Assessing the recovery of communities following removal of the introduced Eastern , Gambusia holbrooki

Z. Tonkin, J. Macdonald, D. Ramsey, A. Kaus, F. Hames, D. Crook, and A. King

2011

Arthur Rylah Institute for Environmental Research

Technical Report Series No. 232

Arthur Rylah Institute for Environmental Research Technical Series No. 232

Assessing the recovery of fish communities following removal of the introduced Eastern Gambusia, Gambusia holbrooki

Zeb Tonkin, Jed Macdonald, David Ramsey, Andrew Kaus, Fern Hames, David Crook and Alison King

Arthur Rylah Institute for Environmental Research 123 Brown Street, Heidelberg, Victoria 3084

June 2011

In partnership with the Murray–Darling Basin Authority

Native fish recovery following Eastern Gambusia removal

Report produced by: Arthur Rylah Institute for Environmental Research Department of Sustainability and Environment PO Box 137 Heidelberg, Victoria 3084 Phone (03) 9450 8600 Website: www.dse.vic.gov.au/ari © Murray-Darling Basin Authority; State of Victoria, Department of Sustainability and Environment 2012 Citation: Tonkin, Z., Macdonald, J., Ramsey, D., Kaus, A., Hames, F., Crook, D. and King, A. (2012). Assessing the recovery of fish communities following removal of the introduced eastern gambusia, Gambusia holbrooki . Arthur Rylah Institute for Environmental Research Technical Report Series No. 232. Department of Sustainability and Environment, Heidelberg, Victoria ISSN 1835-3827 (print) ISSN 1835-3835 (online) ISBN ISBN 978-1-74287-410-4 (Print)

Disclaimer: This publication may be of assistance to you but the State of Victoria and its employees do not guarantee that the publication is without flaw of any kind or is wholly appropriate for your particular purposes and therefore disclaims all liability for any error, loss or other consequence which may arise from you relying on any information in this publication. This work is shared copyright between Murray–Darling Basin Authority and Department of Sustainability and Environment. Graphical and textual information in the work (with the exception of photographs and the MDBA logo) may be stored, retrieved and reproduced in whole or in part, provided the information is not sold or used for commercial benefit and its source (Murray–Darling Basin Authority and Department of Sustainability and Environment) is acknowledged. Such reproduction includes fair dealing for the purpose of private study, research, criticism or review as permitted under the Copyright Act 1968 . Reproduction for other purposes is prohibited without prior permission of the Murray–Darling Basin Authority or the individual photographers and artists with whom copyright applies. To the extent permitted by law, the copyright holders (including its employees and consultants) exclude all liability to any person for any consequences, including but not limited to all losses, damages, costs, expenses and any other compensation, arising directly or indirectly from using this report (in part or in whole) and any information or material contained in it. The contents of this publication do not purport to represent the position of the Murray–Darling Basin Authority. They are presented to inform discussion for improved management of the Basin's natural resources.

Funding for this project was provided by the Murray-Darling Basin Authority’s Native Fish Strategy program Accessibility: If you would like to receive this publication in an accessible format, such as large print or audio, please telephone 136 186, or through the National Relay Service (NRS) using a modem or textphone/teletypewriter (TTY) by dialling 1800 555 677, or email [email protected]

Front cover photo: Ovens River wetland (background), Seine net haul of Eastern Gambusia and adult female Carp Gudgeon (inset: L-R). Authorised by: Victorian Government, Melbourne

iii Native fish recovery following Eastern Gambusia removal

Contents List of tables and figures...... vi Acknowledgements...... x Non-technical summary...... 11 Project background and objectives...... 11 Literature review ...... 11 Cross sectional study...... 12 Trial of Eastern Gambusia removal...... 13 Cost-effectiveness and logistics of Eastern Gambusia removal...... 14 Conclusions, future research and management recommendations...... 15 1 General Introduction and project objectives...... 16 2 A review of the impact of Eastern Gambusia on native of the Murray–Darling Basin 18 2.1 Eastern Gambusia: biology, ecology and distribution ...... 18 2.1.1 and morphology...... 18 2.1.2 Biology and ecology...... 19 2.1.3 Global distribution and spread into and the MDB ...... 22 2.2 Impacts of Gambusia as alien ...... 23 2.2.1 Impacts on native fishes...... 23 2.2.2 Impacts on native fishes of the MDB ...... 26 2.3 Mitigating the impacts of Eastern Gambusia ...... 29 2.4 Knowledge gaps and the current project...... 31 3 Phase 1: Cross-sectional study of wetland fish communities across the mid-Murray region of the MDB ...... 32 3.1 Introduction...... 32 3.2 Methods...... 33 3.2.1 Study area and site selection...... 33 3.2.2 Water quality and habitat assessment ...... 34 3.2.3 Fish sampling and processing...... 34 3.2.4 Statistical analysis and development of correlative models ...... 37 3.3 Results...... 39 3.3.1 Assessment of wetland fish communities of the mid-Murray River ...... 39 3.3.2 The influence of Eastern Gambusia on species occupancy and abundance ...... 40 3.3.3 The influence of Eastern Gambusia on juvenile condition of native species ...... 46 3.4 Discussion ...... 49 3.4.1 Assessment of wetland fish communities of the mid-Murray River ...... 49 3.4.2 The influence of Eastern Gambusia on species occupancy, abundance and condition ...... 54

iv Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

3.4.3 Hypotheses on the influence of Eastern Gambusia on wetland fish communities ..59 4 Phase 2: Trial of Eastern Gambusia control ...... 60 4.1 Introduction...... 60 4.2 Methodology...... 60 4.2.1 Site description and treatments...... 61 4.2.2 Removal of Eastern Gambusia ...... 64 4.2.3 Monitoring of fish biota...... 69 4.2.4 Mark-recapture assessment ...... 69 4.2.5 Analysis...... 72 4.3 Results...... 76 4.3.1 Site parameters ...... 76 4.3.2 Methodology assessment...... 79 4.3.3 Eastern Gambusia population parameters and effectiveness of removal ...... 84 4.3.4 Fish community monitoring and condition indices ...... 90 4.4 Discussion...... 103 4.4.1 Physical control of Eastern Gambusia...... 103 4.4.2 Eastern Gambusia as an invasive species: colonisation and population dynamics107 4.4.3 Fish community response to Eastern Gambusia removal...... 108 4.4.4 Conclusions ...... 111 5 Phase 3: Cost-effectiveness and logistics of Eastern Gambusia removal...... 112 5.1 Introduction...... 112 5.2 Cost-effectiveness and logistics of Eastern Gambusia removal using physical control programs ...... 113 5.3 Social benefits of Eastern Gambusia removal: participation and education...... 116 5.4 Evaluating the benefits to native fish and cost-effectiveness of controlling other potentially harmful alien species across the MDB: applicability of the processes used in the current study...... 118 6 Conclusion ...... 121 6.1 Management / Research Recommendations ...... 121 7 References...... 123 Appendices...... 137 Appendix 1: Eastern Gambusia marking trial...... 137 Appendix 2: Project communications ...... 144

v Native fish recovery following Eastern Gambusia removal

List of tables and figures List of tables

Table 2.1 Potential macro-habitat, micro-habitat and dietary niche overlaps between Eastern Gambusia and native freshwater fish of the MDB...... 27 Table 3.1 Small native fish species and all alien species recorded in floodplain environments (billabongs, anabranches and inundated floodplain) of the mid-Murray River and tributaries (Albury – Barham)...... 35 Table 3.2 Raw abundances of each species collected during the 2009/2010 cross-sectional study from each of the four regions surveyed (Gunbower, Lower Ovens, Albury (Wonga wetlands) and Barmah-Millewa/Goulburn)...... 39 Table 3.3 The relative influence of each predictor variable on abundances of Eastern Gambusia, Australian , Carp Gudgeon, Flat-headed Gudgeon and carp in the final BRT models for each species...... 43 Table 4.1. Treatment type, habitat and water quality parameters for each of experimental sites recorded at the commencement of the experiment in 2009/2010 and 2010/2011 seasons.. .. 78 Table 4.2. Raw abundances of each species collected from removal and control sites, for each trip using all removal methodologies during the field depletion experiment in season one...... 79 Table 4.4 Eastern Gambusia mark-recapture parameters collected during the start of the experiment (spring) during each of the two years (two sites in each year) and end of the experiment in the first year only (end of summer)...... 86 Table 4.5 Parameter estimates from the Bayesian SSM for Eastern Gambusia (Gh), Carp Gudgeon (Hsp), Australian Smelt and carp (Cc)...... 87 Table 4.6 Total numbers of each species collected from all sites during each year of the removal trial...... 91 Table 4.7 Total numbers of each species collected from each site for all trips and during both seasons of the experiment...... 93

List of figures

Figure 2.1 Mature female Eastern Gambusia. Photo: Tarmo Raadik...... 18 Figure 2.2 High densities of Eastern Gambusia in a shallow backwater environment...... 20 Figure 2.3 Fin-nipping and weight loss to dwarf galaxias as a result of Eastern Gambusia aggression and competition...... 25 Figure 3.2.1 Cross-sectional study site locations (red squares) sampled during February / March 2009/ 2010 including detailed locations of those sites sampled in the Lower Ovens River floodplain...... 33 Figure 3.2.2 Example of (a) low; (b) Medium; and (c) High levels of snags and/or debris in site littoral zones...... 34 Figure 3.2.3 Sweep net electrofishing snag / debris habitat in the littoral zone of a billabong on the Ovens River floodplain...... 36 Figure 3.2.4. Examples of different levels of caudal fin damage including fins scored as ‘1’ (top), ‘2’ (middle) and ‘3’ (bottom) for (a) Carp Gudgeon; and (b) Australian Smelt...... 37

vi Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Figure 3.3.1 Co-occurrence of sampled species with the presence or absence of Eastern Gambusia (Gh) and aquatic vegetation (Veg) in the wetlands...... 40 Figure 3.3.2 Probability of Eastern Gambusia occupying a wetland when aquatic vegetation is present or absent...... 41 Figure 3.3.3 The influence of different environmental covariates: wetland size (size), water temperature (temp), conductivity (cond), dissolved oxygen (DO) and turbidity (turb) on the probability of occupancy by Eastern Gambusia...... 41 Figure 3.3.4 Fitted functions describing relationships between Eastern Gambusia abundance and all predictor variables for the final BRT model...... 42 Figure 3.3.5 Fitted functions describing relationships between Australian Smelt abundance and all predictor variables for the final BRT model...... 43 Figure 3.3.6 Fitted functions describing relationships between Carp Gudgeon abundance and all predictor variables for the final BRT model...... 44 Figure 3.3.7 Fitted functions describing relationships between Flat-headed Gudgeon abundance and all predictor variables for the final BRT model...... 45 Figure 3.3.8 Fitted functions describing relationships between carp abundance and all predictor variables for the final BRT model...... 46 Figure 3.3.9 Juvenile condition of Australian Smelt, Carp Gudgeon and Flat-headed Gudgeon plotted against Eastern Gambusia abundance...... 47 Figure 3.3.10 Plots showing the probability of fin damage for Australian Smelt, Carp Gudgeon, and Flat-headed Gudgeon in relation to Eastern Gambusia abundance, the extent of aquatic vegetation and the debris load across all sampled wetlands...... 48 Figure 4.1.1 Examples of removal sites undertaken in the field depletion experiment ...... 63 Figure 4.2.1 General timeline of the experiment for each of the two years, including exercises carried out for each of the treatment (repeated and single removal) and control types (control and reference)...... 64 Figure 4.2.2 (a) Standard collapsible bait trap containing a solar light which automatically operates between dusk and dawn, and recharges throughout the day; (b) Single-wing fine-mesh fyke net...... 66 Figure 4.2.3 (a) Adult Eastern Gambusia collected during a targeted seine shot; and (b) Trial of artificial light and heat using 100 watt, halogen globe spotlights as a means of attracting Eastern Gambusia...... 68 Figure 4.2.4 Picture of adult southern pygmy-perch. Photographs were taken of rarer species for assessment of caudal fin condition ...... 69 Figure 4.2.5 (a) Field set-up for Eastern Gambusia osmotic induction marking with calcein solution with example of resulting marked fish; and (b) field-based detection of marked Eastern Gambusia with example of a marked fish alongside an unmarked individual...... 71 Figure 4.3.1 Ovens river height at Peechelba bridge between July 2009 and March 2010. The red dashed line represents the approximate river height at which experimental sites connect to adjoining river or anabranch habitats. Grey blocks represent the removal and monitoring periods of the first year of the field-depletion experiment...... 76 Figure 4.3.2 Example of site variability during the course of the project. (a) Site 2 in late October, and 3 months later at the end of January during the 2009/2010 season. (b) Hillgrove’s site at

vii Native fish recovery following Eastern Gambusia removal

the commencement of the experiment, and following localised flooding in the area resulting in site connection with adjoining creek and control site...... 77 Figure 4.3.3 Eastern Gambusia catch per unit effort (CPUE) predictions across the four methods utilised in the study for (a) all trips and; (b) trip one only...... 80 Figure 4.3.4 Catch per unit effort (CPUE) predictions from the four methods utilised in the study across all trips for (a) Carp Gudgeon; (b) Australian Smelt; and (c) southern pygmy-perch. 81 Figure 4.3.5 Catch per unit effort (CPUE + SE; fish per hour) predictions for Eastern Gambusia from each of the trapping variables applied for (a) bait traps and; (b) fyke netting (number of observations = 33) across all trips...... 82 Figure 4.3.6 Bait trap (left column) and fyke net (right column) catch per unit effort (CPUE+ SE; fish per hour) predictions for (a) Carp Gudgeon; (b) Southern pygmy-perch and; (c) Australian Smelt for each of the trapping variables across all trips...... 83 Figure 4.3.7 Mean (± SE) number of Eastern Gambusia extracted from removal sites each consecutive day of trip one during the 2009/10 and 2010/11 seasons...... 85 Table 4.3 Raw numbers of Eastern Gambusia collected from each of the removal and control sites using physical removal techniques during the first trip (Spring) of each of the two years.... 86 Figure 4.3.8 Predicted trajectories for Eastern Gambusia abundance at ‘control’ (no removal) and ‘treatment’ (removal) sites from November – March (time 1-5) predicted for 1000 simulated ‘sites’ from the SSM model (includes predictions for one additional time period)...... 87 Figure 4.3.9 (a) Female Eastern Gambusia collected during the first trip of the experiment (before the onset of spawning)...... 88 Figure 4.3.10 Length frequency histograms (% frequency) of Eastern Gambusia collected for each trip (October/November; December; January and February/March) for both the 2009/10 and 2010/11 seasons of the experiment...... 89 Figure 4.3.11 Total raw number of fish extracted (bars) and the cumulative percentage of population removed (line) based on mark-recapture data from Colbinabbin wetland over eight days of removal exercises before the onset of spawning (trip 1, season 2)...... 89 Figure 4.3.12 Examples of species collected at sites during the field removal experiment. Native species (LHS) from top to bottom – Murray–Darling , Carp Gudgeon, southern pygmy-perch and flat-headed galaxias; and alien species (RHS – carp, Eastern Gambusia, and oriental weatherloach)...... 92 Figure 4.3.13 Length frequency histograms (% frequency) for the five most common native species collected for each trip (October/November; December; January and February/March) of the experiment (both seasons combined)...... 94 Figure 4.3.14 Length frequency histograms (% frequency) of carp and goldfish collected for each trip (October/November; December; January and February/March) of the experiment (both seasons combined)...... 95 Figure 4.3.14 The observed abundance of Eastern Gambusia (Gh), Carp Gudgeon (Hsp), Australian Smelt (Rs) and carp (Cc) at each of the 4 time periods (November – February). 96 Figure 4.3.15 Trajectories of Eastern Gambusia (Gh), Carp Gudgeon (Hsp), Australian Smelt (Rs) and carp (Cc) abundance from November – February (trip 1-4) predicted by the SSM (dashed lines and small solid circles) overlaid on the observed trajectories (solid lines and open circles)...... 97 Figure 4.3.16 Mean trajectories from 1000 simulated ‘sites’ for Eastern Gambusia (Gh), Carp Gudgeon (Hsp), Australian Smelt (Rs) and carp (Cc) abundance from November – March

viii Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

(trip 1-5) predicted by the SSM including (includes predictions for one additional time period)...... 98 Figure 4.3.17 Predicted effects of different abundances of Eastern Gambusia on the population growth of Carp Gudgeon (Hsp), Australian Smelt (Rs) and carp (Cc) based on 1000 simulated ‘sites’...... 99 Figure 4.3.18 General linear model outputs (± 95% confidence intervals) examining the

probability of fin damage (Top) and; relative condition (K rel ; Bottom) of juvenile Australian Smelt, Carp Gudgeon and Flat-headed Gudgeon in relation to Eastern Gambusia abundance

(log 10 abundance) at all sites from December - February...... 101 Figure 4.3.19 Juvenile southern pygmy-perch collected in January 2010. Note the damage of the caudal fin...... 102 Figure 4.3.19 Individual southern pygmy-perch lengths (mm) collected in each month (trips 1-4) from Site 19 during the first year of the study. Fish < 30mm represent cohort of juvenile fish...... 102 Figure 5.1 Decision support tool aimed at prioritising sites for physical control or Eastern Gambusia on the basis of maximising the ecological benefits per dollar invested...... 114 Figure 5.2 Demonstration and information on Eastern Gambusia and wetland fish communities presented to Colbinabbin primary school...... 118 Figure 5.3 Template of the processes utilised in the current study to maximise the benefits to native fish arising from an alien fish removal program...... 120

ix Native fish recovery following Eastern Gambusia removal

Acknowledgements The project team thanks members of the steering committee, Heleena Bamford (MDBA), Renae Ayres, John Koehn (DSE), Dale McNeil (SARDI), Mark Lintermans (UC), Jamie Knight (DII), Dave Maynard (AMC Tasmania), Wayne Fulton and Kylie Hall (Vic DPI) for their valuable guidance of the project. Thanks also to Justin O’Mahoney, Scott Raymond, Dean Hartwell, Jason Lieschke, Joanne Kearns and David Semmens for contribution to field work; Thanks to John and Glenys Avard, for access to private land and their generous hospitality. Thanks also to Steve Saddlier and Wayne Koster for internal review of this document. This work was funded by the Native Fish Strategy section of the Murray-Darling Basin Authority, MDBA Project No. MD1043. This study was conducted under the following permits: • DSE Care and Ethics Approval (Permit No. AEC 08/07) • NSW DPI Scientific Collection Permit (Permit No. P07/0115-1.0) • Vic DPI Fisheries Collection Permit (Permit No. RP827)

x Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Summary Project background and objectives • Alien fish species have been recognised as one of eight major threats to native fish in the Murray–Darling Basin (MDB), and the control of these species is one of the key drivers of the Native Fish Strategy. There is growing evidence of detrimental impacts of Eastern Gambusia Gambusia holbrooki on native fish fauna globally, and it has been identified as one of the key alien species contributing to the decline of a number of native fish within the MDB, where it is widespread. • Understanding the ecological impacts of an alien species is an essential component of vertebrate pest management. Unfortunately, the detrimental ecological impacts of the Eastern Gambusia in the MDB remain uncertain. If we also consider the present lack of effective control options, there is an urgent need for research into the feasibility of reducing Eastern Gambusia populations to densities where measurable improvements to native fish communities can be detected. • This project addressed these research needs by integrating surveys and quantitative experimental work in natural billabong systems throughout the MDB. The specific objectives of the project were to: - review current knowledge of the impacts of Eastern Gambusia on native fishes of the MDB, - provide information on the response of native fish communities following the reduction of Eastern Gambusia populations, and - provide a framework to evaluate the feasibility and effectiveness of such control actions, and form a template for evaluating control options for other alien fishes across the MDB.

