Environmental Impacts on Spawning and Survival of Fish Larvae and Juveniles in an Upland River System of the Murray-Darling Basin

Transcription

1 Environmental Impacts on Spawning and Survival of Fish Larvae and Juveniles in an Upland River System of the Murray-Darling Basin Larvae of carp gudgeon (Hypseleotris sp.), 12 mm (Illustrated by F. J. Neira in Neira et al., 1998, p. 385) Kylie Peterson B. Sc. Hons. Applied Ecology Research Group Cooperative Research Centre for Freshwater Ecology University of Canberra A thesis submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy. November, 2003.

2 Statement of originalitv This thesis is my original work and has not been submitted, in whole or in part, for degree at this or any other university. Nor does it contain, to the best of my knowledge and belief, any material published or written by another person, except as acknowledged in the text. Signature of Date 20 mawfl z-?2...

3 Copyright in relation to this thesis: Under Section 35 of the Copyright Act of 1968, the author of this thesis is the owner of any copyright subsisting in the work, even though it is unpublished. Under Section 3 l(i)(a)(i), copyright includes the exclusive right to 'reproduce the work in a material form'. Thus, copyright is infringed by a person who, not being the owner of the copyright, reproduces or authorises the reproduction of the work, or of more than a reasonable part of the work, in a material form, unless the reproduction is a 'fair dealing' with the work 'for the purpose of research or study' as further defined in Sections 40 and 41 of the Act. This thesis must therefore be copied or used only under the normal conditions of scholarly fair dealing for the purposes of research, criticism or review, as outlined in the provisions of the Copyright Act In particular, no results or conclusions should be extracted from it, nor should it be copied or closely paraphrased in whole or in part without the written consent of the author. Proper written acknowledgement should be made for any assistance obtained from this thesis. Copies of the thesis may be made by a library on behalf of another person provided the officer in charge of the library is satisfied that the copy is being made for the purposes of research or study.

4 Acknowledgments I would like to thank the following; My supervisor, Professor Robert Kearney, for endless support and advice, believing in the research and teaching me to be positive - I will always be indebted. Associate Professor Martin Thoms, for making constructive and helphl comments on several chapters in this thesis. Dr Paul Hurnphries, for much needed advice regarding sampling techniques and larval identification. The three anonymous examiners of this thesis, for providing constructive feedback regarding ways to improve the final product. The many private landowners, who allowed me access to the rivers studied. Special thanks to the Chisholm, Oliver, Parker and Tait families. Mr Craig Richardson, for his time and patience when generating the maps used in this thesis. Mrs Louise Amos, for assistance in gaining financial support for this study. Mrs Stephanie Wells, for proof reading several chapters of this thesis. ACT Fisheries Action Program, which contributed funding for this study. Applied Ecology Research Group and the Cooperative Research Centre for Freshwater Ecology for supporting this research. My managers at Environment Australia; Martin, Theo and Mark, for allowing me time off to complete this thesis. My PhD 'cohort'; Cathy, Lisa, Melissa and Patrick, for all your advice and support and for just understanding the many stages of a PhD. Special thanks to Cath, for proof reading the majority of this thesis. My fellow members of the 'Dam Fish Section', Sal and Pete, for keeping me sane. My family and friends, who have often taken a back seat to this research, my thanks and apologies. Alex, for supporting and encouraging me every step of the way and for making many important improvements to this work.

5 Table of Contents Abstract List of Figures List of Tables List of Appendices 1. General Introduction Status of native fish communities in the Murray-Darling Basin Fish larvae and juveniles - why are they important ix xi xvi xvii State of knowledge on spawning and recruitment of fish in the Murray-Darling Basin Effects of river regulation and their impact on spawning and recruitment The usefulness of fish larvae as indicators Identification of the knowledge gap Specific objectives of this research 2. Pilot study 2.1 Introduction 2.2 Study area 2.3 Methods 2.4 Results 2.5 Discussion 2.6 Conclusions and suggestions for next phase

6 3. Main study 3.1 Introduction 3.2 Methods 3.3 Results 3.4 Discussion 3.5 Conclusions and focus of upcoming chapters 4. Mountain galaxias (Galaxias olidus) 4.1 Introduction 4.2 Methods 4.3 Results 4.4 Discussion 5. Carp (Cyprinus carpio) 5.1 Introduction 5.2 Methods 5.3 Results 5.4 Discussion 6. Redfin perch (Percafluviatilis) 6.1 Introduction 6.2 Methods 6.3 Results 6.4 Discussion

7 vii 7. Australian smelt, western carp gudgeon and gambusia 7.1 Australian smelt (Retropinna semoni) Introduction Methods Results Discussion 7.2 Western carp gudgeon (Hypseleotris klunzingeri) Introduction Methods Results Discussion 7.3 Gatnbusia (Gambusia holbrooki) Introduction Methods Results Discussion 8. Experimental validation of aging method 8.1 Introduction 8.2 Methods 8.3 Results Mountain galaxias Carp 8.4 Discussion

8 9. Final discussion Addressing a knowledge gap Applying a new technique Variability in species composition and abundance Spawning cues Absence of larger native species Growth, survival and recruitment of larvae and juveniles Livebearers - the potential use of otolith microstructure Potential use of presence of absence of larval fish as indicators Usefulness of findings for river management Final conclusions 10. References 11. Appendices

9 Abstract Six rivers within the upper Mumbidgee catchment were sampled for larval and juvenile fish. The rivers represented both regulated and unregulated flow regimes and varied widely in size. There was wide variation in the larval fish communities supported by each river, both in terms of the species diversity and total abundance of fish sampled. The highly regulated reach of the Mumbidgee River sampled during this study had the highest numbers of native species and native individuals of any river sampled. In the two rivers selected for further study, the Murmmbidgee and Goodradigbee, there was a high level of inter-annual consistency in the species composition within the reaches sampled, despite considerable change in the temperature and flow regimes of both rivers. This indicates that at least some spawning of those species sampled may occur each year, regardless of environmental conditions. Estimates of the relative abundance of each species sampled changed markedly between years, and it is argued, on the basis of growth information contained in the otoliths, that differential survival of larvae and juveniles was largely responsible for this shift in relative abundance. Otolith microstructure provided information on the date of spawning and early growth patterns of all species sampled in the upper Mumumbidgee catchment. In addition to determining the age and thus 'birth-date' of an individual, the effect of a particular event or series of events has on growth, and subsequent survival, is permanently recorded in the otolith microstructure. This enables accurate back-calculation and correlation to management actions or natural events. No other research tool has this ability to retrospectively assess, on a daily basis, the impacts of management actions on condition and subsequent survival of fish larvae. Species sampled could be separated into three groups based on spawning requirements; those linked with flow, those linked with temperature and generalist species that appear to have river independent cues, such as photoperiod or moon phase. Patterns in growth rate during the early life history stages enabled quantification of the consequences of variation in environmental conditions on the survival and recruitment of

10 various species. Growth was not always highly correlated with water temperature, in fact, for mountain galaxias, high temperatures appear to negatively affect larval condition and subsequent survival. Conversely, carp exhibited a strategy more consistent with common perceptions, with growth and survival increasing with increasing temperature. The study uncovered spawning and growth patterns that were unexpected. Age analysis of western carp gudgeon demonstrated that they had undertaken a mid-winter spawning, when the water temperature in the main channel was far lower than that at which spawning was previously recorded for this species. Redfin perch from the unregulated Goodradigbee River exhibited growth rates exceeding the published upper limits for this and other closely related species. This growth could not be correlated with either temperature or flow, indicating that there are additional factors that dominate growth rates of redfin perch in the Goodradigbee River. The proportion and abundance of native species alone is not necessarily indicative of a 'healthy' or pristine system; some native species may be positively affected by river regulation, at least as juveniles. Comparison of the current larval fish community with likely pre-european fish communities does provide an indication of change to the system. The results of this study suggest that larval fish growth rates can be strongly influenced by environmental conditions, thus providing a powerful tool for monitoring future change and the factors which cause it. This study has demonstrated the value of larval and juvenile fish age and growth information, derived from otolith microstructure techniques, for many aspects of river management. Current river management priorities for which these techniques provide unique information include the determination of environmental flow regimes and the control of undesirable exotic species such as carp.

11 xi List of Figures Figure 1 : Average monthly flows of the Murrumbidgee River at Gundagai from 1890 to 1997 (P. 8). Figure 2: Map illustrating the six rivers sampled during the pilot study in (p. 19). Figure 3: Photograph of site sampled on the Goodradigbee River in (p. 20). Figure 4: Photograph of main channel at the site sampled on the Murrumbidgee River in (p. 20). Figure 5: Photograph of the large backwater at the site sampled on the Murrumbidgee River in (p. 21). Figure 6: Photograph from mouth of backwater at the site sampled on the Murrumbidgee River in (p. 21). Figure 7: Photograph of site sampled on the Queanbeyan River in (p. 22). Figure 8: Map illustrating the reaches of the Goodradigbee and Murrumbidgee rivers sampled in (p. 44). Figure 9: Sagitta of a mountain galaxias larva as seen through the image analysis system (p. 46). Figure 10: Detail of the increment structure of the sa'gitta of a mountain galaxias larva (p. 47). Figure 1 1 : Log plot of the number of each species sampled in the Murrumbidgee River in (p. 49). Figure 12: Log plot of the spatial variability of species captured in the Murrumbidgee River in (p. 51). Figure 13: Comparison of hydrolab readings with water temperatures predicted by the regression model for the Mumbidgee River in (p. 56). Figure 14: Comparison between predicted average daily water temperature in and average daily water temperature in for the Murrumbidgee River (p. 57). Figure 15: Average daily water temperature in the Goodradigbee River in and (p. 58). Figure 16: Correlation between mean daily water temperature and mean daily flow in the Murrumbidgee River in (p. 62).

12 xii Figure 17 Correlation between mean daily water temperature and mean daily flow for the Murmmbidgee River in (p. 62). Figure 18: Correlation between mean daily water temperature and mean daily flow for the Goodradigbee River in (p. 63). Figure 19: Correlation between mean daily water temperature and mean daily flow for the Goodradigbee River in (p. 63). Figure 20: Frequency histogram of standard lengths of mountain galaxias captured in the Murrumbidgee River during (p. 73). Figure 2 1 : Relationship between maximum otolith radius and standard length of mountain galaxias larvae collected from the Murrumbidgee River in (p. 74). Figure 22: Relationship between standard length and age of larval mountain galaxias collected from the Murrumbidgee River in (p. 75). Figure 23: Number of mountain galaxias larvae sampled in each weekly cohort in the Murmmbidgee River, (p. 76). Figure 24: Point of disappearance of mountain galaxias larvae from samples, in relation to temperature and flow in the Murmmbidgee River in (p. 78). Figure 25: Individual growth histories of five mountain galaxias larvae illustrated by daily increment width plotted over time (p. 80). Figure 26: Comparison of mean increment width and mean daily water temperature for the entire sample of mountain galaxias from the Murmmbidgee River in (p. 8 1). Figure 27: Relationship between mean daily water temperature and mean daily increment width for mountain galaxias in the Mumunbidgee River in (p. 82). Figure 28: Relationship between main spawning activity of mountain galaxias and mean daily water temperature and mean daily flow in the Murmmbidgee River in (p. 83). Figure 29: Relationship between spawning activity of mountain galaxias in the Murmmbidgee River in and the predicted daily water temperature and mean daily flow (p. 85). Figure 30: Relative size and position of otoliths in representative teleosts and ostariophysan fish such as carp (p. 92). Figure 3 1: Lapillus of a carp larva as seen through the image analysis system (p. 93). Figure 32: Frequency histogram of standard lengths of carp larvae captured in the Goodradigbee River during (p. 95).

13 ... Xlll Figure 33: Relationship between maximum otolith radius and standard length of carp larvae collected from the Goodradigbee River in (p. 96). Figure 34: Relationship between standard length and age of larval carp collected from the Goodradigbee River in (p. 97). Figure 35: Number of carp larvae sampled over the hatching period in the Goodradigbee River in (p. 99). Figure 36: Relationship between mean daily water temperature and mean daily increment width for carp larvae from the Goodradigbee River in (p. 100). Figure 37: Relationship between spawning activity of carp and mean daily water temperature and mean daily flow in the Goodradigbee River in (p. 102). Figure 38: Frequency histogram of standard lengths of carp larvae captured in the Murmmbidgee River during (p. 104). Figure 39: Relationship between maximum otolith radius and standard length of carp larvae collected from the Murrumbidgee River in (p. 104). Figure 40: Relationship between standard length and age of larval carp collected from the Murmmbidgee River in (p. 105). Figure 41: Number of carp larvae sampled from the hatching period in the Mumbidgee River in (p. 106). Figure 42: Regression of mean increment width of carp larvae against mean daily water temperature, as modelled, in the Murrumbidgee River in (p. 108). Figure 43: Timing of spawning activity in carp in relation to mean daily flow and the predicted water temperature of the Murrumbidgee River in (p. 109). Figure 44: Lapillus of a redfin larva as seen through the image analysis system (p. 119). Figure 45: Frequency histogram of standard lengths of redfin perch captured in the Mumbidgee River in (p. 121). Figure 46: Relationship between maximum otolith radius and standard length of redfin perch collected from the Mumbidgee River in (p. 122). Figure 47: Relationship between standard length and age of larval redfin perch collected from the Murmmbidgee River in (p. 123). Figure 48: Relationship between mean daily water temperature, as modelled, and mean daily increment width for redfin perch in the Murrumbidgee River in (p. 125).

14 xiv Figure 49: Relationship between spawning activity of redfin perch and mean daily water temperature, as modelled, and mean daily flow in the Murmmbidgee River in (p. 127). Figure 50: Frequency histogram of standard lengths of redfin perch juveniles captured in the Goodradigbee River during (p. 129). Figure 5 1 : Relationship between maximum otolith radius and standard length of redfin perch juveniles collected from the Goodradigbee River in (p. 129). Figure 52: Relationship between standard length and age of juvenile redfin perch collected from the Goodradigbee River in (p. 130). Figure 53: Relationship between the calculated spawning activity of redfin perch and mean daily water temperature and mean daily flow in the Goodradigbee River in (p. 132). Figure 54: Sagitta of an Australian smelt larva as seen through the image analysis system (p. 140). Figure 55: Frequency histogram of standard lengths of Australian smelt captured in the Murmmbidgee River in (p. 142). Figure 56: Relationship between maximum otolith radius and standard length of Australian smelt collected from the Murmmbidgee River in (p. 143). Figure 57: Relationship between standard length and age of Australian smelt collected from the Murrumbidgee River in (p. 143). Figure 58: Distribution of hatch dates of Australian smelt sampled from the Murrumbidgee River in (p. 144). Figure 59: Relationship between spawning activity of Australian smelt in the Murrumbidgee River in and mean daily water temperature and mean daily flow (p. 146). Figure 60: Lapillus of a western carp gudgeon larva as seen through the image analysis system (p. 153). Figure 61: Frequency histogram of standard lengths of western carp gudgeon captured in the Murrumbidgee River during (p. 155). Figure 62: Relationship between maximum otolith radius and standard length of western carp gudgeon collected from the Murmmbidgee River in (p. 156). Figure 63: Relationship between standard length and age of western carp gudgeon collected from the Murrurnbidgee River in (p. 156).

15 Figure 64: Calculated hatch dates of western carp gudgeon sampled from the Murmmbidgee River in (p. 157). Figure 65: Relationship between main spawning activity of western carp gudgeon and mean daily water temperature and mean daily flow in the Mumunbidgee River in (p. 160). Figure 66: Standard lengths of a sub-sample of western carp gudgeon sampled from the Murrumbidgee River in (p. 16 1). Figure 67: Estimated spawning periods of western carp gudgeon and mean daily flow of the Murrumbidgee River from November 1997 to March 1999 (p. 163). Figure 68: Comparison of relationship between standard length and otolith size of western carp gudgeon spawned in and those spawned during winter 1998 (p. 164). Figure 69: Comparison of relationship between age and standard length of western carp gudgeon spawned in and winter spawned fish (p. 165). Figure 70: Lapillus of a gambusia larva as seen through the image analysis system (p 172). Figure 71: Standard length of gambusia sampled from the Murmmbidgee River in plotted against date sampled (p. 174). Figure 72: Comparison of water temperatures in the main channel and backwater habitat sampled in the Murrumbidgee River in (p. 175). Figure 73: Relationship between standard length and number of increments counted on the lapilli of gambusia sampled from the Murmmbidgee River in (p. 176). Figure 74: Regression of expected number of daily increments against observed number of increments for mountain galaxias larvae (p. 183). Figure 75: Sagitta of a mountain galaxias larva seen under normal light (p. 184). Figure 76: Mean increment widths of mountain galaxias larvae reared at three different temperatures (p. 186). Figure 77: Regression of expected number of daily increments against observed number of increments for carp larvae (p. 188). Figure 78: Mean increment widths of carp larvae reared at three different temperatures (p. 190).

16 List of Tables Table 1: Location and habitat attributes of sites sampled during the pilot study in (p. 16). Table 2: Comparison of the physical characteristics and expected fish fauna of each river sampled during the pilot study (p ). Table 3: Fish assemblage sampled in each river during the spring and summer of (p. 27). Table 4: Number of occasions sampled and temporal range of each species captured in (p. 29). Table 5: Summary of hydrolab data collected fiom each river sampled in (p. 30). Table 6: Summary of flow data for each river sampled in (p. 3 1). Table 7: Methods used to sample species and developmental stage (p. 32). Table 8: Number of occasions sampled and temporal range of each species captured in (p. 50). Table 9: Standardised catch data for the Mummbidgee and Goodradigbee Rivers in (p. 53). Table 10: Comparison of larval fish communities of the Mummbidgee and Goodradigbee Rivers with pre-european fish community (p. 54). Table 1 1 : Summary of the major flow characteristics for the Murrumbidgee and Goodradigbee Rivers in and (p. 59). Table 12: Daily growth, standard deviation and number of larvae aged for eight cohorts of mountain galaxias larvae in the Mumbidgee River, (p. 77). Table 13: Daily growth, standard deviation and number of larvae aged for two cohorts of carp larvae in the Goodradigbee River, (p. 99). Table 14: Daily growth, standard deviation and number of larvae aged for two cohorts and late hatches of carp larvae in the Murrurnbidgee River, (p. 107). Table 15: Methods used to obtain sufficient quantities of larvae for experimental rearing for a number of different species (p. 180).

17 xvii List of Appendices Appendix 1: Regression model used to predict Murmmbidgee water temperatures (p. 227). Appendix 2: Paired t-test comparing increment counts from left and right sagittae of mountain galaxias (p. 227). Appendix 3: One-way, fixed, ANOVA, mountain galaxias larvae, Murmmbidgee River , comparison of growth rates in the day cohort (p. 228). Appendix 4: One-way, fixed, ANOVA, mountain galaxias larvae, Murmmbidgee River, comparison of daily growth rate between and (p. 228). Appendix 5: Paired t-test comparing increment counts from left and right sagittae of carp (p. 228). Appendix 6: One-way, fixed, ANOVA, carp Goodradigbee River , comparison of daily growth rate cohort 1 and cohort 2 (p. 228). Appendix 7: One-way, fixed, ANOVA, comparing the growth of carp larvae between the Mumunbidgee and Goodradigbee Rivers in (p. 229). Appendix 8: One-way, fixed, ANOVA, carp Mumumbidgee River, 0-10 day age class, comparison of daily growth rate cohort 2 and late hatches (p. 229). Appendix 9: One-way, fixed, ANOVA, carp Murmmbidgee River, day age class, comparison of daily growth rates of cohorts (p. 229). Appendix 10: Tukey HSD test, unequal N carp Murmmbidgee River, day age class, comparison of daily growth rates of cohorts (p. 229). Appendix 11: Paired t-test comparing increment counts from redfm sagittae and lapilli (p. 230). Appendix 12: Paired t-test comparing increment counts from left and right lapilli of redfin (p. 230). Appendix 13: One-way, fixed, ANOVA, redfin, comparison of daily growth rates between the Mumumbidgee and Goodradigbee Rivers (p. 230). Appendix 14: Paired t-test comparing increment counts from left and right sagittae of Australian smelt (p. 230). Appendix 15: Paired t-test comparing increment counts from left and right lapilli of western carp gudgeon (p. 23 1).

18 xviii Appendix 16: One-way, fixed, ANOVA comparing daily growth rate of early and late hatched western carp gudgeon in the Murrumbidgee River in (p. 23 1). Appendix 17: Variation of a model utility test used to validate daily increment formation in mountain galaxias larvae (p. 23 1). Appendix 18: One-way, fixed, ANOVA comparing mortality of mountain galaxias larvae from three treatments (p. 232). Appendix 19: Tukey HSD test, mountain galaxias experiment, mortality between treatments (p. 232). Appendix 20: One-way, fixed, ANOVA comparing standard lengths of mountain galaxias from three treatments (p. 232). Appendix 21: ANCOVA comparing mountain galaxias mean increment widths between three treatments (p. 233). Appendix 22: Tukey HSD test, unequal N, mountain galaxias experiment, comparing mean increment widths between three treatments (p. 233). Appendix 23: Variation of a model utility test used to validate daily increment formation in carp larvae (p. 233). Appendix 24: One-way, fixed, ANOVA comparing carp mortality between three treatments (p. 234). Appendix 25: Tukey HSD test, carp experiment, comparing mortality between three treatments (p. 234). Appendix 26: One-way, fixed, ANOVA comparing standard lengths of carp from three treatments (p. 234). Appendix 27: ANCOVA comparing carp mean increment widths between three treatments (p. 234). Appendix 28: Tukey HSD test, unequal N, carp experiment, comparison of mean increment widths between three treatments (p. 235).

19 1. General Introduction Maintaining the world's fi-eshwater resources is one of the most critical elements of global sustainability (Postel, 2001). In 1997 it was estimated that one-third of the world's population was water stressed. United Nations projections suggest that if current trends persist up to 80% of people will be living with serious water shortages by 2050 (United Nations, 2003). Freshwater systems worldwide are under great threat and Australia's are no exception. Australia has a dry and variable climate, and this has created some of the most hydrologically variable river systems in the world (Puckridge et al., 1998; Thoms and Sheldon, 2000b). Australia stores more water per head of population than any other country, so as to provide security of supply for agricultural and urban use (State of the Environment Advisory Council, 1996). Substantial infrastructure has been constructed to provide this storage. As well as having a direct impact on the longitudinal connection of rivers, it has also led to the alteration of natural flow and temperature patterns (Thoms and Sheldon, 2000a; Lugg et al., 2001). The ten key threats to the sustainability of Australia's inland waters outlined in the State of the Environment Report (1996) included: dryland salinity, deterioration of wetlands, overallocation of water to consumption, irrigation, declines of many native aquatic species, nutrient enrichment, water weeds, sediments, lack of monitoring programs and lack of basic data. Since 1996, the pressure on inland water systems has continued to increase with; unsustainable water extraction in many systems, continued clearing of catchment and riparian vegetation and increases in dryland salinity and pesticide use (State of the Environment Advisory Council, 2001). The Murray-Darling Basin, one of Australia's largest freshwater systems, forms the agricultural heartland of the nation (Fitzpatrick, 1990). It is estimated that three million Australians depend heavily upon the resources of the basin. Approximately 80% of accessible water in the Murray-Darling Basin is diverted for consumptive use, especially irrigation and urban supply. This has created many consequences including; an overall reduction of river flows, changes to flood flows (particularly small and medium sized events) and changes to seasonal flows (Thoms and Sheldon, 2002). These changes have had a dramatic impact on the

20 environmental health of the Basin's aquatic ecosystems (Cadwallader and Lawrence, 1990; MDBC, 2003). This deterioration in the health and condition of inland systems such as the Murray-Darling Basin prompted the Council of Australian Governments (COAG) to implement a strategic water reform framework, the aim of which was to achieve an efficient and sustainable water industry (COAG, 1994). The environment forms a key component of the Water Reform Framework (l994), with the requirement to acknowledge the environment as a legitimate user of water during allocation processes and to ensure the ecological sustainability of new water developments. The National Competition Policy was endorsed by COAG in 1995 and directs financial payments to States and Territories that successfully implement a range of important reforms, including the COAG Water Reform Framework. As such, all jurisdictions in Australia are formulating water management plans that incorporate environmental flow allocations. Progress in slowing the firther deterioration of inland systems has also been made by bodies such as the Murray-Darling Basin Ministerial Council, who in 1995 made a landmark decision to cap water diversions in the Murray-Darling Basin at levels (Independent Audit Group, 1996). Currently the Ministerial Council is investigating the provision of environmental flows for the River Murray (MDBC, 2002a). Other priorities include finalising a native fish strategy, the overall goal of which is to, over 50 years of implementation, rehabilitate native fish communities back to 60 per cent of their estimated pre-european settlement levels (MDBC, 2002b). Status of native fish communities in the Murray-Darling Basin Of particular concern to many, especially in rural and regional areas, is the decline in native fish populations across the Murray-Darling Basin. Native freshwater fishing has been an important recreational past-time for many Australians living inland (Roberts and Sainty, 1996). Historical records of huge catches of Murray cod, golden perch and freshwater catfish

21 highlight exactly how poor the current state of native inland fisheries are (Cadwallader and Lawrence, 1990; Roberts and Sainty, 1996; Reid et al., 1997). However, it is not only the larger, recreational and commercial species that have declined. Many of the smaller native species such as pygmy perch, gudgeons and galaxiids are on various threatened species lists. Eight of the Basin's 35 fish species are classified as threatened by the Australian Society for Fish Biology, while 16 species are listed as threatened under State government legislation (MDBC, 2002b). While most of the evidence for declining native fish populations has been anecdotal, the results of the New South Wales Rivers Survey (Harris and Gehrke, 1997) confirmed that native fish, especially in the Murray region, have suffered severe declines in many rivers. In regulated rivers in the Murray region, native species constituted just 20 per cent of the total fish catch. Reasons for native fish declines The health of Australia's fish communities has declined as the rivers in which they live have deteriorated. In a government commissioned report to identify the key threats to Australia's freshwater fisheries, Kearney et al., (1999) list the following: habitat degradation; pollution/water qualitylwater temperature; reduced environmental flows; barriers to migration; introduced specieslcarp; and fishing. An expert panel convened to identify the key interventions required to increase native fish populations in the Murray-Darling Basin, identified similar issues - environmental flows, habitat restoration, carp management, installation of fishways and the mitigation of thermal pollution. It was also stressed that these actions must be undertaken in a coordinated and integrated way to achieve the best results; any action alone would be unlikely to restore native fish populations beyond 25 per cent of their pre-european levels (MDBC, 2002b). These key threats and the interventions required to reverse them are widely accepted. However, given there is still a great deal unknown about the biology and life-history

22 requirements of many of our native species (Cadwallader and Lawrence, 1990), there remains little quantification of the linkages between the listed threats and fish declines. These linkages are especially important if water management agencies are to develop techniques that rehabilitate native fish populations while still allowing for the consumptive use that has come to typify the Murray-Darling Basin. Key to understanding some of these linkages is gaining knowledge about the ecology and population dynamics of various native and exotic fish species. Information on age structures of populations, growth, spawning requirements and the early-life history stage of fish are critical for successfbl management and in many cases the successful rehabilitation of native fish populations. Likewise, such information has the potential to offer benefits in the control of noxious or pest fish species. While there is little understanding of many of the factors affecting the population structures of fish, this study has chosen to focus on spawning requirements and survival during the earlylife history stage, by examining fish larvae and juveniles in an upland catchment of the Murray-Darling Basin. Fish larvae andjuveniles - why are they important Traditionally, studies investigating spawning in fish have been based on the collection of adult fish and the identification of their spawning condition, with the presence of ripe individuals indicating that spawning is imminent while the presence of spent individuals confirms spawning has occurred (O'Connor and Koehn, 1991). However, it has been demonstrated that some species, such as golden perch (Macquaria ambigua), can maintain ripe gonads for several months and if the correct environmental conditions do not eventuate, can reabsorb their gonads (Cadwallader and Lawrence, 1990). This ability has the potential to confbse spawning studies that are based on adult reproductive condition alone. However, the very presence of fish larvae provides positive evidence that a particular species has spawned and the eggs have successfblly hatched (Humphries and Lake, 2000).

