributions of Hyrtl’s tandan in the Georgina (a) and Cooper (b) catchments in April 2007 and March/April 2008. Plots of catchment discharge (ML/day) are taken from gauge data at Roxborough Downs (a) and Retreat (b). Source: Queensland Department of Environment and Resource Management.

169 6.3.4. Silver tandan ( Porochilus argenteus )

Samples of silver tandan were taken throughout the Lake Eyre and Bulloo-Bancannia basins in April, August and November 2007 and in March/April 2008, and throughout the greater Cooper catchment from September 2006 to March/April 2008. All size classes were present during all sampling occasions, however fish >200mm SL were rare (Figure 6.8). The largest silver tandan sampled during the study (225mm SL) was caught at Waterloo waterhole in the Thomson catchment in November 2007. Multivariate analysis of the different size classes of silver tandan demonstrates similarity of length-frequency distributions for April 2007, August 2007 and March/April 2008 and a slight differentiation of length-frequency distribution for November 2007 (Figure 6.9).

25 n = 2336

20

15

10 % Frequency %

5

0 100 150 200 Standard length (mm)

Figure 6.8 Total percentage frequency of size classes of silver tandan summed from all sampled catchments in April 2007 (light bars), August 2007 (dark bars), November 2007 (clear bars) and March/April 2008 (hatched bars).

170 Stress 0.14

Figure 6.9 Two Dimensional NMS ordination plot of log 10 (x + 1) total catch data for silver tandan in four size classes (100, 150, 200 and >200mm SL), across four sampling periods in eight catchments. Triangles = April 2007, squares = August 2007, diamonds = November 2007 and circles = March/April 2008; Grey = Bulloo catchment, pink = Kyabra sub-catchment, red = Cooper sub-catchment, orange = Thomson sub-catchment, purple = Barcoo sub-catchment, blue = Diamantina catchment, green = Georgina catchment and yellow = Mulligan sub-catchment. Grouping of sites in November 2007 is indicated in the bottom right.

Analysis of Similarities (ANOSIM) indicates that there are differences in the size structure of silver tandan populations across the Lake Eyre and Bulloo-Bancannia basins and that these differences are best explained by season (or time of year) rather than antecedent flow (Table 6.5). Samples taken in early summer (late in the dry season) had high average abundances of large fish (>150mm SL), whereas samples taken in both late summer and winter had high average abundances of smaller fish (<150mm SL) (Table 6.6).

171 Table 6.5 Summary of One-Way ANOSIM results comparing the size structure of silver tandan populations throughout the Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008. All transformations log 10 (x + 1).

Factor Global R P Significant pairwise tests Season 0.119 0.017 Early summer vs late summer (0.009) Early summer vs winter (0.001) Antecedent 0.063 ns flow

Table 6.6 SIMPER analysis comparing size frequency of silver tandan in relation to season (or sampling time) in the Queensland Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008. (Early summer = September – December, late summer = January – April, winter = May – August). SL (mm) Mean abundance per sample Percent contribution to observed differences Early Late Winter Early summer vs late Early summer vs summer summer summer winter <100 - 19.97 9.31 31.27 33.89 100-150 8.67 16.29 1.56 32.53 28.78 151 - 200 7.8 - - 26.19 26.59 >200 1.13 - - 10.01 10.74

Samples of silver tandan taken throughout the study in Kyabra Creek demonstrate that this species breeds in summer in catchments that do not receive overbank flows (Figure 6.10a). Although Kyabra Creek is an un-gauged waterway, landowners attest to the prolonged dry period between September 2006 and March/April 2008 (Bob Morrish, Springfield, personal communication). In the Thomson catchment, where flooding occurred from December 2007 – February 2008, juvenile silver tandan were more abundant in the sample from March/April 2008 (following flood recession), but were also present in samples taken in April and August 2007 (prior to flooding: Figure 6.10b).

172 a) Early summer Late summer Winter 80 50 60

70 September 2006 45 April 2007 50 40 August 2007 60 n=13 n=35 35 40 n=11 50 30 40 25 30

30 20 20 15 20 10 10 10 5 0 0 0 100 150 200 100 150 200 100 150 200

60 80 December 2006 n=103 70 50 March 2008 60 40 n=22 50 30 40

20 30 20 10 10 0 0 100 150 200 100 150 200

70 60 November 2007 x axes: Standard length (mm) 50 n=21 y axes: % Frequency 40

30

20

10

0 100 150 200

b) 250000

Flood peak Jan 24 – Jan 28 2008

200000

150000 ML/day 100000

80 50000 70 March 2008 n=216 60

50 0 April 06 Jan 07 Dec 07 – Feb 08 40 30 70 December 2006 60 20 60 August 2007 n=21 50 10 50 n=8 0 40 40 100 150 200 30 30

20 20

10 10

0 0 100 150 200 80 100 150 200 60

70 April 2007 50 November 2007 60 n=19 40 n=65 50 40 30

30 20 20 10 10 0 0 100 150 200 100 150 200

Figure 6.10 Length frequency distributions of silver tandan through time in Kyabra Creek (a: top) and the Thomson River (b: bottom). No hydrograph is available for Kyabra Creek as this is an un-gauged river. Kyabra Creek did not flood during the study period (Bob Morrish, Springfield, personal communication).

173 Silver tandan populations from the western catchments such as the Georgina and Mulligan were almost exclusively composed of fish <150mm SL for the duration of the study and were common in samples taken following the recession of floodwaters (April and August 2007) (Figure 6.11).

a) 250000

200000

Flood peak 22 Jan – 2 Feb, 2007

150000 ML/day

100000

50000

0 April 2007 August 2007 November 2007 March/April 2008 b) 200 180 n =n=348 348 160

140 Apr-07 120 Aug-07 Nov-07 100 March/April 2008 80

60

40

20

0

180 c) 160 n = 285 140 Apr-07 120 Aug-07

Numbers of individuals of Numbers Nov-07 100

80

60

40

20

0 100 150 200

Standard length (mm)

Figure 6.11 Hydrograph of the Georgina River at Roxborough Downs (a), silver tandan sampled in the Georgina catchment from April 2007 to March/April 2008 (b), and silver tandan sampled in the Mulligan catchment from April 2007 to November 2007 (c). The Mulligan dried completely shortly after the November sample was taken.

174 6.3.5. Australian smelt ( Retropinna semoni )

Australian smelt were sampled on all occasions from September 2006 to March/April 2008 within their distributional range in the greater Cooper catchment. Multivariate analysis of the different size classes of Australian smelt populations display an obvious grouping of early summer sampling times (September and December 2006, and November 2007) in the top left quadrant of the ordination space (Figure 6.12).

Stress = 0.11

Figure 6.12 Two Dimensional NMS ordination plot of log 10 (x + 1) total catch data for Australian smelt in three size classes (20, 40, >40mm SL), across seven sampling periods in three catchments. Crosses = September 2006, Stars = December 2006, Plus symbols = January 2007, \triangles = April 2007, squares = August 2007, diamonds = November 2007 and circles = March/April 2008; Pink = Kyabra sub-catchment, red = Cooper sub-catchment, orange = Thomson sub-catchment.

Analysis of Similarities (ANOSIM) and SIMPER suggest that the presence of any antecedent over-bank flows appears to have a negative impact on Australian smelt populations or their catchability (Table 6.7). Juvenile smelt were more abundant in no-flow areas and smelt between 20 and 40 mm (SL) were more abundant in areas that had received within-channel flows (Tables 6.7 and 6.8).

175 Significant variation in the size structure of Australian smelt populations could be explained by season (sampling time), with juvenile fish <20mm (SL) more common in samples taken in early summer than at any other times (Tables 6.7 and 6.9). The absence of Australian smelt <20mm (SL) in samples taken in April 2007, August 2007 and March/April 2008 suggests that this species breeds immediately following the winter period in the Cooper Creek catchment and that juvenile fish are therefore likely to be present from September until mid-summer (Figure 6.13). The observation that adult Australian smelt displaying breeding colouration were only sampled in September 2006 and August 2007 gives further support to the notion that this species has a breeding season in the Cooper Creek catchment commencing in mid to late winter. Juvenile smelt were absent in April 2007 and March/April 2008, indicating that the recruitment period is over by late summer for this species (Figure 6.13).

Table 6.7 Summary of One-Way ANOSIM results comparing the size structure of Australian smelt populations throughout the greater Cooper Creek catchment from

September 2006 – March/April 2008. All transformations log 10 (x + 1).

Factor Global R P Significant pairwise tests Season 0.156 .003 September 06 vs April 07 (0.001) (Time) September 06 vs August 07 (0.003) September 06 vs March/April 08 (0.001) December 06 vs August 07 (0.004) December 06 vs March/April 08 (0.016) January 07 vs August 07 (0.046) Antecedent 0.106 0.016 Within-channel flow vs no flow (0.008) flow

176 Table 6.8 SIMPER analysis comparing size frequency of Australian smelt in relation to antecedent flow in the Cooper Creek catchment September 2006 – March/April 2008. Abbreviations: NF: No flow, WCF: Within-channel connecting flow, MiMoF: Minor to moderate flooding, MaF: Major flooding. SL (mm) Average abundance per sample Percent contribution to observed differences (>5%). NF WCF MiMoF MaF NF vs WCF <20 18.42 5.96 - - 39.24% 20 - 40 16.08 28.48 - 5.33 40.03% >40 2.04 2.7 - - 20.73%

Table 6.9 SIMPER analysis comparing size frequency of Australian smelt in relation to sampling time in the Cooper Creek catchment September 2006 – March/April 2008. Average abundance per sample SL (mm) Sep 06 Dec 06 Jan 07 Apr 07 Aug 07 Nov 07 Mar 08 <20 46.14 9.00 8.14 - - 13.63 - 20 - 40 15.29 12.6 27.00 62.00 21.88 16.13 4.2 >40 - - - - 3.13 - 1.4 Percent contribution to observed differences (>5%) SL (mm) Sep vs Apr Sep vs Aug Sep vs Mar Dec vs Aug Dec vs Mar Jan vs Aug <20 51.43 53.93 61.91 30.64 39.66 16.20 20 - 40 41.12 36.71 28.64 61.10 48.08 65.63 >40 - - - - 12.26 18.17

177 60 n=599

50

40

30

%Frequency 20

10

0 Sep 06 Dec 06 Jan 07 Apr 07 Aug 07 Nov 07 Mar/Apr 08

Figure 6.13 Seasonal distribution of Australian smelt ≤20mm (SL) from combined sites in the greater Cooper catchment from September 2006 – March/April 2008.

6.3.6. Desert rainbowfish ( Melanotaenia splendida tatei )

Samples of desert rainbowfish taken between April 2007 and March/April 2008 in the Queensland Lake Eyre and Bulloo-Bancannia basins included representatives of all size classes on all sampling occasions (Figure 6.14). Samples taken in March/April 2008 included the highest number of juvenile fish (Figure 6.14). The overwhelming majority of desert rainbowfish were sampled in the Georgina catchment (Figure 6.15).

178 20 n = 3148 18 Apr-07 16

14 Aug-07

12 Nov-07

10 March and April 2008 8

% Frequency % 6

4

2

0 20 40 60 Standard length (mm)

Figure 6.14 Total percentage frequency of size classes of all desert rainbowfish collected from all sampled catchments in April 2007 (light bars), August 2007 (dark bars), November 2007 (clear bars) and March/April 2008 (hatched bars).

600

500

400

300

200 Number of individuals of Number 100

0 20 40 60

Standard length (mm)

Figure 6.15 A comparison of the number of desert rainbowfish sampled during March/April 2008 from all sites in the greater Cooper catchment (including sites in the Barcoo, Thomson, Cooper and Kyabra catchments; grey bars, n=36), and from all sites in the Georgina catchment (black bars, n=1432).

179 Multivariate analysis of the different size classes of desert rainbowfish showed no clear grouping of sites in relation to either flow or season, however sites in the Georgina, Mulligan, Bulloo and Kyabra catchment formed distinct groups in the centre of the plot (Figure 6.16). Analysis of Similarities (ANOSIM) suggested there were no differences in the size structure of desert rainbowfish populations across the Lake Eyre and Bulloo-Bancannia basins that could be explained by either flow or season (Table 6.10).

Stress 0.13

Figure 6.16 Two Dimensional NMS ordination plot of log 10 (x + 1) total catch data for desert rainbowfish in four size classes (20, 40, 60 and >60mm SL), across four sampling periods in eight catchments. Triangles = April 2007, squares = August 2007, diamonds = November 2007 and circles = March/April 2008; Grey = Bulloo catchment, pink = Kyabra sub-catchment, red = Cooper sub-catchment, orange = Thomson sub-catchment, purple = Barcoo sub-catchment, blue = Diamantina catchment, green = Georgina catchment and yellow = Mulligan catchment.

180 Table 6.10 Summary of One-Way ANOSIM results comparing the size structure of desert rainbowfish populations throughout the Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008. All transformations log 10 (x + 1).

Factor Global R P Season 0.039 Not significant Flow 0.026 Not significant

The population of desert rainbowfish in the Georgina catchment was characterised by comparatively large numbers of most size classes on all sampling occasions except August 2007, when a total of only seven (7) individuals was collected. Desert rainbowfish <20mm SL were most common in samples taken in March/April 2008, at least 13 months after the recession of floodwaters from the January 2007 flood event (Figure 6.17).

250000

200000

Flood peak 22 Jan – 2 Feb, 2007

150000 ML/day

100000

50000

0 April 2007 August 2007 November 2007 March/April 2008

80 80 45

70 n=641 70 n=381 40 n=1432 35 60 60 30 50 50 25 40 40 20 30 30 15 20 20 10

% Frequency 10 10 5

0 0 0 20 40 60 20 40 60 20 40 60 Standard length (mm)

Figure 6.17 Length frequency distribution graphs of desert rainbowfish populations in the Georgina catchment in April 2007 (left), November 2007 (centre) and March/April 2008 (right) and a hydrograph of the Georgina catchment at Roxborough Downs during the study period (Courtesy Queensland Department of Natural Resources and Water).

181 6.3.7. Glassfish or Northwest Ambassis ( Ambassis sp.)

Glassfish in three size categories (<20mm, 20 – 40mm and >40mm SL) were collected during all sampling events (Figure 6.18), however glassfish were absent from the Diamantina and Cooper catchments during the current study and were also rare in Kyabra Creek (1 individual between April 2007 and March/April 2008). Neither the Diamantina nor Kyabra catchments experienced over-bank flooding during the study, and the Cooper catchment flooded during summer 2007/2008. In contrast, glassfish were most commonly sampled in the Georgina, Mulligan and Bulloo catchments, where flooding occurred in January 2007 (and again in January 2008 in the Bulloo) (Figure 6.19).

30 n = 1456 25 Apr-07 Aug-07 20 Nov-07 March/April 2008 15

10 % Frequency %

5

0 20 40

Standard length (mm)

Figure 6.18 Total percentage frequency of size classes of glassfish summed from all sampled catchments in April 2007 (light bars), August 2007 (dark bars), November 2007 (clear bars) and March/April 2008 (hatched bars).

182 a)

250000

200000 Flood peak 21 -22 Jan 2008

150000 Flood peak 25 – 26 Jan 2007 ML/day 100000

50000

0 Jan 05 Jan 06 Jan 07 Jan 08

80 100 60 70 n=3790 n=29 n=32 80 50 60 70 40 50 60 40 50 30 30 40 30 20 20 20 10 10 % Frequency % 10 0 0 0 20 40 20 40 20 40

Standard length (mm)

b) 250000

200000

Flood peak 22 Jan – 2 Feb, 2007

150000 ML/day

100000

50000

0 April 2007 August 2007 November 2007 March/April 2008

100 60 90 n=361 x axes: % Frequency 80 50 n=78 70 y axes: Standard length (mm) 60 40

50 30 40 30 20 20 10 10 0 0 20 40 20 40

100 70 90 n=486 60 n=139 80 70 50 60 40 50 40 30 30 20 20 10 10 0 0 20 40 20 40

Figure 6.19 Percentage frequency of size classes of glassfish in relation to antecedent hydrology in the Bulloo (a; top) and Georgina (b; bottom) catchments during the current study. (Only 6 glassfish were sampled from the Bulloo in August 2007 and these results have been omitted from this diagram).

183 Multivariate analysis of the different size classes of glassfish demonstrates the abundance of this species in the Georgina, Mulligan and Bulloo catchments compared with all other locations (Figure 6.20). Sites from the Thomson and Kyabra catchments, where collections of this species were low, are grouped together in the lower right corner of the plot (Figure 6.20). Analysis of Similarities (ANOSIM) indicated that there were differences in the size structure of glassfish across the Lake Eyre and Bulloo-Bancannia basins that could be explained by flow but not by season (Table 6.11). SIMPER results demonstrate that in catchments where major flooding or within-channel flows occurred, recruitment of juvenile glassfish <20mm SL was higher, whereas no-flow areas were dominated by adult fish >40mm SL (Table 6.12).

Stress 0.09

Figure 6.20 Two Dimensional NMS ordination plot of log 10 (x + 1) total catch data for glassfish in three size classes (20, 40 and >40mm SL), across four sampling periods in six catchments. Triangles = April 2007, squares = August 2007, diamonds = November 2007 and circles = March/April 2008; Grey = Bulloo catchment, pink = Kyabra sub-catchment, orange = Thomson sub-catchment, purple = Barcoo sub- catchment, green = Georgina catchment and yellow = Mulligan catchment.

184 Table 6.11 Summary of One-Way ANOSIM results comparing the size structure of glassfish populations throughout the Lake Eyre and Bulloo-Bancannia basins from

April 2007 – March/April 2008. All transformations log 10 (x + 1).

Factor Global R P Significant pairwise tests Season 0.005 ns Antecedent 0.178 0.023 Within-channel flow vs major flood: 0.013 flow Within-channel flow vs minor/moderate flood: 0.018 Within-channel flow vs no flow: 0.001

Table 6.12 SIMPER analysis comparing size frequency of glassfish populations in relation to antecedent flow in the Queensland Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008. Abbreviations: NF: No flow, WCF: Within- channel connecting flow, MiMoF: Minor to moderate flooding, MaF: Major flooding. SL (mm) Average abundance per sample Percent contribution to observed differences (>5%). NF WCF MiMoF MaF WCF vs MaF WCF vs MiMoF WCF vs NF <20 - 5.67 - 5.78 26.41 41.77 16.93 20-40 21.79 5.00 2.4 38.56 35.45 29.05 37.74 >40 31.57 5.33 - 7.67 38.14 29.17 45.33

6.3.8. Yellowbelly ( Macquaria sp.)

Samples of yellowbelly were taken throughout the Lake Eyre and Bulloo-Bancannia basins in April, August and November 2007 and in March/April 2008, and throughout the greater Cooper catchment from September 2006 to March/April 2008. Yellowbelly were present in all catchments except the Mulligan. All size classes were present during all sampling occasions (Figure 6.21). Multivariate analysis of the different size classes of yellowbelly showed no clear clustering of sites (Figure 6.22), and Analysis of Similarities (ANOSIM) suggested that there were no differences in the size structure of yellowbelly populations across the Lake Eyre and Bulloo- Bancannia basins that could be explained by either flow or season (Table 6.13).

185 18 n = 1493 16

14 Apr-07 Aug-07 12 Nov-07

10 March/April 2007

8

6 % Frequency %

4

2

0 50 100 150

Standard length (mm)

Figure 6.21 Total percentage frequency of size classes of yellowbelly summed from all sampled catchments in April 2007 (light bars), August 2007 (dark bars), November 2007 (clear bars) and March/April 2008 (hatched bars).

Stress 0.14

Figure 6.22 Two Dimensional NMS ordination plot of log 10 (x + 1) total catch data for yellowbelly in four size classes (50, 100, 150 and >150mm SL), across four sampling periods in seven catchments. Triangles = April 2007, squares = August 2007, diamonds = November 2007 and circles = March/April 2008; Grey = Bulloo catchment, pink = Kyabra sub-catchment, red = Cooper sub-catchment, orange = Thomson sub-catchment, purple = Barcoo sub-catchment, blue = Diamantina catchment, green = Georgina catchment.

186 Table 6.13 Summary of One-Way ANOSIM results comparing the size structure of yellowbelly populations throughout the Lake Eyre and Bulloo-Bancannia basins from

April 2007 – March/April 2008. All transformations log 10 (x + 1). Factor Global R P Season 0.014 Not significant Flow 0.013 Not significant

In the Diamantina catchment, where no overbank flows occurred throughout the study period, yellowbelly of all size classes were present during all four sampling occasions (Figure 6.23). Across a longer time period, yellowbelly size structure remained relatively constant throughout the Thomson catchment from September 2006 to March/April 2008 despite major flooding occurring in this catchment prior to the last sampling event (Figure 6.23).

187 a) 250000

200000

150000 ML/day

100000

50000

0 Jan/Feb 2007 July 2007 Dec 2007 – Jan 2008

70 50 45 60 (n=73) (n=37) 40 50 35 40 30 25 30 20 20 15 10 10 5 0 0 50 100 150 50 100 150 70

35 60 (n=144) 30 n=35) ( 50 25 x axes: standard length (mm) 40 20 y axes: % frequency 30 15 20 10 10 5 0 0 50 100 150 50 100 150

b)

250000

Flood peak Jan 24 – Jan 28 2008

200000

150000

ML/day 40 100000 35 (n=99) 30

25 50 50000 45 (n=76) 20 40 15 35 10 30 0 5 25 April 06 Jan 07 Dec 07 – Feb 08 20 0 15 50 100 150 10 5 0 50 100 150 60 35 60

30 50 (n=44) 50 (n=36) 25 40 (n=46) 40 20 30 30 15 20 10 20

10 5 10

0 0 0 50 100 150 50 100 150 50 100 150 60 50 45 50 x axes: standard length (mm) (n=39) 40 (n=63) y axes: % Frequency 40 35 30 30 25 20 20 15 10 10 5 0 50 100 150 0 50 100 150

Figure 6.23 Yellowbelly length frequency distributions and antecedent hydrology in the Diamantina catchment in (left to right) April 2007, August 2007, November 2007 and March/April 2008 (a: top diagram), and in the Thomson catchment in (left to right) September 2006, December 2006, January 2007, April 2007, August 2007, November 2007 and March/April 2008 (b: bottom diagram). Note the major flood in the Thomson catchment December – February 2007/2008 and the absence of major flooding in the Diamantina for the entire study period. Source: Queensland Department of Natural Resources and Water

188 6.3.9. Banded grunter ( Amniataba percoides )

Banded grunters were common in the Georgina, rare in the Mulligan and absent from all other catchments during the study period. Although very few individuals (5) were collected during August 2007, samples from April and November 2007 and March/April 2008 included representatives of all size classes (Figure 6.24). Samples of banded grunter taken in both November 2007 and March/April 2008 were dominated by adult fish >90mm SL (Figure 6.24). Analysis of Similarities (ANOSIM) suggested that there were no differences in the size structure of banded grunter populations within the Georgina catchment that could be explained by either flow or season, and juvenile fish <30mm SL were present during all sampling occasions (Table 6.14; Figure 6.24).

250000

200000

Flood peak 22 Jan – 2 Feb, 2007

150000 ML/day

100000

50000

0 April 2007 August 2007 November 2007 March/April 2008

90 80 70

80 n=6970 n=197 60 n =204 70 60 50 60 50 50 40 40 40 30 30 30 20 20 20 10 10 10

% Frequency % 0 0 0 30 60 90 30 60 90 30 60 90

Standard length (mm)

Figure 6.24 Percentage frequency of size classes of banded grunter from the Georgina catchment in April 2007 (left), November 2007 (centre) and March/April 2008 (right).

189 Table 6.14 Summary of One-Way ANOSIM results comparing the size structure of banded grunter populations in the Georgina catchment from April 2007 – March/April

2008. All transformations log 10 (x + 1). Factor Global R P Season -0.035 Not significant Flow -0.032 Not significant

6.3.10. Welch’s grunter ( Bidyanus welchi )

Samples of Welch’s grunter were taken at sites in the Thomson, Barcoo, Kyabra and Cooper catchments on all sampling occasions from September 2006 to March/April 2008 and in all catchments except the Mulligan and Bulloo from April 2007 to March/April 2008. Juvenile Welch’s grunter <100mm SL were only sampled in March/April 2008 in the greater Cooper catchment(s) following flooding, and fish between 100 and 150mm SL were also most common in samples taken during late summer (Figure 6.25).

50 n = 87 45

40 Apr-07

35 Aug-07

30 Nov-07

25 March/April 2008

20 % Frequency % 15

10

5

0 100 200 300

Standard length (mm)

Figure 6.25 Total percentage frequency of size classes of Welch’s grunter collected from all sampled catchments from September 2006 to March/April 2008.

190 6.3.11. Spangled perch ( Leiopotherapon unicolor )

Samples of spangled perch were taken throughout the Lake Eyre and Bulloo- Bancannia basins in April, August and November 2007 and in March/April 2008, and throughout the greater Cooper catchment from September 2006 to March/April 2008. All size classes were present during all seasons, however fish between 50 and 150mm SL were far more common in late summer samples (April 2007 and March/April 2008) than at other times and all spangled perch of all sizes were uncommon in the winter samples (August 2007) (Figure 6.26). Multivariate analysis of the different size classes of spangled perch demonstrates grouping of sites from April 2007 and March/April 2008 in the centre of the plot (Figure 6.27). Sites where spangled perch were rare appear as outliers in the ordination, and are most commonly winter samples (Figure 6.27).

25 n = 1765 Late summer (April 2007)

Winter (August 2007)

20 Early summer (November 2007)

Late summer (March & April 2008)

15

10 % Frequency %

5

0 50 100 150

Standard length (mm)

Figure 6.26 Total percentage frequency of size classes of spangled perch summed from all sampled catchments in April 2007 (light bars), August 2007 (dark bars), November 2007 (clear bars) and March/April 2008 (hatched bars).

191 Stress 0.13

Figure 6.27 Two Dimensional NMS ordination plot of log 10 (x + 1) total catch data for spangled perch in four size classes (50, 100, 150 and >150mm SL), across four sampling periods in eight catchments. Triangles = April 2007, squares = August 2007, diamonds = November 2007 and circles = March/April 2008; Grey = Bulloo catchment, pink = Kyabra sub-catchment, red = Cooper sub-catchment, orange = Thomson sub-catchment, purple = Barcoo sub-catchment, blue = Diamantina catchment, green = Georgina catchment and yellow = Mulligan sub-catchment.

Analysis of Similarities (ANOSIM) indicates that there are differences in the size structure of spangled perch populations across the Lake Eyre and Bulloo-Bancannia basins and that these differences are explained by both season (or time of year) and antecedent flow (Table 6.15). Samples taken in early summer (late in the dry season) and following major flooding had high average abundances of fish between 50 and 150mm SL (Tables 6.16 and 6.17). In contrast, samples taken in winter and late summer, or those taken following all hydrological conditions except major flooding, were characterised by comparatively small numbers of fish of all (or most) size classes (Tables 6.16 and 6.17).

192 Table 6.15 Summary of One-Way ANOSIM results comparing the size structure of spangled perch populations throughout the Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008. All transformations log 10 (x + 1).

Factor Global R P Significant pairwise tests Season 0.135 0.008 Late summer vs winter 0.002 Antecedent 0.146 0.001 Major flood vs within-channel flow 0.001 flow Major flood vs minor/moderate flood 0.008 Major flood vs no flow 0.003

Table 6.16 SIMPER analysis comparing size frequency of spangled perch in relation to season (or sampling time) in the Queensland Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008. (Early summer = September – December, late summer = January – April, winter = May – August. Insignificant differences not included in this table.)

SL (mm) Average abundance per Percent contribution to observed sample differences. Early Late Winter Late summer vs winter summer summer <50 - - 3.44 15.89 50 - 100 3.71 20.00 3.00 29.00 101 - 150 7.71 19.55 2.44 36.00 >150 2.07 3.45 - 19.11

193 Table 6.17 SIMPER analysis comparing size frequency of spangled perch populations in relation to antecedent flow in the Queensland Lake Eyre and Bulloo-Bancannia basins from April 2007 – March/April 2008. Abbreviations: NF: No flow, WCF: Within-channel connecting flow, MiMoF: Minor to moderate flooding, MaF: Major flooding. SL (mm) Average abundance per sample Percent contribution to observed differences (>5%). NF WCF MiMoF MaF MaF vs WCF MaF vs MiMoF MaF vs NF <50 5.79 - - - 11.39 13.20 17.01 51 - 100 4.43 1.81 2.5 33.33 34.15 33.57 31.22 100 - 150 6.07 2.14 11.13 29.67 36.82 33.84 33.91 >150 - 0.86 3.75 4.61 17.64 19.39 17.86

The influence of flooded catchments was particularly evident in samples of spangled perch from the Georgina catchment (which flooded in summer 2007 but not summer 2008), and the Cooper catchment (which flooded in summer 2008 but not summer 2007: Figure 6.29). A large sample of spangled perch dominated by fish between 50 and 150mm was recorded from sites in the Georgina catchment in April 2007 following a flooding event, compared with a small sample of fish of all size classes in March/April 2008 following an extended period of no flows (Figure 6.29a). A nil sample of spangled perch was recorded from all sites in the Cooper catchment in April 2007 following a comparatively dry summer, whereas fish between 50 and 150mm SL dominated the samples following flooding in March/April 2008 (Figure 6.29b).

194 a) 250000

200000

Flood peak 22 Jan – 2 Feb, 2007

150000 ML/day

100000

50000

0 April 2007 August 2007 November 2007 March/April 2008

70 90 n=476 n=48 80 60 70 50 60 40 50

30 40 30 20

% Frequency % 20 10 10 0 0 50 100 150 50 100 150 Standard length (mm)

b)

250000

Flood peak Jan 24 – Jan 28 2008

200000

150000 ML/day 100000

50000

0 April 06 Jan 07 Dec 07 – Feb 08 60 n=177 50

Nil sample, 40 (April 07) 30

20 % Frequency % 10

0 50 100 150 Standard length (mm) Figure 6.29 Length frequency of spangled perch populations in the Georgina catchment in April 2007 and March/April 2008 (a; above) and in the Cooper catchment during the same sampling event (b; below).

195 Evidence of recent recruitment of juvenile spangled perch in no flow areas was noticeable in the Mulligan catchment (Figure 6.30). At S Bend Gorge, spangled perch <50mm SL dominated the sample of this species taken in August 2007, approximately two weeks before the waterhole dried completely (Len Rule, Craven’s Peak station, personal communication: Figure 6.30a). In November 2007, spangled perch <50mm SL were the only representatives of this species present at Pulchera waterhole shortly before it dried (personal observation: Figure 6.30b).

45 a) 40 35 30 25 20 15 10 5 0 50 100 150

b) 30 25

20 Number of individuals of Number

15

10

5

0 50 100 150 Standard length (mm)

Figure 6.30 The size structure of spangled perch populations in drying waterholes in the Mulligan catchment at S Bend Gorge in August 2007 (a; top) and at Pulchera waterhole in November 2007 (b; bottom).

196 6.3.12. Barcoo grunter ( Scortum barcoo )

Samples of Barcoo grunter were taken at sites in the Thomson, Barcoo, Kyabra and Cooper catchments on all sampling occasions from September 2006 to March/April 2008 and in all catchments except the Bulloo from April 2007 to March/April 2008. Juvenile Barcoo grunter <100mm SL were only sampled in April 2007 and March/April 2008 in catchments that had experienced a major flood (the Georgina in 2007 and the Cooper/Thomson/Barcoo in 2008). Fish between 100 and 150mm SL were also most common in samples taken during late summer in previously flooded catchments (Figure 6.31).

30 n = 116

25

Apr-07

20 Aug-07

Nov-07

15 March/April 2008 % Frequency % 10

5

0 100 150 200

Standard length (mm)

Figure 6.31 Total percentage frequency of size classes of Barcoo grunter populations from all sampled catchments from September 2006 to March/April 2008.

6.3.13 Golden goby ( Glossogobius aureus )

Samples of golden goby were taken at sites in the Georgina catchment in April 2007, November 2007 and in March/April 2008 and single individuals were sampled from the Diamantina catchment in April 2007 and November 2008. Juvenile golden goby <50mm SL were only collected in April 2007 and March/April 2008, and fish between 50 and 100mm SL were also most common during both late summer sampling events (Figure 6.32). The Georgina flooded prior to April 2007 but then

197 remained on a drying trajectory for the remainder of 2007 and up until the March/April and November 2008 samples were taken.

35 n = 66

30 Apr-07 25 Aug-07 Nov-07 20 March/April 2008

15 % Frequency %

10

5

0 50 100 150

Standard length (mm)

Figure 6.32 Total percentage frequency of size classes of golden goby populations from sites in the Georgina catchment from April 2007 to March/April 2008.

6.3.14 Carp gudgeon ( Hypseleotris sp.)

Samples of carp gudgeon were taken throughout the greater Cooper catchment from September 2006 to March/April 2008 and from the Bulloo catchment from April 2007 to March/April 2008. All size classes were present during all sampling occasions (Figure 6.33). Multivariate analysis of the different size classes of carp gudgeon showed no clear grouping of sites (Figure 6.34), and Analysis of Similarities (ANOSIM) suggested that there were no differences in the size structure of carp gudgeon populations across the greater Cooper Creek catchment (Thomson/Barcoo/Cooper/Kyabra) and Bulloo-Bancannia basins that could be explained by either flow or season (Table 6.18).

198 35 n = 229

30 Apr-07 25 Aug-07 Nov-07 March/April 2008 20

15 % Frequency % 10

5

0 15 25

Standard length (mm)

Figure 6.33 Total percentage frequency of size classes of carp gudgeon populations from all sampled catchments in April 2007 (light bars), August 2007 (dark bars), November 2007 (clear bars) and March/April 2008 (hatched bars).

Stress: 0.12

Figure 6.34 Two Dimensional NMS ordination plot of log10(x + 1) total catch data for carp gudgeon in three size classes (15, 25 and >25mm SL), across seven sampling periods in five catchments. Triangles = April 2007, squares = August 2007, diamonds = November 2007 and circles = March/April 2008; Grey = Bulloo catchment, pink = Kyabra sub-catchment, red = Cooper sub-catchment, orange = Thomson sub- catchment and purple = Barcoo sub-catchment.

199

Table 6.18 Summary of One-Way ANOSIM results comparing the size structure of carp gudgeon populations throughout the greater Cooper Creek catchment and Bulloo-Bancannia basin from September 2006 – March/April 2008. All transformations log 10 (x + 1). Factor Global R P Season -0.004 Not significant Flow 0.033 Not significant

In the Kyabra catchment, where no overbank flows occurred throughout the study period, juvenile carp gudgeon <15mm SL were present during all seven sampling occasions with the exception of November 2007 (Figure 6.35a). Across a shorter time period, carp gudgeon size structure remained relatively constant throughout the Barcoo catchment from April 2007 to March/April 2008 despite major flooding occurring in this catchment prior to the first and last sampling events (Figure 6.35b).

200

50 70 45 Sep 06 Aug 07 a) 60 40 n=50 n=17 35 50

30 40 25 20 30 15 20 10 10 5 0 0 60 Dec 06 100 50 90 Nov 07 n=21 80 n=9 40 70 60 30 50 20 40 30 10 20 10 0 0 45 40 Jan 07 80 35 n=5 70 Mar 08 % Frequency % 30 60 n=10 25 50 20 40 15 30 10 5 20 0 10 80 0 70 Apr 07

60 n=34 15 25

50 40 Standard length (mm) 30

20

10

0 15 25

b)

250000

200000 Flood peak 21 -22 Jan 2008

150000 Flood peak 25 – 26 Jan 2007 ML/day 100000

50000

0 Jan 05 Jan 06 Jan 07 Jan 08

100 80 80 80 90 70 70 70 80 60 60 60 70 60 Apr 07 50 Aug 07 50 Nov 07 50 Mar 08 50 40 40 40 40 n=39 30 n=38 30 n=7 30 n=7 30 20 20 20 20 10 10 10 10 0 0 0 0 % Frequency 15 25 15 25 15 25 15 25

Standard length (mm)

Figure 6.35 Carp gudgeon length frequency in the Kyabra catchment on all sampling occasions from September 2006 to March/April 2008 (a: top diagram), and in the Bulloo catchment in (left to right) April 2007, August 2007, November 2007 and March/April 2008 (b: bottom diagram). Note the major floods in the Bulloo catchment during summer 2006/2007 and summer 2007/2008. No floods occurred in the Kyabra Creek catchment during the study period (Bob Morrish, Springfield, personal communication).

201 6.4 Discussion

Results from the present study generally support the findings of previous work in Australian dryland rivers suggesting that fish reproduction and recruitment occurs during periods of reduced flows for many species (Humphries et al. 1999; Balcombe et al. 2006; Balcombe and Arthington 2009; Figure 6.36). Most of the fish species sampled were detected as juveniles in isolated waterholes throughout the study, suggesting that these species are capable of completing their life histories within an individual waterhole even when dry conditions persist, and that reproductive effort is most likely to occur over an extended timeframe irrespective of channel or flood flows. Although this finding is not surprising given the unpredictable hydrology of the catchments, it nevertheless brings into sharp focus a major difference between Australian fish in arid and semi-arid systems and fish species of the northern hemisphere and tropics (Welcomme 1985; Junk et al . 1989). These differences have been alluded to in prior Australian studies in both the Lake Eyre and Murray-Darling basins (Puckridge 1999, Humphries et al. 1999 and 2002, Meredith et al. 2002, Balcombe et al . 2006; Balcombe and Arthington 2009), however the present study is significant due to its wide geographic range, the number of catchments sampled, the variability in hydrology among these catchments over time and the number of species considered. Nowhere is the non-reliance on flooding better exemplified than in Kyabra Creek, a comparatively small drainage in the eastern Cooper catchment. Despite an absence of minor to major flooding between September 2006 and March 2008, populations of most of the common species, such as bony bream, silver tandan, yellowbelly, spangled perch and carp gudgeons, were found to recruit independently of the antecedent flow hydrograph. In the Mulligan River, an ephemeral watercourse situated on the north-eastern edge of the Simpson Desert, species such as bony bream and spangled perch were found in juvenile size classes shortly before waterholes dried completely. These observations indicate that these common and widespread species possess an opportunistic life history strategy enabling them to breed independently of channel and flood flows and in all seasons. It is noteworthy that a study in the temperate Brazos River, Texas, also demonstrated that low-flow recruitment and the importance of isolated habitats as fish nurseries characterise fish communities in temporarily disconnected river systems (Zeug and Winemiller 2008).

202 It should be re-iterated that the rivers studied are unregulated and that alien fish species have not yet become established in potentially threatening numbers in far western Queensland. 9 Between September 2006 and March/April 2008 only 10 individual alien fish were collected, comprising six gambusia and four goldfish. The Lake Eyre and Bulloo-Bancannia basins therefore represent the most intact natural aquatic ecological systems in inland Australia and ideal areas in which to conduct ecological and recruitment research (such as that described here). In light of this, the Georgina flood in 2006/07 and the Cooper flood in 2007/08 are illuminating due to their differences rather than their similarities. In the Georgina, populations of species such as desert rainbowfish, glassfish, spangled perch and bony bream increased following the flooding period, but this result was not obvious for species such as yellowbelly, Barcoo grunter and Welch’s grunter. In the Cooper a year later, populations of desert rainbowfish and glassfish did not appear to increase, but there was an obvious response from Hyrtl’s tandan and less dramatic population rises for Welch’s grunter and Cooper Creek catfish. These results demonstrate the variable responses to hydrologic history and events that may be displayed by individual species and the full fish assemblage, both within and among catchments and within and between waterholes. This data suggests that the biological response to flooding evinced by the fishes of different catchments in different years is possibly as stochastic as the flood events themselves, and that predicting the responses of individual species may be very difficult, especially given the limited temporal data currently available. Similar variability in fish responses to flooding has been reported by other research conducted within the Cooper catchment (Balcombe and Arthington 2009).

9 Although areas of gambusia infestation are known (such as within the spring complex at Edgbaston and in the Wilson River at Nocundra, neither gambusia nor goldfish can be considered common widespread species in far western Queensland.

203 Bony bream

Size classes of bony bream sampled throughout the Queensland Lake Eyre and Bulloo-Bancannia basins during the study revealed no obvious recruitment period, with representatives of all size classes present in all seasonal samples. The results suggest that this species breeds year-round in far western Queensland, and that recruitment is an on-going process that is not reliant on channel flows or flooding. These results support previous studies in both the Lake Eyre and north-western Murray-Darling drainages (Balcombe et al. 2006; Balcombe and Arthington 2009). It is notable that the current study found a similar peak in bony bream abundance following a period of prolonged drying across all Lake Eyre Basin catchments to that recorded by Balcombe and Arthington (2009) in Cooper Creek. Results from the study of a temperate river in Texas similarly found that recruitment was strong in isolated habitats for fecund species (Zeug and Winemiller 2008). The degree to which extended wet or dry periods would affect bony bream populations and recruitment is difficult to determine, as this species is often extremely common in both recently- flooded areas (Costello et al. 2004), and those that have experienced an extended drying period. Bony bream recruitment during extended periods of drying was most noticeable in the Mulligan catchment. Despite a limited and shrinking habitat in an isolated desert waterhole (Pulchera), juvenile bony bream were abundant in a sample taken in November 2007 shortly before the waterhole dried completely. This evidence suggests that bony bream certainly breed year-round in far western Queensland irrespective of either flow conditions or season and that this behaviour continues even as waterhole depths and volumes recede (Figure 6.36).

Cooper Creek catfish

Although Cooper Creek catfish were sampled in comparatively small numbers (total 153) from September 2006 – March/April 2008, this figure is far higher than the numbers collected during prior temporal studies in the greater Cooper catchment (Costelloe et al . 2004; Arthington et al . 2005). Cooper Creek catfish in size classes <100mm SL and between 100 – 150mm SL were present in samples taken during both post-summer periods (April 2007 and March/April 2008), suggesting that this species recruits on a seasonal cycle and that breeding activity occurs during (or

204 immediately preceeding) summer. Although juvenile Cooper Creek catfish were more abundant in March/April 2008 following widespread flooding in the Cooper catchment(s), their presence following both a wet summer (07/08) and a dry summer (06/07) indicates that absence of flooding is not likely to prevent this species from breeding (Figure 6.36). Nevertheless, the results from the study do suggest that major flooding may have a positive impact on the recruitment success of this species as individuals 100 – 150mm SL were most common in March/April 2008 following widespread flooding in Cooper Creek.

Silver tandan

Despite a comparatively large total (2336 individuals), the largest silver tandan sampled was 22.5cm SL, suggesting that the maximum size this species attains in the Lake Eyre and Bulloo-Bancannia basins is likely to be considerably less than existing literature indicates (35cm: Allen et al. 2002; 30cm: Wager and Unmack 2000). Although flooding probably enhanced the recruitment success of silver tandan in the Georgina and Mulligan catchments in late summer 2007, this result was not replicated to the same degree in the greater Cooper catchment a year later. The Cooper flood from December 2007 to February 2008 lasted for a longer time period than the Georgina flood in January/February 2007. It is possible that this situation may have created a dilution effect in the greater Cooper during the March/April 2008 sampling occasion that may have made detection of large cohorts of juvenile silver tandan difficult (if they were present). Nevertheless, the ambiguity of the results for this species from both flooded catchments is in contrast to those from non-flood catchments during the study. Despite prolonged periods of no inflows or very slight within-channel connection events, silver tandan populations in catchments such as Kyabra Creek and the Diamantina River were characterised by the presence of large fish in early summer (September to December), and smaller fish from January onwards. It therefore seems highly likely that this species breeds on an annual cycle, with spawning probably occurring in early summer, and that breeding events take place irrespective of antecedent hydrology and flow conditions (Figure 6.36). It seems equally likely that stochastic flow events and the increased habitat and migration pathways they provide almost certainly increase the recruitment success of this species. These results mirror the patterns described for this species by Balcombe and

205 Arthington (2009) in Cooper Creek, and suggest that a similar reproductive strategy is employed by silver tandan throughout far western Queensland.