Literature review • The literature review was compiled by searches of the Web of Science, Scopus and Aquatic Sciences and Fisheries Abstract databases, the FishBase database, and from the knowledge of experts in government and non-government organisations throughout Australia. These searches found several hundred published scientific articles and unpublished documents pertaining to the process of invasion by alien fishes, the biology, ecology, distribution and ecological impacts of Gambusia species around the world, in Australia and in the MDB, and options for, and effects of mitigating such impacts. The detailed literature review has been published as MDBA Publication No. 38/09 (Macdonald and Tonkin 2008). Three of the most significant finsings are outlined here. • Sixteen of the 37 native freshwater species have major niche overlaps (habitat, diet, or both) with the Eastern Gambusia, and thus are at greatest risk. The review highlighted a number of key families, in particular the ambassids (glassfish), nannopercids (pygmy- perches), melanotaenids (), atherinids (hardyheads), eleotrids (gudgeons) and retropinnids (smelt), which make up the majority of the MDB’s wetland fish communities. • The Eastern Gambusia is likely to have contributed to the decline in the distribution or abundance (or both) of the olive perchlet, southern pygmy-perch, Murray–Darling rainbowfish and purple-spotted gudgeon. Conversely, some species that have major niche

Arthur Rylah Institute for Environmental Research 11 Native fish recovery following Eastern Gambusia removal

overlaps with Eastern Gambusia, such as Australian Smelt Retropinna semoni and Carp Gudgeons Hypseleotris spp., remain relatively widespread throughout the MDB, most likely because of their highly flexible trophic niches that buffer them from the impacts of competition. Such species are still likely to suffer localised impacts, particularly in areas of limiting resources such as habitat availability. • In addition to the direct impacts that Eastern Gambusia is likely to have on the MDB’s smaller species (occupying, at least in part, slow or still water habitats), there could be indirect impacts on larger riverine species, particularly if Eastern Gambusia creates energetic blocks within aquatic food webs. Cross-sectional study • Accurately defining the level of risk (by quantifying the potential detrimental, neutral or even positive impacts of invasion on ecosystem processes or particular species) can provide a strong platform for generating and testing hypotheses about the level of control (if any) required to achieve an acceptable level of risk. Therefore, the first phase of the project involved a broad-scale, cross-sectional assessment of wetland fish communities using standardised electrofishing and netting surveys throughout the mid-Murray region of the MDB. This information was used to: - examine how the presence or absence of Eastern Gambusia influences the probability of occupancy of native and alien fish species in wetlands across the MDB - explore patterns of co-occurrence of native fish species and Eastern Gambusia in these systems, and - quantify the relationships between Eastern Gambusia abundance and the abundance and condition of sympatric fish species in differing habitat types and with variability in environmental covariates. • A total of 93 sites were sampled over a two-year period. Carp Gudgeon (native species) and Eastern Gambusia were the dominant species recorded, comprising of 43 and 44% of the catch and occupying 90 and 83% of the sites respectively. Some shifts in fish community composition were observed when compared with past surveys, including the absence of flat-headed galaxias Galaxias rostratus at all sites, the restriction of southern pygmy-perch Nannoperca australis to the lower Ovens River floodplain, and the presence (and dominance at some sites) of Flat-headed Gudgeon grandiceps at particular Ovens River sites. • The results suggest that Eastern Gambusia does not have a negative influence on the occurrence or abundance of common species such as Carp Gudgeon and Flat-headed Gudgeon , even where Eastern Gambusia density is high. Increasing Eastern Gambusia abundance does, however, appear to decrease the body condition of juveniles of these species. There is also some indication that greater habitat complexity (in the form of aquatic vegetation and debris) limits the number of attacks of Eastern Gambusia on Carp Gudgeon, as measured by the degree of fin damage in that species. • There was a slight negative relationship between the abundances of Australian Smelt and Eastern Gambusia. In addition, an assessment of the condition of recruits of the common native species (using two indices: length–weight relationships and fin condition) indicates that there is a negative relationship between each condition index for juvenile Australian Smelt and the abundance of Eastern Gambusia. The effects on rarer species such as

12 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

pygmy-perch and rainbowfish could not be assessed because too few of these species were collected. • One of the more interesting results from the study was that juvenile Cyprinus carpio occurred in low numbers with Eastern Gambusia in the sampled wetlands. We suggest that this may be due to predatory or competitive interactions between the species, which would peak during late spring when early-stage carp larvae would be exposed to pressures from an increasing density of Eastern Gambusia. • The results of this exploratory study enabled us to develop hypotheses about how native fish might respond if Eastern Gambusia densities were reduced in wetland systems. These hypotheses were explored in greater detail in a field removal trial. Specifically, we tested the following hypotheses: - Removing Eastern Gambusia (to any level) will benefit Australian Smelt populations, as measured by (a) increases in relative abundance, and (b) improvement in morphometric condition. - Removing Eastern Gambusia (to any level) will impart minimal changes to Carp Gudgeon and Flat-headed Gudgeon abundances. - Removing Eastern Gambusia (to any level) will improve morphometric condition for carp and Flat-headed Gudgeon. - Removing Eastern Gambusia (to any level) will impart minor increases to carp abundances. - Removing Eastern Gambusia (to any level) will benefit rarer wetland species such as southern pygmy-perch, flat-headed galaxias and rainbowfish populations, as measured by (a) increases in relative abundance, and (b) improvement in morphometric condition.

Trial of Eastern Gambusia removal • In the second phase of the project, Eastern Gambusias were removed from small isolated billabongs to test the hypotheses derived from the cross-sectional study, and to provide information on control options and Eastern Gambusia population dynamics. Unfortunately, extremes in environmental variables limited the analysis and conclusions that could be drawn from the field removal trial, but it still provided important information on Eastern Gambusia removal, population dynamics and native fish responses. • Eastern Gambusias were removed over a 5–10 day period (before the onset of the Eastern Gambusia spawning season), resulting in major reductions in abundance (generally over 40%), and even eradication at several sites. This indicates that, under certain conditions, direct removal could achieve major reductions in Eastern Gambusia populations. However, determining the likelihood of success would require a thorough consideration of various aspects of a site’s hydrology, climate, habitat and size. • During this trial, Eastern Gambusia displayed an astonishing capacity to rapidly colonise habitats; a few individuals were able to re-establish populations of thousands in three or four months. The fact that the intrinsic rate of increase of Eastern Gambusia populations was far higher than even the most common native species in the region emphasises the species’ ability to out-compete native species.

13 Native fish recovery following Eastern Gambusia removal

• Most importantly, the results indicate that reductions of Eastern Gambusia abundances will result in improvements to small-bodied native fish populations. The negative impacts on the more common generalist species were relatively minimal and were probably a result of the intact nature of the floodplain wetland sites used. There was some indication that negative impacts may be far greater on species with limited trophic niches. (A large proportion of these species have already suffered major reductions in range and abundance.) This suggests that, in the short term, management and control of Eastern Gambusia should focus on sites containing these native species, or sites with highly uniform habitats (which are often degraded sites). Further field studies assessing the response of the rarer specialist species and longer-term monitoring of fish communities at more degraded sites after Eastern Gambusia removal is required to determine whether these predictions are valid. • The results also suggest that Eastern Gambusia removal may result in small increases in Common Carp populations, and it is possible that removal may have unexpected benefits for other alien species. Site-specific ecosystem function in the absence of Eastern Gambusia must therefore also be considered before undertaking a removal program.

Cost-effectiveness and logistics of Eastern Gambusia removal • The third phase identified strategies to maximise the improvement to native fish communities through Eastern Gambusia control, given a fixed budget (benefit maximisation), and to minimise the cost of achieving a defined significant improvement in the native fish community (cost minimisation). The factors influencing overall costs and the effectiveness of a physical control program at a specific site could be generalised to four simple variables relating to a site’s hydrological connectivity, ecological value, habitat complexity and size. Considering these factors in a methodical manner enables managers to identify sites where the control of Eastern Gambusia would achieve the maximum ecological benefit per dollar invested, or would minimise the cost of achieving a defined ecological benefit.

Very Low All sites Permanently connected Permanently sites facilitating constant connected immigration of pest fish Does the site facilitate permanent immigration and emigration?

High frequency Low Isolated sites Low ecological value of connection Isolated sites of low ecological value and frequent connection Is the site of high ecological value How frequently does the site connect to to adjoining habitats facilitates (species / habitat)? adjoining waterbodies? frequent immigration of pest Low frequency fish of connection

Medium High structural Isolated sites of high ecological High ecological High frequency of complexity value but frequent connection value connection to adjoining habitats. Value of investment increases with How frequently does the site connect How much structural habitat reduced structural habitat to adjoining waterbodies? does the site contain? Low structural complexity and surface area complexity

High Large surface Low frequency of High structural Isolated sites of high ecological area connection complexity value and infrequent Benefits per $ invested $ Benefits per connection to adjoining habitats. Value of investment How much structural habitat What is the size of the site? does the site contain? increases with reduced Small surface structural habitat complexity area and surface area due to a reduction in required effort; increased ability to undertake active netting methods and Low structural Large surface increased negative interaction area complexity between pest and native species due to a reduction in habitat niche partitioning. What is the size of the site? Small surface area

14 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

• We developed a simple decision support tool for managers who might be considering investing in a control program in their jurisdiction. By considering a few primary factors, the tool enables managers to assign individual sites a general rating (from very low to high) based on ecological benefit per dollar invested, ultimately enabling a decision on whether or not to employ a control program, and the sites that a program should target. If sites are likely to experience very low or low ecological benefits for a given investment, alternative mitigation activities such as habitat restoration should still be investigated. • We found that there is an alarming lack of community awareness about Eastern Gambusia. We strongly recommend that community awareness and education is incuded in all Eastern Gambusia management programs. Community participation is a possibility because the methods for removing Eastern Gambusia could easily be utilised by community bodies such as Landcare and school groups. Even if this did not result directly in immediate ecological benefits, it would still result in substantial social benefits, particularly in light of the lack of education on the threats posed by Eastern Gambusia.

Conclusions, future research and management recommendations • Understanding the ecological impacts of an alien species is an essential component of pest management, based on the concept of managing impacts rather than simply reducing numbers. This project has taken a step forward in researching the feasibility of controlling Eastern Gambusia populations to achieve measurable improvements to native fish communities. The project provided important information on Eastern Gambusia removal, population dynamics and native fish responses to the removal. • Reductions of Eastern Gambusia will result in improvements to native fish populations, but management actions should focus on sites containing native species with limited trophic niches, or sites containing highly uniform habitats, to maximise the ecological benefits. • The removal of Eastern Gambusia, if conducted under certain conditions, can be used as an effective management tool to achieve major reductions in Eastern Gambusia populations, but the degree of success in reducing the density of Eastern Gambusia and maximising the ecological response requires a thorough consideration of various aspects of a site, hydrology, ecological value, climate, habitat and size. • Further field studies to assess the response of rarer specialist species, and longer-term monitoring of fish communities across different habitat conditions after Eastern Gambusia removal, are required to refine these predictions. • Because the removal of Eastern Gambusia may have unexpected benefits for other alien fish species, ecosystem function in the absence of Eastern Gambusia must also be considered before implementing removal programs. • Including community awareness, education and participation in Eastern Gambusia management programs will result in substantial social benefits, particularly in light of the lack of general public awareness about this alien fish species. • We suggest that the simple process used in the study could also prove valuable if applied to other established alien fish species, particularly those with limited socio-economic value.

15 Native fish recovery following Eastern Gambusia removal

1 General introduction and project objectives Invasion by alien species is a primary threat to global biodiversity (Clavero and Garcia-Berthou 2005; Leprieur et al . 2008). Growth in international trade and concurrent increases in transport capacity have accelerated the rate of introduction of alien species worldwide, with freshwater ecosystems and their native fish communities being particularly susceptible (Vitousek et al . 1997; Sala et al . 2000; Marchetti et al . 2004; Gozlan 2008). Alien fishes are implicated in the displacement, reductions in abundance, distributional range and condition, local extirpation and extinction of many native fish species worldwide (see Moyle and Light 1996; Amundsen et al . 1999; Irons et al . 2007). They can alter ecosystem function, affect genetic integrity (Weigel et al . 2002) and transmit parasites and disease (Gozlan et al . 2005), potentially resulting in high ecological, economic and social damage (Pimentel et al . 2000, 2005, see Rowe et al . 2008). In the last 30 years the number of freshwater fish species translocated outside their natural range has more than doubled (see Williamson 1996), with the increase in freshwater aquaculture production identified as a major cause (De Silva 2006; Gozlan 2008). In an analysis of the Food and Agriculture Organization’s database and FishBase, Gozlan (2008) reported that 624 species of fish throughout the world have now been introduced outside their natural range. Fifty-one per cent of these introductions were for aquaculture production, 21% for the ornamental market, 12% for angling or sport fishing, and the remainder for biological control or to fill an ecological niche. Gozlan (2008) estimated the risk of ecological impact from the introduction of a freshwater fish species is less than 10% for 84% of introduced fishes. This estimate is quite similar to that provided by Simberloff (2007) or the ‘tens rule’ of Williamson (1996), which predicts that 10% of introductions will become established, and 10% of these in turn will have an ecological impact on the recipient ecosystem (but see Gherardi 2007). In their synthesis of global patterns of freshwater fish invasion, Leprieur et al . (2008) identified Australian freshwater systems as one of six major invasion hotspots where alien fish species represent more than a quarter of the total number of species present. Australian freshwater systems currently harbour 34 known alien fish species (Lintermans 2004), 11 of which are established in the Murray–Darling Basin (MDB) (Lintermans 2007). Four of these 11 (including eastern gambusia, Gambusia holbrooki ) are classified in the top eight ‘worst’ invasive fish taxa by the International Union for Conservation of Nature (Lowe et al . 2000; see also Koehn and Mackenzie 2004; Fausch 2007). The ecological consequences of invasion by these alien fishes have been a major focus of research in Australian freshwater systems (e.g. Roberts et al . 1995; Arthington and McKenzie 1997; King et al . 1997; Howe et al . 1997; Jackson et al . 2004). Yet until quite recently the potential economic and social costs and benefits of such invasions have been mostly overlooked (but see Bomford and Hart 2002; McLeod 2004; West et al . 2007). In a recent comprehensive review, Rowe et al . (2008) used a triple-bottom-line approach to assess the environmental, economic and social impacts of six species of alien fish present in Australian freshwaters, including Eastern Gambusia. The Murray–Darling Basin Authority’s Native Fish Strategy promotes the principles of integrated pest management to mitigate such threats, and recently funded a three-year study (Project MD1043: Native fish recovery following the removal of alien fish species) to explore the relative economic costs of alien species control actions in relation to the ecological benefits to native fish communities. The Arthur Rylah Institute for Environmental Research was commissioned to undertake this project, with the following objectives: 1. Review current knowledge of the impacts of Eastern Gambusia on native fishes of the MDB.

16 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

2. Provide information on the response of native fish communities following the reduction of Eastern Gambusia populations. 3. Provide a framework to evaluate the feasibility and effectiveness of such control actions and, form a template for evaluating control options for other alien fishes across the MDB.

17 Native fish recovery following Eastern Gambusia removal

2 A review of the impact of Eastern Gambusia on native fishes of the Murray–Darling Basin This chapter is a shortened version of the comprehensive review of the impact of Eastern Gambusia on native fishes of the Murray–Darling Basin that was undertaken to address the first objective of the project (Macdonald and Tonkin 2008). 2.1 Eastern Gambusia: biology, ecology and distribution There is an extensive array of literature available on Eastern Gambusia and Western Gambusia (Gambusia affinis ) because of their widespread distribution, high abundance, ease of capture and maintenance, and mixed attitudes towards them (Pyke 2005). McKay et al . (2001), Pyke (2005, 2008), Maynard et al . (2008), and Rowe et al . (2008) have presented extensive reviews on Eastern Gambusia, including its taxonomy, biology and ecology. These reviews complement numerous other publications about this species (e.g. McKay 1984; Meffe and Snelson 1989; Lintermans 2007). The following is a brief summary of this information to help understand the subsequent chapters of the report. 2.1.1 Taxonomy and morphology Family: Scientific name: Gambusia holbrooki (Girard, 1859) Common names: Eastern Gambusia, gambusia, mosquito fish, plague , top minnow Eastern Gambusia is a small (< 60 mm) poeciliid fish distinguished by a stout body, large cycloid scales and a flattened upper head with a small up-turned mouth (Karolak 2006). It has strong conical teeth and a shortened oesophagus and intestine, which are typical traits of predatory fish (Pyke 2005). The colour is usually olive-brown on the back, blue-grey on the sides and white- silver on the underside (Lintermans 2007). It is a sexually dimorphic species, with females larger and much deeper bodied than males and have a large dark spot near the vent (Figure 1). Males, which stop growing after reaching maturity, are generally slimmer and have a slender, elongated anal fin that is used in copulation (McKay et al . 2001; Karolak 2006).