23 Fish larvae also have the potential to provide much more information. This has long been recognised by researchers working in marine systems, thus most of the literature cited below is from marine studies. Since the early 1970's, the technology and knowledge have been available to accurately estimate daily age of fish larvae using their otoliths (Pannella, 1971 ; Brothers et al., 1976; a detailed discussion about otoliths and their potential uses is included later in this chapter). Knowledge about the age of an individual can, with supporting biological data, be transformed into a hatch date and also a spawning date. Thus, by aging larval fish, it is possible to correlate spawning times with environmental conditions such as temperature, river flow and moon phase. Undoubtedly, this information could be of great benefit to fisheries and river managers. In addition to providing information about the environmental cues associated with spawning, determining the growth of larval fish is a powerful tool for assessing the impact of environmental conditions on entire cohorts and thus year-class strength. Fish larvae are subject to enormous mortality, with estimates that as much as 99.99% of mortality in a cohort can occur during the larval stage (Campana, 1996). This mortality occurs in various ways. Displacement from favorable nursery areas can occur during the larval stage and has been shown to cause mortality (Iles and Sinclair, 1982; Fortier and Gagne, 1990). Likewise the condition or health of larval and juvenile fish is an important factor regulating mortality during the first year of life (Theilacker, 1986). Indeed, it has long been proposed that levels of recruitment and subsequent year class strength are determined during the 'critical period' which occurs between hatching and first-feeding (Hjort, 1914), with survival related to the period between when fish require exogenous nourishment and when such nourishment becomes available (Ellertsen et al., 1989). This concept was later extended to include development throughout the larval stage and is referred to as the 'match/mismatch hypothesis' (Cushing, 1990). According to this hypothesis, fish in temperate waters tend to spawn at a fixed period each year. In contrast, there is great variability in the timing of production of larval food. The probability of either an exact temporal match or mismatch is unlikely, leading to substantial variation in year class strength.

24 A key component of the 'match/mismatch hypothesis' (Cushing, 1990) is the assumption that increased growth rates during the larval stage will increase survival. Certainly, growth rates are often used as a surrogate for condition in larval fish and have been shown to be inversely related to mortality for a number of species (Jenkins et al., 1993; Rice et al., 1993; Letcher et al., 1996). Rapid growth during the larval stage is believed to enhance survival, by increasing a larva's ability to locate food and avoid predators (Campana, 1996). A number of theories, collectively called the 'growth-predation hypothesis' (Anderson, 1988) suggest that mortality is size selective and is greater in slower growing larvae. These theories are based on the assumption that faster growing larvae will be vulnerable to predation pressure for a shorter period of time because they either remain pelagic for less time or their larger size provides a physical advantage (Miller et al., 1988; Houde, 1989; Leggett and DeBlois, 1994). Larval growth rates have been used to predict year class strength by the end of the first year of life. In Atlantic cod (Gadus morhua) variability in year class strength of populations is due to differential growth rates of young fish (Campana, 1996). Similarly, the settlement and recruitment of the juvenile of a Carribbean reef fish (Acanthurus chirurgus) was positively correlated with variation in larval growth rates (Bergenius et al., 2002). Most of the above-mentioned research has been conducted in marine systems. Do Murray- Darling Basin species use similar strategies or have the evolutionary pressures associated with inhabiting such a variable ecosystem produced differing life-history traits? State of knowledge on spawning and recruitment ofjsh in the Murray-Darling Basin Most early work on spawning behaviour and early life history development of native fish in the Murray-Darling Basin was done under hatchery or experimental pond conditions (Milward, 1965; Lake, 1967a; Lake, 1967b). From this, conclusions were drawn about the importance of rises in water levels for stimulating spawning in important commercial and recreational species, such as Murray cod and golden perch. These conclusions were seen as being supportive of the flood-pulse concept (Junk et al., 1989), that flooding created a burst of productivity, including fish spawning. Ideas regarding the importance of flooding and the inundation of the floodplain as a spawning and nursery ground then grew from this knowledge

25 (Cadwallader and Lawrence, 1990). Indeed, flooding has been shown to be important to the spawning and recruitment of some native species, with the 'flood recruitment model' being proposed to explain the importance of flooding for golden perch (Harris and Gehrke, 1994). A different perspective was introduced by Humphries et al., (1999), who advocated that there was limited evidence to support the importance of flooding and the floodplain for fish spawning and recruitment in the majority of native species. Furthermore, they suggested that the main channel habitat of lowland river systems may play an important role in the development of many native species. The 'low flow hypothesis' was proposed as an explanation for why some native species spawn and recruit during periods when flows are low. Humphries et al., (1999) also questioned the validity of the 'flood-pulse concept' to the entire Murray-Darling Basin, concluding that it describes river systems where high flow periods are accompanied by high temperatures, a condition that may only occur in some lower reaches of the River Murray. The ideas proposed by Humphries et al., (1999) are further explored in this study, as in upland areas of rivers systems, fish must spawn and larvae must develop in the confines of the channel as there is generally little floodplain habitat available. Temperature has also been illustrated to be an important factor in spawning of many native species. Temperature cues usually involve the crossing of a specific temperature threshold. For example, silver perch and freshwater catfish require temperatures in excess of 23 C to spawn and golden perch, temperatures above 20 C (Koehn and O'Connor, 1990). While in other species, a rise in temperature may be required. For example, Murray cod have been observed to spawn in spring after a rise in temperature and while 20 C is often considered a minimum temperature for this species (Humphries et al., 1999), spawning has been recorded at temperatures as low as 16 C (Koehn and O'Connor, 1990). EJtjTects of river regulation and their impact on spawning and recruitment The effects of river regulation are thought to have had a major impact on both spawning and recruitment of many native fish species via two main mechanisms: altered flow regimes and thermal pollution (Cadwallader and Lawrence, 1990; Lugg et al., 2001).

26 The effect of river regulation on the flow regime of a river is illustrated in Figure 1, which averages historical flow records of the Murrumbidgee River at Gundagai dating from 1890 to While Burrinjuck Reservoir was completed in 1927, the capacity of the storage was increased in 1957 and again in 1994, so the regulated versus natural scenario may be exacerbated now, and the long-term records should be interpreted with this in mind. As Figure 1 illustrates, there has been a seasonal reversal in the flow regime, with the highest flows now occurring in summer, whereas under natural conditions flows peaked in winter and early spring. This has also resulted in the loss of periods of low flow, that naturally occurred in summer. 0-1 I month Figure 1 : Average monthly flows of the Murrumbidgee River at Gundagai using data from 1890 to represents natural conditions, o represents current conditions. Taken from Thoms and Dyer, 2001, page 276. River regulation can also have a direct impact on the thermal regimes of rivers. Many of the reservoirs in the headwaters of the southeast Murray-Darling Basin were originally designed to release deep, cold water. The water released during the spring and summer irrigation period is much cooler than the water in the feeder river. In New South Wales, significant thermal pollution occurs downstream of 17 major reservoirs. These storages have the potential to

27 decrease the temperature downstream by as much as 15OC, however, a decrease of 8-12 C is more common (Environment Protection Authority, 2000). The effects of thermal pollution can be felt for hundreds of kilometres downstream of a reservoir. For example, the thermal pollution effects of Hume Dam on the River Murray can be apparent for 200 km downstream (Cadwallader and Lawrence, 1990), although generally thermal pollution effects are greatest within the first 100 to 150 km downstream of the impoundment (Lugg et al., 2001). Effects on spawning The decreased occurrence of flood events during winter and spring, especially smaller and medium size events, as well as reversed seasonality of flow could act to disrupt spawning cues for those species which are reliant on the hydrological regime, for example, golden perch (Harris and Gehrke, 1994). Likewise, reversed seasonality of flow has created the highest river flows over the summer period, when under natural conditions flows would have been low (Figure 1). The decrease in the duration of low flow periods over summer may have negatively affected those species which commonly spawn and recruit under low flow conditions (Humphries et al., 1999). Thermal pollution has the potential to prevent or significantly decrease spawning in some native species, as it often decreases temperatures to below the minimum spawning temperatures for a great proportion time during spring and early summer. This can significantly reduce the spawning window available, especially to species such as silver perch and freshwater catfish that are reported to have high minimum spawning temperatures (Koehn and OYConnor, 1990). Effects on recruitment It has been suggested that in two lowland Australian rivers, spawning of many native fish is a predictable event that occurs regardless of variations in environmental conditions such as hydrology (Humphries and Lake, 2000). This suggests that recruitment failures originate from poor larval survival rather than a lack of spawning activity.

28 Larval growth generally has a positive relationship with temperature, due to the direct metabolic advantage of warmer water and also because increasing temperatures are often associated with increased primary and secondary production. Therefore, in many native species thermal pollution may decrease growth during the larval phase. As previously discussed, this has implications for survival and recruitment. In addition, it is highly likely that the altered flow and temperature regimes of regulated rivers increase the possibility of a 'mismatch' between the appearance of larvae and appropriate prey availability. This may impact significantly on larval growth and condition, also contributing to recruitment failure. The usefulness ofjish larvae as indicators The use of fish larvae as possible indicators of river health has largely been overlooked in Australia's freshwater systems, with the few major monitoring programs that have been undertaken (NSW Rivers Survey, Harris and Gehrke, 1997) or are currently underway (for example the Murray-Darling Basin Commission's Sustainable Rivers Audit) using adult fish. Given the wide scale fish stocking programs that continue throughout rivers and impoundments of the Murray-Darling Basin, monitoring of adult populations may only be providing an indication of stocking success for some key species and in doing so, revealing satisfactory environmental conditions for advanced juveniles and adult fish. However, the presence of fish larvae provides positive evidence that a successful natural spawning event has taken place. Taking this a step fbrther, comparing the larval fish assemblage with the pre- European or 'natural' fish community, provides an indication of change to the system. Indeed, this technique was used successfully by Humphries and Lake (2000) who concluded that fish larvae were a useful indicator of the effect of river regulation in the Campaspe River. However, as previously discussed, fish larvae have the potential to provide even more information, by assessment of their age and growth characteristics. This potential is the subject of much of this current study.

29 The potential value of larval a&e and growth information Determination of the age of a larval fish allows a back-calculation into a hatch date and when combined with basic biological data such as incubation time, a spawning date. Historically, the age of fish has been analysed indirectly, by estimating age from length or by using generalised age-length data that may have been derived from a different population (Clark, 1981 ; Allen, 1992). When used for young-of-year fish, these earlier methods have the potential problem of confusing spawning cohorts with differential growth rates of larval fish or area specific accumulation of size, not age, classes. Knowledge of the exact age of an individual can precisely identify spawning cohorts. Spawning can then be accurately correlated to physical factors such as hydrological regimes, temperature or moon phase thus providing important information on the cues and triggers associated with spawning in fish species, information that for many species, is still poorly understood. Knowledge of age, coupled with information such as the size of a species at hatch, can provide an average growth rate over the life of a fish to the date of capture. This can be used to compare growth rates in a species between cohorts or even between rivers and catchments. Even more powerful, knowledge of the daily growth history of a fish larvae enables correlation of periods of high and low growth with environmental factors. Given the importance of growth rates in determining mortality of year classes, gaining an understanding of the environmental conditions favouring particular species, is vital information for assessing the effectiveness of river management. Methods of determining age and growth Age andlor growth parameters of larval fish can be obtained by several methods including histological, biochemical and morphometric techniques and the examination of bony structures which encode age information (scales, operculum, vertebrae and otoliths) (Jones, 1992).

30 There are several methods of obtaining an index of growth, including histological (muscle and tissue), biochemical (RNA:DNA) and morphometric (length and body depth) techniques. However, no information regarding the age of fish is available from any of these methods. Age and growth data can be obtained by examination of the bony structures of fish. Scales and vertebrae are used, but one of the most popular bones used is the otolith (Jones, 1992). Otoliths are small bones in the inner ear of fish. There are three pairs of otoliths, the sagittae, lapilli and asterisci. Generally, the sagittae and lapilli are the largest and most often used in agelgrowth studies. However, selection of otoliths is dependent on the species being investigated (Jones, 1992). The otolith microstructure consists of a series of incremental rings. Each increment is composed of a protein zone (dark band) and a calcium carbonate zone (light band). Otoliths have been used in agelgrowth studies since the late 1800's. These early studies observed yearly increment formation (Jones, 1992). In the early 19707s, daily increment formation in the otoliths of several temperate water fish was observed (Panella, 1971). Although originally identified in adult fish, the daily increment technique is now more commonly used for the early life history stages (Jones, 1992). The otolith microstructure provides accurate age information by counting the number of daily increments present. It can also be used to provide a daily growth record, as the width of daily increments has been shown to be well correlated to somatic growth in a number of species (Volk et al., 1984; Bailey and Stehr, 1988; Hovenkamp, 1990; Jenkins et al., 1993). It has also been illustrated that the width of daily increments responds rapidly, within one or two days, to environmental changes (such as temperature and food availability) (Eckmann and Rey, 1987; Maillet and Checkley, 1990; Peterson, 1996). In some tropical fish species a short time lag between otolith growth and somatic growth can occur (Molony and Choat, 1990; Milicich and Choat, 1992). Other advantages of otolith microstructure include cost effectiveness, speed with which results can be obtained and the fact that otoliths do not suffer fiom the shrinkage or reabsorption problems often associated with the other techniques mentioned above (Suthers, 1991).

31 There are several assumptions and relationships that need to be determined when using otolith microstructure for age and growth studies. There are two basic assumptions that must be acknowledged before using otolith microstructure to accurately determine age. Firstly, the age at initial increment formation must be known or assumed. For the majority of species, deposition of the first increment generally occurs at hatch or at the onset of exogenous feeding but this can vary, even between closely related species. The only method of validating this assumption is to obtain fertilized eggs and hatch them under laboratory conditions. Secondly, the timing of increment deposition must be determined or assumed. Daily increment formation has been validated for a large number of marine and freshwater fish species (review in Jones, 1986). Daily increment formation is thought to occur almost universally amongst fish species, only the most severe environmental conditions preventing it (Jones, 1992). When using otolith microstructure to estimate growth of fish, the allometric relationship between otolith size and fish size must be determined. This relationship can be anything from a simple linear to a complex curvilinear relationship, depending on the life history of the species (Jones, 1992). Generally for most fish larvae, this relationship is linear. Once this relationship is described, the growth rate and subsequent size at earlier ages can be backcalculated using information contained in the otolith (Lough et al., 1982; Nishimura and Yamada, 1984). The use of otolith microstructure in freshwater studies in Australia has been extremely limited. The hatch and spawning patterns of carp in lowland reaches of the River Murray have been assessed by daily increment counts of otoliths (Vilizzi, 1998). However, no studies have investigated the potential for using the width of daily increments to provide a measure of growth in larval fish. IdentiJication of the knowledge gap There is a good understanding of the impacts of river regulation on the flow and temperature regimes of rivers (Cadwallader and Lawrence, 1990; Lugg et al., 2001) and recognition of how this may impact spawning and recruitment of native fish. However, for many species,

32 little information is known about the precise environmental conditions required to initiate spawning or the environmental conditions required subsequently, to increase the survival and recruitment of larvae. Therefore, it is difficult to quantify the linkages between environmental change and declines of native fish. These linkages are especially important as native fish rehabilitation will occur in the context of the consumptive use that typifies the Murray-Darling Basin. The work that has been done in the Murray-Darling Basin in relation to spawning and recruitment has largely been experimental (Milward, 1965; Lake, 1967a; Lake, 1967b) or focused on lowland river systems (Harris and Gehrke, 1994; Humphries et al., 1999; Humphries and Lake, 2000). This research addresses several of the major questions for upland river systems, using the upper Murmmbidgee catchment as an example. The potential broader implications for system-wide river management are also investigated. Specific objectives of this research: To determine age and growth relationships for several species of native and exotic fish larvae in selected rivers of the upper Murrumbidgee catchment using otolith microstructure. To investigate the variability in growth and survival of native and exotic fish larvae and the relationship to environmental variability in selected rivers in the upper Murmmbidgee catchment. To validate the effect of these environmental factors on growth rates of native and exotic fish larvae in laboratory experiments. To back-calculate the birth dates of fish larvae, and investigate the value, to management, of precise knowledge of the factors influencing successful spawning of various fish species. To assess the usefulness of larval fish growth rates as indicators of river condition.

33 2. Pilot study 2.1 Introduction Prior to the commencement of k s study in March 1997, there was little scientific information available on the abundance, distribution and general health of fish communities in the Murray- Darling Basin. The results of the NSW Rivers Survey (Hanis and Gehrke, 1997) provided the first comprehensive assessment of the distribution and abundance of adult fish communities across New South Wales. Work has also been conducted on the larval fish communities of two lowland rivers, the Campaspe and Broken Rivers in Victoria (Hurnphnes and Lake, 2000). However, no work had been carried out on the early life-history stages of fish in upland reaches of the Murray-Darling Basin. Therefore, the pilot study of this project was necessary to provide basic data on the composition, abundance and variability of larval and juvenile assemblages and the correct methods to sample these assemblages in an upland system. The specific aims of this chapter are to: Investigate the larval and juvenile fish communities present in six streams in the upper Murnunbidgee catchment. Verify that the methods used to sample larval fish in lowland rivers can be successfully used in upland streams. 2.2 Study area Sampling was conducted in six rivers within the upper Murrumbidgee catchment (Figure 2). These rivers were selected, with the major research aims in mind, on the basis of: 1. flow regime - examples of both regulated and unregulated 2. fish fauna - at least some of the expected species common to several rivers 3. size - covering a range of channel and sub-catchment sizes 4. access to continuous flow data

34 Sites on each river were selected on the basis of 1. being privately owned, with limited use by the general public (permission to access the river was granted by all landowners prior to selection of the site) 2. having a range of in-stream habitat types, including riffles, runs, pools, backwaters, aquatic vegetation and snags. 3. all weather vehicle access The sites selected for study are indicated in Figure 2. The exact location and habitat attributes of each site sampled are shown in Table 1. Table 1: Location and habitat attributes of sites sampled during the pilot study in River Queanbeyan Mumbidgee Tumut Gudgenby (ACT) Goodradigbee Goobagandra Latitude 35" 24' 80" 34" 56' 50" 35" 05' 00" 35" 00' 00" 35" 06' 80" 35" 19' 25" Longitude 149" 15' 50" 148" 18' 60" 148" 11 ' 00" 149" 04' 00" 148" 41' 30" 148" 18' 60" Nearest town Queanbeyan Jugiong Gundagai Tharwa Wee Jasper Tumut In-stream habitats present Riffles, runs, pools, overhanging vegetation Runs, backwaters Runs, backwaters Riffles, runs, pools, overhanging vegetation, aquatic vegetation, snags Riffles, runs, pools, backwaters, overhanging vegetation, snags Runs, backwaters, overhanging vegetation, snags

35 The expected fish communities of each river at the site sampled were derived from the NSW River Survey (Hanis and Gehrke, 1997), fish experts within the Cooperative Research Centre for Freshwater Ecology and Environment ACT, land owners and personal observations. A summary of each river including its main physical characteristics and expected fish fauna at the site sampled is included in Table 2. Examples of sites chosen for sampling during the pilot study are shown in Figures Methods Field work Six rivers were sampled during the pilot study in (Figure 2). A description of each river sampled is included in section 2.2, Study Area. One sampling site, ranging in length between two and six river kilometers was selected on each river. Each site contained areas with backwaters and what was assumed to be good in-stream habitat (for example snags, macrophyte beds). Each river was sampled on a fortnightly basis from September 1997 through March 1998, except during major public holiday periods (for example Christmas and New Year) as specified by the NSW Fisheries Scientific Collection Permit obtained for this work. It was also felt that the increased number of people visiting the rivers over holiday periods would increase the possibility of sampling equipment being stolen or damaged. Each river was sampled on 14 occasions. Sampling was conducted at night, with most devices set during the afternoon and collected the following morning. Two main methods of sampling larval fish were used - perspex light traps (modified Quatrefoil; Humphries and Lake, 2000) with a yellow CyalurneB 12 hour light stick and drift nets (500pm mesh). Dip nets (1 mm mesh) were also used on occasions when larvae and juveniles were visible during set-down and collection periods. Sampling was carried out as close to the new moon as possible in order to increase the effectiveness of the light traps.

36 On each sampling occasion, five light traps were placed in backwater and pool habitats, in close proximity to snags and other in-stream habitat and two drift nets were placed in areas of higher flow. The exact placement of light traps and nets within each site was dependent on river height, as this affected available in-stream habitat, and thus varied on each sampling occasion.

37 Figure 2: Map illustrating the six rivers sampled during the pilot study in Sampled river reaches are highlighted and approximate positioning of sampling sites are marked with an arrow. Associated storage reservoirs and closest towns are also labeled. Inset shows position of region sampled in relation to the rest of Australia.

38 Figure 3: Photograph of site sampled on Goodradigbee River, looking upstream. Taken 30 September Figure 4: Photograph of main channel of the Murrurnbidgee River, looking upstream. Taken 22 September 1997.

39 Figure 5: Photograph of the large backwater sampled in the Mummbidgee River, looking in from main channel. Backwater is approximately 50 m long. Taken 9 September Figure 6: Photograph from mouth of backwater, looking out towards the main channel of the Mumbidgee River, which flows towards the top of the page. Taken 9 September 1998.

40 Figure 7: Photograph of site sampled on Queanbeyan River, looking downstream. Light traps are deployed below coloured floats. Taken 9 October 1997.

41 Table 2: Comparison of the physical characteristics and expected fish fauna of each river sampled during the pilot study. CA refers to catchment area upstream of sampling site. 1 NSW River Survey, personal observations, M. Lintermans (pers. comm.), 410cal landowner River Plow Regime Management Purpose CA (km2) Altitude (m) Expected fish fauna Queanbeyan Regulated Urban water supply mountain galaxias' (Googong Res.) brown trout1 rainbow trout1 Murrumbidgee Regulated Irrigation releases call"' (Burrinjuck Res.) brown trout1 rainbow trout1 redfin perch1 golden perch1 Murray cod4 Australian smelt2 Tumut Regulated Power generation and carp1 (Blowering Res.) irrigation releases golden perch1 Murray cod4 brown trout1 rainbow trout1 redfin perch1 Australian smelt1' western carp gudgeon1

42

43 One TinytagTM data logger with external temperature probe (recording range from -40 C- 125OC) was deployed within the sampling site at each river. Of the six data loggers deployed, only three were recovered. Therefore hydrolab readings were used for comparison of temperature data between the six rivers in this chapter. Hydrolab readings of temperature (OC), ph, dissolved oxygen (% saturation and mgll), conductivity (ps/cm) and turbidity (NTU) were taken during set-down and collection of light traps and drift nets. On several occasions during the sampling season the hydrolab was unavailable, so physical data could not be collected on these sampling trips. Flow data for the rivers were obtained from the relevant state water agency. For NSW rivers (Murmmbidgee, Tumut, Goodradigbee and Goobagandra) the NSW Department of Infrastructure, Land, Planning and Natural Resources (DIPNR) provided mean daily flow in megalitres per day (MLJday). For the river reaches in the ACT (Queanbeyan and Gudgenby) ActewAGL provided mean daily flow in megalitres per day. Flow data were obtained from the upstream gauging station closest to the sampling site on each river. Fish collected from light traps and dip nets were euthanased in an ice slurry. Fish were then placed in a 90% ethanol solution and transported back to the laboratory. Samples from drift nets were immediately placed in a 90% ethanol solution and transported back to the laboratory. This work was carried out under a University of Canberra Animal Ethics Permit. Laboratorv work Samples were sorted using a Leica MZ8 binocular microscope. All fish were extracted and identified to species (where possible) using the draft larval fish key of Humphries and Serafini (1997) or the various family keys in McDowall(1996). The number of each species sampled was calculated for each river. A note was also made of the developmental stage of each fish.

44 2.4 Results Species composition and abundance A total of 559 fish, representing eight species, was sampled in (Table 3). The developmental stage of individuals sampled mostly ranged from egg-sac larvae to early juveniles. For some of the smaller species, adults were occasionally captured. Four exotic species, comprising 91% of the total fish sampled, were found - carp, redfin perch, gambusia and trout (no distinction was made between brown trout and rainbow trout). Four native species were also sampled - Australian smelt, mountain galaxias, western carp gudgeon and flathead gudgeon; all small-bodied species. There was wide variability in the number of larvae sampled from different rivers, with total abundance varying from 373 fish in the Murrumbidgee River to three fish in the Queanbeyan River (Table 3). The Murrumbidgee River had the largest number of species, totalling seven. While four of these species were natives, the three exotic species; gambusia, carp and redfin, were numerically the most dominant (Table 3). Of the native species, only western carp gudgeon was caught in reasonable numbers (Table 3). The Goodradigbee River was the next most diverse river, with four species sampled. Australian smelt was the only native species sampled and carp was the most dominant species in the Goodradigbee River (Table 3). The four remaining rivers had low numbers of species and total abundance. No carp or gambusia were found in any of these rivers, but trout were found in all four. Small numbers of mountain galaxias were found in both the Queanbeyan and Gudgenby Rivers.