Hyrtl’s tandan

Hyrtl’s tandan populations demonstrated the most obvious and pronounced response to flooding, with juveniles present in the Georgina catchment following major flooding in summer 2006/2007 and in the Cooper catchment following major flooding in summer 2007/2008. It is notable that no Hyrtl’s tandans smaller than 100mm (SL) were recorded from either the Diamantina catchment or the Kyabra Creek sub- catchment for the duration of the study. Both these areas received only within-channel connection flows as opposed to major flooding. Evidence from the current study therefore suggests that recruitment of Hyrtl’s tandan is likely to be triggered by major flooding, and that if breeding and juvenile production occurs at other times the number of successful recruits is probably extremely low. These results are in broad agreement with similar work conducted in both the Warrego River catchment in the north-western Murray-Darling Basin (Balcombe et al . 2006) and Cooper Creek (Balcombe and Arthington 2009), and indicate that Hyrtl’s tandan is most likely to recruit successfully during the summer months when flows occur in Australian dryland systems (Figure 6.36). Consequently, a series of wet summers would probably result in increased recruitment success for Hyrtl’s tandan, whereas a series of dry summers would lead to poor recruitment.

Australian smelt

Populations of Australian smelt sampled throughout the greater Cooper Creek catchment between September 2006 and March/April 2008 were characterised by the presence of juvenile fish (<20mm SL) in September 2006, December 2006, January 2007 and November 2007. Sampling undertaken during the current study did not detect an increase in Australian smelt larval abundance in relation to flow as noted in a South Australian study of the downstream Cooper at Coongie Lakes in South Australia (Puckridge et al. 2000), and samples from both 2007 (preceded by a dry summer) and 2008 (preceded by a wet summer) displayed similar length-frequency patterns. Evidence gained from the current study therefore suggests that Australian

206 smelt probably commence spawning in mid-winter (late July/August) in the Cooper catchment and that the timing of breeding events is likely to be linked with seasonal cues rather than flooding (Figure 6.36). The success of spawning and early recruitment of Australian smelt in the Cooper catchment appeared to be enhanced by stable, low-flow conditions, however an abundance of fish in the 20 – 40mm (SL) size class may be related to the occurrence of within-channel connection flows (as opposed to minor to major flooding). These results therefore indicate that maintenance of Australian smelt populations probably occurs through annual spawning irrespective of antecedent or subsequent flow events, but that post-spawning within-channel flows may result in enhanced survivorship of juveniles. Additionally, flows certainly facilitate Australian smelt migration and permit movement throughout connected areas of the system by juvenile cohorts (Puckridge et al. 2000). The data obtained from the current study therefore supports previous studies of Australian smelt in other areas of Australia and indicates that this species has a similar life history response to seasonal phenomena throughout its range (Milton and Arthington 1985; Pusey et al. 2004). The evidence gathered here suggests that wet years may enhance Australian smelt recruitment in the greater Cooper Creek catchment but that breeding itself is not contingent on elevated flows.

Desert rainbowfish

Desert rainbowfish were almost always frequently sampled in the western catchments, such as the Georgina and Mulligan, and always rarely sampled in all other catchments during the current study. In the Georgina, the strongest recruitment response for desert rainbowfish was recorded in March/April 2008, and this sampling period occurred following an extended drying phase in the catchment as floodwaters from the January 2007 flood had receded by April 2007. At the very least, these results indicate that desert rainbowfish breeding events occur during periods of no flow, and possibly suggest that recruitment during periods of no flow may be beneficial to the survivorship of this species. Alternatively, it also seems likely that major flooding (such as occurred in the Georgina in early 2007) may create conditions that allow desert rainbowfish to recruit over subsequent months regardless of ensuing climatic and flow conditions (Figure 6.36). Although it seems likely that recruitment of this species may increase during the warmer months, small numbers of juvenile

207 rainbowfish were sampled on all occasions, suggesting that this species probably breeds year-round. The sample of desert rainbowfish from the Georgina catchment provides the strongest evidence for any of the studied species that major floods may have a long-term (>12 months) effect on the recruitment success of fish species in the rivers of far western Queensland, and further that long-term monitoring and evaluation of the health of arid-zone rivers needs to be assessed over a comparatively extended timeframe (Sheldon 2005). The length of this timeframe is necessarily highly variable as major flooding can be considered an infrequent event, especially in the Georgina system and its sub-catchments (such as the ephemeral Mulligan). Puckridge et al. (2000) proposed that the flow context or flood history of a catchment was likely to exert more influence on biological processes than the size of individual flood events, and essentially that a series of floods would lead to elevated fish recruitment and general population abundance. Although the results from the Georgina catchment in 2007/08 do not contrast with this proposal, they also seem to add a caveat: that major flooding in a desert catchment will similarly elevate recruitment and general activity for an unspecified period, despite the subsequent absence of flow.

Monitoring populations of desert rainbowfish along an extended timeframe in the western catchments would be beneficial with regard to tracking the relationship between population changes and antecedent hydrology, as the results from the current study are noticeably different between catchments. Despite strong populations of large desert rainbowfish in the Mulligan catchment, juvenile cohorts were not detected in either August or November 2007. This result is in contrast to the results from the Georgina catchment, where juvenile rainbowfish were always present. It therefore seems likely that desert rainbowfish recruitment was very poor if it occurred in isolated desert waterholes, and this may possibly be due to the impact of predation by piscivorous species (such as spangled perch) or other unknown factors.

Glassfish

The results from the current study generally indicate that glassfish recruitment is linked to the occurrence of antecedent flooding, and therefore support previous work in the Lake Eyre Basin that has documented population rises and falls of this species

208 (Puckridge 1999; Arthington et al . 2005). During the current study, the apparent absence of glassfish in the Diamantina catchment and rarity in the Kyabra catchment where no major flows occurred, demonstrate that lack of elevated flows may underpin recruitment failure for this species in far western Queensland. Nevertheless, the results can be considered slightly ambiguous and warrant further explanation. Firstly, the flood responses evinced by glassfish in the Bulloo, Georgina and Mulligan catchments following flooding in summer 2007 were not mirrored in any of the greater Cooper catchments (with the possible exception of the Barcoo) following prolonged flooding from December 2007 – February 2008. It is possible that the timing of the March/April sampling in 2008 occurred at a relatively earlier time in the flood cycle than sampling in April 2007, and therefore detection of new recruits may have been hampered by receding (as opposed to receded) floodwaters. Secondly, although glassfish samples following within-channel connection flows and major flooding revealed the presence of juvenile fish, samples following minor to moderate flooding did not. This is likely explained by the comparative absence of minor to moderate flooding throughout the study area where glassfish were present from April 2007 to March/April 2008, and indicates that sampling along a longer timeframe (or across a larger area) is necessary in order to better explain the influence of flows of all types on the life cycle of this species. Thirdly, although glassfish exhibited a strong flood response, it must be noted that this species continued to recruit within the Georgina catchment during a prolonged drying period following the recession of floodwaters early in 2007. This result is very similar to that recorded for desert rainbowfish, and consequently also suggests that following a major flood event, populations of some species are elevated and are likely to continue breeding irrespective of abiotic factors such as season or flow (Figure 6.36). This result is most obvious in samples from Lake Idamea in the Georgina catchment in November 2007, where juvenile glassfish <20mm SL were present despite a long (~10 month) drying period (Chapter 4 and Appendix 3).

Yellowbelly

The results obtained for yellowbelly during the current study give further support to prior work conducted in the Lake Eyre Basin indicating that this species may breed throughout the year in the western rivers (Pritchard et al. 2004; Balcombe et al. 2007;

209 Balcombe and Arthington 2009; Figure 6.36). Indeed, it would appear that reliance on flood-pulse recruitment for the closely related M. ambigua , the Murray-Darling species/sub-species, is also unlikely (Mallen-Cooper and Stuart 2003; Pusey et al. 2004; Balcombe et al. 2006). These results are in contrast to earlier studies on Murray-Darling populations of yellowbelly that concluded flooding was necessary in order to trigger spawning in this species (Lake 1967; Cadwallader 1978, 1979).

During the current study, juvenile yellowbelly <50mm SL were sampled on all occasions, including after major flooding events and after prolonged dry periods. Despite major flooding occurring in the Georgina catchment in January/February 2007 and in the Thomson, Barcoo and Cooper catchments from December 2007 – February 2008, there was no noticeable increase in the numbers of juvenile yellowbelly in these catchments. It seems likely that yellowbelly have a similar recruitment strategy to bony bream in the Lake Eyre and Bulloo-Bancannia basins, and that juvenile fish recruit to the population relatively constantly. Evidence of no or low flow spawning of this species in the Lake Eyre Basin is most obvious in the Diamantina catchment, where flows were small and sporadic throughout the study period but the size structure of sampled populations varied little from those in flooded catchments. Despite major flooding occurring in the Georgina catchment in January 2007, yellowbelly populations did not demonstrate a detectable recruitment response, thus suggesting, as a Murray-Darling otolith study has previously, that over-bank flooding may not necessarily provide advantageous conditions for the recruitment of this species (Mallen-Cooper and Stuart 2003).

Banded grunter

Results from the current study suggest that banded grunter is likely to breed opportunistically in the Georgina system in far western Queensland. This finding is similar to previous studies of this species in both the Alligator Rivers region (Bishop et al. 2001) and Burdekin River (Pusey et al. 2004), and yields further evidence indicating that banded grunter, unlike some other Terapontid fishes, do not require elevated flows in order to reproduce (Figure 6.36). Results from the Georgina River suggest that banded grunter adults exist in high densities in isolated waterholes and that recruitment is likely to occur constantly, but that reproductive activity – and

210 catchability/activity generally – may decrease in winter. Despite evaporation and habitat shrinkage in the Georgina system in late 2007 and early 2008, all size classes of this species, including juveniles and a large number of adults, were common in the samples. The abundance of this species in all size classes is further evidence of its adaptability to the arid-zone conditions that exist in the Georgina catchment, where waterholes frequently remain on a drying trajectory for periods in excess of 12 months. Evidence from Warburton Creek (downstream Diamantina) in South Australia provides similar evidence of the persistence of this species (Costelloe et al. 2004): banded grunter colonised the area during a flood in 1991, yet the population was enduring in 2003 despite a prolonged period of drought.

Welch’s grunter

Welch’s grunter was rare in all samples taken during the current study with the exception of the post-flood samples from Cooper Creek. Despite a low sample size (87), this result supports previous assertions regarding summer-flood cued reproduction for this species (Merrick and Schmida 1984; Wager and Unmack 2000; Allen et al. 2002; Figure 6.36). Although the results from the Cooper system in March/April 2008 suggest that it is highly likely that Welch’s grunter breed during summer and that recruits are advantaged by elevated flows, cohorts of juveniles were absent in the Georgina catchment during sampling undertaken in April 2007, despite adults being present. Gathering accurate field data relating to populations of Welch’s grunter is likely to be difficult due to the low catchability of this species using the suite of sampling equipment employed during the current study, or possibly due to the uncommon status of this species within its range. Welch’s grunter were similarly rarely sampled during the Cooper Creek Dryland Refugium project which also employed a number of sampling methods (Arthington et al. 2005), and were only slightly more common in samples taken during ARIDFLO (Costelloe et al. 2004). Given the combined results of the three most recent studies (Dryland Refugium, ARIDFLO and the current study), determining the life history of Welch’s grunter in far western Queensland clearly requires a more targeted approach with increased sampling effort including use of a wider range of methods, and sampling at shorter temporal intervals in waterholes where adults are known to be present. This approach was beyond the scope of the present study.

211 Spangled perch

The presence of juvenile spangled perch in rapidly drying waterholes in the Mulligan catchment at different times of the year suggests that this species breeds opportunistically in the rivers of far western Queensland and that reproductive activity itself can occur along a comparatively constant timeframe in Australian desert rivers (Figure 6.36). Nevertheless, the increased numbers of fish between 50 – 150mm SL in all samples taken in late summer (January – April), and in areas that had experienced a major flood, indicate that successful recruitment of this species is almost certainly increased for cohorts spawned in early summer that are then able to capitalise on the conditions afforded by summer floods. Alternatively, it seems equally likely that spangled perch may possess an adaptable recruitment strategy that is generally linked to summer spawning (and thus potential flows and floods), but can also occur at other times of the year in isolated waterholes in order to capitalise on dwindling resources. The results from the current study therefore deliver similar conclusions to studies from other areas of Australia indicating that flooding, though certainly enabling an improved recruitment response, is not an essential pre-requisite for spangled perch breeding (Pusey et al. 2004).

Barcoo grunter

The sample of Barcoo grunter was comparatively small during the current study and similarly small totals were recorded by prior temporal studies in the Lake Eyre Basin (Costelloe et al. 2004; Arthington et al. 2005). Gaining accurate data relating to the life history of Barcoo grunter is likely to be achieved only through more specifc studies aimed at both this species and Welch’s grunter, a closely related Terapontid fish with a similar range. During the current study juvenile Barcoo grunter were most common in late summer samples (April 2007 and March/April 2008). The presence of small Barcoo grunter in the Mulligan catchment in April 2007 and in the Cooper catchment in March/April 2008 tends to indicate that flooding may play an important role in recruitment success for this species, as both sampling periods occurred following major floods in the respective catchments (Figure 6.36). The success of Barcoo grunter recruitment in non-flood catchments appears to be low based upon the data gained from this study, as juvenile Barcoo grunter were noticeably rare or absent

212 in all samples taken in Kyabra Creek and the Diamantina River. Thus it would appear most likely that Barcoo grunter recruitment is greatly advantaged by, if not dependent upon, summer flooding, but again, the small size of these samples precludes further speculation.

Golden goby

Although the sample of golden goby was too small to allow statistical comparisons, the noticeable presence of fish smaller than 100mm (SL) in both late-summer sampling periods indicates that this species probably breeds during summer in the Georgina catchment and that flooding is unlikely to be a pre-requisite for reproduction (Figure 6.36). The large individual sizes of golden gobies recorded from the Georgina far exceed the maxima recorded for this species in other Australian waterways (Wager and Unmack 2000; Allen et al. 2002; Pusey et al. 2004). Golden gobies from the Georgina catchment frequently exceeded 150mm SL and ranged to 195mm SL. Results from other Australian catchments have recorded golden goby to 157mm SL (Akihito and Meguro 1975, ‘northern’ Australia), 80mm SL (Kennard 1995, ) and 125mm TL (Bishop et al. 2001, Alligator Rivers). The results from the Georgina River therefore suggest that inland populations of this species are likely to reach a much larger maximum size than populations in coastal rivers, and that this species can complete its entire life cycle in freshwater. In contrast, it is generally accepted that this species has a marine larval stage in coastal catchments (Pusey et al. 2004).

Carp gudgeon

The samples of carp gudgeon collected from September 2006 to March/April 2007 give a strong indication that this species breeds constantly in the greater Cooper catchment and Bulloo River, and this result is in agreement with studies of this species in the Murray-Darling Basin suggesting that carp gudgeon recruitment success is possibly advantaged by low or zero flows (King et al. 2003). Indeed, no results for this species from any of the Cooper catchments found that carp gudgeon recruitment was enhanced by flooding, as previous studies have in South Australia (Puckridge et al . 2000). Carp gudgeon are a small-bodied benthic species and it seems very likely

213 that large and prolonged flooding events (such as the summer 2007/08 flood in the greater Cooper catchment) have the potential to dislocate rather than concentrate populations of adults. In Kyabra Creek, a comparatively small waterway in the eastern Cooper catchment that did not experience over-bank flows for the duration of the study, populations of carp gudgeon included fish <15mm SL on all but one occasion (November 2007). This suggests that typically low or no flow catchments in far western Queensland may provide stable and suitable environmental conditions that benefit the recruitment strategy of this species (Figure 6.36).

Possible implications in the Murray-Darling Basin

One of the most obvious results from the study was the constant presence of at least small numbers of most species in most waterholes on all sampling occasions and across all seasons. This suggests that small ‘source’ populations potentially survive dry periods in isolated waterholes, and that most species can breed and thus maintain their numbers despite occasionally adverse conditions (such as prolonged dry periods). This finding has several species-related ramifications, particularly for fish that naturally occur in both the far western rivers and the Murray-Darling Basin.

Small-bodied fish such as olive perchlet, Ambassis agassizii , and Murray-Darling rainbowfish, Melanotaenia fluviatilis , are generally considered to have reduced ranges in the Murray-Darling Basin, and this is most often attributed to habitat degradation, river regulation and the impact of alien species (Lintermans 2007). During the current study, populations of closely related species, Ambassi s sp. and Melanotaenia splendida tatei were occasionally found in large numbers, but in catchments unaffected by large flow events were present in most waterholes in small numbers. This indicates that under natural flow regimes, small-bodied species appear capable of surviving in isolated habitats in far western Queensland. It seems entirely plausible that small populations of similar species in Murray-Darling Basin catchments may naturally exhibit similar behaviour, but that additional stress factors (such as alien fish and flow regulation), could have negative impacts on such ‘maintenance’ populations and contribute to their local extinction. If this situation occurred throughout a reach or catchment, it is just as possible that populations of such small-bodied species could become locally extirpated and eventually extinct over a wider geographical area.

214 Olive perchlet is listed as a threatened species in the Murray-Darling Basin, and the Murray-Darling rainbowfish is considered uncommon and declining (Lintermans 2007). Results from the western Queensland catchments for closely related species suggest that alteration of the natural flow regime in the Murray-Darling Basin may be a major causal factor preventing these species from experiencing population booms such as those observed in the Georgina and Mulligan catchments in April 2007. More localised effects on source populations (such as water abstraction and the presence of alien species) may affect the ability of remnant – or parent/source – populations of such species in the Murray-Darling Basin to survive drying conditions, whereas in the Lake Eyre and Bulloo-Bancannia basins these perturbations generally do not occur. Testing such an hypothesis could be achieved by comparative studies of Ambassid and Melanotaeniid species in regulated and unregulated areas of the Murray-Darling and Lake Eyre/Bulloo-Bancannia basins through time.

Of the four Terapontid species extant in the Lake Eyre Basin, both spangled perch and banded grunter can be considered vagile and adaptable ecological generalists. The current study has also demonstrated that both of these species also possess flexible recruitment strategies including the ability to breed independently of flow events and/or detectable seasonal cues. This does not appear to be the case for either Barcoo or Welch’s grunter, as juveniles of both species were comparatively uncommon and were only sampled in late summer following over-bank flooding. The positive impacts of elevated flows on recruitment have been identified for many species of freshwater Terapontid fishes in Australia (Merrick and Schmida 1984), including silver perch, Bidyanus bidyanus a close relative of both Barcoo grunter and Welch’s grunter and the only Terapontid species from the Murray-Darling Basin. Silver perch are a threatened species (Lintermans 2007) with impacts on populations ranging from commercial fishing (Reid et al. 1997) to the EHN virus carried by alien redfin perch, Perca fluviatilis (Lintermans 2007). The survey and recruitment data from the Lake Eyre Basin for both Welch’s and Barcoo grunter suggest that these species are generally present but often uncommon, with recruitment linked to the unpredictable and sporadic summer flood cycle. Given the factors underlying the current threatened status of silver perch in the Murray-Darling Basin, it seems highly likely that any human-induced disruption(s) to the natural flow regimes of the rivers of far western

215 Queensland would have a negative impact on the ability of these Terapontid species to maintain populations.

Throughout their range, yellowbelly are a popular angling species and in the Murray- Darling Basin were a target species for commercial fishers (Reid et al. 1997) until the closure of the inland commercial fishery in 2003. Despite recreational fishing pressure, commercial fishing pressure and potential impacts associated with river regulation and alien species, yellowbelly populations did not experience as severe a decline as other large-bodied freshwater species in the Murray-Darling, such as silver perch, Murray cod, Macchullochella peelii peelii , trout cod, Macchullochella macquariensis and freshwater catfish, Tandanus tandanus . The results from the current study in rivers of the Queensland Lake Eyre and Bulloo-Bancannia basins, other studies conducted in Cooper Creek (Balcombe and Arthington 2009) and a growing amount of research in other areas (Mallen-Cooper and Stuart 2003; Pusey et al . 2004) indicate that this species is almost certainly not as reliant on flood events for successful recruitment as previous studies have reported (Lake 1967; Cadwallader 1979), and this may account for its comparative resilience. During the current study juvenile yellowbelly were present on all sampling occasions and in both flow and non-flow affected locations. The fact that yellowbelly populations in far western Queensland appear capable of relatively constant recruitment irrespective of hydrological conditions also demonstrates that natural populations of this species are robust and should not be compromised by the introduction of farmed stock under any circumstances.

Summary

The evidence gathered by the work presented above indicates that un-regulated dryland catchments that are comparatively free of alien fish species provide valuable study areas in which to observe natural recruitment patterns and population ebbs and flows for native fish species. For the majority of species, and with the exception of Hyrtl’s tandan and possibly Barcoo and Welch’s grunter, the hypothesis that fish species in the Australian arid zone breed during periods of no flow can be accepted based on the presence of juveniles (and all size classes) of most species during all sampling events conducted between 2006 and 2008. By undertaking similar temporal

216 work in similarly un-altered catchments, researchers in Australia and worldwide have the best chance of gaining fundamental data relating to the life histories of extant fish species and the ecology of the areas in which they live.

The work presented in this chapter invites comparison to previously-stated concepts describing riverine processes and their relationship to flow events. Fish recruitment in Australian arid-zone rivers has been demonstrated to be linked to flooding for some fish species, and so certainly demonstrates linkages with the Flood Pulse Concept (Junk et al. 1989). However, the biological response to the impact of elevated flows was variable among fish species, and was only demonstrable for a comparatively small number of present species. Additionally, the flood cycle was extremely irregular, with major flooding non-existent in two of the study catchments for the duration of the study (the Diamantina River and Kyabra Creek) and both spatially and temporally patchy elsewhere (see Chapter 4). Furthermore, although population booms of some species occurred following major flooding, recruitment of many species (such as glassfish and desert rainbowfish) appeared to continue following the recession of floodwaters, suggesting that major flooding in the Australian arid zone may create a long-term (as opposed to, or as well as, a short term) disturbance that may create favourable recruitment conditions for an extended period (Puckridge et al. 2000; King et al . 2003). Therefore, although flood events increase the productivity of waterholes and the availability of food resources comparatively rapidly (Bunn et al . 2003; Balcombe et al. 2005), fish recruitment - and recruitment success – seems likely to occur over a far longer timeframe. In most cases, recruitment does not appear to be solely a reactionary response to flooding itself but an on-going process that may continue entirely in the absence of flooding. Regular sampling of post-flood waterholes in arid zone catchments therefore has the potential to deliver useful data regarding the possibility of extended temporal effects of flooding on fish recruitment.

In common with the low-flow recruitment hypothesis (Humphries et al. 1999) fish recruitment in Australian arid-zone rivers was demonstrated to occur along a temporal gradient for most species that included periods of reduced flows. This was particularly obvious for species that were demonstrated to recruit seasonally, such as Australian smelt and silver tandan. Perhaps unsurprisingly, the study also supports Puckridge’s flow-pulse model (1999), and finds variability in fish recruitment response related to

217 the magnitude of flow events, such as the flood response of Hyrtl’s tandan and the more variable response of species such as glassfish and spangled perch. However, the importance of processes occurring at individual waterhole scale, particularly during drying periods, cannot be under-estimated when considering Australian desert fishes (Arthington et al . 2005), and nor can the importance of the waterholes themselves in sustaining dryland river ecosystems (Sheldon et al. in press).

A conceptual model of the recruitment behaviour of fish species in the Queensland Lake Eyre and Bulloo-Bancannia basins is presented in Figure 6.36. Most species in the Australian arid-zone were found to have a no-flow recruitment strategy and to reproduce successfully in isolated waterholes. Given the stochasticity of flow events, this seems unremarkable with regard to population maintenance, but it may also serve other life history functions. A species that has cohorts of larvae and juveniles ready to capitalise on random flow events when they occur (such as small local flows linking backwaters and waterholes), is likely to be far more successful in the arid-zone than a species that requires a flow stimulus to spawn, for flow events often involve short periods of time (days or hours). It seems highly likely that many fish species in the Australian arid zone therefore reproduce along a relatively constant timeframe such that their progeny are best-placed to utilise recently-flooded ephemeral habitats when movement and migration pathways become available. Such a strategy allows fish to capitalise on the rich food resources of flooded habitats if and when flooding occurs (Balcombe et al. 2005).

218 No-flow recruitment Flow-dependent recruitment

Seasonal Continuous

Winter Summer

Australian smelt Silver tandan Bony bream Hyrtl’s tandan Golden goby* Banded grunter Barcoo grunter* Cooper Creek catfish* Carp gudgeon Welch’s grunter* Yellowbelly Rainbowfish Glassfish** Spangled perch***

Figure 6.36 A conceptual model of the recruitment behaviour of fish species in the Lake Eyre and Bulloo-Bancannia basins based on evidence from the current study. * indicates species where the total sample was <200 individuals, ** indicates species that were found to recruit under no-flow conditions as well as demonstrating flow- related recruitment, and *** indicates species demonstrating no-flow, seasonal and flow-related recruitment.

219 7. Movement, colonisation and extirpation of fish in four catchments in far western Queensland

7.1 Introduction

Movement allows fish to locate and utilise suitable resources and to complete their life cycles in favourable habitats. Knowledge of the movement behaviour of species and the consequences for communities is therefore crucial for effective management and conservation of biodiversity (Gowan et al. 1994; Schlosser and Angermeier 1995). Nevertheless, with the exception of species such as salmonids that display extremely obvious mass movement associated with particular life cycle stages, the movement capability and variability of most freshwater fish species has remained comparatively unknown.

Studies of the community structure of fishes have demonstrated that variation occurs along stream gradients from variable shallow to stable deep habitats (Schlosser 1987), and that species richness typically increases with depth, water persistence and channel size (Capone and Kushlan 1991; Dekar and Magoulick 2007). Specifically, permanent pools have been shown to provide habitat for larger, predatory species (Schlosser 1987), whereas smaller more tolerant species inhabit shallower areas in systems prone to seasonal drying (Capone and Kushlan 1991). Capone and Kushlan (1991) therefore identified several ‘pioneer species’ based on their ability to colonise and survive in comparatively harsh conditions, and also suggested that these species are often likely to be particularly successful in areas where populations of other species are low. These observations suggest that ephemeral areas in rivers prone to drying are therefore likely to provide habitat for a subset of riverine species rather than complete assemblages similar to those from more permanent areas, and that colonisation or migration ability, combined with broad environmental tolerances, will characterise such species. The important role played by periodically-inundated or temporary habitats as nursery and feeding areas has also been demonstrated (Cucherousset et al. 2007; Zeug and Winemiller 2008), as has the role of ephemeral areas as refugia from disturbance (Magoulick and Kobza 2003; Dekar and Magoulick 2007). In an

220 experimental study, Davey et al. (2006) found that 85% of fish use refugia during periods of low flow, and that species-specific refuge-use strategies, such as upstream migration or burrowing into the substrate, may influence the structure of fish communities in these low flow areas. Davey and Kelly (2007) demonstrated that the position of refugia within the landscape plays a crucial role in determining biological responses, with vagile fish species most likely to recolonise ephemeral areas close to permanent refugia comparatively quickly. In general, the impact of disturbance (such as drought: Lake 2000; Bond et al . 2008), has been suggested to elicit biotic responses in fish communities such as migration (Love et al. 2008), which frequently also increases exposure to risk, such as predation (Thome-Souza and Chao 2004). Combinations of disturbance, responses and risk therefore lead to alterations in the structure of fish assemblages in affected systems and particularly to changes in species richness and abundance.

Mark-recapture studies have been undertaken over the last 40 years in Australian freshwater systems and have demonstrated that individual fish (such as yellowbelly, Macquaria ambigua ), may move up to 1000 or more kilometres in a given direction (Reynolds 1983). Although the reliability of data gained from this method is often compromised by the comparatively small percentage of re-captures (Fausch et al. 2002), generalisations have developed and management actions have been adopted based on the findings of such studies, an example being the presumed requirement of golden perch or yellowbelly for an upstream migration prior to spawning (Reynolds 1983). Recognition that broader recruitment strategies may exist for this species, such as spawning in response to lower or no flows, has recently emerged as a result of work completed in areas with variable flow regimes, such as the Fitzroy Basin (Roberts et al. 2008) and Cooper Creek (Balcombe and Arthington 2009; this study, Chapter 6).

The recognition that barriers to fish movement are likely to impact on the life cycles of freshwater fish has led to an increase in studies aimed at quantifying the upstream and downstream movement of fish species at such structures, particularly when fish ladders and similar engineering works have been installed (Mallen-Cooper et al. 1995). Consequently there is now a growing body of knowledge relating primarily to the upstream movement of large-bodied species of different size classes through

221 fishways (Mallen Cooper et al. 1995; Stuart and Berghuis 2002), but also investigating downstream dispersal and movement of adults and juveniles (Baumgartner 2004).

Both mark-recapture and sampling at barriers have disadvantages when attempting to draw conclusions relating to the overall movement of fish species in that they record movement at either a single point or multiple points in time but yield no information on the movement of individuals when sampling is not actively undertaken. Thus, although tagging studies have demonstrated that an individual fish may physically move over 1000 kilometres from where it was tagged (Reynolds 1983), detail regarding finer scale movement requires increased effort (Crook et al . 2001). Similarly, migrations through a fishway demonstrate that a particular cohort of fish may move during a particular sampling event (Mallen-Cooper et al. 1995), but often yield no data concerning either antecedent or subsequent movement.

Advances in tagging techniques have to some degree ameliorated the disadvantages of traditional tag-recapture and barrier sampling techniques. In particular, the use of radio and acoustic telemetry has demonstrated that large-bodied species in the Australian Murray-Darling system such as yellowbelly, Murray cod and the alien common carp, exhibit high site fidelity most of the time, and are likely to return to their home ranges even after longer migrations (Crook et al . 2001; Koehn 1996; Koehn 2004). Electronic tagging of fish species is also practiced more widely, with studies relating to commercially-important species such as Atlantic cod, Gadus morha, occurring in the North Atlantic Ocean (Righton et al. 2007) and radio-tracking of zulega, Prochilodus argenteus, in Brazil (Godinho and Kynard 2006).

It is salient to point out that initiating movement studies of fish, especially using advanced technology, is an expensive enterprise and therefore has predominantly been used on either commercially-important, endangered or pest species. Consequently, movement studies of freshwater fish from arid-zone or desert systems is particularly difficult due to the logistics of tracking animals in isolated areas where distances are frequently large, fish retrieval is unlikely and costs are prohibitive. Additionally, tagging and tracking of fish necessarily involves using a comparatively small number of individuals, and the results of such work may not reveal typical responses and

222 behaviour. Nevertheless, determining the movement behaviour of fish species and community members in the Australian arid zone is possible by monitoring ephemeral sites before and after inundation events.

Fish communities in arid and hydrologically variable environments persist for the majority of the time in isolated waterholes or sections of the main channel (or channels) of rivers that retain water following extended periods of no rainfall or flow (Hamilton et al . 2005; Arthington et al . 2005; this study). Movement possibilities for these species are thus limited to opportunities within the waterbody they inhabit for extended periods, and if re-wetting does not occur extirpation will inevitably result. Survival of fish species in such areas is thus contingent on the ability to rapidly disperse when movement is facilitated through re-connection of habitats via flow, and to colonise new areas despite suffering local extirpation in others (Fausch and Bramblett 1991). In arid zone rivers, movement facilitates colonisation of newly- inundated areas following localised rainfall or flooding, and may simultaneously facilitate local extirpation if migration pathways lead to ephemeral, rapidly drying waterholes.

Spatial distribution of fishes in the arid zone is the determining factor driving recolonisation potential (Fagan et al. 2005). If a fish species is present at catchment scale, it can therefore be considered to have re-colonisation potential at finer scales – such as reaches and individual waterholes – when connection events such as flows and flooding occur. This has been demonstrated in studies in the American south-west (Scheurer and Fausch 2003; Eby et al. 2003) where single species or groups of species were found to be present in previously dry environments following a cycle of drying, extirpation and subsequent inundation. This ‘boom and bust’ ecosystem dynamic is frequently reported for animal communities in rivers with highly variable hydrology (Sheldon and Walker 1998; Arthington et al. 2005) and is supported by observations of fish colonisation and extirpation events recorded for waterways in the Lake Eyre Basin (Ruello 1976; Puckridge 1999). In the case of fish in arid zone systems, colonisation potential is therefore contingent on populations persisting in ‘refuge’ waterholes (for example permanent main channel waterholes) such that re- colonisation can occur following re-connection, and movement is contingent on migration pathways being created by typically-unpredictable flow events. Movement

223 and colonisation of fishes in arid zone rivers are therefore linked, for without replenishing flows and renewed connectivity, extirpation is likely to occur in many localised areas.

Although the majority of present fish species have been demonstrated to move to floodplain habitats in Cooper Creek in the Queensland Lake Eyre Basin (Balcombe et al. 2007), the ability of species to move and colonise newly inundated areas when migration pathways become available requires further investigation. In particular, migration distances and the use of habitats at the extreme limits of fish habitation, such as ephemeral desert streams, remain poorly understood. Such studies, quite apart from yielding specific information that can be added to the life history data of the studied species in arid areas, may also be illuminating at a wider scale, for many of the Lake Eyre Basin species are also found in other Australian drainage basins (Allen et al . 2002; Pusey et al . 2004; Lintermans 2007), and an identified need exists to more-accurately understand ephemeral river systems (Hughes 2005; Larned et al . 2009; Sheldon et al. in press).

Presently, the vast majority of research on inland freshwater fish in Australia is concentrated in the Murray-Darling Basin, primarily because rivers in this system are comparatively highly populated, have been affected by flow regulation and are inhabited by large populations of alien fish species (Roberts et al. 1995; Lintermans 2007; Koehn and Mackenzie 2004; MDBC 2008; Rayner et al. 2009). Consequently, research effort has been focussed on ameliorating the effects of barriers to fish movement, rehabilitation of habitats, provision of environmental flows and removal or control of alien fish species (Mallen-Cooper et al. 1995; Crook and Robertson 1999; MDBC 2003 and 2004). Unfortunately, pre-disturbance data is not available for the majority of Murray-Darling Basin rivers, and defining reference points – a crucial step in river restoration - is therefore difficult (Lake 2001; Humphries and Winemiller 2009). The common forms of river perturbation (barriers to movement, flow regulation, habitat loss, alien species) are generally not present in the Lake Eyre Basin, and it therefore presents ideal opportunities for the study of movement, colonisation and extirpation of fish species and communities in a predominantly natural system (Balcombe and Arthington 2009).

224 Investigating the colonisation and extirpation of fish species in natural or near-natural Australian inland systems has the potential to shed light on many of the key questions relating to dispersal patterns and biogeography of the species concerned. ‘Pioneer species’ (Capone and Kushlan 1991) that can be shown to undertake speculative migrations to recently-inundated ephemeral systems are likely to be capable of extending their ranges; other less vagile species are likely to evince far more restricted ranges. Within the Lake Eyre Basin the fish fauna is comprised of a combination of widespread species (occurring in most catchments) and locally endemic species generally occurring in fewer catchments (see Chapter 5). Investigating the migration patterns of the fish fauna following the opening of migration pathways therefore has the potential to demonstrate which species are most likely to have actively moved through opportunistic colonisation, and which have been unable, or less able, to disperse in this fashion. Additionally, establishing the dispersal distances of colonising species has been identified as a necessary step in predicting colonisation patterns in ephemeral systems (Larned et al. 2009). Given that waterholes in ephemeral systems are only likely to hold water for a limited time, the age/size structure of colonising populations may be skewed towards juveniles (suggesting usage as a nursery area) or adults (suggesting the importance of resource and/or reproductive requirements). In order to address these questions, the current study aims to test the following hypotheses regarding fish colonisation in the rivers of far western Queensland:

1. Fish species and/or members of communities will migrate to ephemeral areas following inundation and re-connection. 2. Colonisation ability in newly-inundated areas will be related to size classes or life-cycle stage of colonising species. 3. Migration distance will influence the composition of fish communities in newly-inundated ephemeral areas. 4. Antecedent hydrology will influence the composition of fish communities at formerly dry sites. 5. Fish assemblage structure will vary among, rather than within, catchments at previously-dry ephemeral sites.

225 7.2 Methods

Studies relating to the colonisation and extinction of fish in the Australian arid zone were completed in permanent and ephemeral waterholes of the Georgina, Mulligan, Barcoo and Thomson catchments, as these catchments provided a combination of areas recently inundated by major flooding (Georgina/Mulligan) and within-channel flows (Thomson/Barcoo). Following a prolonged dry period in 2006, the Georgina and Mulligan rivers both experienced major flooding in January 2007 and presented opportunities to sample fish in previously dry areas and monitor these populations as the waterholes dried down. In the Barcoo and Thomson catchments, ephemeral sites were selected due to their geographic proximity to permanent waterholes (<5 km) and monitored following filling and drying events in order to investigate colonisation patterns following smaller connection events. Chapter 4 provides a detailed description of study sites used.

Calculation of distances between ephemeral and permanent waterholes in the Georgina and Mulligan catchments was effected using aerial photographs and maps made available by the Queensland Department of Environment and Resource Management.

Field methods were identical to those described in Chapter 5.

7.2.1 Data analysis

Data relating to capture of each fish species at each site on each sampling occasion was combined for the three sampling methods used in order to calculate catch-per- unit-effort (CPUE). All fyke net samples were standardised to a 19 hour set time (as per Arthington et al. 2005) and larval trawls were standardised to 5 minutes at each site on each sampling occasion. To examine species abundance at catchment scale, data from permanent and ephemeral waterholes in the Georgina and Mulligan rivers in April 2007, August 2007, November 2007 and March/April 2008 was analysed, whereas datasets from ephemeral and permanent waterholes in the Thomson and Barcoo catchments were selected in order to demonstrate periods when connection

226 flows replenished ephemeral waterholes (September 2006, April and August 2007 and November 2008). To investigate differences in fish assemblages at ephemeral sites following variable antecedent hydrology, recently-inundated ephemeral sites in the Georgina and Mulligan catchments following major flooding in early 2007 were compared to sites in the Barcoo and Thomson catchments that filled following smaller flow events in September 2006, April and August 2007 and November 2008. Species that were absent from one or other of the Georgina/Mulligan or Thomson/Barcoo catchments were removed from the dataset in order to prevent species absence influencing the analyses. Consequently, banded grunter and golden goby were removed from Georgina/Mulligan data set and Cooper Creek catfish, Australian smelt and carp gudgeons were removed from the Thomson/Barcoo data set. Samples collected in the Georgina and Mulligan systems in April 2007, August 2007 and November 2007 were examined to investigate differences in fish assemblage composition in ephemeral sites located at various distances from permanent waterholes.

Bray-Curtis similarity matrices (Bray and Curtis 1957) were constructed using CPUE totals using PRIMER-E Version 5 on each data-set. All data-sets were log 10 (x+1) transformed. Ordination analyses (Clarke 1993) were performed using hybrid non- metric multi-dimensional scaling in PRIMER-E Version 5 in order to identify obvious patterns of similarity between fish communities in permanent and ephemeral waterholes in the Georgina/Mulligan and Thomson/Barcoo catchments. Similar matrices and ordinations were performed in order to identify patterns of similarity between ephemeral sites following major flooding or smaller flows and patterns of similarity at various distances from main channel permanent waterholes in the Georgina and Mulligan catchments. One-way analysis of similarities (ANOSIM) using the Bray-Curtis matrices was then used in order to test for the influence of waterhole permanence and antecedent hydrology on fish communities in both the Georgina/Mulligan and Thomson/Barcoo catchments. ANOSIM was also used to identify differences in fish communities at distances from main channel sites in the Georgina and Mulligan catchments. Results from ANOSIM calculate a test statistic ‘R’ identifying the observed differences between treatments compared with the differences among replicates within treatments (Clarke and Warwick 1994).

227 To test for the influence of distance in the Georgina and Mulligan catchments, sites were categorised as either permanent, ephemeral within 10 kilometres of permanent water or ephemeral greater than 100 kilometres from permanent water. Although categorisation of waterholes along a more even gradient would have been more desirable, this was not possible in the study area as no waterholes existed between 10 and 100 kilometres from the main channel where fish colonisation could be evaluated. In instances where ANOSIM revealed significant pairwise differences between fish communities explained by distance, hydrology or waterhole permanence, SIMPER analysis in PRIMER-E Version 5 was used to calculate the average dissimilarity between paired samples and allocate the contribution each species made to this dissimilarity (Clarke and Warwick 1994).

For data from the Georgina and Mulligan catchments, one-way analysis of variance (ANOVA) was used in SPSS Version 14 to test for differences in the abundance of individual fish species CPUE at permanent waterholes, ephemeral waterholes <10 kilometres from permanent water and ephemeral waterholes >100 kilometres from permanent water (fixed factors). All data was log 10 (x+1) transformed and Levene’s Test was used to satisfy assumptions of homogeneity. Significant results were accepted at P≤0.05. Kolmogorov-Smirnov 2-sample tests were used to test for differences in the size structure of all measured individuals collected at the same distances from permanent water. The same statistical tests were used on data from the Thomson and Barcoo catchments in order to investigate differences in the abundance and length frequency of fish species in permanent and ephemeral waterholes where sufficient data was available.

228 7.3 Results

7.3.1 Fish communities at permanent and ephemeral sites in the Georgina and Mulligan catchments, 2006 - 2008

Multivariate analysis of the fish assemblages at all sites in the Georgina and Mulligan catchments demonstrated separation based on fish species abundance between permanent and ephemeral sites (Global R: 0.201, p = 0.001) (Figure 7.1). Fish communities from permanent waterholes were concentrated in the top and left of the ordination space, whereas fish communities from ephemeral habitats were concentrated in the bottom half of the ordination (Figure 7.1).

Stress: 0.17 Pa/Au MC/MA MC/Au MC/No

WJ/N2 Id/Au

Lo/Au Pa/No LM/N2 Pa/MA SB2/Ap Id/No MC/Ap SB/Ap Pu/No Id/Ap SB/Au Lo/Ap Pu/Ap Id/MA Pu/Au

Figure 7.1 Two Dimensional NMS ordination plot of log 10 (x + 1) CPUE data for fish communities in the Georgina and Mulligan catchments. Filled circles = permanent waterholes, clear circles = ephemeral waterholes. Waterhole codes: SB = S Bend, Pu = Pulchera, Id = Idamea, Lo = Lower, MC = Georgina Main Channel, Pa = Parapituri, LM = Lake Mary, WJ = Walkaba/Jimberella. Sampling time codes: Ap = April 2007, Au = August 2007, No = November 2007, MA = March/April 2008, N2 = November 2008.