Figure 2.1 Mature female Eastern Gambusia. Photo: Tarmo Raadik.

18 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

2.1.2 Biology and ecology Eastern Gambusia generally live for only a few months and die in the same season in which they mature (McKay et al . 2001). However, females maturing towards the end of the season have been recorded living up to 15 months (Cadwallader and Backhouse 1983, McKay et al . 2001), and Karolack (2006) reported a lifespan of up to three years for the species. Males and females generally mature when the body length is 17–20 mm, at one to two months of age (Pyke 2005). Growth and maturation rates are highest when water temperatures are 25–30 °C (Pyke 2005). Like most poeciliines, fertilisation in Eastern Gambusia is internal and young develop inside the mother until they are born as free swimming fish (Parenti and Rauchenberger 1989; Pyke 2005). Both male and female fish have an annual reproductive cycle with a distinct breeding season from spring until autumn, peaking during the warmest times of the year. Females can store viable sperm in their oviducts for several months, giving them considerable flexibility in the timing of egg fertilisation (Pyke 2005). A single previously fertilised female can colonise a new site, but seasonal timing and duration of spawning is strongly governed by water temperatures and day length: temperatures over 16 °C and day lengths over 12–13 hours are required to initiate spawning (Pyke 2005). Female Eastern Gambusia can have several broods in a single breeding season; large females are known to have up to nine broods per season in the wild (Milton and Arthington 1983), although 2– 5 broods is more typical (Pyke 2005). The clutch size of each brood is extremely variable and depends on age, time of season, food availability, female size, and geographic location (Pyke 2005). Clutches are typically around 50, but clutches as large as 375 and as small as one have been reported (Cadwallader and Backhouse 1983; Milton and Arthington 1983; Rowe et al . 2008). The gestation period for each brood can range from 15 to 50 days, depending on the water temperature, but is usually around 22–25 days (Milton and Arthington 1983). Because females cannot be fertilised until after a litter is released and there is a delay of 2–14 days between birth and fertilisation, broods can be produced every three to four weeks (Pyke 2005). Maglio and Rosen (1969) reported that on average, females produced a brood every 25 days. The average number of broods and clutch size of a population can be used to estimate its total life-time fecundity (Rowe et al . 2008). Ignoring factors such as predation and resource availability, Maglio and Rosen (1969) calculated that 10 adult females could produce 5 million individuals in six months. Eastern Gambusia often assumes plague proportions in favourable habitats because of its rapid breeding ability (Lintermans 2007). Rowe et al . (2008) reported large aggregations of hundreds of fish per square metre can occur in surface waters of lakes and ponds during summer (Figure 2). Much lower densities are observed in the winter, although it is unclear whether this is a result of reduced population sizes caused by mortality, or sheltering behaviour that makes the fish less observable (Pyke 2005). Like may successful alien fish species, Eastern Gambusia can exist in a range of habitat types, including large rivers, creeks, wetlands, lakes, channels and bores. It is a poor swimmer and prefers still waters to flowing waters (Rowe et al . 2008), and therefore is most commonly found in areas such as wetlands, weir pools, lakes and backwaters (Figure 3). Eastern Gambusia tends to prefer shallow areas (often less than 15 cm deep) within these macrohabitats, mostly around the littoral margins, in surface waters or among freshwater plants (Karolak 2006; Lintermans 2007; Rowe et al . 2008). Stoffels and Humphries (2003) reported that larger fish preferred the benthic areas around macrophyte beds. Water velocity barriers such as those formed by rapids, chutes and falls limit its upstream penetration because it is not able to tolerate fast-flowing areas (Rowe et al . 2008).

19 Native fish recovery following Eastern Gambusia removal

Figure 2.2 High density of Eastern Gambusia in a shallow backwater environment. Photo: Tarmo Raadik.

Eastern Gambusia is not known to undertake active migrations; individuals generally remain within relatively small areas (Pyke 2005; Rowe et al . 2008). Exceptions include downstream displacement during flooding, and perhaps seasonal movement to deeper water before the onset of winter (Pyke 2005). However, Lyon et al . (2010) reported movement of Eastern Gambusias between the main river channel and off-channel environments, with the highest numbers of fish moving during the day. It is not known whether this was an active migration or, given their diurnal feeding pattern (see below), just part of daily feeding behaviour. Eastern Gambusia have a remarkable ability to withstand adverse conditions (McKay et al . 2001). The species is extremely tolerant of poor water quality, particularly high turbidity, extremes of temperature and salinity ranges, and low dissolved oxygen (Karolak 2006), reflecting its success as an invasive species in Australian floodplain habitats. While Eastern Gambusia can tolerate a wide range of temperatures (1.8–38 °C), its preference and reproductive requirement for high temperatures (see above) indicate it is a warm water species (Pyke 2005; Rowe et al . 2008). McNeil and Closs (2007) reported Eastern Gambusia’s extreme tolerance of low dissolved oxygen levels in both controlled conditions and MDB floodplain habitats, demonstrating the species’ ability to comfortably utilise aquatic surface respiration in conditions ranging from severe hypoxia to anoxia. Those authors suggested that the species’ flattened head and upturned mouth indicate Eastern Gambusia is morphologically adapted for this aquatic surface respiration. Disturbed habitats are particularly susceptible to Eastern Gambusia invasion because of their extreme tolerance of poor water quality (Arthington et al . 1983; Kennard et al . 2005; King and Warburton 2007). This is evident in many urban systems that have suffered extreme habitat alteration. For example, habitat alteration and subsequent water pollution have contributed to the decline of native fishes and successful establishment of Eastern Gambusia in urban Brisbane waterways (Arthington et al . 1983). The construction of dams and weirs reduces water discharge and subsequent flow velocities, thus also creating additional favourable habitat for Eastern

20 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Gambusia (McKay et al . 2001). On the other hand, undisturbed lotic systems with naturally variable discharge regimes are not favoured by Eastern Gambusia, and higher river discharges reduce their populations and may even almost eliminate them (Arthington and Lloyd 1989; Arthington et al . 1990; Chapman and Warburton 2006). Eastern Gambusia exhibits both social and anti-social behaviour (Pyke 2005). On one hand it is a schooling fish, often occurring in large aggregations. On the other hand it is well known for both intraspecific and interspecific aggression towards other fish, often chasing and nipping the fins of fish much larger than itself (Lintermans 2007). The effect of this behaviour is discussed in section 3.3. Eastern Gambusia feeds during daylight hours and relies on sight to detect and attack prey (Swanson et al . 1996; McKay et al . 2001). It is primarily carnivorous (as indicated by the dentition and digestive organs), with a generalist diet that includes a range of aquatic macro-invertebrates, terrestrial insects and arachnids, and the early life stages of fish and anurans (Arthington and Marshall 1999; Ivantsoff and Aarn 1999; Stoffels and Humphries 2003; Pyke 2005). Cannabalism of young is also known to occur (Maglio and Rosen 1969; Maynard et al . 2008). Eastern Gambusia is best described as an opportunistic or generalist omnivore, as the species has also been reported to consume filamentous algae, fragments of fruit and other plant tissue (Arthington and Marshall 1999; McKay et al . 2001; Maynard et al . 2008). McDowall (1996) described Eastern Gambusia as an adaptable generalist predator than is able to vary its diet according to the available prey. For example, researchers conducting seine netting surveys in a floodplain billabong observed large female Eastern Gambusia attacking and then ingesting juvenile Carp Gudgeon that were recovering after being released after sampling (Tonkin pers. obs.). This generalist or opportunistic nature of Eastern Gambusia’s feeding habits is another reason for its success as an invasive species. The flattened head and upturned mouth of Eastern Gambusia, coupled with its tendency to occupy surface waters, indicates that a large proportion of its diet is sourced from or near the surface. Because of its relatively small mouth, its prey is typically small, but prey size increases with increasing fish size (Pyke 2005) and so larger females can be expected to have a wider diet. Although mosquito larvae are consumed by Eastern Gambusia, they make up a very small proportion of their overall diet (Arthington and Lloyd 1989). Furthermore, although Eastern Gambusia is used extensively throughout the world as a mosquito control agent, there are no striking examples of its effectiveness in reducing mosquito populations (Lloyd 1990). In its native range Eastern Gambusia is susceptible to numerous predators, parasites and diseases (Swanson et al . 1996). In Australia, however, it appears there are very few control agents. Other than the introduced copepod Lernia , few species parasitise Eastern Gambusia in Australia compared to native species, which may carry at least two or three parasite species . This relatively light parasite burden may have also contributed to its success as an invader in Australia (Lloyd 1990). There has been little work on predation of Eastern Gambusia in Australia, but predators are likely to include birds such as cormorants and egrets (e.g. Boulton and Brock 1999, cited in Rowe et al . 2008), aquatic mammals (e.g. Lloyd 1987), invertebrates such as crayfish (e.g. Beatty 2006) and fish. In particular, another introduced fish, redfin perch ( Perca fluviatilis ), is thought to prey heavily on Eastern Gambusia, and Stoffels and Humphries (2003) and McNeil (2004) suggested that the presence of redfin perch may govern the densities of Eastern Gambusia in Australian floodplain billabong habitats. The level of predation on Eastern Gambusia by these various organisms is largely unknown. For example, while the introduced redfin perch has been documented to influence Eastern Gambusia numbers by direct predation, both native and alien fish

21 Native fish recovery following Eastern Gambusia removal predators prefer other prey when given a choice, so that Eastern Gambusia could still establish large populations in the presence of large predator populations (Lloyd 1990). 2.1.3 Global distribution and spread into Australia and the MDB Eastern Gambusia is native to south-eastern United States and northern Mexico (McKay et al . 2001), but it was not until the was recognised as a potential biological control agent that it became one of the most widespread in the world. The genus is now present on all continents except Antarctica (Courtenay and Meffe 1989). After the discovery that mosquitoes were responsible for the transmission of diseases such as malaria and yellow fever, health authorities began searching for mosquito control techniques, including the new concept of biological control (Lloyd 1990; McKay et al . 2001). During the early 1900s, scattered anecdotal evidence of Eastern Gambusia successfully controlling mosquito populations led to their widespread introduction as a biological control, despite limited research into their capabilities or the utility of native species as mosquito control agents (Lloyd 1990; McKay et al . 2001). This resulted in the spread of Eastern Gambusia into freshwater environments throughout the world, particularly from 1920 to 1940, including the first introduction into Australia in 1925 (Rowe et al . 2008). Shortly after the initial introduction of Eastern Gambusia into the Botanic Gardens, Sydney, the species spread throughout New South Wales to the point that it was established in most of the state by the early 1940s (McKay et al . 2001). At this time it was released in northern , parts of the Northern Territory, South Australia, Victoria and Western Australia, all for the purpose of mosquito control (McKay 1984; Maynard et al . 2008). Eastern Gambusia was recorded in Tasmania in 1992, believed to be a result of an unauthorised release into a private dam (Maynard et al . 2008). Arthington and McKenzie (1997) reported that the species was still being spread about Australia for mosquito control during the late 1990s, despite substantial evidence that native species were more effective at mosquito reduction than Eastern Gambusia (Lloyd 1990). Most states have since adopted legislation that has meant the species is no longer advocated as a mosquito control agent (see Chapter 5.1) and therefore unlikely to be intentionally distributed. Further spread of the species is most likely to occur during widespread flooding, dispersing individuals from existing populations (e.g. McKay 1984) and from populations in irrigation channels and bore drains (McKay et al . 2001). Eastern Gambusia is now present in all states and territories, having established populations throughout most of the major drainage divisions in the country (Rowe et al . 2008). This includes most of the eastern drainage division, from Port Douglas in northern Queensland to Adelaide in South Australia (McDowall 1996; Arthington and McKenzie 1997; Rowe et al . 2008). In Western Australia they are extremely abundant in the south-west and southern Pilbara, and a small population has been discovered in an isolated region of the Kimberley (Morgan et al . 2004). Several populations have been discovered in Darwin, several eastern catchments draining into the Gulf of Carpentaria, and in central Australia’s Lake Eyre drainage region (Rowe et al . 2008). Isolated populations have also become established in northern Tasmania’s Tamar basin (Maynard et al . 2008). Eastern Gambusia is now present in all river basins of the MDB (MDBC 2008) and is common and frequently extremely abundant in farm dams, slowly flowing waters and shallow wetlands (Lintermans 2007). They have been recorded at altitudes from 20 to 1120 m, although most populations have been recorded below 300 m (Faragher and Lintermans 1997; McKay et al . 2001), which is not surprising given their preference for warm water and slow flows. The only areas where they have not been recorded are the higher altitudes in the south-eastern alpine areas, and the upper reaches of parts of the Condamine and Warrego basins (Lintermans 2007).

22 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

2.2 Impacts of Gambusia as an alien species Eastern Gambusia’s aggressive nature, high reproductive potential, fast maturation rate, flexible behaviour and broad environmental tolerances have contributed to its success as an invader, and the species poses a serious threat to native fishes in Australia and overseas (Courtney and Meffe 1989; Howe et al . 1997; Rowe et al . 2008). It can also have a detrimental effect (see Webb and Joss 1997; Gillespie and Hero 1999; Komak and Crossland 2000), influence macroninvertebrate, zooplankton and phytoplankton communities (Hurlbert et al . 1972; Margaritora et al . 2001; Angeler et al . 2007) and enhance primary productivity by increasing allochthonous nutrient loads (Hargrave 2006). Such impacts can also indirectly affect native fishes by reducing or removing available prey resources and altering physicochemical properties such as turbidity, water temperature and phosphorus cycles (Hurlbert et al . 1972). Changes to these variables may interfere with normal feeding, sheltering or reproductive strategies of native fishes and alter ecosystem level processes (Cardona 2006). Recently, Pyke (2008) completed a thorough review of the impacts of eastern and Western Gambusias on aquatic systems, giving particular attention to the documented impacts of both species on mosquitos, other invertebrates, amphibians, planktonic communities and ecosystem functions. Accordingly, we have decided to focus the following discussion on the impacts of Gambusia species on native fishes alone, and draw attention to Pyke (2008) for further information on indirect effects and ecosystem level responses, and to Gillespie and Hero (1999) and Rowe et al . (2008) for reviews on the effects of Eastern Gambusia on Australian native amphibians. A substantial portion of the literature devoted to the impacts of Gambusia species has been based on studies conducted in the USA and Australasia on Western Gambusia (e.g. Laha and Mattingly 2007; Rowe et al . 2007). Because the two species are similar in body size, diet, feeding rates, reproductive capacity and habitat requirements (Rehage et al . 2005), the mechanisms of impacts may be similar, notwithstanding the potential for such mechanisms to be strongly mediated by the local physical environment, and the diversity and abundances of fish species present (Courtney and Meffe 1989; Rowe et al . 2008). The following sections provide a global perspective on the impacts of Gambusia species on native fishes. 2.2.1 Impacts on native fishes Western and Eastern Gambusias have been introduced to more than 60 countries (Garcia-Berthou et al . 2005), and populations of Eastern Gambusia are now established in 21 countries (Froese and Pauly 2007; Fishbase). Direct predation on native fishes (Myers 1965; Lloyd et al . 1986; Laha and Mattingly 2007), competitive exclusion from food resources and habitat (Arthington et al . 1983; Howe et al . 1997; Caiola and de Sostoa 2005) and aggressive interactions in confined environments (Howe et al . 1997; Knight 1999; Laha and Mattingly 2007) have been well documented. The implications of such interactions range from reduced condition of native fishes, such as stunted growth, reduced ovarian weight and low fecundity (Howe et al . 1997; Mills et al . 2004), and increased susceptibility of individuals to secondary infection and damage to skin and fins (i.e. via fin-nipping) (Meffe et al . 1983), to mortality or more or less competitive interference- driven reductions in population size and distribution (see Galat and Roberston 1992). Combinations of these mechanisms may be acting simultaneously, and the magnitude of impacts on a given species may vary between day and night, seasonally, and with ontogeny (Mills et al . 2004; Ayala et al . 2007). The nature of interactions between Gambusia species and sympatric fish species may also depend strongly on the relative densities, and on the behaviour of the species present (e.g. Knight, 1999; Breen, 2000; Conte 2001), the availability of vacant trophic or habitat niches (e.g. Lloyd and Walker 1986; Lloyd 1987, 1990; Keller and Brown 2008), variation in physicochemical and environmental parameters (e.g. McNeil 2004; Rincón et al . 2002), and seasonal alteration to natural or managed flow regimes and associated changes to aquatic ecosystem processes (e.g. McNeil 2004; Fairfax et al . 2007).