45 Table 3: Fish species and numbers captured in each river during the spring and summer of Development stage categories: L, Larvae, J, Juvenile, A, Adult. * gambusia give birth to juveniles. RIVER SPECIES NUMBER DEVELOPMENT TOTAL CAPTURED CAPTURED STAGE carp 75 LY J Redfin perch 37 L Gambusia 227 J*, A Mumbidgee Australian smelt 4 L 373 Western carp gudgeon 25 Ly J Flathead gudgeon 1 J Mountain galaxias 4 Ly Jy A carp 113 Ly J Goodradigbee Redfin Australian smelt 14 5 J J 134 Trout 2 J Trout 2 J Tumut Western carp gudgeon 2 A 5 Australian smelt 1 A Goobagandra Trout 3 2 J 3 2 Gudgenb y Trout Mountain galaxias 6 6 J L l 3 1 Queanbeyan Mountain galaxias L

46 Temvoral variability in svecies There was also variability in the temporal range of species sampled (Table 4). Gambusia had the widest temporal range, being sampled seven times in the Murmmbidgee River between November 1997 and February The majority of species were sampled only once or twice from each river during the pilot study (Table 4). Temporal patterns within a species were not consistent between rivers, with carp being sampled five times between November 1997 and January 1998 in the Murmmbidgee River but only twice in the Goodradigbee River over the December-January period. Similarly, redfin were sampled once in mid-november in the Murmmbidgee River and three times over the November-December period in the Goodradigbee River (Table 4). Australian smelt were the first larvae to be sampled in late September Gambusia and western carp gudgeon were the last larvae to be sampled in late February 1998 (Table 4). Physical data Hydrolab readings illustrated a wide range of temperature profiles in the rivers sampled, the Turnut and Queanbeyan Rivers being the coldest (Table 5). The Murmmbidgee River was the warmest. The ph of all rivers was slightly alkaline. Dissolved oxygen was high at all sites, the lowest average value of mgll was recorded in the Gudgenby River (Table 5). Conductivity was variable between sub-catchments, but was generally less than 100 ps/cm. The highest conductivity readings were recorded in the Mummbidgee River, which had an average of ps/cm. Turbidity was low in all rivers (Table 5).

47 Table 4: Number of occasions sampled and temporal range of each species captured in I GB - Goobagandra, GD - Goodradigbee, GU - Gudgenby, MB - Munumbidgee, QU - Queanbeyan, TU - Tumut. SPECLES RIVER NUMBER OF TEMPORAL RANGE OCCASIONS SAMPLED carp MB GD Nov Jan Dec Jan 98 Redfin MB Nov 97 GD 3 14 Nov Dec 97 Garnbusia MB 7 11 Nov Feb 98 Smelt MB 2 22 Sept 97 & 25 Nov 97 GD 2 10 Dec 97 & 10 Jan 98 TU 1 23 Oct 97 Western carp gudgeon MB 2 30 Jan 98 & 20 Feb 98 TU 1 13 Nov 97 Flathead gudgeon MB 1 25 Nov 97 Mountain galaxias MB 1 25 Nov 97 GU 2 24 Oct 97 & 8 Nov 97 QB 1 11 Dec 97 Trout GD 1 30 Sept 97 TU 1 23 Oct 97 GB 2 30 Oct 97 & 15 Nov 97 GU 1 24 Oct 97

48 Table 5: Summary of hydrolab data collected from each river sampled in Mean value is given with the standard deviation. The range is shown in brackets. GB - Goobagandra, GD - Goodradigbee, GU - Gudgenby, MB - Munumbidgee, QU - Queanbeyan, TU - Tumut. GB GD GU MB QU TU Temperature ("c) /- 3.8 ( ) /- 4.9 ( ) /- 4.0 ( ) /- 4.3 ( ) /- 2.6 ( ) /- 4.3 ( ) PH 7.3 +/- 0.2 ( ) 7.4 +/- 0.3 ( ) 7.1 +/- 0.1 ( ) 7.6 +/- 0.2 ( ) 7.7 +/- 0.2 ( ) 7.3 +/- 0.3 ( ) Dissolved /- 1.5 ( ) 8.8 +/- 1.3 ( ) 7.6 +/- 0.9 ( ) 8.8 +/- 1.1 ( ) /- 0.9 ( ) /- 1.5 ( ) Conductivity (P S/cm) /- 6.1 ( ) / ( ) / ( ) / ( ) /- 1.7 ( ) /- 2.9 ( ) Turbidity (ntu) /- 6.6 (0-15.9) 1.8 +/- 8.2 (0-29.3) 2.0 +/- 1.9 (0-5.0) 4.0 +/- 5.3 (0-15.0) 2.0 +/- 6.5 (0-21.1) 8.3 +/- 4.9 ( )

49 Flow data The median daily flow in rivers sampled ranged from 12 MLIday in the Gudgenby River to 9225 MLlday in the Tumut River (Table 6). The Murrumbidgee and Tumut Rivers had lower coefficients of variation than the other rivers (Table 6), illustrating the stable nature of flows imposed in reaches directly downstream of major reservoirs in these two rivers. The third regulated river, the Queanbeyan River had one of the lowest discharges over the springlsummer period. However, this stream also had a relatively high co-efficient of variation, comparable to that of the unregulated rivers sampled during this study (Table 6). Table 6: Summary of flow data for each river sampled in Data have been averaged over the sampling period, from September to February. GB - Goobagandra, GD - Goodradigbee, GU - Gudgenby, MB - Munumbidgee, QB - Queanbeyan, TU - Tumut. ' data kindly provided by NSW DIPNR; data kindly provided by ACTEWAGL. (MLIday) SD CV

50 Sampling methods There were differences in species and developmental stages caught by the various methods (Table 7). Young carp and redfin larvae were sampled in drift nets, however, as carp developed they were sampled using light traps. Mountain galaxias, Australian smelt and western carp gudgeon were exclusively sampled using light traps. Trout and gambusia juveniles were sampled by light traps and dip nets. carp Red fin Table 7: Methods used to sample species and developmental stage SPECIES DEVELOPMENT STAGE SUCCESSFUL METHOD(S) Egg-sac larvae Drift net Later larvae Light trap Juveniles Light trap Larvae Drift net and Light trap Gambusia Trout Mountain galaxias Australian smelt Western carp gudgeon Flathead gudgeon Juveniles Juveniles Larvae Juveniles Larvae Larvae Juveniles Juveniles Dip net and Light trap Dip net and Light trap Light trap Light trap Light trap Light trap Light trap Light trap

51 2.5 Discussion To date, freshwater larval fish studies undertaken in Australia have mainly focused on lowland rivers (Gehrke, 1992; Vilizzi, 1998; Humphries et al., 1999; Humphries and Lake, 2000). Prior to the commencement of this study in 1997, no published information existed on the larval fish assemblages in upland river reaches of the Murray-Darling Basin. This pilot study was necessary to provide a baseline assessment of the species composition and abundance of larval fish in six rivers in the upper Murmmbidgee catchment so that appropriate sampling sites could be selected for investigations into the spawning, growth and survival of fish larvae in this upland system. As specific fish species are discussed in detail in Chapters 4-7, this discussion is focused on investigating the differences between the expected and observed fish faunas; the variability in species composition and abundance between rivers; and the utility of the methods used for sampling larval and juvenile fish in this pilot study. The six rivers selected were representative of the various types of rivers present in this area. Two are large and heavily regulated for irrigation (the Murmmbidgee and Tumut), while the Queanbeyan River is much smaller and regulated for the purpose of urban water supply. The Goodradigbee and Goobagandra are the largest relatively pristine (in terms of physical condition) rivers left in the region. However, these rivers are much smaller than both the Murrumbidgee and Tumut Rivers. The Gudgenby River is a small, unregulated tributary of the Murmmbidgee River. A total of 559 fish was sampled during the pilot study (Table 3). While the ratio of native species to exotic species was equal, the abundance data illustrates that 91% of the total fish caught were exotic species. Similar abundance data was found in the NSW Rivers Survey (Hams and Gehrke, 1997) with native species constituting only 20% of the catch in the Murray region of New South Wales (which includes the Murmmbidgee system). The temporal range of species was variable, both between species and within a species between rivers (Table 4). Gambusia, carp and redfin had the longest temporal ranges,

52 however, the majority of species were only sampled once or twice from each river during the pilot study. Observed versus expected fish fauna Of the nine species expected (combining brown and rainbow trout) from the six rivers (Table 2), six were sampled (Table 3). Species expected but not observed were three large native species - golden perch, Murray cod and Macquarie perch. The failure to sample golden perch is a concern as this species is still believed to be relatively abundant, at least in the Murrumbidgee River. Golden perch larvae have been sampled with drift nets in the Broken River, (Humphries and Lake, 2000) therefore the absence of golden perch is not believed to be due to inadequate sampling methods. Adult Murray cod were not sampled by the NSW Rivers Survey in any of the rivers included in this study, therefore, the probability of sampling larvae was always low. Likewise, while Macquarie perch are known to occur in the Goodradigbee River (Harris and Gehrke, 1997) they are not common and therefore it was not surprising that no larvae were sampled. Two unexpected species were sampled, gambusia and flathead gudgeon. It was uncertain whether gambusia would be collected from any of the rivers sampled during this study, as they were not sampled in any of these rivers by the NSW Rivers Survey, although they were sampled in other lowland rivers in the Murray region (Harris and Gehrke, 1997). Their distribution was thought to be restricted to lakes and slow-flowing rivers (McDowall, 1996). In this study, gambusia were only sampled from the Murrumbidgee River, however, they were the most numerically abundant species sampled, indicating that this species is also able to inhabit larger, faster flowing rivers. Flathead gudgeon were also unexpected in the study area, as their known distribution range is mher west (or downstream) in the Murmmbidgee system (McDowall, 1996). Only one juvenile flathead gudgeon was recorded during and this was sampled from the

53 Murmmbidgee River, suggesting that the presence of this species in the region sampled is not common. Variabilitv in species comvosition and abundance The variability between rivers, both in terms of species composition and abundance, was a feature of the pilot study. There was no negative relationship between the number of native species or native fish abundance and river regulation, with the heavily regulated Murmmbidgee River having the highest numbers of native species and native individuals, of any river sampled. The three unregulated rivers all had equal or less numbers of native species compared to exotic species, suggesting that factors other than flow regime may be affecting spawning and larval survival of native fish in these upland streams. Murmmbidnee River Despite having the most native fish species and the highest abundance of native fish of any river sampled, the fish fauna of the Murmmbidgee River was still heavily dominated by exotic species, which constituted 91 % of the total fish sampled from this section of the river. This result is not surprising and is consistent with the findings that regulated rivers of the Murray region are heavily dominated by exotic species (Harris and Gehrke, 1997; Gehrke and Harris, 2001). It has been observed that Burrinjuck Reservoir can create thermal pollution for 100 km downstream of the dam (Lugg et al., 2001). Despite this, during the spring and summer of the site on the Murrumbidgee River was the warmest of the six rivers sampled (Table 5). River regulation and poor land management practices have also created habitat problems in the Murmmbidgee River. The large extremes in flow rate (Table 6) combined with clearing of the riparian zone have also caused massive erosion and bank slumping in this section of the Murrumbidgee River.

54 Noticeably, the coefficient of variation for flow in the Murmmbidgee River is much lower than that of the three unregulated streams (Table 6) indicating a loss of flow variability in this reach. While most attention is often focused on the loss of high flow events and the negative effects this may have on spawning in some native species (Harris, 1991 ; Gehrke and Harris, 200 l), periods of low flow may also be important for spawning and recruitment of other native species (Humphries et al., 1999). The altered flow regime and degraded habitat of the Mumbidgee may combine to make this river more suitable for recruitment of exotic species than native species. Regardless, spawning is occurring in at least four native species, suggesting that the spawning cues of some native species are being satisfied. The low numbers of native species sampled indicates that mortality during the larval phase may be a major factor in this river. These ideas are further developed in Chapters 4-7. Goodradigbee River The study site on the Goodradigbee River was the second most diverse sampled, with a total of four species. Carp larvae dominated the sample from this river, comprising 84% of the total fish sampled (Table 3). The Goodradigbee was the third warmest river sampled, however, it also had the highest variability in temperature (Table 5). In respect to flow, the coefficient of variation was the second highest of all rivers sampled (Table 6). Both these factors illustrate the highly variable nature of this stream. This variability is not caused by a regulating structure, but is due to the nature of the catchment this river drains. The headwaters of the Goodradigbee River and many of its tributaries occur in the Kosciusko National Park, the snowfields of New South Wales. The high flow rates during spring are due to snowmelt, thus the cooler temperatures. During summer, when the flow of the river is naturally low and there is little runoff from the upper catchment, the water warms quickly. While the low probability of sampling Murray cod and Macquarie perch has already been discussed, it was felt that if they were to be sampled, it would most likely be from the Goodradigbee River. Adults of both species are known from this river and the relatively

55 pristine condition of the upper and mid reaches of the Goodradigbee River support the notion that these species should be successfully spawning in this river. However, the lower reaches of the Goodradigbee River, downstream from Wee Jasper, are also affected by river regulation. Burrinjuck Reservoir can back-up a significant distance of the Goodradigbee River. The condition of the river in this area is poor, with a lack of riparian vegetation and bank erosion creating a shallow channel with heavy sedimentation. The poor state of the river in this area could account for the decline in Murray cod upstream of Wee Jasper. Large species like Murray cod may be reluctant or unable to move through such degraded areas to the more pristine habitat upstream. As recreational fishers selectively target and remove large fish from the upper areas, a lack of migration could result in the loss of such species to upstream areas. A similar situation occurs in the Murmmbidgee River above Burrinjuck Dam at Tharwa, where a massive sand slug has prevented upstream fish migration for some time. Rock groynes have been built into the side of the bank to scour out deep pools which the fish can utilise as they move upstream. Although still in the early stages of the project, trout cod have been sampled utilising these scoured out areas (Lintermans, pers. comm.). Other rivers The sites on the four other rivers sampled produced few species (four in total) and a low total number captured (52, Table 3). Interestingly, trout was the only exotic species present in any of these rivers. Neither carp nor gambusia were sampled from the other regulated rivers, the Tumut and Queanbeyan, despite these species often being positively associated with river regulation (Gehrke and Harris, 2001). Obviously, factors other than the flow regime affect the distribution and abundance of these species. These factors will be mher considered in Chapters 5 (carp) and 7 (gambusia). Interestingly, the Tumut and Queanbeyan Rivers were the two coldest sampled (Table 5), the thermal pollution aspects of river regulation were apparent in these rivers, with average springlsumrner temperatures failing to increase above 14OC. Certainly, this is well below the suggested threshold temperatures for a variety of larger

56 native species (Koehn and O'Connor, 1990; McDowall, 1996) as well as exotic species other than trout (Chapter 5, this study; Smith, 2003). One interesting difference between the Queanbeyan River and the other two regulated rivers is the flow variation. Regulated for urban water supply rather than irrigation, the Queanbeyan River had a coefficient of variation comparable to that of an unregulated stream and much higher than that of both of the other regulated rivers (Table 6). This suggests that Googong Reservoir, while having a major impact on the water temperature and discharge volume, is having little effect on the timing of flows down the Queanbeyan River. However, this relatively 'natural' flow variability has not produced a good result in terms of the fish fauna. The Queanbeyan River had the lowest species diversity and the lowest abundance of any river sampled during this pilot study (Table 3) indicating that temperature andfor total flow may be important for recruitment of both native and exotic fish species. Given the limited data available from this river, further investigation is required to precisely determine any relationship between environmental variables and recruitment of larvae. Appropriateness of methods The results of the pilot study indicate that the methods used to sample larval and juvenile fish in lowland rivers, successfblly sample a wide range of species in upland systems. There was both species and developmental stage differentiation with respect to methods (Table 7). Young carp were sampled in drift nets, however, light traps became the dominant successfbl sampling method as carp developed. Mountain galaxias, Australian smelt and western carp gudgeon were exclusively sampled in light traps. Trout and gambusia were sampled in light traps and by dip nets. The results of the pilot study confirm that all three sampling techniques are required to obtain all species and all developmental stages. The pilot study also confirmed that these methods maintain the sample in good condition, a factor that is important when undertaking otolith work. Of the three methods used, drift nets were the most hazardous in this respect, given the large amount of additional organic material

57 collected. However, rapid sorting of samples and re-freshing of the preservative after approximately 24 hours were sufficient to prevent breakdown of otoliths. 2.5 Conclusions and suggestions for next phase The pilot study successfully recorded, for the first time, the larval and juvenile fish assemblages of six rivers in the upper Murmmbidgee catchment. A total of 559 individuals was sampled, comprising eight species. In respect to species composition, the number of native species sampled was equivalent to the number of exotic species sampled. However, exotic species were numerically dominant, representing 91% of the total fish sampled. The pilot study successfully used sampling equipment that had hitherto not been trialled in upland systems. The combined use of light traps, drift nets and dip nets successfully sampled a range of developmental stages of both native and exotic fish species. It was also shown that these techniques have the ability to sample large numbers of larvae and juveniles while simultaneously maintaining samples in excellent condition. This is an important requirement when the aim of the research is to undertake otolith work and provides a sound basis for future work of this type. A feature of the pilot study was the variability, both in terms of species composition and abundance, between rivers. The Murmmbidgee River produced seven species and a total of 373 individuals. In contrast, the Queanbeyan River produced a single species and a total of three individuals. The immense variation in both species composition and abundance of larval fish sampled from the six rivers, while extremely interesting in itself, indicated that continuing to sample all six rivers did not represent a cost-effective method of investigating the major aims of this study. The two rivers that had yielded the most species and individuals, the Murmmbidgee and Goodradigbee Rivers, were selected for further study on the basis that they offered the most effective experimental design that still enabled comparison of regulated and unregulated rivers. The decrease in the number of rivers being sampled created the opportunity to increase

58 sampling effort on the two remaining rivers. The change in sampling design resulting from this pilot study is described in Chapter 3.

59 3. Main study 3.1 Introduction The pilot study (Chapter 2) illustrated high variability in the larval and juvenile fish assemblages, both in terms of species diversity and abundance, between rivers in the upper Murmmbidgee catchment. Based on these results, and with the major aims of the study in mind, it was determined that the most effective sampling design would be obtained by concentrating sampling effort in those rivers where catch rates were highest. Therefore, sampling for the main study focused on the Mumumbidgee and Goodradigbee Rivers. The reduction in the number of rivers sampled provided the opportunity to significantly increase the sampling effort in the two remaining rivers. The details of the re-worked sampling strategy are included in section 3.2, Methods. A brief description of the two focus rivers was provided in Chapter 2. Further details of both the Mumumbidgee and Goodradigbee Rivers are included below. Murrumbidrzee River The Mumunbidgee River has a total length of 1690 krn and is the third longest river in Australia and the largest river in the Canberra region. It originates in the upper Snowy Mountains where it flows south-east before turning to the north at Cooma. It flows through the Australian Capital Territory and turns west near Yass. From there, it continues in a westerly direction until it joins the River Murray upstream of Euston Weir. Along its length, the Mumumbidgee is regulated by two main storages and several weirs. The headwaters of the Murmmbidgee are regulated by Tantangra Reservoir, which has a storage capacity of 254 GL. Some of the water captured by this reservoir is transferred to Lake Eucumbene as part of the Snowy Mountains Scheme. The largest storage on the Murmmbidgee is Burrinjuck Reservoir, which has a capacity of 1026 GL. Burrinjuck Reservoir is some 250 km from the source of the Murmmbidgee and thus has the potential to impact the 1440 km or 85% of river that occurs downstream. Studies carried out on the thermal regime of the Mumunbidgee River

60 illustrate that the effects of Burrinjuck Dam are felt for approximately 100 km downstream (Lugg et al., 2001). Burrinjuck Reservoir has also had a significant impact on the flow regime of the Murmmbidgee River, since its completion in 1927 (Figure 1, Chapter 1). One major problem created by Burrinjuck Reservoir is the reversal in flow seasonality. Large water releases intended for the Mumbidgee Irrigation Area throughout spring and summer act to maintain high, often constant flow rates when under natural conditions, flows should be decreasing. In contrast, many of the natural winter flow events originating higher in the catchment are captured by Burrinjuck, and not passed on downstream. This creates low flow conditions over winter, when under natural conditions flows would have been at their highest levels. A reach stretching approximately 40 km, between Burrinjuck Reservoir and the confluence of the Murmmbidgee with the Tumut River at Gundagai, was investigated in The predominant land use in the area sampled is agricultural, including grazing and cropping of lucerne, asparagus and sorghum. The in-stream and riparian habitat of this stretch of the Murnunbidgee is highly variable. Some areas have fair riparian cover of mainly native species with good in-stream habitat. Other areas have severe bank erosion with little or no riparian habitat remaining. The bedload of the stream is also highly variable, with areas of exposed bedrock, cobbles and also large silt and sand slugs. Goodradigbee River The Goodradigbee River is a smaller, northerly flowing stream, with a total length of approximately 80 km. There are no storages or weirs on the Goodradigbee to regulate the flow, however, the natural discharge has been reduced due to diversions from the upper reaches to the Snowy Mountains Scheme. The source of the Goodradigbee is also in the alpine areas of New South Wales. The river flows on the western side of the Brindabella Range and enters Burrinjuck Reservoir downstream of Wee Jasper. Depending on the level of water behind the dam, the lower reaches of the Goodradigbee can back up to the township of Wee Jasper.

61 The lower reaches of the Goodradigbee are severely degraded, with severe bank erosion and minimal riparian habitat. The predominant sediments in this area of the river are silt and sand and the channel is shallow. Further upstream more natural riverine habitat is found, riparian strips have been protected and are often fenced to prevent stock access. In these areas of the Goodradigbee, pool, riffle and run sequences are common. The reach sampled during was 2 to 16 km upstream of Wee Jasper. Upstream of the sampled reach, a large gorge prevents the movement of larger species such as Murray cod and carp. The flow regime of the Goodradigbee River still follows a natural seasonal pattern, with high flows over the winter and early spring period and low flows during the summer. The specific aims of this chapter are to: Determine the effectiveness of a new sampling design. Investigate inter-annual consistency in species composition and abundance in the Mumunbidgee and Goodradigbee Rivers. Investigate the relationships between flow and temperature in the Mumunbidgee and Goodradigbee Rivers, and how these relate to larval fish assemblages. 3.2 Methods The methods outlined below were used to sample and analyse all species. Particular methods, specific to a species, are outlined in the relevant species chapter. Field work Each river reach was sampled weekly from September 1998 through March As in the pilot study, public holiday periods were avoided as a requirement of the NSW Fisheries Scientific Collection Permit. This restriction also reduced the possibility of loss of equipment. Three sites were sampled in each reach (Figure 8). These sites were selected using the same criteria outlined in Chapter 2.

62 At each site five light traps were set. One drift net was also deployed at the first and third site. As in the pilot study, dip nets were also used. Each site was sampled on 30 occasions in Given the three different methods used it is difficult to calculate the exact change in sampling effort between and For the analysis, a six-fold increase was estimated for , based on increases in the numbers of light traps and drift nets set. TinytagTM data loggers with external temperature probes were set at each site on both rivers from 10 September 1998 in the Mumbidgee and 22 September 1998 in the Goodradigbee to the 31 March Temperatures were recorded every four hours. Apart from the changes outlined above, sampling was conducted as outlined in section 2.3, Methods - field work. Figure 8: Map illustrating the reaches of the Goodradigbee and Murumbidgee Rivers sampled in The position of the three sites sampled on each river are indicated with numbers, with 1 denoting the site firthest upstream and 3 the site furthest downstream.

63 Laboratow Work Samples were sorted under a Leica MZ8 binocular microscope. All fish were extracted and identified as in Chapter 2. Each larva, including those larvae and juveniles sampled from the Mumbidgee and Goodradigbee Rivers in the pilot study, was assigned a number and stored in a separate glass vial. For each larva, the total length (TL) and standard length (SL) were measured to the nearest 0.5 mm. Wet weight was also obtained to the nearest 0.1 mg. Otoliths were removed under a Leica MZ8 binocular microscope. Otoliths were mounted, convex side up, on separate microscope slides using clear commercial nail varnish (Secor et al., 1992). Slides holding otoliths were numbered. All specimens of a single species were analysed together as this increased consistency between readings. However, otoliths from each river were analysed randomly to minimise bias. Details of what particular otoliths were used for each species and associated statistics are provided in the relevant species chapters (Chapters 4 to 7). As the majority of otoliths were from larvae and juveniles they required minimal preparation. However, when necessary otoliths were polished with 3M lapping film (grades from 3-12 pm). Otoliths were viewed under oil immersion (100x power) using a Zeiss compound microscope. Age was estimated by counting the total number of growth bands, which were proved to be daily growth increments for mountain galaxias and carp (Chapter 8) and assumed to be daily growth increments for the remaining species. For each otolith, increment counts were made in triplicate and the mean value taken. Otolith diameter and radius and increment widths were measured using image analysis (Figures 9 and 10). All measurements were made in micrometres (pm). A Leica CCD video camera was attached to the Zeis compound microscope, and connected to a monitor. The public domain image analysis software Scion Image was used to process results. All measurements were taken along the same axis of the otolith. The axis used varied for different species, but generally the maximum radius was used. Measurements were made on as many

64 increments as possible. The estimated error of this technique, from re-measuring a small sample of otoliths, was pm. Figure 9: Sagitta of a mountain galaxias larva (SL = 13.5 mm) as seen through the image analysis system. Normal light, 40x magnification.

65 Figure 10: Detail of the increment structure of the sagitta shown in Figure 9, as seen through the image analysis system. Increment widths were measured on a horizontal line along the maximum otolith radius, from the core to the outside. Each increment was measured from the beginning of the protein (dark) band, as illustrated. Normal light, 100x magnification with oil.