229 In the Georgina and Mulligan catchments, bony bream and desert rainbowfish displayed high average abundances in both permanent and ephemeral habitats, whereas spangled perch, glassfish and silver tandan were more common in ephemeral habitats. Hyrtl’s tandan and banded grunter were common in samples from permanent habitats but rare in samples from ephemeral habitats (Table 7.1).

Table 7.1 SIMPER analysis comparing fish assemblages in the Georgina and Mulligan catchments at all permanent and ephemeral sites during the study. Species Mean abundance per sample Percent contribution to observed differences. Permanent Ephemeral Permanent vs ephemeral waterholes waterholes Spangled perch 33.75 49.62 13.58 Glassfish 41.50 129.62 11.30 Bony bream 225.5 160 14.01 Desert rainbowfish 182.75 123.23 13.91 Hyrtl’s tandan 66.88 8.15 12.43 Banded grunter 42.88 5.31 10.97 Silver tandan 3.25 46.85 8.99

7.3.2 Fish assemblages in recently-filled ephemeral sites and permanent sites in the Thomson and Barcoo catchments

Multivariate analysis of the fish assemblages at sites in the Thomson and Barcoo catchments (pooled from sampling events following connection flows in September 2006, April and August 2007 and November 2008) demonstrate a clear separation of ephemeral sites and permanent sites (Figure 7.2).

230 Stress: 0.11

MC/N2

C2/Sep Is/Ap Wl/N2

MC/Se Co/Ap Wl/Se MC/Au Co/Se Wl/Au C2/Ap

TT/Au

Permanent sites Ephemeral sites

Figure 7.2 Two Dimensional NMS ordination plot of log 10 (x + 1) CPUE data for fish assemblages at sites in the Thomson and Barcoo catchments, April 2007 – November 2008. Clear circles = ephemeral sites close (<5 kilometres) to the main channel and filled circles = permanent sites within the main channel that did not dry during the study. Arrows represent the most likely colonisation pathways from permanent to ephemeral waterholes during within-channel connection events. Coding of sites: site/sampling time. Abbreviations (site): MC = Main Channel, C2 = Coolagh 2, TT = Thomson Tiny, Co = Coolagh, Wl = Waterloo, Is = Isisford; (sampling time): Se = September 2006, Ap = April 2007, Au = August 2007, N2 = November 2008.

Analysis of Similarities (ANOSIM) indicates there were significant differences between the structure of fish assemblages at ephemeral and permanent waterholes within the Thomson and Barcoo catchments (Global R: 0.525; p = 0.005). Bony bream, spangled perch and yellowbelly were always more common at ephemeral sites than permanent sites in the Thomson and Barcoo catchments (Table 7.2), desert rainbowfish was only found at ephemeral sites and Hyrtl’s tandan was only common at permanent sites (Table 7.2).

231 Table 7.2 SIMPER analysis comparing fish assemblages at sites in the Thomson and Barcoo catchments categorised by presence in permanent and ephemeral waterholes. Species Average abundance per sample Percent contribution to observed differences. Ephemeral Permanent Ephemeral vs permanent waterholes waterholes waterholes Spangled perch 12.14 - 12.25 Hyrtl’s tandan - 31.00 17.65 Bony bream 72.86 9.17 12.74 Desert rainbowfish 4.29 - 7.45 Yellowbelly 25.56 9.17 12.26 Silver tandan 5.86 9.17 8.92 Australian smelt - 12.50 11.75

7.3.3 The influence of antecedent hydrology on fish communities in ephemeral waterholes

Multivariate analysis of the fish assemblages present at ephemeral sites in both the Georgina/Mulligan and Thomson/Barcoo catchments demonstrates separation of sites that experienced major flooding from those that experienced bankfull flows or lower flows (Figure 7.3). Species endemic to either area were removed from the dataset in order to complete this analysis, and hence it represents species present in both areas (Georgina/Mulligan and Thomson/Barcoo).

232 Stress: 0.12 MC/Th/N2

C2/Ba/Se MC/Th/Se

Pu/Mu/Ap

Lo/Ge/Ap

SB/Mu/Ap MC/Th/Au Id/Ge/Ap

SB2/Mu/Ap C2/Ba/Ap

Figure 7.3 Two Dimensional NMS ordination plot of log 10 (x + 1) CPUE data for fish assemblages at sites experiencing major flooding (overbank flows) in the 3 months prior to sampling (circles) and bank-full flows or less (stars) during the sampling period. Coding of sites: site/catchment/sampling time. Abbreviations (site): Lo = Lower, Id = Idamea, SB = S Bend, Pu = Pulchera, MC = Main Channel, C2 = Coolagh 2; (catchment): Ge = Georgina, Mu = Mulligan, Th = Thomson, Ba = Barcoo; (sampling time): Se = September 2006, Ap = April 2007, Au = August 2007, N2 = November 2008.

Analysis of Similarities (ANOSIM) indicates that there were significant differences between the structure of fish assemblages at ephemeral waterholes filled from major flooding compared with those filled from smaller flows (Global R: 0.467; p = 0.009). All species were more abundant in ephemeral waterholes filled from major flooding than they were in waterholes filled from smaller flows, with the most pronounced differences occurring in populations of desert rainbowfish, glassfish and silver tandan (Table 7.3).

233 Table 7.3 SIMPER analysis comparing fish assemblages at ephemeral sites filled from major flooding with those filled from smaller flows (All transformations log 10 (x+1). Species Mean abundance per sample Percent contribution to (ephemeral waterholes) observed differences. Smaller flows Major flooding Major flooding vs smaller (

7.3.4 Location of permanent waterholes in the Georgina and Mulligan catchments

The Georgina catchment in far western Queensland is characterised by relatively few permanent waterholes (Figure 7.4), and the Mulligan River has no permanent waterholes. Following rainfall events and flooding in January 2007, the Georgina and Mulligan waterholes were replenished (Figure 7.5), but by November 2008 only Lake Mary, Lake Nash (Northern Territory), Walkaba, the Lake Katherine complex, Basin, Parapituri and Tommydonka Waterhole at Glengyle within the Georgina main channel system and Bulla Bulla waterhole on the Hamilton River retained water (Figure 7.4; personal observation November 2008). Similarly dry conditions prevailed prior to flooding in January and February 2007 (Shane McGlinchey, Badalia Station; Stephen Bryce, Glenormiston Station, personal communications).

234 1. Lake Mary 1 2. Lake Nash 3. Walkaba 4. Lake Katherine 5. Basin 2 6. Parapituri 7. Glengyle 8. Bulla Bulla

3 4 8 5 6

7

kilometres

100 200 300

Figure 7.4 The location of 8 permanent waterholes in the Georgina catchment in November 2008.

235 Lake Idamea Basin

Lower Lake Pituri Ck/Georgina junction

S Bend Gorge Ocean Bore Kunnamuka Swamp Dune Pond Pulchera

Mulligan/Eyre Ck junction

Figure 7.5 Ephemeral waterholes in the Pituri Creek/Georgina catchment and the Mulligan catchment in relation to Basin Waterhole (permanent), the Pituri Ck/Georgina confluence and the Mulligan/Eyre Creek confluence following the recession of floods in early 2007 (April 2007).

7.3.5 Fish migration distances following flooding in the Georgina and Mulligan catchments in January and February 2007

Riverine distances and distances across dunefields (where relevant) for sites sampled in the Mulligan and Georgina catchments in early 2007 are given in Table 7.4. The confluence of the Mulligan River with Eyre Creek is situated at the northern end of Kalidewarry Waterhole (David Brook, Adria Downs Station, personal communication; Figure 2). The confluence of Pituri Creek with the Georgina River occurs 3 kilometres south of Basin Waterhole (Jenny Silcock, Queensland Environmental Protection Agency, personal communication; Figure 7.5). Sites located

236 on Pituri Creek, such as Lake Idamea and Lower Lake, all occur on the main channel, whereas sites in the Mulligan occur on both the main channel (Pulchera and S Bend) and at remote locations separated by dunefields (Dune Pond and Kunnamuka Swamp)(Table 7.4).

Table 7.4 Migration distances* to previously dry waterholes in the Georgina and Mulligan catchments in January/February 2007 (Silcock 2009). ‘NA’ = not applicable. Sub- Waterhole Confluence Upstream Dunefield Total catchment location Distance distance distance (km) (km) (km) Pituri Creek Lake 3 km south of Basin 9 NA 9 Idamea Waterhole, Georgina River Pituri Creek Lower Lake 3 km south of Basin 6 NA 6 Waterhole, Georgina River Mulligan Pulchera Kalidewarry Waterhole north, 105 NA 105 Eyre Creek Mulligan Dune Pond Kalidewarry Waterhole north, 140 20 (west) 160 Eyre Creek Mulligan S Bend Kalidewarry Waterhole north, 250 NA 250 Gorge Eyre Creek Mulligan Kunnamuka Kalidewarry Waterhole north, 240 40 (west) 280 Swamp Eyre Creek * Shortest wetted distance/most likely minimum migration distance during flooding.

7.3.6 Fish species presence/absence at distances from confluence sites in the Georgina and Mulligan catchments following major flooding

Fish species present at the sampled sites in the Georgina and Mulligan catchments in April 2007 are given in Table 7.5. Sites <10 kilometres from the confluence of ephemeral rivers with permanent waterbodies consistently contained a greater number of species than sites 100km, 160km and 250km distant. Hyrtl’s tandan, yellowbelly and golden goby were not detected further than 10 kilometres from permanent

237 waterholes, whereas bony bream, silver tandan, desert rainbowfish, glassfish, Barcoo grunter, banded grunter and spangled perch were found 100 kilometres upstream of confluence sites (Table 7.5). Only bony bream, silver tandan, desert rainbowfish, glassfish and spangled perch were sampled 250 kilometres upstream of confluence sites (Table 7.5). A spangled perch was sampled 140 kilometres upstream of the confluence site and 20 kilometres west across the Simpson Desert dunefields, but no fish were sampled from Kunnamuka Swamp, a distance of 240 kilometres upstream of the confluence site but a further 40 kilometres west across the dunefields (Table 7.5; Figure 7.6).

Table 7.5 Fish species sampled at distances from confluence sites with main channel waterholes in the Georgina and Mulligan catchments, April 2007. Dots indicate presence, spaces indicate absence. Species Distance between main channel waterholes and ephemeral sites <10 km 100km 160km* 250k 280km** m Basin Idamea Lower Pulchera Dune Pond S Bend Kunnamuka Nematolosa erebi ● ● ● ● ● Bony bream Neosiluris hyrtlii ● ● ● Hyrtl’s tandan Porochilus argenteus ● ● ● ● ● Silver tandan M. s. tatei ● ● ● ● ● Desert rainbowfish Ambassis sp. ● ● ● ● ● Glassfish Macquaria sp. ● Yellowbelly Amniataba percoides ● ● ● ● Banded grunter Leiopotherapon unicolor ● ● ● ● ● ● Spangled perch Scortum barcoo ● ● Barcoo grunter Glossogobius aureus ● ● ● Golden goby TOTAL NUMBER OF 9 8 9 7 1 5 0 SPECIES *includes 20 kilometre westward migration across dunefields. ** includes 40 kilometre westward migration across dunefields

238

Figure 7.6 A typical sand dune in the eastern Simpson Desert close to Kunnamuka Swamp. Dunes in this vicinity are generally 6 - 8 metres high and therefore represent a considerable barrier to fish movement.

7.3.7 Fish communities in isolated waterholes in the Georgina/Mulligan catchments: the influence of migration distance on assemblage structure

Multivariate analysis of the fish assemblages at sites in the Georgina and Mulligan catchments demonstrates grouping of ephemeral sites >100 kilometres from the main channel and a separation of these sites from both permanent sites and ephemeral sites <10 kilometres from the main channel (Figure 7.7).

239 Stress: 0.17 Pu/Au Lo/Au

>100km

Pu/Ap

Id/Ma Pu/No

SB/Au SB1/Ap

Lo/Ap SB2/Ap MC/Au <10km

MC/Ap Id/Ap Id/Au LM/N2 Id/No Pa/No MC/No Pa/Au Pa/MA Permanent

WJ/N2 MC/MA

Figure 7.7 Two Dimensional NMS ordination plot of log 10 (x + 1) CPUE data for fish assemblages at sites in the Georgina and Mulligan catchments, April 2007 – November 2008. Green triangles = ephemeral sites >100 kilometres from main channel, blue triangles = ephemeral sites <10 kilometres from main channel, and black squares = permanent sites within the main Georgina channel that did not dry during the study. Waterhole codes: SB = S Bend, Pu = Pulchera, Id = Idamea, Lo = Lower, MC = Georgina Main Channel, Pa = Parapituri, LM = Lake Mary, WJ = Walkaba/Jimberella. Sampling time codes: Ap = April 2007, Au = August 2007, No = November 2007, MA = March/April 2008, N2 = November 2008.

Analysis of Similarities (ANOSIM) indicates there are significant differences between the structure of fish assemblages at ephemeral sites >100 kilometres from the main channel and permanent waterholes within the Georgina main channel, and also significant differences between fish assemblages >100 kilometres from the main channel and assemblages at ephemeral sites <10 kilometres from the main channel

240 (Table 7.4). However, fish assemblages were not significantly different between permanent sites and ephemeral sites <10 kilometres from the main channel (Table 7.6).

Table 7.6 Summary of One-Way ANOSIM results comparing fish assemblages at distances from the main channel of the Georgina River in the Georgina and Mulligan catchments, April 2007 – November 2008. All transformations log 10 (x + 1). Factor Global R P Significant pairwise tests Distance 0.275 0.004 >100 km vs <10 km (0.035) >100 km vs permanent (0.002) <10km vs permanent (n.s.)

Bony bream and desert rainbowfish were always more common at main channel sites and ephemeral sites within 10 kilometres of the main channel than they were at sites >100 kilometres from the main channel (Table 7.7), and species that were absent or rare at sites >100 kilometres from the main channel, such as Hyrtl’s tandan, banded grunter and yellowbelly, always contributed to the differences between distant sites (>100 kilometres from the main channel) and sites close to the main channel (Table 7.7). Spangled perch were always common at distant sites and this made a large contribution to the separation of these sites from sites closer to, or within, the main channel.

241 Table 7.7 SIMPER analysis comparing fish assemblages at sites in the Georgina and Mulligan catchments categorised by distance from the main channel of the Georgina. Species Average abundance per sample Percent contribution to observed differences. Ephemeral Ephemeral Permanent >100km vs >100km vs waterholes waterholes waterholes <10km permanent >100km from <10km from main channel main channel Spangled perch 54.5 53 - 15.18 17.09 Glassfish 45 235.67 37 12.43 9.89 Bony bream 55.83 285.17 204.22 18.23 11.35 Desert rainbowfish 85.17 177.50 165.33 18.32 13.27 Banded grunter - 11.17 38.11 7.29 10.16 Hyrtl’s tandan - - 59.56 5.92 11.73 Yellowbelly - - 5 - 7.35 Silver tandan - - - 14.31 10.72

7.3.8 Fish species colonisation in the Mulligan, Georgina, Thomson and Barcoo catchments

Bony bream

Bony bream were more numerous at previously dry sites <10 kilometres from the main channel than at sites >100 kilometres from the main channel in April 2007 in the Georgina and Mulligan rivers following major flooding (df: 1, F = 20.283, p = 0.011; Figure 7.8). The size structure of bony bream populations was significantly different between sites <10 kilometres from the main channel and >100 kilometres from the main channel (KS-Z 5.348, p < 0.001; Figure 7.8). Bony bream at sites <10 kilometres from the main channel included both smaller and larger individuals than at sites >100 kilometres from the main channel (Figure 7.8). In the Thomson and Barcoo catchments, juvenile bony bream were consistently more abundant in ephemeral habitats than in permanent habitats following recent inundation, (df: 1, F = 6.865, p = 0.024) and the size structure of bony bream was significantly different between ephemeral and permanent sites (KS-Z 2.148, p < 0.001; Figure 7.9). Juvenile bony

242 bream were always common at ephemeral sites in the Thomson and Barcoo catchments, whereas adult fish (~200mm SL) were generally only present at permanent sites (Figure 7.9).

450 450

360 360

270 270

t t

n n

u u

o o C 180 C 180

90 90

0 0 0 50 100 150 0 50 100 150 >100km (n=116) <10km (n=1220)

Standard length (mm)

Figure 7.8 Size frequency histograms of populations of bony bream pooled from sites >100 kilometres from the main channel (left) and <10 kilometres from the main channel following flooding in the Georgina and Mulligan catchments in April 2007.

50 50

40 40

30 30

t t

n n

u u

o o C 20 C 20

10 10

0 0 0 100 200 300 0 100 200 300 Ephemeral (n=505) Permanent (n=54)

Standard length (mm)

Figure 7.9 Size frequency histograms of populations of bony bream pooled from ephemeral and permanent sites in the Thomson and Barcoo catchments in September 2006, April and August 2007 and November 2008.

243 Hyrtl’s tandan

Adult Hyrtl’s tandan populations demonstrated a clear habitat preference for permanent waterholes in the Thomson and Barcoo catchments following connection events (Figure 7.10). No Hyrtl’s tandan over 120mm SL were sampled in ephemeral waterholes of the Thomson and Barcoo catchments, nor in previously dry sites <10 kilometres from the main channel in the Georgina catchment throughout the study (Figure 7.10).

30 30 30

20 20 20

t t t

n n n

u u u

o o o

C C C

10 10 10

0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 <10 km (Georgina, n=108) Ephemeral (Thomson/Barcoo n=8) Permanent (Thomson/Barcoo n=185)

Standard length (mm)

Figure 7.10 Size frequency histograms of populations of Hyrtl’s tandan pooled from sites <10 kilometres from the main channel following flooding in the Georgina catchment in April 2007 (left), from ephemeral sites in the Thomson and Barcoo catchments (centre) and from permanent sites in the Thomson and Barcoo catchments in September 2006, April and August 2007 and November 2008 (right).

Silver tandan

There was no significant difference in the abundance of silver tandan at previously dry sites <10 kilometres from the main channel than at sites >100 kilometres from the main channel in April 2007 in the Georgina and Mulligan rivers following major flooding (df: 1, F = 0.889, p = 0.399; Figure 7.11). The size structure of silver tandan populations was significantly different between sites <10 kilometres from the main channel and >100 kilometres from the main channel (KS-Z 6.126, p < 0.001; Figure 7.11). The sample of silver tandan at sites >100 kilometres from the main channel was

244 dominated by fish in the 70 – 80mm SL classes, whereas the sample from sites <10 km from the main channel included a broader range of size classes (Figure 7.11). In the Thomson and Barcoo catchments, there was no significant difference in the abundance of silver tandan in ephemeral habitats and permanent habitats following recent inundation, (df: 1, F = 0.043, p = 0.727) but the size structure of silver tandan was significantly different between ephemeral and permanent sites (KS-Z 3.067, p < 0.001; Figure 7.12). Juvenile silver tandan occurred in both ephemeral and permanent sites in the Thomson and Barcoo catchments, but adults were only common at permanent sites.

80 80

60 60

t t

n n

u 40 u 40

o o

C C

20 20

0 0 0 50 100 150 200 0 50 100 150 200 >100km (n=181) <10km (n=340)

Standard length (mm)

Figure 7.11 Size frequency histograms of populations of silver tandan pooled from sites >100 kilometres from the main channel (left) and <10 kilometres from the main channel following flooding in the Georgina and Mulligan catchments in April 2007.

245 15 15

10 10

t t

n n

u u

o o

C C

5 5

0 0 0 50 100 150 200 0 50 100 150 200 Ephemeral (n=39) Permanent (n=55)

Standard length (mm)

Figure 7.12 Size frequency histograms of populations of silver tandan pooled from ephemeral (left) and permanent (right) sites in the Thomson and Barcoo catchments during sampling events undertaken in September 2006, April and August 2007 and November 2008.

Desert rainbowfish

There was a significant difference in the abundance of desert rainbowfish at previously dry sites <10 kilometres from the main channel and at sites >100 kilometres from the main channel in April 2007 in the Georgina and Mulligan rivers, with far larger populations of this species present closer to the junction with the main channel (df: 1, F = 10.102, p = 0.034; Figure 7.13). The size structure of desert rainbowfish populations was significantly different between sites <10 kilometres from the main channel and >100 kilometres from the main channel, with populations at distant sites populated by larger individuals (> 40mm SL) and populations at closer sites demonstrating a far wider range of size classes (KS-Z 1.189, p = 0.003; Figure 7.13). Desert rainbowfish were rare in ephemeral waterholes of the Thomson and Barcoo catchments following recent inundation, with only 30 individuals recorded from combined sampling events in September 2006, April and August 2007 and November 2008. However, during the same sampling periods no desert rainbowfish were recorded from permanent waterholes in the same general vicinity.

246 170 170

136 136

102 102

t t

n n

u u

o o C 68 C 68

34 34

0 0 0 20 40 60 80 0 20 40 60 80 >100km (n=54) <10 km (n=641)

Standard length (mm)

Figure 7.13 Size frequency histograms of populations of desert rainbowfish pooled from sites >100 kilometres from the main channel (left) and <10 kilometres from the main channel following flooding in the Georgina and Mulligan catchments in April 2007.

Glassfish

There was no significant difference in the abundance of glassfish at previously dry sites <10 kilometres from the main channel than at sites >100 kilometres from the main channel in the Georgina and Mulligan rivers in April 2007 (df: 1, F = 0.136, p = 0.731; Figure 7.14). The size structure of glassfish populations was significantly different between sites <10 kilometres from the main channel and >100 kilometres from the main channel (KS-Z 7.165, p < 0.001; Figure 7.14). The sample of glassfish at sites >100 kilometres from the main channel was dominated by individuals >40mm SL and contained no fish <30mm SL. In contrast, the sample from sites <10 km from the main channel included a broader range of size classes including juveniles <20mm SL (Figure 7.14). Glassfish were rare in ephemeral waterholes of the Thomson and Barcoo catchments following recent inundation, with only 20 individuals recorded from combined sampling events in September 2006, April and August 2007 and November 2008. However, during the same sampling periods no glassfish were recorded from permanent waterholes in the same general vicinity.

247

130 130

104 104

78 78

t t

n n

u u

o o C 52 C 52

26 26

0 0 0 20 40 60 0 20 40 60 >100km (n=84) <10km (n=361)

Standard length (mm)

Figure 7.14 Size frequency histograms of populations of glassfish pooled from sites >100 kilometres from the main channel (left) and <10 kilometres from the main channel (n = 361) following flooding in the Georgina and Mulligan catchments in April 2007.

Yellowbelly

Yellowbelly were absent from all ephemeral waterholes in the Mulligan and Georgina catchments for the duration of the study. In contrast, juvenile yellowbelly were always common in recently-inundated waterholes in the Thomson and Barcoo catchments, whereas a broader range of size classes inhabited permanent waterholes (Figure 7.15).

248 20 20

15 15

t t

n n

u 10 u 10

o o

C C

5 5

0 0 0 200 400 600 0 200 400 600 Ephemeral (n=181) Permanent (n=55)

Standard length (mm)

Figure 7.15 Size frequency histograms of populations of yellowbelly pooled from ephemeral and permanent sites in the Thomson and Barcoo catchments during sampling events undertaken in September 2006, April and August 2007 and November 2008.

Spangled perch

There was no significant difference in the abundance of spangled perch at previously dry sites <10 kilometres from the main channel than at sites >100 kilometres from the main channel in April 2007 in the Georgina and Mulligan rivers (df: 1, F = 1.408, p = 0.301; Figure 7.16). The size structure of spangled perch populations was significantly different between sites <10 kilometres from the main channel and >100 kilometres from the main channel (KS-Z 2.755, p < 0.001; Figure 7.16). At distant sites, spangled perch <50mm SL were comparatively common, whereas at sites close to the main channel all collections were dominated by larger individuals (Figure 7.16). In the Thomson and Barcoo catchments, samples of spangled perch were dominated by juveniles in all ephemeral habitats during sampling undertaken following inundation in September 2006, April and August 2007 and November 2008, and very few spangled perch were recorded from permanent waterholes (Figure 7.17).

249 100 100

80 80

60 60

t t

n n

u u

o o C 40 C 40

20 20

0 0 0 50 100 150 200 0 50 100 150 200 >100km (n=219) <10km (n=523)

Standard length (mm)

Figure 7.16 Size frequency histograms of populations of spangled perch pooled from sites >100 kilometres from the main channel (left) and <10 kilometres from the main channel following flooding in the Georgina and Mulligan catchments in April 2007.

15 15

10 10

t t

n n

u u

o o

C C

5 5

0 0 0 50 100 150 200 0 50 100 150 200 Ephemeral (n=84) Permanent (n=7)

Standard length (mm)

Figure 7.17 Size frequency histograms of populations of spangled perch pooled from ephemeral and permanent sites in the Thomson and Barcoo catchments during sampling events undertaken in September 2006, April and August 2007 and November 2008.

250 7.3.9 Waterhole drying in the Georgina and Mulligan catchments

In the Georgina and Mulligan catchments, major flooding filled all waterholes in January and February 2007 (Stephen Bryce, Scott Morrison, David Brook, Len Rule, property managers in the district, personal communications). Following the recession of floodwaters in March 2007 (Scott Morrison, Ethabuka Station, personal communication) fish habitats became dry in April 2007 (Dune Pond; Scott Morrison, Ethabuka Station, personal communication), August 2007 (S Bend Gorge; personal observation), September 2007 (Lower Lake; Stephen Bryce, Glenormiston Station, personal communication), by March 2008 (Pulchera Waterhole; personal observation) and by winter 2008 (Lake Idamea; Stephen Bryce, Glenormiston Station, personal communication)(Figure 7.18).

250000 Georgina/Mulligan flood peak 22 Jan – 2 Feb, 2007

200000

150000 ML/day

100000

Kunnamuka Swamp, June 07

50000 Lower Lake, September 2007 Lake Idamea winter 2008

0

Pulchera Waterhole, by March 2008 Dune pond, April 07 S Bend Gorge, August 2007

Figure 7.18 Drying dates of waterholes within the Georgina and Mulligan catchments following flooding in January and February 2007. Hydrograph from Roxborough Downs gauging station courtesy Queensland Department of Environment and Resource Management.

251 7.3.10 Localised extinctions at waterhole scale in the Georgina and Mulligan catchments

Dune Pond

A solitary 150mm SL spangled perch was sampled in Dune Pond in April 2008. The site was dry one week later.

S Bend Gorge

Representatives of five fish species (bony bream, silver tandan, spangled perch, desert rainbowfish and glassfish) were present at S Bend Gorge in both April and August 2007 (Figure 7.19). The waterhole dried completely two weeks after the August samples were taken (Len Rule, Craven’s Peak Station, personal communication). Numbers of both desert rainbowfish and glassfish were higher in August than in April, and this is most likely due to the waterhole receding and the fish assemblage becoming more concentrated (Figure 7.19).

500

450

400

350

300

250

200

150 Numbers of individuals of Numbers 100

50

0 Bony bream Silver tandan Spangled perch Desert rainbowfish Glassfish

Figure 7.19 Fish populations sampled at S Bend Gorge in April 2007 (light bars) and two weeks prior to complete drying in August 2007 (dark bars).

252 Lower Lake

There was a large reduction in the numbers of all fish species sampled at Lower Lake between April 2007 and 3 weeks’ prior to the complete drying of Lower Lake in August 2007 (Figure 7.20). In August 2007, only small numbers of spangled perch, banded grunter and glassfish were detected at Lower Lake (Figure 7.20).

350

300

250

200

150

100 Numbers of individuals of Numbers 50

0

n r an e h d rch fis nda unter unt s a pe r r s d g e d gl e Gla d Bony bream an n Golden goby Hyrtl's tan Silver t p a S Barcoo g B Desert rainbowfish

Figure 7.20 Fish populations sampled at Lower Lake in April 2007 (light bars) and three weeks prior to complete drying in August 2007 (dark bars).

Pulchera Waterhole

Populations of spangled perch, bony bream and glassfish survived at Pulchera waterhole until at least November 2007, whereas populations of silver tandan appeared to diminish (Figure 7.21). Pulchera waterhole dried prior to the March/April 2008 samples being taken. Desert rainbowfish, Barcoo and banded grunter were rare in Pulchera waterhole (Figure 7.21).

253 200

180

160

140

120

100

80

60

40 Numbers of individuals of Numbers

20

0 Bony bream Silver tandan Spangled Barcoo Banded Desert Glassfish perch grunter grunter rainbowfish

Figure 7.21 Fish populations sampled at Pulchera waterhole in April 2007 (light bars), August 2007 (dark bars) and prior to drying in November 2007 (clear bars).

254 7.4 Discussion

The results from the current study provide support for the notion that spatial distribution of fishes in the arid zone is the determining factor driving recolonisation potential (Fagan et al. 2005), however this assertion is not uniform amongst Australian desert species in far western Queensland. Australian freshwater fishes inhabiting arid zone rivers exhibit variable vagility depending upon species, and long- distance (>100 kilometres) colonisation is more likely to occur for different life-stages of certain species (Table 7.8). This suggests that presence at catchment scale does not necessarily lead to colonisation of distant areas, and that less vagile species, such as Hyrtl’s tandan and golden goby, are less likely to leave safe habitats within the study area. Consequently, the hypothesis that fish species and communities will migrate to ephemeral areas following inundation (Hypothesis 1, Section 7.1) requires clarification with regard to species, migration distance and antecedent hydrology if it is to be accepted for the river systems included in this study.

Table 7.8 Summary table of vagility (maximum migration distance) for 10 fish species from the Georgina/Mulligan and/or Barcoo/Thomson catchments during the study.

Vagility Species Migration Overland Colonising size Evidence of distance distance classes recruitment in (channel) (dunefields) ephemeral migration destinations Extreme Spangled perch ~250 km ~20 km All Yes Very High Bony bream ~250 km Nil Mostly juveniles Yes High Desert rainbowfish ~250 km Nil Definitely adults, Yes, but only <10 km possibly juveniles from permanent habitats Glassfish ~250 km Nil Definitely adults, Yes, but only <10 km possibly juveniles from permanent habitats Silver tandan ~250 km Nil Juveniles No Moderate Banded grunter ~100 km Nil Juveniles No Barcoo grunter ~100 km Nil Juveniles No Yellowbelly ~10 km Nil All size classes No direct evidence – juveniles most likely derived from permanent populations. Low Golden goby ~10 km Nil All size classes Yes, but only <10 km from permanent habitats Hyrtl’s tandan ~10 km Nil Mostly juveniles No

255 The results presented above similarly support the proposition that probability of survival of fish species in arid areas is enhanced by the ability to rapidly disperse when movement is facilitated (Fausch and Bramblett 1991), but this also appears to be species dependent for fish in Australian desert areas. Long-distance dispersal was not observed for three fish species (Hyrtl’s tandan, golden goby and yellowbelly) that shared permanent waterhole refuges with seven more vagile species in the Georgina and Mulligan catchments and – presumably – enjoyed similar opportunities for colonisation and dispersal. It therefore seems highly likely that certain Australian species, such as bony bream, spangled perch, desert rainbowfish and glassfish, may represent ‘pioneer species’ ( sensu Capone and Kushlan 1991) in an Australian context, that are able to capitalise on shallow and comparatively harsh conditions, whereas species such as Hyrtl’s tandan and golden goby are less tolerant of such conditions.

Demonstrating variability in the composition of colonising assemblages between post- flood periods (Scheurer and Fausch 2003; Eby et al. 2003) was not possible during the current study as only a single flood occurred in the Georgina and Mulligan catchments in April 2007, and these catchments have been shown to contain a different species assemblage from other catchments in the study area (see Chapter 5). Nevertheless, the results from sampling multiple sites at distant locations ( >100 km from main channel areas) strongly suggest that less-vagile species are unlikely to undertake such long migrations. These results support the hypothesis that migration distance influences the composition of fish communities in recently-inundated areas (Hypothesis 3, Section 7.1). However, testing this hypothesis during a similar post-flood year in western Queensland in the Mulligan and Georgina catchments of western Queensland would obviously be a valuable extension of this study with regard to understanding the migratory abilities and habits of present species. 10

The results suggest that following major flooding, fish communities in the Australian arid zone are likely to be similar at permanent sites and at ephemeral sites close to permanent sites ( <10 kilometres), but that a smaller sub-group of species is likely to

10 Sampling of the Mulligan catchment in October 2009 using similar methods produced similar results but included one additional species (a hardyhead) that is currently being identified by staff at the South Australian Museum (see Appendix 5).

256 migrate further (>100 kilometres). These results indicate that the size or the volume of flow events is likely to have a bearing on migration ability or tendency (Poff and Allan 1995), and that major flooding is more likely to elicit a migratory response from a larger range of species and size classes than smaller flows. These results support the hypothesis that antecedent hydrology influences the composition of fish communities in formerly dry areas (Hypothesis 4, Section 7.1). In contrast, following smaller flows, fish populations in ephemeral habitats were characterised by smaller size classes, with juvenile bony bream, yellowbelly and spangled perch frequently sampled in areas that had recently experienced connection flows. It is therefore suggested that these periodically-inundated areas function as nurseries following inundation, and that the hydrological longevity of such areas determines whether these nurseries succeed (thus allowing cohorts to migrate again upon re-connection) or fail (resulting in the extirpation of the community). These results are supportive not only of prior work in the same geographical area (Balcombe et al. 2007), but also other studies concluding that large, adult fish are more likely to remain in deeper more permanent areas (Schlosser 1987; Capone and Kushlan 1991). The results also suggest acceptance of the hypothesis that migration ability or likelihood is influenced by the size class or life-cycle stage of potential colonists (Hypothesis 2, Section 7.1).

Bony bream and silver tandan

Bony bream and silver tandan both demonstrated the ability to colonise areas up to 250 kilometres upstream from the most likely source populations following flooding, and for both species there was a strong migratory response from smaller size classes. Davey and Kelly (2007) found a similar species-dependent upstream migration occurred during drying periods in the Selwyn River in New Zealand. The results from the present study suggest that large adult bony bream and silver tandan possibly avoid potentially dangerous migrations to ephemeral areas in the Australian arid zone, but when connectivity is re-instated juvenile representatives of each population migrate to such areas. Evidence of the role played by periodically-inundated ephemeral waterholes as nursery areas for juveniles of both species was particularly evident in the Thomson and Barcoo catchments, where samples taken following connection events yielded strong cohorts of small individuals. Such a strategy could have numerous benefits for the populations of bony bream and silver tandan, including the

257 preservation of broodstock in deeper, more permanent waterholes and the possibility of biogeographic range extension if juvenile colonists are successful in reaching ephemeral areas that retain water for several months, or connect at later times to other less ephemeral habitats. Under ideal circumstances (following a series of floods or wetter-than-average years), colonists would obviously be in a position to become breeding (source) populations capable of delivering juvenile colonists to more distant ephemeral environments.

During the current study there was no evidence that silver tandan colonists persisted for more than seven months following migration to waterholes ~100km upstream from main channel habitats, and no evidence of colonists recruiting in isolated waterholes despite a comparatively large cohort of juveniles undertaking the initial migration. In contrast, populations of colonising bony bream appeared to breed in the desert waterholes, and a cohort of juveniles was detected in Pulchera waterhole (some 100+ kilometres from the closest permanent water) in November 2007 shortly before the waterhole dried. If a rainfall event had occurred in the Mulligan catchment and replenished Pulchera waterhole, it therefore seems likely that bony bream populations originating from a cohort of colonising juveniles may have endured, whereas silver tandan populations may have already become locally extinct.

Desert rainbowfish and glassfish

In contrast to the populations of bony bream and silver tandan, populations of desert rainbowfish and glassfish in the ephemeral Mulligan system were almost exclusively sampled as adults rather than juveniles, but for both species these migrations were still comparatively extensive (up to 250 kilometres; Table 7.8). It is possible that juveniles of both species colonised the Mulligan system and reached maturity prior to the first samples being taken in April 2007. Nevertheless, despite comparatively large numbers of both desert rainbowfish and glassfish being present in samples taken in April and August in the Mulligan catchment, there was no evidence that the colonising individuals had successfully recruited prior to the extirpation of the S Bend community (September 2007) or shortly before the extirpation of the Pulchera community (November 2007), as only large fish were sampled on all occasions. This data suggests that, like silver tandan, colonising populations of both desert

258 rainbowfish and glassfish possibly require a longer establishment time in previously- dry waterholes than bony bream. Alternatively, it seems equally likely that any pulse of recruitment in isolated waterholes for both of these species – if it occurred - may have been impacted by predation or competitive influences, and that these factors did not impact to the same degree on bony bream recruits. Neither species was detected colonising recently-inundated ephemeral waterholes close to permanent waterholes in the Thomson and Barcoo catchments in large numbers throughout the study. Differences in the colonisation behaviour of both desert rainbowfish and glassfish in the Georgina/Mulligan and Thomson/Barcoo are possibly related to flow volume, for in the Georgina/Mulligan recruitment may have been enhanced by the occurrence of a major flood. Alternatively, the general rarity of both desert rainbowfish and glassfish in the Thomson and Barcoo rivers may explain their low utilisation of ephemeral waterholes in these catchments, as both species were also generally absent from permanent waterholes. Again, it is likely that both species represent Australian ‘pioneer species’ (Capone and Kushlan 1991) and that their success is heightened in areas where there are fewer competitive or predatory species (such as the Georgina/Mulligan) and lowered in areas with a robust predatory fauna (such as the Thomson/Barcoo). With specific reference to glassfish, the results from the current study in the Diamantina catchment are interesting when compared to those in the lower Cooper (South Australia) in the period prior to 1989 (Puckridge 1999). Despite a regular sampling regime carried out over five years, Puckridge (1999) found no glassfish in the lower Cooper until 1989, when colonists were detected following flooding. In subsequent years (1990 and 1991) glassfish were abundant in the lower Cooper (Puckridge 1999) and it appears just as likely that this species could re- establish widespread populations in the Queensland reaches of the Diamantina River following a series of wet years.

Yellowbelly

Although yellowbelly were present in permanent waterholes of the Georgina River, no evidence was found of this species migrating to close (<10km) or distant (>100km) areas in the Georgina and Mulligan catchments during major flooding. This is a surprising result, especially given the existence of a study that has documented migrations in excess of 1000km for Murray-Darling yellowbelly, Macquaria ambigua

259 (Reynolds 1983). In contrast, juvenile yellowbelly were always present in recently- inundated ephemeral waterholes in the Barcoo and Thomson rivers, suggesting that this species colonises nearby nursery habitat in these catchments following connection events (Balcombe et al. 2007). The presence of large yellowbelly (>300mm SL) in permanent waterholes of the Thomson and Barcoo rivers combined with the absence of large fish in ephemeral waterholes indicates that large adults of this species are generally likely to remain in permanent waterholes and that juveniles are most likely to disperse opportunistically to less-safe/more risky habitats. This result is similar to the patterns observed for both bony bream and silver tandan in all catchments during the current study, and also to results for fish species from other areas (Schlosser 1987; Zeug and Winemiller 2008). Evidence from southern Australia indicates that adult yellowbelly generally occupy relatively small home ranges (Crook et al . 2001), and the results from the current study suggest that this behaviour may be replicated in rivers of the Lake Eyre Basin. Nevertheless, the apparent aversion to colonisation by this species in the far western catchments (the Georgina and Mulligan) is notable, 11 particularly as juvenile yellowbelly were present in ephemeral areas of the Thomson and Barcoo catchments. It seems highly likely that the Georgina River may represent the western extent of yellowbelly distribution in Queensland (see Chapters 5 and 6), and that this large-bodied and long-lived species would probably derive no ecological benefit from migrating to ephemeral desert areas. Given the demonstrated success of yellowbelly breeding events in flood years in the Lake Eyre Basin (Pritchard 2004; Balcombe and Arthington 2009) and also in the Fitroy catchment (Roberts et al. 2008), the absence of a detectable yellowbelly recruitment response to major flooding in the Georgina and Mulligan catchments remains curious. An animal behaviour question arises if it is indeed true that yellowbelly never – or at least very rarely – migrate into desert systems during flooding: how do they ‘know’ not to migrate to certain areas? Unfortunately answering this question was beyond the scope of this study.

11 During sampling undertaken in October 2009, and following inundation and connection in early 2009, no yellowbelly were sampled from ephemeral waterholes of the Mulligan catchment, a similar result to 2007.

260 Spangled perch

The current study provides strong evidence to support general assumptions regarding the vagility and hardiness of spangled perch (Pusey et al . 2004; Kingsford et al. 2006a). This species was the only fish that was detected in a rapidly-drying site at the base of a sand dune some 20 kilometres west of the Mulligan channel and 140 kilometres north of the Mulligan/Eyre Creek confluence (Table 7.8). Access to this site is most likely to have occurred via migration through overland flow during rainfall events as there are no channels in the area. Additionally, both adult and juvenile spangled perch were sampled at all Mulligan River sites throughout 2007, indicating that this species undertakes migrations of at least 250 kilometres upstream following inundation events. Like bony bream, spangled perch successfully recruited in the isolated habitat of Pulchera Waterhole following colonisation, as a cohort of juveniles was detected in November 2007 shortly before the waterhole dried. The presence of comparatively strong populations of spangled perch at recently-inundated ephemeral sites in the Thomson and Barcoo catchments provides further evidence of the colonising behaviour of this species throughout the study area. It is notable that although adult spangled perch were rarely detected in permanent waterholes in these catchments, juveniles frequently colonised ephemeral habitats. This evidence strongly suggests that spangled perch possess a life history strategy that is geared towards capitalising on the unpredictability of the hydrological regime and exploiting new habitats as soon as they become available. As spangled perch frequently colonised all available areas as fully-grown adults, it seems highly likely that the presence of this piscivorous species in isolated waterholes may exert a strong influence on the composition and age structure of fish communities in such areas. This may account for the apparent lack of juvenile desert rainbowfish and glassfish in the isolated desert waterholes sampled, and suggests that bony bream may recruit more successfully due to comparatively higher fecundity. Confirming such assertions would obviously require more detailed study of individual desert fish communities along temporal gradients of wetting and drying. Finally, it is noteworthy that spangled perch were both more abundant and often larger in the Mulligan and Georgina systems – both areas where they occupied the apex of the food chain with little competition from either other Terapontid species or yellowbelly.

261 Banded and Barcoo grunter

Small samples of juvenile banded and Barcoo grunter detected at Pulchera Waterhole indicate that these species are also likely to undertake lengthy (approximately 100 kilometre) migrations to formerly dry ephemeral watercourses (Table 7.8). However, the absence of both of these species from any samples at S Bend Gorge in either April or August 2007 indicates that they are unlikely to have colonised this area (~250 kilometres upstream from permanent sites) following flooding in January and February 2007. Evidence from South Australia indicates that banded grunter are likely to undertake downstream migrations into the lower Diamantina (downstream of Goyder’s Lagoon) during flooding, and source populations are suggested to originate from the Georgina/Eyre Creek system (Costelloe et al . 2004). Further sampling effort in Queensland’s desert systems following flooding is therefore recommended in order to yield a larger data-set regarding the colonisation abilities of this Terapontid species.12

Hyrtl’s tandan and golden goby

Neither Hyrtl’s tandan nor golden goby were sampled in previously-dry waterholes >10 kilometres from the main channel of the Georgina, suggesting that these species are the least likely to undertake long colonisation migrations to ephemeral habitats in the Georgina/Mulligan catchments following flooding (Table 7.8). Evidence from the Thomson and Barcoo catchments similarly indicates that adult Hyrtl’s tandan are unlikely to move to recently-inundated ephemeral habitats, a trait that appears to be shared with other large-bodied species such as yellowbelly, bony bream and silver tandan, but not with spangled perch. As golden goby are predominantly benthic and slow-moving, these attributes may explain the comparative lack of vagility evinced by this species. Nevertheless, it is also possible that long distance migrations by these species did not occur on this occasion for other, unknown reasons and that future sampling events undertaken following flooding may yield a different result. If, for example, Hyrtl’s tandan, like silver tandan, undertakes long-distance migration only

12 During sampling undertaken in October 2009, and following inundation and connection in early 2009, no Terapontids were sampled from ephemeral waterholes of the Mulligan catchment. This further indicates that these species are occasional rather than regular colonists of desert systems in far western Queensland.