23 Native fish recovery following Eastern Gambusia removal

Also of concern is the potential for Eastern Gambusia to reduce or fragment populations of native species to levels at which negative factors associated with small population size, such as inbreeding depression, and loss of allelic diversity and heterozygosity, exist (Rowe et al . 2008). The negative effects of inbreeding are well documented (Gall 1987; Rowe et al . 2008) and include decreases in fitness and increases in deformed offspring (Kincaid 1976), and extinction probability (Saccheri et al . 1998). In Australia, a large number of studies have indicated that Eastern Gambusia has negative ecological impacts on small-bodied native fish species, but others have documented no detrimental effects; see Macdonald and Tonkin (2008) for list of references. Rowe et al . (2008) identified 23 native species (for which published information was available) that were adversely impacted. These species included several members of the families , Gobiidae, , Melanotaenidae as well as an ambassid (Olive Perchlet) and a retropinnid (Australian Smelt) among others (see section 4.4). Rowe et al . (2008) pointed out that much of this evidence is based solely on correlative field data (11 of 23 species) or on controlled aquarium experiments in which predatory or competitive interactions may be intensified (4 of 23). Evidence from both field studies and aquarium experiments was available for only eight species. The mechanisms and consequences of impacts documented in these studies reflect those reported in stuides overseas on for small-bodied native species (mechanisms — direct predation, competitive exclusion from essential food and habitat resources, aggressive interactions; consequences — increased mortality rates, decreased growth, condition, reproduction, population declines and population fragmentation) and are most likely driven by large overlaps in habitat use and trophic niches, mediated by the local physical environment. Quantitative information regarding the magnitude of impacts in Australia is still lacking, and caution is warranted in extrapolating aquarium experiments to natural systems (Ling 2004). Eastern Gambusia is one of the most widely distributed freshwater fishes in Australia, and its range is still expanding. Because of its broad physiological tolerances, high reproductive capacity and proven ability to rapidly colonise degraded freshwater environments (Arthington et al . 1983; Kennard et al . 2005; King and Warburton 2007), the impact on Australian native fishes may not have peaked and could increase (Rowe et al . 2008). Because invasions by alien species are generally population-level processes, the most reliable evidence for impacts is likely to come from density manipulations under natural conditions, whereby Eastern Gambusia populations are depleted or removed entirely and the recovery of native species is then monitored and compared to control treatments (e.g. Peterson and Fausch 2003). Such experiments have never been attempted in Australia (Rowe et al . 2008).

24 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Figure 2.3 Fin-nipping and weight loss in Dwarf Galaxias as a result of Eastern Gambusia aggression and competition. The fish below and on the left (respectively) was collected from an area with high densities of Eastern Gambusia. The other fish was collected from a healthy population with low densities of Eastern Gambusia. Source: Pitman and Tinkler (2007).

25 Native fish recovery following Eastern Gambusia removal

2.2.2 Impacts on native fishes of the MDB The native fish fauna of the MDB comprises approximately 46 species. This low species richness is typical of Australia’s depauperate freshwater fish fauna, which is a legacy of the continent’s long isolation, low rainfall and high proportion of arid areas (Unmack 2001; Lintermans 2007). Nevertheless, the MDB contains a variety of aquatic systems, including upland and lowland rivers, wetlands, billabongs and lakes, each of which contain their own array of habitat types and associated fish communities. Over half of the native species in the MDB are listed as rare or threatened in state, territory or national listings (Lintermans 2007), and overall abundances are believed to be around 10% of those prior to European settlement (MDBC 2004). Alien species make up one quarter of fish species diversity and in many areas account to 80–90% of the fish biomass, so it is not surprising that alien species have been listed as one of the key threatening processes to the native fish fauna of the MDB (MDBC 2004). Because of the growing evidence of its detrimental impacts on native fish fauna globally and its widespread distribution throughout the MDB, Eastern Gambusia has been identified as one of the key alien species contributing to the decline of a number of native fish within the MDB (MDBC 2004). Despite the common view that interspecific interaction between Eastern Gambusia and MDB native fish is detrimental to the native species, the evidence is largely circumstantial and, like the Australia-wide evidence (see Rowe et al . 2008), is based largely on a limited number of speculative correlative studies and aquarium experiments. Although evidence from individual cases is not strong, particularly for the development of control strategies (see chapter 4.2), the collation and combination of evidence with biological information can provide a better understanding of the potential impacts of an alien species. Therefore we must not only understand the biology of the alien species in question (identifying the reasons for its success as an invader) but also have an equally thorough knowledge of the biology and ecology of the native fish community it has invaded (Rowe et al . 2008). This enables the identification of mechanisms by which native species may or may not be directly affected, and the recognition of potential indirect impacts on individual species and on the fish community as a whole. In the following sections we combine the documented evidence with biological information for each native fish species in the MDB to assess the impact of Eastern Gambusia on the fish fauna of the MDB. The major mechanisms through which Eastern Gambusia directly affect native fish around the world involves interspecific competition for resources such as food and habitat, aggression, and predation of eggs, larvae and juveniles. For Eastern Gambusia to directly affect native species by one of these mechanisms, there must be some overlap in ecological niches. We know that Eastern Gambusia prefer still and slow-flowing waters such as wetlands and backwaters and predominantly occupy the upper water column and littoral zone, although large individuals are associated with benthic areas of macrophyte beds (Stoffels and Humphries 2003). We also jnow that they have a broad, omnivorous diet that includes detritus, aquatic and terrestrial invertebrates, and the eggs and larvae of fish and amphibians. By overlaying this information about diet and macrohabitat and microhabitat niches with those of each native species (using the available knowledge of each developmental stages of each native species), we found that 16 of the 37 native freshwater species of the MDB may have major niche overlaps in a number of developmental stages with Eastern Gambusia (Table 2.1). They included, in particular, ambassids (glassfish), nannopercids (pygmy-perches), melanotaenids (rainbowfishes), atherinids (hardyheads), eleotrids (gudgeons), and retropinnids (smelt), which occupy (at least in part) slowly flowing or still water and together make up the majority of the MDB’s wetland fish communities.

26 Arthur Rylah Institute for Environmental Research

Table 2.1 Potential macrohabitat, microhabitat and trophic niche overlaps between Eastern Gambusia and native freshwater fish in the MDB. Macrohabitat niche = still and slow flowing areas such as back waters, weir pools and wetlands; Microhabitat niche = aquatic vegetation, upper water column, and shallow littoral areas;Trophic overlap includes detritus, microinvertebrates and macroinvertebrates, terrestrial invertebrates and fish eggs and larvae; E = eggs, L = larvae, J = juveniles, A = adult developmental stages, - = minimal overlap of all developmental stages; * Carp Gudgeon species complex. Grey shading indicates species at high risk, defined as species for which there is overlap with Eastern Gambusia for all four developmental stages of the native species in at least one niche, and overlap in at least three developmental stages of the native species in at least one other niche. Macrohabitat niche Microhabitat niche Trophic niche Family Species name Common name overlap overlap overlap

Ambassidae Ambassis agassizii Olive Perchlet E,L,J,A 4,18 E,L,J 4,18 L,J,A 4,15 Anguillidae Anguilla australis Short-finned Eel J,A 6,7,18 - -

Anguilla reinhardtii Long-finned Eel J,A 6,7,18 - -

Atherinidae stercusmuscarum fulvus Unspecked Hardyhead E,L,J,A 1,5,6,18 E,L,J,A 1,5,6,18 L,J,A 6

Craterocephalus fluviatilis Murray Hardyhead E,L,J,A 5,6,18 E,L,J,A 5,6,18 L,J,A 6

Craterocephalus amniculus Hardyhead E,L,J,A 6,18 E,L,J,A 6,18 L,J,A an.

Eleotridae Hypseleotris spp.* Carp Gudgeon* E,L,J,A 1,2,8,9,18 E,L,J,A 1,2,8,9,18,34 L,J,A 8,9,10,15,34

Mogurnda adspersa Southern Purple-spotted Gudgeon E,L,J,A 6,16 E,L,J,A 6,16 L,J 16

Philypnodon grandiceps Flat-headed Gudgeon E,L,J,A 1,2,10,18 E,L,J,A 1,2,10,18 L,J 1,2,10

Philypnodon macrostomus Dwarf Flat-headed Gudgeon E,L,J,A 1,3,18 E,L,J,A 1,3,18 L,J an.

Bovichthyidae Pseudaphritis urvillii Tupong, Congolli - - -

Clupeidae Nematalosa erebi Bony Herring E,L,J,A 11,6 E,L,J,A 11,6 J,A 6,15

Gadopsidae Gadopsis marmoratus River Blackfish J,A 32 - Lan

Gadopsis bispinosus Two-spined Blackfish J,A 32 - Lan

Galaxiidae Galaxias rostratus Flat-headed Galaxias E,L,J,A 6,17,18 E,L,J,A 6,17,18 L,J,A 6,17

Galaxias olidus Mountain galaxias - - -

Galaxias brevipinnis Climbing galaxias - - -

Galaxias truttaceus Spotted Galaxias - - -

(continued on next page)

Arthur Rylah Institute for Environmental Research 27 Native fish recovery following Eastern Gambusia removal

Table 2.1 (continued)

Macrohabitat niche Microhabitat niche Trophic niche Family Species name Common name overlap overlap overlap

Galaxias maculatus Common galaxias J,A 6,17,18,33 J,A 6,17,18 J,A 6,17 Geotriidae Geotria australis Pouched lamprey L6,21,22 - -

Nannopercidae Nannoperca australis Southern pygmy-perch E,L,J,A 1,6,19,20,24 E,L,J,A 1,6,19,20,24 L,J,A 6,19,24

Nannoperca obscura Yarra pygmy-perch E,L,J,A 6,24 E,L,J,A 6,24 L,J,A 6,24

Melanotaeniidae Melanotaenia splendida tatei Desert rainbowfish E,L,J,A 6,15 E,L,J,A 6 L,J,A 6,15

Melanotaenia fluviatilis Murray–Darling rainbowfish E,L,J,A 1,2,25,37 E,L,J,A 1,2,25,37 L,J,A 1,2,9,25,37

Mordaciidae Mordacia mordax Shortheaded lamprey L6,23 - -

Percichthyidae Maccullochella peelii peelii Murray cod L1,2 - L9,30,31

Maccullochella macquariensis Trout cod L1,2 - L13

Macquaria ambigua Golden perch E,L,J 1,2,26 L,J 18,27 L13,14,30

Macquaria australasica Macquarie perch L28 L28 L,J 6,27

Macquaria colonorum Estuary perch - - -

Plotosidae Tandanus tandanus Freshwater catfish E,L,J,A 6,18,29 - L,J 6,29

Neosilurus hyrtlii Hyrtl's tandan E,L,J,A 6 - L,J 6,an.

Porochilus rendahli Rendahl's tandan E,L,J,A 6 - L,J 6,an.

Retropinnidae Retropinna semoni Australian Smelt E,L,J,A 1,2 E,L,J,A 1,2 L,J,A 9,15

Terapontidae Bidyanus bidyanus Silver perch L,J 1,26 L, 1,18,26 L,J 12,13,30

Leiopotherapon unicolor Spangled perch E,L,J,A 6,18,32 E,L,J,A 6,32 L,J,A 6,15,32

1King et al. 2007, 2Humphries et al. 2002, 3Tonkin and Rourke 2008, 4Allen 1996a, 5Wedderburn et al. 2007, 6Lintermans 2007, 7Koehn and O'Connor 1990, 8Stoffels and Humphries 2003, 9King 2005, 10 Gehrke 1992, 11 Puckridge and Walker 1990, 12 Warburton et al. 1998, 13 Lake 1967, 14 Arumugam and Geddes 1987, 15 Balcome et al. 2005, 16 Larson and Hoese 1996, 17 McDowall and Fulton 1996, 18 Treadwell and Hardwick 2004, 19 Llewellyn 1974, 20 Tonkin et al. 2008., 21 Potter 1996a, 22 Potter et al. 1986, 23 Potter 1996b, 24 Kuiter et al. 1996, 25 Allen 1996b, 26 Geddes and Puckridge 1989, 27 Cadwallader and Douglas 1986, 28 Jason Thiem personal communication, 29 Clunie and Koehn 2001, 30 Tonkin et al. 2006, 31 Rowland 1992, 32 Llewellyn 1973, 33 SRA 2007, 34 Balcome and Humphries 2006. 35 Arthington et al. 1983, 36 Lloyd and Walker 1986, 37 Lukies 2004, an Anecdotal from similar species.

28 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Although the direct detrimental impacts of Eastern Gambusia are likely to be less severe in riverine environments, there still may be an impact on larger riverine species. For example, Eastern Gambusia may interact directly with the early developmental stages of these species, which often occupy slackwater habitats (e.g. Barlow and Bock 1981, cited in Howe et al . 1997) or there may be indirect interactions. The indirect interactions of Eastern Gambusia on the MDB’s native fish community have been largely overlooked, mostly because of our poor understanding of the species’ overall ecological impact on Australian environments. Fletcher (1986) suggested that large populations of Eastern Gambusia are likely to alter invertebrate communities through predation. This was documented by Hurlbert et al . (1972), who found algal densities in experimental ponds increased as a result of Eastern Gambusia feeding selectively on zooplankton. Localised reductions in zooplankton can directly limit the availability of food and subsequent survivorship of larval stages and in situ small native fish species (Wilson 2006), and is likely to also reduce the amount of food that is transported to other environments, such as from wetland to river channel (e.g. the flood recruitment model: see Harris and Gehrke 1994). Large populations of small native fish are also likely to affect zooplankton communities, particularly in wetlands. High native fish densities are a natural occurrence, acting as a key source of dispersive offspring for the broader catchment (Wilson 2006; Stuart and Jones 2002). For example, Lyon et al . (2010) reported large numbers of Carp Gudgeon (> 175 000) as well as Australian Smelt, Flat-headed Gudgeon and Eastern Gambusia, moving between off-channel sites and the main river channel. The smaller native species are important forage fish for larger riverine species such as Murray Cod and Golden Perch, which actively avoid Eastern Gambusia as a prey item (Lloyd 1990). Wilson (2006) suggested that alien species could create energetic blocks within aquatic food webs, which would be pertinent for Eastern Gambusia if it resulted in a reduction or replacement of native fish or their prey. Such energetic blocks could have an indirect impact on larger riverine species of recreational significance such as Murray Cod and Golden Perch by reducing the amount of food (such as zooplankton for larvae, and forage fish for adults) entering the broader environment. Because these species are the basis of a large inland recreational fishery, Eastern Gambusia could have both social and economic impacts, along with their environmental impacts. 2.3 Mitigating the impacts of Eastern Gambusia If we consider the extensive literature relating to Eastern Gambusia invasion and global recognition of the threats that Gambusia species pose to freshwater ecosystem function outside their native range, it is surprising that there is very little information available on mitigating their impacts. Pyke (2008) suggests there are two ways to reduce the negative impacts of invasive Gambusia species on native species: lowering their numbers (control), and reducing the impact per individual. Wilson (2006) found that virtually no information was available to guide managers in choosing the most appropriate control strategies in Australia, largely because traditional techniques such as poisoning, exclusion, egg dehydration, direct removal, commercial harvest and habitat restoration have little chance of success in controlling smaller species such as Eastern Gambusia, which occupy more cryptic habitats. Nevertheless, there is some information about Gambusia control from both overseas and in Australia. This control focuses predominantly on chemical techniques and drying of habitats, and always requires extremely thorough treatment and an extensive knowledge of the hydrology of the area. In Australia there have been few attempts to control Eastern Gambusia; most activities on pest fishes, particularly in the MDB, have focused on Common Carp (Wilson 2006). Most states and territories have listed Eastern Gambusia as a noxious or controlled species, so that it is illegal for Eastern Gambusia to be kept, returned to the water or translocated ( Queensland

29 Native fish recovery following Eastern Gambusia removal

Fisheries Management Act 1995; New South Wales Fisheries Management Regulation 2002; Victorian Fisheries Regulations 1998; Northern Territory of Australia Fisheries Act 2005; Tasmania Inland Fisheries Act 1995; ACT Fisheries Act 2000; South Australian Fisheries Management Act 2007 ). New South Wales also developed a threat abatement plan for Eastern Gambusia that proposed further research, a review of existing legislation, and control actions at feasible sites (NSW 2003). While such legislation and planning is an important step in Eastern Gambusia control, particularly in slowing the establishment of new populations, there is little, if any, on-ground activities occurring in areas of established populations, particularly in the MDB. Those activities that have occurred have focused on two relatively new invasions, one in the Northern Territory and the other in Tasmania. Although the success of these programs is dependent on the results of ongoing monitoring (and perhaps further treatment), Macdonald and Tonkin (2008) suggested that, with extensive biological and hydrological knowledge of an area and a coordinated approach, the control of Eastern Gambusia within closed systems or in areas of new invasion is possible. Successful eradication of Eastern Gambusia using current methods is not likely to be feasible in larger open systems, such as many of the areas of the MDB. Minimising the impact of Eastern Gambusia on native fish may be an important strategy, particularly in areas containing threatened fish fauna, until new control strategies such as harvesting techniques and daughterless technology are developed. Pyke (2008) suggested that reducing any negative impacts of Gambusia species on native species could be achieved by a reduction in their numbers, and by reducing the impacts per individual. We know that lentic habitats that contain Eastern Gambusia and provide cover such as aquatic vegetation and snags, contain more native fish than areas without cover and Eastern Gambusia (Morgan et al . 2004). Furthermore, it appears that Eastern Gambusia has the greatest impact on native fish during times of low water levels (e.g. Fairfax et al . 2007). Therefore, habitat maintenance such as enhancing thick aquatic vegetation (Pyke 2008) and watering during times of extreme low levels may reduce the impact of Eastern Gambusia on rare native fish fauna until improved control techniques are available. The key objective of an alien species removal program is to reverse the negative impacts that the species has had on native biota, to benefit native biological diversity (Zavaleta et al . 2001). One question that is often not considered in pest species management programs is ‘What negative effects will a reduction or complete removal of the target alien species have on ecosystem function?’ There is evidence that successful eradications of alien species can have unexpected and undesirable impacts on native species and ecosystems, particularly in areas that the alien species has occupied for a long time and is an established species in the food chain (e.g. Murphy et al . 1998). Maezono and Miyashita (2004) suggested two ways in which such undesirable impacts may occur. First, the removal of an alien species can enhance secondary establishment, or increase the impact of other alien species. Secondly, negative impacts to native biota may occur if the alien species performs functions similar to those of native species that are no longer in the system. Ultimately, the type of species being removed, the degree to which it has replaced native taxa, and the presence of other non-native species can all affect the eventual impacts of removal of an alien species (Zavaleta et al . 2001). Thus, to restore native biodiversity it seems essential to clarify the community-wide impacts of alien species, including interactions with other alien species, prior to eradication (Maezono and Miyashita 2004). Zavaleta et al . (2001) suggested a pre-eradication assessment, including qualitative evaluation of trophic interactions between alien and native species to anticipate the need for special planning, and post-eradication monitoring. They concluded that, while invasive species eradication is an increasingly important component of conservation management, in natural systems a shift in focus is required from pure alien species control towards broader ecosystem restoration goals.