66 3.3 Results Larval fish com~osition a) Mumunbidgee River In the larval and juvenile fish community of the Murmmbidgee River was dominated by exotic species - gambusia (60.9%), carp (20.1%) and redfin perch (9.9%) (Table 3). Four native species were sampled - Australian smelt, western carp gudgeon, flathead gudgeon and mountain galaxias. The combined proportion of native species sampled was 9.1 % (Table 3). In the larval and juvenile fish assemblage of the Mumbidgee River was dominated by native species - western carp gudgeon (92. I%), mountain galaxias (4.9%) and Australian smelt (2%) (Figure 11). Few exotic species were sampled, a combined proportion of less than 1% (Figure 11). b) Goodradigbee River In the larval and juvenile fish community of the Goodradigbee River was dominated by exotic species - carp (84.3%) and redfin perch (10.4%) (Table 3). A small number of juvenile trout were also sampled. One native species, Australian smelt, was present in this river, comprising 3.7 % of the sample (Table 3). In fish were captured in the Goodradigbee River. Four species were sampled - Australian smelt (1 I), redfin perch (5), Murray cod (2) and trout (2).

67 western carp mountain Australian redfm perch gambusia Carp flathead gudgeon galaxias smelt Species gudgeon Figure 11: Log plot of the number of each species sampled in the Mumbidgee River in A log scale was used due to the large variation in the number of larvae of different species captured. The actual number of each species sampled is printed above the corresponding bar.

68 Temporal variability in Variability in the temporal range of species sampled in was also examined (Table 8). Table 8: Number of occasions sampled and temporal range of each species captured in MB - Mumbidgee, GD - Goodradigbee. SPECIES RIVER NUMBER OF TEMPORAL RANGE OCCASIONS SAMPLED Western carp gudgeon ME3 6 9 Sept March 99 Mountain galaxias ME Sept Jan 99 Australian smelt ME Sept March 99 GD 3 20 Nov 98-4 Dec 98 Redfin perch ME Oct 98 & 29 Oct 98 GD 2 20 Nov 98 & 27 Nov 98 Gambusia ME3 2 10Feb99 & 18 Feb99 carp ME Feb 99 Flathead gudgeon ME Oct 98 Murray cod GD 1 11 Feb 99 Trout GD 1 24 Sept 98 There were only three species captured more than three times during Australian smelt, western carp gudgeon and mountain galaxias (Table 8). Australian smelt and western carp gudgeon had the widest temporal range, both being sampled from September 1998 through until March 1999.

69 Spatial variability in Spatial variability was examined by comparing the nurnber of larvae of each species sampled at each site in (Figure 12). This analysis was only undertaken for the Murmmbidgee River due to the low number of larvae sampled from the Goodradigbee River in Site Site 2 Site 3 western carp mountain Australian redfin perch gambusia CarP flathead gudgeon galaxias smelt gudgeon Species Figure 12: Log plot of the spatial variability of species captured in the Murmmbidgee River in Site numbers correspond to those shown in Figure 8, 1 being furthest upstream and 3 being furthest downstream. Two of the three most abundant species (western carp gudgeon and Australian smelt) were sampled across all three sites in the Murmmbidgee River (Figure 12). Mountain galaxias was the exception, only being sampled at sites one and two.

70 Comparisons between and To enable comparison between sampling years, catch data were standardised (Table 9). Numbers sampled in were divided by a factor of six to take into account the additional sampling effort used in this year (section 3.2, Methods). The standardised data further illustrate the dramatic decrease in the relative abundance of exotic species in in both rivers. Gambusia and carp showed the largest changes, decreasing by factors of 284 and 250 respectively in the Mumbidgee River (Table 9). Redfm perch were more persistent, decreasing by a factor of 25 in the Munumbidgee River and 17.5 in the Goodradigbee River. Two native species in the Munumbidgee River showed increases in western carp gudgeon and mountain galaxias, increased by factors of 11 and four respectively. Australian smelt showed little change in abundance between years in the Mumunbidgee and decreased slightly in the Goodradigbee River (Table 9).

71 Table 9: Standardised catch data for the Murmmbidgee and Goodradigbee Rivers in and Catches in have been divided by six to take into account the extra sampling effort used in this year (see section 3.2, Methods). Brackets show actual number sampled.

72 Table 10: Comparison of larval fish communities of the Murrumbidgee and Goodradigbee Rivers with pre-european fish community. Pre-European fish community determined from McDowall, 1996 and Australia New Guinea Fishes Association information. The pre-european fish communities of the Murmmbidgee and Goodradigbee Rivers were derived from distribution records (McDowall, 1996; Australia New Guinea Fishes Association). The number of species sampled as larvae or juveniles in the current study was then calculated as a percentage of the pre-european fish community. In the Murmmbidgee River, 33% of the predicted pre-european fish fauna was sampled as larvae or juveniles during this study (Table 10). In the Goodradigbee River, 40% of the predicted pre-european fish fauna was sampled as larvae or juveniles during this study (Table 10).

73 Temperature data The data from all three loggers in each river were averaged to calculate mean daily water temperature for each river. Data were examined to detect if loggers had been interfered with or removed from the water. Any such data were discarded from the analysis. This only applied to loggers in the Murmmbidgee River that would occasionally be left out of the water during sudden drops in water level. In the logger placed in the Murmmbidgee was washed downstream during a large flow event late in the season. The logger was found however the data could not be recovered. Therefore temperature data in the Murmmbidgee in this year was modelled. Loggers from the Murmmbidgee in and the Goodradigbee in both and were successfully recovered. However, the logger recording temperature in the Goodradigbee River from January to March 1998 was stolen from the sampling site. Therefore, recorded temperature data in the Goodradigbee River are available only until January a) Murmmbidgee River The average daily water temperature for the reach sampled in the Murmmbidgee River in was predicted using standard painvise multiple regression analysis in Statistica. The R~ of the model was 0.79 (p < , Appendix 1). Of the independent variables used, mean daily flow, maximum daily air temperature and water temperature of the Goodradigbee River (a main upstream tributary) were significant (Appendix 1). The predicted water temperatures were similar to the hydrolab readings, however, they were on average higher. This difference was significant (Figure 13, n = 8, paired t-test, t = 2.757, p = 0.028). This result is not surprising as hydrolab measurements were often taken early in the morning, when the water temperatures would have been lower than the average daily temperature. However, as Figure 13 shows, there is good correlation between the hydrolab readings and the modeled temperatures and the lower readings from the hydrolab do not cause any bias in subsequent analysis.

74 G E 3 Q) +., h e- I lo- ""; t I m m - Predicted mean daily temperature Hydrolab reading Figure 13: Comparison of hydrolab readings (dark diamonds) with water temperatures predicted by the regression model for the Murmmbidgee River in Date The predicted average daily water temperatures for and the average daily water temperature obtained from data logger readings in are shown in Figure 14. The predicted water temperature in (calculated over the period from 16 September to 5 January) was on average 3.2OC warmer than the corresponding temperature recorded in The maximum temperature predicted in was 26.2OC, compared with a maximum recorded temperature of 23.8"C in There were also temporal differences between years for the water temperature to reach 20 C, a temperature associated with spawning in a number of large native species including golden perch (Kohen and O'Connor, 1990; McDowall, 1996; Lugg et al., 2001). The average daily water temperature reached 20 C on 10 November in 1997 and 24 December in 1998.

75 Figure 14: Comparison between predicted average daily water temperature in and average daily water temperature in for the Murmmbidgee River. b) Goodradigbee River The average daily water temperatures for the Goodradigbee River in and are shown in Figure 15. The Goodradigbee River was warmer in compared to by an average of 2.9"C. The maximum temperature recorded in the Goodradigbee River in was 31 C, compared with 25.5"C in As in the Murmmbidgee there was also a temporal difference, with the average daily water temperature reaching 20 C on 11 November in and 10 December in

76 Figure 15: Average daily water temperature in the Goodradigbee River in and The average water temperatures in the Goodradigbee River in and are similar until mid to late October (Figure 15). After this point the temperatures diverge by an average of 5 C. This trend continues until the end of recorded data in

77 Flow data A summary of the flow data for the Murmmbidgee and Goodradigbee Rivers in and is shown in Table 1 1. Table 11: Summary of the major flow characteristics for the Murrumbidgee and Goodradigbee Rivers in and Data have been averaged over the sampling season, from September to February. Historical data is averaged over September to February from to for the Goodradigbee River and to for the Murmmbidgee River. Flow data kindly provided by NSW DIPNR. Murrumbidgee River Historical Goodradigbee River Historical Minimum flow (MLlday) Maximum flow (MLlday) Median flow (MLlday) Mean flow (MLIday) SD CV

78 The trends in flow characteristics between the two seasons were the same for both rivers. Discharges in were higher than in This is illustrated by increases in all flow rate parameters including minimum, maximum, median and mean daily flow in (Table 11). The co-efficient of variation was lower in , suggesting that flows were less variable than in in both rivers (Table 11). Discharges in the Goodradigbee were much lower than those in the Murmmbidgee, as indicated by the minimum, maximum, median and mean flow data (Table 11). This in itself is not an indication of river regulation as the Goodradigbee is a much smaller river. However, one strong indication of river regulation is the coefficient of variation, which is considerably higher in the Goodradigbee River (Table 11). Of particular note in the Murmmbidgee River was the release of a 13 day environmental flow, which peaked at just over ML during early September The hydrograph of this event is shown in Figure 28 (Chapter 4). A comparison with historical flow data (Table 11) illustrates that the Murmmbidgee, while having comparable mean and median flow discharges had far less variability in the two studied years than the longer term average. In the Goodradigbee River the mean and median discharge volumes in were similar to the longer term average, however, the pattern of flow delivery was less variable, resulting in a lower co-efficient of variation (Table 11). Relationships between temperature and flow a) Murrumbidgee River The correlations between discharge and water temperature in the Murmmbidgee River in and are shown in Figures 16 and 17 respectively. In both years correlations are weak (r = 0.19 and respectively). Generally, the lowest temperatures are associated with the larger discharges. The correlation is confused by the relatively low winter temperatures, which are associated with periods of low flow.

79 b) Goodradigbee River The correlation between discharge and water temperature in the Goodradigbee River in and is shown in Figures 18 and 19 respectively. There is a relatively strong negative relationship between mean daily water temperature and mean daily flow in the Goodradigbee River in both years (Figure 18 and 19). The correlation strength, r is for and for A comparison of the figures provided for the Murrumbidgee and Goodradigbee Rivers illustrates the problems of river regulation in the Murrumbidgee, both in terms of changed seasonality of flow and thermal pollution. Many of the low flow periods in the Murmmbidgee are associated with low temperatures, due to the fact they are occurring in winter, not summer as in the Goodradigbee River.

80 TEMP = ' MB97-98 Correlation: r =.I9450 z 5 o Regression MEAN DAILY FLOW (MUday) 95% confid. Figure 16: Correlation between mean daily water temperature (OC) and mean daily flow (MLlday) in the Murmmbidgee River Correlation r = Data used corresponded to the sampling season, September - February. TEMP = ' FLOW Correlation: r = MEAN DAILY FLOW (MUday) o Regression 95% confid. Figure 17: Correlation between mean daily water temperature (OC) and mean daily flow (MLlday) for the Murmmbidgee River Correlation r = Data used corresponded to the sampling season, September - February.

81 TEMP = ,0158 ' FLOW Correlation: r = Q 10 z Q W ' MEAN DAILY FLOW (MUday) o Regression 95% confid. Figure 18: Correlation between mean daily water temperature (OC) and mean daily flow (MLIday) for the Goodradigbee River in The correlation co-efficient, r = Data used in correlation corresponded to the sampling season, September - February. TEMP = ' FLOW Correlation: r = MEAN DAILY FLOW (MUday) o Regression 95% confid. Figure 19: Correlation between mean daily water temperature (OC) and mean daily flow (MLIday) for the Goodradigbee River in The correlation co-efficient, r = Data used in correlation corresponded to the sampling season, September - February.

82 3.4 Discussion Compared with marine and estuarine studies of larval fish (Tmski, 2001; Leis et al., 2002), the sample numbers obtained in this study appear relatively low. However, low sample numbers are characteristic of larval fish studies in the Murray-Darling Basin. Larval fish studies carried out on the Broken and Campaspe Rivers (Humphries and Lake, 2000) and the Ovens River floodplain (King et al., 2003) also display highly variable sample numbers. While some species such as Australian smelt were often sampled in large numbers (i.e. hundreds), other species including mountain galaxias were rarely caught (i.e. 57 individuals over three years), despite relatively intensive sampling effort. The data presented in this study, for carp and gambusia in and western carp gudgeon and mountain galaxias in , are based on large sample numbers compared to other studies in this field. There was evidence of inter-annual consistency in species present in both rivers. However, the relative abundance of species was markedly different between years, in both rivers. In both the Munumbidgee and Goodradigbee Rivers were dominated by exotic fish species, while in native species dominated the Murmmbidgee River and while few fish were sampled in the Goodradigbee River, native species were dominant. This shift in the relative abundance of larval fish was accompanied by shifts in both the temperature and flow characteristics of the rivers. These results indicate that the larval fish assemblages of rivers in the mid and upper Murmmbidgee catchment are geographically diverse (Chapter 2) and extremely variable within the same river from year to year. Species composition between years The species composition of the Murmmbidgee River remained unchanged between sampling years, with the same seven species being sampled (Table 3 and Figure 11). Inter-annual consistency in species composition was not as strong in the Goodradigbee River, with three of the five species sampled from this river sampled in both years. Strong inter-annual consistency in the species composition of larval fish has been demonstrated in the Broken and Campaspe Rivers by Humphries and Lake (2000), who went on to suggest that the occurrence

83 of fish larvae in these two rivers is predictable, with most species spawning regardless of the variations in the hydrologic regime. Certainly, the results from the Murrumbidgee River support this concept. However, it should be noted that species for which the hydrologic regime is believed to be important, such as golden perch (Harris and Gehrke, 1994) were not sampled during this study, and this could be due to insufficient spawning cues. Relative abundance of species between years Despite consistency in species composition, there were major changes in the relative abundance of species sampled between years, in both rivers (Table 9). In the Murmmbidgee River in , gambusia, carp and redfin comprised 91 % of the larval fish sampled. In contrast, in the Mummbidgee was dominated by native species including western carp gudgeon, mountain galaxias and Australian smelt. These three native species comprised more than 99% of the larval fish sampled from this area of the Mummbidgee in The three exotic species that had dominated the sample in the previous year accounted for less than 1% of the larvae sampled in Similar patterns were seen in the Goodradigbee River. In , exotic species accounted for 96% of the larvae sampled, with carp and redfin being the most abundant species. In catches in the Goodradigbee were low, with 20 larvae representing four species being sampled. Native species constituted 60% of fish sampled, with Australian smelt and Murray cod both present. The decrease in total catch between years in the Goodradigbee River can be attributed to the complete disappearance of carp from samples. They had constituted 84% of fish sampled in Links between fish assemblages. flow and temperature The shift in the relative abundance of species was matched by changes in both the temperature and flow regimes of both rivers sampled.

84 Discharges were higher in both systems in as illustrated by the increases in almost all of the flow parameters, the one exception being the coefficient of variation, whch even decreased in the unregulated Goodradigbee River (Table 11). Of particular interest in the Murrumbidgee River was the managed environmental flow release, which peaked at just over ML in early September 1998 (Figure 24, Chapter 4). In response to this increase in discharge, there was a decline in the average daily water temperature by approximately 3OC in both rivers in (Figures 14 and 15). There was also a major temporal difference in the rivers reaching 20 C, a temperature associated with spawning in many larger native species such as golden perch (Koehn and O'Connor, 1990; McDowall, 1996; Lugg et al., 2001). There was a six week lag in reaching this temperature in the Murrumbidgee River and a four week lag in the Goodradigbee River. This lag could have a serious negative effect on spawning and larval development of fish species that have temperature related flow cues andlor require warm temperatures during the larval phase to assist in growth and development. The loss of flow variability in Australia's river systems has been suggested as one of the key threats to native species (Walker and Thoms, 1993; Thoms and Sheldon, 2000b). Therefore, it appears noteworthy that the numbers of larvae and juvenile native fish increased in , despite a decrease in the coefficient of variation, a measure of the variability of stream flow. This parameter was determined for the entire sampling period, from September to March. The period during which variability would be most vital for a fish is in the lead up to spawning, when changes to the flow and temperature regime are needed to initiate spawning in many species (Koehn and O'Connor, 1990; Harris and Gehrke, 1994; McDowall, 1996). Stability over this period may result in spawning cues not being triggered and may lead to reduced spawning activity and lower numbers of native larvae. However, once spawning has occurred and larvae have hatched, variability in the system could negatively affect larvae. Compared to adults, larvae have a limited ability to cope with changing environmental conditions. Periods of strong flow could displace larvae into potentially unfavourable habitats downstream, while

85 sudden shifts in temperature, of even modest amplitude, could cause stress and increased mortality of young larvae. The correlation of temperature and flow in the Mumunbidgee and Goodradigbee Rivers provides a good illustration of the effect of regulation on the temperature and discharge regimes of the Murrumbidgee River (Figures 16, 17, 18 and 19). Prior to regulation it is llkely that the Murrumbidgee River would have had a temperaturelflow correlation of a similar nature to that seen in the Goodradigbee. However, the effect of Burrinjuck Reservoir, in terms of reversed seasonality of flow and thermal pollution, has created the different relationship that is observed. In the Murrumbidgee River, low flow periods are associated with the lowest temperatures, a signal of the low flows maintained in this river over winter. Likewise, the thermal pollution effects of large releases of cold water, which act to depress temperatures during summer, are also obvious. Comments on s~ecific species For the majority of species successfully sampled in this study, entire chapters are devoted to discussing their spawning requirements and larval characteristics (Chapters 4-7). However, there was a lack of larger native species sampled. Murray cod were rarely sampled during this study, while golden perch and Macquarie perch were not sampled at all. A discussion regarding the lack of these species was included in Chapter 2, and this is expanded below. The successll sampling of Murray cod larvae in the Goodradigbee River in provides positive evidence that this species is still present and successfully spawning in this river. It has been suggested that Murray cod is one species that does not require a flood peak to initiate spawning but that spawning follows a circa-annual pattern that is regulated by a minimum temperature (Humphries et al., 1999). The required temperature range to initiate spawning of Murray cod has been suggested as 16 to 21 C (Koehn and O'Connor, 1990; McDowall, 1996; Lugg et al., 2001). Given this, it is surprising that Murray cod larvae were not sampled from the Goodradigbee River in when the temperature was warmer. Possibly, the breeding population of Murray cod is now so restricted, spawning is an uncommon event. The fact that

86 cod larvae were sampled in may have been due to the extensive sampling effort used rather than any specific environmental event. The small number of cod sampled supports this theory. The failure to sample golden perch larvae from the Murrumbidgee River in either year is a concern. Adult golden perch were sampled from upstream of the sampled reach in the NSW Rivers Survey (Harris and Gehrke, 1997). The literature regarding this species suggests that flow and temperature cues may be necessary for this species (Harris and Gehrke, 1994; Hurnphries et al., 1999). The large environmental flow release in early September 1998 in the Murmmbidgee River should have met any flow cue. However, the lower water temperatures associated with this release and with the onset of the irrigation flows shortly after, may have been a major factor in preventing successful spawning. The inability of golden perch to undertake a significant spawning migration, due to the presence of Burrinjuck Reservoir, may also be hampering the successful spawning of this species in the sampled reach of the Murmmbidgee River. It is also possible that spawning did occur but eggs and larvae were displaced downstream beyond the sampling area because of the high flow events seen in the Murrumbidgee during the spring and summer of Success of new samvling desim The increased sampling effort used in successfully demonstrated that while the composition of species may be relatively stable between years in a river, there can be extreme inter-annual variation in the abundance of each species in a river. The results from the Murrumbidgee River highlight this, with gambusia and carp being the most dominant species in , while few individuals of either species were sampled in Likewise, there was a huge increase in the numbers of western carp gudgeon sampled from the Murrumbidgee River between and The low total catch in the Goodradigbee River in is believed to be a true reflection of the larval fish assemblage in that particular year. The methods used in this study have been proven to successfully sample nine species occurring in the upper Murrumbidgee catchment,

87 therefore, if larvae had been present in the Goodradigbee, it is probable that they would have been sampled. 3.5 Conclusions and focus of upcoming chapters These results indicate that there is generally a high level of inter-annual consistency in the species composition of the Murmmbidgee and Goodradigbee Rivers. However, there was an extreme shift in the relative abundance of different species between and in both rivers sampled. In both rivers were dominated by exotic species such as carp, gambusia and redfin. By contrast, in , the Murmmbidgee River was dominated by native species such as mountain galaxias, western carp gudgeon and Australian smelt, while in the Goodradigbee River the proportion of native species sampled increased considerably. This change in the relative abundance of species was accompanied by a shift in both the temperature and flow regimes of both rivers between sampling years. In the following four chapters the possible reasons behind this shift in the larval and juvenile fish assemblages is further investigated on a species by species basis. Age and growth data, determined using otolith microstructure, is used to assess the factors affecting spawning patterns and condition of several native and exotic fish species, during the larval and juvenile phase.

88 4. Mountain galaxias (Galaxias olidus) 4.1 Introduction Mountain galaxias is a small, short-lived, native fish of the family Galaxiidae (O'Connor and Koehn, 1991). It grows to a maximum size of approximately 140 mm, but often only reaches mm in length (McDowall, 1996). Mountain galaxiids are widespread throughout eastern Australia, both in coastal drainages and the Murray-Darling Basin. They are primarily found in small streams at moderate and high altitudes, up to about 1800 m, where the water temperature remains relatively cool over the summer months (O'Connor and Koehn, 1991). Mountain galaxias remains relatively widespread and abundant over its range, however it has been reported as having a reduced abundance and fragmented distributions in areas also inhabited by trout (Lintermans, 2000). The spawning season of mountain galaxias varies between late winter and autumn across its distribution range. The cues required for spawning in mountain galaxias are uncertain (Humphries et al., 1999). There is wide variation in the temperatures associated with spawning, ranging from 6.5"C to 20 C, although temperatures between 7 and 11 C are most common (Cowden, 1988; Marshall, 1989; O'Connor and Koehn, 1991). Female mountain galaxias spawn relatively few eggs (around 200), which take approximately 21 days to hatch at water temperatures between OC (O'Connor and Koehn, 1991). In laboratory incubated eggs, newly hatched larvae ranged in size from 9 to 9.8 mm total length (average of 9.4 mm total length) (O'Connor and Koehn, 1991). However, this may vary across regions, as larvae with total lengths as low as 7 mm were sampled in this study. Several honours theses have looked at otoliths in adult mountain galaxias, with one study suggesting that annular patterns on the otolith were variable and therefore unsuitable for aging purposes (Cowden, 1988; Drayson, 1989). No published work could be found which has examined the otolith microstructure in larval and juvenile mountain galaxias. Work has been done examining the growth and condition of mountain galaxias larvae using cell cycle analysis (Bromhead et al., 2000). Variable results were found, which illustrated that in larvae held at

89 20 C, brain cell division was in one case higher and in another case lower than in larvae held at 12OC. The later finding was contributed to the possible effects of disease and increased mortality of larvae in this experiment. This highlights the potential use of otolith microstructure in condition studies, as it provides a growth history, rather than an instantaneous growth rate, that is permanently recorded and therefore not subject to possible sampling effects. This chapter examines both the effect of environmental factors such as temperature and flow on the growth of young mountain galaxias and the timing and cues associated with spawning in this species in an upland river system, using otolith microstructure. 4.2 Methods Aside from the species specific methods mentioned below, all other methods are as outlined in Chapter 3, section 3.2. Both the sagittae and lapilli otoliths were removed fiom 10 mountain galaxias larvae and analysed with the image analysis system. The sagittae was selected for further analysis for two reasons - the resolution of the ring structure was better and they were more uniform in shape (Figure 9). The sagittae was the only otolith removed from the remainder of the mountain galaxias analysed. The left and right sagittae of 10 individuals were examined to observe variations. There was no significant difference in increment counts between the left and right (paired t-test, t = -1, p = Appendix 2), so only one sagitta was used from each larva. Daily increment formation in mountain galaxias was validated in this study (Chapter 8) and under the assumption that the first increment is laid down at hatch, ages were calculated from counts of otolith increments (Figure 10).

90 4.3 Results Mountain galaxias larvae were sampled in the Murmmbidgee, Gudgenby and Queanbeyan Rivers (Table 3 and Figure 11). Only the larvae captured from the Murmmbidgee River have been used for age and growth investigations. The low numbers of larvae sampled in the other rivers would not have yielded statistically rigorous results. In , four mountain galaxias larvae were caught in the Murmmbidgee River (Table 3). The remainder of this section will focus on the larvae sampled in , however, some reference will be made to the larvae at the conclusion of this section. In , 92 mountain galaxias larvae were captured in the Murmmbidgee River (Figure 11). Of these fish, four were removed from the sample as they were badly damaged and could not be measured accurately. 88 fish were suitable for further analysis. Otoliths of 83 fish were analysed, due to loss and breakage of a small number of otoliths during the extraction process. Lenrrth data The standard length of mountain galaxias larvae captured in ranged from 6.5 mm to 35 mm, with an average of 14.5 rnrn (Figure 20). Developmentally, individuals sampled ranged from early larvae to late juveniles. The maximum otolith radius was linearly related with standard length of the fish (SL, Figure 21) for mountain galaxias between 6.5 and 35 mm standard length in : Radius = SL , (? = 0.96) Age data The relationship between standard length and age was also linear (Figure 22) for mountain galaxias between 6.5 and 35 mm standard length in : SL = age , (r2 = 0.92)

91 The average age of mountain galaxias sampled from the Murrumbidgee River in was 40 +I- 18 days (range from 15 to 120 days) Standard length (mm) Figure 20: Frequency histogram of standard lengths (mm) of mountain galaxias captured in the Murrumbidgee River during

92 o i I 1 t r 1 1 I Standard length (mm) Figure 21 : Relationship between maximum otolith radius (ym) and standard length (rnm) of mountain galaxias larvae sampled in the Murmmbidgee River in

93 Figure 22: Relationship between standard length (mm) and age (days) of larval mountain galaxias collected from the Murmmbidgee River in Hatch dates Hatch dates were back-calculated from the ages measured. Mountain galaxias larvae captured in the Mumunbidgee River hatched over a long period, from 20 August 1998 to 2 November However, the majority of larvae hatched after 4 September 1998, only one individual hatched prior to this. Larvae were assigned into weekly cohorts starting on 4 September. Eight cohorts were formed, the abundance of larvae in each cohort is illustrated in Figure 23. The number of larvae by cohort was normally distributed, with the majority of larvae being hatched in the three weeks between 20 September and 13 October (Figure 23).