262 as juveniles of a certain size class (70 – 80mm SL for silver tandan) it is possible that cohorts of juvenile Hyrtl’s tandan were not large enough to complete a long upstream migration into the Mulligan before connectivity ceased. 13

Summary

The effective limit of fish migration within the eastern Simpson Desert can generally be regarded as the dunefields west of the Mulligan River channel. Although a solitary spangled perch was sampled 20 kilometres across the dunefield in April 2007, no fish were sampled at Kunnamuka Swamp (a further 20 kilometres west). When combined with results from the Thomson and Barcoo catchments, the results from this study can therefore be summarised as follows. a) Five Australian fish species are capable of migrating up to 250 kilometres in order to colonise previously dry waterholes when connection flows occur in ephemeral arid-zone rivers. b) Bony bream and silver tandan are most likely to migrate as juveniles. c) There was no direct evidence of desert rainbowfish and glassfish migrating as juveniles to ephemeral desert areas. d) Spangled perch migrate as juveniles and adults. e) Bony bream and spangled perch successfully recruit in isolated desert waterholes following colonisation. f) Yellowbelly colonise ephemeral habitats – mostly in juvenile size classes - but this is dependent on location relative to geographic range. Yellowbelly colonised ephemeral areas in the Thomson/Barcoo catchments but not in the Mulligan/Georgina.

These results reveal that for three species of Australian desert fish (spangled perch, desert rainbowfish and glassfish), there is unlikely to be an ecological or biological disadvantage associated with comparatively large numbers of adults colonising ephemeral systems, but that for up to four larger-bodied species (bony bream, silver tandan, yellowbelly and Hyrtl’s tandan) this lack of disadvantage appears to apply for juveniles only. Therefore, although these results generally support conclusions

13 During the sampling mentioned above, no Hyrtl’s tandan were found, again providing supporting evidence that this species is unlikely to migrate into ephemeral desert systems.

263 demonstrating that stochastic events like floods control and influence the subsequent distribution patterns/extent of fish species assemblages (Resh et al. 1988), species life history strategies appear to play an equally important role in structuring these assemblages in the Australian arid zone. Although Australian freshwater fish communities demonstrably colonise recently-inundated ephemeral habitats, it is important to remember that these events may often be species, size-class or area dependent, as detailed in the above discussion and demonstrated in studies elsewhere (Davey and Kelly 2007). Longer term studies of the colonisation of Australian desert river systems following inundation, connection and drying are required in order to separate identifiable patterns from general stochasticity with regard to the response of individual fish species and communities.

Determining patterns of extinction was not possible using the data collected during the current study for waterholes in the Georgina and Mulligan catchments, however these results have the potential to inform and supplement future studies aimed at determining these patterns using larger data-sets. Although some waterholes, such as Lower Lake, evinced almost 100% extinction of all species and individuals between April and August 2007, this was not the case at S Bend Gorge, where the full suite of extant species persisted until at least two weeks prior to total drying. This result supports the evidence from North American studies and indicates that even very crowded concentrations of fish are able to persist in reduced habitats as dry periods continue (Matthews and Marsh-Matthews 2003). Differences in the rate of waterbird predation at large lake habitats (such as Lower Lake and Pulchera Waterhole) may have accounted for the differences in extinction rates between open-water habitats and more sheltered channel areas (such as S Bend), and the concentration effect caused by waterhole drying may have favourably biased the detection of more species from S Bend than from open-water habitats. Nevertheless, predicting extinction rates of desert-adapted fish in Australian arid zone ephemeral systems is likely to remain difficult due to variation of both waterholes and fish assemblages following the cessation of connectivity. Fish species such as golden goby, Hyrtl’s tandan and yellowbelly, that were less likely to migrate long distances from main channel waterholes, were less likely to suffer local extirpation than highly vagile colonists such as bony bream, silver tandan, spangled perch, glassfish and desert rainbowfish. Thus it would appear that avoidance of long migrations to ephemeral systems may be

264 an extinction mitigation strategy for the less vagile species. In both dry and wet periods, the fish assemblages at permanent main channel sites that did not go dry remained extremely similar, suggesting that adaption to dry conditions (McMahon and Finlayson 2003; Bond et al. 2008; Chapter 5) is almost certainly underpinned by species’ persistence at waterhole scale. Analysis of previous studies along a much longer timeframe (Matthews and Marsh-Matthews 2003) similarly indicates that the composition of fish communities remains similar in drought-prone systems, and that localised wetting and drying events, whilst exerting large short-term impacts on abundance and species richness, do not affect the long term trajectory of fish community structure in such systems.

This chapter has demonstrated that migration to ephemeral areas by fish species in the Australian arid zone is dependent on several factors. The magnitude of hydrological disturbance appears to exert a considerable influence on the migration abilities of species and communities (Poff and Allan 1995), as following major flooding species richness and abundance were both elevated in ephemeral habitats (see also Chapters 5 and 6). Fish assemblages at ephemeral sites were also influenced by distance from permanent waterholes, with a subset of highly vagile species making long distance migrations (>100 kilometres) and some species displaying reduced likelihood of movement from safe (permanent) environments. Given the paucity of such studies (Davey and Kelly 2007; Larned et al . 2009) this finding is likely to be instructive in predicting the colonisation patterns of fish species in Australian arid-zone systems in the future. As within-catchment colonisation was primarily influenced by distance, the hypothesis stating that fish assemblage structure will vary among rather than within catchments is refuted (Hypothesis 5, Section 7.1).

265 8. Fish populations in the rivers of far western Queensland: discussion and conclusions

The preceding chapters have investigated the distribution of fish across the Lake Eyre and Bulloo-Bancannia basins, the recruitment patterns of these fish species and evidence of their ability to colonise new and/or different areas when migration pathways exist. This chapter will summarise the information presented in chapters 4, 5, 6 and 7 and consider this information in relation to existing ecological theories of river ecosystem function in order to conceptualise the behaviour of arid-zone fish communities and species in central Australian watersheds. Finally, it will draw general conclusions regarding the ecological health of catchments in far western Queensland and comment on the challenges and opportunities that exist in relation to the management and conservation of these unique and unpredictable arid-zone river systems.

8.1 Summary of data from the current study

Chapter 4 traced the hydrological history of all study sites in the Queensland Lake Eyre and Bulloo-Bancannia basins from September 2006 to March/April 2008, and presented temporal water quality data. These datasets reveal that water quality parameters such as turbidity and electrical conductivity are the most variable across the rivers of western Queensland, whereas pH, dissolved oxygen and temperature vary within more limited and expected ranges. In general, the Georgina River is characterised by comparatively clear and often saline water by virtue of contributions from a combination of saline and non-saline aquifers that underly the permanent waterholes. In the ephemeral Mulligan catchment, conductivity is also higher and turbidity lower than in the more easterly catchments. In contrast, all rivers in the greater Cooper catchment (Thomson/Barcoo/Kyabra/Cooper), the Bulloo catchment and the Diamantina catchment are characterised by highly turbid water and low conductivity levels. Perhaps most importantly, the data presented in Chapter 4 demonstrates the hydrological variability of the rivers of far western Queensland. Major flooding occurred in the Georgina and Mulligan catchments in early 2007, but

266 a period of minimal or zero flows then commenced and continued through the following summer. In contrast, the Barcoo, Thomson and Cooper catchments experienced major flooding in early 2008, but not in the preceding year. In both the Diamantina and Kyabra catchments, no overbank flooding occurred for the duration of the study, whereas in the Bulloo catchment, overbank flows occurred in both summer periods - 2006/2007 and 2007/2008. This marked spatial and temporal variability of flow regimes within the study area was highly conducive to conducting the subsequent recruitment and movement studies.

8.1.1 Fish distribution patterns

In Chapter 5, the distribution patterns of fish across the Queensland Lake Eyre and Bulloo-Bancannia basins were investigated using field data collected between 2006 and 2008. The scientific and common names of the species collected are re-iterated in Table 8.1, along with range extension or range confirmation details. Widespread species, such as bony bream, silver tandan, desert rainbowfish and spangled perch, were found across the entire study area, including isolated waterholes of the most hydrologically variable system, the ephemeral Mulligan River in the Simpson Desert (Table 8.1). Although yellowbelly, Hyrtl’s tandan and glassfish were also widely distributed, neither yellowbelly nor Hyrtl’s tandan were detected in the Mulligan catchment following flooding in early 2007, and glassfish were never sampled from the Diamantina catchment, which experienced a prolonged drying period for the duration of the study (Table 8.1). This investigation has confirmed the presence of glassfish in the Bulloo catchment, and it is possible that this species may be morphologically intermediate between Ambassid fish from the Lake Eyre and Murray-Darling basins (Jeff Johnson, Queensland Museum, personal communication) (Table 8.1). Welch’s and Barcoo grunter were distributed widely, with the exception of the Bulloo catchment for both species and the Mulligan for Welch’s grunter only, whereas banded grunters were only detected in the far west of the study area (the Georgina and Mulligan catchments) (Table 8.1). The general assumption that both Welch’s and Barcoo grunter are present in the Bulloo catchment (Wager and Unmack 2000; Allen et al . 2002) requires clarification by future surveys (Table 8.1). Cooper Creek catfish and Australian smelt were found within their known distributional

267 ranges in the greater Cooper catchment, however Cooper Creek catfish were absent from Kyabra Creek, as noted also by Arthington et al . (2005) (Table 8.1). The absence of Australian smelt from the Bulloo catchment indicates that the biogeographic origin of this species in the Lake Eyre Basin is unlikely to have been via colonisation across the Bulloo-Bancannia Basin from the north-western Murray- Darling (Hammer et al. 2007). In contrast, the presence of carp gudgeons in both the Bulloo and greater Cooper catchments indicate that this species may have accessed the Lake Eyre Basin from the east (Table 8.1). This study is the first to detect golden goby in the Diamantina catchment. This previously unknown population appears to be rare in waterholes of the Diamantina River, with only two specimens recorded between April 2007 and November 2008. In contrast, golden goby were found in comparatively large numbers in the Georgina catchment. Prior surveys of the Diamantina conducted during ARIDFLO (Costelloe et al . 2004) did not detect golden goby in either the Queensland or South Australian reaches of the Diamantina River, and golden goby were not detected in the Georgina during the only previous survey of this catchment (Bailey and Long 2001). Although golden goby presence in the Georgina catchment is mentioned in existing literature (Wager and Unmack 2000; Pusey et al. 2004), this species can currently be regarded as amongst the least known of the Lake Eyre Basin fish assemblage (Table 8.1). A single sleepy cod sampled in the Thomson catchment in November 2008 is also a new record for the Lake Eyre Basin, however this specimen is more likely to be the result of translocation than a natural range extension, as the known range of this species does not include any central Australian waterways (Allen et al . 2002; Pusey et al . 2004). Two alien fish species, goldfish and gambusia, were sampled in very small numbers from the greater Cooper catchment between 2006 and 2008, however no alien fish species were detected in any of the western catchments or the Bulloo catchment to the east (Table 8.1). The detection of golden goby in the Diamantina, sleepy cod in the Thomson and seven species in the ephemeral Mulligan catchment are significant range extensions and indicate that further fish survey work is required throughout the river systems of far western Queensland to be more confident of the full species complement (Table 8.1).

268 Table 8.1 Fish species presence in the studied catchments 2006 – 2008. C = presence confirmed, NC = presence not confirmed, E = range extension. Blank spaces indicate that the species is not known to occur in the catchment.

Species Catchment Mulligan Georgina Diamantina Thomson Barcoo Cooper Kyabra Bulloo

Nematolosa erebi E C C C C C C C Bony bream Neosiluroides cooperensis C C C Cooper Creek catfish Neosiluris hyrtlii C C C C C C C Hyrtl’s tandan Porochilus argenteus E C C C C C C C Silver tandan Retropinna semoni C C C C Australian smelt Melanotaenia splendida tatei E C C C C C C C Desert rainbowfish Ambassis sp. E C NC C C C C C/E* Northwest Ambassis or Glassfish Macquaria sp. C C C C C C C Yellowbelly Amniataba percoides E C Banded or Barred grunter Bidyanus welchi C C C C C C NC Welch’s grunter Leiopotherapon unicolor E C C C C C C C Spangled perch Scortum barcoo E C C C C C C NC Barcoo grunter Glossogobius aureus C E Golden goby Hypseleotris sp. C C C C C Carp gudgeon Oxyeleotris lineolatus E Sleepy cod (TRANSLOCATED) Carassius auratus C C Goldfish (ALIEN) Gambusia holbrooki C Gambusia (ALIEN) TOTAL NUMBER OF SPECIES 7 11 9 14 13 13 11 8 * The distribution of glassfish in the Bulloo was uncertain prior to the current study (see Chapter 3).

269 The hypotheses from Chapter 5 are presented below followed by a brief statement regarding their acceptance or refutation based on the results.

1. Fish communities will exhibit spatial differences due to catchment barriers. Specifically: a) The fish communities in the Bulloo catchment will be different from all others as it is an endorheic catchment separated from the Lake Eyre Basin rivers. This hypothesis is supported as the Bulloo contained a subset of fish species from the greater Cooper catchments. b) The fish communities in the greater Cooper catchment (Thomson, Barcoo, Cooper and Kyabra sub-catchments) will be similar to each other, but different from all other catchments. This hypothesis is supported as all greater Cooper catchments contained a similar fish community that was different from the other catchments. c) The fish communities in the Diamantina and Georgina catchments will be similar to each other as the rivers join sporadically at Goyder’s Lagoon, but different from the communities in the Bulloo and greater Cooper catchments. This hypothesis is refuted as the Diamantina fish communities were more similar to those from the greater Cooper catchment. d) The fish communities in the ephemeral Mulligan catchment are likely to be the same, or a sub-set, of those in the Georgina, the Mulligan’s parent catchment. This hypothesis is supported as the Mulligan fish communities were indeed a subset of the Georgina fish communities. 2. Fish communities will exhibit temporal variability associated with sampling time or season, and species richness will be lower in winter. This hypothesis is supported. Species richness increased in the late summer periods (often associated with flooding) and was always low in winter (as was abundance). 3. Fish presence/absence will exhibit variability depending upon antecedent hydrology, with species richness increasing following flooding.

270 This hypothesis is supported. Fish species richness was highest in the Georgina catchment following flooding in early 2007 and in the Cooper catchment following flooding in early 2008. 4. Fish presence/absence patterns will exhibit variability between permanent and ephemeral habitats (waterholes). This hypothesis can be accepted as certain species exhibited a preference for permanent or ephemeral areas, however these results were also influenced to a large degree by the magnitude of antecedent flow events and resulting migration distances to ephemeral habitats.

8.1.2 Fish recruitment

In Chapter 6, results from analyses of length-frequency patterns for each fish species sampled through time were presented and discussed with reference to the recruitment strategies these species may employ in order to maintain populations in the hydrologically variable and unpredictable environments of the Queensland Lake Eyre and Bulloo-Bancannia basins. Given the geographically large study area, the presence of rare species and the goal of comparing recruitment patterns between species, catchments and hydrological regimes, it was considered that length frequency analysis would yield a more useful and general set of data than destructive techniques.

The hypotheses from Chapter 6 are presented below followed by a brief statement regarding their acceptance or refutation based on the results.

1. The recruitment of some or all present species of fish will occur at local scale (within waterhole) during no-flow periods in the rivers of far western Queensland. This hypothesis is accepted for all species except Hyrtl’s tandan, Barcoo grunter and Welch’s grunter. 2. Fish (some or all present species) recruitment will be enhanced by periods of flow and/or flooding in the rivers of far western Queensland. This hypothesis is accepted for spangled perch, glassfish, Hyrtl’s tandan and Barcoo and Welch’s grunter.

271 3. There will be a seasonal recruitment response by fish species in the rivers of far western Queensland. This hypothesis is accepted for Australian smelt (winter), silver tandan, spangled perch, golden goby and Cooper Creek catfish (all early summer).

Species such as bony bream, desert rainbowfish, banded grunter, carp gudgeon and yellowbelly were demonstrated to be present in waterholes as juveniles throughout the year, irrespective of flow conditions (Table 8.2). In contrast, Hyrtl’s tandan appeared to be highly flood-dependent, and large cohorts of juveniles were only detected following major flooding events. Silver tandan and Australian smelt displayed a seasonally-cued breeding cycle, with silver tandan juveniles likely to be present from summer and Australian smelt from winter (Table 8.2). For smelt, this reflects their predominantly southern distribution and adaptation to breeding at low temperatures (Pusey et al . 2004). Spangled perch demonstrated strong patterns of juvenile abundance following both flooding and post-summer dry periods, however this species also spawned successfully in an isolated drying waterhole in early summer (November 2007) (Table 8.2).

Flooding appeared to enhance the recruitment potential of most species, particularly glassfish, Hyrtl’s tandan and spangled perch. These patterns of enhanced recruitment in response to flow conditions are consistent with other studies in Cooper Creek (Puckridge et al. 2000; Pritchard 2004; Balcombe and Arthington 2009).

Small sample sizes made determining the recruitment patterns of Cooper Creek catfish, golden goby and two Terapontid species (Barcoo and Welch’s grunter) more difficult. The results available indicate that Cooper Creek catfish and golden goby probably breed annually in summer, thus capitalising on summer flows and floods if they occur, whereas the Terapontid species are likely to be more flow-dependent as juvenile cohorts were only detected following flooding (Table 8.2). The beneficial effects of flooding for juvenile cohort success has been noted for many Terapontid species in northern Australia (Pusey et al . 2007), and for silver perch, Bidyanus bidyanus , in the Murray-Darling Basin (Lintermans 2007). The extremely small numbers of goldfish, gambusia and sleepy cod sampled during this study did not

272 permit any further speculation on recruitment patterns of these species in far western Queensland.

Table 8.2 Summary of no-flow and flow dependent recruitment for fish species from the study area. No-flow recruitment Flow-dependent recruitment Seasonal Continuous Winter Summer Australian smelt Silver tandan Bony bream Hyrtl’s tandan Golden goby Carp gudgeon Welch’s grunter Cooper Creek catfish Banded grunter Barcoo grunter Spangled perch Yellowbelly Rainbowfish Glassfish

8.1.3 Fish movement

In Chapter 7 the migration and colonisation abilities of extant fish species were investigated. Prohibitively high costs and the spatial scale of the study prevented the effective use of tag-recapture and tracking fish using telemetry, however the location of fish sampling sites in ephemeral catchments and areas before and after flow and flood events provided accurate data relating to fish colonisation and extinction events. Results from the Georgina, Mulligan, Thomson and Barcoo catchments were analysed in order to examine the composition of fish assemblages in permanent and ephemeral waterholes and the distances fish species migrated in ephemeral catchments following re-connection of waterholes.

The hypotheses from Chapter 7 are presented below followed by a brief statement regarding their acceptance or refutation based on the results.

273 1. Fish species and/or communities will migrate to ephemeral areas following inundation and re-connection. This hypothesis cannot be broadly accepted, however qualification for individual species and communities is required as outlined below. 2. Colonisation ability in newly-inundated areas will be related to size classes or life-cycle stage of colonising species. This hypothesis is supported, as many of the studied species were more likely to colonise as juveniles, particularly following connection flow events. 3. Migration distance will influence the composition of fish communities in newly- inundated ephemeral areas. This hypothesis can be accepted, as smaller subsets of fish species were shown to colonise ephemeral waterholes at increasing distance from permanent waterholes. 4. Antecedent hydrology will influence the composition of fish communities at formerly dry sites. This hypothesis is supported as major flooding was demonstrated to elicit increased species richness, more varied size classes and increased overall abundance of fish when compared to less extensive connection flow events. 5. Fish assemblage structure will vary among, rather than within, catchments at previously-dry ephemeral sites. This hypothesis is refuted as migration distance influenced the composition of fish assemblages within catchments.

Silver tandan, desert rainbowfish, glassfish, bony bream and spangled perch were all demonstrated to undertake speculative or opportunistic upstream migrations of up to 250 kilometres into formerly dry desert river habitats, and spangled perch were also shown to be capable of making use of sporadic overland flows in order to colonise highly ephemeral pools in the vicinity of desert sand dunes. Species such as desert rainbowfish and glassfish were shown to colonise as adults, whereas larger-bodied species such as silver tandan, bony bream and yellowbelly were more likely to colonise ephemeral habitats as juveniles. This suggests that larger, adult fish of these species are more likely to remain in secure and potentially permanent habitats such as deeper waterholes in river channels (Schlosser 1987; Capone and Kushlan 1991). Interestingly, colonisation of ephemeral habitats by yellowbelly was catchment- dependent, with no evidence of this behaviour in the westerly Mulligan catchment

274 following major flooding. Similarly, Hyrtl’s tandan did not appear to colonise ephemeral catchments as adults, suggesting that for some species the opportunities afforded by newly-created aquatic habitat were overshadowed by the need to maintain populations of breeding-age fish in permanent waterholes. Although it is possible that some species may have abandoned opportunistic migration behaviour due to the unpredictability of flow in western Queensland, more thorough sampling of the more westerly catchments in both wet and dry years is required, especially with regard to yellowbelly and Hyrtl’s tandan, before such a notion can be accurately tested. Both Barcoo grunter and banded grunter were found to be highly vagile species likely to undertake migrations of at least 100 kilometres upstream as juveniles, however the sample sizes demonstrating this behaviour were small. Small samples of Welch’s grunter, Cooper Creek catfish and golden goby also rendered predictions regarding their colonising behaviour speculative, but it appears likely that all three species undertake at least short migrations (<10 km) to ephemeral habitats when connection flows occur. Spangled perch of all size classes were detected in ephemeral waterholes, indicating that this species almost certainly possesses the most opportunistic life history of all species studied here.

8.2. Concepts of fish ecology in Australian arid-zone rivers

The drawbacks of ascribing theories developed in temperate and tropical river systems to Australian dryland rivers have been described by Australian scholars and principally relate to the hydrological unpredictability of Australia’s inland systems (Puckridge et al . 1998; Humphries et al. 1999; Puckridge et al. 2000). Consequently, concepts underpinned by the presence of regular flow events (FPC: Junk et al . 1989) or the presence of a continually connected river (RCC: Vannote et al. 1980) generally do not describe the functioning of Australia’s inland rivers particularly well (Walker et al . 1995; Leigh 2008). In the simplest terms, neither constantly wet river channels nor predictable flow events occur in the rivers considered during this study. In terms of primary production, the Riverine Productivity Model of Thorp and Delong (1994 and 2002) is applicable as it emphasises the importance of localised within-stream productivity underpinning aquatic food webs. In terms of models particularly relevant

275 to fish, the most suitable is the ‘flow-pulse’ concept described by Puckridge (1999), where variable flows, from zero to major flooding, elicit particular biological responses ranging from localised recruitment within an individual waterhole or river reach to widespread migration and enhanced recruitment success following floodplain inundation.

The current study demonstrates the highly variable hydrology of the rivers of far western Queensland and gives credence to the relevance of the flow pulse concept. A major flood occurred in January and February 2007 in the Georgina catchment, but during the following summer (2008) no waterholes in the Georgina valley became connected or were recharged by within-channel flow. This situation was reversed in the Thomson/Barcoo/Cooper, with flooding occurring in summer 2007/08 but not the previous year. More extreme examples include the Kyabra catchment, which received no overbank flows for the duration of the study period (September 2006 – March/April 2008), and the Mulligan system, a desert river which is generally dry but which flowed briefly following flooding in January and February 2007. These hydrological patterns support the flow pulse concept advanced by Puckridge (1999), as habitats that filled following small flows were demonstrated to act as fish nurseries, whereas areas experiencing major flooding demonstrated increased species richness, more varied size classes and increased abundance of all present species. It is noteworthy that recruitment also continued to occur in both drying catchments (Diamantina, Kyabra) and drying waterholes (Pulchera) for a range of species.

The unpredictable nature of Australian inland rivers is extreme (Puckridge et al . 1998), and therefore the life histories of extant species exhibit general or specific traits and strategies consistent with such variable abiotic factors. In recognition of this, work completed on Australian fish species has sought to investigate the role played by variable flow in structuring fish assemblages and fish recruitment patterns. Humphries et al . (1999) found little evidence of reliance on flows and flooding for the recruitment of certain fish species in the southern Murray-Darling. Puckridge (1999) suggested that all flows, regardless of size, are likely to influence recruitment and movement behaviour of fish in the downstream Cooper at Coongie Lakes in South Australia. More recently, research from Queensland reaches of Cooper Creek and the Thomson River have demonstrated that the majority of fish species do not rely on

276 flooding for recruitment (Arthington et al. 2005; Balcombe and Arthington 2009). Perhaps unsurprisingly, completed studies have found that increased flows and the inundation of floodplains and formerly dry areas increase primary production and result in population ‘booms’ and that prolonged dry periods result in the opposite (population busts: Arthington et al . 2005; Bunn et al. 2006). Nevertheless, it is salient to remember that prolonged dry periods – ‘busts’ – occur far more frequently than flow events in central Australia. Consequently, it may be useful to interpret Australian arid zone rivers and their ecology as predominantly ‘dry’ (a series of disconnected waterholes) and only very occasionally wet when brief and sporadic connection flows or floods occur. As alluded to above, both the RPM (Thorp and Delong 2002) and the flow pulse model (Puckridge 1999) have application to such systems, despite the fact that the RPM is in essence a riverine-based model, as opposed to an individual waterhole-based productivity concept. Nevertheless, it is tempting to consider the aridity of central Australian catchments through to their logical end-point – extreme drying – and re-consider the most suitable way to represent such systems. During multi-year dry periods, a conceptual model to explain fish ecology in the arid zone should be built upon a foundation of processes occurring at waterhole scale only rather than from the perspective of the conditions created by variable and unreliable flooding. This consideration renders theories predicated on the effects of flows and flood pulses as far less applicable, and instead suggests that fish communities in Australian arid-zone aquatic systems are driven by and derived from the populations that exist in isolated permanent habitats, with flow of whatever magnitude a stochastic event.

Larger waterholes within arid zone river systems retain water for longest and are therefore often termed ‘refuge’ waterholes (Hamilton et al . 2005; Costelloe et al. 2004; Arthington et al . 2005; Dekar and Magoulick 2007). In the context of the current study, refuge waterholes include Springfield on Kyabra Creek, Murken and Shed waterholes on Cooper Creek, Hunter’s Gorge in the Diamantina and Parapituri in the Georgina catchment (see Chapter 4). These waterholes have not dried since September 2006 (personal observation) and had not dried for many years previously. As examples, none of the waterholes mentioned have dried in living memory – a period in excess of 100 years (Silcock 2009; Bob Morrish, Sandy Kidd, David Smith, John Clemments, Shane McGlinchey, personal communications). In general, larger

277 representatives of larger-bodied fish species, such as yellowbelly, the larger Terapontids, Hyrtl’s tandan and Cooper Creek catfish, were more common in these deeper, more permanent waterholes during the current study. Nevertheless, Chapter 7 demonstrated that migration to ephemeral waterholes following flooding is a common behavioural trait for many fish species, suggesting that the permanence of a waterhole or channel does not preclude movement and subsequent colonisation for a number of species. Fish survival in the Australian arid zone is therefore underpinned by the ability to complete life cycles in permanent waterholes and the ability to migrate to and benefit from colonisation of non-permanent (and other permanent) environments when stochastic flow events occur. Perhaps unsurprisingly, similar strategies have been noted in arid zone fish communities elsewhere (Fausch and Bramblett 1991; Labbe and Fausch 2000). Consequently, rather than being reliant on random flow events, it seems far more likely that the life history strategies of fish in such systems are in fact geared towards maintaining populations during periods of no flows. Chapter 6 demonstrated that this is indeed the case, as almost all fish species recruited independently of high flows or flooding.

When flow events occur, colonising fish are derived from source populations in large or refuge waterholes and disperse to new habitats in previously dry areas. This was demonstrated in the Mulligan catchment during the current study, when up to seven species migrated over 100 kilometres and five species migrated up to 250 kilometres (see Chapter 7). The waterholes in the Mulligan became disconnected upon flood recession, and the colonists became extirpated within the entire Mulligan catchment in less than a year. Despite the rapidly-drying environment, bony bream and spangled perch bred in the waterholes as they receded, thus providing further supporting evidence for Australian native species breeding under no-flow conditions, and reinforcing one of the central themes of this thesis relating to the inherent opportunism of fish species in the Australian arid zone. Ecologically, the long migrations and subsequent extirpation of fish assemblages, particularly in the waterholes of the Mulligan River, demonstrate that opportunism is extreme for many species. It seems likely that these species either produce a surplus of eggs and recruits when flooding occurs, or that elevated flows increase the number of individuals that successfully recruit from recently-spawned cohorts. The presence of an abnormally- sized and formed silver tandan in Pulchera Waterhole (Mulligan catchment) in April

278 2007 suggests that the latter is more likely than the former explanation, and that when new aquatic habitat is created, the chances of survival increase for all species, cohorts and individuals (including those that are deformed) capable of migration. Nevertheless, this behaviour results in massive losses to many species through extirpation in drying waterholes (Chapter 7), suggesting that the biological disadvantage to such species is minimal, and that the risks associated with migrating long distances to ephemeral systems must be outweighed by the benefits (Kozakiewicz and Szacki 1995). These benefits include access to resources and – presumably – the possibility of reaching waterholes that may persist until the next re- filling event and thereby permit the establishment of long-term populations, or connection to more favourable habitats. It can be inferred, therefore, that the biogeographic expansion and contraction of fish species’ distributional ranges in Australian desert systems continually changes due to the presence of permanent and temporary habitat, and that over longer time periods a combination of habitat availability and migration pathways has shaped the catchment and basin-wide fish faunas that currently exist. However, populations must exist – and must persist - at waterhole scale in order for such processes to continue to occur.

Despite the over-arching influence of flow in sustaining aquatic systems in inland Australia along extended temporal scales (Walker et al. 1995), describing the fish communities of Australian arid-zone aquatic ecosystems using population models presents a useful alternative to theories predicated on flow at smaller scales. Under such models, reliance on flow, or flow-driven processes, is not necessarily a pre- requisite for the continued survival of fish communities, and flow instead becomes a random abiotic factor similar to drought. The fundamental idea of a metapopulation, where local extirpation is balanced by local recolonisation to the benefit (and for the on-going maintenance) of the regional population (Levins 1969; Hanski and Gilpin 1997), was investigated with reference to fish communities in Oklahoma by Gotelli and Taylor (1999). Their research demonstrated that fish assemblage colonisation and extinction probabilities at ten sites over ten years in the Cimarron River were related more to the position of sites within the stream gradient than they were to the proportion of sites occupied (Gotelli and Taylor 1999). Similarities are immediately obvious within the dataset compiled during the current study, for the most distant sites in ephemeral rivers (such as those close to the headwaters of the Mulligan River) were

279 colonised by a comparatively smaller subset of present fish species than sites closer to the main channel of the Georgina River, and these populations were also the first to become extinct. Thus it would appear that, at the least, the position of a particular site within a river system should be incorporated into metapopulation models if they are to be applied to stream fish communities (Gotelli and Taylor 1999). This consideration has particular relevance to Australian arid zone systems, where the spatial and temporal distribution of aquatic habitats is – or can be - highly fragmented.

During the current study, extirpation of entire fish assemblages was recorded in isolated waterholes of the Mulligan catchment during a drying phase, yet populations always persisted through time in permanent waterholes that became connected to these areas during flows or flooding. Following re-wetting, fish were present in ephemeral areas, however these fragmented populations were demonstrated to contain differing species (and size classes), and the composition of these communities was demonstrated to be dependent upon the distance of the ephemeral habitat from permanent water and the magnitude of the flow event. This evidence suggests that the concept of a metacommunity (as opposed to a metapopulation: Larned et al. 2009) may be more applicable to aquatic systems in the Australian arid zone, where the populations are too fragmented to reach ‘equilibrium’ (Levins 1969) and instead present as diversified communities spread out within an extended landscape. Larned et al. (2009) suggest that the disconnected refuge habitats that exist in ephemeral rivers comprise a metacommunity that is frequently impacted by variable hydrology, and that colonisation patterns will be driven by the dispersal capability of extant species – a conclusion reinforced by the results from this study. Dispersal from permanently wet areas to less favourable habitat areas frequently results in extirpation of colonising individuals and entire assemblages, yet it is the very presence of these permanent habitats that enables the metacommunity to exist.

Modifying source/sink population models (Pulliam 1996) is a potentially useful and straightforward way of conceptually representing fish communities in areas affected by drought or aridity (Magoulick and Kobza 2003). Source populations were originally defined as those habitat patches where recruitment exceeds mortality, and sink habitats as patches where mortality exceeds recruitment (Pulliam 1996). In the context of Australian arid zone rivers, source populations of fish species can be said to

280 exist in permanent waterholes, as these areas provide the highest quality (and frequently the only) aquatic habitats in a given catchment or river reach, whereas ephemeral waterholes usually act as sink habitats, where many migrants face an uncertain future (Chapter 7). The fact that large adult fish were most commonly sampled in permanent waterholes and juvenile fish were often sampled in ephemeral habitats during the current study lends support to this source-sink concept, as it suggests that safer, source habitats are more likely to be inhabited by species or individuals best able to reproductively maintain their populations. When considered in this way, source waterholes in the arid zone can be said to mimic the ecological role of mainland populations as proposed by McArthur and Wilson (1967). Extending the island biogeography analogy, it is relatively straightforward to consider source waterholes scattered throughout the landscape as a series of ‘mainlands’ that are occasionally linked to ephemeral systems (‘islands’) and from which the ephemeral habitats inherit populations (Harrison and Hastings 1996). More specifically, in Australian arid zone source waterholes, the majority of fish species (such as desert rainbowfish, bony bream and yellowbelly) appear to breed on a year-round basis or seasonally (such as Australian smelt and silver tandan). If no connection flows occur, recruitment may be comparatively poor due to the increased likelihood of cannibalism, predation and lack of access to resources afforded by a single, isolated habitat. However, recruitment is almost certainly successful enough to allow individual species to maintain their populations through time, albeit at occasionally low or ‘maintenance’ abundances. The multiple size-classes of many species detected in permanent waterholes throughout the field survey suggests that this is indeed the case in the catchments studied (Appendix 3). During extreme extended dry periods it is conceivable that all ephemeral waterholes (and many ‘permanent’ waterholes) may themselves become sink habitats, but it is likely that some permanent waterholes would persist, and that subsequent colonisation of river reaches and catchments could occur upon re-wetting. Although the degree to which groundwater inputs sustain permanent waterholes in far western Queensland is unknown, some waterholes in the Georgina catchment are sustained by saline and non-saline aquifers (Silcock 2009; Shane McGlinchey, Badalia Station; Stephen Bryce, Glenormiston Station; personal communications) and the documented presence of Great Artesian Basin aquifers throughout the Lake Eyre Basin suggest that some areas and ecological communities, such as the spring complexes at Edgbaston and Elizabeth Springs, are sustained in the

281 absence of surface water run-off (Fairfax et al . 2007). Alternatively, if connection flows or floods occur, emigration from the source population is possible, and fish move to a combination of other source habitats (thus providing opportunity for genetic transfer) and ephemeral – or sink – habitats, where they frequently become extirpated within a comparatively short period of time (Chapter 7). Consequently, fish from source habitats and surrounding floodplains provide a ‘surplus’ of recruits and individuals during ‘booms’, and many of these undertake migrations to less- favourable sink habitats. In arid Australia, sink habitats often dry before they are re- wetted, and the colonists are extirpated. Sink habitats are then re-colonised again by another suite of recruits and individuals originating from the same – or different - source populations when sporadic flow events next occur. In the Georgina catchment, as an example, only ten waterholes are known to have remained wet during living memory (Jenny Silcock, Queensland Environmental Protection Authority, personal communication), yet fish are present in highly isolated parts of the catchment that are usually dry following flood recession (Chapter 7), and these habitats evaporate rapidly following disconnection (personal observation). Under abnormal conditions (such as a series of wet years, or a wet winter), sink habitats could be re-filled and become source habitats for more distant ephemeral areas. The working model for fish ecology in the Australian arid-zone proposed in Chapter 1 adequately conceptualises the source/sink nature of waterholes and migration pathways documented during the study as it includes the notion of multiple spatial scales (waterhole, reach and catchment) and the unpredictability of flow events (Figure 8.1).

282 Waterhole scale andWaterhole processes responses

Dry

Flood Waterholes disconnected Life-cycles completed Migration across floodplain within-waterhole and along migration pathways Catchment/sub-catchment scaleCatchment/sub-catchment processesresponses and

Late Dry Connection flow Waterholes disconnected Reach scale processes Reach scale andresponses Life-cycles completed Migration of adults/juveniles within-waterhole along migration pathways Local extirpations

Figure 8.1 A working model for fish ecology in the Australian arid-zone (from Chapter 1)

At a much larger temporal scale, it is possible to envisage this source/sink model accounting for biogeographic alterations to fish faunas as regions become wetter or drier. The results from the current study suggest that a suite of species that are both highly vagile and in possession of flexible recruitment strategies exhibit the most pronounced colonisation ability within the Queensland Lake Eyre Basin (Figure 8.2). It can therefore be assumed that biogeographic range expansion is likely to occur for spangled perch, bony bream, desert rainbowfish, silver tandan and glassfish when migration pathways are open, as they are the most capable colonising species (Figure 8.2). Conversely, species such as Hyrtl’s tandan, Welch’s grunter and Cooper Creek catfish appear to possess more conservative colonisation and recruitment strategies and therefore would not be expected to fill the roles of ‘pioneer species’ (Capone and Kushlan 1991) to the same extent (Figure 8.2).

283 250 Spangled perch Bony bream Glassfish Rainbowfish Silver tandan

100 Barcoo grunter Banded grunter

10 Hyrtl’s tandan Golden goby Yellowbelly Welch’s grunter Australian smelt Carp gudgeon Cooper Creek catfish Migration distance from from Migration permanent (km) water distance

Flow-dependent No-flow

Recruitment strategy

Figure 8.2 The relationship between migration capability or vagility and recruitment strategies for fish species from the rivers of far western Queensland.

By simplifying the stated working model further (Figure 8.1), a generalised source/sink model that represents fish communities accurately in the arid zone can be created (Figure 8.3). Fundamentally, this model is wholly underpinned by biological processes occurring at waterhole (‘source’) scale but also includes the effects of unpredictable connection events that influence processes at larger spatial scales (Figure 8.3). This concept has been suggested previously in relation to fish communities affected by drought (Magoulick and Kobza 2003), and is therefore likely to apply to systems where ‘drought’ conditions persist for the majority of the time. Flow events, no matter how large or small, are unreliable and unpredictable as a cue for spawning and recruitment. If flows occur, fish species in the source waterholes can take advantage of them and move to other areas – frequently less-safe sink areas, but occasionally to other and possibly better refuge habitats. If they do not move, the source populations maintain themselves within the waterhole until stochastic flow events occur or until the source waterhole itself becomes a sink and possibly becomes dry (Figure 8.3).

284 a) No or low-flow scenario

Seasonal and continual breeding continues

Recruitment success dependent on waterhole size, resources, predation. Renewed source Source waterhole waterhole population population Recruitment compromised for flow- dependent species

b) Connection event to major flooding scenarios

Sink waterhole

Flow event population adults)Migration (juveniles and/or (ephemeral)

Seasonal and continual breeding continues

Recruitment success enhanced for most species.

Source Recruitment facilitated for flow- waterhole dependent species Renewed source population waterhole population Migration pathways open for varying lengths of time and distance

Other source waterhole populations

Time

Figure 8.3 A modified source/sink model for fish communities in Australian arid zone rivers.

8.3. Fitting the conceptual model: examples from the Queensland Lake Eyre Basin

The following examples display the differences between source and sink waterholes within the study area through time (Figure 8.4). Waterloo, a source waterhole in the Thomson catchment, had experienced no flows in the three months prior to September 2006, and did not receive any flows until after sampling was completed in January 2007. During this drying period, Waterloo provided demonstrated habitat (refuge) for adult Hyrtl’s tandan and Cooper Creek catfish, and both refuge and recruitment opportunities for bony bream, yellowbelly, Australian smelt, carp gudgeon and silver tandan. Following within-channel connection flows in January, July and November

285 2007, Waterloo provided refuge habitat for a similar suite of species, and new cohorts of species such as bony bream, yellowbelly and smelt were sampled consistently. Barcoo grunter and spangled perch were first sampled in November 2007, possibly due to migration from other areas during a within-channel flow event. Following major flooding in early 2008, recruitment of almost all species was detected, including Barcoo grunter, spangled perch and Hyrtl’s tandan. During all connection events (January, July and November 2007 and December 2007 – February 2008), migration from Waterloo to other areas in the Thomson catchment became possible.

Pulchera, an ephemeral waterhole in the Mulligan, demonstrates the role played by sink waterholes in the Queensland Lake Eyre Basin (Figure 8.4). Pulchera was dry in November 2006, and filled and re-connected with the Georgina River during major flooding in January and February 2007. A suite of colonising species moved into Pulchera waterhole during flooding and subsequently became isolated by early April (see Chapter 7). By August, species such as banded grunter and desert rainbowfish were absent from the samples, and by November only four of the seven colonising species remained: bony bream, spangled perch, glassfish and silver tandan. Although new cohorts of both bony bream and spangled perch were sampled in Pulchera waterhole in November 2007, the entire community became extirpated by March/April 2008. Thus, for Pulchera, migration into flooded areas was only possible on one occasion (April 2007), and the absence of subsequent flow events led to the evaporation of the waterhole and associated mortality of the entire fish community.