30 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

2.4 Knowledge gaps and the current project A central tenet of integrated pest management is the reduction of impacts on ecosystem processes, not simply a reduction in numbers of the target species (Lodge and Schraeder- Frechette 2003; Koehn and Mackenzie 2004). Because of the ongoing threat Eastern Gambusia pose to native fish communities (Lintermans 2007), the lack of current effective control options (see McKay et al . 2001), and the uncertainty about the long-term detrimental impacts of the species on ecosystem function in the MDB, research into the feasibility of controlling Eastern Gambusia populations to densities where measurable improvements to native fish communities can be detected should be a priority. The core component of this project (Project MD1043) addressed these research needs by taking a holistic approach. It integrated surveys and quantitative experimental work in natural billabong systems throughout the MDB with cost-effectiveness approaches, to: • provide information on the response of native fish communities following the reduction of Eastern Gambusia populations, and • provide a framework to evaluate the feasibility and effectiveness of such control actions, and form a template for evaluating control options for other alien fishes across the MDB. To achieve these objectives the project was divided into three phases. The first phase involved a broad-scale, cross-sectional study of wetland fish communities to develop hypotheses about the effect of Eastern Gambusia on native fish communities in these enclosed systems. The second phase was a field trial of Eastern Gambusia control in small isolated billabongs, to test the hypotheses through density manipulation experiments and to provide information on control options and Eastern Gambusia population dynamics. The third phase identified strategies to maximise the level of improvement to the native fish community through Eastern Gambusia control given a fixed budget (benefit maximisation), and to minimise the cost of achieving a defined significant improvement in the native fish community (cost minimisation). Finally, the project provided a template for evaluating control options for other alien fishes across the MDB.

31 Native fish recovery following Eastern Gambusia removal

3 Phase 1: Cross-sectional study of wetland fish communities across the mid-Murray region of the MDB 3.1 Introduction The invasion, establishment and successful integration of an alien species into an ecosystem often presents some form of ecological risk to that system (Copp et al. 2005). Accurately defining the level of risk, by quantifying the potential detrimental, neutral or even positive impacts of invasion on ecosystem processes or particular species, can provide a strong platform for generating and testing hypotheses regarding what level of control (if any) is required to achieve an acceptable level of risk; see Copp et al (2005) and Gozlan et al. (2010) for discussion. But what constitutes a detrimental impact, and how do we measure it? A recent paper by Gozlan et al. (2010) suggested that the focus should not be on the ecological changes that inevitably occur after invasion, but on whether such changes can be correlated with measurable reductions in native species diversity, abundance, reproduction, condition or shifts in ecosystem function that compromise the long-term integrity of native species populations. In Chapter 2 we detailed the mechanisms by which Eastern Gambusia can impact Australian native fish species, the potential consequences of these mechanisms, and how the nature of the physical environment can mediate the magnitude of such impacts at multiple spatial and temporal scales (Kennard et al. 2005; Pyke 2005; Fairfax et al. 2007; Rowe et al. 2008; Costelloe et al. 2010). The collation of biological information on wetland fishes in the MDB, which was a key component of our literature review, suggests that Eastern Gambusia are likely to have the greatest direct impact on smaller species, particularly the ambassids (glassfish), nannopercids (pygmy-perches), melanotaenids (rainbowfishes), atherinids (hardyheads), eleotrids (gudgeons) and retropinnids (smelt) that comprise the majority of the MDB’s wetland and billabong fish communities (Macdonald and Tonkin 2008). The review found that 16 of the 37 native freshwater species present exhibit major habitat or trophic niche overlaps with Eastern Gambusia, and we postulate that these 16 species are at greatest risk of negative impacts in systems where they coexist with Eastern Gambusia. In order to quantify the current level of impact, we designed a broad-scale, cross-sectional study of wetland fish communities throughout the mid-Murray region of the MDB. The overarching objective of the study was to examine the current influence of Eastern Gambusia on native fish abundance, diversity and condition in ‘closed’ wetland or refugia systems, where dispersal opportunities are limited and any impacts on native species are likely to be magnified. In situations where baseline data are limited, correlative or species distribution models (SDMs) can enable an investigation of variables that cannot be directly manipulated, and reveal patterns between a species’ occurrence or abundance and environmental characteristics (Elith and Leathwick 2009; Capinha and Anastácio 2011), although they cannot identify the causes of such patterns. We collected correlative data on fish species composition, abundance and several environmental covariates across a large spatial scale and constructed such models to: • examine how the presence or absence of Eastern Gambusia influences the probability of occupancy of native and alien fish species in wetlands across the MDB • explore patterns of co-occurrence of native fish species and Eastern Gambusia in these systems • quantify the relationships between Eastern Gambusia abundance and the abundance of other sympatric fish species in differing habitat types and with variability in environmental covariates.

32 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

We also examined relationships between Eastern Gambusia abundance and morphometric condition in juveniles of the three most common native fish species captured during the study: Australian Smelt, Carp Gudgeon and Flat-headed Gudgeon. These models were exploratory and were designed to inform the development of hypotheses on how native fish respond when Eastern Gambusia populations are reduced in wetland systems. These hypotheses are introduced here, and the ecological basis for each is discussed. Phase 2 of the project (see Chapter 4) provided a direct test of each hypothesis in a controlled field removal experiment at selected wetlands in the Ovens River, Victoria. 3.2 Methods 3.2.1 Study area and site selection The cross-sectional study focused on off-channel habitats such as billabongs and anabranches in the mid-Murray region of the MDB. Because of the prolonged drought conditions in the region, which persisted into summer 2009, these habitats were restricted to the Ovens River and Reedy Creek floodplain, Gunbower Forest, Wonga Wetlands (Albury) and the Barmah– Millewa Forest (Figure 3.2.1).

146ºE New South Wales 36ºS Yarrawonga

Lake Mulwala Murray River

# #### # # # # # # # # # # # ## # ##### N ### # ### W E ### ## Ovens River # # # # S # ## ## ## #### Victoria 0 10 20 km # Wangaratta

# ### Wonga Wetlands Gunbower Forest # ## ## ############# ###### ########### Barmah-Millewa/Goulburn # Ovens River floodplain

Melbourne

Figure 3.2.1 Cross-sectional study site locations (red squares) sampled during February– March 2009 and 2010, including detailed locations of sites sampled in the Ovens River floodplain.

33 Native fish recovery following Eastern Gambusia removal

A total of 93 sites were sampled in February–March (later summer – early autumn) 2009 and 2010. To ensure that the assessment of each site’s fish community was not influenced by recent immigrations or emigrations, sites were selected on the basis that they were isolated from the nearby main river channel or other waterbodies for several months prior to sampling. This was achieved by ground-truthing potential sites in December 2008 and 2009, coupled with close monitoring of flow data in the associated rivers throughout spring, summer and autumn each year, and knowledge of ‘commence-to-fill’ levels of anabranches and wetlands in the region. All sites were sampled on a single occasion during February or March, timed to coincide with the peak abundance of Eastern Gambusia and the conclusion of spawning of most native species. We anticipated that any impacts of Eastern Gambusia on the native fish community would be highest during this period (i.e. high gambusia densities coupled with high densities of larval or juvenile native species in restricted habitats). 3.2.2 Water quality and habitat assessment Prior to the commencement of fish sampling, measures of the wetland size (length × width, to the nearest metre) and water quality parameters (pH, conductivity, turbidity, dissolved oxygen and temperature) at a depth of approximately 30 cm were recorded. The mean and maximum depth of the site as a whole, and of each replicate sampling unit, was also recorded. Habitat complexity was also broadly assessed by an estimation of (a) the amount of aquatic vegetation in the littoral zone (none, low, high); and (b) the amount of snags and other debris in the littoral zone (none, low, medium or high) (Figure 3.2.2). Habitat complexity was only assessed for the littoral zone, as all sampling was undertaken within 7 m of the water’s edge.

(a) (b)

(c)

Figure 3.2.2 Example of (a) low (b) medium, and (c) high levels of snags and other debris in site littoral zones.

3.2.3 Fish sampling and processing Recent surveys throughout the study region recorded 10 small-bodied native species as well as five alien species (including Eastern Gambusia) in floodplain environments (Table 3.1). This assemblage composition, in conjunction with behavioural traits and habitat preferences of these species, informed the development of a fish sampling protocol which aimed to provide the most accurate representation possible of the small fish community at a given site.

34 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Table 3.1 Small native fish species and all alien species recorded in floodplain environments (billabongs, anabranches and inundated floodplains) of the mid-Murray River and tributaries between Albury and Barham. Common name Scientific name Reference

Native

Australian Smelt Retropinna semoni 1,2,3,4,5,6,7,8

Carp Gudgeon Hypseleotris spp. 1,2,3,4,5,6,7,8

Flat-headed Gudgeon 2,3,4,5,6,7,8

Craterocephalus stercusmuscarum Un-specked Hardyhead fulvus 3,4,6,7,8

Southern Pygmy-perch Nannoperca australis 1,2,3,4

Murray–Darling Rainbowfish Melanotaenia fluviatilis 3,4,6,7

Dwarf flat-headed Gudgeon Philypnodon macrostomus 3,4,6,7

Flat-headed Galaxias Galaxias rostratus 1,2

Mountain Galaxias Galaxias olidus 2

Bony Bream Nematalosa erebi 5,6,8

Exotic

Eastern Gambusia Gambusia holbrooki 1,2,3,4,5,6,7,8

Carp Cyprinus carpio 1,2,3,4,5,6,7,8

Goldfish Carassius auratus 1,2,3,4,5,6,7,8

Redfin Perch Perca fluviatilis 1,2,3,4,5,6,7,8

Oriental Weatherloach Misgurnus anguillicaudatus 1,2,3,4,5,6,7,8

1 McNeil 2004; 2 King et al. 2003; 3 King et al. 2007; 4Tonkin and Rourke 2008; 5 McKinnon 1997; 6 Rehwinkel and Sharp 2009; 7 Richardson et al. 2005; 8 Jones and Stuart 2004. To allow a rapid yet comprehensive assessment of the fish community at each site, only active sampling methods were used. Passive methods such as fyke netting and bait traps which rely on fish movement and give high interspecies and intraspecies variability in catch data were deemed unsuitable. We restricted the sampling to the littoral zone in each wetland, defined here as the area extending from the water’s edge to not more than 7 m from the bank. We stratified the sampling by habitat type, using standardised fine-mesh seine netting (7 m length × 1.5 m drop; 1 mm mesh diameter; 500 mm square purse) in open water littoral habitats and sweep net electrofishing (SNE) in complex littoral habitats. ‘Open water’ habitats were defined as containing no aquatic vegetation and no small or large woody debris. Habitats in which aquatic vegetation or woody debris were present were classified as ‘complex’. Seine netting and SNE were both effective at capturing fish and larvae as small as 6 mm in length, but seine netting was more effective at capturing mobile species such as Australian Smelt and Eastern Gambusia, and SNE was better at capturing species associated with complex habitats (e.g. gudgeons and Southern Pygmy-perch).

35 Native fish recovery following Eastern Gambusia removal

The SNE method (King and Crook 2002) used a modified standard backpack electrofishing unit (Smith-Root Model 12b) fitted with an anode ring 15 cm in diameter. A moulded plastic rectangular frame (25 × 30 × 2 cm) holding a 250 µm mesh sampling net was attached near the anode ring, with the net tapering to a removable collection jar. Each SNE sample was obtained by approaching the selected complex habitat, activating the anode and moving at a constant speed throughout the habitat for 20 seconds of electrofishing time (Figure 3.2.3). Fish were then concentrated into the collection jar and emptied into an aerated 30 L bucket for processing. Each seine sample was obtained using a standardised point–point method, in which one end of the seine was held in a fixed position at the water’s edge while the other was walked at a steady pace in a tear-drop shape, returning back to the same point; in effect, covering an area of approximately 3 m 2. All fish captured were then concentrated into the square purse and quickly transferred to a 300 L aerated bin for processing.

Figure 3.2.3 Sweep net electrofishing snag / debris habitat in the littoral zone of a billabong on the Ovens River floodplain.

The number of replications required to achieve an accurate representation of the fish community at a given site was determined by conducting a power analysis, based on data collected from three sites for each sampling method. This involved sampling the entire littoral zone, with complex habitats sampled by SNE and open water habitats sampled by seine netting (usually up to 12 samples for each site and method). The results indicated that six replicate SNE samples in complex habitat and six seine samples in open water habitat would provide an adequate assessment of species diversity and abundance in the littoral zone of larger sites. Therefore, following the initial habitat assessment at each site, any complex habitats in the littoral zone were sampled with a maximum of six SNE passes, and all other open water areas of the littoral zone were sampled with a maximum of six seine passes. At sites that were too small to achieve such replication we made as many passes as possible without overlapping the areas sampled. In addition, if a site contained no complex habitat in its littoral zone, only seine netting was undertaken; and conversely, if a site only contained complex habitat, only SNE was undertaken. At the completion of each replicate SNE and seine shot, all fish were identified and counted, and twenty individuals of each species were randomly selected and measured for total length (TL; nearest mm). Additionally, the first 20 individuals of each species measured at each site were euthanised in an overdose of anaesthetic (alfaxalone 40 mg/L) for 10 minutes, and then preserved in 95% ethanol to enable measures of condition (i.e. Krel , fin damage) and ageing to be undertaken in the laboratory.

36 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

To assess morphometric condition, preserved juvenile Australian Smelt, Carp Gudgeon and Flat-headed Gudgeon collected during the sampling events were measured for standard length (nearest 0.1 mm) and weight (nearest 0.001 g). These measurements were made after a minimum of 10 days to allow shrinkage associated with preservation to stabilise (Fey and Hare 2005). Fish were then positioned on their right side and examined for fin condition. The caudal, dorsal, anal and left hand pelvic and pectoral fins were given a score from 1 to 3, where: 1 = no or minimal damage — fin has no sign of damage or only minor splitting of rays 2 = moderate damage — part of fin is missing, but more that 50% remains 3 = major damage — less than 50% of the fin remains. Fin scores were subsequently converted to either 0 (scores of 1), or 1 (scores of 2 or 3) i.e. ‘no damage’ or ‘damaged’ (Figure 3.2.4).

(a) (b)

Figure 3.2.4. Examples of different levels of caudal fin damage including fins scored as ‘1’ (top), ‘2’ (middle) and ‘3’ (bottom) for (a) Carp Gudgeon; and (b) Australian Smelt.

3.2.4 Statistical analysis and development of correlative models

3.2.4.1 Species occupancy models Models were fitted in a Bayesian framework in WinBUGS (Lunn et al. 2000) to explore how the presence or absence of Eastern Gambusia or other covariates — presence or absence of aquatic vegetation (veg), wetland size (size), water temperature (temp), conductivity (cond), DO, pH, turbidity (turb) — can influence the probability of a site being occupied by native and other alien fish species. Analyses was performed for 81 wetland sites using five samples from each site (combined SNE and seine passes). The data consisted of the presence or absence of seven species — Eastern Gambusia (Gh), Carp Gudgeon (Hsp), Flat-headed Gudgeon (Pg), Australian Smelt (Rs), Unspecked Hardyhead (Cs), Common Carp (Cc) and Goldfish (Ca) in each of the samples per site. The low detection of other native species (e.g. Southern Pygmy-perch, Murray–Darling Rainbowfish) precluded their inclusion in the models. These models estimate the true occupancy probability of a given species, accounting for errors in detection from sampling methods. In addition, we used them to estimate the patterns of co-occurrence of native species and Eastern Gambusia.