94 Cohort hatching date Figure 23: Number of mountain galaxias larvae sampled in each weekly cohort in the Mumunbidgee River, (n = 83). Growth data The daily growth rate for each fish was calculated by the following formula: Daily growth rate (mm d-') = total length (mm) - 5 mm (estimated size at hatch) age (days) On average mountain galaxias larvae between 6.5 and 35 mm standard length grew 0.24 mm d-' (range from mm d-') in Larvae in each cohort were assigned into one of seven age classes, based on their age at capture: days, days, days, days, days, days and 7 1 days and over.

95 The results were then averaged to provide a mean growth rate for each age class in each cohort (Table 12). This removed the impact of age on growth, allowing the growth rates of different cohorts to be compared. The oldest fish sampled were hatched in the earliest cohorts, with fish older than 70 days exclusively found in cohorts 1 and 2. Fish that hatched in later cohorts disappeared out of the sample at a younger age. Table 12: Daily growth (mrn d-i), standard deviation and number of larvae aged (in brackets) for eight cohorts of mountain galaxias larvae in the Murmmbidgee River, Cohort Number and hatching date Age at capture (days) /9/ /9/98-19/9/ /9/98-27/9/ /9/9& 5/10/98 5 6/10/98-13/10/ /10/98-21/10/ /10/98-29/10/ /10/98-6/11/ / (5) 0.16+/ (4) 0.2 I+/ (4) 0.25+/ (5) 0.28+/ (2) 0.22+/ (7) 0.25+/ (3) 0.27+/- 0.03(9) 0.28+/ (8) 0.24 (1) 0.29 (1) 0.22+/ (2) 0.25+/- 0.02(5) 0.24+/- 0.02(4) 0.25+/ (2) 0.24+/- 0.02(5) 0.24+/- O.Ol(3) 0.22 (1) 0.24+/ (6) 0.25+/- 0.02(3) 0.24+/ (2) 0.25 (1)

96 Analysis of Variance (ANOVA) was carried out on each age class to investigate differences between cohorts. In the day age class, cohort 5 had significantly higher growth rates than cohort 8 (ANOVA, one-way, fixed, F = 7.81, p = 0.03, Appendix 3). No other significant differences were detected within other age classes. Interestingly few larvae from cohorts 3-8 were present in samples taken after 19 November, with just nine larvae sampled on 26 November. After this date only advanced fish fi-om cohorts 1 and 2 were sampled, in mid-january (note no sampling was conducted in the Mumunbidgee between 22 December 1998 and 7 January 1999, see section 3.2). There was a sudden increase in temperature after 19 November from 14.6OC to 17OC within two days. The temperature of the Murmmbidgee continued to increase throughout December (Figure 24). Sharp increases in temperature were also experienced in late September, due to low flows at this time (Figure 24). However, the maximum temperature reached at this time was 14'C. only 9 lame wre sampled after this date 1 previous increases in teqmture Date Figure 24: Point of disappearance of mountain galaxias larvae from samples, in relation to temperature (OC) and flow (MLIday) in the Murmmbidgee River in

97 Growth histories were also examined using daily increment widths. The otolith increment technique allows identification of the daily growth history of individual fish, as illustrated for several mountain galaxias larvae (Figure 25). From this, daily variations in growth can be correlated to environmental factors. In order to assess the broad aims of the study, daily increment widths of all fish were averaged to provide a population response to environmental conditions. The mean increment width of all fish analysed (n = 83) was well correlated to temperature (Figure 26). As the temperature increases past 14 C the relationship absolves, indicating that increased temperatures reduce increment widths, thus slowing growth and negatively affecting condition of larvae. The regression of mean increment width and temperature also illustrates that a temperature threshold is reached, at approximately 14"C, beyond which increment widths stabilise (Figure 27). A weak negative correlation exists between mean daily water flow and mean daily increment width (2 = 0.09), suggesting that temperature is an over-riding factor in the condition of mountain galaxias larvae.

98 (wn) q ~p!~ ~nau~a~~n! &tra

99 - 4 Incremmt width I I I 1 I I I 0 I 1 0 I 24/9/98 1/10/ /10/98 W / / / /98 Date Figure 26: Comparison of mean increment width (pm) and mean daily water temperature (OC) for the entire sample of mountain galaxias from the Murmmbidgee River in (n = 83). The standard deviation of mean increment width ranged from 0.2 to 0.8 pm (average of 0.5 pm).

100 Mean daily water kqerature ec) Figure 27: Relationship between mean daily water temperature (OC) and mean daily increment width (pm) for mountain galaxias in the Murmmbidgee River in (n = 83). S~awning; activity The timing of spawning activity was also investigated for this species. Estimated spawning date was back-calculated from hatch date, using an incubation time of 21 days (O'Connor and Koehn, 1991 ; McDowall, 1996). Spawning is thought to have occurred in close proximity to the sampled reach as mountain galaxias larvae are exclusively sampled with light traps, not drift nets (Table 7).

101 For the season it is estimated that all spawning occurred between 30 July and 11 October The majority of larvae were spawned after the 24 August (77 of 83 fish analysed). There was a peak in spawning from 17 to 19 September, with 16 fish spawned over this three day period (Figure 28) w I miin spavming activity, peak is highlighted Date Figure 28: Relationship between main spawning activity of mountain galaxias (shown with dashed line, peak spawning period is solid) and mean daily water temperature ("C) and mean daily flow (MLJday) in the Mumbidgee River in The commencement of the main spawning period in mountain galaxias occurred three days after the new moon in September, with the peak three day period starting four days after the last quarter. The average water temperature throughout the main spawning period was 11 C. Spawning activity of mountain galaxias in was correlated to several flow peaks. The

102 initial flow peak on the 25 August appears to have initiated spawning, while the peak spawning period occurred immediately following the commencement of the large environmental flow release in early September Mountain galaxias sampled in While only four fish were captured in , some interesting comparisons can be made between the two seasons. The standard length of mountain galaxias larvae captured in ranged from 10 mm to 12 mm with an average of mm. The maximum otolith radius was linearly related with standard length for mountain galaxias between 10 and 12 mm standard length from the Mumbidgee River in : Radius = SL , (r2 = 0.77) The relationship of standard length and age was also linear for mountain galaxias between 10 mm and 12 mm standard length from the Mumbidgee River in : SL = age , (2 = 0.90) The larvae sampled in ranged in age from 25 to 3 1 days old, and had an average growth rate of 0.27 * 0.03 mm d-l. This compares with the value of 0.22 * 0.08 mm d" growth for similar aged larvae spawned in However, this difference was not significant (ANOVA, one-way, fixed, F = 1.4, p = 0.27, Appendix 4) due to the large variability in the data. The hatching period for larvae in was from 25 October to 31 October Assuming a 21 day incubation period, spawning was estimated to have occurred between 4 October and 10 October 1997 (Figure 29). Again, the spawning period corresponds to a spring flow peak created by the initial irrigation releases of the season. Up until ths point, base flows had been maintained (Figure 29). The spawning period occurred two days after the new moon and was associated with average temperatures of 12 C. The temperature rose

103 substantially after this point and despite firther flow peaks in early and late November, no spawning was recorded (Figure 29). Date Figure 29: Relationship between spawning activity of mountain galaxias in the Murmmbidgee River in (shown by solid line) and the predicted mean daily water temperature ( OC) and mean daily flow (MWday).

104 4.4 Discussion Mountain galaxias (Galaxias olidus) are widespread throughout rivers of the upper Murrumbidgee catchment, being successhlly sampled in the Murmmbidgee, Gudgenby and Queanbeyan Rivers during the pilot study in (Chapter 2). Abundance of mountain galaxias was variable between rivers and between sampling seasons. The majority of mountain galaxias were sampled from the Murrumbidgee River, with relatively few individuals sampled from the other two rivers. More mountain galaxias were sampled in than in , the relative abundance increasing from four to 15.3 (Chapter 3). Possible reasons for this includes changes to the level of spawning activity andlor larval survival between sampling years. These ideas are hrther developed later in this section. A broad size range of mountain galaxias was sampled from the Mumunbidgee River in , ranging developmentally from early larvae to late juveniles. Previous laboratory studies have indicated that the size of mountain galaxias larvae at hatch exceeds 9 mm total length (O'Connor and Koehn, 1991). Given that the smallest larvae sampled in this study were 7 mm total length and had approximately 14 daily increments, it was estimated that the size at hatch of larvae from the Murmmbidgee River was approximately 5 mm. Certainly, from these inconsistent results, it appears that size at hatch is variable across the geographical range of mountain galaxias, and may be related to incubation temperature or other environmental factors. The use of otolith microstructure proved to be a good measure of larval growth in mountain galaxias, as the size of the otolith was strongly correlated with standard length (Figure 21). A strong linear relationship was also demonstrated between standard length and age (Figure 22), supporting the assumption that increments are deposited daily. Daily increment formation in this species was subsequently validated under laboratory conditions (Chapter 8). Similar strong relationships were seen in the larvae sampled in , despite the low number of individuals.

105 Back-calculation of hatch dates demonstrated that mountain galaxias larvae in the Mumbidgee in hatched over a prolonged period, in excess of two months. The temporal range in hatch dates provided the opportunity to compare average growth rates of early and later hatched fish. The analysis demonstrated that for young larvae, those hatched earlier in the season had significantly higher growth than late hatched larvae (Table 12). This is extended to assume that earlier hatched larvae had a greater chance of survival than later hatched larvae. Interestingly, the larvae from the six later cohorts largely disappeared from samples after 19 November, with only a small number of larvae captured on 26 November. After this date, only well-developed juveniles from the first two cohorts were sampled. The period of disappearance was associated with a sudden increase in average daily temperature of more than 2OC, to 17OC over two days (Figure 24). This disappearance could be due to movement away from the study area or mortality. Dispersal of larger, well developed individuals is definitely a possibility. However, this is unlikely for younger larvae that are not advanced enough to actively seek out new habitat. Displacement due to a flow event is also unlikely given the relatively stable flows over this period. Therefore, mortality is the most likely reason why young larvae disappeared from the sample. This could be due to two factors related to the observed temperature increase. Mountain galaxias larvae may have had a negative response to the crossing of some temperature threshold or a negative response to the sudden change in the temperature. This second factor is unlikely, as sharp increases in temperature were also experienced earlier in the season. In these cases the average daily temperature reached did not exceed 15OC. This evidence indicates that larval mountain galaxias may be susceptible to increasing temperatures, above a threshold temperature of approximately 14-15OC. This theory was further investigated using growth histories, obtained from the otolith microstructure. As demonstrated in Figure 25, the otolith provides a powerful tool, from which the daily growth history of individual fish can be gained. The otolith offers a precise tool for determination of the effect of environmental factors on the daily growth rate of an individual. This level of detail regarding condition cannot be determined from any other

106 technique currently used. This has been illustrated for mountain galaxias alone, simply as an indication of the wide applicability of daily increments in fisheries science. With regard to the broad aims of this study, daily growth histories of all mountain galaxias sampled were averaged to provide a population response, rather than an individual response, to environmental change. There was a strong positive correlation between mean increment width and temperature, however, as the temperature rose past 14OC this relationship appears to weaken (Figures 26 and 27). This is evidence that as temperature rises the otolith growth and therefore somatic growth, of mountain galaxias larvae slows, indicating declining health of larvae. This could result in increased mortality during this critical phase, and may offer an explanation for the disappearance of larvae from samples and the relatively slow growth of younger larvae hatched later in the season when temperatures were starting to increase, both previously discussed. This theory was examined experimentally, with the results indicating that mortality of mountain galaxias larvae and water temperature are positively correlated (Chapter 8). Another study, which used cell cycle analysis to determine condition of mountain galaxias larvae at various temperatures, had variable results (Bromhead et al., 2000). In one experiment, brain cell division of larvae at 20 C was lower than in larvae held at 12OC, in a second experiment this finding was reversed. The authors suggest that the results of the first experiment were affected by the high mortality of larvae in this treatment, possibly due to disease or temperature shock, further evidence that mountain galaxias larvae may be negatively affected by increased water temperatures. Spawning activity of mountain galaxias occurred at average daily water temperatures between 11 and 12 C in both seasons sampled, similar temperatures to that previously found (O'Connor and Koehn, 1991). There was a pattern between moon phase and spawning, with the onset of spawning periods occurring within days of the new moon in both years sampled. However, the strongest correlation was with flow, with spawning periods and peaks within these periods occurring immediately after increases in flow (Figures 28 and 29). The

107 environmental flow release in the Murrumbidgee in early September corresponded to the highest peaks in spawning of mountain galaxias calculated in this study. Conclusions The change in relative abundance of mountain galaxias between and , noted earlier in this section, could be due to changes in the level of spawning activity and/or larval survival. The presence of larvae in alone provides positive evidence that spawning did occur. However, it is impossible from the data collected in this study to compare the levels of spawning activity between years. The effect of the environmental flow release on spawning activity in the Murrumbidgee in , while accepted as being positive, may not have been the only reason for increased abundance in The results of this study have indicated that there is differential growth and survival of larvae with changing temperatures (Figure 27; Chapter 8). Given the warmer water temperatures experienced in and the fact they occurred earlier in the season (Chapter 3), survival of larvae may have been poor in In , the cooler temperatures over more of the season may have positively affected larval survival, thus increasing the relative abundance between sampling years. By spawning in response to flow peaks, adults effectively match the appearance of larvae with relatively cool water temperatures and therefore increase the chances of larval survival. There could be evolutionary pressure placed on mountain galaxias in a river like the Murrumbidgee, which is not characteristic habitat for this species and can warm considerably during summer. Differential survival of the offspring from those fish which spawn in response to flows may result in this trait becoming more dominant in the mountain galaxias population of the Murrumbidgee River.

108 5. Carp (Cyprinus carpio) 5.1 Introduction Carp (Cyprinus carpio) is a native of Asia. This species has been spread by humans throughout the world and now appears on every continent except Antarctica, making it the world's most widely distributed freshwater fish (McDowall, 1996). Four strains of carp have been released into Australia. The two strains that cause the most concern to fisheries managers are the 'Boolara' or River carp and the ornamental Koi carp (Driver et al., 1997). The 'Boolara' stain was imported for aquaculture in Victoria in Some fish escaped into Lake Hawthorn near Mildura in These fish bred and spread rapidly up the Murray and Darling Rivers, interbreeding with the Koi strain in the Mumbidgee to produce a stock of carp with a broad genetic make-up. Their spread was hastened by floods, especially during the 1970's. It is now estimated that across the Murray- Darling Basin, carp constitute 80% of the fish biomass (MDBC, 2002b). Carp are found in a diverse array of habitats, however the optimal habitat for carp is considered to be low-altitude, slow-flowing waters with access to shallow, vegetated habitat for spawning (Driver et al., 1997). Carp have a high tolerance of low oxygen levels, an adaptation that allows them to inhabit almost stagnant waters from which other fish are excluded (McDowall, 1996). Carp mature early, males at 1 year, females at 2 years. Fecundity is high, a 6kg female may spawn up to 1.5 million eggs. Spawning time depends on water temperature, but usually occurs in spring with the water temperature between C. Eggs are deposited on plant material in shallow water. They hatch after a few days and the young can grow rapidly in warm waters with prolific plankton (McDowall, 1996). When present in high densities, carp contribute to the loss of native fish species, reductions in aquatic macrophyte abundance and the deterioration of water quality, especially elevated turbidity (Gehrke and Harris, 1994; Roberts et al., 1995; Driver et al., 1997). The natural

109 tolerances of carp make them well equipped to thrive in areas where human impacts are high. Indeed, carp dominate the fish community in many regulated reaches in the Murray-Darling Basin, while the proportion of native species improves in unregulated reaches (Gehrke and Harris, 2001). Gaining an understanding of the early life history processes of a pest species such as carp could offer potential solutions for control. For example, work has been done on using river flow rate manipulation as a method of carp control (Wilson, 2000). In a pond experiment, carp were induced to spawn with an increase in water level, accessing the shallow, newly inundated areas to lay their eggs. The pond was then drawn-down, desiccating and killing the eggs. The most detailed study carried out to date on larval carp looked at the age, growth and cohort composition of young-of-the-year carp from the River Murray in South Australia (Vilizzi, 1998). In this study, daily increment formation was validated in carp for the first five weeks of life. This study also showed that the mean peaks of carp spawning activity in the lower Murray occurred in October and November. However, these spawning peaks were not linked to environmental factors. The present study takes this earlier research one step further, by investigating the environmental triggers for spawning and the effect of different physical river conditions on the growth and condition of carp larvae in an upland river system. 5.2 Methods Aside from the species specific methods detailed below, all other methods are as outlined in Chapter 3, section 3.2. As carp are ostariophysan fish, the position and size of the otoliths are different to that seen in other teleosts (Figure 30, Secor et al., 1992). In ostariophysan fish the astericus is the largest otolith, followed by the lapillus and then the sagitta. Secor et al. (1992) expected that studies on ostariophysan fishes would use the asteriscus, however, most previous studies (including Vilizzi, 1998) have used the lapillus.

110 Figure 30: Relative size and position of otoliths in representative teleosts (T) and ostariophysan fish such as carp (0). Labels refer to otoliths as follows: Sagitta (Sag), Lapillus (Lap) and Asteriscus (Ast). Modified from Secor et al., All three pairs of otoliths were extracted from 10 carp larvae and analysed with the image analysis system. Only the lapillus was found to be usehl in determining age and differentiating daily increment widths (Figure 31). Lapilli were the only otoliths removed fiom the remainder of the carp larvae. The left and right lapilli of 10 individuals were examined to observe variations. There was no significant difference in increment counts (paired t-test, t = -0.6, p = 0.59, Appendix 5) so only one lapillus was analysed from each carp larva.

111 Figure 3 1: Lapillus of a carp larva (TL = 6.5 mm) as seen through the image analysis system. Normal light, loox magnification with oil. Daily increment formation in carp larvae has been validated (Vilizzi, 1998) and is again proved in this study (Chapter 8) and under the assumption that the first increment is laid down at hatch (Vilizzi, 1998), ages were calculated fi-om counts of otolith increments.

112 5.3 Results Carp larvae were sampled in the Goodradigbee and Murmmbidgee Rivers (Table 3 and Figure 11). Carp were the most abundant larvae sampled in the Goodradigbee River in and the second most abundant larvae sampled in the Murmmbidgee in Larvae from both rivers were used for age and growth investigations. In no carp were found in the Goodradigbee River and only two individuals were sampled from the Mumunbidgee (Figure 11). The remainder of this section will focus on larvae sampled in In ,75 carp were sampled in the Murmmbidgree River and 113 carp were sampled in the Goodradigbee River (Table 3). Of these fish, 74 were removed from the sample because; they were badly damaged and could not be measured accurately or otoliths were lost or broken during extraction. 59 fish sampled in the Goodradigbee River and 55 fish sampled in the Murmmbidgee River were suitable for further analysis. Goodradigbee River a) Length data The standard length of carp larvae captured in the Goodradigbee in ranged from 5.5 mm to 18.5 mm, with an average of 6.6 +I- 1.2 mm (Figure 32). The maximum otolith radius was linearly related with standard length (SLY Figure 33) for carp between 5.5 and 18.5 mm standard length from the Goodradigbee River in : Radius = SL , (r2 = 0.96) If the single point at 18.5 mrn is removed from the analysis the relationship weakens considerably, r2 = 0.2.

113 b) Age data The relationship of standard length and age was also linear (Figure 34) for carp between 5.5 mm and 18.5 mm standard length from the Goodradigbee River in : SL = age , (3 = 0.95) If the single point at 18.5 mm is removed from the analysis, the strength of the relationship is reduced, r2 = 0.7. The average age of carp sampled from the Goodradigbee River in was 13 +I- 5 days (range from 9 to 49 days). Standard length (rnrn) Figure 32: Frequency histogram of standard lengths (mm) of carp larvae captured in the Goodradigbee River during

114 0 0 I I I I Standard length (rnm) Figure 33: Relationship between maximum otolith radius (pm) and standard length (mm) of carp larvae collected from the Goodradigbee River in

115 Figure 34: Relationship between standard length (mm) and age (days) of larval carp collected from the Goodradigbee River in c) Hatch dates Hatch dates were back-calculated from the ages measured. Carp larvae captured in the Goodradigbee River in hatched over a relatively short period, from 22 November 1997 to 1 December The peak hatching period occurred over three days, from 28 November to 30 November (Figure 35). d) Growth rates The daily growth rate for each fish was calculated by the following formula: Daily growth rate (mm d") = total lenpth (mm) - 3 mrn (estimated size at hatch) age (days)

116 On average carp larvae between 5.5 and 18.5 mm standard length grew 0.34 mm d-' (range from mm d-') in the Goodradigbee River in Larvae were separated into two cohorts to investigate whether larvae hatched during the peak period had different growth rates. Cohort 1 included those larvae hatched between 22 November and 27 November, cohort 2 included larvae hatched between 28 November and 1 December. Larvae in each cohort were assigned into one of three age classes, based on their age at capture: 0-10 days, days and 21 days and over. The results were then averaged to provide a mean growth rate for each age class in each cohort (Table 13). This removed the impact of age on growth, allowing the growth rates of cohorts to be compared.

117 22/11/ / /97 26/11/ / / / /97 1/12/97 Hatching date Figure 35: Number of carp larvae sampled over the hatching period in the Goodradigbee River in Table 13: Daily growth (mm d"), standard deviation and number of larvae aged (in brackets) for two cohorts of carp larvae in the Goodradigbee River, Cohort number and hatching date / / /12/97 Age at capture (days) I (19) I I (11) (29) (1) One individual was aged greater than 20 days at capture, indicating that conditions in this reach of the Goodradigbee River sampled may not favour carp survival. Analysis of Variance (ANOVA) was carried out on the day age class, to test for differences between the two cohorts. Larvae from cohort 2 had significantly higher mean daily growth than larvae from cohort 1 (ANOVA, one-way, fixed, F = 33.2, p = , Appendix 6).

118 Growth histories of larvae were also examined using daily increment widths. Although the majority of carp sampled from the Goodradigbee River were less than 20 days old, a strong positive linear relationship was found between mean increment width and mean daily water temperature (Figure 36, r2 = 0.73). During the time over which growth was analysed, the temperature remained fairly constant, averaging /- 0.9OC, therefore this relationship could not be examined at higher or lower temperatures. 2 1 I I I I I I I Mean daily water temperature ('C) Figure 36: Relationship between mean daily water temperature (OC) and mean daily increment width (pm) for carp larvae from the Goodradigbee River in

119 A weak positive correlation existed between mean daily water flow and mean daily increment width (3 = 0.21), suggesting that temperature is a more important variable in determining the growth rates of carp larvae. After 10 December 1997 only one carp was sampled; on 10 January 1998, this was the eldest fish sampled from the Goodradigbee River, aged 49 days. This period also corresponded to decreasing flows; the last major flow peak occurring on 10 December During this period the temperature of the Goodadigbee River rose sharply, reaching a maximum of 31 C (average of I- 3.4OC). e) Spawning activity The timing of spawning activity was also investigated for this species. Estimated spawning date was calculated using an incubation time of three days (McDowall, 1996). It is estimated that all spawning occurred between 19 November and 28 November 1997, with a peak spawning period between the 25 and 27 November 1997 (Figure 37). The temperature during the spawning period ranged from 17OC to 23.2OC (average /- 1.9OC). The spawning period also corresponded to a steady decline in flow, from MLIday to MLIday (average 183 +I MLIday). Spawning commenced within three days of a small flow peak. Larger flow peaks also occurred in the beginning of November, however, no spawning was detected as a result of these events. The flow peak on 1 November 1997 was associated with a mean daily water temperature of 15.9OC. During the week following this flow event the average daily water temperature was I- 1.5OC (range from OC). The spawning period commenced four days after the full moon.

120 -.- - Flow Temperature spawning activity, peak is highlighted T 25 Date Figure 37: Relationship between spawning activity of carp (shown with dashed line, peak spawning period is solid) and mean daily water temperature (OC) and mean daily flow (MLIday) in the Goodradigbee River in

121 Mumbidnee River a) Length data The standard length of carp captured in the Murmmbidgee River in ranged from 6 rnrn to mm, with an average of /- 2.9 mm (Figure 38). The maximum otolith radius was linearly related with standard length (Figure 39) for carp between 6 and 21.5 rnrn standard length in the Mumbidgee River in : Radius = SL , (3 = 0.94) b) Age data The relationship between standard length and age was also linear (Figure 40) for carp between 6 and rnrn standard length in the Mumunbidgee River in : SL = 0.496age , (3 = 0.95) The average age of carp sampled from the Murmmbidgee River in was 18 +/- 6 days (range from 7 to 39 days).

122 Standard length (mm) Figure 38: Frequency histogram of standard lengths (mm) of carp larvae captured in the Munumbidgee River during Standard length (mm) Figure 39: Relationship between maximum otolith radius (pm) and standard length (mm) of carp larvae collected from the Murrumbidgee River in

123 Figure 40: Relationship between standard length (mm) and age (days) of larval carp collected from the Murmmbidgee River in c) Hatch dates Hatch dates were back-calculated from the ages measured. Carp larvae captured in the Murmmbidgee River in hatched over a six week period, from 14 October 1997 to 3 December 1997 (Figure 41). The hatch dates separate out into two main cohorts, the first fi-om 21 October to 27 October and the second fi-om 3 November to 12 November (Figure 41). d) Growth data The daily growth rate of each fish in both cohorts was calculated as outlined in section (d) of the Goodradigbee River results.

124 On average carp larvae between 6 and 21.5 mm standard length grew 0.53 mm d-' (range from mm d-l) in the Murmmbidgee River in Growth of carp larvae in the Murmmbidgee was significantly higher than that in the Goodradigbee River (ANOVA, one-way, fixed, F = , p = 0.00, Appendix 7). Larvae for each of the two cohorts, described in section (c) above, were assigned into one of three age classes, based on their age at capture: 0-10 days, days and 30 days and over. The results were then averaged to provide a mean growth rate for each age class in each cohort (Table 14). This removed the impact of age on growth, allowing the growth rates of different cohorts to be compared. This analysis was also done for fish spawned after the second cohort (13 November to 3 December), to see if there was any effect of late hatching. Hatch date Figure 41: Number of carp larvae sampled from the hatching period in the Murmmbidgee River in

125 Table 14: Daily growth (rnm d-i), standard deviation and number of larvae aged (in brackets) for two cohorts and late hatches of carp larvae in the Murmmbidgee River, Cohort hatching date Cohort 1 21/10/97-27/10/97 Cohort /97 Late hatches 13111/97-3/12/97 Age at capture (days) / (18) / (5) / (2) / (24) 0.66 (1) / (2) Analysis of Variance (ANOVA) was carried out to test for differences between cohorts for two age classes, less than 10 days and days. In the less than 10 days old class, late hatched larvae had significantly higher average daily growth rates than larvae hatched in cohort 2 (ANOVA, one-way, fured, F = 12.14, p = 0.02, Appendix 8). In the day old class there was also a significant difference detected (ANOVA, one-way, fixed, F = 9.5, p = , Appendix 9). Apost-hoe comparison (Tukey HSD test, unequal N) showed that the late hatched larva had significantly higher growth rates than both cohorts 1 and 2 (p = 0.05 and p = 0.02 respectively, Appendix 10). However, given there was only one late hatched larva, these results should be treated cautiously. There was no significant difference in average daily growth rates between larvae from cohort 1 and cohort 2. Growth histories were also examined using daily growth increments. There was a positive correlation between mean increment width and water temperature (Figure 42, r2 = 0.61). The daily increments used in this figure corresponded to dates between 15 October 1997 and 25 November The predicted temperature during this period ranged from 12.8 to 20.7'C, with an average of /- 2.4"C.