286

A source waterhole through time – Waterloo, Thomson catchment

September 2006 December 2006 January 2007 April 2007 August 2007 November 2007 March/April 2008

Connection Connection Connection Major flood flow late flow November, flow July, Isolated Isolated Isolated January, and connected Dec - Jan then isolated then isolated during sampling

Recent recruitment: Recruitment: Recruitment: Recruitment: Recruitment: Recruitment: Recruitment + refugia Bony bream Bony bream Bony bream Bony bream Yellowbelly Bony bream Silver tandan Yellowbelly Yellowbelly Yellowbelly Yellowbelly Silver tandan Yellowbelly Barcoo grunter Smelt Smelt Smelt Smelt Smelt Bony bream Silver tandan Carp gudgeon Refugia: Cooper Ck catfish Refugia for adults: Carp gudgeon Refugia: Bony bream Refugia: Glassfish Bony bream Refugia: Bony bream Yellowbelly Silver tandan Hyrtl’s tandan Yellowbelly Refugia: Bony bream Hyrtl’s tandan Hyrtl’s tandan Hyrtl’s tandan Spangled perch Smelt Bony bream Yellowbelly Cooper Ck catfish Silver tandan Yellowbelly Yellowbelly Hyrtl’s tandan Yellowbelly Hyrtl’s tandan Smelt Cooper Ck catfish Carp gudgeon Hyrtl’s tandan Cooper Ck catfish Barcoo grunter Refugia only: Silver tandan Smelt Smelt Carp gudgeon Cooper Ck catfish Carp gudgeon Spangled perch Desert rainbowfish Smelt, gudgeon

A sink waterhole through time – Pulchera, Mulligan catchment

November 2006 April 2007 August 2007 November 2007 March/April 2008

Major flooding Dry January and Dry February Isolated Isolated 2007

Colonisers: Recruitment: Recruitment: All species extinct Bony bream No new recruits of any Bony bream Spangled perch species detected Spangled perch Silver tandan Desert rainbowfish Refugia: Refugia: Glassfish Bony bream Bony bream Barcoo grunter Spangled perch Spangled perch Banded grunter Glassfish Glassfish Barcoo grunter Silver tandan Silver tandan

Figure 8.4 Case studies demonstrating the source/sink concept by reference to the temporal changes in the fish assemblage at a source waterhole (Waterloo; top) and a sink waterhole (Pulchera; bottom).

287 Waterloo and Pulchera waterholes provide instructive examples of the way in which source and sink waterholes provide habitat for fish in the Australian arid zone. Source habitats, such as Waterloo, are populated by adult members of extant species, and recruitment continues in both dry and wet periods for many species. Sink habitats, such as Pulchera, are colonised by vagile species following the opening of migration pathways, and although recruitment may occur if these areas hold water for long enough, the end result is extirpation of all colonists and their progeny unless subsequent rainfall or flooding occurs. For these vagile species, the possibility of benefit from new and/or better conditions must outweigh the risk of encountering worse conditions. Biogeographic range expansion is likely to occur for spangled perch, bony bream, desert rainbowfish, silver tandan and glassfish when migration pathways are open, as they are the most capable colonising species.

288 8.4. Fish communities within the Queensland Lake Eyre and Bulloo-Bancannia basins: current status, threats, management implications and recommendations

The work presented in this dissertation demonstrates that native fish populations within all catchments west of the Murray-Darling Basin in Queensland are in good condition. These results present a stark contrast to those from the neighbouring (and highly modified) Murray-Darling Basin, where alien fish species are consistently common and several native species are threatened or declining (Faragher and Harris 1994; Harris and Gehrke 1997; Faragher and Lintermans 1997; Koehn and Mackenzie 2004; Balcombe et al . 2006; Lintermans 2007; Rayner et al. 2009). In particular, populations of alien fish species appear to be present in very small numbers and in a limited range of catchments (only the Thomson, Barcoo and Cooper) in the Queensland Lake Eyre and Bulloo-Bancannia basins.

Nevertheless, it is prudent to remember that the spatial distribution of fish should still be considered poorly known in far western Queensland, as many tributaries and reaches have not been surveyed, particularly in the upper Thomson and Barcoo catchments, the Bulloo and Wilson catchments, the Diamantina River below Diamantina Lakes, and areas in the Georgina catchment not considered during this study. The number of range extensions for fish species documented during the current study (including golden goby in the Diamantina catchment and the suite of species recorded from the Mulligan catchment (see Chapter 5 and Table 8.1 above) demonstrates the lack of basic biogeographic knowledge that exists regarding aquatic systems in outback Queensland. Work that has the potential to fill some of these knowledge gaps has recently (November 2009) been instigated by the Queensland Department of Environment and Natural Resources in the Cooper catchment and similar broad-scale surveys are planned for the entire Queensland Lake Eyre Basin, commencing in 2010 (Jonathon Marshall, Department of Environment and Natural Resources, personal communication). However, these surveys need to be conducted across a wide enough geographical area and along a suitable temporal scale such that they may yield meaningful data comparable with existing records of species

289 distributions, abundance and temporal dynamics (Arthington et al . 2005; Balcombe et al. 2007; this study).

The rivers of the Queensland Lake Eyre and Bulloo-Bancannia basins remain comparatively un-affected by river regulation and the associated suite of factors that has been implicated in the decline of native fish species in other Australian rivers. These factors include flow regulation (Walker et al. 1995; Arthington and Pusey 2003), barriers to movement such as dams and weirs (Reynolds 1983; Mallen-Cooper et al . 1995), cold water pollution resulting from deep discharge from reservoirs (Astles et al. 2003), the impact of commercial fishing (Reid et al . 1997; Humphries and Winemiller 2009) and the impact of alien species (Koehn and Mackenzie 2004). Results from this study tend to indicate that in areas where native fish populations remain intact, and where ecosystems remain comparatively unaltered, opportunities for alien fish establishment may be limited, possibly due to the presence of native predators and the highly variable nature of the hydrological and related conditions (Bunn and Arthington 2002; Costelloe et al . 2004). Nevertheless, populations of both gambusia and goldfish continue to be detected, particularly in the Thomson/Barcoo/Cooper catchments, and remain an ever-present threat to the aquatic ecosystems of far western Queensland. Gambusia occur in infestation-level populations in the spring complex at Edgbaston, where they threaten the survival of one of Queensland’s smallest and rarest freshwater fish, the red-finned blue-eye, Scaturiginichthys vermeilipinnis (Fairfax et al . 2007; Kerezsy 2009). Consequently, the possibility that gambusia can and may infest aquatic systems in the arid and semi- arid zones, with potentially catastrophic results, should be considered by agencies charged with the management of these systems.

Other alien species such as the cane toad, Bufo marinus , and translocated species such as the redclaw crayfish, Cherax quadricarinatus , are also present in the catchments of far western Queensland (personal observation), however their impacts on local ecosystems are currently poorly known. The presence of carp, Cyprinus carpio , in the Murray-Darling Basin (to the east of the study area) and tilapia (various species) in the Gulf of Carpentaria to the north represent serious threats (Koehn 2004; Canonico et al. 2005). These species have the potential to cause immeasurable damage to the rivers of the Lake Eyre and Bulloo-Bancannia basins should they become liberated

290 and spread (Koehn 2004; Canonico et al. 2005). The presence of a sleepy cod in the Thomson catchment during sampling in November 2008 indicates that translocated, non-endemic fish species may also pose a threat to the aquatic ecosystems of far western Queensland. Research initiatives aimed at investigating the impacts of alien and translocated species, as well as potentially controlling such species, should be undertaken within the studied catchments.

It is important to point out that human-induced vectors for the liberation of alien fish species include a range of activities undertaken (or likely to be undertaken) in the study area. Recreational fishing is a popular pastime throughout Queensland’s outback rivers, and aquaculture is a popular addition to farm businesses that has been promoted by the Queensland Department of Primary Industries and Fisheries (Vanessa Bailey, Queensland Department of Environment and Resource Management, personal communication). Both practices, as well as illegal fishing, are likely to have a negative effect on the biota of the rivers of far western Queensland, because few anglers, aquaculturists and government agency employees possess a detailed knowledge of the endemic fish fauna. As examples, many anglers are known to carry imported live bait, such as redclaw crayfish and goldfish, through the area (Gary Muhling, Queensland Department of Primary Industries and Fisheries, Longreach, personal communication), and at least two translocated species (Murray cod, Maccullochella peelii , and yellowbelly, Macquaria ambigua , from the Murray- Darling Basin) have been stocked in Lake Eyre Basin waterways by the Queensland Department of Primary Industries and Fisheries (Bailey and Long 2001).

Prevention of future liberation of alien and translocated fish species in outback Queensland can only be achieved through a simple, persistent information and communication campaign targeted at local fishers, local graziers, and occasional visitors. As well as communicating information regarding the threats associated with translocated and alien species, such a campaign could also actively counter perceived attitudes relating to the benefits of stocking fingerlings in large waterholes, and instead encourage catch-and-release rather than retention of caught broodstock (Peter Kind, Queensland Department of Primary Industries and Fisheries, Brisbane, personal communication). Additionally, the corollary to pastoral prosperity (Durack 1959; Bowen 1987) has almost certainly been an occasionally irresponsible attitude by some

291 landholders towards caring for the aquatic environments in what is often marginal grazing land. Consequently, examples of refuge waterholes and springs that are over- used by domestic stock are not uncommon in outback Queensland (personal observation). If restoration is to be considered a genuine option these areas must be identified and positive remedial actions prioritised.

Populations of commonly occurring fish species within the Queensland Bulloo- Bancannia and Lake Eyre basins were sampled in comparatively large numbers throughout this study (see Chapter 5). Additionally, the observed populations frequently included representatives of all size-classes, suggesting that recruitment of many species occurs along a comparatively constant timeframe. This indicates that the majority of fish species from far western Queensland have adaptable or opportunistic life history strategies and that they can reproduce independently of flow events (see Chapter 6). However, this study also found that all but a few species were highly vagile and capable of long migrations to previously-dry areas when connectivity pathways became open following flow events (see Chapter 7). It therefore seems likely that a combination of an opportunistic life history strategy and an un-altered hydrograph may be the most important factors contributing to the success of these species. Indeed, alteration of the natural flow regime in the Bulloo, greater Cooper, Diamantina and Georgina catchments could possibly impact negatively on the populations of all extant fish species and create a degraded environment similar to that existing in the western rivers of the Murray-Darling Basin, where native fish communities are in obvious decline (Harris and Gehrke 1997; Rayner et al. 2009). It is therefore strongly recommended that no further water abstraction or flow alteration devices are installed on any of the aforementioned rivers, and it is suggested that these recommendations be recognised formally in legislation (such as the revised Cooper Water Resource Plan – Mark Foreman, Queensland Department of Environment and Natural Resources, personal communication). It is relevant to mention that although current legislation and governance of the western rivers appears to be becoming more influenced by conservation objectives, mining and pastoralism remain the most common and lucrative land-use activities in the area considered. Consequently, although the outcome of the current 10-year review of the Cooper Creek Water Resource Plan is to be announced in mid-2010, possibly along with a ‘Wild Rivers’ declaration for the Cooper catchment (Mark Foreman, Department of Environment

292 and Resource Management, personal communication), plans for a large coal mine in the eastern Lake Eyre Basin have also been announced (Rupert Quinlan, The Wilderness Society, personal communication). Given that a tailings dam spill in February 2009 from the Lady Annie mine in the upper Georgina catchment resulted in 100% mortality of the riverine fauna for a distance of approximately 50km downstream (Shelley Curr, Yelvertoft Station, and Neil Maver, Queensland Department of Environment and Resource Management, personal communications), balancing the environmental requirements of river systems and the industrial requirements of governments and communities in far western Queensland is unlikely to be straightforward or devoid of future controversy. Nevertheless, given the example of the Murray-Darling Basin, it would seem unwise and irresponsible to allow the development of any water-reliant infrastructure in the drainages of outback Queensland without extremely stringent environmental controls.

Although this study has demonstrated useful information relating to the life-histories of many extant species in the Bulloo-Bancannia and Lake Eyre basins (Chapters 6 and 7), knowledge gaps remain that can only be addressed through investigation at either expanded spatial and temporal scales and/or through species-based research with a strong laboratory and/or experimental component. Opportunities certainly exist within far western Queensland to restore populations of the three extant threatened fish species: the red-finned blue-eye and Edgbaston goby, Chlamydogobius squamigenus , in the Cooper catchment and the Elizabeth Springs goby, Chlamydogobius micropterus , in the Diamantina catchment. All three species are listed under state (NCA 1992) and national (EPBC 1999) endangered species legislation, and also under the International Union for the Conservation of Nature criteria as critically endangered (IUCN 2010). Falling under the jurisdiction of the Queensland Department of Environment and Resource Management, the Elizabeth springs goby currently occupies approximately 15 springs and is in a comparatively secure position (Alicia Whittington, Queensland Department of Environment and Resource Management, Longreach, personal communication). In contrast, populations of both red-finned blue-eye and Edgbaston goby are under threat from a range of factors including competition with gambusia (Fairfax et al. 2007). Individual populations of red-finned blue-eye and Edgbaston goby are limited to four and eight springs, respectively, and the remaining numbers of both species are estimated to comprise

293 fewer than 3000 individuals (Kerezsy 2009). Although Edgbaston is managed by the private not-for-profit conservation company Bush Heritage Australia, maintaining targeted fish restoration programs will be dependent on sourcing funding from state and national entities charged with protecting and enhancing Australia’s biodiversity. It is notable that both the first stage of a gambusia control project at Edgbaston (Kerezsy 2009) and a more wide-ranging study investigating the permanence of waterholes in the Lake Eyre Basin (Silcock 2009) have been funded by South Australian Arid Lands (SAAL) rather than by agencies or catchment management authorities based in Queensland. In the future, it is hoped that Queensland-based entities themselves may consider initiating and supporting similar recovery and research programs.

In addition to securing the populations of endangered species, the distribution and status of species such as golden goby in the Georgina and Diamantina drainages and Cooper Creek catfish in the greater Cooper catchment require attention, as both are benthic species with limited ranges in the Australian inland. Although this study sampled a greater number of both species than has previously been reported from the Lake Eyre Basin (Costelloe et al . 2004; Arthington et al . 2005; Balcombe and Arthington 2009), these species can be regarded as particularly susceptible to potential threats such as incursion from alien species and reduction in habitat quality and quantity. It is recommended that monitoring programs in the Lake Eyre Basin should pay particular attention to assessing the geographic range and security of both species, and that specific research aimed at gaining further data relating to their life histories be undertaken by research institutions and/or management agencies.

In conclusion, it should be re-iterated that far western Queensland offers an exceptional study area to investigate the structure and function of near-natural aquatic ecosystems in the Australian arid zone. In particular, the current study makes a significant contribution to the identified paucity of information relating to ephemeral river systems (Hughes 2005; Larned et al. 2009), and increases current knowledge of fish distribution, recruitment and movement patterns in an isolated area of outback Australia. If identified priority areas are addressed, such as conducting useful and valid public education programs, securing endangered species, establishing a sensible broad-scale monitoring project and targeting research at known problems (such as alien species) and known knowledge gaps (such as unknown sub-catchments and

294 reaches, and population ecology), there is every possibility that the Bulloo, Cooper, Diamantina and Georgina catchments will remain in reasonably good condition. However, it must also be acknowledged that the health of the rivers in outback Queensland is largely due to good luck rather than good management and is almost certainly a by-product of their isolation, lack of intensive human settlement and the continuance of their natural flow regimes. Management and monitoring of these systems is now required in order to maintain natural flow regimes, prevent any future degradation and preserve the rivers of the Queensland Lake Eyre and Bulloo- Bancannia basins as the most ecologically intact arid-zone aquatic ecosystems in Australia.

295 296 Appendices

Appendix 1. Techniques for sampling larval and juvenile fish in waterholes of the Queensland Lake Eyre Basin: A preliminary field study, June 2006.

Abstract

Small-meshed fyke nets (2mm mesh), unbaited bait traps (2mm mesh) and larval trawl nets (500 micron mesh) were used at eight sites in the Coolagh waterhole complex in the Barcoo River catchment to evaluate their effectiveness for sampling larval and juvenile fish. The techniques were employed at 4 sites in the main waterhole at Coolagh, and at 4 semi-permanent or ‘satellite’ waterholes. The satellite waterholes had become disconnected from the main waterhole 3 weeks before the sampling commenced following a flow event. Both small-meshed fyke nets and larval trawl nets sampled seven species of fish within the Coolagh complex. Unbaited bait- traps sampled no fish species. The larval trawl technique could not be standardised across all sites due to differences in the size, shape and depth of the satellite waterholes. Consequently, in the main waterhole, larval nets were trawled behind a boat, whereas in the satellite waterholes they were dragged by hand. Larval trawl nets sampled a similar number of species in both the main and satellite waterholes despite these different deployment methods. Both small-meshed fyke nets and larval trawl nets were found to be successful techniques for sampling larval and juvenile fish within the waterholes at Coolagh, and could have wider application for sampling small fish within the river systems of the Lake Eyre Basin.

297 1. Introduction

1.1 Selecting trial sampling techniques for larval and juvenile fish in the Lake Eyre Basin.

The waterways of the Lake Eyre Basin are generally turbid and present logistical difficulties associated with isolation and geography, such as an absence of access to mains power and steep muddy banks which can preclude the launching of trailer boats (personal observation). Thus, although the most commonly-employed method of fish sampling in eastern Australia and the Murray-Darling Basin has become electro- fishing using either boat or backpack mounted units (Faragher and Rodgers 1997), neither method is suitable for fish sampling work in the more remote rivers of the Lake Eyre Basin. Successful electro-fishing is contingent upon operators being able to see stunned fish in order to net them, and the waterways of the Lake Eyre Basin are generally too turbid to facilitate visual recognition of stunned fish. Although a modified sweep-net backpack electro-fishing system has been developed (King and Crook 2002), the waterways of the Lake Eyre Basin are also not suitable for backpack electro-fishing due to the increased danger presented to operators by unseen underwater obstacles, the lack of mains power from which to recharge batteries and the presence of a layer of unconsolidated silt making wading difficult or dangerous.

The most frequently employed passive netting techniques used for fish sampling are gill nets and fyke nets. Gill nets comprise a float line, a lead line and a ‘wall’ of mesh between the two. Although gill nets are a highly effective technique, and a commonly employed method by commercial fishers, they were excluded from the current study for ethical reasons. As their name suggests, gill nets frequently trap fish by enmeshing their operculum, or the hard outer gill covering. Removing trapped fish can sometimes result in injuries or death, neither of which were consistent with the aims of the current study. Fyke nets comprise either a single or double ‘wing’ from which fish are directed into a tunnel suspended by hoops. As fyke nets do not harm fish, they were selected for use in the current study. Fyke nets with a 2mm mesh were chosen as the current study aimed to trial fishing methods suitable for sampling larval and juvenile fish.

298 Passive trapping techniques used to sample small fish in Australian rivers include small rectangular bait traps (either baited or unbaited), and modified light traps (Humphries et al. 2002). Light traps were excluded from the current study due to the high expected turbidity of waterways of the Lake Eyre Basin. Bait traps with a 2mm mesh and 40mm opening were selected for the current study as they have been used previously in waterways of the Murray-Darling Basin and in eastern and southern coastal drainages to sample small fish (Faragher and Rodgers 1997).

Active netting techniques used to sample fish include a variety of trawled nets and manually-dragged seine nets. Like gill nets, seine nets comprise a lead line and a float line with mesh between the two. Seine-netting has been used in previous studies in the Cooper Creek area of the Lake Eyre Basin (Arthington et al. 2005). For the purposes of the current study, larval trawl nets with a 500 micron mesh and 58cm diameter opening were chosen as it was considered that these nets could possibly secure representative samples of larval and juvenile fish.

The Barcoo River was selected for the current study as it the easternmost catchment in the Lake Eyre Basin (and thus the closest geographically to Brisbane). The Coolagh waterhole complex was chosen as it is a typical waterhole complex within the Lake Eyre Basin, comprising a main (permanent) waterhole and several satellite (semi- permanent) waterholes. Permission to sample at Coolagh was sought and attained from station owners Joe and Anita Taylor. Queensland Boating and Fisheries Patrol officer Gary Muhling (Longreach) was informed of the location and duration of the sampling period.

299 1.2. Aims The aims of the current study can be summarised as follows: • To field-trial and evaluate the effectiveness of small-meshed fyke nets, bait traps and larval trawl nets for sampling larval and juvenile fish species in a waterhole complex within the Queensland Lake Eyre Basin. • To gather preliminary data relating to the recruitment of fish species in or close to the Coolagh waterhole complex on the Barcoo River in late autumn, 2006.

2. Methods

2.1 Study Area

The Coolagh waterhole complex is situated immediately downstream of the junction of the Barcoo and Alice Rivers in western Queensland (Figure 1). The main Coolagh waterhole was approximately 2 kilometres long, up to 120 metres wide and up to 4 metres deep during the current study. Four satellite waterholes which had recently (within the preceeding 3 weeks) become disconnected from the main waterhole were also included in the study. 250 metres north of the main waterhole, Satellite waterhole 1 was approximately 800 metres long, 10 metres wide and up to 1.5m deep. 100 metres west of the main waterhole, Satellite waterhole 2 was 50 metres long, 15 metres wide and up to 0.8 metres deep. Directly adjacent to the south-western end of the main waterhole, Satellite waterhole 3 was 25 metres long, 12 metres wide and up to 0.65 metres deep, and 800 metres north of the main waterhole, Satellite waterhole 4 was 1.2 kilometres long, 22 metres wide and up to 1.5 metres deep (Figure 1). The area had received considerable rain and run-off in the period before the study took place, resulting in over-bank flows and connectivity between all waterbodies in early April, late April and early May (Joe Simpson personal communication).

300

Figure 1 . Site map of the Coolagh waterhole complex, showing Sites 1 – 4 in the main waterhole and illustrating the geographical position of the satellite waterholes.

2.2 Sampling Design

The study was undertaken at four sites in the main Coolagh waterhole and in each of the satellite waterholes. Due to the comparatively recent rainfall and flow events, all waterbodies were at close to full capacity.

Sampling was undertaken during the last week of May, 2006. Site 1 was situated on the northern shore of the main waterhole at a point where it first begins to change from a narrow, sheltered habitat to a more open environment (Figure 2). Site 2 was situated at the western extremity of the main waterhole where the water was very still

301 and there was an abundance of snags (Figure 2). Sites 3 and 4 were spaced approximately 500 metres apart on the southern shore of the main Coolagh waterhole in more open habitat, however overhanging vegetation and/or in-stream woody debris were present at all sites (Figure 2).

Figure 2 . Examples of sites utilised in the main waterhole at Coolagh. Left, Site 1 looking south-east. Centre, Site 2 at the western terminus of the main waterhole, and right, Site 4 on the southern shoreline of the main waterhole.

Each satellite waterhole was treated as a separate site. Satellite waterholes 1 and 4 were comparatively long and narrow and ran parallel to the main waterhole, whereas satellite waterholes 2 and 3 were smaller and more oblong in shape (Figure 3). Woody debris and riparian vegetation were present at all sites, and were most prominent at Satellite waterhole 2 (Figure 3).

Figure 3 . (l – r) Satellite waterholes 1, 2, 3 and 4 in the current study.

302 2.3. Field Sampling Techniques

Water quality meters were used at each site in order to measure temperature, dissolved oxygen, pH and conductivity. In deep sites (Sites 1, 3 and 4 in the main waterhole) dissolved oxygen was measured both at the surface and at depth (3 metres). Turbidity was measured at each site using a Secchi disc.

Fish were surveyed using a combination of mini fyke nets, bait traps and trawled larval nets. All sampling was carried out under General Fisheries Permit No: PRM03315D issued by the Queensland Department of Primary Industries and Griffith University Ethics Permit.

Two double-winged fyke nets with a stretched mesh of 2mm were used to sample fish at all sites (Figure 4). Fyke nets were set with funnels facing in opposite directions from a central post, and the cod-ends of each fyke net were secured above the water surface. Fyke nets were set at 4pm and then retrieved at 9am the following morning.

Ten unbaited bait traps with a mesh size of 2mm and an opening of 45mm were set at each site at 4pm and retrieved at 9am the following morning (Figure 4). Bait traps were set within 30 metres of each pair of fyke nets and were set at a variety of depths depending upon the characteristics of each site.

Figure 4. Fyke nets utilised during the current study (left and centre), and a bait trap utilised during the current study (right).

Larval trawl nets were used at each site, however the physical characteristics of each site required in-field adaptation of this technique. The larval nets used at all sites were constructed of 500 micron mesh and had a hoop width of 580mm (Figure 5). At Sites 1, 3 and 4 in the main waterhole, 2 larval nets were trawled on either side of a boat

303 using a purpose built boom for 5 minutes during the day and also at night (Figure 5). At Site 2 in the main waterhole, the waterway was too narrow to deploy 2 larval nets, and consequently one net was towed for 10 minutes at each sampling period. In all satellite waterholes, a single larval net was towed by hand for the length of each waterhole during the day only, as it was not possible to use a boat in these comparatively small waterways (Figure 5). During all sampling events utilising the larval nets, a Swoffler flow meter was used in order to measure the amount of water filtered and to enable calculation of speed and distance.

Fish sampled by all methods were held in water-filled buckets following the emptying of nets. Fish were identified, counted, measured (fork length for fork-tailed species and total length for other species) and returned to the water.

Figure 5. A larval net utilised during the current study (left), a larval net deployed behind a boat (centre), and a larval net being towed by hand (right).

2.4 Data Analysis

In order to calculate the distance travelled during larval trawls, the difference in counts from the Swoffler flow meters and the published standard speed rotor constant (General Oceanics) were used in the following equation: Distance (m) = Difference in counts (x) Rotor constant 99999

In order to calculate the speed of larval trawls, the following equation was used:

304 Speed (cm/sec) = Distance in metres (x) 100 Time (sec)

In order to calculate the volume of water filtered during larval trawls, the following equation was used, and the result multiplied by 2 in instances where two larval trawl nets were deployed (Sites 1, 3 and 4):

Volume (m 3) = Net diameter 2 (x) distance (x) 3.14 4

In order to investigate differences in fish presence/absence between the main waterhole and the satellite waterholes, presence/absence data was combined from all sampling methods and analysed in Primer 5.2.9 (Plymouth Marine Laboratory).

Similarities between fish communities in the main waterhole and the satellite waterholes were calculated using the Bray-Curtis similarity measure (Bray & Curtis 1957). One-way ANOSIM (Analysis of Similarities) was performed in order to compare the fish community composition between the 4 sites in the main waterhole and the 4 sites in satellite waterholes, and a multi-dimensional scaling (MDS) ordination in two dimensions was performed in order to create a visual representation of the relationships between sites (Clarke 1993).

Results from ANOSIM calculate a test statistic ‘R’ identifying the observed differences between treatments compared with the differences among replicates within treatments (Clarke & Warwick 1994). Species contributing to the differences between areas were examined using the SIMPER analysis, which calculates the average dissimilarity between paired samples and then allocates the contribution each species makes to this dissimilarity (Clarke & Warwick 1994).

305 3. Results

3.1 Water Quality All water quality parameters exhibited a high degree of similarity between all sites (Table 1). Temperature remained slightly lower in the shallower satellite waterholes where daily air temperature fluctuations presumably exerted a stronger influence on the temperature of the waterbodies (Table 1). The water at all sites was characteristically turbid, and Secchi disc visibility was never greater than 5cm (Table 1).

Table 1. Water quality at all sites in the Coolagh waterhole complex in May 2006. Depth profiles (at 3 metres) are given for dissolved oxygen in brackets.

Parameter Unit of Main waterhole Satellite waterholes measurement 1 2 3 4 1 2 3 4

Temperature oC 18.2 16.5 17.9 17.4 15.3 16.4 14.9 15.5 Dissolved % saturation 67.1 70.8 67.1 68 73.3 71 69.5 75 oxygen (36.2) (42) (51.4) Conductivity mS/cm -1 0.165 0.173 0.167 0.170 0.176 0.169 0.174 0.167 pH 7 6.45 6.31 5.84 6.37 5.72 6.12 6.87

Turbidity Secchi disc 3.5 4 4 5 5 4.5 4.5 5 (cm)

3.2 Fish species

A total of 223 individual fish from 9 species were sampled in the Coolagh waterhole complex. Desert rainbowfish were the most abundant species, accounting for 35.9% of all fish caught (Table 2a). Carp gudgeons (71 individuals) and Lake Eyre golden perch (30) were also sampled in comparatively large numbers (Table 2a). The remainder of the species were sampled in smaller numbers and included spangled perch (14), bony bream (12), Hyrtl’s tandan (6), Australian smelt (6), silver tandan (2) and northwest ambassis (2)(Table 2a).

306 In addition to the species sampled by the methods used in this study, three further species were caught by an aquaculture farmer fishing in the vicinity of Site 1 using 50mm gill nets and hook and line on May 23 (Table 2a). These species were the Cooper Creek catfish (2), Welch’s grunter (2) and Barcoo grunter (21)(Table 2a).

The total number of native species caught by all methods is consistent with sampling events conducted in waterholes of Cooper Creek (Arthington et. al. 2005, Balcombe pers comm.), however no alien species were sampled in the Coolagh waterhole complex.

307 Table 2a. Total abundance of fish caught in four areas of the main Coolagh waterhole and four satellite waterholes using mini-fyke nets and larval trawl nets in May 2006 Scientific Name Common Name Main Coolagh waterhole Satellite waterholes Total species Proportional Site 1 Site 2 Site 3 Site 4 1 2 3 4 abundance abundance (individuals) (%) Nematolosa erebi Bony bream 2 - - - 5 3 1 1 12 5.4% Neosiluris hyrtlii Hyrtl’s tandan 1 2 3 - - - - - 6 2.7% Porochilus argenteus Silver tandan 2 ------2 0.9% Retropinna semoni Australian smelt 2 1 1 - - 2 - - 6 2.7% Melanotaenia splendida tatei Desert rainbowfish 41 16 6 5 1 2 9 - 80 35.9% Ambassis spp. Northwest ambassis ------1 1 2 0.9% Macquaria spp. Lake Eyre golden perch 7 5 1 2 4 9 - 2 30 13.5% Leiopotherapon unicolor Spangled perch - - - - 3 6 - 5 14 6.3% Hypseleotris spp . Carp gudgeons 36 23 2 2 2 - 2 4 71 31.8% Total individuals 91 47 13 9 15 22 13 13 Total individuals: % contribution to total 40.8% 21% 5.8% 4% 6.7% 9.9% 5.8% 5.8% 223

Table 2b. Additional specimens caught in the vicinity of Site 1 in the main Coolagh waterhole using 50mm mesh gill nets and/or line fishing by Bruce Sambell on June 23. Scientific Name Common Name Number Nematolosa erebi Bony bream Approx. 100 Neosiluroides cooperensis Cooper Creek catfish 2 Macquaria spp. Lake Eyre golden perch 5 Bidyanus welchi Welch’s grunter 2 Scortum barcoo Barcoo grunter 21

308 3.3 Species selectivity of sampling methods

No fish were caught during the current study using bait traps. Seven species were sampled using larval trawl nets and seven species were also sampled using fyke nets (Table 3). Australian smelt and northwest ambassis were sampled exclusively in larval trawl nets, and Hyrtl’s tandan and spangled perch were sampled exclusively in fyke nets (Table 3). All other species were sampled using both methods (Table 3). Although larval trawl nets sampled seven species during the day, only desert rainbowfish, carp gudgeons, Lake Eyre golden perch and Australian smelt were caught during the night samples.

Table 3. Species sampled using fyke nets and larval trawl nets in the Coolagh waterhole complex in May 2006 Species Fyke Nets Larval trawl nets Day Night Bony bream ● ● Hyrtl’s tandan ● Silver tandan ● ● Australian smelt ● ● Desert rainbowfish ● ● ● Northwest ambassis ● Lake Eyre golden perch ● ● ● Spangled perch ● Carp gudgeon ● ● ●

3.4 Size selectivity of sampling methods

The smallest fish sampled during the current study was a carp gudgeon (10mm) and the largest was a Hyrtl’s tandan (270mm)(Table 4). Fish smaller than 15mm TL were only caught in larval trawl nets, and fish longer than 200mm TL were only caught in fyke nets (Table 5). Both fyke nets and larval trawl nets sampled fish between 15 and 60mm (TL). Larval trawl nets were generally ineffective at sampling fish longer than

309 60mm, however trawls at Site 1 sampled a comparatively large bony bream (142mm) and a silver tandan (132mm).

Table 4. Size ranges of sampled fish species from the Coolagh waterhole complex in May 2006. Species Size range- total length (mm) Bony bream 49 – 142 (n=12) Hyrtl’s tandan 194 – 270 (n=6) Silver tandan 132 – 207 (n=2) Australian smelt 15 – 50 (n=6) Desert rainbowfish 19 – 55 (n=80) Northwest ambassis 34 – 35 (n=2) Lake Eyre golden perch 11 – 60 (n=30) Spangled perch 35 – 95 (n=14) Carp gudgeon 10 – 39 (n=71)

Table 5. Size selectivity of fyke nets and larval trawl nets during sampling in the Coolagh waterhole complex, May 2006.

Total length (mm) of sampled fish Fyke nets Larval trawl nets 10 – 15 ● 15 – 20 ● ● 20 – 25 ● ● 25 – 30 ● 30 – 40 ● ● 40 - 50 ● ● 50 - 60 ● ● 60 – 100 ● 100 – 200 ● ● >200mm ●

310 3.5 Variation within larval trawls

Larval trawls varied according to distance travelled, speed, volume of water filtered, number of nets and day or night sampling events (Table 6). The longest larval trawl was 163.9 metres at Site 3 (Day), and the shortest was 46.6 metres at Satellite 3 (Table 6). The fastest larval trawl was 54.63 cm/sec at Site 3 (Day) and the slowest was 7.77 cm/sec at Satellite 3 (Table 6). The largest volume of water filtered during a larval trawl was 86.56 m 3 at Site 3 (Day) and the smallest was 12.3 m 3 at Satellite 3. Despite these differences, a similar number of species were sampled at both Site 3 (Day) and Satellite 3 (4 and 3 respectively). The mean number of species sampled in Satellite waterholes (3) was similar to the mean number sampled in the main waterhole (3.125), despite the longer distances, higher speeds and larger volumes sampled in the main waterhole (Table 6). The mean number of species sampled during the day in the main waterhole (3.5) was higher than the mean number of species sampled during night trawls (2.75).

Table 6. Larval trawls conducted in the Coolagh waterhole complex in May 2006. Site Day/ Time No. of Distance Speed Volume No. of Night (seconds) nets (metres) cm/sec (m 3) species 1 Day 300 2 68.82 22.94 36.35 6 1 Night 300 2 120.07 40.02 63.41 3 2 Day 600 1 142.88 23.78 37.33 3 2 Night 600 1 149.12 24.85 39.38 4 3 Day 300 2 163.9 54.63 86.56 4 3 Night 300 2 77.02 25.67 40.68 1 4 Day 300 2 94.97 31.66 50.16 1 4 Night 300 2 94.97 25.72 40.75 3 S1 Day 300 1 61.2 20.37 16.14 2 S2 Day 480 1 51.52 10.74 13.6 3 S3 Day 600 1 46.6 7.77 12.3 3 S4 Day 900 1 85.67 9.52 22.62 4

311 3.6 Fish presence/absence in the main waterhole and satellite waterholes

Seven species were sampled in both the main waterhole (Sites 1 – 4) and the combined satellite waterholes (Sites 1s – 4s), however the two areas were distinct (Global R = 0.708). Bony bream, Australian smelt, desert rainbowfish, Lake Eyre golden perch and carp gudgeons were present in both the main waterhole and satellite waterholes (Table 2). Hyrtl’s tandan and silver tandan were sampled in the main waterhole whereas northwest ambassis and spangled perch were sampled only in the satellite waterholes (Table 2).

In the MDS plot, Site 4 in the main waterhole is separated from Sites 1, 2 and 3 due to the absence of any catfish species from the samples (Figure 5). Sites 1s, 2s, 3s and 4s are separated from the main waterhole sites due to the presence of spangled perch and northwest ambassis and the absence of catfish species (Figure 5, Table 7). Sites 1s, 2s, 3s and 4s are separated from one another due to smaller differences in species presence/absence between these sites (Figure 5, Table 2). In the satellite waterholes, bony bream were the only species present at all four sites (Table 2). Desert rainbowfish, Lake Eyre golden perch, spangled perch and carp gudgeons were present at three of the four satellite waterholes, and northwest ambassis were present in two of the four satellite waterholes (Table 2). Australian smelt were only sampled in the satellite waterholes in Satellite 2 (Table 2).

312 Satellite 2

Satellite 4

Site 1

Satellite 1

Sites 2 & 3

Satellite 3

Site 4 Stress = 0.1

Figure 5 . MDS ordination of fish species presence/absence at sites in main waterhole (Sites 1 – 4) and satellite waterholes (Satellite 1, 2, 3 and 4) in the Coolagh waterhole complex, May 2006.

Table 7. SIMPER analysis comparing species presence/absence between the main

waterhole at Coolagh and the satellite waterholes in May 2006.

Average abundance per sample Percent contribution to observed differences Species Main waterhole Satellite waterholes (>5%) Bony bream 0.5 2.5 18.58 Hyrtl’s tandan 1.5 - 16.13 Silver tandan 0.5 - - Australian smelt 1 0.5 13.93 Desert rainbowfish 17 3 5.66 Northwest ambassis - 0.5 11.99 Lake Eyre golden perch 3.75 3.75 6.32 Spangled perch - 3.5 16.99 Carp gudgeons 15.75 2 5.66

313 3.7 Larval and juvenile fish sampled at Coolagh in May 2006.

Juvenile spangled perch (<50mm TL) and bony bream (<90mm TL) were sampled in the Coolagh waterhole complex in May 2006 in small numbers (9 individuals of both species). Although carp gudgeons were present in both adult and juvenile size classes, the majority (50 out of a total of 71) were between 10 and 15mm in total length (Figure 6).

60 n = 71 50

40

30

20 Number of individuals of Number

10

10 – 15 15 – 20 20 – 25 25 – 30 30 – 35 35 – 40 Total length (mm)

Figure 6 . Length frequency of carp gudgeons from sites within the Coolagh waterhole complex in May 2006.

Desert rainbowfish were the most frequently sampled species in the Coolagh system, and were present in a range of size classes from juvenile to adult (Figure 7). Juvenile Lake Eyre golden perch were also present in a range of size classes in the Coolagh system, however the majority of Lake Eyre golden perch were between 10 and 15mm in total length (Figure 8).

314 30 n = 80 25

20

15

10 Number of individuals of Number 5

10 – 15 15 – 20 20 – 25 25 – 30 30 – 35 35 – 40 40 – 45 45 – 50 50 - 55 Total length (mm)

Figure 7 . Length frequency of desert rainbowfish from sites within the Coolagh waterhole complex in May 2006.

18

16 n = 30

14

12

10

8

6

4 Number of individuals of Number 2

10 – 15 15 – 20 20 – 25 25 – 30 30 – 35 35 – 40 40 – 45 45 – 50 50 – 55 55 - 60 Total length (mm)

Figure 8 . Length frequency of Lake Eyre golden perch from sites within the Coolagh waterhole complex in May 2006.

One 15mm Australian smelt metalarva was present in the sample from Satellite waterhole 2. Adult Hyrtl’s tandan, silver tandan and northwest ambassis were sampled using the methods described in this study, and adult Barcoo grunter, Welch’s grunter and Cooper Creek catfish were caught by an aquaculture farmer in the Coolagh

315 waterhole on May 23, 2006. Larvae and juveniles of these species were not sampled in the current study.

4. Discussion

4.1 Water quality

Water quality parameters such as dissolved oxygen, conductivity and pH measured in the Coolagh waterhole complex were similar across all sites and consistent with the period of flow connectivity that had existed 3 weeks prior to the sampling period. Temperature was higher in the main waterhole than the satellite waterholes. It is likely that the influence of low air temperatures on the small and shallow (up to 1.2m deep) satellite waterholes was responsible for lower temperatures in all satellite waterhole sites than main waterhole sites. Turbidity was uniformly high at all sites, with Secchi disc visibility never greater than 5cm.

4.2 Fish species

Fish species sampled within the Coolagh waterhole complex comprised all native species known to occur in the north-eastern Lake Eyre Basin (Wager and Unmack 2000), however Barcoo grunter, Welch’s grunter and Cooper Creek catfish were only caught by an aquaculture farmer working in the same area using gill nets and lines (pers. obs.). No alien species were sampled, which contrasts with samples taken in morphologically similar waterhole systems in Cooper Creek, where both gambusia and goldfish were sampled (Arthington et al. 2005, Balcombe pers comm.).

Given the suite of native fish species which occur at Coolagh, the area has positive attributes as a future site from which to conduct research into the aquatic ecology of the Lake Eyre Basin, particularly if this research requires fish samples to be taken.

316 4.3 Species selectivity of sampling methods

Unbaited bait traps caught no fish at any site during the current study. Although bait traps are frequently used as a sampling method for small and juvenile fish species in the Murray-Darling Basin and in coastal drainages, they are an ineffective method for sampling fish in the Lake Eyre Basin. The reason for the ineffectiveness of bait traps as a sampling method is almost certainly related to the turbidity of the water in areas such as the Coolagh waterhole complex, where fish are less likely to utilise bait traps for cover.

Although commonly sampled species such as desert rainbowfish, carp gudgeons and Lake Eyre golden perch were caught using both small-meshed fyke nets and larval trawl nets, both methods were ineffective for sampling some species. Fyke nets did not sample northwest ambassis or Australian smelt, and larval trawl nets did not sample Hyrtl’s tandan or spangled perch. As these four species were sampled in small numbers (6, 2, 6 and 14), it is possible that their absence in the samples may have been due to low population density at the sites chosen rather than failure of certain gear types. Nevertheless, both methods should be retained for use in fish sampling effort aimed at larval and juvenile fishes in the Lake Eyre Basin based on the results of this study and the possibility that northwest ambassis, Australian smelt, Hyrtl’s tandan and spangled perch may be difficult to sample without a combination of sampling techniques.

Neither larval trawl nets or small-meshed fyke nets sampled Cooper Creek catfish, Barcoo grunter and Welch’s grunter in the Coolagh waterhole complex, however adult fish of all 3 species were sampled by an aquaculture farmer using gill nets and lines. Determining whether the absence of these species is due to absence of larval and juvenile specimens from the sampled sites or aversion to the sampling gear may be resolved by continued sampling of the sites utilised at Coolagh at different seasonal and climatic periods.

Larval trawl nets sampled seven species during the day but only four species at night. This suggests that larval trawl sampling is more effective during the day than at night

317 in waterhole complexes such as Coolagh and that night sampling is unnecessary as an additional sampling event.

4.4 Size selectivity of sampling methods

Both larval trawl nets and small-meshed fyke nets regularly sampled fish from 15 – 60mm TL during the current study, suggesting that both methods are well-suited to sampling juvenile fish and adult small fish such as desert rainbowfish. Fish less than 15mm TL were only sampled using larval trawl nets, indicating that this method is likely to be an effective way of sampling larval fish in waterholes such as the Coolagh complex. Small-meshed fyke nets also sampled adult fish such as Hyrtl’s tandan and silver tandan. Larval trawl nets occasionally sampled larger fish such as bony bream and silver tandan (up to 142 mm TL) in small numbers.

By combining the two methods described with a third method aimed at sampling adult fish (such as larger fyke nets), representative samples of present fish species from larvae to adult size classes could be taken from waterholes such as those in the Coolagh complex.

4.5 Using the larval trawl method in different waterholes

Standardising larval trawl procedures is difficult in waterholes such as the Coolagh complex due to the different physical size of each satellite waterhole. At Sites 1,3 and 4 in the main waterhole, a standard approach was instigated by suspending a larval net on either side of a boat-mounted boom and trawling for 5 minutes, however variation still existed with regard to distance travelled, trawl speed and volume of water filtered. Additionally, at Site 2, at the western extremity of the Coolagh waterhole, the narrowness of the waterway prevented trawling two nets, so only one was trawled for double the time (10 minutes). In the satellite waterholes, boat-mounted larval trawls could not be undertaken due to the shallowness of the water. In satellite waterholes, a single larval trawl net was dragged by hand from one end of the waterhole to the other and the time recorded. Accordingly, there is substantial variation in speed, distance and volume of the larval nets utilised in the current study.