37 Native fish recovery following Eastern Gambusia removal

3.2.4.2 Relationships between species’ abundances and environmental covariates The purpose of this analysis was primarily exploratory, in that we were attempting to determine whether there were any relationships (whether linear or non-linear) between the abundance of Eastern Gambusia, the abundance of native and other alien fish species, and several environmental covariates. We used a relatively new machine-learning technique called boosted regression trees (BRT) to explore the nature of these relationships in 58 of the wetlands sampled in the cross-sectional study. Conventional regression models produce a single ‘best’ model, whereas BRT combines large numbers of relatively simple tree models adaptively to optimise the predictive performance (Elith et al. 2008). The optimal number of trees in the final model is estimated using cross-validation techniques. For each wetland, the data used was the average counts of fish species from six samples (combined SNE and seine net passes). These averages were log( x + 1) transformed before analysis to stabilise the variance. An analysis was undertaken for the same seven species used in the occupancy modelling analysis (see section 1.2.4.1), but here results are presented for Eastern Gambusia and the three most abundant native fish species based on our survey data: Australian Smelt, Carp Gudgeon and Flat-headed Gudgeon. Abundances of species that were not subject to analysis (i.e. response variable) were treated as potential explanatory variables. Other explanatory variables included in the models were environmental parameters — wetland size (size), aquatic vegetation cover (veg), debris load (debris), pH, dissolved oxygen (DO), conductivity (cond), turbidity (turb) and water temperature (temp). 3.2.4.3 Condition indices for juvenile native fishes An exploratory assessment of body and fin condition was undertaken for the juveniles of Australian Smelt, Carp Gudgeon and Flat-headed Gudgeon. The body condition index for juvenile fish was estimated by applying the length/weight polynomial model for the juveniles of these species. The residuals from each model were used as an index of condition for each individual. A second measure of condition was undertaken using the relative condition factor (see Froese 2006). Length and weight of juvenile fish were combined to give an indicator of general body form of juvenile Australian Smelt, Carp Gudgeon and Flat-headed Gudgeon, using the equation b W = aL where W is the weight in grams, L is the TL in mm, and a and b are constants. Length–weight relationships could then be used to compare body condition using the relative condition factor

Krel , using the formula

b Krel = W / aL

The relationship between juvenile condition of each species and the Eastern Gambusia abundance (measured as catch per unit effort, CPUE) at each sampling site was then assessed using simple linear regression. An initial examination of the fin condition data for the three aforementioned native species indicated that the caudal fin sustained by far the most damage of any fin type. As a result we decided to limit our analysis to this fin type. We constructed logistic regression models in R (version 2.12.2, The R Foundation for Statistical Computing) to examine the impact of Eastern Gambusia CPUE on caudal fin condition in juveniles of each of the three species. We were also interested in the potential influence of habitat complexity and its interaction with Eastern Gambusia CPUE in mediating the prevalence of aggressive encounters and subsequent effects on the degree of caudal fin damage. Aquatic vegetation cover and the

38 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal debris load at each site were added as categorical variables in the models, and fin condition was assessed in relation to changes in these variables. 3.3 Results 3.3.1 Assessment of wetland fish communities of the mid Murray River A total of 61 575 fish comprising six native and five alien species were captured during the cross-sectional study (Table 3.2). The most abundant and frequently occurring species were Carp Gudgeon (27 346; 44% of abundance and occupying 90% of sites), Eastern Gambusia (26 932; 43% of abundance and occupying 83% of sites), Australian Smelt (4040, 7% of abundance and occupying 49% of sites) and Flat-headed Gudgeon (2186, 4% of abundance and occupying 48% of sites) respectively. Four of the ten small-bodied native species previously recorded in off-channel habitats in these regions — Dwarf Flat-headed Gudgeon, Flat-headed Galaxias, Mountain Galaxias and Bony Bream — were not recorded during the study: see Table 3.1). Although three of the four species occur in higher abundances outside the study region or are more riverine species, the absence of the Flat-headed Galaxias is of concern. Very few Southern Pygmy-perches were captured in the Lower Ovens, and none from other sites previously known to harbour healthy populations (McNeil 2004, Tonkin et al. 2008). Small numbers of Unspecked Hardyhead and Murray–Darling Rainbowfish were captured, but both these species are known to persist in good numbers in the middle and lower reaches of the MDB (Lintermans 2007). All of the alien species previously recorded in off- channel habitats in the regions were collected in the surveys.

Table 3.2 Raw abundances of each species collected during the 2009–2010 cross-sectional study from each of the four regions surveyed (Gunbower, Lower Ovens, Albury (Wonga wetlands) and Barmah–Millewa/Goulburn). The number of sites surveyed within each region is shown in parentheses. Gunbower Ovens Albury B–M/G Common name Scientific name (n = 5) (n = 83) (n = 2) (n = 3) Total Native Australian Smelt Retropinna semoni 1234 3170 48 – 4452 Carp Gudgeon Hypseleotris spp. 2100 23985 974 287 27346 Philypnodon Flat-headed Gudgeon – 2208 55 28 2291 grandiceps Craterocephalus Unspecked Hardyhead stercusmuscarum – 18 – 1 19 fulvus Nannoperca Southern Pygmy-perch – 16 – – 16 australis Murray–Darling Melanotaenia 3 – – 3 6 rainbowfish fluviatilis Alien Eastern Gambusia Gambusia holbrooki 1266 25042 455 169 26932 Carp Cyprinus carpio – 194 1 – 195 Goldfish Carassius auratus – 293 1 – 294 Redfin perch Perca fluviatilis – 9 – – 9 Misgurnus Oriental weatherloach – 15 – – 15 anguillicaudatus Total 4603 54950 1534 488 61575

39 Native fish recovery following Eastern Gambusia removal

3.3.2 The influence of Eastern Gambusia on species occupancy and abundance 3.3.2.1 Species occupancy models The output from the occupancy modelling suggests that the presence of Eastern Gambusia and the presence or absence of aquatic vegetation at a particular wetland can interact to influence the probability of occupancy by several sympatric fish species (Figure 3.3.1). The three most common native fish species captured in the surveys — Carp Gudgeon, Australian Smelt and Flat-headed Gudgeon — were most likely to coexist with Eastern Gambusia in wetlands lacking aquatic vegetation, with the probability of occupancy for these species highest when Eastern Gambusia were present. Unspecked Hardyhead and Common Carp were least likely to occur in wetlands containing Eastern Gambusia and no vegetation. Moreover, Common Carp were significantly more likely to coexist with Eastern Gambusia in wetlands containing aquatic vegetation, or in the absence of Eastern Gambusia. The likelihood of Goldfish being present at a particular wetland appears to be largely unaffected by the presence or absence of Eastern Gambusia and aquatic vegetation. At sites where Eastern Gambusia were not captured, the presence of aquatic vegetation had minimal influence on the probability of that wetland being occupied by every other species examined (Figure 3.3.1).

Veg present: Gh present Veg present: Gh absent Veg absent: Gh present Veg absent: Gh absent

Occupancyprobability

0.0 0.2 0.4 0.6 0.8 1.0

Hsp Pg Rs Cs Cc Ca Species

Figure 3.3.1 Co-occurrence of sampled species with the presence or absence of Eastern Gambusia (Gh) and aquatic vegetation (Veg) in the wetlands. Species codes are Carp Gudgeon (Hsp), Flat-headed Gudgeon (Pg), Australian Smelt (Rs), Unspecked Hardyhead (Cs), carp (Cc) and Goldfish (Ca).

Eastern Gambusia exhibited substantial flexibility in the types of habitats in which they persisted. The probability of occupancy was greater than 0.75 regardless of the presence of aquatic vegetation, although the species was more likely to occupy wetlands devoid of aquatic vegetation (Figure 3.3.2). In addition, larger wetlands sampled were more likely to be occupied by Eastern Gambusia than smaller ones, and the odds of occupancy were decreased as temperature and turbidity increased (Figure 3.3.3).

40 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

probability

Occupancy

0.0 0.2 0.4 0.6 0.8 1.0 Veg present Veg absent Aquatic vegetation

Figure 3.3.2 Probability of Eastern Gambusia occupying a wetland when aquatic vegetation was present or absent.

Log (odds ofoccupancy) (odds Log

-6 -4 -2 0 2 4 6

size temp cond DO turb

Environmental variable Figure 3.3.3 The influence of different environmental covariates: wetland size (size), water temperature (temp), conductivity (cond), dissolved oxygen (DO) and turbidity (turb) on the probability of occupancy by Eastern Gambusia.

41 Native fish recovery following Eastern Gambusia removal

3.3.2.2 Relationships between species’ abundances and environmental covariates The primary aim of the BRT analysis was to explore the nature of any relationships existing between the abundances of Eastern Gambusia and other sympatric species, and the potential influence of environmental covariates in sampled wetlands. As stated in the Methods (section 3.2.4.2), we ran separate BRT models on all seven species (Gh, Hsp, Pg, Rs, Cs, Cc, Ca) examined in the occupancy modelling section, but here present results only for Eastern Gambusia, Australian Smelt, Carp Gudgeon, Flat-headed Gudgeon and Common Carp. The output from the BRT models reports the relative contribution of each potential predictor variable for the final model of each species (Table 3.3) and provides fitted functions of each predictor in relation to the abundance of each species.

Eastern Gambusia abundance The final BRT model fitted to the combined SNE and seine netting data had 1750 trees, as estimated by cross-validation, and explained 51% of the deviance. The most influential predictor variable was the abundance of Carp Gudgeon, which contributed 51% of the variance of the final BRT (Table 3.3). The fitted functions of each predictor variable relative to Eastern Gambusia abundance are shown in Figure 3.3.4. The plots illustrate the strong positive relationship between Eastern Gambusia and Carp Gudgeon abundance, and very weak negative relationships with turbidity, pH and Australian Smelt abundance.

fitted function fitted function fitted function

-0.5 0.0 0.5 -0.5 0.0 0.5 -0.5 0.0 0.5 -0.5 0.0 0.5

0 1 2 3 4 5 6 0 100 200 300 6.5 7.0 7.5 8.0 8.5 9.0 9.5 15 20 25 30 35 Hsp (51.1%) Turb (8.5%) pH (7.1%) Temp (6.9%)

fitted function fitted function fitted function

-0.5 0.0 0.5 -0.5 0.0 0.5 -0.5 0.0 0.5 -0.5 0.0 0.5

HLMN 0 200 400 600 0 5000 10000 20000 2 4 6 8 10 12 debris (6.8%) Cond (5.7%) size (4.9%) DO (4.9%)

gambusiaEastern abundance

fitted function fitted function fitted function -0.5 0.0 0.5 -0.5 0.0 0.5 -0.5 0.0 0.5 -0.5 0.0 0.5

0 1 2 3 4 5 HLN 0 1 2 3 4 0.0 0.5 1.0 1.5 2.0

Rs (2.9%) veg (1%) Pg (0.2%) Cc (0%)

Predictor variable

Figure 3.3.4 Fitted functions describing relationships between Eastern Gambusia abundance and all predictor variables for the final BRT model. Refer to Figure 6 for species codes. The value in brackets after each predictor variable shows the relative influence of that predictor in the final model (see also Table 3.3). For variables ‘debris’ and ‘veg’, N = none, L = low, M = medium, H = high levels.

42 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Table 3.3 Relative influence (I) of each predictor variable (PV) on abundances of Eastern Gambusia, Australian Smelt, Carp Gudgeon, Flat-headed Gudgeon and Common Carp in the final BRT models for each species. Eastern Australian Carp Flat-headed Common Gambusia Smelt Gudgeon Gudgeon Carp PV I (%) PV I (%) PV I (%) PV I (%) PV I (%)

Hsp 51.1 Hsp 28.5 Gh 36.8 Hsp 39.9 Hsp 29.1 Turb 8.5 size 25.0 Rs 19.2 DO 20.9 pH 18.9 pH 7.1 turb 9.6 DO 16.4 pH 10.9 size 12.5 temp 6.9 cond 8.8 turb 10.2 size 7.8 cond 10.2 debris 6.8 Gh 7.1 size 4.1 cond 6.0 Gh 7.6 cond 5.7 pH 5.8 Cc 3.7 turb 4.8 turb 6.1 size 4.9 temp 5.3 temp 2.7 Gh 4.4 temp 4.9 DO 4.9 DO 4.5 Pg 2.2 temp 4.0 Rs 3.7 Rs 2.9 Cc 2.9 cond 2.1 veg 0.7 DO 2.1 veg 1.0 Pg 1.1 pH 1.5 Rs 0.3 Ca 1.6 Pg 0.2 debris 0.8 veg 0.6 debris 0.3 veg 1.6 Cc 0.0 Veg 0.6 debris 0.3 Cc 0.0 debris 1.0 Ca 0.0 Ca 0.1 Ca 0.0 Ca 0.0 Pg 0.5

Australian Smelt abundance The optimal BRT model contained 1900 trees and explained 50% of the deviance. Carp Gudgeon abundance was the most influential variable, followed by wetland size, with positive relationships noted between these two variables and Australian Smelt abundance (Figure 3.3.5). Australian Smelt abundance was found to increase slightly at more turbid and saline wetlands, and there was a weak negative relationship with Eastern Gambusia abundance.

fitted function fitted function fitted function

-0.4 -0.2 0.0 0.2 0.4 -0.4 -0.2 0.0 0.2 0.4 -0.4 -0.2 0.0 0.2 0.4 -0.4 -0.2 0.0 0.2 0.4

0 1 2 3 4 5 6 0 5000 10000 20000 0 100 200 300 0 200 400 600

Hsp (28.5%) size (25%) Turb (9.6%) Cond (8.8%)

fitted function fitted function fitted function

-0.4 -0.2 0.0 0.2 0.4 -0.4 -0.2 0.0 0.2 0.4 -0.4 -0.2 0.0 0.2 0.4 -0.4 -0.2 0.0 0.2 0.4 0 1 2 3 4 5 6 6.5 7.0 7.5 8.0 8.5 9.0 9.5 15 20 25 30 35 2 4 6 8 10 12

Gh (7.1%) pH (5.8%) Temp (5.3%) DO (4.5%)

abundance Australian smelt

fitted function fitted function fitted function

-0.4 -0.2 0.0 0.2 0.4 -0.4 -0.2 0.0 0.2 0.4 -0.4 -0.2 0.0 0.2 0.4 -0.4 -0.2 0.0 0.2 0.4

0.0 0.5 1.0 1.5 2.0 0 1 2 3 4 HLMN HLN Cc (2.9%) Pg (1.1%) debris (0.8%) veg (0.6%)

Predictor variable Figure 3.3.5 Fitted functions describing relationships between Australian Smelt abundance and all predictor variables for the final BRT model.

43 Native fish recovery following Eastern Gambusia removal

Carp Gudgeon abundance The final model contained 2800 trees and explained 73% of the deviance. The most important variables in the model were Eastern Gambusia abundance, which contributed 37% of the variance, followed by Australian Smelt abundance, DO and turbidity. Carp Gudgeon abundance was positively related to each of these variables (Figure 3.3.6).

fitted function fitted function fitted function -0.5 0.0 0.5 -0.5 0.0 0.5 -0.5 0.0 0.5 -0.5 0.0 0.5

0 1 2 3 4 5 6 0 1 2 3 4 5 2 4 6 8 10 12 0 100 200 300 Gh (36.8%) Rs (19.2%) DO (16.4%) Turb (10.2%)

fitted function fitted function fitted function -0.5 0.0 0.5 -0.5 0.0 0.5 -0.5 0.0 0.5 -0.5 0.0 0.5

0 5000 10000 20000 0.0 0.5 1.0 1.5 2.0 15 20 25 30 35 0 1 2 3 4

size (4.1%) Cc (3.7%) Temp (2.7%) Pg (2.2%)

gudgeonCarp abundance

fitted function fitted function fitted function -0.5 0.0 0.5 -0.5 0.0 0.5 -0.5 0.0 0.5 -0.5 0.0 0.5

0 200 400 600 6.5 7.0 7.5 8.0 8.5 9.0 9.5 HLN HLMN Cond (2.1%) pH (1.5%) veg (0.6%) debris (0.3%) Predictor variable Figure 3.3.6 Fitted functions describing relationships between Carp Gudgeon abundance and all predictor variables for the final BRT model.

44 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Flat-headed Gudgeon abundance The final BRT model contained 500 trees and explained 15% of the deviance. Carp Gudgeon abundance was by far the most influential variable, evidenced by a strong positive relationship with Flat-headed Gudgeon abundance. DO and pH were also positively related to Flat-headed Gudgeon abundance, whereas Eastern Gambusia abundance had minimal little influence (Figure 3.3.7).

fitted function fitted function fitted function

-0.10 0.00 0.10 -0.10 0.00 0.10 -0.10 0.00 0.10 -0.10 0.00 0.10 0 1 2 3 4 5 6 2 4 6 8 10 12 6.5 7.0 7.5 8.0 8.5 9.0 9.5 0 5000 10000 20000 Hsp (39.9%) DO (20.9%) pH (10.9%) size (7.8%)

fitted function fitted function fitted function

-0.10 0.00 0.10 -0.10 0.00 0.10 -0.10 0.00 0.10 -0.10 0.00 0.10 0 200 400 600 0 100 200 300 0 1 2 3 4 5 6 15 20 25 30 35 Cond (6%) Turb (4.8%) Gh (4.4%) Temp (4%)

Flat-headedgudgeon abundnace

fitted function fitted function fitted function -0.10 0.00 0.10 -0.10 0.00 0.10 -0.10 0.00 0.10 -0.10 0.00 0.10

HLN 0 1 2 3 4 5 HLMN 0.0 0.5 1.0 1.5 2.0 veg (0.7%) Rs (0.3%) debris (0.3%) Cc (0%) Predictor variable Figure 3.3.7 Fitted functions describing relationships between Flat-headed Gudgeon abundance and all predictor variables for the final BRT model.

45 Native fish recovery following Eastern Gambusia removal

Common Carp abundance The optimal BRT model contained 200 trees and explained 43% of the deviance. The most influential variable in determining Common Carp abundance was Carp Gudgeon abundance, with a clear negative association between the two variables. The pH of the wetland and its size were the next most important predictor variables. Common Carp abundance showed a slight negative relationship with Eastern Gambusia abundance (Figure 3.3.8).

fitted function fitted function fitted function -0.10 0.00 0.10 -0.10 0.00 0.10 -0.10 0.00 0.10 -0.10 0.00 0.10 0 1 2 3 4 5 6 6.5 7.0 7.5 8.0 8.5 9.0 9.5 0 5000 10000 20000 0 200 400 600 Hsp (30.5%) pH (17.3%) size (12.5%) Cond (9.7%)

fitted function fitted function fitted function -0.10 0.00 0.10 -0.10 0.00 0.10 -0.10 0.00 0.10 -0.10 0.00 0.10 0 1 2 3 4 5 6 0 100 200 300 15 20 25 30 35 0 1 2 3 4 5

Carp abundanceCarp Gh (7.5%) Turb (5.6%) Temp (4.8%) Rs (4.1%)

fitted function fitted function fitted function -0.10 0.00 0.10 -0.10 0.00 0.10 -0.10 0.00 0.10 -0.10 0.00 0.10 2 4 6 8 10 12 0.0 0.5 1.0 1.5 2.0 HLN HLMN DO (2.8%) Ca (1.7%) veg (1.7%) debris (1.3%) Predictor variable

Figure 3.3.8 Fitted functions describing relationships between carp abundance and all predictor variables for the final BRT model.