126 2 1 I I I I I i Mean daily water temperature ec) Figure 42: Regression of mean increment width (pm) of carp larvae against mean daily water temperature (OC), as modelled, in the Murmmbidgee River in Mean daily increment width was also regressed against mean daily flow. There was a weak, negative correlation (r2 = 0.34), suggesting that temperature is a more important influence on mean daily increment width in carp larvae. e) Spawning activity The timing of spawning was also investigated for carp larvae in the Murmmbidgee River in , using a incubation time of three days (McDowall, 1996).

127 Spawning b T 25 Date Figure 43: Timing of spawning activity in carp (shown by dashed line) in relation to mean daily flow (MLlday, represented by bars) and the mean daily water temperature (as modelled, "C, represented by line) of the Murmmbidgee River in Two peak periods of spawning were identified, these are marked as 1 and 2 with a bold line. Spawning in the Munwnbidgee River in occurred over a six week period from 11 October 1997 to 30 November Predicted temperatures during the spawning period ranged from a minimum of 9.3"C to a maximum of 20.7"C. The average temperature during this period was I- 2.7"C. Spawning commenced two days after the first quarter of the moon. Two peak spawning periods were identified; October and 31 October - 9 November (Figure 43). Prior to both peak spawning periods, there was a rise in temperature which was associated with a decrease in the flow rate. Only two individuals sampled had hatch dates after 27

128 November 1997 (Figure 41). This date corresponded to the highest flow release of the season, MLIday (Figure 43). The first peak spawning period occurred two days after the full moon, while the second peak spawning commenced on the day of the new moon. The site on the Murmmbidgee River sampled during included an area of main channel and also a significant backwater habitat (Figures 5 and 6). The water temperatures used in the above analysis, refer to what temperature the main channel was predicted to be. However, many carp were sampled in this backwater habitat (30 of 55 larvae analysed). Adult carp were exhibiting typical spawning behaviour in the backwater on several sampling occasions. The narrow mouth (Figure 6) and the orientation of the backwater, which was angled away from the direction of stream flow suggest that it is unlikely larvae passively drifted into this area. Therefore, it is highlyprobable that these larvae were spawned and hatched in this backwater. If the larvae sampled from the backwater habitat are removed from the analysis, the spawning period of larvae sampled (and presumably spawned) in the main channel occurs over a short period from 3 November 1997 to 6 November 1997 (n = 25). The predicted temperature of the main channel over this period ranged from 16.7 to 18.4OC with an average of I- 0.6OC. All of the larvae sampled from the main channel had hatch dates that fell in cohort 2, only seven of the 31 fish assigned to cohort 2 were sampled fi-om the backwater. This indicates that the remainder of the spawning activity (cohort 1 and late hatched larvae) may have been confined to the backwater habitat. Hydrolab measurements of the backwater suggest that it was warmer than the main channel by an average of 4.5OC (range from 2.6-7OC), and spot measurements suggest that the temperature of the backwater ranged between 19 and 24 C over the calculated spawning period.

129 5.4 Discussion Despite carp being the most widespread and abundant fish species throughout the Murray- Darling system (MDBC, 2002b), carp larvae were only found in two of the six rivers sampled during this study, the Mumunbidgee and Goodradigbee Rivers (Chapter 2). The majority of carp larvae were sampled in , when they were the most abundant species sampled in the Goodradigbee River and the second most abundant species sampled in the Murrumbidgee River (Table 3). In the following year, , no carp were sampled from the Goodradigbee River and only two individuals were sampled in the Murrumbidgee River, causing a massive reduction in the relative abundance of carp larvae caught between sampling years (Table 9). As discussed for mountain galaxias, this dramatic change in relative abundance may be due to decreased spawning activity andlor differential survival of larvae, both of which are further discussed in this section. Age and m-owth Carp larvae sampled from the Goodradigbee River were on average, much smaller and younger than those sampled from the Murmmbidgee (Figures 34 and 40). Comparing between rivers (using the equations generated in Figures 34 and 40) a 14 day old larva had a standard length of 7.1 mm in the Goodradigbee River and 9 mm in the Murrumbidgee River. Results from a study conducted in the lower Murray (Vilizzi, 1998) indicate that 14 day old larvae range in standard length between 8 and 10 mm. Thus larvae in the Murrumbidgee have a similar developmental pattern to those found in the lower Murray, while those in the Goodradigbee River developed slower than larvae found in both the Murmmbidgee or lower Murray. The increased developmental time required by carp larvae in the Goodradigbee could have led to increased mortality of carp larvae through increased predation (Campana, 1996), a theory supported by the collection of only one juvenile carp. This indicates that longevity of carp larvae in the reach of the Goodradigbee River sampled was not good. It is also possible that carp larvae were carried downstream into Burrinjuck Reservoir. In contrast, a full size range

130 of carp larvae and juveniles were sampled from the Murmmbidgee River (Figure 38), indicating that conditions in this river in were beneficial to the survival of carp larvae, leading to successful transition to the juvenile stage. Otolith size was strongly correlated with standard length (Figures 33 and 39) indicating that changes in somatic growth of carp larvae and juveniles are likely to be reflected in the lapillus. The strong linear relationships between age and standard length (Figures 34 and 40) are consistent with the formation of daily increments in carp larvae (Chapter 8; Vilizzi, 1998). As expected from the developmental stage analysis discussed above, the daily growth rate of carp sampled from the Murmmbidgee River (0.53 mm d-l) was significantly higher than in larvae sampled from the Goodradigbee River (0.34 mm d-i). Two main reasons may account for this difference. Firstly, the size and age differences of larvae between the two rivers may have lead to differential growth rates, as generally both are positively correlated with growth. However, as the size at age comparisons detailed above illustrate, this still does not account for the difference observed. Secondly, water temperature was lower in the Goodradigbee by an average of 3OC (Chapter 3), which may have reduced metabolic activity and thus growth rates of larvae in the Goodradigbee. Carp larvae in the Goodradigbee River hatched over a nine day period in mid - late November Larvae hatched during the second half of this window had significantly higher growth than larvae hatched earlier (Table 13). This is most likely associated with the rapidly increasing water temperature of the Goodradigbee during late November and early December (Figure 19, however, it was not associated with increased survival of larvae hatched during this period. Carp larvae in the Murmmbidgee River hatched over a six week time period from mid October to early December 1997, with the majority of hatch dates separating out into two obvious cohorts (Figure 41). While the few larvae hatched late in the period (mid-november to early December) had higher average growth rates than larvae from the earlier cohorts (Table 14), overall larval growth and survival of carp was strong in (compared with ) and

131 probably resulted in strong year-class recruitment as suggested by the number of juveniles sampled. Daily increment width of carp larvae was positively correlated with water temperature in both rivers (Figures 36 and 42), the strongest relationship was demonstrated in larvae from the Goodradigbee River. In the Goodradigbee River this relationship was based on a narrow temperature range, spanning approximately 3.5OC, therefore, slight variations in temperature were sufficient to create a response in the increment width of larvae in this river. In accordance with the agellength comparisons and the average growth rate data, average daily increment width was considerably lower in the Goodradigbee than in the Mumunbidgee (Figures 36 and 42). Consistently, all data have indicated that carp larvae in the Mumunbidgee were in better condition than those in the Goodradigbee River. This is suggested to explain the apparent difference in survivability between the two rivers, illustrated by the lack of later-stage larvae and juvenile carp in the reach of the Goodradigbee River sampled. S~awninactivitv From the existing data, spawning activity in the Mumunbidgee appeared to have occurred at a remarkably low temperature (below 10 C, Figure 43), well below the suggested minimum spawning temperature of around 17OC suggested for this species (McDowall, 1996; Smith, 2003). However it is proposed that habitat differentiation was responsible for this apparent anomaly. The large backwater habitat sampled as part of the larger reach in the Murmmbidgee (Figures 5 and 6) provided carp with the opportunity to spawn at higher, more stable temperatures than those being experienced in the main channel. Indeed, carp took advantage of this opportunity and were observed exhibiting typical spawning behaviour in the backwater on several sampling occasions. This permitted carp to spawn weeks before the main channel was of a

132 suitable temperature and explains the large temporal variation seen in carp spawning in the Murrumbidgee in (Figure 43). Larvae spawned early in the spawning window were exclusively sampled (and it is suggested spawned) in the backwater habitat. When these 'backwater' larvae are removed from the analysis, the spawning period condenses to a short, five day window in early November. It is suggested that this spawning did occur in the main channel, presumably somewhere upstream of the sampling site. Temperature data supports this theory, with average daily water temperatures in the main channel rapidly climbing to 17OC at this precise time (Figure 43). In the Goodradigbee River, carp spawned over a narrow temporal window in 1997, with all spawning calculated to have occurred over a nine day period in late November. Spawning commenced when the average daily water temperature reached 17OC, with a two day peak in spawning occurring at water temperatures exceeding 20 C (Figure 37). Spawning continued, with increasing water temperatures, until the temperature peaked at 23OC and then started decreasing. Spawning in the Goodradigbee River did commence within three days of a small flow peak. Carp spawning in the lower Murray and Darling has been correlated with the downstream progression of a flow event, the same study illustrating that carp can be induced to spawn in ponds with an increase in water level (Wilson, 2000). There was also a much larger flow peak in the Goodradigbee River at the beginning of November 1997 (Figure 37), however no spawning was correlated to this event. In this instance, the temperature of the river was only slightly colder (15.9OC) and had increased to 19OC, albeit briefly, during the week following. Therefore, water temperature was in the correct range for spawning to occur. The absence of larvae sampled around this period may be due to several factors; spawning may have occurred but eggs and larvae may have suffered large mortality due to a decrease in water temperature shortly after; or carp may also have secondary spawning cues that are independent of river conditions such as moon phase. The majority of spawning in this study occurred around the full moon, apart from one peak spawning period in the Murrumbidgee River, which occurred on the new moon.

133 The results of this study strongly suggest that temperature is an over-riding factor affecting spawning in carp, with 17OC the likely threshold. There was little evidence to suggest that carp spawned after a flood peak. This is not surprising given that most flow peaks are contained within channel in the rivers sampled and therefore do not inundate shallow, vegetated areas, as they would in a lowland river system. The results also suggest that there may be secondary cues, potentially moon phase, however further investigation into this is required. The ecological benefit of carp spawning with increasing temperatures is further explored in Chapter 8, where a positive relationship was found between temperature and both growth and survival of carp larvae. Conclusions The dramatic decrease in the relative abundance of carp larvae and juveniles in both the Murmmbidgee and Goodradigbee Rivers between and is likely to have been due to the decreased water temperatures experienced during the second year. The results have shown that this would have affected the duration of the spawning window, as well as impacting the survival of larvae, as temperature is a key factor for both processes. This chapter has successfully demonstrated how differential larval growth rates can influence survival and eventual year-class strength. The poor growth of carp larvae in the Goodradigbee River was associated with poor transition to the juvenile stage, thus negatively impacting yearclass strength. In the Murmmbidgee River, strong growth of larvae was correlated to a number of larvae successfully entering the juvenile phase, thus increasing the eventual size of the year-class.

134 6. Redfin perch (Pevcafluviatilis) 6.1 Introduction Redfin perch (Percafluviatilis) belongs to the Percidae family, which includes approximately 60 species native to the northern hemisphere. It is the only species from the Percidae family known in Australia. Its natural range includes western Europe and across eastern Europe into Siberia. Redfin perch was introduced to Australia from Europe around Since then it has become widespread thoughout Victoria, New South Wales (McDowall, 1996) and southern Western Australia (Morgan et al., 2002). It has also been recorded in South Australia and Tasmania (McDowall, 1996). Redfin perch are more sensitive to environmental conditions than many other alien species,. exhibiting poor survival at temperatures exceeding 31 C (Weatherley, 1963) and salinities in excess of 10 parts per thousand (Privolnev, 1970 in Morgan et al., 2002). This separates them from more generalist alien species such as carp and gambusia who both exhibit wider tolerances to such factors (McDowall, 1996; Morgan et al., 2002). Redfin perch are often associated with still and slow flowing waters (McDowall, 1996), believed to be due to their poor swimming ability (Weatherley, 1977). The presence of aquatic vegetation is also important, as this provides spawning habitat (Weatherley, 1963; McDowall, 1996). In Australia, spawning usually occurs in spring (McDowell, 1996), with incubation time and size of larvae at hatch dependent on water temperature (Guma'a, 1978a; Guma'a, 1978b). Each female is highly fecund, releasing several hundred thousand eggs onto aquatic vegetation and submerged rocks and logs. Newly hatched larvae form shoals but become solitary as they get older (McDowall, 1996). Highly regarded as a sporting and table fish in Europe, redfin perch does not have the same reputation in Australia. Redfin perch probably have less impact upon aquatic ecosystems than other exotic species such as carp, however, they are the main host for epizootic haematopoietic necrosis virus (EHNV). Experimental work has demonstrated that a number of native fish species (including the nationally threatened Macquarie perch, silver perch and mountain

135 galaxias) are extremely susceptible to the virus, which is characterised by sudden large-scale mortality (MDBC, 2002b). Fears have also been expressed about the predatory effect redfin perch may have on some native fish including Macquarie perch (McDowall, 1996). Abundance of redfin perch has been illustrated to be highly variable both within and between populations (Heibo and Vollestad, 2002). Year class strength of perch can be determined by a number of factors including, year-to-year variations in temperature, wind, predation, and inter and intra-specific competition. Differential growth rates in redfin perch can alter the age at sexual maturity with males maturing in a range between one and three years in fast and slow growing populations respectively (Morgan et al., 2002). Such growth differences have been attributed to size and density-dependent factors, including competition and predation, with abiotic factors playing a more minor role (Holmgren and Appelberg, 2001). Morphological variations of redfin perch in response to available prey resources have also been illustrated; with mouth size, number of gill rakers and body height changing in relation to the biomass of zooplankton, macroinvertebrates and planktivorous fish present (Hjelm et al., 2000). Research undertaken on spawning cues and larval development of redfin perch in Australia (Lake, 1967a; Lake, 196%) suggested that water temperature is an important variable, with temperatures in excess of 11.5OC required to initiate spawning. Results from southern Western Australia corroborated this early work, with peak spawning taking place at temperatures between 11 and 15OC (Morgan et al., 2002). However, research from the northern hemisphere suggests spawning can commence at temperatures as low as 8OC (Guma'a, 1978a). Photoperiodism was also suggested by Lake (1967a) as being a significant factor affecting spawning in redfm perch. In terms of larval development, it was suggested that this was negatively affected by increased temperature, a factor that was seen to be related to the control in distribution of this species (Lake, 1967b). Similar work has also been carried out in the northern hemisphere, including studies on the effect of temperature on the mortality and development of eggs and the early

136 growth of young of the year redfin perch in Windermere, a large English lake (Guma'a, 1978a; Guma'a, 1978b). Redfin perch eggs were shown to be sensitive to temperature, with optimal egg survival occurring between 6 and 16OC. Guma'a's work also demonstrated a hyperbolic relationship between water temperature and length of larvae at hatching, with the largest and healthiest larvae occurring at 14OC. Several more recent studies have also found a positive relationship between juvenile growth and water temperature (Romare, 2000; Kjellman et al., 2001). It has also been suggested that growth of young of the year perch can be affected by the abundance, type and size of prey available (Romare, 2000). This is the first study to provide a detailed analysis of the otolith microstructure of larval and juvenile redfin perch in Australia. Daily increments have been used to identify hatch periods in yellow perch, a closely related species, in Canada (section 6.2, Powles and Warlen, 1988). 6.2 Methods Aside from the species specific methods detailed below, all other methods are as outlined in Chapter 3, section 3.2. I Both the sagittae and lapilli were removed from a sub-sample of 10 redfin perch. There was no significant difference between age counts from sagittae and lapilli (paired t-test, t = -0.9, 1 p = 0.4, Appendix 11). Based on this result, only the lapillus was analysed as it was smaller I and only required minimal preparation (Figure 44). Both left and right lapilli from a sub- sample of 10 redfin larvae were aged to observe any variations. There was no significant difference in increment counts (paired t-test, t = 1.4, p = 0.2, Appendix 12), so only one lapillus was analysed from each individual. Daily increment formation in redfm perch larvae has not been validated. However, daily increment formation has been validated for the closely related yellow perch (Percaflavescens) (Powles and Warlen, 1988). Powles and Warlen also found that, in the majority of yellow perch analysed, daily increment formation commenced at hatch. Based on these findings, it is assumed that increment formation in redfin perch commences at hatch and subsequent increments are laid daily.

137 system. NO& light, 40x magnification.

138 6.3 Results Redfin perch larvae and juveniles were sampled in the Murmmbidgee and Goodradigbee Rivers (Table 3 and Figure 11). In ,37 redfin perch were sampled in the Murmmbidgee River, making it the third most abundant species sampled from this river. In the Goodradigbee River, 14 individuals were sampled, making it the second most abundant fish sampled in this river. In few redfin perch were sampled, with only nine individuals sampled from the Mumbidgee River and five from the Goodradigbee River. These differential catch rates translate to large decreases in the relative abundance of redfin perch from to in both rivers (Table 9). The remainder of this section will concentrate on those individuals sampled in individuals were suitable for further analysis from the Mumbidgee River in individuals were suitable for further analysis from the Goodradigbee River in Mumunbidnee River a) Length data The standard length of redfin perch captured in the Murmmbidgee River in ranged from 10 mm to 15 mm, with an average of 12 mm (Figure 45). Developmentally, individuals sampled ranged from late stage larvae to early juveniles. The maximum otolith radius was linearly related with standard length (SL, Figure 46) for redfm perch between 10 and 15 mm standard length from the Murmmbidgee River in : Radius = SL , (8 = 0.73)

139 b) Age data The relationship of standard length and age was also linear (Figure 47) for redfin perch between 10 and 15 mm standard length in : SL = 0.29age , (? = 0.81) The average age of redfin perch sampled from the Murrumbidgee River was 32 +I- 4 days (range from 24 to 4 1 days). Standard length (mm) Figure 45: Frequency histogram of standard lengths (mm) of redfm perch captured in the Murmmbidgee River in

140 0 I I Standard length (mm) Figure 46: Relationship between maximum otolith radius (pm) and standard length (mm) of redfin perch collected from the Murmmbidgee River in

141 Figure 47: Relationship between standard length (rnrn) and age (days) of larval redfin perch collected from the Murrumbidgee River in c) Hatch dates Hatch dates were back-calculated from the ages measured. Redfin perch sampled in the Murmmbidgee River hatched over a relatively short period, from 2 October 1997 to 19 October The majority of fish sampled hatched over the 10 day period between 6 October and 16 October 1997 (28 of 3 1 larvae analysed).

142 d) Growth data The daily growth rate for each individual was calculated by the following formula (size at hatch was estimated using data from Guma'a, 1978a): Daily growth rate (mm d-') = total length (mm) - 6 mm (estimated size at hatch) On average redfin perch between 10 and 15 mm standard length grew 0.25 mm d-' (range from mm d-') in the Murmmbidgee River in Growth histories were also examined using daily increment widths. The average increment width of redfin in the Murrumbidgee River was 3.3 +/- 0.4 pm. There was a strong positive correlation between the mean daily increment width and temperature (Figure 48, r2 = 0.88). This relationship is maintained over a wide temperature range from 9.25 to 20.74'C (average /- 2.85'C). A weak negative correlation exists between mean daily water flow and mean daily increment width (2 = 0.45), suggesting that temperature is an over-riding factor influencing the condition of redfrn perch larvae.

143 1.0 I I I I I I Mean daily water temperature (OC) Figure 48: Relationship between mean daily water temperature, as modelled (OC), and mean daily increment width (pm) for redfin perch in the Murrumbidgee River in d) Spawning activity The timing of spawning activity was also investigated for this species. There is considerable variation in the literature regarding the incubation time of redfin perch and the closely related yellow perch. Powles and Warlen (1988) suggest incubation times of 7 days at 15OC and 6 days at 20 C for yellow perch. Guma'a (1978a) suggests incubation of redfin perch eggs can take from 6 to 44 days, depending on temperature: at 14.2OC eggs took 14 days to hatch. Thls is approximately double that suggested for yellow perch by Powles and Warlen (1988). It was thought more appropriate to use the data from the study on the same species, therefore, incubation times based on Guma'a (1978a) were used. The average daily water temperature in the Murmmbidgee in the days preceding hatch was 14S C, therefore it was assumed an incubation period of 14 days would be most likely.

144 The predicted period of spawning activity occurred from 18 September to 5 October 1997 and was associated with a period of base flows and increasing water temperature (Figure 49). Spawning commenced one day after the full moon. The water temperature was 13.9"C at the commencement of the spawning period and reached a peak of 16 C on 27 September The water temperature decreased steadily in the second half of the spawning period, as irrigation flows commenced. The last day of the spawning period coincided with a large flow peak of 9052 ML and a temperature of approximately 11 "C.

145 0.rl - Flow Temperature Date Figure 49: Relationship between spawning activity of redfin perch and mean daily water temperature, as modelled (OC) and mean daily flow (ML/day) in the Murrumbidgee River in Peak spawning period is highlighted with solid line. Spawning date is based on a 14 day incubation time.

146 Goodradigbee River a) Length data The standard length of redfin perch captured in the Goodradigbee River in ranged from 3 1 mm to 51 mm, with an average of 38 mm (Figure 50). All individuals sampled were classified as juveniles. The maximum otolith radius was linearly related to standard length (SL, Figure 5 1) for redfin perch between 3 1 and 5 1 mm standard length in the Goodradigbee River in : Radius = 7.436SL , (r2 = 0.82) b) Age data The relationship of standard length and age was also linear (Figure 52) for redfin perch between 3 1 and 5 1 mrn standard length in : SL = 1.3age , (? = 0.71) The average age of redfin perch sampled from the Goodradigbee River was 47 +I- 10 days (range from 36 to 65 days).

147 Standard length (mm) Figure 50: Frequency histogram of standard lengths (mm) of redfin perch juveniles captured in the Goodradigbee River during Standard length (mm) Figure 51 : Relationship between maximum otolith radius (pm) and standard length (mm) of redfin perch juveniles collected from the Goodradigbee River in

148 Figure 52: Relationship between standard length (mm) and age (days) of juvenile redfin perch collected from the Goodradigbee River in c) Hatch dates Redfin perch juveniles captured in the Goodradigbee River hatched over the period from 5 October 1997 to 29 October During this period, hatching was evenly spread with no obvious cohorts. d) Growth data The mean growth rate for each juvenile was calculated by the formula outlined for Murmmbidgee fish. On average redfin perch juveniles between 3 1 and 51 mm standard length grew 0.83 mm d-' (range from mm d-l) in the Goodradigbee River in

149 Growth of redfin perch larvae in the Goodradigbee was significantly higher than that in the Murrumbidgee River (ANOVA, one-way, fixed, F = =, p = 0.00, Appendix 13). Growth histories were also examined using daily increment widths. The average increment width of redfin from the Goodradigbee River was 8 +I- 0.7 pm. There was a weak negative correlation between the mean daily increment width and temperature (2 = 0.16). This relationship was tested over a wide temperature range from to 23.52OC. A weak positive correlation existed between mean daily water flow and mean daily increment width (r2 = 0.13), indicating that neither temperature or flow over the ranges experienced during this study were having a major impact on the condition of redfin perch juveniles in the Goodradigbee. e) Spawning activity Determination of exact spawning time in the Goodradigbee River in was difficult because of the range of temperatures experienced during the possible spawning window and the impact of temperature on incubation period. Again, based on Guma'a (1978a), incubation time would have been approximately 17 days (at 11 C). This incubation time equates to a spawning window from 19 September to 13 October 1997 (Figure 53). The onset of this spawning window occurred two days after the full moon in September 1997.

150 Estimated spawning period -v m Flow i Date Figure 53: Relationship between the calculated spawning activity of redfin perch and mean daily water temperature (OC) and mean daily flow (MLIday) in the Goodradigbee River in Spawning dates are based on an incubation time of 17 days. Predicted spawning in the Goodradigbee River in commenced at a temperature of 9.1 C after a period of decreasing temperature and flow. During the spawning period the temperature remained relatively constant, averaging I- 0.9OC. The last day of the predicted spawning window corresponded to a sharp increase in temperature from 9.6OC to 13.8OC over 24 hours (Figure 53).