318 Despite these variations, there is comparatively little difference in the number of species sampled by larval trawl nets either dragged by hand through satellite waterholes, or used in pairs or singly in the main waterhole. In the current study, the effectiveness of larval trawl nets appears to be more related to their presence in a waterhole than to the speed at which they are trawled, the distance over which they are trawled or the volume of water through which they are trawled.

4.6 Geographical distribution of fish species within the Coolagh waterhole complex.

The current study suggests that variation of fish communities may exist between main waterholes and satellite waterholes in the Coolagh waterhole complex following periods of flow and connectivity. Catfish, such as Hyrtl’s tandan, silver tandan and Cooper Creek catfish were not recorded from any satellite waterholes, and spangled perch and northwest ambassis were not recorded from the main waterhole. The sample sizes of these species were comparatively small (6, 2, 2, 14 and 2, respectively), indicating that future sampling effort is required in order to investigate these perceived differences more accurately. Studies in morphologically similar waterhole complexes in the Cooper Creek system did not evince such differences in fish community composition (Balcombe pers comm.). In extended dry periods, it is likely that satellite waterholes such as Satellite 2 and Satellite 3 would become dry, and without an intervening period of renewed flow this would result in localised extinction events for fish inhabiting these waterholes. Results from the current study indicate that sampling effort should be directed towards as many satellite waterholes as possible when sampling fish in waterhole complexes of the Lake Eyre Basin, as it is possible that differences in fish communities may exist following flow or flood events. Additionally, in main or large waterholes, it is unlikely that an increased number of replicate sites will result in additions to the species sampled during a given sample period.

319 4.7 Evidence of recent fish recruitment in the Coolagh vicinity

4.7.1 Australian smelt

A single Australian smelt metalarva (Serafini and Humphries 2003) was sampled in Satellite waterhole 2. The majority of information relating to the spawning and recruitment of Australian smelt comes from research conducted in the Murray Darling Basin and along the east coast (Pusey et al . 2004). It is suggested that Australian smelt in these areas spawn from winter onwards and that recruitment is enhanced by low flow conditions (Pusey et al. 2004). No studies of Australian smelt spawning and recruitment exist from the Lake Eyre Basin, however the presence of a metalarva in the sample from Satellite waterhole 2 at Coolagh suggests that this species may spawn in autumn in the Barcoo River catchment, and that spawning may be associated with flow events.

4.7.2 Carp gudgeons

Fifty carp gudgeons between 10 and 15mm TL were recorded from the Coolagh waterhole complex. This size range is consistent with the metamorphosis from larvae to juvenile fish of this species complex from Murray Darling Basin specimens examined by Serafini and Humphries (2003). Existing information relating to the spawning and recruitment of Hypseleotris spp. gudgeons is confined to populations in the Murray Darling Basin and from east coast drainages as the life history of these fishes is unstudied in the Lake Eyre Basin (Pusey et al. 2004). General information relating to this species complex is further hindered by taxonomic confusion resulting from an uncertain number of potential species, hybrid species and sub-species which are yet to be described throughout their geographic range (Kerezsy 2005). Although the reproductive biology of carp gudgeons (all species) is poorly understood (Pusey et al. 2004), various species have been spawned in aquaria when water temperatures were between 20° and 30°C (Unmack 2000). The current study indicates that recruitment within wild populations in the Lake Eyre Basin may occur at temperatures at the lower end of this range, or perhaps even lower, as the highest water temperature recorded during the current study was 18.2 degrees in the main waterhole. However, the presence of juvenile, immature and adult carp gudgeon in the Coolagh samples

320 also suggests that this species complex may spawn over an extended period or may spawn more than once during a breeding season. The current study indicates that carp gudgeon recruitment in the Lake Eyre Basin may not be contingent upon periods of low or no flows, a phenomenon which is thought to exist in the Murray Darling Basin (King et al. 2003), as high flows had occurred in the Coolagh waterhole complex only 3 weeks prior to the sampling period.

4.7.3 Lake Eyre golden perch

Juvenile Lake Eyre golden perch were sampled in both the main waterhole and satellite waterholes at Coolagh. 22 individual Lake Eyre golden perch were between 10 and 20mm TL, 3 were between 20 and 30mm TL and 5 were between 40 and 60mm TL. This suggests that up to three cohorts of juvenile Lake Eyre golden perch were present in the sites sampled at Coolagh, and that the youngest cohort may have been spawned only weeks prior to the current study. Lake Eyre golden perch were suggested to be a genetically separate species from Murray Darling golden perch in 1992 (Musyl and Keenan 1992), and a current study is continuing to describe the genetic differences between the two populations (Faulks pers comm.). The vast majority of literature pertaining to golden perch recruitment and life history has been undertaken in the Murray Darling Basin (Reynolds 1983; Mallen-Cooper at al. 1995), and the first study relating to otolith-based age determination of this species in the Lake Eyre Basin has only recently been completed (Pritchard 2004).

Golden perch in the Diamantina River, Neales River and Cooper Creek appear to spawn in both low and high flow years, however recruitment is most successful in flood years or in years of higher flow (Pritchard 2004). The results of the current study suggest that golden perch in the Barcoo River near (or at) Coolagh may have recently (ie: within the last 6 months) spawned twice or three times prior to the sampling period, and that there may be a relationship between the periods of heightened flow in April and May and the presence of juvenile golden perch. The presence of such young golden perch at Coolagh also suggests that the thermal limit for golden perch spawning in the Lake Eyre Basin may be comparatively low, given that the highest water temperature recorded during the current study was 18.2 degrees in the main waterhole. This supports the findings of Gilligan and Schiller (2003) in

321 the Murray Darling Basin, and also suggests that flow rather than temperature may be the primary trigger for golden perch spawning in the Lake Eyre Basin. A current study of golden perch recruitment being undertaken in the western Darling River in the Murray Darling Basin also indicates that golden perch spawn in response to flow events (Sharpe pers comm.). Further sampling of larval and juvenile fishes within the Coolagh waterhole complex therefore has the potential to yield information relating to seasonal and flow-related recruitment events of golden perch in the Barcoo River catchment which may increase the body of knowledge relating to this species both in the Barcoo and other rivers within the Lake Eyre Basin. Additionally, there may be recruitment similarities between Lake Eyre golden perch and Murray Darling golden perch, and knowledge of these similarities has the potential to contribute positively to the management of both populations.

4.7.4 Bony bream

The presence of immature bony bream (<150mm TL) and juvenile bony bream (<50mm TL) in the Coolagh waterhole complex during the current study suggests that this species may have an extended spawning season in the Lake Eyre Basin. A study of the recruitment behaviour of bony bream by Puckridge and Walker (1990) demonstrated that bony bream spawn in open-water environments within the southern Murray-Darling Basin. It is likely that this species also spawns in riverine waterholes or intermittently flowing rivers and creeks within the Lake Eyre Basin based on the results of the current study.

4.7.5 Spangled perch

Fourteen spangled perch were sampled in the satellite waterholes of the Coolagh complex during the current study. Nine were <50mm TL suggesting that this cohort of juvenile spangled perch may have been spawned comparatively recently. Tank rearing trials (Llewellyn 1973) of this species indicate that spangled perch from Tottenham in the Murray Darling Basin mature at 58mm (males) and 78mm (females), however the recruitment behaviour of spangled perch in the Lake Eyre Basin currently remains unstudied.

322 5. Conclusions

The Coolagh waterhole complex in the Barcoo River catchment is a typical conglomeration of main and satellite waterholes within the Lake Eyre Basin. All native fish species (12) which were predicted to be possibly present in the Coolagh complex were sampled using small-meshed fyke nets, larval trawl nets or by gill nets and line fishing by an aquaculture farmer who was fishing in the main waterhole at the same time. No alien fish species were caught by any method. Water quality parameters within the Coolagh complex were consistent with the time of year, the recent period of flow and connectivity 3 weeks earlier and similar sites in the Lake Eyre Basin sampled by other researchers.

Both larval trawl nets and small-meshed fyke nets were found to sample larval, juvenile and adult fish from a total of 9 species. Bait traps caught no fish and were found to be an ineffective sampling method in the Coolagh waterhole complex. Larval trawl nets were deployed either behind a boat in the main waterhole or manually dragged through the satellite waterholes. However, despite large differences in trawl speed, trawl distance and trawled volume of water, larval trawl nets sampled a similar number of fish species in both the main and satellite waterholes. A similar adaptive approach to deployment of trawl nets is recommended for any future fish sampling effort utilising this technique in waterways of the Lake Eyre Basin.

There was a difference in the fish communities of the main waterhole and the satellite waterholes at Coolagh due to differences in presence/absence of sampled fish species. However, the species contributing to this difference were sampled in small numbers and increased sampling effort and occurrence would be required in order to more fully investigate this pattern.

Australian smelt, carp gudgeons and Lake Eyre golden perch were sampled as either larvae or at larval/juvenile metamorphosis, indicating that these species had been spawned recently either in the Coolagh complex or in the surrounding Barcoo catchment when connectivity existed between the main and satellite waterholes. Juvenile spangled perch and bony bream were also sampled, suggesting these species

323 had also spawned comparatively recently either within the Coolagh complex or in the surrounding Barcoo catchment.

Future sampling of larval and juvenile fish in the Lake Eyre Basin should utilise small-meshed fyke nets and adaptive larval trawls in order to gain representative samples of fish communities in main and satellite waterholes.

324 Appendix 2. Water quality data September 2006 – November 2008, Lake Eyre and Bulloo-Bancannia Basins

Site name Catchment Sampling time Surfac Depth Dissolved Dissolved Electrical pH Turbi e temp. (2m) oxygen (% oxygen conductivity dity (°C) temp. saturation) (2m) (µS/cm) (cm) (°C) (% saturation) One Mile Kyabra September 2006 21.4 Not 43.2 Not 112.1 8.3 6 Springfield Kyabra September 2006 21.6 recorded 54 recorded in 150.2 8.2 3 Springfield South Kyabra September 2006 26.7 in Sept 32.1 September 143 7.5 2 Durham Downs Cooper September 2006 22 2006 74.7 2006 100.2 7.21 5 Durham Downs 2 Cooper September 2006 17.5 80 97.7 7.2 4 Durham Downs East Cooper September 2006 21.3 63.4 101.7 7.3 3 Shed Cooper September 2006 24.5 80.2 80.9 8.95 5 Currareva Cooper September 2006 22.5 51 149.8 7.33 3 Murken Cooper September 2006 23.9 39.8 129.4 8.6 2 Waterloo Thomson September 2006 21.4 74.3 193 8.43 8 Thomson Main Thomson September 2006 27.8 17.4 255 8.87 40 Coolagh Barcoo September 2006 25.8 32.9 193.7 8.33 10 Coolagh 2 Barcoo September 2006 21 41.6 212.3 7.92 3 One Mile South Kyabra December 2007 25.5 NA 31.5 NA 384 7.68 6 Halfway Kyabra December 2007 25.7 NA 46 NA 254 8.57 5 Outside Kyabra December 2007 28 26 49 NA 178.7 7.8 6 One Mile Kyabra December 2007 30 26 46 37.5 352 7.7 4 Springfield South Kyabra December 2007 32.7 27.1 71 NA 346 7.93 2 Springfield Kyabra December 2007 34.9 27.5 69 NA 129.6 7.76 6 Shed Cooper December 2007 30.9 27.6 82 65 129.3 7.95 5 Currareva Cooper December 2007 28 25.2 50 NA 274 7.33 3 Murken Cooper December 2007 34.3 27.2 85 52 184.5 8.16 5 Thomson Main Thomson December 2007 31.8 30.2 Equipment malfunction. 465 8.84 35 Waterloo Thomson December 2007 28 25.7 No further dissolved 265 8.3 7 Vergemont 1 Thomson December 2007 29.4 NA oxygen records for 197.2 7.36 2 Vergemont 2 Thomson December 2007 24.2 26.2 December 2006 112.9 7.8 2 Native Thomson December 2007 25.5 23 143.8 5.4 5 Isisford Barcoo December 2007 32.2 27 307 8.62 15 8 Mile Barcoo December 2007 29.1 25 254 8.31 35 Coolagh Barcoo December 2007 32.9 27 225.9 8.38 15 No water quality readings for January 2007

Pump Bulloo April 2007 22 21.8 52 44.7 98.5 7 20 Bulloo Shed Bulloo April 2007 24.7 20.5 60.3 15.4 30.7 7 5 House Bulloo April 2007 25 22 42 35 29 7 6 Springfield Kyabra April 2007 23 22 90 62 58 6.5 8 Springfield South Kyabra April 2007 25 23 99 56 99.7 6.7 6 Halfway Kyabra April 2007 28 NA 91 NA 130 6.6 4 One Mile Kyabra April 2007 26 24 91 70 75.8 6.7 7 Shed Cooper April 2007 26.2 23.6 93.6 72.8 179 6.7 7 Currareva Cooper April 2007 24.6 24.5 80.2 73.6 130 6.8 6 Murken Cooper April 2007 26.9 26.7 98.1 83.3 125.3 7 6 Thomson Main Thomson April 2007 26.7 23.1 92.2 72 160.5 6.98 8 Waterloo Thomson April 2007 24.3 24.3 89.9 83 107.6 7.2 7 Vergemont Thomson April 2007 22.3 20.9 54.9 43.1 90.8 6.96 3 Native Thomson April 2007 23 22.4 31 56 63 6.4 7 Isisford Barcoo April 2007 23.1 21.2 50 28.5 158.2 6.62 3 Coolagh Barcoo April 2007 25.2 22.4 43.6 13.2 144 6.78 5 Coolagh 2 Barcoo April 2007 23.9 21.9 50 50 174 6.96 6 Coolagh 3 Barcoo April 2007 25.5 NA 108.5 NA 328 8.36 2 Pulchera North Mulligan April 2007 28 NA 105 NA 775 6.96 6 Pulchera South Mulligan April 2007 24.9 24.9 98.2 78.3 714 7.2 6 Dune Pond Mulligan April 2007 32 NA 124.6 NA 1158 8.11 NA Kunnamuka Swamp Mulligan April 2007 24.3 NA 81.7 NA 56.5 6.9 10 Ocean Bore Mulligan April 2007 28.5 NA 109.2 NA 2850 7 80 S Bend Mulligan April 2007 24 22,7 38 14.4 120.8 7 6

325 Idamea Georgina April 2007 26.8 27.6 162.2 102.7 142.7 8.4 50 Lower Georgina April 2007 27.6 25 92.6 110.2 141.1 8.26 35 Georgina Main Georgina April 2007 27.6 27.5 107.1 101.7 350 7.52 17 Billyer Diamantina April 2007 22.6 21.4 53.5 60.1 215 7 3 2 Mile Diamantina April 2007 24.3 22.3 42.6 79 114.2 7 2 Diamantina 1 Diamantina April 2007 23.2 NA 91.2 NA 203.1 7.21 2 Diamantina 2 Diamantina April 2007 22.5 NA 76.3 NA 152.7 7.14 2 Pump Bulloo August 2007 12 14 63 51 No data No data 5 Bulloo Shed Bulloo August 2007 16.5 NA 58 NA No data No data 3 House Bulloo August 2007 15 NA 55 35 No data No data 5 Coolagh Barcoo August 2007 16 18 57 NA No data No data 5 Coolagh 2 Barcoo August 2007 18.6 17 112 63 No data No data 30 Isisford Barcoo August 2007 17 NA 70 NA No data No data 3 Little Oma Barcoo August 2007 18 NA 72 NA 203 8.6 10 Oma Barcoo August 2007 16 NA 72.4 NA 160 8.4 5 Springfield Kyabra August 2007 18 NA 57 NA 80 7 4 Springfield South Kyabra August 2007 16 NA 63 NA 140 7 4 Halfway Kyabra August 2007 10 NA 58 NA 190 8 3 One Mile Kyabra August 2007 15 NA 50 NA 99 7.5 5 Shed Cooper August 2007 18 NA 62 NA 163 8.5 7 Currareva Cooper August 2007 17 NA 64 NA 90.9 7.6 3 Murken Cooper August 2007 16 NA 55 NA 95 8 5 Thomson Channel 1 Thomson August 2007 17 NA 56 NA 93.9 7 5 Thomson Small Thomson August 2007 20 NA 63 NA 131 7.5 5 Thomson Tiny Thomson August 2007 21 NA 63 NA 116.5 8.3 5 Waterloo Thomson August 2007 18.5 16.5 68 50 90 7.3 3 Vergemont Thomson August 2007 16.5 17 62 50 45.7 7.8 5 Native Thomson August 2007 18 17.4 67 58 96 8 6 Hunter’s Diamantina August 2007 18 NA 75 NA 57 7.5 3 Lake Constance Diamantina August 2007 23 NA NA NA 277 8.5 5 Warracoota Diamantina August 2007 14.5 NA 66 NA 170.5 8.4 4 Billyer Diamantina August 2007 15.7 NA 60.5 NA 405 7 3 2 Mile Diamantina August 2007 13 NA 60 NA 358 7 2 Parapituri Georgina August 2007 15 NA 72.3 NA 1020 8.7 45 Idamea Georgina August 2007 20.3 17.5 85 78 158 8.7 12 Lower Georgina August 2007 22.7 NA 110 NA 161 7.8 17 Georgina Main Georgina August 2007 17 16.7 81 66 185 7 30 Pulchera Mulligan August 2007 23 NA 49 NA 1150 7 10 S Bend Mulligan August 2007 19 NA 62 NA 146 6.5 5 Pump Bulloo November 2007 23.7 23 26 19 40.1 7 3 Bulloo Shed Bulloo November 2007 23.4 23.2 30.9 31.6 32.4 7 3 House Bulloo November 2007 24 23.8 26.9 20.8 29 7 4 Coolagh Barcoo November 2007 28 26 60 41 176 7.5 3 Coolagh 2 Barcoo November 2007 25.2 NA 50 NA 200 7.3 4 Isisford Barcoo November 2007 27 25.4 50 51 158 7 4 Springfield Kyabra November 2007 26 24 62.1 56.6 59.7 7 5 Springfield South Kyabra November 2007 25.6 NA 70 NA 331 7 2 One Mile Kyabra November 2007 24.7 24.5 48 44.7 68.7 7 7 Shed Cooper November 2007 27.5 27 100.9 58.8 260 7.6 10 Currareva Cooper November 2007 25.3 24.8 68 74.4 113.6 7 4 Murken Cooper November 2007 27.9 27.7 65.4 69.2 119.5 7.4 3 Thomson backflow Thomson November 2007 25 NA 48.4 NA 131 7 3 Waterloo Thomson November 2007 28 26.5 68 55 129 9 3 Vergemont Thomson November 2007 31 30 51 50 75 9 3 Native Thomson November 2007 27 NA 32.5 NA 104 8 5 Hunter’s Diamantina November 2007 30 29.5 68 61 58 7.2 1 Lake Constance Diamantina November 2007 30 NA 60 NA 332 6.8 1 Warracoota Diamantina November 2007 27.5 NA 34.9 NA 154 7.3 2 Billyer Diamantina November 2007 32 NA 60 NA 153 6.8 2 Parapituri Georgina November 2007 27 26 49 42 9180 7.3 85 Idamea Georgina November 2007 29 NA 86.5 NA 258 8 17 Georgina Main Georgina November 2007 33 29 55 48 464 8.9 45 Pulchera Mulligan November 2007 29 NA 107 NA 2615 6.1 15 Pump Bulloo March/April 2008 24.3 25.6 29 23 72 7 10 Bulloo Shed Bulloo March/April 2008 20 NA 55 NA 60 7 3 House Bulloo March/April 2008 26 27 45 37 32 8 10 Springfield Kyabra March/April 2008 24 25 75 65 168 8.5 4

326 Springfield South Kyabra March/April 2008 24 NA 70 NA 380 9 3 One Mile Kyabra March/April 2008 25 NA 63 NA 155 9 5 Shed Cooper March/April 2008 22 24 79 66 144 8 3 Currareva Cooper March/April 2008 25 NA 64 NA 139 8 7 Murken Cooper March/April 2008 26.8 24.8 66 56 125 7.5 7 Parapituri Georgina March/April 2008 26 NA 91 NA 24000 8 22 Idamea Georgina March/April 2008 25 NA 111 NA 1032 5.8 8 Georgina Main Georgina March/April 2008 22 23 65 47 942 5.8 38 Warracoota Diamantina March/April 2008 23 NA 64 NA 280 8 2 Lake Constance Diamantina March/April 2008 24 NA 69 NA 386 9 2 Hunter’s Diamantina March/April 2008 26 23 90 63 115 8 3 Billyer Diamantina March/April 2008 24 NA 88 NA 300 8 1 Waterloo Thomson March/April 2008 29 25 62 45 137 7 5 Thomson Main Thomson March/April 2008 26 NA 72 NA 148 8 8 Vergemont Thomson March/April 2008 24.5 21 60 49 70 7 5 Native Thomson March/April 2008 27 24 80 32 96 7 7 Isisford Barcoo March/April 2008 26 24 75 31 200 7 5 Coolagh Barcoo March/April 2008 25 23 78 35 208 8 8 Coolagh 2 Barcoo March/April 2008 27 NA 55 NA 160 8 10 Coolagh 3 Barcoo March/April 2008 25 NA 72 NA 220 8 20 Lake Mary Georgina November 2008 32 NA 90 NA 258 7.7 15 Walkaba/Jimberella Georgina November 2008 29 NA 63.6 NA 345 8.4 70 Waterloo Thomson November 2008 27.6 24.2 62.6 52.1 195 8.7 5 Lake Dunn Thomson November 2008 27.5 NA 70 NA 272 8 5 Thomson Main Thomson November 2008 28.5 NA NA NA 200 7.1 3 Rocky Crossing Diamantina November 2008 28 24.3 52 34.6 64 8.7 6 Conn Diamantina November 2008 29.5 NA 72.9 NA 165 8 4

327 Appendix 3. Length frequency histograms – all fish species, all sites, Queensland Lake Eyre and Bulloo-Bancannia basins, 2006 - 2008.

Length frequency histograms for all species at each sampling site through time are presented on the following pages. In cases where samples were particularly low, sites have been condensed (eg: some species in the Diamantina and Barcoo). Standard length (mm) is presented on the x axes and counts are presented on the y axes.

In general, data from September 2006 – March/April 2008 is presented for the Kyabra and Cooper Creek catchments, and data from December 2006 – March/April 2008 is presented for the Thomson catchment. Data from April 2007 – March/April 2008 is presented for the Barcoo, Diamantina and Georgina catchments. In the Mulligan catchment, data is presented for April 2007 – November 2007 for Pulchera waterhole, for April 2007 – August 2007 for S Bend Gorge and April 2007 only for Dune Pond. The Mulligan waterholes dried down quickly following flooding in January and February 2007.

Abbreviations for sites and sampling times can be interpreted as follows: Bulloo catchment PUMP = Pump SHED = Shearing shed HOUSE = House Kyabra catchment SFIELD = Springfield 1MILE (V1MILE) = 1 Mile SSOUTH = Springfield South

Barcoo catchment COOLA(COOL) = Coolagh ISIS = Isisford C2 = Coolagh 2 Thomson catchment WLOO = Waterloo THOM(THOMS) = Thomson Main Channel

328 VERG = Vergemont Creek NAT = Native Cooper Catchment MURK = Murken CREVA = Currarreva SHED = Shed Diamantina catchment HG = Hunter’s Gorge WCOOTA = Warracoota BILL = Billyer CONST = Lake Constance Georgina catchment PL = Parapituri/Lower IDA = Lake Idamea MC = Georgina Main Channel Mulligan catchment PUL = Pulchera waterhole SBEND = S Bend Gorge DP = Dune Pond Sampling times SEP = September 2006 DEC = December 2006 JAN = January 2007 APR = April 2007 AUG = August 2007 NOV = November 2007 MA = March/April 2008

329 Bony bream, Bulloo catchment, April 2007 – March/April 2008

10 10 10 10

8 8 8 8

6 6 6 6

t t t t

n n n n

u u u u

o o o o C 4 C 4 C 4 C 4

2 2 2 2

0 0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 PUMPAPR PUMPAUG PUMPNOV PUMPMA

10 10 10 10

8 8 8 8

6 6 6 6

t t t t

n n n n

u u u u

o o o o C 4 C 4 C 4 C 4

2 2 2 2

0 0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 SHEDAPR SHEDAUG SHEDNOV SHEDMA

10 10 10 10

8 8 8 8

6 6 6 6

t t t t

n n n n

u u u u

o o o o C 4 C 4 C 4 C 4

2 2 2 2

0 0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 HOUSEAPR HOUSEAUG HOUSENOV HOUSEMA

Standard length (mm)

330 Bony bream, Kyabra catchment, September 2006 – March/April 2008

60 60 60 60 60 60 60

40 40 40 40 40 40 40

t t t t t t t

n n n n n n n

u u u u u u u

o o o o o o o

C C C C C C C

20 20 20 20 20 20 20

0 0 0 0 0 0 0 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 SFIELDSEP SFIELDDEC SFIELDJAN SFIELDAPR SFIELDAUG SFIELDNOV SFIELDMA

60 60 60 60 60 60 60

40 40 40 40 40 40 40

t t t t t t t

n n n n n n n

u u u u u u u

o o o o o o o

C C C C C C C 20 20 20 20 20 20 20

0 0 0 0 0 0 0 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 V1MILESEP V1MILEDEC1MI V1MILEJAN V1MILEAPR V1MILEAUG V1MILENOV V1MILEMA

60 60 60 60 60 60 60

40 40 40 40 40 40 40

t t t t t t t

n n n n n n n

u u u u u u u

o o o o o o o

C C C C C C C 20 20 20 20 20 20 20

0 0 0 0 0 0 0 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 SSOUTHSEP SSOUTHDEC SSOUTHJAN SSOUTHAPR SSOUTHAUG SSOUTHNOV SSOUTHMA

Standard length (mm)

331 Bony bream, Barcoo catchment, April 2007 – March/April 2008

80 80 80

60 60 60

t t t

n n n

u 40 u 40 u 40

o o o

C C C

20 20 20

0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 COOLAPR COOLNOV COOLMA

80 80 80 80

60 60 60 60

t t t t

n n n n

u 40 u 40 u 40 u 40

o o o o

C C C C

20 20 20 20

0 0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 ISISAPR ISISAUG ISISNOV ISISMA

80 80 80

60 60 60

t t t

n n n

u 40 u 40 u 40

o o o

C C C

20 20 20

0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 C2APR C2AUG C2MA Standard length (mm)

332 Bony bream, Thomson catchment, December 2006 – March/April 2008

60 60 60 60 60

40 40 40 40 40

t t t t t

n n n n n

u u u u u

o o o o o

C C C C C

20 20 20 20 20

0 0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 WLOODEC WLOOJAN WLOOAPR WLOOAUG WLOONOV

60 60 60 60 60

40 40 40 40 40

t t t t t

n n n n n

u u u u u

o o o o o

C C C C C

20 20 20 20 20

0 0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 THOMDEC THOMAPR THOMAUG THOMNOV THOMMA

60 60 60 60 60 60

40 40 40 40 40 40

t t t t t t

n n n n n n

u u u u u u

o o o o o o

C C C C C C

20 20 20 20 20 20

0 0 0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 VERGDEC VERGJAN VERGAPR VERGAUG VERGNOV VERGMA

60 60 60 60 60 60

40 40 40 40 40 40

t t t t t t

n n n n n n

u u u u u u

o o o o o o

C C C C C C

20 20 20 20 20 20

0 0 0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 NATDEC NATJAN NATAPR NATAUG NATNOV NATMA Standard length (mm)

333 Bony bream, Cooper catchment, Sep 2006 – March/April 2008.

40 40 40 40 40 40

30 30 30 30 30 30

t t t t t t

n n n n n n

u 20 u 20 u 20 u 20 u 20 u 20

o o o o o o

C C C C C C

10 10 10 10 10 10

0 0 0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 MURKSEP MURKDEC MURKJAN MURKAPR MURKAUG MURKMA

40 40 40 40 40 40 40

30 30 30 30 30 30 30

t t t t t t t

n n n n n n n

u 20 u 20 u 20 u 20 u 20 u 20 u 20

o o o o o o o

C C C C C C C

10 10 10 10 10 10 10

0 0 0 0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 CREVASEP CREVADEC CREVAJAN CREVAAPR CREVAAUG CREVANOV CREVAMA

40 40 40 40 40 40 40

30 30 30 30 30 30 30

t t t t t t t

n n n n n n n

u 20 u 20 u 20 u 20 u 20 u 20 u 20

o o o o o o o

C C C C C C C

10 10 10 10 10 10 10

0 0 0 0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 SHEDSEP SHEDDEC SHEDJAN SHEDAPR SHEDAUG SHEDNOV SHEDMA

Standard length (mm)

334 Bony bream, Diamantina catchment, April 2007 – March/April 2007

10 10

8 8

6 6

t t

n n

u u

o o C 4 C 4

2 2

0 0 0 100 200 300 0 100 200 300 HGAPR HGMA

10 10 10 10

8 8 8 8

6 6 6 6

t t t t

n n n n

u u u u

o o o o C 4 C 4 C 4 C 4

2 2 2 2

0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 BILLAPR BILLAUG BILLNOV BILLMA

10 10 10

8 8 8

6 6 6

t t t

n n n

u u u

o o o C 4 C 4 C 4

2 2 2

0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 WCOOTAUG WCOOTNOV WCOOTMA

10 10

8 8

6 6

t t

n n

u u

o o C 4 C 4

2 2

0 0 0 100 200 300 0 100 200 300 CONSTAUG CONSTNOV Standard length (mm)

335 Bony bream, Georgina catchment, April 2007 – March/April 2008

200 200 200 200

150 150 150 150

t t t t

n n n n

u 100 u 100 u 100 u 100

o o o o

C C C C

50 50 50 50

0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 PLAPR PLAUG PLNOV PLMA

200 200 200

150 150 150

t t t

n n n

u 100 u 100 u 100

o o o

C C C

50 50 50

0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 IDAAPR IDAAUG IDANOV

200 200 200

150 150 150

t t t

n n n

u 100 u 100 u 100

o o o

C C C

50 50 50

0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 MCAPR MCNOV MCMA

Standard length (mm)

336 Bony bream, Mulligan catchment, April 2007 – November 2007 (all sites dry prior to March/April 2008)

50 50 50

40 40 40

30 30 30

t t t

n n n

u u u

o o o C 20 C 20 C 20

10 10 10

0 0 0 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 PULAPR PULAUG PULNOV

50 50

40 40

30 30

t t

n n

u u

o o C 20 C 20

10 10

0 0 0 50 100 150 200 0 50 100 150 200 SBENDAPR SBENDAUG

Standard length (mm)

337 Cooper Creek catfish, all sampled fish pooled from Barcoo/Thomson/Cooper sites, September 2006 – March/April 2008

20 20 20 20 20 20 20

15 15 15 15 15 15 15

t t t t t t t

n n n n n n n

u 10 u 10 u 10 u 10 u 10 u 10 u 10

o o o o o o o

C C C C C C C

5 5 5 5 5 5 5

0 0 0 0 0 0 0 0 100 200 300 400 500 0 100 200 300 400 500 0 100 200 300 400 500 0 100 200 300 400 500 0 100 200 300 400 500 0 100 200 300 400 500 0 100 200 300 400 500 SEP DEC JAN APR AUG NOV MA08

Standard length (mm)

338 Hyrtl’s tandan, Bulloo catchment, April 2007 – March/April 2008

80 80 80 80

60 60 60 60

t t t t

n n n n

u 40 u 40 u 40 u 40

o o o o

C C C C

20 20 20 20

0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 SHEDAPR SHEDAUG SHEDNOV SHEDMA

80 80 80 80

60 60 60 60

t t t t

n n n n

u 40 u 40 u 40 u 40

o o o o

C C C C

20 20 20 20

0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 HOUSEAPR HOUSEAUG HOUSENOV HOUSEMA

Standard length (mm)

339 Hyrtl’s tandan, Kyabra catchment, September 2006 – March/April 2008

30 30 30 30 30 30

20 20 20 20 20 20

t t t t t t

n n n n n n

u u u u u u

o o o o o o

C C C C C C

10 10 10 10 10 10

0 0 0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 SFIELDSEP SFIELDDEC SFIELDJAN SFIELDAPR SFIELDNOV SFIELDMA

30 30 30 30 30 30

20 20 20 20 20 20

t t t t t t

n n n n n n

u u u u u u

o o o o o o

C C C C C C 10 10 10 10 10 10

0 0 0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 V1MILESEP V1MILEDEC1MI V1MILEJAN V1MILEAPR V1MILENOV V1MILEMA

30

20

t

n

u

o

C 10

0 0 100 200 300 SSOUTHNOV

Standard length (mm)

340 Hyrtl’s tandan, Barcoo catchment, April 2007 – March/April 2008

20 20 20

15 15 15

t t t

n n n

u 10 u 10 u 10

o o o

C C C

5 5 5

0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 COOLAPR COOLNOV COOLMA

20 20

15 15

t t

n n

u 10 u 10

o o

C C

5 5

0 0 0 100 200 300 0 100 200 300 ISISAPR ISISMA

Standard length (mm)

341 Hyrtl’s tandan, Thomson catchment, December 2006 – March/April 2008

80 80 80 80 80 80

60 60 60 60 60 60

t t t t t t

n n n n n n

u 40 u 40 u 40 u 40 u 40 u 40

o o o o o o

C C C C C C

20 20 20 20 20 20

0 0 0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 WLOODEC WLOOJAN WLOOAPR WLOOAUG WLOONOV WLOOMA

80 80 80 80

60 60 60 60

t t t t

n n n n

u 40 u 40 u 40 u 40

o o o o

C C C C

20 20 20 20

0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 THOMDEC THOMAPR THOMNOV THOMMA

80 80 80 80 80

60 60 60 60 60

t t t t t

n n n n n

u u u u u

o 40 o 40 o 40 o 40 o 40

C C C C C

20 20 20 20 20

0 0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 VERGDEC VERGJAN VERGAPR VERGNOV VERGMA

Standard length (mm)

NB: Thomson Main Channel sample in March/April 2008: total count <110mm = 2684.

342 Hyrtl’s tandan, Cooper catchment, September 2006 – March/April 2008

40 40 40 40 40

30 30 30 30 30

t t t t t

n n n n n

u 20 u 20 u 20 u 20 u 20

o o o o o

C C C C C

10 10 10 10 10

0 0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 MURKSEP MURKDEC MURKJAN MURKAPR MURKMA

40 40 40 40 40 40

30 30 30 30 30 30

t t t t t t

n n n n n n

u 20 u 20 u 20 u 20 u 20 u 20

o o o o o o

C C C C C C

10 10 10 10 10 10

0 0 0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 CREVASEP CREVADEC CREVAJAN CREVAAPR CREVANOV CREVAMA

40 40 40 40 40 40

30 30 30 30 30 30

t t t t t t

n n n n n n

u 20 u 20 u 20 u 20 u 20 u 20

o o o o o o

C C C C C C

10 10 10 10 10 10

0 0 0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 SHEDSEP SHEDDEC SHEDJAN SHEDAPR SHEDNOV SHEDMA

Standard length (mm)

343 Hyrtl’s tandan, all Diamantina sites, April 2007 – March/April 2008

20 20 20 20

15 15 15 15

t t t t

n n n n

u 10 u 10 u 10 u 10

o o o o

C C C C

5 5 5 5

0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 DIAAPR DIAAUG DIANOV DIAMA

Standard length (mm)

344 Hyrtl’s tandan, Georgina catchment, April 2007 – March/April 2008

50 50 50

40 40 40

30 30 30

t t t

n n n

u u u

o o o C 20 C 20 C 20

10 10 10

0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 PLAPR PLAUG PLMA

50 50

40 40

30 30

t t

n n

u u

o o C 20 C 20

10 10

0 0 0 100 200 300 0 100 200 300 IDAAPR IDAMA

50 50 50 50

40 40 40 40

30 30 30 30

t t t t

n n n n

u u u u

o o o o C 20 C 20 C 20 C 20

10 10 10 10

0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 MCAPR MCAUG MCNOV MCMA

Standard length (mm)

NB: Total sample March/April 2008 = 451, length range 120 – 185 SL (mm)

345 Silver tandan, Bulloo catchment, April 2007 – March/April 2008

15 15 15

10 10 10

t t t

n n n

u u u

o o o

C C C

5 5 5

0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 PUMPAUG PUMPNOV PUMPMA

15 15 15 15

10 10 10 10

t t t t

n n n n

u u u u

o o o o

C C C C

5 5 5 5

0 0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 SHEDAPR SHEDAUG SHEDNOV SHEDMA

15 15 15 15

10 10 10 10

t t t t

n n n n

u u u u

o o o o

C C C C

5 5 5 5

0 0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 HOUSEAPR HOUSEAUG HOUSENOV HOUSEMA

Standard length (mm)

346 Silver tandan, Kyabra catchment, September 2006 – March/April 2008

20 20 20 20 20 20 20

15 15 15 15 15 15 15

t t t t t t t

n n n n n n n

u 10 u 10 u 10 u 10 u 10 u 10 u 10

o o o o o o o

C C C C C C C

5 5 5 5 5 5 5

0 0 0 0 0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 SFIELDSEP SFIELDDEC SFIELDJAN SFIELDAPR SFIELDAUG SFIELDNOV SFIELDMA

20 20 20 20 20 20

15 15 15 15 15 15

t t t t t t

n n n n n n

u 10 u 10 u 10 u 10 u 10 u 10

o o o o o o

C C C C C C

5 5 5 5 5 5

0 0 0 0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 V1MILESEP V1MILEDEC V1MILEJAN V1MILEAPR V1MILENOV V1MILEMA

20 20 20 20 20

15 15 15 15 15

t t t t t

n n n n n

u 10 u 10 u 10 u 10 u 10

o o o o o

C C C C C

5 5 5 5 5

0 0 0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 SSOUTHSEP SSOUTHJAN SSOUTHAPR SSOUTHAUG SSOUTHMA

Standard length (mm)

347 Silver tandan, Barcoo catchment, April 2007 – March/April 2008

5 5 5

4 4 4

3 3 3

t t t

n n n

u u u

o o o C 2 C 2 C 2

1 1 1

0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 COOLAAPR COOLANOV COOLAMA

5 5

4 4

3 3

t t

n n

u u

o o C 2 C 2

1 1

0 0 0 50 100 150 200 250 0 50 100 150 200 250 ISISAPR ISISAUG

5

4

3

t

n

u

o C 2

1

0 0 50 100 150 200 250 C2MA

Standard length (mm)

348 Silver tandan, Thomson catchment, December 2006 – March/April 2008

60 60 60 60

40 40 40 40

t t t t

n n n n

u u u u

o o o o

C C C C

20 20 20 20

0 0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 WLOODEC WLOOAUG WLOONOV WLOOMA

60 60 60 60 60

40 40 40 40 40

t t t t t

n n n n n

u u u u u

o o o o o

C C C C C

20 20 20 20 20

0 0 0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 THOMDEC THOMJAN THOMAUG THOMNOV THOMMA

60 60 60 60

40 40 40 40

t t t t

n n n n

u u u u

o o o o

C C C C

20 20 20 20

0 0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 VERGDEC VERGJAN VERGAPR VERGMA

60 60

40 40

t t

n n

u u

o o

C C

20 20

0 0 0 50 100 150 200 250 0 50 100 150 200 250 NATJAN NATAPR

Standard length (mm)

349 Silver tandan, Cooper catchment, September 2006 – March/April 2008

30 30 30 30 30 30

20 20 20 20 20 20

t t t t t t

n n n n n n

u u u u u u

o o o o o o

C C C C C C

10 10 10 10 10 10

0 0 0 0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 MURKSEP MURKDEC MURKJAN MURKAPR MURKAUG MURKMA

30 30 30 30 30 30 30

20 20 20 20 20 20 20

t t t t t t t

n n n n n n n

u u u u u u u

o o o o o o o

C C C C C C C

10 10 10 10 10 10 10

0 0 0 0 0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 CREVASEP CREVADEC CREVAJAN CREVAAPR CREVAAUG CREVANOV CREVAMA

30 30 30 30 30

20 20 20 20 20

t t t t t

n n n n n

u u u u u

o o o o o

C C C C C

10 10 10 10 10

0 0 0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 SHEDSEP SHEDDEC SHEDJAN SHEDAPR SHEDMA

Standard length (mm)

350 Silver tandan, all Diamantina sites, April 2007 – March/April 2008

80 80 80 80

60 60 60 60

t t t t

n n n n

u 40 u 40 u 40 u 40

o o o o

C C C C

20 20 20 20

0 0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 APR AUG NOV MA

Standard length (mm)

351 Silver tandan, Georgina catchment, April 2007 – March/April 2008

50

40

30

t

n

u

o C 20

10

0 0 50 100 150 200 250 PLAPR

50 50 50

40 40 40

30 30 30

t t t

n n n

u u u

o o o C 20 C 20 C 20

10 10 10

0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 IDAAPR IDANOV IDAMA

50 50 50

40 40 40

30 30 30

t t t

n n n

u u u

o o o C 20 C 20 C 20

10 10 10

0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 MCAPR MCAUG MCMA

Standard length (mm)

352 Silver tandan, Mulligan catchment, April 2007 – November 2007

80 80 80

60 60 60

t t t

n n n

u 40 u 40 u 40

o o o

C C C

20 20 20

0 0 0 0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250 PULAPR PULAUG PULNOV

80 80

60 60

t t

n n

u 40 u 40

o o

C C

20 20

0 0 0 50 100 150 200 250 0 50 100 150 200 250 SBENDAPR SBENDAUG

Standard length (mm)

353 Australian smelt, Kyabra catchment, September 2006 – March/April 2008

10 10 10 10 10 10

8 8 8 8 8 8

6 6 6 6 6 6

t t t t t t

n n n n n n

u u u u u u

o o o o o o C 4 C 4 C 4 C 4 C 4 C 4

2 2 2 2 2 2

0 0 0 0 0 0 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 SFIELDSEP SFIELDDEC SFIELDAPR SFIELDAUG SFIELDNOV SFIELDMA

10 10 10 10

8 8 8 8

6 6 6 6

t t t t

n n n n

u u u u

o o o o C 4 C 4 C 4 C 4

2 2 2 2

0 0 0 0 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 V1MILESEP V1MILEDEC V1MILEAUG V1MILENOV

10 10 10 10

8 8 8 8

6 6 6 6

t t t t

n n n n

u u u u

o o o o C 4 C 4 C 4 C 4

2 2 2 2

0 0 0 0 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 SSOUTHSEP SSOUTHDEC SSOUTHJAN SSOUTHNOV(1)

Standard length (mm)

354 Australian smelt, all Barcoo sites, April 2007 – March/April 2008

5 5

4 4

3 3

t t

n n

u u

o o C 2 C 2

1 1

0 0 0 10 20 30 40 50 0 10 20 30 40 50 APR NOV

Standard length (mm)

355 Australian smelt, Thomson catchment, December 2006 – March/April 2008

15 15 15 15 15

10 10 10 10 10

t t t t t

n n n n n

u u u u u

o o o o o

C C C C C

5 5 5 5 5

0 0 0 0 0 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 WLOODEC WLOOJAN WLOOAPR WLOOAUG WLOONOV

15 15 15 15 15

10 10 10 10 10

t t t t t

n n n n n

u u u u u

o o o o o

C C C C C

5 5 5 5 5

0 0 0 0 0 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 THOMDEC THOMJAN THOMAPR THOMAUG THOMNOV

15

10

t

n

u

o

C

5

0 0 10 20 30 40 50 VERGDEC

15 15 15 15

10 10 10 10

t t t t

n n n n

u u u u

o o o o

C C C C

5 5 5 5

0 0 0 0 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 NATDEC NATJAN NATAUG NATNOV

Standard length (mm)

356 Australian smelt, Cooper catchment, September 2006 – March/April 2008

100 100 100 100 100 100 100

80 80 80 80 80 80 80

60 60 60 60 60 60 60

t t t t t t t

n n n n n n n

u u u u u u u

o o o o o o o C 40 C 40 C 40 C 40 C 40 C 40 C 40

20 20 20 20 20 20 20

0 0 0 0 0 0 0 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 MURKSEP MURKDEC MURKJAN MURKAPR MURKAUG MURKNOV MURKMA

100 100 100 100 100 100

80 80 80 80 80 80

60 60 60 60 60 60

t t t t t t

n n n n n n

u u u u u u

o o o o o o C 40 C 40 C 40 C 40 C 40 C 40

20 20 20 20 20 20

0 0 0 0 0 0 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 CREVASEP CREVADEC CREVAJAN CREVAAUG CREVANOV CREVAMA

100 100 100 100 100 100 100

80 80 80 80 80 80 80

60 60 60 60 60 60 60

t t t t t t t

n n n n n n n

u u u u u u u

o o o o o o o C 40 C 40 C 40 C 40 C 40 C 40 C 40

20 20 20 20 20 20 20

0 0 0 0 0 0 0 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 SHEDSEP SHEDDEC SHEDJAN SHEDAPR SHEDAUG SHEDNOV SHEDMA

Standard length (mm)

357 Desert rainbowfish, all Bulloo sites, April 2007 – March/April 2008

5 5 5

4 4 4

3 3 3

t t t

n n n

u u u

o o o C 2 C 2 C 2

1 1 1

0 0 0 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 APR AUG MA

Standard length (mm)

358 Desert rainbowfish, Kyabra catchment, September 2006 – March/April 2008

10 10 10 10

8 8 8 8

6 6 6 6

t t t t

n n n n

u u u u

o o o o C 4 C 4 C 4 C 4

2 2 2 2

0 0 0 0 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 SFIELDDEC SFIELDJAN SFIELDAPR SFIELDAUG

10 10 10 10 10

8 8 8 8 8

6 6 6 6 6

t t t t t

n n n n n

u u u u u

o o o o o C 4 C 4 C 4 C 4 C 4

2 2 2 2 2

0 0 0 0 0 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 V1MILEDEC V1MILEJAN V1MILEAPR V1MILEAUG V1MILEMA

10 10 10 10 10 10

8 8 8 8 8 8

6 6 6 6 6 6

t t t t t t

n n n n n n

u u u u u u

o o o o o o C 4 C 4 C 4 C 4 C 4 C 4

2 2 2 2 2 2

0 0 0 0 0 0 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 SSOUTHSEP SSOUTHDEC SSOUTHJAN SSOUTHAPR SSOUTHAUG SSOUTHMA

Standard length (mm)

359 Desert rainbowfish, all Thomson sites, December 2006 – March/April 2008

20 20 20 20 20 20

15 15 15 15 15 15

t t t t t t

n n n n n n

u 10 u 10 u 10 u 10 u 10 u 10

o o o o o o

C C C C C C

5 5 5 5 5 5

0 0 0 0 0 0 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 DEC JAN APR AUG NOV MA

Standard length (mm)

360 Desert rainbowfish, all Cooper sites, April 2007 – March/April 2008 (Desert rainbowfish not sampled at Cooper sites prior to

April).