3.3.3 The influence of Eastern Gambusia on juvenile condition of native species 3.3.3.1 Body condition The assessment of body condition in relation to Eastern Gambusia abundance at particular sites was undertaken for 223 juvenile Australian Smelts, 336 juvenile Carp Gudgeons and 106 juvenile Flat-headed Gudgeons. Condition was expressed in two forms: the residuals derived from the population’s length–weight relationship and the relative condition factor ( Krel ). The linear regression models showed significant negative relationships between both indices of juvenile condition of all three species as Eastern Gambusia abundance increased (all p < 0.001: Figure 3.3.9). Although marked variation in condition scores for each of the species was apparent at some sites, resulting in relatively low R2 values for these relationships, the data suggest a general decline in body condition for the three native species in sites containing high Eastern Gambusia densities.

46 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Australian smelt (n=223) Carp gudgeon(n=336) Flat-headed gudgeon (n=107)

p < 0.001 p < 0.001 p < 0.001 80 2 60 R2 = 0.1856 30 R = 0.114 R2 = 0.239 60 20 40 40 20 10 20

0 0

Wresiduals 0 -

L -20 W residuals W -10 -

L -40 -20 -20 -60

p < 0.001 p < 0.001 p < 0.001 1.2 1.4 R2 = 0.074 R2 = 0.116 R2 = 0.199 1.4 1.2 1.0

1.0 1.2

rel rel 0.8 K K 0.8 1.0 0.6 0.6 0.8 0.4 0.4 010 20 30 40 01020 30 40 05 10 15 20 Eastern gambusia CPUE Figure 3.3.9 Juvenile condition of Australian Smelt, Carp Gudgeon and Flat-headed Gudgeon plotted against Eastern Gambusia abundance. Condition indices are expressed as residuals of each species’ length–weight (L–W) relationship (top), where 0 = average condition, and the relative condition factor ( Krel ; bottom). Fitted and observed relationships of the regression analysis are presented with 95% confidence intervals.

3.3.3.2 Fin condition The probability of caudal fin damage in juvenile Australian Smelt rose significantly as Eastern Gambusia abundance increased: Australian Smelt were 1.4 times more likely to sustain damage with every one unit increase in Eastern Gambusia CPUE (logistic regression: odds ratio = 1.407, p = 0.029) (Figure 3.3.10). The extent of aquatic vegetation cover and debris load had little effect on the likelihood of fin damage in juvenile Australian Smelt, and the interactions between these variables and Eastern Gambusia CPUE were also not significant (all p > 0.100). Carp Gudgeon fin condition was largely unaffected by increases in Eastern Gambusia abundance (Figure 3.3.10). The logistic model actually predicted a slight decrease in likelihood of fin damage as Eastern Gambusia CPUE increased, although this effect was not significant. Similarly, a negative relationship was observed between the extent of aquatic vegetation cover and the probability of fin damage, but again this relationship was relatively weak. Higher debris loads significantly reduced the probability of fin damage across our sites however, with fin damage estimated to be almost four times less likely in sites with medium versus low debris, and high versus medium debris loads (odds ratio = 0.283, p = 0.029). Interactions between Eastern Gambusia CPUE, and both aquatic vegetation cover and debris load were not significant (all p > 0.100). Flat-headed Gudgeon showed minimal fin damage across all sites, and the probability of damage remained low regardless of increases in Eastern Gambusia CPUE or variation in aquatic vegetation cover and debris load.

47 Native fish recovery following Eastern Gambusia removal

Australian smelt (n=183) Carp gudgeon (n=173) Flat-headed gudgeon (n=69)

1.0 1.0 1.0

1.0 p = 0.029 p = 0.128 p = 0.279 0.8 0.8 0.8

0.8 0.6 0.6 0.6

0.6 Value Value Value 0.4 0.4 0.4 DAMAGE DAMAGE DAMAGE 0.4

0.2 0.2 0.2 0.2

findamage ofProbability 0.0 0.0 0.0 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 LOGCPUE Log (eastern LOGCPUEgambusia CPUE+1) LOGCPUE 1.0 1.0 1.0 p = 0.336 p = 0.377 p = 0.807 0.8 0.8 0.8 0.6 0.6 0.6 Value Value Value 0.4 0.4 0.4 DAMAGE DAMAGE DAMAGE 0.2 0.2 0.2

Probability of findamage ofProbability 0.0 0.0 0.0 None1 Low 2 High3 None1 Low 2 High3 None1 Low 2 High3 AQUAVEG AQUAVEG AQUAVEG Aquatic vegetation cover

1.0 1.01.0 1.0

0.8 p = 0.545 0.80.8 p = 0.029 0.8 p = 0.235

0.6 0.60.6 0.6

Value Value Value 0.4 0.40.4 0.4 DAMAGE DAMAGE DAMAGE 0.2 0.20.2 0.2

Probability of findamage ofProbability 0.0 0.00.0 0.0 Low2 Med 3 High4 Low2 Med 3 High4 Low2 Med 3 High4 DEBRIS DEBRIS DEBRIS Debris load

Figure 3.3.10 Plots showing the probability of fin damage for Australian Smelt, Carp Gudgeon, and Flat-headed Gudgeon in relation to Eastern Gambusia abundance, the extent of aquatic vegetation and the debris load across all sampled wetlands. Fitted curves from the logistic regression analyses are shown with 95% confidence intervals. Data points represent observations of damage (coded as 1) or no damage (coded as 0).

48 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

3.4 Discussion 3.4.1 Assessment of wetland fish communities of the mid-Murray River The cross-sectional study provided a snapshot of wetland fish community structure and associated environmental variables across a large portion of the southern MDB. The dominance of Carp Gudgeon, Eastern Gambusia and (to a lesser extent) Australian Smelt across the four regions surveyed was expected, given the timing of sampling (late summer – early autumn), and the results from previous work (e.g. Humphries et al. 2002; McNeil 2004; McMaster and Bond 2008). Yet there were some differences in fish community composition when compared with past surveys in the MDB (e.g. King et al. 2003; McNeil 2004; King et al. 2007; Tonkin and Rourke 2008; Rehwinkel and Sharp 2009), namely the total absence of Flat-headed Galaxias from any of our sites, the restriction of Southern Pygmy-perch to the lower Ovens River floodplain, and in very low numbers, and the presence and in some cases dominance of Flat-headed Gudgeon at particular Ovens River sites. In designing this phase of the study, we hypothesised that the control exerted by Eastern Gambusia on the diversity, abundance and condition of sympatric fishes would be maximised during late summer and early autumn in MDB wetland systems. We predicted that peak Eastern Gambusia densities would coincide with low water levels at this time, subsequently increasing ecological niche overlap with sympatric species and raising both the intensity and frequency of negative interactions. However, the interplay of many biotic and abiotic forces could shape fish community structure in freshwater ecosystems (Winemiller et al. 2000; Matthews and Marsh-Matthews 2003; Magoulick and Kobza 2003; and Prince 2010). In hydrologically variable environments such as floodplain wetlands, the timing, magnitude and duration of flooding and drying plays a pivotal role in mediating dispersal, immigration and reproductive opportunities and controlling physicochemical conditions, resource availability and habitat use — factors that can ultimately define fish species diversity, abundance, and the extent and nature of intraspecific and interspecific interactions (e.g. Jackson et al. 2001; McNeil 2004, Arthington et al. 2005; Beesley and Prince 2010). Factors associated with dispersal and colonisation may be more influential during or shortly after inundation, with predation, competition and environmental stressors becoming more important as drying proceeds (Matthews and Marsh Matthews 2003; Magoulick and Kobza 2003, McNeil 2004). During the survey the wetlands we sampled were rapidly contracting, exposing the local fish communities to increasingly harsh physicochemical conditions (e.g low DO and high water temperatures) and at the same time intensifying potential competition for ever-diminishing resources. Such conditions are likely to favour the persistence of species with high physiological tolerances and morphological or behavioural adaptations that enable them to exploit these situations, species that can successfully buffer competition effects by occupying different ecological niches, and ‘generalist’ species that are highly flexible in relation to resource use (see Poff and Allan 1995; Magoulick and Kobza 2003; McNeil and Closs 2007). Our data support this prediction. The two most abundant species captured during the surveys, Eastern Gambusia and Carp Gudgeon, have high tolerances to physiological stress (McNeil and Closs 2007; McMaster and Bond 2008), high reproductive capacities, and protracted spawning periods over the summer months. These species coexisted in over 80% of wetlands sampled and appeared to occupy similar habitats in the littoral zone, often in extremely high densities. Low capture rates of certain key wetland species (Southern Pygmy-perch and Murray–Darling Rainbowfish) and the total absence of Flat-headed Galaxias during the study unfortunately precluded their inclusion in the correlative modelling component. In the following discussion we focus on these rarer species and discuss the mechanisms that may have contributed to their low abundances in our surveys.

49 Native fish recovery following Eastern Gambusia removal

Adult Carp Gudgeon collected from lower Goulburn wetland.

50 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Southern Pygmy-perch The southern pygmy-perch is patchily distributed in rivers and wetlands in northern Victoria and has undergone a major decline in distribution and abundance throughout the MDB, particularly throughout New South Wales, where it is endangered (Tonkin et al. 2008). Loss of habitat and the effects of river regulation have been cited as the primary reasons for the observed declines (Arthington et al. 1983; Lintermans 2007; Tonkin et al. 2008). Although only very few fish were captured in our study, the off-channel habitats within Barmah– Millewa Forest (BMF) and adjacent to the unregulated lower Ovens River, where most our effort was focused, have supported healthy Southern Pygmy-perch populations in the past, particularly during years of substantial floodplain inundation (McNeil 2004; Tonkin et al. 2008). A study examining trends in abundance and recruitment of the species in wetlands of the BMF over five years and under varying hydrological conditions found that dispersal and recruitment was enhanced in a year of prolonged spring–summer floodplain inundation (associated with a major environmental watering event in 2005–6), coinciding with the highest abundances (Tonkin et al. 2008). The authors concluded that reduced flooding frequency associated with river regulation may be an important driver of the species’ decline in the MDB. In light of these findings, it is possible the lack of significant flooding in BMF in the years preceding our cross-sectional study may have diminished the Southern Pygmy-perch population in the area, resulting in the lack of detection in BMF. Our sampling was limited to late summer – early autumn in 2008–9 and 2009–10, a time when we predicted that overlap in trophic niches and the intensity of antagonistic interactions between Eastern Gambsuia and native fishes would be greatest. Southern Pygmy-perch is known to occupy habitat and trophic niches that are very similar to those of Eastern Gambusia (see Table 2.1), and despite its apparently high tolerance to physicochemical extremes (McNeil and Closs 2007; McMaster and Bond 2008; but see Morrongiello et al. in press) the species may be particularly susceptible in shrinking wetlands during late summer. Following large-scale floods that inundated the entire lower Ovens floodplain in 1996, McNeil (2004) documented substantial shifts in wetland fish assemblages as drying progressed from spring to late summer. During spring sampling in November 1996, he noted that Carp Gudgeon dominated floodplain fish communities, Southern Pygmy-perch was abundant and widely distributed, and Eastern Gambusia was found in large numbers only in a few wetlands. As wetlands contracted through summer, numerical dominance switched from Carp Gudgeon to Eastern Gambusia. By the late dry season (February–March 1997), captures of Southern Pygmy-perch were rare and all billabongs were arranged along a gradient based solely on the relative dominance of Carp Gudgeons and Eastern Gambusia. Notwithstanding the strong evidence relating altered flow regimes to declines in populations across the MDB (e.g. Tonkin et al. 2008), seasonal decreases in population size in relatively unmodified systems, as reported by McNeil (2004), point more to interactions of local physical and biological factors associated with the progression of drying in these habitats (Matthews and Marsh-Matthews 2003; Morrongiello et al. in press). In addition, the weight of correlative evidence from our study and that of McNeil (2004) and others (e.g. Lloyd and Walker 1986; Koster 1997; SRA 2007) suggests that Eastern Gambusia has at least some level of negative impact on Southern Pygmy-perch throughout the MDB. Clearly, further data is need to define the level of impact and assess the relative risk to the species at key sites. We discuss these issues further in relation to Phase 3 of the project (see Chapter 5).

51 Native fish recovery following Eastern Gambusia removal

Flat-headed Galaxias The Flat-headed Galaxias has undergone substantial range contraction across much of the southern MDB in recent decades (Morris et al. 2001; Lintermans 2007; Fisheries Scientific Committee 2008), and in 2008 the species was listed as critically endangered in New South Wales under the state’s Fisheries Management Act 1994 . Apart from aspects of its reproductive and larval biology (see Llewellyn 2005), little information is available on the life history of the species or the causes of its decline. Habitat degradation and limitations on dispersal associated with altered flow regimes, and recruitment failure caused by releases of cold water from dams in addition to predatory and competitive interactions with alien (e.g. redfin, Common Carp and Eastern Gambusia) and native fish species (e.g. climbing galaxias Galaxias brevipinnis ) have been suggested as possible factors (McNeil 2004; Lintermans 2007). Billabongs and wetlands associated with north-central Victorian rivers have harboured healthy populations of the species in the recent past (McNeil 2004; Lintermans 2007; McNeil and Closs 2007). McNeil (2004) captured large numbers of Flat-headed Galaxias on the lower Ovens floodplain following widespread flooding of the region in winter–spring 1996. In explaining the declines in abundance observed from spring to late summer, McNeil (2004) suggested that reduced DO in small wetlands and predation by Redfin in larger ones may constrain the species distribution across the floodplain (see also Stoffels and Humphries 2003; McNeil and Closs 2007). He attributed the persistence of Flat-headed Galaxias in relatively marginal habitats to a trade-off between environmental harshness and predation pressure from Redfin, and suggested that in the absence of Redfin, the distribution of Flat-headed Galaxias would expand and its and recruitment potential would increase. Because we captured just nine small Redfin from 6 of the 83 sites sampled in the Lower Ovens region, we suggest that alternative factors must have contributed to the absence of flat-headed galaxias on the floodplain across the two years of the study. Opportunities for colonisation of our sites from the Ovens River main channel did exist in both years, as was the case during the 1999 and 2000 surveys by King et al. (2003) who captured only two individuals across both years, so the extent of connectivity does not appear be a limiting factor. Similarly, anthropogenic disturbance effects and habitat degradation are likely to be minimal at the majority of these sites, as the unregulated Ovens River provides a relatively natural cycle of flooding and drying to the floodplain, which is in state forest dominated by River Red Gum (Eucalyptus camaldulensis ). Although we did not sample these wetlands in spring when physicochemical and other environmental stressors are likely to be less severe, factors associated with wetland drying over summer (see section 3.4.1) and concurrent peaks in (1) abundance of Eastern Gambusia and (2) competitive pressure from juvenile Common Carp, Carp Gudgeons and Australian Smelt, which occupy similar dietary niches, may interact to suppress spawning and increase mortality in these closed systems. Further targeted sampling is required to assess the current status of Victorian Flat-headed Galaxias populations, but our data suggest that the species may be at risk of local extinction in a region that once supported healthy populations.

Murray–Darling Rainbowfish The absence of Murray–Darling Rainbowfish from the lower Ovens River wetlands in our study was not unexpected, because previous studies in the region had failed to detect the species (King et al. 2003; McNeil 2004). The few that we did capture in the remaining wetlands in Gunbower and BMF were at least 42 mm long (TL) and most likely in their second year of life (Milton and Arthington 1984). It is significant that we did not capture juveniles at any of our sites, as this suggests that: (a) local spawning did not occur during the previous spring – early summer, which is the known spawning period of the species

52 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

(Backhouse and Frusher 1980; Humphries et al. 2002), (b) we were unable to detect it with our sampling techniques, or (c) that biotic or physical factors (or both) increased mortality during the early life stages. Murray–Darling Rainbowfish are still collected within creek and main river channel environments throughout the Barmah–Millewa and lower Goulburn region (e.g. King et 2007; Tonkin and Rourke 2008), indicating this species has a much broader trophic niche than the two species previously mentioned. In terms of biological factors, the threats posed by Eastern Gambusia to melanotaeniids have been well documented (e.g. Arthington 1991; Arthington and Marshall 1991). Predation on eggs and larvae has been reported (Ivanstoff and Aarn 1999) and suggested (Lukies 2004), and — because of the substantial overlap in dietary and habitat niches identified for Eastern Gambusia and Murray–Darling Rainbowfish across all life stages (see Table 2.1; Lukies 2004; Kellaway et al. 2010) — we suggest that competitive and predatory impacts of Eastern Gambusia may go some way to explaining the increasing patchiness of the species across the southern MDB. If this is the case, then the late and restricted spawning time of species such as the Murray–Darling Rainbowfish may magnify the impacts (Pen et al. 1993). Spawning during October, November and December (Backhouse and Frusher 1980; Humphries et al. 2002) exposes newly hatched larvae and juvenile stages to rapidly expanding Eastern Gambusia populations, which thus could exert increasing control where emigration opportunities are limited; for example, in rapidly drying wetlands such as those sampled during our study. Little information exists on the effects of environmental factors on rainbowfish abundance. Flooding does not appear to affect spawning time or larval production in the species (King et al. 2010), and larvae rarely drift (Humphries et al. 2002). The species was never collected during a seven-year period in the highly regulated Campaspe River in central Victoria, yet it was captured in the mildly regulated Broken River (Humphries et al. 2002). But because temperature — the suggested spawning stimulus (Backhouse and Frusher 1980; Milton and Arthington 1984) — is often independent of flow, and egg development requires aquatic vegetation (Milton and Arthington 1984), the absence of the species from the Campaspe likely reflects other factors such as habitat quality and availability rather than river regulation. Because of small samples sizes in the present study it is not possible to accurately define the relative importance of each of these processes in structuring populations of Murray–Darling rainbowfish. It may be that the simple absence of the species from so many sites containing high densities of Eastern Gambusia is enough to infer that there is a negative impact (without identifying the mechanism), but clearly more data is needed to validate this inference.

Murray–Darling Rainbowfish collected from a wetland in the Gunbower Forest.