151 6.4 Discussion Larval and juvenile redfm perch (Percafluviatilis) were not widespread throughout the upper Murumbidgee catchment, being found in just two of the six rivers sampled, the Murmmbidgee and Goodradigbee, during the pilot study in (Chapter 2). This species was again sampled in these two rivers in , however, their relative abundance had decreased considerably (Table 9), a pattern seen for both carp and gambusia as well. While intra-annual variations in abundance of redfin perch are not uncommon (Heibo and Vsllestad, 2002), the possible reasons behind this decline are investigated throughout this discussion. Age and growth Young larvae were not sampled in either river. In the Murmmbidgee, the average age of redfm perch sampled was 32 days, with the youngest fish sampled 24 days old. In the Goodradigbee, the average age of redfin perch sampled was 47 days, with the youngest fish sampled being 36 days old. This could be the result of two factors. Firstly, larvae could have been spawned and hatched outside of the sampling area and then as they develop may move into the sampling area. Secondly, the methods used in this study; light traps and drift nets, may not successfully capture the younger individuals. Gurna'a (1978b) reported that larval redfin perch remain planktonic in water up to five metres deep in Windermere for the first four to six weeks of life before moving to the shallow edge habitats of the lake. Therefore, it may be possible that young redfin may not be physically able to proactively enter light traps, but are not drifting either. Therefore a seine net dragged along the edge and in deeper backwaters may be the best method to sample the smaller sizes of redfin perch. There was a difference found in the length at age relationships of redfin perch sampled from the Murmmbidgee and Goodradigbee rivers. Redfin sampled from the Murrumbidgee were much smaller at a given age than those sampled from the Goodradigbee. For example, in the Murmmbidgee a 35 day old fish was 13 mrn standard length (using the equation in Figure 47). The same age fish had a standard length of 31 mrn in the Goodradigbee (using the equation in Figure 52). There is evidence from the literature that supports varying size at age relationships

152 for members of the Percidae family in the larval, juvenile and adult stages (Powles and Warlen, 1988; Morgan et al., 2002). For example, yellow perch individuals aged 35 days had standard lengths varying between 13 and 20 mm for fish captured in different lakes (Powles and Warlen, 1988). This provides some evidence that fish in the Murmmbidgee were not growing well, their size at 35 days being at the lower levels of recorded limits for similar species. Conversely, juveniles sampled in the Goodradigbee are considerably larger at a given age than other studies have suggested. The use of otolith microstructure proved to be a good measure of growth for redfin larvae and juveniles sampled in both the Murmmbidgee and Goodradigbee Rivers. There was strong, positive linear relationships between maximum otolith radius and standard length for redfin sampled from both rivers (Figures 46 and 5 I), indicating that changes in somatic growth are reflected in changes to otolith growth. Strong linear relationships also existed between standard length and age for redfm from the Murmmbidgee and Goodradigbee rivers (Figures 47 and 52). This supports the assumption that otolith increments are deposited daily in this species. Validation of daily increments for redfin perch was unable to be conducted during this study because of the low number of individuals caught during the experimental periods in and late However, daily increment formation has been validated for the closely related yellow perch (Powles and Warlen, 1988). There were large differences in the average growth rates of redfin between the two rivers. Individuals sampled from the Murmmbidgee River had an average growth rate of 0.25 mm d-l, significantly lower than that measured from the Goodradigbee River, 0.83 mm dm'. Data recalculated from Guma'a (1978b) indicated the average growth of redfin perch larvae in Windermere was approximately 0.40 mm d-'. This supports the conclusions drawn from the size at age data, that redfin in the Murrumbidgee were growing slower than expected, while in the Goodradigbee the growth of redfin perch was much higher than expected. There were also differences between redfin perch sampled in the two rivers, in average width of daily increments and the relationship of increment width to environmental factors. As expected from the average growth rate differences between the two rivers, the average

153 increment width of redfin from the Goodradigbee River was considerably higher than that from the Murrumbidgee River, 8 and 3 ym respectively. In redfin perch sampled from the Murmmbidgee, there was a strong correlation between the mean daily increment width and temperature (Figure 48). In contrast, the mean daily increment width of redfin sampled from the Goodradigbee was negatively, but weakly, correlated to water temperature. These differing relationships were observed despite a similar temperature range being used, suggesting that different factors affect growth of redfin in the Murrumbidgee and Goodradigbee Rivers. All of the data compiled for redfin suggests that the growth of redfin perch in the Mumbidgee River is slower than in both the Goodradigbee River and other studies on redfin and closely related species. This suggests they may be in poorer condition and susceptible to mortality. The strong positive correlation between mean increment width and temperature is somewhat of a surprise, however, Guma'a (1978a) also demonstrated a positive, linear relationship between embryo and larval development and temperature. Given that temperatures in the Murrumbidgee and Goodradigbee were similar during the identified incubation times (Chapter 3) larvae would have hatched at a similar time and theoretically been in similar condition (in relation to structural deformities noted with increased incubation temperature by Guma'a, 1978a). Therefore, the reason for the decreased growth rate must be due to the environment the larvae inhabited post hatch. This is not a surprising result as growth in redfin perch is thought to be highly variable and affected by numerous factors including, water temperature, population density, intra-specific competition and prey availability (Heibo and Vsllestad, 2002). Increased competition in the Murmmbidgee could be a factor in decreasing growth rates. The large number of gambusia and carp sampled in the Murmmbidgee had overlapping temporal occurrence with redfin; all were present as larvae in October 1997 (Chapters 5 and 7). While adults of these species exhibit different foraging strategies (McDowall, 1996), it is likely that during the larval and juvenile phase there would be considerable overlap in prey items. This assumes however, that food is a limiting factor in the Munumbidgee River. The increased

154 turbidity of the Mumunbidgee (Chapter 2) may have also negatively effected both prey production and acquisition of prey. Redfm perch from the Goodradigbee River are much larger at a given age than any other published data. The remarkably consistent increment counts between sagittae and lapilli (section 6.2) and the clear ring structure of redfin perch otoliths indicate that mis-calculation of age is unlikely. There was no strong relationship between growth and temperature, but, the relationship between mean increment width and temperature was slightly negative, supporting suggestions that warmer water can reduce the condition of redfin perch larvae and juveniles. However, it is unlikely that the cooler temperatures experienced in the Goodradigbee alone caused the extreme growth rates exhibited. Other possible factors could be; decreased competition - no gambusia were present in this river (Chapter 3) and the identified carp hatching period was temporally distinct, occurring later in the season (Chapter 5); increased food availability; access to prey of a particularly high nutritional quality; or low turbidity increasing the ease of prey acquisition. Interestingly, redfm would have been four to six weeks of age when carp larvae (the most abundant species) began hatching in the Goodradigbee River. Perhaps the abundance of these larvae allowed redfin to shift from a zooplankton and invertebrate diet to fish larvae, leading to a rapid increase in growth. Svawning activity There was a strong temporal similarity in calculated spawning periods, despite the use of different incubation times for each river. In the Mumunbidgee River, the estimated spawning period occurred from 18 September to 5 October Spawning of redfin perch in the Mummbidgee was not triggered by a flow event, commencing during a period of base flow at a water temperature of 13.g C (Figure 49). In the Goodradigbee River, the estimated spawning period occurred fiom 19 September to 13 October Unlike the Murmmbidgee results, spawning commenced directly after a period of decreasing water temperatures, at an average daily temperature of 9.6OC (Figure 53).

155 This similarity in hatch and spawn periods between the two rivers occurs despite differing temperature and flow regimes, and indicates that other factors such as photoperiod or moon phase could be playing a significant role in the spawning behaviour of redfin perch. Certainly, the spawning periods were closely correlated to the full moon in September There was no obvious pattern in the cessation of spawning activities, occurring after a large flow event in the Murrumbidgee and a sudden increase in temperature in the Goodradigbee. The cessation of spawning in the Murrumbidgee is not surprising given the low affinity of adult redfin to high flows, suggested to be due to their poor swimming abilities (Weatherley, 1977). However, the reason for cessation of spawning in the Goodradigbee River is not so apparent, given that temperatures were still well within the limits for this species. It may be that the sudden increase in temperature had a negative effect on developing embryos rather than changing the spawning behaviour of adults. Conclusions Intra-annual variation in abundance of redfin perch in itself is not unusual (Heibo and Vollestad, 2002). However, the reasons why the relative abundance of redfin perch decreased considerably in are not so apparent. While the water temperatures were lower, they were still well within the demonstrated spawning range of redfin perch and given that the results of this study indicate that moon phase plays a major role in initiating spawning, it is unlikely that spawning cues were not met. High river flows have been suggested to be a factor limiting the distribution of redfm perch, due to the poor swimming abilities of this species (Weatherley, 1977). Thus, the observed decrease in relative abundance observed in may be due to the high flows experienced in this year decreasing spawning activity, rather than differential survival of larvae. The slow growth rates of redfm perch larvae and juveniles in the Murrumbidgee River are thought to be due to the environment the larvae encountered post-hatch, as a number of density-dependent and density-independent factors are known to influence growth rates of

156 redfin perch (Holmgren and Appelberg, 2001; Heibo and Vsllestad, 2002). High competition with the larvae and juveniles of other species is thought to be a major factor. The extremely high growth rates observed in the Goodradigbee deserve further investigation. While the decreased temperatures experienced in this river would assist in improving the health of redfin embryos and larvae (Gurna'a, 1978a) it is unlikely this is the only factor affecting growth. Decreased competition, superior prey quality (carp larvae) and reduced turbidity are all proposed as explanatory factors, however, further research is required to determine the relationship of these factors to growth rates.

157 7. Australian smelt, western carp gudgeon and gambusia 7.1 Australian smelt (Retropinna semoni) Introduction Australian smelt (Retropinna semoni) is one of four species of the family Retropinnidae. Two retropinnids are known to Australia, the Tasmanian smelt (R. tasmanica) is known only to Tasmania. The retropinnids are closely related to the osmerid smelts of the northern hemisphere (McDowall, 1996). Australian smelt are widespread in south-eastern Australia, in both coastal and inland drainages. In the Murray-Darling Basin, they are mainly found in rivers entering from the south-east, however, they are also found in the catchment of the Cooper Creek (McDowall, 1996). They are usually associated with still or slow moving streams. While some other species of the Retropinnidae family are anadromous, migration to the sea is not an essential part of the life cycle of Australian smelt, and may not occur at all. Upstream migrations of juvenile and adult smelt (15-40 mrn long) have been observed at Torrumban-y fishway on the River Murray (McDowall, 1996). Throughout the southern part of its range, spawning of Australian smelt occurs in spring when the water temperatures exceed 15OC. A study conducted in Brisbane, Queensland found that spawning commenced in winter, but was still associated with water temperatures above 15OC, which occur earlier in this northern area (Milton and Arthington, 1985). Spawning of Australian smelt, at relatively low temperatures compared with other native species, reflects its salmonifom affinities and temperate distribution (Milton and Arthington, 1985). Some work has also been carried out on the development of the eggs and early larvae of Australian smelt, after natural breeding in an experimental pond (Milward, 1965). At temperatures between 15.5 and lg C, hatching commenced nine days after fertilization and

158 was completed within nine hours. Newly hatched larvae averaged 4.6 mm total length (Milward, 1965) Methods Aside from the species specific methods detailed below, all other methods are as outlined in Chapter 3, section 3.2. The sagittae of eight Australian smelt were extracted and analysed to determine variations. There was no significant difference in increment counts between left and right otoliths (paired t-test, t = 0.31, p = 0.76, Appendix 14) so only one sagitta from each larva was analysed (Figure 54). Figure 54: Sagitta of an Australian smelt larva (TL = 8.5 mm) as seen through the image analysis system. Normal light, 40x magnification.

159 Daily increment formation and the timing of first increment deposition has not been validated for Australian smelt. The Australian smelt families have close affinities with the northern osmerid smelts (McDowall, 1996) and daily increment formation, commencing at hatch has been proven for rainbow smelt (Osmerus mordax) (Sirois et al., 1998). Based on this, it was assumed in this study that in Australian smelt, the first increment is laid at hatch and subsequent increments are deposited daily. Therefore, ages were calculated from the number of increments counted Results Australian smelt were sampled in the Murrumbidgee, Goodradigbee and Tumut Rivers in (Table 3). All individuals sampled in the Tumut River were classified as adults. Australian smelt were also sampled in the Murmmbidgee and Goodradigbee River in (Figure 11). In the Murrumbidgee River, only four Australian smelt were sampled in , while 38 were sampled in In the Goodradigbee River, five Australian smelt were sampled in and 11 individuals were sampled in Due to the low numbers of fish sampled from the Goodradigbee, the majority of the following analysis focuses on those fish sampled from the Mumunbidgee River. While most of the analysis focuses on fish sampled in , a brief comparison is made to Australian smelt sampled in the Murmmbidgee River in and in the Goodradigbee River in both years. Length data The standard length of Australian smelt captured in the Murmmbidgee River in ranged from 6 mm to 52 mm, with an average of 14 +I- 13 mm (Figure 55). Developmentally, individuals sampled ranged from early larvae to adults. The maximum otolith radius was linearly related with standard length (SL, Figure 56) for Australian smelt between 6 and 52 mm standard length in : Radius = 11.66SL , (2 = 0.97, n = 25)

160 Age data The relationship of standard length and age was also linear (Figure 57) for Australian smelt between 6 and 23 mm standard length from the Murrumbidgee River in (note the three adult Australian smelt sampled were not aged): SL = 0.32age , (? = 0.98, n = 22) The average age of Australian smelt larvae and juveniles sampled from the Murrumbidgee River in was 22 +I- 14 days (range from 9 to 63 days). Standard length (mm) Figure 55: Frequency histogram of standard lengths (mm) of Australian smelt captured in the Mumbidgee River in

161 Standard length (mm) Figure 56: Relationship between maximum otolith radius (pm) and standard length (mm) of Australian smelt collected from the Mumunbidgee River in Figure 57: Relationship between standard length (rnm) and age (days) of Australian smelt collected from the Mumbidgee River in

162 Hatch dates Hatch dates for Australian smelt were back-calculated from the ages measured. Australian smelt captured in the Murmmbidgee River in hatched over a prolonged period, from 12 November 1998 to 9 March 1999 (Figure 58). Only two individuals were spawned in November Their identification was re-checked and confirmed to be Australian smelt. The majority of individuals hatched between 22 February 1998 and 9 March 1999 (Figure 58). Hatch date Figure 58: Distribution of hatch dates of Australian smelt sampled from the Murmmbidgee River in

163 Growth data The daily growth rate for each individual was calculated by the following formula (size at hatch taken from Milward, 1965): Daily growth rate (mm d-') = total length (mm) mm (estimated size at hatch) On average Australian smelt between 6 and 23 mm standard length grew 0.25 mm d-' (range from mm d-') in the Murrumbidgee River in Growth histories were also examined using daily increment widths. There was a weak negative linear relationshp between mean daily increment width and temperature (3 = 0.36) and a very weak positive linear relationship with mean daily flow (r2 = 0.17). Svawning; activitv The timing of spawning activity was also investigated for this species. Estimated spawning date was back-calculated from hatch date using an incubation time of nine days. This was based on data presented in Milward (1965) for Australian smelt, which indicated a hatchng time of 9 to 9.5 days at temperatures between 15 and 18OC. In Australian smelt in the Munumbidgee River spawned during two periods (Figure 59). A small spawning event took place on 3 November A more significant spawning event also occurred between 13 and 28 February These two spawning periods were quite dissimilar in terms of environmental conditions. The first spawning period commenced one day after the fill moon, when the mean daily water temperature was approximately 12.5"C and the mean daily flow was approximately 7000 MLIday. The second spawning period commenced three days before the new moon, the mean daily water temperature was approximately 22OC and the mean daily flow was approximately 2000 MLlday.

164 Date Figure 59: Relationship between spawning activity of Australian smelt in the Murmmbidgee River in (as indicated by numbers 1 and 2) and mean daily water temperature (OC) and mean daily flow (MLIday). Larvae from the Murmmbidgee in While only four Australian smelt were sampled in , some interesting comparisons can be made between the two seasons. Of the four individuals sampled, three were late larvae and one was an adult. The standard length of Australian smelt captured in ranged from 12mm to 50mm. The average age of Australian smelt sampled from the Murmmbidgee in (excluding the adult) was 3 1 days (range from 28 to 34 days).

165 The larvae sampled hatched over a narrow window, from 22 October to 28 October This corresponded to a spawning period between 13 October and 19 October The spawning period commenced three days before the full moon. The mean daily water temperature during this period was 13S C and the mean daily flow was approximately 6500 MLIday. The spawning period immediately followed a sharp increase in temperature from 9.25OC to 13.68OC over three days. Larvae from the Goodradigbee River in and Very few Australian smelt were sampled from the Goodradigbee River with five individuals sampled in and 1 1 in Larvae sampled in were all late juveniles and ranged in size from 24.5 mm to 31 rnrn standard length. Spawning dates fell into two periods, between 9 and 11 September 1997 and between 22 and 24 September The first period occurred on the first quarter of the moon, the second period correlated to the last quarter of the moon. During the earlier period, the mean daily water temperature was approximately 9OC and the mean daily flow approximately 1500 MLIday. In the second period the mean daily temperature had risen to 1 l C, with the mean daily flow approximately 1200 MLIday. Larvae sampled in were also late juveniles, ranging in size from 24 to 26.5 mm standard length. Spawning dates occurred over a narrow window from 22 to 27 September The spawning period commenced one day after the new moon. At the beginning of the spawning period the mean daily flow was approximately 1500 MLlday. On 24 September 1998 a small flow peak of approximately 5300 MLIday occurred. The spawning period occurred immediately after an increase in the water temperature, from an average of 9S C to and average of 11 C during the spawning period.

166 7.1.4 Discussion Despite being reported as a widespread and abundant species in south-eastern Australia (McDowall, 1996), Australian smelt larvae and juveniles were only sampled in two rivers in the upper Murmmbidgee catchment, the Murmmbidgee and Goodradigbee. Abundance of Australian smelt was variable between rivers and between sampling seasons. More Australian smelt were captured in the Murmmbidgee River than the Goodradigbee River. In the Murmmbidgee River the relative abundance of Australian smelt increased slightly between sampling years, from four in to 6.3 in , while it fell in the Goodradigbee River from five in to 1.8 in (Table 9). A broad size range of Australian smelt was sampled in the Murmmbidgee River, with individuals ranging developmentally from early larvae to adult. In the Goodradigbee River, Australian smelt were exclusively sampled as late juveniles. This trend was also noted with redfin perch (Chapter 6) and further supports the theory that spawning in the Goodradigbee River could be taking place further upstream, perhaps in a tributary, with fish migrating into the main stream once they have developed. The size of the otolith was strongly correlated with standard length (Figure 56) indicating that changes in somatic growth of Australian smelt are likely to be reflected in the sagitta. The strong linear relationship between age and standard length (Figure 57) also supports the assumption that otolith increments are laid down daily. Attempts were made to validate the formation of daily increments in Australian smelt (Chapter 8) however this was not successful. Daily increment formation and timing of first increment deposition (at hatch) has been validated for rainbow smelt (Osmerus mordax, Family Osmeridae) (Sirois et al., 1998), a species that belong to a family that has close affinities with the Australian smelt families (McDowall, 1996). Therefore it is appropriate to make these assumptions for Australian smelt.

167 Analysis of mean daily increment widths provided no strong relationships with either temperature or flow. However, the weak negative relationship found between daily increment width and temperature offers further support to Australian smelt having salmonifom affinities, as concluded by Milton and Arthington (1985). Spawning activity of Australian smelt in the Murmmbidgee and Goodradigbee Rivers did not follow the pattern suggested by the literature of spawning occurring at temperatures exceeding 15OC (Milton and Arthington, 1985). There was no relationship between spawning period and moon phase, with spawning commencing at varying times between the new and full moon. In the Goodradigbee River, estimated spawning periods occurred in September in both sampling years, at water temperatures between 9 and 11 C. In the Murmmbidgee River estimated spawning periods were highly variable, occurring in October 1997 at a water temperature of 13S C, November 1998 at a water temperature of 12.5"C and February 1999 at a water temperature of 22OC. Interestingly, the spawning event in February 1999 produced more larvae than any other Australian smelt sampling event in either or In all cases, spawning of Australian smelt occurred immediately after an increase in water temperature, suggesting that it may be this increase that stimulates spawning, rather than any specific threshold temperature. However, it should be noted that these conclusions were drawn from analyses of relatively small data sets. Multiple spawning periods were observed in both the Murmmbidgee and Goodradigbee Rivers in offering support to the suggestion that Australian smelt may be serial or repeat spawners (Humphries et al., 1999). Of added interest is the temporal variation seen in the spawning periods in the Murmmbidgee River in Other spawning periods could have taken place during this time, but larvae may have been displaced downstream because of the high flows experienced.

168 Conclusions The relative stability in abundance of Australian smelt in the Murrurnbidgee River, occurred despite the release of an environmental flow in September 1998, indicating that large flows alone do not necessarily result in strong recruitment and thus strong year class strength of Australian smelt. Periods of increasing temperature appeared to initiate spawning, rather than the crossing of a specific temperature threshold. Indeed, spawning was calculated to have occurred on several occasions at temperatures below the previously reported optimum of 15OC. It is curious why a species which is thought to have salmoniform affinities (Milton and Arthington, 1985), and certainly the increment width data presented in this study supports this, would have its most successful spawning event in late summer, when temperatures are at their highest. Certainly, the variety of environmental conditions under which Australian smelt can successfully spawn and recruit offers some evidence as to why this species remains widespread and abundant across its natural range.

169 7.2 Western carp gudgeon (Hypseleotris klunzingeri) Introduction The Australian gudgeons belong to the Gobiidae family, occurring in two sub-families, Eleotridinae or Butinae. Twelve species of gudgeon occur in south-eastern Australia (McDowall, 1996). Western carp gudgeon is one of seven formally described species that belong to the Hypseleotris genus of the Eleotrids (Bertozzi et al., 2000). Western carp gudgeon are tiny fish, often only reaching approximately 45 mm in total length. They are widely distributed, occurring throughout the Murray-Darling Basin and coastal drainages of south-east Queensland and north-east New South Wales (McDowall, 1996). While not important as a target species commercially or recreationally itself, the species is seen as being an important part of the ecosystem, especially as prey for larger species llke Murray cod. Given this, some early work was carried out on the spawning requirements and subsequent larval development of western carp gudegon (Lake, 1967a ; Lake, 1967b). However, at the time of Lake's research, some species of the Hypseleotris genus with overlapping distributions had not been described, so it is possible that there may be some inconsistencies with species classification (McDowall, 1996). There is still some confusion over the taxonomy of the Hypseleotris genus, with two species with a similar distribution to Hypseleotris klunzingeri remaining undescribed (McDowall, 1996; Bertozzi et al., 2000). However, juvenile and adult gudgeons sampled during this study were identified as western carp gudgeon using the identification keys in McDowall, Therefore, it is assumed that the gudgeon larvae sampled during this study were predominantly western carp gudgeon (Hypseleotris klunzingeri). While research has been carried out on other members of the Hypseleotris genus (Anderson et al., ; Auty, 1978; Konagai and Rimer, 1985), Lake's early work remains the only record of spawning and recruitment of western carp gudgeon.

170 Lake (1967a) found that western carp gudgeon spawned in earthen ponds when the temperature reached approximately 22S C, and concluded that this species required temperature cues alone to stimulate spawning. It was also noted that this species laid its eggs in extremely shallow water and thus was thought to be highly susceptible to the fluctuating water levels caused by river regulation (Lake, 1967b). Work on the development of the eggs and early larval stages of western carp gudgeon suggests that at temperatures between 18 and 23OC eggs hatched in 47 to 53 hours (Lake, 196%). The total length of the newly hatched larvae averaged 1.94 rnm (ranging from 1.76 to 2.10 mm). These data were used as a baseline for comparison with larvae caught in the Murmmbidgee River during this study. Of particular interest was whether western carp gudgeon employed different spawning strategies in a highly regulated river system as opposed to an earthen pond Methods Aside from the species specific methods detailed below, all other methods are as outlined in Chapter 3, section 3.2. Both sagittae and lapilli of 11 western carp gudgeon were extracted and analysed with the image analysis system. Due to otolith thickening, age counts could not be obtained from the sagitta. Therefore the lapillus was used (Figure 60). There was no significant difference in increment counts between left and right lapilli (paired t-test, t = -1, p = 0.34, Appendix 15) so only one lapillus from remaining western carp gudgeon was analysed.

171 Figure 60: Lapillus of a western carp gudgeon larva (SL = 17.5 mm) as seen through the image analysis system. Normal light, 25x magnification. Daily increment formation and the timing of first increment deposition has not been validated for western carp gudgeon or any related species. However, given that daily increment formation appears to occur universally in fish (Jones, 1992), it is assumed that increment formation is daily. The deposition of the first increment is assumed to occur at hatch. Therefore age was calculated from the number of increments counted.

172 7.2.3 Results In western carp gudgeon were sampled from the Murmmbidgee and Tumut Rivers (Table 3). Western carp gudgeon was the most abundant native species sampled in the Murmmbidgee River in The developmental stage of the 25 individuals sampled from the Murrumbidgee River ranged from late larvae to adults. Only two individuals were sampled from the Tumut River, both adults. In western carp gudgeon was again the most abundant native species sampled in the Murrumbidgee River, with 1717 individuals sampled (Figure 11). The developmental stage of these individuals ranged from larvae to adults. Of the 25 individuals sampled from the Murmmbidgee River in ,24 were suitable for further analysis. Due to the large number of western carp gudgeon sampled from the Murmmbidgee in , those individuals thought to be young of the year (based on their standard length) were extracted from the sample and used for otolith analysis (n = 12). The standard length of a sub-sample (n = 200) of remaining western carp gudgeon was also measured. The remainder of this analysis will focus on western carp gudgeon sampled from the Murmmbidgee River in , with a brief comparison to larvae spawned in Len& data The standard length of western carp gudgeon captured in the Murmmbidgee River in ranged from 11 mm to 22 mm, with an average of /- 3.6 mm (Figure 61). The maximum otolith radius was linearly related with standard length (SL, Figure 62) for western carp gudgeon between 1 1 and 22 mm standard length in : Radius = 10.22SL , (r2 = 0.93)

173 Aae data The relationship of standard length and age was also linear (Figure 63) for western carp gudgeon between 11 and 22 rnm standard length from the Mumunbidgee River in : SL = 0.2lage , (r2 = 0.97) Standard length (mm) Figure 6 1: Frequency histogram of standard lengths (mm) of western carp gudgeon captured in the Murrumbidgee River during

174 0 I I I I Standard length (mm) Figure 62: Relationship between maximum otolith radius (pm) and standard length (mm) of western carp gudgeon collected from the Murmmbidgee River in Figure 63: Relationship between standard length (mm) and age (days) of western carp gudgeon collected from the Murmmbidgee River in

175 Hatch dates Hatch dates were back-calculated from the ages measured. Western carp gudgeon captured in the Murmmbidgee River hatched over a long period, from 26 November 1997 to 5 January However, the majority of individuals sampled hatched after 25 December 1997 (Figure 64). Calculated hatch date Figure 64: Calculated hatch dates of western carp gudgeon sampled fiom the Murmmbidgee River in

176 Growth data The daily growth rate for each fish was calculated by the following formula (size at hatch taken from Lake, 1967a): Daily growth rate (mm d-') = total length (mm) - 2 mm (size at hatch) age (days) On average western carp gudgeon between 11 and 22 mm standard length grew mm d-' (range fi-om mm dm'). The mean growth rate of those individuals hatched prior to 25 December 1997 (average 0.31 mm d-') was compared with the mean growth rate of individuals hatched after 25 December 1997 (average 0.31 mm d-') to investigate whether there was any advantage to hatching later in the season. There was no significant difference in mean growth rates between early and late hatches (ANOVA, F = 0.01, p = 0.91, Appendix 16). Growth histories were also examined using daily increment widths. There was a relatively weak positive linear relationship between mean increment width and temperature (2 = 0.26). The lack of a stronger relationship is probably due to the narrow temperature range over which this relationship could be tested (temperature range from to 26.19"C). There was no observable relationship between mean increment width and mean daily flow. Spawning activitv The timing of spawning activity was also investigated for this species. Estimated spawning date was back-calculated from hatch date using an incubation time of 48 hours. This was based on data presented in Lake (1967a) for western carp gudgeon that illustrated a hatching time between 47 and 53 hours at temperatures between 18 and 23 C.