3 3 3

2 2 2

t t t

n n n

u u u

o o o

C C C

1 1 1

0 0 0 0 20 40 60 0 20 40 60 0 20 40 60 APR NOV MA

Standard length (mm)

361 Desert rainbowfish, all Diamantina sites, April 2007 – March/April 2008

5 5 5 5

4 4 4 4

3 3 3 3

t t t t

n n n n

u u u u

o o o o C 2 C 2 C 2 C 2

1 1 1 1

0 0 0 0 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 APR AUG NOV MA

Standard length (mm)

362 Desert rainbowfish, Georgina catchment, April 2007 – March/April 2008

200 200 200 200

150 150 150 150

t t t t

n n n n

u 100 u 100 u 100 u 100

o o o o

C C C C

50 50 50 50

0 0 0 0 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 PLAPR PLAUG PLNOV PLMA

200 200 200 200

150 150 150 150

t t t t

n n n n

u 100 u 100 u 100 u 100

o o o o

C C C C

50 50 50 50

0 0 0 0 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 IDAAPR IDAAUG IDANOV IDAMA

200 200 200

150 150 150

t t t

n n n

u 100 u 100 u 100

o o o

C C C

50 50 50

0 0 0 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 MCAPR MCNOV MCMA

Standard length (mm)

363 Desert rainbowfish, Mulligan catchment, April 2007 – March/April 2008

130 130

104 104

78 78

t t

n n

u u

o o C 52 C 52

26 26

0 0 0 20 40 60 80 0 20 40 60 80 PULAPR PULAUG

130 130

104 104

78 78

t t

n n

u u

o o C 52 C 52

26 26

0 0 0 20 40 60 80 0 20 40 60 80 SBENDAPR SBENDAUG

Standard length (mm)

364 Glassfish, Bulloo catchment, April 2007 – March/April 2008

10

8

6

t

n

u

o C 4

2

0 0 20 40 60 PUMPNOV

10 10 10

8 8 8

6 6 6

t t t

n n n

u u u

o o o C 4 C 4 C 4

2 2 2

0 0 0 0 20 40 60 0 20 40 60 0 20 40 60 SHEDAPR SHEDAUG SHEDMA

10 10 10 10

8 8 8 8

6 6 6 6

t t t t

n n n n

u u u u

o o o o C 4 C 4 C 4 C 4

2 2 2 2

0 0 0 0 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 HOUSEAPR HOUSEAUG HOUSENOV HOUSEMA

Standard length (mm)

365 Glassfish, all Kyabra sites, September 2006 – March/April 2008

5 5 5 5 5

4 4 4 4 4

3 3 3 3 3

t t t t t

n n n n n

u u u u u

o o o o o C 2 C 2 C 2 C 2 C 2

1 1 1 1 1

0 0 0 0 0 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 SEP DEC JAN APR MA

Standard length (mm)

Glassfish, all Barcoo sites, April 2007 – March/April 2008

10 10 10

8 8 8

6 6 6

t t t

n n n

u u u

o o o C 4 C 4 C 4

2 2 2

0 0 0 0 20 40 60 0 20 40 60 0 20 40 60 APR AUG MA

Standard length (mm)

366 Glassfish, all Thomson sites, December 2006 – March/April 2008

5 5 5 5 5 5

4 4 4 4 4 4

3 3 3 3 3 3

t t t t t t

n n n n n n

u u u u u u

o o o o o o C 2 C 2 C 2 C 2 C 2 C 2

1 1 1 1 1 1

0 0 0 0 0 0 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 DEC JAN APR AUG NOV MA

Standard length (mm)

Glassfish, all Cooper sites, September 2006 and January 2007 (Glassfish absent at all other times)

5 5

4 4

3 3

t t

n n

u u

o o C 2 C 2

1 1

0 0 0 20 40 60 0 20 40 60 SEP JAN

Standard length (mm)

NB: Glassfish absent from the Diamantina catchment on all sampling occasions.

367 Glassfish, Georgina catchment, April 2007 – March/April 2008

150 150 150 150

100 100 100 100

t t t t

n n n n

u u u u

o o o o

C C C C 50 50 50 50

0 0 0 0 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 PLAPR PLAUG PLNOV PLMA

150 150 150 150

100 100 100 100

t t t t

n n n n

u u u u

o o o o

C C C C 50 50 50 50

0 0 0 0 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 IDAAPR IDAAUG IDANOV IDAMA

150 150 150 150

100 100 100 100

t t t t

n n n n

u u u u

o o o o

C C C C 50 50 50 50

0 0 0 0 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 MCAPR MCAUG MCNOV MCMA

Standard length (mm)

368 Glassfish, Mulligan catchment, April 2007 – November 2007

50 50 50

40 40 40

30 30 30

t t t

n n n

u u u

o o o C 20 C 20 C 20

10 10 10

0 0 0 0 20 40 60 0 20 40 60 0 20 40 60 PULAPR PULAUG PULNOV

50 50

40 40

30 30

t t

n n

u u

o o C 20 C 20

10 10

0 0 0 20 40 60 0 20 40 60 SBENDAPR SBENDAUG

Standard length (mm)

369 Yellowbelly, Bulloo catchment, April 2007 – March/April 2008

5 5 5 5

4 4 4 4

3 3 3 3

t t t t

n n n n

u u u u

o o o o C 2 C 2 C 2 C 2

1 1 1 1

0 0 0 0 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 PUMPAPR PUMPAUG PUMPNOV PUMPMA

5 5

4 4

3 3

t t

n n

u u

o o C 2 C 2

1 1

0 0 0 130 260 390 520 0 130 260 390 520 SHEDAPR SHEDNOV

5 5 5

4 4 4

3 3 3

t t t

n n n

u u u

o o o C 2 C 2 C 2

1 1 1

0 0 0 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 HOUSEAPR HOUSEAUG HOUSENOV

Standard length (mm)

370 Yellowbelly, Kyabra catchment, September 2006 – March/April 2008

10 10 10 10 10 10 10

8 8 8 8 8 8 8

6 6 6 6 6 6 6

t t t t t t t

n n n n n n n

u u u u u u u

o o o o o o o C 4 C 4 C 4 C 4 C 4 C 4 C 4

2 2 2 2 2 2 2

0 0 0 0 0 0 0 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 SFIELDSEP SFIELDDEC SFIELDJAN SFIELDAPR SFIELDAUG SFIELDNOV SFIELDMA

10 10 10 10 10 10 10

8 8 8 8 8 8 8

6 6 6 6 6 6 6

t t t t t t t

n n n n n n n

u u u u u u u

o o o o o o o C 4 C 4 C 4 C 4 C 4 C 4 C 4

2 2 2 2 2 2 2

0 0 0 0 0 0 0 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 V1MILESEP V1MILEDEC V1MILEJAN V1MILEAPR V1MILEAUG V1MILENOV V1MILEMA

10 10 10 10 10 10

8 8 8 8 8 8

6 6 6 6 6 6

t t t t t t

n n n n n n

u u u u u u

o o o o o o C 4 C 4 C 4 C 4 C 4 C 4

2 2 2 2 2 2

0 0 0 0 0 0 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 SSOUTHDEC SSOUTHJAN SSOUTHAPR SSOUTHAUG SSOUTHNOV SSOUTHMA

Standard length (mm)

371 Yellowbelly, Barcoo catchment, April 2007 – March/April 2008

50 50 50 50

40 40 40 40

30 30 30 30

t t t t

n n n n

u u u u

o o o o C 20 C 20 C 20 C 20

10 10 10 10

0 0 0 0 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 COOLAAPR COOLAAUG COOLANOV COOLAMA

50 50 50 50

40 40 40 40

30 30 30 30

t t t t

n n n n

u u u u

o o o o C 20 C 20 C 20 C 20

10 10 10 10

0 0 0 0 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 ISISAPR ISISAUG ISISNOV ISISMA

50 50 50 50

40 40 40 40

30 30 30 30

t t t t

n n n n

u u u u

o o o o C 20 C 20 C 20 C 20

10 10 10 10

0 0 0 0 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 C2APR C2AUG C2NOV C2MA

Standard length (mm)

372 Yellowbelly, Thomson catchment, December 2006 – March/April 2008

10 10 10 10 10 10

8 8 8 8 8 8

6 6 6 6 6 6

t t t t t t

n n n n n n

u u u u u u

o o o o o o C 4 C 4 C 4 C 4 C 4 C 4

2 2 2 2 2 2

0 0 0 0 0 0 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 WLOODEC WLOOJAN WLOOAPR WLOOAUG WLOONOV WLOOMA

10 10 10 10 10 10

8 8 8 8 8 8

6 6 6 6 6 6

t t t t t t

n n n n n n

u u u u u u

o o o o o o C 4 C 4 C 4 C 4 C 4 C 4

2 2 2 2 2 2

0 0 0 0 0 0 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 THOMDEC THOMJAN THOMAPR THOMAUG THOMNOV THOMMA

10 10 10 10 10 10

8 8 8 8 8 8

6 6 6 6 6 6

t t t t t t

n n n n n n

u u u u u u

o o o o o o C 4 C 4 C 4 C 4 C 4 C 4

2 2 2 2 2 2

0 0 0 0 0 0 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 VERGDEC VERGJAN VERGAPR VERGAUG VERGNOV VERGMA

10 10 10 10 10 10

8 8 8 8 8 8

6 6 6 6 6 6

t t t t t t

n n n n n n

u u u u u u

o o o o o o C 4 C 4 C 4 C 4 C 4 C 4

2 2 2 2 2 2

0 0 0 0 0 0 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 NATDEC NATJAN NATAPR NATAUG NATNOV NATMA

Standard length (mm)

373 Yellowbelly, Cooper catchment, September 2006 – March/April 2008

90 90 90 90 90 90

60 60 60 60 60 60

t t t t t t

n n n n n n

u u u u u u

o o o o o o

C C C C C C

30 30 30 30 30 30

0 0 0 0 0 0 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 MURKSEP MURKDEC MURKJAN MURKAPR MURKAUG MURKMA

90 90 90 90 90 90 90

60 60 60 60 60 60 60

t t t t t t t

n n n n n n n

u u u u u u u

o o o o o o o

C C C C C C C

30 30 30 30 30 30 30

0 0 0 0 0 0 0 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 CREVASEP CREVADEC CREVAJAN CREVAAPR CREVAAUG CREVANOV CREVAMA

90 90 90 90 90 90 90

60 60 60 60 60 60 60

t t t t t t t

n n n n n n n

u u u u u u u

o o o o o o o

C C C C C C C

30 30 30 30 30 30 30

0 0 0 0 0 0 0 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 SHEDSEP SHEDDEC SHEDJAN SHEDAPR SHEDAUG SHEDNOV SHEDMA

Standard length (mm)

374 Yellowbelly, Diamantina catchment, April 2007 – March/April 2008

30 30 30

20 20 20

t t t

n n n

u u u

o o o

C C C 10 10 10

0 0 0 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 HGAPR HGNOV HGMA

30 30 30 30

20 20 20 20

t t t t

n n n n

u u u u

o o o o

C C C C

10 10 10 10

0 0 0 0 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 BILLAPR BILLAUG BILLNOV BILLMA

30 30 30

20 20 20

t t t

n n n

u u u

o o o

C C C 10 10 10

0 0 0 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 WCOOTAAUG WCOOTANOV WCOOTAMA

30 30 30

20 20 20

t t t

n n n

u u u

o o o

C C C

10 10 10

0 0 0 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 CONSTAUG CONSTNOV CONSTMA

375 Yellowbelly, all Georgina sites, April 2007 – March/April 2008

10 10 10 10

8 8 8 8

6 6 6 6

t t t t

n n n n

u u u u

o o o o C 4 C 4 C 4 C 4

2 2 2 2

0 0 0 0 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 0 130 260 390 520 APR AUG NOV MA

Standard length (mm)

376 Banded grunter, Georgina catchment, April 2007 – March/April 2008

50 50 50 50

40 40 40 40

30 30 30 30

t t t t

n n n n

u u u u

o o o o C 20 C 20 C 20 C 20

10 10 10 10

0 0 0 0 0 50 100 150 0 50 100 150 0 50 100 150 0 50 100 150 PLAPR PLAUG PLNOV PLMA

50 50

40 40

30 30

t t

n n

u u

o o C 20 C 20

10 10

0 0 0 50 100 150 0 50 100 150 IDAAPR IDAMA

50 50 50

40 40 40

30 30 30

t t t

n n n

u u u

o o o C 20 C 20 C 20

10 10 10

0 0 0 0 50 100 150 0 50 100 150 0 50 100 150 MCAPR MCAUG MCNOV

Standard length (mm)

NB: Two banded grunter (33 and 34mm SL) were sampled in the Mulligan catchment in April 2007.

377 Welch’s grunter, all Kyabra catchments, September 2006 – November 2007

5 5 5 5 5

4 4 4 4 4

3 3 3 3 3

t t t t t

n n n n n

u u u u u

o o o o o C 2 C 2 C 2 C 2 C 2

1 1 1 1 1

0 0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 SEP JAN APR AUG NOV

Standard length (mm)

Welch’s grunter, all Barcoo catchments, November 2007

5

4

3

t

n

u

o C 2

1

0 0 100 200 300 NOV

Standard length (mm)

378 Welch’s grunter, all Thomson catchments, April 2007 – March/April 2008

5 5 5

4 4 4

3 3 3

t t t

n n n

u u u

o o o C 2 C 2 C 2

1 1 1

0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 APR NOV MA

Standard length (mm)

Welch’s grunter, all Cooper catchments, September 2006 – March/April 2008

20 20 20 20

15 15 15 15

t t t t

n n n n

u 10 u 10 u 10 u 10

o o o o

C C C C

5 5 5 5

0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 SEP APR NOV MA

Standard length (mm)

379 Welch’s grunter, all Diamantina catchments, November 2007

5

4

3

t

n

u

o C 2

1

0 0 100 200 300 NOV

Standard length (mm)

Welch’s grunter, all Georgina catchments, August 2007

5

4

3

t

n

u

o C 2

1

0 0 100 200 300 AUG

Standard length (mm)

380 Spangled perch, Bulloo catchment, April 2007 – March/April 2008

35 35 35 35

) 28 ) 28 ) 28 ) 28

s s s s

l l l l

a a a a

u u u u

d d d d

i i i i

v v v v i 21 i 21 i 21 i 21

d d d d

n n n n

i i i i

( ( ( (

y y y y

c c c c

n 14 n 14 n 14 n 14

e e e e

u u u u

q q q q

e e e e

r r r r

F 7 F 7 F 7 F 7

0 0 0 0 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 PUMPAPR PUMPAUG PUMPNOV PUMPMA

35 35 35 35

) 28 ) 28 ) 28 ) 28

s s s s

l l l l

a a a a

u u u u

d d d d

i i i i

v v v v i 21 i 21 i 21 i 21

d d d d

n n n n

i i i i

( ( ( (

y y y y

c c c c

n 14 n 14 n 14 n 14

e e e e

u u u u

q q q q

e e e e

r r r r

F 7 F 7 F 7 F 7

0 0 0 0 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 SHEDAPR SHEDAUG SHEDNOV SHEDMA

35 35 35 35

) 28 ) 28 ) 28 ) 28

s s s s

l l l l

a a a a

u u u u

d d d d

i i i i

v v v v i 21 i 21 i 21 i 21

d d d d

n n n n

i i i i

( ( ( (

y y y y

c c c c

n 14 n 14 n 14 n 14

e e e e

u u u u

q q q q

e e e e

r r r r

F 7 F 7 F 7 F 7

0 0 0 0 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 HOUSEAPR HOUSEAUG HOUSENOV HOUSEMA

Standard length (mm)

381

Spangled perch, Kyabra catchment, September 2006 – March/April 2008

15 15 15 15 15

) ) ) ) )

s s s s s

l l l l l

a a a a a

u u u u u

d 10 d 10 d 10 d 10 d 10

i i i i i

v v v v v

i i i i i

d d d d d

n n n n n

i i i i i

( ( ( ( (

y y y y y

c c c c c

n n n n n

e e e e e

u 5 u 5 u 5 u 5 u 5

q q q q q

e e e e e

r r r r r

F F F F F

0 0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 SFIELDSEP SFIELDDEC SFIELDJAN SFIELDNOV SFIELDMA

15 15 15 15 15 15 15

) ) ) ) ) ) )

s s s s s s s

l l l l l l l

a a a a a a a

u u u u u u u

d 10 d 10 d 10 d 10 d 10 d 10 d 10

i i i i i i i

v v v v v v v

i i i i i i i

d d d d d d d

n n n n n n n

i i i i i i i

( ( ( ( ( ( (

y y y y y y y

c c c c c c c

n n n n n n n

e e e e e e e

u 5 u 5 u 5 u 5 u 5 u 5 u 5

q q q q q q q

e e e e e e e

r r r r r r r

F F F F F F F

0 0 0 0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 V1MILESEP V1MILEDEC V1MILEJAN V1MILEAPR V1MILEAUG V1MILENOV V1MILEMA

15 15 15 15 15 15

) ) ) ) ) )

s s s s s s

l l l l l l

a a a a a a

u u u u u u

d 10 d 10 d 10 d 10 d 10 d 10

i i i i i i

v v v v v v

i i i i i i

d d d d d d

n n n n n n

i i i i i i

( ( ( ( ( (

y y y y y y

c c c c c c

n n n n n n

e e e e e e

u 5 u 5 u 5 u 5 u 5 u 5

q q q q q q

e e e e e e

r r r r r r

F F F F F F

0 0 0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 SSOUTHDEC SSOUTHJAN SSOUTHAPR SSOUTHAUG SSOUTHNOV SSOUTHMA

Standard length (mm)

382 Spangled perch, Barcoo catchment, April 2007 – March/April 2008

30

20

t

n

u

o

C

10

0 0 50 100 150 200 COOLAGHMA

30

20

t

n

u

o

C

10

0 0 50 100 150 200 ISISAPR

30 30 30

20 20 20

t t t

n n n

u u u

o o o

C C C

10 10 10

0 0 0 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 C2APR C2AUG C2MA

Standard length (mm)

383 Spangled perch, Thomson catchment, December 2006 – March/April 2008

10 10

8 8

6 6

t t

n n

u u

o o C 4 C 4

2 2

0 0 0 50 100 150 200 0 50 100 150 200 WLOONOV WLOOMA

10 10 10 10 10

8 8 8 8 8

6 6 6 6 6

t t t t t

n n n n n

u u u u u

o o o o o C 4 C 4 C 4 C 4 C 4

2 2 2 2 2

0 0 0 0 0 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 THOMSJAN THOMSAPR THOMSAUG THOMSNOV THOMSMA

10 10 10 10 10

8 8 8 8 8

6 6 6 6 6

t t t t t

n n n n n

u u u u u

o o o o o C 4 C 4 C 4 C 4 C 4

2 2 2 2 2

0 0 0 0 0 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 VERGDEC VERGJAN VERGAPR VERGAUG VERGNOV

10 10 10 10

8 8 8 8

6 6 6 6

t t t t

n n n n

u u u u

o o o o C 4 C 4 C 4 C 4

2 2 2 2

0 0 0 0 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 NATDEC NATJAN NATAUG NATMA Standard length (mm)

384 Spangled perch, Cooper catchment, Jan 2007 – March/April 2008

20 20 20

15 15 15

t t t

n n n

u 10 u 10 u 10

o o o

C C C

5 5 5

0 0 0 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 SHEDJAN SHEDNOV SHEDMA

20 20 20

15 15 15

t t t

n n n

u 10 u 10 u 10

o o o

C C C

5 5 5

0 0 0 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 MURKJAN MURKAUG MURKMA

20 20 20

15 15 15

t t t

n n n

u 10 u 10 u 10

o o o

C C C

5 5 5

0 0 0 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 CREVAJAN CREVANOV CREVAMA

Standard length (mm)

385 Spangled perch, all Diamantina sites, April 2007 – November 2007

10 10 10

8 8 8

6 6 6

t t t

n n n

u u u

o o o C 4 C 4 C 4

2 2 2

0 0 0 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 ALLSITESAP ALLSITESAU ALLSITESNO

Standard length (mm)

386 Spangled perch, Georgina catchment, April 2007 – March/April 2008

50 50

40 40

30 30

t t

n n

u u

o o C 20 C 20

10 10

0 0 0 50 100 150 200 0 50 100 150 200 PLAPR PLAUG

50 50

40 40

30 30

t t

n n

u u

o o C 20 C 20

10 10

0 0 0 50 100 150 200 0 50 100 150 200 IDAAPR IDAMA

50 50

40 40

30 30

t t

n n

u u

o o C 20 C 20

10 10

0 0 0 50 100 150 200 0 50 100 150 200 MCAPR MCAUG

Standard length (mm)

387 Spangled perch, Mulligan catchment, April 2007 – November 2007

15 15 15

10 10 10

t t t

n n n

u u u

o o o

C C C 5 5 5

0 0 0 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 PULAPR PULAUG PULNOV

15 15

10 10

t t

n n

u u

o o

C C

5 5

0 0 0 50 100 150 200 0 50 100 150 200 SBAPR SBAUG

15

10

t

n

u

o

C 5

0 0 50 100 150 200 DPAPR Standard length (mm)

388 Barcoo grunter, all Kyabra sites, September 2006 – March/April 2008

5 5

4 4

3 3

t t

n n

u u

o o C 2 C 2

1 1

0 0 0 100 200 300 0 100 200 300 DEC NOV

Standard length (mm)

Barcoo grunter, all Barcoo sites, September 2006 – March/April 2008

5 5 5 5

4 4 4 4

3 3 3 3

t t t t

n n n n

u u u u

o o o o C 2 C 2 C 2 C 2

1 1 1 1

0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 JAN APR NOV MA

Standard length (mm)

389 Barcoo grunter, all Thomson sites, September 2006 – March/April 2008

5 5 5 5 5

4 4 4 4 4

3 3 3 3 3

t t t t t

n n n n n

u u u u u

o o o o o C 2 C 2 C 2 C 2 C 2

1 1 1 1 1

0 0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 DEC JAN APR NOV MA

Standard length (mm)

Barcoo grunter, all Cooper sites, September 2006 – March/April 2008

10 10 10 10 10 10

8 8 8 8 8 8

6 6 6 6 6 6

t t t t t t

n n n n n n

u u u u u u

o o o o o o C 4 C 4 C 4 C 4 C 4 C 4

2 2 2 2 2 2

0 0 0 0 0 0 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300 DEC JAN APR AUG NOV MA

Standard length (mm)

390 Barcoo grunter, all Diamantina sites, April 2007 – March/April 2008

5 5

4 4

3 3

t t

n n

u u

o o C 2 C 2

1 1

0 0 0 100 200 300 0 100 200 300 APR NOV

Standard length (mm)

Barcoo grunter, all Georgina sites, April 2007 – March/April 2008

5 5

4 4

3 3

t t

n n

u u

o o C 2 C 2

1 1

0 0 0 100 200 300 0 100 200 300 APR MA Standard length (mm)

391 Barcoo grunter, all Mulligan sites, April 2007 – March/April 2008

2.0 2.0

1.5 1.5

t t

n n

u 1.0 u 1.0

o o

C C

0.5 0.5

0.0 0.0 0 100 200 300 0 100 200 300 APR AUG

Standard length (mm)

Golden goby, all Georgina sites, April 2007 – March/April 2008

10 10 10

8 8 8

6 6 6

t t t

n n n

u u u

o o o C 4 C 4 C 4

2 2 2

0 0 0 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 APR NOV MA Standard length (mm) A single golden goby (100mm) was sampled at Lake Billyer in the Diamantina catchment in April 2007.

392 Carp gudgeon, all Bulloo sites, April 2007 – March/April 2008

10 10 10 10

8 8 8 8

6 6 6 6

t t t t

n n n n

u u u u

o o o o C 4 C 4 C 4 C 4

2 2 2 2

0 0 0 0 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 APR AUG NOV MA Standard length (mm)

Carp gudgeon, all Kyabra sites, September 2006 – March/April 2008

20 20 20 20 20 20 20

15 15 15 15 15 15 15

t t t t t t t

n n n n n n n

u 10 u 10 u 10 u 10 u 10 u 10 u 10

o o o o o o o

C C C C C C C

5 5 5 5 5 5 5

0 0 0 0 0 0 0 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 SEP DEC JAN APR AUG NOV MA

Standard length (mm)

393 Carp gudgeon, all Barcoo sites, September 2006 – March/April 2008

10 10 10 10 10 10 10

8 8 8 8 8 8 8

6 6 6 6 6 6 6

t t t t t t t

n n n n n n n

u u u u u u u

o o o o o o o C 4 C 4 C 4 C 4 C 4 C 4 C 4

2 2 2 2 2 2 2

0 0 0 0 0 0 0 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 SEP DEC JAN APR AUG NOV MA

Standard length (mm)

Carp gudgeon, all Thomson sites, September 2006 – March/April 2008

20 20 20 20 20 20 20

15 15 15 15 15 15 15

t t t t t t t

n n n n n n n

u 10 u 10 u 10 u 10 u 10 u 10 u 10

o o o o o o o

C C C C C C C

5 5 5 5 5 5 5

0 0 0 0 0 0 0 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 SEP DEC JAN APR AUG NOV MA

Standard length (mm)

394 Carp gudgeon, all Cooper sites, September 2006 – March/April 2008

10 10 10 10 10

8 8 8 8 8

6 6 6 6 6

t t t t t

n n n n n

u u u u u

o o o o o C 4 C 4 C 4 C 4 C 4

2 2 2 2 2

0 0 0 0 0 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 SEP DEC JAN APR NOV

Standard length (mm)

395 Appendix 4. Field identification of catfish in the Queensland Lake Eyre and Bulloo Bancannia Basins.

1. General characteristics

Three Plotosid catfish are known to inhabit the Queensland Lake Eyre Basin, the silver tandan, Porochilus argenteus , Hyrtl’s tandan, Neosiluris hyrtlii and the Cooper Creek catfish, Neosiluroides cooperensis . A fourth species is endemic to the Dalhousie mound spring complex and is not discussed here.

Identifying the three catfish species is relatively straightforward in most instances. Silver tandan are a slender, silver or whitish fish with a forward-pointing mouth (Figure 1). Female silver tandan displayed distended bellies consistent with egg production during all samples taken in early summer during the current study. Silver tandan were sampled throughout the Queensland Lake Eyre and Bulloo-Bancannia Basins from June 2006 – April 2008.

Figure 1. A 90mm (SL) silver tandan from the Mulligan catchment. Photo: Adam Kerezsy

Hyrtl’s tandan are a comparatively stocky, golden or blotched fish with a slightly more downward-pointing mouth than silver tandan (Figure 2). Hyrtl’s tandan also

396 generally display a more sloping head profile and larger eyes than silver tandan. Female Hyrtl’s tandan displayed distended bellies consistent with egg production during all samples taken in early summer during the current study. Hyrtl’s tandan were sampled throughout the Queensland Lake Eyre and Bulloo-Bancannia Basins from June 2006 – April 2008, however no Hyrtl’s tandan were recorded from waterholes in the Mulligan catchment.

Figure 2. A 170mm (SL) Hyrtl’s tandan from the Thomson River catchment. Photo Angus Emmott.

Cooper Creek catfish are a large-bodied benthic species with a pronounced downward-pointing mouth and anguilliform (eel-like) body (Figure 3). Cooper Creek catfish are generally a mottled grey colour, although they can occasionally be very pale (ie: almost white). Female Cooper Creek catfish displayed distended bellies consistent with egg production during all samples taken in early summer during the current study. Cooper Creek catfish were sampled throughout the Cooper Creek catchment (Thomson, Barcoo, Cooper sub-catchments) from June 2006 – April 2008, however no Cooper Creek catfish were recorded from waterholes in Kyabra Creek.

397

Figure 3. A juvenile (125mm) Cooper Creek catfish from the Barcoo catchment Photo Adam Kerezsy.

2. Other catfish

During sampling in Pulchera waterhole in the Mulligan catchment in April 2007 a single small (43mm) catfish was sampled amongst a much larger number of silver tandans (Figure 4). This catfish is suspected to be a mutant silver tandan that probably survived due to enhanced migration and habitat options provided by antecedent flooding in January 2007.

398

Figure 4. Mutant silver tandan (above) and silver tandan (below) from the Mulligan catchment in April 2007. Photo Angus Emmott.

During sampling throughout the Georgina catchment in April 2007 a number of catfish (approximately 100) were sampled that were initially identified as Hyrtl’s tandan. Subsequent morphological examination of this sample suggests that these individuals may be a geographically distinct form of this species (Figure 5, 6 and 7). In the photographs below these individuals are referred to as ‘Georgina catfish’. A morphologically similar catfish was also sampled at Lake Dunn in the upper Thomson catchment in November 2008.

399

Figure 5. Georgina catfish (above) and silver tandan (below) Photo Angus Emmott.

Figure 6. Georgina catfish Photo Angus Emmott.

400 Figure 7. Head details of Georgina catfish (left) Hyrtl’s tandan (below left) and Silver tandan (below). Photos Angus Emmott.

401 Appendix 5. Hardyhead ( Craterocephalus spp.) in the Queensland Lake Eyre Basin

Abstract

A survey of fish and aquatic fauna was undertaken in the Mulligan catchment in October 2009 following the detection of a species of hardyhead in Pulchera waterhole in August 2007. Hardyhead ( Craterocephalus spp.) are known from the South Australian Lake Eyre Basin (Glover and Sim 1978; Glover 1979; Glover 1982), and may be present in a spring near Aramac, north-east of Longreach (Wager and Unmack 2000), however hardyhead have not been recorded from the Queensland Lake Eyre Basin during previous large-scale surveys (Bailey and Long 2001; Costelloe et al . 2004; Arthington et al. 2005; Kerezsy, unpublished data). The presence of hardyhead in far western Queensland suggests that source populations may exist in permanent waterholes of the Georgina catchment in the south-western corner of Queensland, or that this species undertakes upstream migrations from South Australia into Queensland waterways during major flooding.

Introduction

The Mulligan is an ephemeral watercourse on the eastern edge of the Simpson Desert that flows through Craven’s Peak, Ethabuka, Carlo, Glenormiston, Marion Downs, Sandringham and Kamaran Downs stations. The Mulligan was sampled for fish in 2007 at Pulchera waterhole (Ethabuka) and S Bend Gorge (Craven’s Peak) and seven Lake Eyre Basin species were found (A. Kerezsy, unpublished data, but see Chapters 5, 6 and 7, this thesis). Sampling events undertaken in 2007 (April, August and November) followed major flooding in the Georgina and Mulligan catchments. A local colour variant of desert rainbowfish, Melanotaenia splendida tatei , was detected in S Bend Gorge (see Chapter 5, this thesis). During opportunistic follow-up sampling of the Mulligan in August 2009 at Pulchera waterhole an eighth species was found – most probably Lake Eyre hardyhead, Craterocephalus eyresii – that has not been previously recorded from Queensland rivers of the Lake Eyre Basin (Michael Hammer, South Australian Museum, personal communication, personal observation).

402 The aquatic sampling in October 2009 aimed to investigate the waterholes and springs of the Mulligan River north of (and including) Pulchera waterhole in order to: a) Establish the northward distribution of hardyhead, and b) Locate possible source populations of fish species in the upper Mulligan.

Methods

Three waterholes were sampled during October 2009 using two large fyke nets (9mm stretched mesh, 8 metre wings), two small fyke nets (2mm mesh, 3 metre wing) and a 5 metre seine net (2mm mesh) dragged through the water for 5 minutes (Figures 1 and 2). Sites at Monkeyducka and Pulchera were set overnight, whereas at Wongitta waterhole nets were deployed for approximately 3 hours in the early afternoon. At two bores - Dribbler Bore and Currabinta Bore - a 5 metre seine net (2mm mesh) was dragged through the water (Figure 1). All sampling was carried out under General Fisheries Permit (No: PRM03315D) and under a Griffith University Ethics Agreement (AES/09/06/AEC). Sampled fish and invertebrates were held in water filled buckets prior to identification, measurement (standard length – SL) and release.

Allawonga and Pitchamurra springs on Marion Downs spring were visually surveyed for 10 minutes by two operators for macroinvertebrates and fish.

403 Monkeyducka

Currabinta Bore

Wongitta

Dribbler Bore

Pulchera

Figure 1. Sites sampled for fish and aquatic invertebrates in the upper Mulligan catchment in October 2009.

Figure 2. Waterholes sampled in the Mulligan catchment in October 2009 included Monkeyducka (left), Wongitta (centre) and Pulchera (right).

Results

Fish were sampled from both Wongitta and Pulchera waterholes in October 2009 but not from Monkeyducka (Table 1). Fish sampled from Wongitta and Pulchera included common species detected during sampling conducted in 2007 (see Chapter 5, this thesis) as well as hardyhead. (Table 1). It is notable that hardyhead were found in comparatively large numbers in Wongitta waterhole, and also that this species was present in a wide range of size classes (Table 1 and Figure 3). Adult hardyhead were

404 in breeding condition (Figure 3). Specimens of hardyhead were retained and subsequently identified as Craterocephalus eyresii by Jeff Johnson (Queensland Museum). Genetic confirmation of this identification is currently being undertaken by Mark Adams (South Australian Museum).

Desert rainbowfish were sampled from Dribbler Bore but no fish were present at Currabinta Bore (Table 1). Desert rainbowfish sampled from Dribbler Bore displayed the vivid colouration first detected in specimens of desert rainbowfish sampled from S Bend Gorge in 2007.

No fish were sampled at either Allawonga or Pitchamurra springs (Table 1). Pitchamurra spring showed evidence of heavy and/or protracted usage and associated trampling and fouling by cattle (Figure 4).

Table 1. Fish species, numbers and size ranges (in brackets) sampled from the upper Mulligan catchment in August 2009. Site Species Ambassis sp. M. s. tatei L. unicolor C. eyresii N. erebi (Glassfish) (Rainbowfish) (Spangled perch) (Hardyhead) (Bony Bream) Monkeyducka No fish Wongitta 19 (35-45mm) 7 (40 – 65mm) 2 (65 – 70mm) 603 (5 – 65mm) 257 (35 – 60mm) Pulchera 3 (33 – 45mm) 1 (50mm) 0 23 (6 – 70mm) 36 (45 – 120) Dribbler Bore 0 >100 (~25 – 70mm)* 00 0 0 Currabinta Bore No fish. Sampled fauna includes tadpoles ( Cyclorana sp.), leeches and a snail. Allawonga Spring No fish. Pitchamurra Spring No fish. Extensive damage to the spring from cattle. * exact measurements and counts not taken from Dribbler Bore rainbowfish.

405

Figure 3. Hardyhead sampled at Wongitta waterhole in October 2009. A juvenile (left), and adults in breeding condition (right).

Figure 4. The effect of cattle usage at Pitchamurra spring in November 2008 (left) and in comparison with Allawonga spring (right).

Discussion and recommendations

Fish migration into the Mulligan River is facilitated by flooding events, and therefore the most likely source populations for all species are populations from further south in Eyre Creek and the Georgina River. The junction of the Mulligan River with Eyre Creek occurs at Kalidewarry waterhole, and this permanent (but saline) waterhole is thought to be the closest source population for fish populations colonising the Mulligan (Jim Smith, Bedourie Hotel, personal communication).

In both 2007 and 2009, the large Mulligan waterhole Pulchera was colonised by 4 common species – glassfish, desert rainbowfish, bony bream and spangled perch. In

406 2007, silver tandan, Porochilus argenteus , Barcoo grunter, Scortum barcoo , and banded grunter, Amniataba percoides , were also sampled, however these species were not present in the samples taken in October 2009 (personal observation). It therefore appears likely that fish migration upstream into the Mulligan River is influenced by factors such as time of flooding and timing of the opening of migration pathways.

The presence of Lake Eyre hardyhead in both Pulchera and Wongitta waterholes in the Mulligan in 2009 (but not 2007) is a curious result, given that connection to Kalidewarry waterhole (and consequently Eyre Creek) occurred during both flood events (David Brook, Adria Downs, and Jim Smith, Bedourie Hotel, personal communications.). It therefore seems most likely that the larger flood (2009) allowed upstream colonisation by this species into the Mulligan from source populations as far away as South Australia. These colonists may have migrated from Lake Eyre itself, as the Lake filled in early 2009, and Lake Eyre hardyhead are known to breed in large numbers during flood events (Michael Hammer, SA Museum, personal communication).

The results from the Mulligan during 2009 are important with regard to the migratory abilities of Australian freshwater fish in desert systems, for they suggest that following floods, species such as hardyhead may opportunistically occupy any available habitat, including that which is >1000 kilometres upstream.

Although it is possible that the hardyhead from Pulchera and Wongitta in 2009 are a local Mulligan species, this seems unlikely, as no representatives were sampled from Pulchera during sampling undertaken in 2007 on three occasions. Furthermore, hardyhead have not been sampled from the permanent water source Dribbler Bore on Ethabuka in either 2007 or 2009. Desert rainbowfish – in contrast - have been sampled from Dribbler Bore, and it seems likely that this population represents a ‘source’ that may opportunistically colonise the upper Mulligan during connection events.

The presence of hardyhead from 5 – 70mm in both Wongitta and Pulchera waterholes indicates that this species is capable of recruitment in receding, still-water environments in the Australian arid zone.

407 The fish results from 2009 indicate that a community comprising five species (hardyhead, desert rainbowfish, spangled perch, glassfish and bony bream) almost certainly migrated into the Mulligan catchment following flooding in early 2009. The fact that no fish were sampled at the most northerly waterhole (Monkeyducka, on Glenormiston) is most likely due to the location of this waterhole and its separation from the main Mulligan channel. It is salient to remember that a sand dune represents a reasonably insurmountable obstacle for a migratory fish, and consequently there may be fish in one swale (the Mulligan), but not in adjacent swales.

The Lake Eyre hardyheads sampled from Wongitta waterhole in August 2009 represent the most northern record for this species. Future audits and surveys of other waterholes in the Mulligan River and waterholes in Eyre Creek and the lower Georgina River are now required in order to establish the range of this species in Queensland. It is recommended that this work be undertaken as soon as possible after, or perhaps during, the next major flood that occurs in the Mulligan and that sampling should be conducted at suitable sites on Craven’s Peak, Carlo, Glenormiston, Marion Downs, Ethabuka and Kamaran Downs.

The fish results from October 2009 tend to confirm the likelihood that the rainbowfish at Dribbler Bore may function as a source population that opportunistically colonise the upper Mulligan River during floods. Rainbowfish sampled from Wongitta waterhole (upstream of Dribbler Bore) were morphologically similar and exhibited vivid colouration. These results suggest that Dribbler Bore may be an important fish habitat in the Mulligan catchment.

408 References

Akihito, P. and Meguro, K. (1975). Description of a new Gobiid Fish, Glossogobius aureus , with notes on related species in the genus. Japanese Journal of Icthyology 22 , 127 – 142.

Allen, G.R., Midgley, S.H. and Allen, M. (2002). Field Guide to the Freshwater Fishes of Australia . Western Australian Museum, Perth.

Allen, M.G., Morgan, D.L. and Gill, H.S. (2005). Distribution, zoogeography and biology of the Murchison River hardyhead ( Craterocephalus cuneiceps Whitley, 1944), an atherinid endemic to the Indian Ocean (Pilbara) Drainage Division of Western Australia. Ecology of Freshwater Fish 14 , 209 – 224.

Amarasinghe, U.S. and Welcomme, R.L. (2002). An analysis of fish species richness in natural lakes. Environmental Biology of Fishes 65 , 327 – 339.

Arthington, A.H., Balcombe, S.R., Wilson, G.A., Thoms, M.C. and Marshall, J. (2005). Spatial and temporal variation in fish assemblage structure in isolated waterholes during the 2001 dry season of an arid-zone river, Cooper Creek, Australia. Marine and Freshwater Research 56, 25 - 35.

Astles, K.L., Winstanley, R.K., Harris, J.H. and Gehrke, P.C. (2003). Regulated Rivers and Fisheries Restoration Project – Experimental study of the effects of cold water pollution on native fish. NSW Fisheries Final Report Series No. 44.

Bailey, V. and Long, P. (2001). Wetland, fish and habitat survey in the Lake Eyre Basin, Queensland: final report. Queensland Department of Natural Resources and Mines, Brisbane.

Balcombe, S.R., Bunn, S.E., McKenzie-Smith, F.J. and Davies, P.M. (2005). Variability of fish diets between dry and flood periods in an arid zone floodplain river. Journal of Fish Biology 67 , 1552 – 1567.

409 Balcombe, S.R., Arthington, A.A., Foster, N.D., Thoms, M.C., Wilson, G.G. and Bunn, S.E. (2006). Fish assemblages of an Australian dryland river: abundance, assemblage structure and recruitment patterns in the Warrego River, Murray-Darling Basin. Marine and Freshwater Research 57, 619 - 633.

Balcombe, S.R., Bunn, S.E., Arthington, A.H., Fawcett, J.H., McKenzie-Smith, F.J. and Wright, A. (2007). Fish larvae, growth and biomass relationships in an Australian arid zone river: links between floodplains and waterholes. Freshwater Biology 52 , 2385 – 2398.

Balcombe, S.R. and Arthington, A.H. (2009). Temporal changes in fish abundance in response to hydrological variability in a dryland floodplain river. Marine and Freshwater Research 60 , 146 – 159.

Baumgartner, L.J. (2004). The effects of Balranald Weir on spatial and temporal distributions of lower Murrumbidgee River fish assemblages. NSW Fisheries Final Report Series No. 65.

Bertozzi, T., Adams, M. and Walker, K.F. (2000). Species boundaries in carp gudgeons (Eleotrididae: Hypseleotris) from the River Murray, South Australia: evidence for multiple species and extensive hybridisation. Marine and Freshwater Research 51 , 805 – 815.

Beumer, J.P. (1979). Reproductive cycles of two Australian freshwater fishes: the spangled perch, Therapon unicolor Gunther, 1859 and the East Queensland rainbowfish, Nematocentris splendida Peters, 1866. Journal of Fish Biology 15 , 111 – 134.

Bishop, K.A., Allen, S.A., Pollard, D.A. and Cook, M.G. (2001). Ecological studies on the freshwater fishes of the Alligator Rivers Region, Northern Territory: Autecology. Office of the Supervising Scientist Report 145, Supervising Scientist, Darwin.

410 Bond, N.R., Lake, P.S. and Arthington, A.H. (2008). The impact of drought on freshwater ecosystems: an Australian perspective. Hydrobiologica 600 , 3 – 16.

Bowen, J. (1987). Kidman: the forgotten king . Angus and Robertson Publishers.

Bowler J. (1990). The Last 500,000 Years. In: The Murray (eds. Mackay N. and Eastburn D), Murray-Darling Basin Commission, Canberra, Australia: 95 – 110.

Box, J.B., Duguid, A., Read, R.E., Kimber, R.G., Knapton, A., Davis, J. and Bowland, A.E. (2008). Central Australian waterbodies: The importance of permanence in a desert landscape. Journal of Arid Environments 72 , 1395 – 1413.

Bray J. R. and Curtis J. T. (1957). An ordination of the upland forest communities of southern Wisconsin. Ecological Monographs 27 , 325-349.

Brown, P. and Wooden, I. (2007). Age at first increment formation and validation of daily growth increments in golden perch ( Macquaria ambigua : Percichthyidae) otoliths. New Zealand Journal of Marine and Freshwater Research 41 , 157 – 161.

Budd, C. (2005). The distribution and abundance of fishes in middle Murrumbidgee River billabongs, NSW. Honours Thesis, Charles Sturt University, Wagga.

Bunn, S.E. and Arthington, A.H. (2002). Basic Principles and Ecological Consequences of altered Flow Regimes for Aquatic Biodiversity. Environmental Management 30(4), 492 – 507.

Bunn, S.E., Thoms, M.C., Hamilton, S. K. and Capon, S.J. (2006). Flow variability in dryland rivers: Boom, bust and the bits in between. River Research and Applications 22 , 179 – 186.

Bunn, S.E., Davies, P.M. and Winning, M. (2003). Sources of organic carbon supporting the food web of an arid zone floodplain river. Freshwater Biology 48 , 619 – 635.

411 Bureau of Meteorology (2008). Online at http.www.bom.gov.au/hydro/flood/flooding.shtml. Viewed December 2008.

Bureau of Meteorology (2009). Online at http.www.bom.au/climate/data/index.shtml. Viewed July 2009.

Caley, M.J., Carr, M.H., Hixon, M.A., Hughes, T.P., Jones, G.P. and Menge, B.A. (1996). Recruitment and the local dynamics of open marine populations. Annual Review of Ecology and Systematics 27 , 477 – 500.

Cadwallader, P.L. (1978). Some causes of the decline in range and abundance of native fish in the Murray-Darling River system. Proceedings of the Royal Society of Victoria 90 , 211 – 224.

Cadwallader, P.L. (1979). Distribution of native and introduced fish in the Seven Creeks River system, Victoria. Australian Journal of Ecology 4, 361 – 385.

Campana, S.E. (2001). Accuracy, precision and quality control in age determination, including a review of the use and abuse of age validation methods. Journal of Fish Biology 59 , 197 – 242.

Canonico, G.C., Arthington, A., McCrary, J.K. and Thieme, M.L. (2005). The effects of introduced tilapias on native biodiversity. Aquatic Conservation: Marine and Freshwater Ecosystems 15 , 463 – 483.

Capon, S.J. (2003). Plant community responses to wetting and drying in a large arid floodplain. River Research and Applications 19 , 509 – 520.

Capon, S.J. and Brock, M.A. (2006). Flooding, soil seed bank dynamics and vegetation resilience of a hydrologically variable desert floodplain. Freshwater Biology 51 , 206 – 223.

Capone, T.A. and Kushlan, J.A. (1991). Fish community structure in dry-season stream pools. Ecology 72(3) , 983 – 992.

Cass, A.J. and Beamish, R.J. (1983). First evidence of validity of the fin-ray method of age determination for marine fishes. North American Journal of Fisheries Management 3, 182 – 188.

412 Chen, X.Y. and Barton, C.E. (1991). Onset of aridity and dune building in central Australia: sedimentological and magnetostratigraphic evidence from Lake Amadeus, central Australia. Palaeogeography, Palaeoclimatology, Palaeoecology 84 , 55 – 73.

Clarke K. R. (1993). Non-parametric multivariate analyses of changes in community structure. Australian Journal of Ecology 18 : 117-143.

Clarke K.R. and Warwick R.M. (1994). Changes in Marine Communities: An Approach to Statistical Analysis and Interpretation. (Natural Environment Research Council: Plymouth Marine Laboratory).

Costelloe, J.F., Hudson, P.J., Pritchard, J.C., Puckridge, J.T. and Reid, J.R.W. (2004). ARIDFLOW Scientific Report: Environmental Flow Requirements of Arid Zone Rivers with Particular Reference to the Lake Eyre . School of Earth and Environmental Sciences, University of Adelaide, Adelaide. Final Report to South Australian Department of Water, Land and Biodiversity Conservation and Commonwealth Department of Environment and Heritage.

Crook, D.A. and Robertson, A.I. (1999). Relationships between riverine fish and woody debris: implications for lowland rivers. Marine and Freshwater Research 50 , 941 – 53.

Crook, D.A., Robertson, A.I. and King, A.J. (2001). The influence of spatial scale and habitat arrangement on diel patterns of habitat use by two lowland river fishes. Oecologia 129 , 525 – 533.

Cross, F.B. and Moss, R.E. (1987). Historic changes in fish communities and aquatic habitats in plains streams of Kansas. In: Community and Evolutionary Ecology of North American Stream Fishes (Eds W.J. Matthews and C.C. Heins), pp. 155 – 165, University of Oklahoma Press, Norman, OK.

413 Cucherousset, J., Carpentier, A. and Paillisson, J.-M. (2007). How do fish exploit temporary waters throughout a flooding episode? Fisheries Management and Ecology 14 , 269 – 276.

Cushing, D.H. (1975). Fisheries Resources of the Sea and Their Management. Oxford University Press.

Davey, A.J.H., Kelly, D.J. and Biggs, B.J.F. (2006). Refuge-use strategies of stream fishes in response to extreme low flows. Journal of Fish Biology 69 , 1047 – 1059.

Davey, A.J.H. and Kelly, D.J. (2007). Fish community responses to drying disturbances in an intermittent stream: a landscape perspective. Freshwater Biology 52 , 1719 – 1733.

Davis, T.L.O. (1977a). Age Determination and growth of the Freshwater Catfish, Tandanus tandanus Mitchell, in the Gwydir River, Australia. Australian Journal of Marine and Freshwater Research 28 , 119 – 137.

Davis, T.L.O. (1977b). Reproductive Biology of the Freshwater Catfish, Tandanus tandanus Mitchell, in the Gwydir River, Australia 1. Structure of the Gonads. . Australian Journal of Marine and Freshwater Research 28 , 139 – 158.

Davis, T.L.O. (1977c). Reproductive Biology of the Freshwater Catfish, Tandanus tandanus Mitchell, in the Gwydir River, Australia II. Gonadal Cycle and Fecundity. Australian Journal of Marine and Freshwater Research 28 , 159 – 169.

Dekar, M.P. and Magoulick, D.D. (2007). Factors affecting fish assemblage structure during seasonal stream drying. Ecology of Freshwater Fish 16 , 335 – 342.

414 Dudgeon, D., Arthington, A.H., Gessner, M.O., Kawabata, ZI, Knowler, D.J., Leveque, C., Naiman, R.J., Prier-Richard, AH, Soto, D., Stiassny, M.L.J. and Sullivan, C.A. (2006). Freshwater Diversity: importance, threats, status and conservation challenges. Biological Reviews 81 , 163 – 182.

Dunning, J.B., Danielson, B.J. and Pulliam, H.R. (1992). Ecological processes that affect populations in complex landscapes. Oikos 65 , 169 – 175.

Durack, M. (1959). Kings in Grass Castles . Transworld Publishers, Milsons Point.

Ebner, B. (2004). A numerical-digestion method for estimating zooplanktivory by a fish population. PhD Thesis, La Trobe University, Bundoora, Victoria.

Eby, L.A., Fagan, W.F. and Minckley, W.L. (2003). Variability and dynamics of a desert stream community. Ecological Applications 13(6), 1566 – 1579

Environment Protection and Biodiversity Conservation Act (1999). Online at http.www.environment.gov.au/epbc/. Viewed January 2010.

Fagan, W.F., Unmack, P.J., Burgess, C and Minckley, W.L. (2002). Rarity, fragmentation, and extinction risk in desert fishes. Ecology 83(12) , 3250 – 3256

Fagan, W. F., Kennedy, C.M. and Unmack, P.J. (2005). Quantifying rarity, losses, and risks for native fishes of the lower Colorado River Basin: Implications for Conservation Listing. Conservation Biology 19(6), 1872 – 1882.

Fairfax, R., Fensham, R., Wager, R., Brooks, S., Webb, A. and Unmack, P. (2007). Recovery of the red-finned blue-eye: an endangered fish from springs of the Great Artesian Basin. Wildlife Research 34 , 156 – 166.

Faragher, R.A. and Harris, J.H. (1994). The Historical and Current Status of Freshwater Fish in New South Wales. Australian Zoologist 29 , 166 – 176.

415 Faragher, R.A. and Lintermans, M. (1997). Alien fish species from the New South Wales Rivers Survey. In: Fish and Rivers in Stress – the NSW Rivers Survey (eds. Harris J.H. and Gehrke P.C.). NSW Fisheries Office of Conservation and the Cooperative Research Centre for Freshwater Ecology, Cronulla and Canberra: pp. 201 – 224.

Faragher, R.A. and Rodgers, M. (1997). Performance of sampling-gear types in the New South Wales Rivers Survey. In: Fish and Rivers in Stress – the NSW Rivers Survey (eds. Harris J.H. and Gehrke P.C.). NSW Fisheries Office of Conservation and the Cooperative Research Centre for Freshwater Ecology, Cronulla and Canberra: pp. 251 - 268.

Faulks, L. (2009). Evolutionary history, landscape genetics and conservation of two Australian freshwater fishes golden perch ( Macquaria ambigua ) and Macquarie perch ( Macquaria australasica ). PhD Thesis, Macquarie University, Sydney.

Fausch, K.D. and Bramblett, R.G. (1991). Disturbance and Fish Communities in Intermittent Tributaries of a Western Great Plains River. Copeia 3, 659 – 674.

Fausch, K.D., Torgersen, C.E., Baxter, C.V. and Li, H.W. (2002). Landscapes to riverscapes: bridging the gap between research and conservation of stream fishes. Bioscience 52 , 483 – 498.

Folkvard, A. (1997). Ontogeny of cannibalism in larval and juvenile fishes with special emphasis on Atlantic cod. In Early Life History and Recruitment in Fish Populations , pp. 252 – 278, Chambers, R.C. and Trippel, E.A. (eds.), Chapman and Hall, London.

Freilich, J.E., Burnham, K.P., Collins, C.M. and Garry, C. A. (2000). Factors affecting population assessments of desert tortoises. Conservation Biology 14(5), 1479 – 1489.

416 Gardner, G.J., Mortlock, A.J., Price, D.M., Readhead, M.L. and Wasson, R.J. (1987). Thermoluminescense and radiocarbon dating of Australian desert dunes. Australian Journal of Earth Sciences 34 , 343 – 357.

Gehrke, P.C., Schiller, C.B. and Brown, P. (1999). Native fish and river flows: the Paroo perspective. In A Free-Flowing River: The Ecology of the . Kingsford, R.T. (ed). National Parks and Wildlife Service, Sydney, pp. 77 – 96.

Georges, A., Guarino, F. and White, M. (2006). Sex-ratio bias across populations of a freshwater turtle (Testudines: Chelidae) with genotypic sex determination. Wildlife Research 33 , 475 – 480.

Gilligan, D. and Schiller, C. (2003). Downstream transport of larval and juvenile fish in the . NSW Fisheries Final Report Series No. 50, NSW Fisheries, Narrandera.

Glover, C.J.M. (1979). Studies on central Australian fishes: further observations and records, Part I. South Australian Naturalist 53 , 58 – 62.

Glover, C.J.M. (1982). Adaptations of fishes in arid Australia. In Evolution of the Flora and Fauna of Arid Australia , pp. 241 – 246, Barker, W.R. and Greenslade, P.J.M. (eds.) Peacock Publications, South Australia.

Glover, C.J.M. and Sim, T.C. (1978). A survey of central Australian ichthyology. Australian Journal of Zoology 15 , 61 – 64.

Godinho, A.L. and Kynard, B. (2006). Migration and spawning of radio-tagged zulega, Prochilodus argenteus , in a dammed Brazilian river. Transactions of the American Fisheries Society 135(3) , 811 – 824.

Gotelli, N.J. and Taylor, C.M. (1999). Testing metapopulation models with stream- fish assemblages. Evolutionary Ecology Research 1, 835 – 845.

417 Gowan, C., Young, M.K., Fausch, K.D. and Riley, S.C. (1994). Restricted movement in resident stream salmonids: a paradigm Lost? Canadian Journal of Fisheries and Aquatic Sciences 51 , 2626 – 2637.

Harris J.H. and Gehrke P.C. (1997). Fish and Rivers in Stress – the NSW Rivers Survey . NSW Fisheries Office of Conservation and the Cooperative Research Centre for Freshwater Ecology, Cronulla and Canberra.

Hamilton, S.K., Bunn, S.E., Thoms, M.C. and Marshall, J.C. (2005). Persistence of aquatic refugia between flow pulses in a dryland river system (Cooper Creek, Australia). Limnology and Oceanography 50(3), 743 – 754.

Hammer, M.P., Adams, M, Unmack, P.J. and Walker, K.F. (2007). A rethink on Retropinna : conservation implications of new taxa and significant genetic sub- structure in Australian smelts (Pisces: Retropinnidae). Marine and Freshwater Research 58(4) , 327 - 341.

Hanski, I.A. and Gilpin, M.E. (1997). Metapopulation Biology: Ecology, Genetics, and Evolution . Academic Press.

Harrison, S. and Hastings, A. (1996). Genetic and evolutionary consequences of metapopulation structure. Trends in Ecology and Evolution 11 , 180 – 183.

Haythornthwaite, A.S. and Dickman, C.R. (2006). Long-distance movements by a small carnivorous marsupial: how Sminthopsis youngsoni (Marsupialia:Dasyuridae) uses habitat in an Australian sandridge desert. Journal of Zoology 270 , 543 – 549.

Howe, E., Howe, C., Lim, R. and Burchett, M. (1997). Impact of the introduced poeciliid Gambusia holbrooki (Girard, 1859) on the growth and reproduction of Pseudomugil signifier (Kner, 1865) in Australia. Marine and Freshwater Research 48 , 425 - 434.

418 Huey, J.A., Hughes, J.M. and Baker, A.M. (2006). Patterns of gene flow in two species of eel-tailed catfish, Neosiluris hyrtlii and Porochilus argenteus (Siluriformes: Plotosidae), in western Queensland’s dryland rivers. Biological Journal of the Linnean Society 87 , 457 – 467.

Hughes, D.A. (2005). Hydrological issues associated with the determination of environmental water requirements of ephemeral rivers. River Research and Applications 21 , 899 – 908.

Hughes, J., Baker, A.M., Bartlett, C., Bunn, S, Goudkamp, K. and Somerville, J. (2004). Past and present patterns of connectivity among populations of four cryptic species of freshwater mussels Velesunio spp. (Hyriidae) in central Australia. Molecular Ecology 13 , 3197 – 3212.

Hughes, J.M. and Hillyer, M.J. (2003). Patterns of connectivity among populations of Cherax destructor (Decapoda: Parastacidae) in western Queensland, Australia. Marine and Freshwater Research 54 : 587 - 596.

Hughes, J., Ponniah, M., Hurwood, D., Chenoweth, S. and Arthington, A. (1999). Strong genetic structuring in a habitat specialist, the Oxleyan pygmy perch, Nannoperca oxleyana . Heredity 83 , 5 – 14.

Humphries, P. (2005). Spawning time and early life history of Murray cod, Maccullochella peelii peelii (Mitchell) in an Australian river. Environmental Biology of Fishes 72 , 393 – 407.

Humphries P., King A.J. and Koehn J.D. (1999). Fish, flows and flood plains: Links between freshwater fishes and their environment in the Murray-Darling River system, Australia. Environmental Biology of Fishes 56 , 129 – 151.

Humphries, P., Serafini, L.G., and King, A.J. (2002). River regulation and fish larvae: variation through space and time. Freshwater Biology 47 , 1307 – 1331.

419 Humphries, P. and Winemiller, K.O. (2009). Historical Impacts on River Fauna, Shifting Baselines, and Challenges for Restoration. BioScience 59 , 673 – 684.

International Union for the Conservation of Nature (2010). Online at http.www.iucnredlist.org. Viewed January 2010.

Ivantsoff, W. and Aarn (1999). Detection of predation of Australian native fishes by Gambusia holbrooki. Marine and Freshwater Research 50, 467 - 468.

Jobling, M. (1995). Environmental Biology of Fishes . Chapman and Hall, London.

Junk, W.J., Bayley, P.B. and Sparks, R.E. (1989). The flood pulse concept in river- floodplain systems. Canadian Special Publication Fisheries and Aquatic Science 106 , 110 – 127.

Kennard, M.J. (1995). Factors influencing freshwater fish assemblages in floodplain lagoons of the Normanby River, : a large tropical Australian river. M. Phil. Thesis, Griffith University, Brisbane.

Kerezsy, A. (2005). The distribution and abundance of fish in the Lake Cargelligo system, New South Wales. Honours thesis, Charles Sturt University, Wagga.

Kerezsy, A. (2009). Gambusia control in spring wetlands. South Australian Arid Lands Natural Resources Management Board.

Kesteven, G.L. (1960). Manual of Field Methods in Fisheries Biology . FAO Manual in Fisheries Sciences. No. 1. FAO, Rome.

King, A.J. and Crook, D.A. (2002). Evaluation of a sweep-net electrofishing method for the collection of small fish and shrimp in lotic environments. Hydrobiologia 472 , 223 – 233.

420 King, A.J., Humphries, P. and Lake, P.S. (2003). Fish recruitment on floodplains: the roles of patterns of flooding and life history characteristics. Canadian Journal of Fisheries and Aquatic Sciences 60 , 773 – 786.

Kingsford, R.T., Georges, A. and Unmack, P.J. (2006a). Vertebrates of desert rivers: meeting the challenges of temporal and spatial unpredictability. In Ecology of Desert Rivers, pp. 154 - 200. Kingsford, R.T. (ed.) Cambridge University Press.

Kingsford, R.T., Lemly, A.D. and Thompson, J.R. (2006b). Impacts of dams, river management and diversions on desert rivers. In Ecology of Desert Rivers, pp . 203 - 247. Kingsford, R.T. (ed.) Cambridge University Press.

Kingsford, R.T. and Thompson, J.R. (2006). Desert or dryland rivers of the world: an introduction. In Ecology of Desert Rivers, pp. 3 - 10. Kingsford, R.T. (ed.) Cambridge University Press.

Knight, J.T. and Arthington, A.H. (2008). Distribution and habitat associations of the endangered Oxleyan pygmy perch, Nannoperca oxleyana Whitley, in eastern Australia. Aquatic Conservation: Marine and Freshwater Ecosystems 18 , 1240 – 1254.

Knighton, D.A. and Nansen, G.C. (1994). Flow transmission along an arid zone anastomosing river, Cooper Creek, Australia. Hydrological Processes 8, 137 – 154.

Koehn, J.D. (1996). Habitats and movements of freshwater fish in the Murray-Darling Basin. In Proceedings of the 1995 Riverine Research Forum . Murray-Darling Basin Commission, Canberra: pp. 27 – 32.

Koehn, J.D. (2004). Carp ( Cyprinus carpio ) as a powerful invader in Australian waterways. Freshwater Biology 49 : 882 – 894.

421 Koehn, J.D. (2006). The ecology and conservation management of Murray cod, Maccullochella peelii peelii . PhD Thesis, University of Melbourne.

Koehn, J. D. and Mackenzie R.F. (2004). Priority management actions for alien freshwater fish species in Australia. New Zealand Journal of Marine and Freshwater Research 38 , 457 – 472.

Kodric-Brown, A., Wilcox, C., Bragg, J. G. and Brown J.H. (2007). Dynamics of fish in Australian desert springs: role of large mammal disturbance. Diversity and Distributions 13 , 789 – 798.

Kozakiewicz, M. and Szacki, J. (1995). Movements of small mammals in a landscape: patch restriction or nomadism? In Landscape approaches in mammalian ecology and conservation : pp. 78 – 94. Lidicker, W.Z. (ed.) Minneapolis: University of Minnesota.

Labbe, T.R. and Fausch, K.D. (2000). Dynamics of intermittent stream habitat regulate persistence of threatened fish at multiple scales. Ecological Applications 10(6) , 1774 – 1791.

Lake Eyre Basin Co-ordinating Group (2000). Georgina/Diamantina catchment Strategic Plan.

Lake, J.S. (1967). Rearing experiments with five species of Australian freshwater fishes. Australian Journal of Marine and Freshwater Research 18, 137 - 173.

Lake, P.S. (2000). Disturbance, patchiness and diversity in streams. Journal of the North American Benthological Society 19(4) , 573 – 592.

Lake P.S. (2001) On the maturing of restoration: Linking ecological research and restoration. Ecological Management & Restoration 2(2) , 110 – 115.

Lake, P.S. (2003). Ecological effects of perturbation by drought in flowing waters. Freshwater Biology 48 , 1161 – 1172.

422

Larned, S.T., Datry, T, Arscott, D.B. and Tockner, K. (2009). Emerging concepts in temporary-river ecology. Freshwater Biology , doi:10.1111/j.1365- 2427.2009.02322.x

Larson, H.K. and Martin, K.C. (1989). Freshwater Fishes of the Northern Territory . Northern Territory Museum of Arts and Sciences, Darwin.

Larson, H.K. and Hoese, D.F. (1996). Gudgeons. In McDowall R.M. (ed.) Freshwater Fishes of South-eastern Australia . Reed, Sydney: pp. 200 – 219.

Leggett, R. and Merrick, J.R. (1987). Australian Native Fishes for Aquariums . Griffin Press Limited, Netley, South Australia.

Leigh, C. (2008). Floodplain river function in Australia’s wet/dry tropics, with specific reference to aquatic macroinvertebrates and the Gulf of Carpentaria. PhD Thesis, Griffith University, Brisbane.

Leigh, C. and Sheldon, F. (2008). Hydrological changes and ecological impacts associated with water resource development in large floodplain rivers in the Australian tropics. River Research and Applications 24(9) , 1251 – 1270.

Levins, R. (1969). Some demographic and genetic consequences of environmental heterogeneity for biological control. Bulletin of the Entomological Society of America 15 , 237 – 240.

Lintermans, M. (2007). Fishes of the Murray-Darling Basin: An Introductory Guide . MDBC Publication No. 10/07.

Llewellyn, L.C. (1973). Spawning, development, and temperature tolerance of the spangled perch, Madigania unicolor (Gunther), from inland waters in Australia. Australian Journal of Marine and Freshwater Research 24 , 73 - 94.

423 Llewellyn, L.C. (1983). The distribution of fish in New South Wales . Australian Society for Limnology Special Publication No.7.

Long, P.E. and Humphery, V.E. (1995). Fisheries Study Lake Eyre Catchment – Thomson and Diamantina Drainages December 1995 . Department of Primary Industries, Brisbane.

Love, J.W., Taylor, C.M. and Warren Jr, M.P. (2008). Effects of Summer Drought on Fish and Macroinvertebrate Assemblage Properties in Upland Ouachita Mountain Streams, USA. The American Midland Naturalist 160 , 265 – 277.

Lovich, J. and Meyer, K. (2002). The western pond turtle ( Clemmys marmorata ) in the Mojave River, California USA: highly adapted survivor or tenuous relict? Journal of Zoology 256 (4) , 537 – 545.

McDowall R.M. (Ed.) (1996). Freshwater Fishes of South-eastern Australia , (2 nd edition) Reed, Sydney. 247 p.

McMahon, T.A. and Finlayson, B.L. (2003). Droughts and anti-droughts: the low flow hydrology of Australian rivers. Freshwater Biology 48 , 1147 – 1160.

McMahon, T.A, Murphy, R., Little, P., Costelloe, J.F., Peel, M.C., Chiew, F.H.S., Hayes, S., Nathan, R. and Kandel, D.D. (2005). Hydrology of Lake Eyre Basin . Sinclair Knight Merz, Armadale.

MacArthur, R.H. and Wilson, E.O. (1967). The Theory of Island Biogeography . Princeton University Press.

Mackay, N. J. (1973). Histological changes in the ovaries of the golden perch, Plectroplites ambiguous, associated with the reproductive cycle. Australian Journal of Marine and Freshwater Research 24 , 95 - 101.

Madigan, C.T. (1946). Crossing the Dead Heart . Georgian House, Melbourne.

424 Magoulick, D.D. and Kobza, R.M. (2003). The role of refugia for fish during drought: a review and synthesis. Freshwater Biology 48 , 1186 – 1198.

Mallen-Cooper, M. and Stuart, I.G. (2003). Age, growth and non-flood recruitment of two potamodromous fishes in a large semi-arid/temperate river system. River Research and Applications 19 , 697 – 719.

Mallen-Cooper, M., Stuart, I.G., Hides-Pearson, F. and Harris, J.H. (1995) Fish Migration in the Murray River and Assessment of the Torrumbarry Fishway. NSW Fisheries Research Institute and the Cooperative Research Centre for Freshwater Ecology . Final report for Natural Resources Management Strategy Project N002.

Maroulis, J.C., Nanson, G.C., Price, D.M. and Pietsch, T. (2007). Aeolian-fluvial interaction and climate change: source-bordering dune development over the past ~100ka on Cooper Creek, central Australia. Quaternary Science Reviews 26 , 386 – 404.

Marshall, J.C., Sheldon, F., Thoms, M. and Choy, S. (2006). The macroinvertebrate fauna of an Australian dryland river: spatial and temporal patterns and environmental relationships. Marine and Freshwater Research 57 , 61 - 74.

Matthews, W.J. and Marsh-Matthews, E. (2003). Effects of drought on fish across axes of space, time and ecological complexity. Freshwater Biology 48 , 1232 – 1253.

Medeiros, E.S.F. (2004). Trophic ecology and energy sources for fish on the floodplain of a regulated dryland river: Macintyre River, Australia. Ph.D. Thesis, Griffith University, Queensland.

Meredith, S., Gawne, B., Sharpe, C., Whiterod, N., Conallin, A. and Zukowski, S. (2002). Dryland Floodplain Ecosystems: influence of flow pattern on fish production. Murray-Darling Freshwater Research Centre, Mildura.

425 Merrick, J.R. and Schmida, G.E. (1984). Australian Freshwater Fishes: Biology and Management . John R. Merrick, North Ryde. 409 p.

Midgley, S.H., Midgley, M. and Rowland, S.J. (1991). Fishes of the Bulloo- Bancannia drainage division. Memoirs of the Queensland Museum 30 , 505 – 508.

Milton, D.A. and Arthington, A.H. (1984). Reproductive strategy and growth of the crimson-spotted rainbowfish, Melanotaenia splendida fluviatilis (Castelnau) (Pisces: Melanotaeniidae) in south-eastern Queensland. Australian Journal of Marine and Freshwater Research 35 , 75 – 83.

Milton, D.A. and Arthington, A.H. (1985). Reproductive strategy and growth of the Australian smelt, Retropinna semoni (Weber) (Pisces: Retropinnidae), and the Olive perchlet, Ambassis nigripinnis (DeVis) (Pisces: Ambassidae), in Brisbane, south-eastern Queensland. Australian Journal of Marine and Freshwater Research 36, 329 - 341.

Murgatroyd, S. (2002). The Dig Tree: the story of Burke and Wills . The Text Publishing House, Melbourne.

Murray-Darling Basin Ministerial Council (2003). Native Fish strategy for the Murray-Darling Basin 2003 – 2013. MDBC Publication No. 25/04.

Murray-Darling Basin Commission (2004). Fish Theme Pilot Audit Technical Report – Sustainable Rivers Audit . MDBC Publication No. 06/04.

Murray-Darling Basin Commission (2008). Murray-Darling Basin Rivers: Ecosystem Health Check, 2004 – 2007. A summary report based on the Independent Sustainable Rivers Audit Group’s SRA Report 1: A Report on the Ecological Health of Rivers in the Murray-Darling Basin, 2004 – 2007. Canberra, Australia.

Musyl, M.K. and Keenan, C.P. (1992). Population genetics and zoogeography of Australian freshwater golden perch, Macquaria ambigua (Richardson

426 1845)(Teleostei: Percichthyidae), and electrphoretic identification of a new species from the Lake Eyre Basin. Australian Journal of Marine and Freshwater Research 43, 1585 - 1601.

Nanson, G.C., Price, D.M., Jones, B.G., Maroulis, J.C., Coleman, M, Bowman, H., Cohen, T.J., Pietsch, T.J. and Larsen, J.R. (2008). Alluvial evidence for major climate and flow regime changes during the middle and late Quaternary in eastern central Australia. Geomorphology 101 , 109 – 129.

Nature Conservation Act (1992). Online at http.www.legislation.qld.gov.au/ LEGISLTN/CURRENT/N/NatureConA92.pdf

Nikolsky, G.V. (1963). The Ecology of Fishes . Academic Press, London and New York.

Nunn, A.D., Cowx, I.G. and Harvey, J.P. (2002). Recruitment patterns of six species of cyprinid fishes in the lower River Trent, England. Ecology of Freshwater Fish 11 , 74 – 84.

Orr, T.M. and Milward, N.E. (1984). Reproduction and development of Neosiluris ater Perugia and Neosiluris hyrtlii Steindachner (Teleostei: Plotosidae) in a tropical Queensland stream. Australian Journal of Marine and Freshwater Research 35 , 187 - 195.

Peterson, K. (2003). Environmental Impacts on Spawning and Survival of Fish Larvae and Juveniles in an Upland River System of the Murray-Darling Basin. PhD Thesis, University of Canberra.

Poff, N.L. and Allan, J.D. (1995). Functional organisation of stream fish assemblages in relation to hydrological variability. Ecology 76(2) , 606 – 627.

Pollard, D.A. (1972). The biology of a landlocked form of the normally catadromous salmoniform fish Galaxias maculatus (Jenyns) III. Structure of the gonads. Australian Journal of Marine and Freshwater Research 23 , 17 – 38.

427

Pritchard, J.C. (2004). Linking fish growth and climate across modern space and through evolutionary time. Otolith chronologies of the Australian freshwater fish, golden perch ( Macquaria ambigua , Percichthyidae). Ph.D. thesis, Australian National University, Canberra.

Propst, D.L., Gido, K.B. and Stefferud, J.A. (2008). Natural flow regimes, nonnative fishes and native fish persistence in arid-land river systems. Ecological Applications 18(5) , 1236 – 1252.

Puckridge, J.T. (1999). The role of hydrology in the ecology of Cooper Creek, Central Australia: implications for the flood pulse concept. Ph.D. Thesis, The University of Adelaide.

Puckridge, J.T., Sheldon, F., Walker, K.F. and Boulton, A.J. (1998). Flow variability and the ecology of large rivers. Australian Journal of Marine and Freshwater Research 49, 55 - 72.

Puckridge, J.T. and Walker, K.F. (1990). Reproductive Biology and Larval Development of a Gizzard Shad, Nematolosa erebi (Gunter)(Dorosomatinae:Teleostei), in the River Murray, South Australia. Australian Journal of Marine and Freshwater Research 41 , 695 – 712.

Puckridge, J.T., Walker, K.F. and Costelloe, J.F. (2000). Hydrological persistence and the ecology of dryland rivers. Regulated Rivers-Research and Management 16 , 385 – 402.

Pulliam, H.R. (1996). Sources, sinks and population regulation . American Naturalist 132 , 652 – 661.

Pusey, B., Kennard, M. and Arthington, A. (2004). Freshwater Fishes of North- Eastern Australia . CSIRO Publishing, Collingwood, Victoria. 684 p.

428 Quammen, D. (1996). The Song of the Dodo, Island biogeography in an age of extinctions , Scribner, USA.

Rahel, F.J. (2007). Biogeographic barriers, connectivity and homogenization of freshwater faunas: it’s a small world after all. Freshwater Biology 52 , 696 – 710.

Rayner, T.S., Jenkins, K.M. and Kingsford, R.T. (2009). Small environmental flows, drought and the role of refugia for freshwater fish in the Macquarie Marshes, arid Australia. Ecohydrology 2, 440 – 453.

Read, J.L. (2003). Red sand green heart: Ecological adventures in the outback. Thomas C. Lothian Pty Ltd, South Melbourne.

Reid, D.D., Harris, J.H. and Chapman D.J. (1997). NSW inland commercial fishery data analysis . Fisheries Research and Development Corporation Project No. 94/027 Report.

Resh, V.H., Brown, A.V., Covich, A.P., Gurtz, M.E., Li, H.W., Minshall, G.W., Reice, S.R., Sheldon, A.L., Wallace, J.B. and Wissmar, R. (1988). The role of disturbance in stream ecology. Journal of the North American Benthological Scoiety 7 , 433 – 455.

Reynolds L. F. (1983). Migration Patterns of Five Fish Species in the Murray-Darling River System. Australian Journal of Marine and Freshwater Research 34 , 857 – 71.

Rice, J.A., Crowder, L.B. and Marschall, E.A. (1997). Predation on juvenile fishes: dynamic interactions between size-structured predators and prey. In Early Life History and Recruitment in Fish Populations , pp. 333 – 356, Chambers, R.C. and Trippel, E.A. (eds.), Chapman and Hall, London.

Righton, D., Quayle, V.A., Hetherington, S. and Burt, G. (2007). Movements and distribution of cod, Gadus morhua , in the Southern North Sea and English

429 Channel. Results from conventional and electronic tagging experiments. Journal of the Marine Biological Association of the United Kingdom 87(2) , 599 – 613.

Roberts, D.T., Duivenvoorden, L.J. and Stuart, I.G. (2008). Factors influencing recruitment patterns of Golden Perch ( Macquaria ambigua oriens ) within a hydrologically variable and regulated Australian tropical river system. Ecology of Freshwater Fish 17 , 577 – 589.

Roberts, J., Chick, A., Oswald, L. and Thompson, P. (1995). Effects of Carp, Cyprinus carpio L., an Exotic Benthivorous Fish, on Aquatic Plants and Water Quality in Experimental . Marine and Freshwater Research 46 , 1171 – 1180.

Robertson, A.L. (2000). Lotic meiofaunal community dynamics: colonisation, resilience and persistence in a spatially and temporally heterogeneous environment. Freshwater Biology 44 , 135 – 147.

Robillard, S.R. and Marsden, J.E. (1996). Comparison of otolith and scale ages for yellow perch from Lake Michigan. Journal of Great Lakes Research 22 , 429 – 435.

Ruello, N.V. (1976). Observations on some massive fish kills in Lake Eyre. Australian Journal of Marine and Freshwater Research 27 , 667 - 672.

Scheurer, J.A. and Fausch, K.D. (2003). Multiscale Processes Regulate Brassy Minnow Persistence in a Great Plains River. Transactions of the American Fisheries Society 132 , 840 – 855.

Schlosser, I.J. (1987). A conceptual framework for fish communities in small warmwater streams. In Community and evolutionary ecology of North American stream fishes, 17 - 24. Matthews, W.J. and Heins, D.C. (eds.) University of Oklahoma Press.

430 Schlosser, I.J. and Angermeier, P.L. (1995). Spatial variation in demographic processes of lotic fishes: conceptual models, empirical evidence, and implications for conservation. American Fisheries Society Symposium 17 , 392 – 401.

Secor, D.H., Dean, J.M. and Campana, S.E. (ed.)(1995). Recent Developments in Fish Otolith Research. University of South Carolina Press, Columbia, South Carolina.

Senior, B.R. and Mabbutt, J.A. (1979). A proposed method of defining deeply weathered rock units based on regional geological mapping in southwest Queensland. Journal of the Geological Society of Australia 46 , 237 – 254.

Serafini, L.G. and Humphries P. (2003). Preliminary Guide to the Identification of Larvae of Fish from the Murray-Darling Basin . Cooperative Research Centre for Freshwater Ecology, Murray-Darling Freshwater Research Centre, Albury.

Sheldon, F. (2005). Incorporating natural variability into the assessment of ecological health in Australian dryland rivers. Hydrobiologia 552 , 45 – 56.

Sheldon, F., Bunn, S.E., Hughes, J.M., Arthington, A.H., Balcombe, S.R. and Fellows, C.S. in press . Dryland river waterholes: A review of the ecological roles and threats to aquatic refugia in arid landscapes . Marine and Freshwater Research

Sheldon, F. and Fellows, C. in press . Water quality and water chemistry in two Australian dryland rivers: spatial and temporal variability and the role of flow. Marine and Freshwater Research

Sheldon, F. and Thoms, M.C. (2006). Relationships between flow variability and macroinvertebrate assemblage composition: Data from four Australian dryland rivers. River Research and Applications 22 , 219 – 238.

Sheldon, F. and Walker, K. (1998). Spatial distribution of littoral invertebrates in the

431 lower Murray-Darling River system, Australia. Marine and Freshwater Research 49 , 171 - 182.

Shephard, M. (1992). The Simpson Desert: Natural history and human endeavour . Reed books, Chatswood.

Silcock, J. (2009). Identification of Permanent Refuge Waterbodies in the Cooper Creek and Georgina-Diamantina River Catchments for Queensland and South Australia. South Australian Arid Lands Natural Resources Management Board.

Stromberg, J.C. (2007). Seasonal reversals of upland-riparian diversity gradients in the Sonoran Desert. Diversity and Distributions 13 , 70 – 83.

Stuart, I.G. and Berghuis, A.P. (2002). Upstream passage of fish through a vertical- slot fishway in an Australian subtropical river. Fisheries Management and Ecology 9, 111 – 122.

Stuart, I.G. and Jones, M. (2006). Large, regulated forest floodplain is an ideal recruitment zone for non-native common carp ( Cyprinus carpio L.) Marine and Freshwater Research 57 , 333 - 347.

Thacker, C. and Unmack P.J. (2005) Phylogeny and biogeography of the eleotrid genus Hypseleotris (Teleostei: Gobioidei: Eleotridae). Records of the Australian Museum 57 , 1-13.

Thoms-Souza, M.J.F. and Chao, N.L. (2004). Spatial and temporal variation of benthic fish assemblages during the extreme drought of 1997-98 (El Nino) in the middle rio Negro, Amazonia, Brazil. Neotropical Icthyology 2(3) , 127 – 136.

Thorp, J.H. and Delong, M.D. (1994). The riverine productivity model: An heuristic view of carbon sources and organic-processing in large river ecosystems. Oikos 70 , 305 – 308.

432

Thorp, J.H. and Delong, M.D. (2002). Dominance of autochthonous autotrophic carbon in food webs of heterotrophic rivers. Oikos 96(3) , 543 – 550.

Unmack, P.J. (1996). The unique Cooper Creek catfish from central Australia. Fishes of Sahul 10(2), 460 – 464.

Unmack, P.J. (2000). The genus Hypseleotris in southeastern Australia: its identification and breeding biology. Fishes of Sahul 14 , 645 – 657.

Unmack, P.J. (2001). Biogeography of Australian freshwater fishes. Journal of Biogeography 28 , 1053 – 1089.

Vannote, R.L., Minshall, G.W., Cummins, K.W., Sedell, K.W. and Cushing, C.E. (1980). The river continuum concept. Canadian Journal of Fisheries and Aquatic Science 37 , 130 – 137.

Wager, R. (1994). The distribution of two endangered fish in Queensland. Endangered Species Unit Project Number 276. Final Report Part B: The distribution and status of the red-finned blue-eye. Australian Nature Conservation Agency, Canberra.

Wager, R. and Unmack, P.J. (2000). Fishes of the Lake Eyre catchment in central Australia. Queensland Department of Primary Industries, Brisbane.

Walker, K.F., Sheldon, F. and Puckridge, J.T. (1995). A Perspective on Dryland River Ecosystems. Regulated Rivers: Research and Management 11 , 85 – 104.

Welcomme, R.L. (1985). River Fisheries. Food and Agriculture Organisation of the United Nations. FAO Fisheries Technical Paper No. 262, 303 pp. Rome, Italy.

Winemiller, K.O. (1989). Patterns of variation in life history among South American fishes in seasonal environments. Oecologia 81 , 225 – 241.

433 Wopfner, H. (1963). Post-Winton sediments of probable Upper Cretaceous age in the Great Artesian Basin. Transactions of the Royal Society of South Australia 86 , 247 – 253.

Zeug, S.C. and Winemiller, K.O. (2008). Relationships between hydrology, spatial heterogeneity, and fish recruitment dynamics in a temperate floodplain river. River Research and Applications 24 , 90 – 102.

434