177 In western carp gudgeon in the Murmmbidgee River spawned over the period from 24 November 1997 to 3 January Spawning commenced five days after the new moon. The majority of individuals were spawned after the 23 December 1997, four days after the new moon. Spawning activity of western carp gudgeon in appeared to be triggered when the temperature increased to 20 C (Figure 65). Initial spawning occurred immediately after an increase in temperature from to C over seven days. A second smaller spawning event (illustrated by hatch dates in Figure 64) started on 13 December and was also closely correlated to an increase in temperature above 20 C (Figure 65). The peak spawning period also occurred directly after a period of increasing water temperature, from to 20.53OC over six days. The average daily water temperature over the entire spawning period was /- 2.56"C. The average daily water temperature during the peak spawning period was /- 1.56OC.

178 T I Flow... Spawning period T 4 & T I Figure 65: Relationship between main spawning activity of western carp gudgeon (shown with dashed line, peak spawning period is solid) and mean daily water temperature (as modelled, "C) and mean daily flow (MLIday) in the Munumbidgee River in Date

179 Fish sampled in While western carp gudgeon were sampled in the Murumbidgee River in , very few of these were young of the year (Figure 66). 0 1 I I I I I 24/7/98 12/9/ /98 21/12/98 9/2/99 3 1/3/99 Date sampled Figure 66: Standard lengths (mm) of a sub-sample of western carp gudgeon sampled from the Murumbidgee River in (n = 206). Note the smallest individuals were not sampled until February Some relatively small western carp gudgeon, around 14 to 16 mm standard length, were sampled at the beginning of the sampling period (early to mid September 1998). The smallest western carp gudgeon (standard length 12 to 14 mm) were sampled during February To determine when western carp gudgeon were spawned in , otolith analysis was carried out on the 12 individuals under 17.5 mm standard length. These individuals were sampled in both September 1998 and February 1999.

180 Age counts for individuals sampled in September 1998 ranged from 52 to 65 days. This corresponds to a spawning period between mid July and early August Temperature data were not available for this period. However, the results from a separate study also being conducted on the Murrumbidgee River suggest that the main channel temperature of the Murrumbidgee River during this period would have been between 10 and 15 C (Lugg et al., 2001). Mean daily flows during this period were approximately 600 MLIday. The second batch of small larvae were sampled in February and March Age counts for these individuals ranged from days. Calculated hatch dates corresponded to a spawning period between 24 and 27 January The water temperature during this period averaged 19OC. The spawning period commenced immediately after a rise in temperature from 16 to 19OC over two days. The mean daily flow rate during this period was 4000 ML and the moon was in the first quarter. The three calculated spawning periods of western carp gudgeon that occurred over this study are shown in Figure 67, in relation to the flow of the Mumbidgee River.

181 Date Figure 67: Mean daily flow of the Murmmbidgee River from November 1997 to March Estimated spawning periods of western carp gudgeon are marked 1,2 and 3. The number of larvae sampled from each spawning period is included below in parentheses. As spawning of western carp gudgeon in winter was not expected, the possibility that daily increment deposition had ceased and these fish had in fact been spawned the previous summer/autumn was investigated. Otolith size and length and length and age relationships were compared to the data presented in Figures 62 and 63 to determine any differences. The otolith size of fish spawned during winter was smaller at a given age than those larvae spawned in spring and summer (Figure 68). The standard length of fish spawned during winter was smaller at a given age than those larvae spawned in spring and summer (Figure 69).

182 Wmter spawned fish (n = 5) Standard length (mm) Figure 68: Comparison of relationship between standard length (mrn) and otolith size (pm) of western carp gudgeon spawned in (shown by trendline as per Figure 62) and those spawned during winter

183 Wmter spawned fish Figure 69: Comparison of relationship between age (days) and standard length (mm) of western carp gudgeon spawned in (shown by trendline as per Figure 63) and winter spawned fish.

184 7.2.4 Discussion Despite beliefs that western carp gudgeon would be extremely susceptible to river regulation because of its habit of spawning eggs in extremely shallow water (Lake, 1967b), this species was only found in the two most highly regulated rivers sampled during this study, the Mummbidgee and Tumut Rivers. Only adult western carp gudgeon were sampled from the Tumut River, however, even their existence here confirms that conditions in this river system must at least occasionally allow spawning and subsequent recruitment of this species. In the Murrumbidgee River, western carp gudgeon was the most abundant native species sampled in both years, with a good size range of individuals sampled. Highly variable river systems, with river levels constantly rising and falling, would have a poor effect on survival of western carp gudgeon eggs. However, highly regulated rivers like the Murmmbidgee and Tumut actually provide relatively stable conditions. Once irrigation releases start in spring, they are often quite constant until late summer. The analysis of the flow characteristics of the six rivers sampled in (Table 6) illustrates this, with the Tumut and Mumbidgee Rivers having the lowest co-efficient of variation for flow. This stability, along with the fact that many of the larger, predatory native species have disappeared from these river systems probably act together to benefit western carp gudgeon. Otolith size was strongly correlated with standard length (Figure 62) indicating that changes in somatic growth of western carp gudgeon are likely to be reflected in the otolith microstructure. A strong linear relationship also existed between standard length and age (Figure 63) offering support to the assumption that increment formation in western carp gudgeon occurs systematically and probably daily. Attempts were made to validate this assumption experimentally (Chapter 8), however, these were not successful. There are no published data which has validated daily increment formation in western carp gudgeon or a related species. Analysis of mean daily increment widths illustrated no strong relationships with either temperature or flow. There was a weak positive relationship with temperature, however this was only able to be tested over a narrow temperature range.

185 Calculation of spawning activity in western carp gudgeon produced a surprising result. Three spawning periods occurred between November 1997 and March 1999 (Figure 67). Spawning during the first period, between November 1997 and January 1998, was sporadic, and appeared to be correlated to periods when the water temperature increased above 20 C (Figure 65). A peak in spawning activity, towards the end of this period was associated with water temperatures approaching 24 C. A second spawning period was calculated to have occurred during winter, between mid July and early August 1998 (Figure 67). Temperature data were not collected as a part of this study over this period as it was not anticipated that any spawning would be occurring at this time of the year. However, data from another study conducted on the Murmmbidgee during 1997 suggests that the temperature at this time would have been between 10 and 15 C (Lugg et al., 2001). Flows were extremely low, averaging 600 MLJday. Data from Chapter 5 illustrated that in mid spring backwater habitats can be up to 7 C warmer than the main channel, however it is unknown whether this also occurs during winter. Therefore, it is difficult to ascertain with the available data whether these fish were spawned in response to a temperature threshold or a flow trigger. The third and final spawning period was calculated to have occurred during late January 1999 (Figure 67). The average water temperature during this period was approximately 19 C and spawning commenced immediately after a rise in temperature of 3 C over two days. Flows were quite high during this period, averaging 4000 ML/day. The first and third spawning periods support the data presented by Lake (1967a) that suggested temperature cues spawning in western carp gudgeon, although spawning in both periods commenced at lower temperatures than the 22.5"C noted by Lake, with 20 C a likely threshold. Obviously, some of the backwater habitats in which western carp gudgeon were sampled may have been some degrees warmer than the average temperature of the main channel (as illustrated in Chapter 5).

186 The mid-winter spawning event is surprising, as it may have occurred at a very low temperature, certainly below what has previously been considered as optimal for this species. The possibility that these fish were actually spawned in late summer/autumn 1998 and had 'over-wintered' was considered. This inferred that ages of these individuals had been underestimated due to the cessation of daily increment formation over winter. While there is a lack of evidence of daily increment formation ceasing with decreasing temperature in any fish species, the possibility that daily increment formation could have been interrupted was investigated. The otolith size of fish spawned in winter was smaller than expected (Figure 68). This is not surprising as growth, both somatic and otolith, would have been greatly reduced during winter due to colder water temperatures. Fish spawned in winter were also smaller for a given age (Figure 69), suggesting that daily increment formation had not ceased. If increment formation had ceased, then age would be greatly underestimated, producing fish that were larger at a given age than expected. The data supports the assertion that these fish were growing slower due to the lower temperatures experienced over winter and sheds some light on the question of whether adults were accessing a warmer habitat, such as a backwater. The slower growth rates indicate that this is unlikely. Spawning and recruitment in several species of carp gudgeon have been proposed to fall under the 'low flow recruitment hypothesis' as these species spawn in mid-summer when the probability of low flow conditions and increased temperature are high (Humphries et al., 1999). Certainly western carp gudgeon in the Mumbidgee undertake spawning at this time, however since the construction of Burrinjuck Reservoir, low flow periods over the summer period are now the exception rather than the rule. It is apparent however, that western carp gudgeon are able to overcome this, perhaps by accessing backwater habitats that are protected from the high main channel flows. It is interesting that a spawning period was calculated during winter, when the flows were low, a common feature of the current Mumbidgee flow regime. Assuming that these fish were not spawned in a backwater that was 5 to 10 C warmer than the main channel, it appears that spawning cues are now confbsed in western carp gudgeon, some individuals spawning in relation to temperature, some in relation to low flow periods.

187 Conclusions The majority of spawning in western carp gudgeon in the Murmmbidgee River appeared to be cued by average main channel temperatures of approximately 20 C, rather than flow. Certainly it is possible that the backwater habitats sampled routinely during this study were offering refuge from the high flow events common over the spawning period, as well as warmer water temperatures. While no strong relationship could be demonstrated between growth and temperature, the fmding that fish that were spawned and developed over the winterlearly spring period were smaller at a given age suggests growth in western carp gudgeon is positively correlated with temperature. The apparent occurrence of a mid-winter spawning event was surprising. It remains unclear whether western carp gudgeon are able to spawn at such low temperatures or whether this spawning was restricted to a habitat, such as a shallow backwater or tributary, that may have been considerably warmer than the main channel. The results indicate that there may be some individual differentiation in western carp gudgeon in respect to spawning cues. While the majority of individuals spawn in relation to temperature, others may be initiated to spawn by low flows. Certainly the occurrence of a mid-winter spawning of western carp gudgeon in the Murmmbidgee warrants fixther investigation.

188 7.3 Gambusia (Gambusia holbrooki) Introduction Gambusia is an introduced species, belonging to the Poeciliidae family, which includes many popular aquarium species. Poeciliids give birth to live young rather than laying eggs, thus the family is commonly referred to as 'livebearers' (McDowall, 1996). There have been several releases (both accidental and deliberate) of various poeciliids into south-eastern Australia, but by far the most widespread and abundant species is gambusia. Gambusia are widespread throughout New South Wales, South Australia and Victoria in inland and coastal drainages (McDowall, 1996). They have also been recorded in coastal drainages of Queensland (McDowall, 1996) and in the upper reaches of the Tamar Estuary in northern Tasmania (Neira, 2001). Gambusia are tiny fish, females may reach 60 mm, males only about 35 mm. It is a highly fecund species, and this along with its reputation of consuming mosquito larvae has led to its introduction into many countries. Release of gambusia for mosquito control has been poorly documented in Australia, but it is believed it was first brought into the country in 1925 and released in the Botanical Gardens in Sydney to control mosquitoes (NSW NPWS, 2002). Gambusia are reported to have been released for the purpose of mosquito control as late as the 1960s (NSW NPWS, 2002). Gambusia are often associated with warm and gently flowing waters, preferring water temperatures between 25OC and 38OC, however, they do tolerate a wide range of temperatures, surviving below ice and up to temperatures of approximately 44OC. The species is also extremely tolerant of a wide range of salinity levels, from pure freshwater to salinities approaching that of sea water (McDowall, 1996). Undisturbed river systems with naturally variable flow regimes are not favoured by gambusia, likewise, this species is not typically found in rapid flowing waters (NSW NPWS, 2002).

189 Reproduction in gambusia has been recorded as occurring at temperatures above 15.5"C (Medlen, 1951 ; Milton and Arthington, 1983). However, the reproductive cycle is thought to be governed more by photoperiod, with reproduction commencing as daylight hours increase to around 11 (Milton and Arthington, 1983). Fertilization of eggs takes place internally and the young take between three and four weeks to develop. On average a female releases about 50 larvae per spawning, and each female may deliver up to nine broods per season (McDowall, 1996). At birth, new born gambusia are approximately 6 to 7 mm long, but are highly developed, taking just another four to six weeks to reach sexual maturity (Milton and Arthington, 1983). Gambusia is now regarded as an aggressive pest. Gut content analysis of gambusia found juveniles of several native species including Australian smelt, several species of rainbowfish, southern blue-eye and fire-tailed gudgeon (Ivantsoff and Am, 1999). The NSW Scientific Committee have identified gambusia as a possible source of decline for several threatened fish species including silver perch, Murray hardyhead and oxleyan pygmy perch (NSW NPWS, 2002). It has also been documented that gambusia preys upon eggs and tadpoles of the threatened green and golden bell frog (Litoria aurea) and other frog species. Based on these findings, Predation by Gambusia holbrooki was listed, in January 1999, as a key threatening process under the Threatened Species Conservation Act Very little research has been carried out on the potential for using otolith microstructure on members of the Poeciliidae family. Work has been done on the relationship between fish somatic growth and otolith growth in adult guppies (Poecilia reticulata) (Reznick et al., 1989) however there is no literature on the periodicity of increment deposition or on the timing of first increment deposition. There is considerable variation in the suggested gestation periods of gambusia, which range from 21 to 35 days depending on temperature, season and locality (Milton and Arthington, 1983; Vargas and Sostoa, 1996). While increments are more than likely laid down daily, without an indication of the number of increments laid down before birth, determination of age is difficult. Likewise, the use of the increment width to infer growth and condition is also hard

190 to justify, as until the embryo is born they may only be reflective of conditions within the mother. There is large variation in embryo size in gambusia both between and within females, these differences remain poorly understood, only being partly explained by factors such as clutch size and maternal size (Meffe, 1987) Methods The standard length and total length of all gambusia captured were measured. Gambusia are ostariophysan fish (Figure 30) thus the two largest otoliths, believed to be the asterisci and lapilli, were removed fiom gambusia believed to be young of the year. Analysis of these otoliths illustrated that both the asteriscus and lapillus exhibited clear microstructure and could be used to determine increment counts. As the lapillus was smaller and required less preparation time it was used preferentially. Counts were made of the number of increments present on each lapillus (Figure 70). Figure 70: Lapillus of a gambusia larva (TL = 8.5 mm) as seen through the image analysis system. Normal light, 40x magnification.

191 7.3.3 Results Gambusia were only sampled fiom the Murrumbidgee River. In ,227 individuals were captured making them the most abundant species sampled in this river. Developmentally, individuals sampled ranged fiom new-born larvae to adults. In only five gambusia were captured in the Murmmbidgee River. All of these fish were classified as adults. The remainder of this analysis will focus on gambusia sampled in The standard length of gambusia sampled in the Murrumbidgee River in ranged from 6 to 25 mm, with an average of I- 5 mm. Several cohorts of the smallest fish (6 to 6.5 mm standard length) were sampled, starting in early November 1997 through until late January 1998 (Figure 71). The majority of the smaller fish sampled were captured using dip nets and light traps in the large backwater referred to in Chapter 5. Adult gambusia were also sampled from this backwater and the main channel. As discussed in Chapter 5 there was a difference in water temperature between the backwater habitat and main channel (Figure 72), with the backwater habitat being warmer by an average of 4.5OC during the entire sampling season.

192 Date Figure 71 : Standard length (mm) of gambusia sampled from the Murrumbidgee River in plotted against date sampled (n = 227). Note that the Murrumbidgee River was not sampled during the period from 17 December 1997 to 14 January 1998.

193 s r. cx Main channel Date Figure 72: Comparison of water temperatures (OC) in the main channel and backwater habitat sampled in the Murmmbidgee River in Note, temperatures for the main channel were derived using the regression model outlined in Chapter 3, temperatures for the backwater are from hydrolab measurements.

194 There was a strong linear relationship between standard length and number of otolith increments for gambusia between 6 and 12.5 mm standard length (Figure 73, r2 = 0.99). The smallest gambusia sampled (standard length 6 mm) had an average increment count of 14 +I- 1. Figure 73: Relationship between standard length (mm) and number of increments counted on the lapilli of gambusia sampled from the Murmmbidgee River in (n = 20).

195 7.3.4 Discussion Appearance of the smallest gambusia sampled (standard length of 6 mm) occurred in early November, when temperatures in the backwater were approximately 23OC. It is difficult to estimate exactly when these fish were born, as there is evidence to suggest that there is great variability in gambusia in gestation times, which range between 21 and 35 days, and embryo size both within and between females (Milton and Arthington, 1983; Meffe, 1987). Length of gambusia offspring at birth has been recorded as being between 5.7 and 7.22 mm (Milton and Arthington, 1983), suggesting that the smallest fish sampled during this study had been born just prior to capture. The youngest gambusia sampled had an average of 14 increments on the lapillus. It is unlikely that increment deposition started at birth, because after two weeks juvenile gambusia would have attained a greater size, given that they can reach sexual maturity by six weeks of age at a length of around 21 mm (McDowall, 1986). Therefore, it is likely that increment deposition occurs before birth, however, the stimulus for and exact timing of first increment formation was not precisely determined by this study. Given that temperatures in the backwater habitat of the Murrumbidgee were quite warm (Figure 72), a gestation period at the lower end of the suggested range, approximately three weeks, is probable. The youngest, presumably newly born, gambusia had on average, 14 increments. Therefore, increment formation may begin approximately one week after fertilization. The capture of fish measuring 10 mm standard length and having approximately 27 increments (Figure 73) in early November indicates that the first brood of gambusia sampled in 1997 were born around 16 October, corresponding to a photoperiod of approximately 13 hours. At this time, the temperature in the backwater, where the juveniles were sampled was approximately 20 C. The temperature of the main stream was approximately 16OC. Other studies confirm that gambusia are capable of spawning at these temperatures, with a range between 15.5 and

196 17 C being noted (Medlen, 1951; Milton and Arthington, 1983). While temperature was within published ranges for gambusia in the main stream, juveniles were exclusively sampled from the backwater habitat. As gambusia are born as highly developed juveniles, actively able to swim, it may be erroneous to assume that spawning was restricted to the backwater, as was suggested for early cohorts of carp (Chapter 5). However, it appears that the still conditions of the backwater habitat were more conducive to survival of newly born gambusia than the high flow conditions of the main stream. It has been suggested that photoperiod plays a more critical role in determining the timing of the reproductive cycle in gambusia than temperature, with reproduction commencing once daylight hours increased past 11 (Milton and Arthington, 1983). As exact age could not be determined this study is unable to either support or refute this suggestion. However, the results of this study suggest that the first brood of gambusia in 1997 were born when daylight hours had increased to approximately 13. Therefore, daylight hours were in the range between 12 and 13 hours during the 21 day gestation period. Given the apparent relationship of reproduction to photoperiod and the fact that reproduction can occur at relatively low temperatures (less than 20 C), the lack of gambusia juveniles sampled in is more likely due to poor survival of juveniles rather than reduced spawning activity. This suggests that this species is not well adapted to recruiting under high flow conditions such as those experienced in However, there is no doubt that gambusia employ a similar strategy to carp in exploiting still and slow flowing backwater habitats to enable recruitment, even if conditions in the main channel are not suitable. The assumptions underpinning the use of otolith microstructure for livebearers such as gambusia could be readily validated by laboratory rearing and examination of the otoliths of developing embryos. Certainly, otolith microstructure offers a unique method of gaining a precise understanding of reproduction and recruitment processes in gambusia. Knowledge of the factors affecting both reproduction and juvenile development of gambusia may provide a breakthrough for control of this species not yet considered. Therefore, the factors affecting increment deposition in this species warrant mher investigation.

197 8. Experimental validation of aging 8.1 Introduction In the species chapters presented to date, a number of assumptions have been made in relation to the formation of daily increments and the timing of first increment deposition. In order to verify some of these assumptions, experimental rearing of a number of different species was attempted. The main aims were to: 1. Validate daily increment formation in those species for which this has not been proven 2. Explore interesting relationships resulting from the field work component of this study A number of species were targeted for experimental work using two different methods, field caught larvae and laboratory spawned larvae (Table 15). It was necessary to have a minimum of approximately thirty larvae available for experimental work. Unfortunately, due to the variability in abundance and difficulties in capturing large numbers of larvae, only carp and mountain galaxias were available in sufficient numbers to be used experimentally. While daily increment formation in carp has previously been validated (Villizzi, 1998) the relationship between temperature and growth noted in the field was Mher explored over a wider temperature range. There are no published studies that verify that formation of increments is daily in mountain galaxias, so the main aim of the experiment was to validate this. Secondly, the interesting relationship between temperature and mean increment width discussed in Chapter 4 was further explored.

198 Table 15: Methods used to obtain sufficient quantities of larvae for experimental rearing for a number of different species. Methods Species carp Mountain galaxias Australian smelt Redfin perch Western carp gudgeon Field caught Yes Yes Yes - but high No - too few No - too few mortality of captured larvae and larvae when juveniles returned to captured Reared from - - laboratory Adults kept for No - adults Yes - several laboratory two years, but too big to successful spawning no successful keep in spawnings, spawnings aquaria but high mortality of larvae 8.2 Methods Larval mountain galaxias and Australian smelt were collected from the Murrumbidgee River, downstream of Burrinjuck Reservoir in early December They were collected using light traps in the main channel. Larval carp were collected from the Murmmbidgee River, downstream of Buninjuck Reservoir in mid December Sampling was concentrated in backwater habitats using light traps. After collection, larvae were returned to the laboratory and placed in 20 litre aquaria at the same temperature as the river or backwater. Live identification of larval mountain galaxias and Australian smelt was extremely difficult. However, the mortality of smelt larvae was very

199 high and often only mountain galaxias remained alive on return to the laboratory. Mountain galaxias larvae were collected over a three day period and then left to settle in aquaria for at least two days before having their otoliths marked. During the six day galaxiid experiment, carp larvae were collected and kept in 20 litre aquaria for a minimum of three days before having their otoliths marked. Otolith marking followed the method outlined in Vilizzi (1998). Larvae were randomly allocated to one of nine, one litre aerated jugs containing 350 mg of oxy-tetracycline (OTC) for six hours. Each jug contained 10 larvae. During the marking process, carp larvae were held at 22 C and mountain galaxias at 15 C. Each one litre jug, containing 10 larvae, was then randomly assigned to a temperature treatment, High, Medium or Low. Temperatures of the treatments were 25"C, 15 C and 10 C for High, Medium and Low treatments respectively. These temperatures were chosen based on data presented in Chapters 4 and 5, which suggested that growth of mountain galaxias slowed above a threshold temperature of approximately 14 C. For carp larvae the temperatures were chosen to firther investigate the strong linear relationship between mean increment width and temperature shown in carp larvae from both the Goodradigbee River and Mumbidgee River. In the Murrumbidgee River this relationship was demonstrated over a temperature range from 13 C to 21 C, while in the Goodradigbee River the temperature range was much narrower, between 21 C and 23.5"C. Each treatment had three replicate, 20 litre, aerated aquaria into which the larvae were placed. Treatments were housed in separate constant temperature rooms. Each aquarium was fitted with a temperature probe, measuring the water temperature every four hours. All aquaria were exposed to a 12L: 12D photoperiod. Three larvae were sampled from each replicate on Days 2,4 and 6 (mountain galaxias) and Days 2,5 and 7 (carp). Mortalities were recorded morning and afternoon. Larvae were fed twice daily with Wardley's liquid fry food at a rate of 5 rnl per aquarium.

200 At the conclusion of the experiment the total and standard length of larvae were measured and otoliths were extracted using the general procedures outlined in Chapter 3 and the species specific methods outlined in Chapter 4 for mountain galaxias and Chapter 5 for carp. Otoliths were viewed under a compound microscope under both normal and ultra-violet light to determine the position of the fluorescent increment. The number of increments after the OTC mark were counted and the width of daily increments for up to one week prior to the OTC mark and during the experimental period were measured (as per Chapter 3). Daily increment formation was tested using linear regression. The slope of the regression was tested using a variation of the model utility test (in this case the null hypothesis was: Slope = 1 (one ring is laid each day) and as the slope could have been greater or less than 1 a twotailed test was used with a significance level of 0.05) (Devore and Peck, 1997; Plaza et al., 2001; Humphrey et al., 2003). Mortality data was tested for significance using a single factor, Analysis of Variance (ANOVA, with treatment as the factor). To test for differences in mean increment width between the three treatments, daily increments for the experimental period were averaged for each individual. These data were then able to be analysed using Analysis of Covariance (ANCOVA, with standard length as the covariate), as the repeated measure effects of using single daily increment widths were removed. 8.3 Results Mountain galaxias Dailv increment formation There was good correlation between the expected number of daily increments and the observed number of increments after the OTC mark (Figure 74,? = 0.93). The slope of the

201 regression was not significantly different from 1 (d.f. 68, t = -0.48, p = 0.618, null hypothesis accepted, Appendix 17), verifying that increment formation in mountain galaxias larvae does occur daily. 0 1 I I I I I Expected number of daily increments Figure 74: Regression of expected number of daily increments against observed number of increments for mountain galaxias larvae. Larvae from High, Medium and Low treatments are combined.

202 A distinct check mark was also visible under normal light, which corresponded to the OTC band (Figure 75). Figure 75: Sagitta of a mountain galaxias larva (SL = 13.5 rnm) seen under normal light. The dark band represents a check mark laid down during OTC marking. This larvae was sacrificed six days and six increments after this mark. Relationshiv between temperature and mortality 13 larvae (equivalent to 14%) died during the experimental period. Mortalities were restricted to the Low and High treatments. A total of nine larvae died in the High treatment - three larvae on Day 1, two on Day 2, three on Day 3 and one on Day 4. This corresponded to 30% of the high treatment. A total of four larvae died in the Low treatment, all on Day 1. This corresponded to 13% of the Low treatment. The ANOVA showed that there was a significant difference in mortality between treatments (ANOVA, F = 61, p < 0.001, Appendix 18). A