53 Native fish recovery following Eastern Gambusia removal

3.4.2 The influence of Eastern Gambusia on species occupancy, abundance and condition Identifying the factors that currently influence fish community composition across southern MDB wetlands, and defining the nature and extent of the impact of Eastern Gambusia, were key objectives of the cross-sectional study. If the wetland fish community is structured by local abiotic factors, then we would expect strong relationships to exist between physicochemical parameters or habitat variables and species abundances. Conversely, if competitive or predatory forces dominate, clear patterns in co-occurrence or relationships between relative abundances of different species are predicted. The use of correlative or species distribution models (SDMs) is an intuitive way to explore and visualise such relationships and develop hypotheses about the mechanisms underlying the patterns observed (Elith and Leathwick 2009; Capinha and Anastácio 2011). Such models can also be used in a predictive capacity to make forecasts on likely changes in species distributions and abundances under a range of future scenarios of ecological change. Boosted regression trees (BRTs) are a relatively new type of SDM, based on traditional regression and more modern machine learning techniques (Elith et al. 2008). Their use is increasing among freshwater and marine fish ecologists who require robust tools to explain and predict patterns of occurrence at numerous spatial and temporal scales (see Elith et al. 2008; Leathwick et al. 2006, 2008). The correlative models we constructed were based on this framework. They were designed to gain an insight into relationships the presence and abundance of Eastern Gambusia and the presence and abundance of other species in the sampled wetlands, and to examine the influence of physicochemical and other environmental parameters on these relationships. The results from the occupancy modelling clearly suggests that the occurrence of Eastern Gambusia in a wetland has a strong effect on the probability of the four most common native species captured in our surveys (Carp Gudgeon, Australian Smelt, Flat-headed Gudgeon and un-specked hardyhead) being present. The magnitude of the effect for these species appears to depend on the amount of aquatic vegetation, although it is notable that all four species were most likely to occur in the presence of Eastern Gambusia, regardless of the presence or absence of aquatic vegetation. The high probability of co-occurrence of these species with Eastern Gambusia is not unexpected, since Eastern Gambusia occupied 83% of all sites sampled. The results, in effect, highlight the success of Eastern Gambusia as an invader and coloniser of these wetland systems. Its broad tolerance of environmental extremes (Karolak 2006; McNeil and Closs 2007), high reproductive capacity (Milton and Arthington 1983) and ability to thrive in disturbed habitats (Arthington et al. 1983; Kennard et al. 2005) is reflected in its wide distribution in the MDB and beyond, and its extraordinary capacity to rapidly dominate fish communities following the onset of spawning (see Chapter 4). The output from the BRT models provides a deeper insight into the relationships between Eastern Gambusia abundance and potential predictor variables in the sampled wetlands. The results suggest that environmental variables have little influence on Eastern Gambusia abundance — a result which aligns well with the characteristics of a broadly distributed, highly tolerant alien species. By far the most significant predictor of Eastern Gambusia abundance was the abundance of Carp Gudgeon, with a strong positive relationship evident. From the data we have it is not possible to say whether or not these species show partial segregation in microhabitat niches at scales undetectable by our sampling, or employ other behavioural traits that limit competition between them at particular life stages (but see Stoffels and Humphries 2003). However, it is clear that both species are thriving in most wetlands across the southern MDB. Our findings also illustrate the resilience of the most common native species to high Eastern Gambusia densities. At sites where Eastern Gambusia was present during late February –

54 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

March, it was almost always very abundant. When we predicted the likely outcomes of the study we expected that such high densities combined with the increasingly harsh physicochemical conditions and other stressors associated with wetland drying would act to decrease the abundance of many common wetland species. This may have been the case with some rarer species (e.g. Southern Pygmy-perch, Murray–Darling Rainbowfish, Flat-headed Galaxias), although the small numbers of individuals collected during our study limit the conclusions we can reach for these species. In any case, the observation that Carp Gudgeon, Australian Smelt and Flat-headed Gudgeon were able to persist at many sites under such conditions (albeit at lower abundances or decreased morphometric condition at some sites) reflects the ability of these species to adapt quickly to changes in ecosystem processes associated with rapid increases in Eastern Gambusia abundance, and to resist some of the effects of drought disturbance (Morrongiello et al. 2006; McNeil and Closs 2007; McMaster and Bond 2008; Crook et al. 2010). Carp Gudgeon typically inhabited wetlands containing high abundances of Eastern Gambusia and Australian Smelt, and tended to prefer those with higher DO and moderate to high turbidity. As noted above, the positive relationships observed with Eastern Gambusia and Australian Smelt may reflect differential use of habitat and dietary resources by one or more of these species at different developmental stages (e.g. Stoffels and Humphries 2003). Any such separation of habitat or trophic niche would ultimately decrease the frequency of interactions between species, and hence competition with Carp Gudgeon. Alternatively, such resources may not be limited in some wetlands, enabling these species to persist even when ecological niches need to be shared (Stoffels and Humphries 2003; Kellaway al. 2010). In analysing species abundance data from the lower Ovens, McNeil (2004) reported that, although Carp Gudgeon and Eastern Gambusia distributions overlapped substantially on the floodplain, higher Carp Gudgeon abundances were generally associated with bigger, deeper, more permanent wetlands, lower macrophyte cover and higher pH. Eastern Gambusia abundance was, by contrast, associated with smaller, shallower, ephemeral habitats with higher macrophyte cover and lower pH. McNeil (2004) also showed that Carp Gudgeon abundance was positively correlated with minimum oxygen concentrations in these wetlands, a finding corroborated by our data. However, apart from DO and a weak positive relationship with turbidity, none of the other environmental parameters that we measured appeared to influence abundances for this species at the time of our sampling. Relative body condition in Carp Gudgeons was shown to decrease at higher Eastern Gambusia densities, but there was no such relationship with fin condition. However, caudal fin damage was considerably less at sites with higher debris loads. We conclude from this data and from previous work that a mix of biotic (i.e. competitive interactions, or their absence) and abiotic controls (i.e. DO levels, habitat complexity) have contributed to the abundance patterns and condition scores observed for Carp Gudgeon. Australian Smelt was widely distributed across our sites and was collected in moderate numbers over the two years of the study. In each year the catch was dominated by recruits spawned in the previous spring, although there were small numbers of adult fish that may have been more than one year old. At sites where Eastern Gambusia abundance was high, many Australian Smelt captured showed visible signs of caudal fin damage and were generally in poor condition. This species is known to occupy trophic and habitat niches that are similar to those of Eastern Gambusia. Combining that with our knowledge of their behaviour and physiological tolerances (see McNeil 2004; McNeil and Closs 2007), we hypothesised that Eastern Gambusia would impart a strong influence on the abundance and body and fin condition indices at sites where the two species coexisted, particularly where habitat complexity was low. The output from the BRT models revealed a weaker negative relationship between Eastern Gambusia and Australian Smelt abundance than we expected.

55 Native fish recovery following Eastern Gambusia removal

Rather, the abundance of Carp Gudgeon and wetland size were most important in explaining the abundance of Australian Smelt, which was more likely to occupy larger wetlands containing a large number of Carp Gudgeons. Interestingly, habitat complexity and physicochemical predictors had very little influence on Australian Smelt abundance, which again suggests a high tolerance to a range of environmental conditions associated with wetland drying. As predicted, the assessment of morphometric condition revealed a significant decline in both body and fin condition with increasing Eastern Gambusia abundances. Our data suggest that these effects may be exacerbated in smaller wetlands, despite the extent of habitat complexity having little effect. The positive correlation between Australian Smelt abundance and wetland size suggests that the permanence of refuge habitats may influence recruitment success and persistence of the species under conditions of environmental stress, and that larger wetlands may provide some release from competitive pressures conferred by other sympatric species that occupy similar trophic niches (Kellaway et al. 2010). Flat-headed Gudgeon was prevalent in the lower Ovens sites during our surveys, a result that contrasts markedly with previous work in region. King et al. (2003) sampled the area over two years, incorporating a year of minimal floodplain inundation during spring and summer (1999–2000) and a year of repeated flooding (2000–1). Although only small numbers of Flat- headed Gudgeon were captured in both years, a slight increase in numbers was recorded in 2000–1 coinciding with regular floods during the sampling period and prolonged periods of floodplain inundation. In late 1996, following widespread floods that inundated all floodplain habitats, McNeil (2004) failed to capture a single individual while sampling shortly after floodwaters had subsided and again later that summer. Our sampling comprised a large range of size-classes and a high proportion of juveniles, suggesting that either spawning had occurred locally or that a large influx of early-stage fish had colonised these sites during a recent connection to the main stem (see Lyon et al. 2010). This is noteworthy because the species does not typically use the floodplain for larval development (Lintermans 2007). Large adult Flat-headed Gudgeon appeared to completely dominate the fish community at some sites, which were generally characterised by low species diversity and very low abundances of other species, including Eastern Gambusia. Although we predicted a high risk of negative impacts from Eastern Gambusia, based on habitat and dietary niche overlaps, our data, combined with knowledge of the species utilisation of benthic habitats (Kilsby and Walker in press) suggests that impacts would most likely be on juvenile stages in situations where habitat and food resources were limited. At the time of our sampling in the lower Ovens region, a positive relationship was found between Flat-headed Gudgeon and Carp Gudgeon abundance, and Flat-headed Gudgeon were more likely to inhabit wetlands with relatively high DO and pH. The relative abundance of Eastern Gambusia or any other habitat or environmental variables that we measured did not substantially affect Flat-headed Gudgeon densities. These findings, in addition to the lack of a relationship between fin damage in the species and Eastern Gambusia abundance at our sites, points to a partitioning of resources between these two species, which is not unexpected given their natural benthic versus pelagic behavioural traits and habitat preferences. Whether resource use is more partitioned at sites with higher Eastern Gambusia abundances we are unable to say from our data, but clearly factors other than Eastern Gambusia abundance are the major determinants of Flat-headed Gudgeon abundance in these systems. Flat-headed Gudgeon is considered highly tolerant to low hypoxic conditions (Gee and Gee 1991; McNeil 2004), so although higher DO habitats were preferred in our study, DO is not likely to constrain the distribution or abundance of the species. McNeil (2004) reported that the sudden appearance of Flat-headed Gudgeon on the lower Ovens floodplain during 2000 (a year of repeated and substantial spring flooding) coincided with a large influx of Common Carp into

56 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal the wetlands, and suggested that the same mechanisms that drive carp recruitment and use of off-channel habitats in the region (i.e. flows) may be also determine the distribution of Flat- headed Gudgeon. We found no correlation with carp abundances in our surveys, but the importance of connectivity with the Ovens main channel for dispersal and colonisation opportunities is obvious (Lyon et al. 2010). One of the more interesting results from the study was that juvenile Common Carp, a highly successful and pervasive alien species that commonly makes use of floodplain habitats in the MDB for spawning and larval development (King et al. 2003; Koehn et al. 2004; Macdonald and Crook 2006; Stuart and Jones 2006), occurred in low numbers with Eastern Gambusia in the sampled wetlands. These species were significantly more likely to coexist in wetlands containing aquatic vegetation, and substantial numbers of Common Carp were collected only in the complete absence of Eastern Gambusia. Common Carp is highly resilient and resistant to the impacts of drought (Crook et al. 2010) and, like Eastern Gambusia, has an extreme tolerance of poor water quality and the effects of anthropogenic disturbances (Kennard et al. 2005), a very high fecundity (Sivakumaran et al. 2003) and a flexible diet (Crivelli 1981). Whether carp and Eastern Gambusia populations exert any simultaneous control upon each other in these wetlands is unknown. However, because the vast majority of carp collected were at least a few months of age, having been spawned during the previous spring, with a small number of fish probably older than one year also captured (Brown et al. 2005; Smith and Walker 2004), it is likely that any impacts associated with predatory or competitive interactions between the species would have peaked during late spring (well before our sampling), so that early-stage carp larvae would have been exposed to pressures from ever- increasing densities of Eastern Gambusia. It is important to note, however, that our sampling techniques would not have captured large adult Common Carp effectively even if they were present at our sites. The BRT models indicate that Common Carp abundance was strongly negatively correlated with Carp Gudgeon abundance. This suggests that either Carp Gudgeon are having a direct or indirect impact on carp populations (most likely through resource competition between early life stages) or that sites with high Carp Gudgeon abundances are in some way less prone to invasion by Common Carp. Our sampling was restricted to higher quality sites on the lower Ovens floodplain, but it is likely that carp may be advantaged in more degraded systems where human disturbance can act to reduce the biological resistance of the native fish fauna (Kennard et al. 2005; Crook et al. 2010). We predict that in those circumstances the influence of Carp Gudgeon and local physical factors such as pH would be lessened, and that other landscape factors such as the frequency and timing of flooding, and access to spawning and nursery habitat on the floodplain would be major factors (Crook and Gillanders 2006; Macdonald and Crook 2006; Stuart and Jones 2006).

57 Native fish recovery following Eastern Gambusia removal

Juvenile carp collected from lower Ovens wetland.

Billabong on the lower Ovens floodplain with healthy riparian and aquatic vegetation and woody debris. Does this diverse habitat availability buffer some native species from negative alien species interactions?

58 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

3.4.3 Hypotheses for the influence of Eastern Gambusia on wetland fish communities Correlative models provide a platform to examine current trends in species diversity, abundance and environmental parameters in MDB wetlands. The models we used were essentially hypothesis-development tools that could allow predictions to be made on the likely response of wetland fish communities to different levels of Eastern Gambusia reduction in these systems, and then tested in a field-based control program. Using the results from the models, we developed four hypotheses relating to the outcomes of Eastern Gambusia removal in disconnected wetlands and refugia for the most common species captured. These predictions were developed under a scenario in which there was no dispersal away from the site and no immigration into it over the course of the removal program. Because of the low numbers of rarer species collected and the dominance of Eastern Gambusia, and information in current literature, an additional hypothesis was included to allow for the absence of model predictions for these species. The hypotheses were as follows: 1. Removal of Eastern Gambusia (to any level) will benefit Australian Smelt populations, as measured by (i) increases in relative abundance and (ii) improvement in morphometric condition. 4. Removal of Eastern Gambusia (to any level) will impart minimal changes to Carp Gudgeon and Flat-headed Gudgeon abundances. 5. Removal of Eastern Gambusia (to any level) will improve morphometric condition for carp and Flat-headed Gudgeon. 6. Removal of Eastern Gambusia (to any level) will impart minor increases to carp abundances. 7. Removal of Eastern Gambusia (to any level) will benefit rarer wetland species such as Southern Pygmy-perch, Flat-headed Galaxias and Murray–Darling Rainbowfish populations, as measured by (i) increases in relative abundance and (ii) improvement in morphometric condition. Our testing the validity of these hypotheses in a field-based removal experiment is the focus of the next chapter.

59 Native fish recovery following Eastern Gambusia removal

4 Phase 2: Trial of Eastern Gambusia control 4.1 Introduction Despite the vast amount of literature devoted to the Gambusia species, there is a surprising lack of knowledge about factors controlling their abundance, impacts on native ecosystems, and the mitigation of these impacts (Pyke 2008). Studies examining trophic interactions (as suggested by Zavaleta et al. 2001) and subsequent ecological effects of small alien species such as Eastern Gambusia have received very little attention, particularly in comparison to larger species such as Common Carp (e.g. Wilson 2006). Whilse we have documented a number of controlled experimental studies in this review, rigorous comparisons of control success under field conditions have not been undertaken. Rowe et al. (2008) recommended full BACI (before/after control/impact) or manipulation studies (or both) for the assessment of impacts of alien species on indigenous species. Such experiments have not yet been undertaken on Eastern Gambusia. Furthermore, information regarding how native species respond to reductions of an alien species in the wild is necessary before large amounts of money are spent on large-scale control strategies. Rowe et al. (2008) noted that ‘scientific proof of the impact of gambusia on indigenous biodiversity is likely to be required in the future as management efforts to control gambusia increase in number and size and therefore attract closer public scrutiny of cost’. Field-based manipulation experiments can provide scientifically defensible information on an alien species’ impact in the natural environment (Rowe et al. 2008) and also valuable information on controlling populations of alien species to a level that results in a measurable improvement in native fish communities. Although there is global recognition of the threats Gambusia pose to freshwater ecosystem function outside their native range, there is very little information available on mitigating the impacts. This is largely due to traditional techniques including poisoning, exclusion, egg dehydration, direct removal, commercial harvest and habitat restoration having minimal chance of success for smaller species such as Eastern Gambusia which occupy more cryptic habitats (Wilson 2006). The few documentations of Gambusia species control both internationally and within Australia focus predominantly on chemical techniques and drying of habitats, (see McKay et al. 2001). More recently, some consideration has also been given to management options in systems where total eradication using chemical treatments is undesirable because of the presence of threatened species or fragile ecosystems, e.g. Maynard et al. (2008) and Brookhouse and Coughran (2010). These techniques all call for extremely thorough treatment and extensive knowledge of the hydrology of the area. With such limited information available for managers, research should be directed towards the feasibility of existing and new techniques for mitigating the impacts of Eastern Gambusia. This chapter reports on Phase 2 of the project — a field-based trial of Eastern Gambusia control in small isolated billabongs. The trial had three broad objectives: 1. Investigate aspects of Eastern Gambusia population dynamics to inform pest management practices for the species. 2. Investigate the effectiveness of a variety of physical removal techniques. 3. Assess the response of native fish populations following reductions of Eastern Gambusia populations, by determining the level of support for each hypothesis (model validation) derived from the predictive models developed in Phase 1 of the project.

4.2 Methodology The trial involved two major components. The first was the removal of Eastern Gambusia at specific sites, ultimately creating a range of abundances across sites. The second was the

60 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal monitoring of the fish community to investigate the impacts of different abundances of Eastern Gambusia and ultimately assess what responses we might expect from native fish communities following Eastern Gambusia removal. Because of the high variability in fish assemblages at small off-channel sites, both between sites and across years within sites (derived from the results presented in Phase 1 of the project; see also King et al. 2007), and, more importantly, a lack of information on methods for physically removing Eastern Gambusia, the trial was undertaken over two seasons in an attempt to minimise the risk of factors such as flooding affecting the experiment, while maximising the project outputs that could be achieved within the budget and time-frame.

4.2.1 Site description and treatments The trial was undertaken in wetland habitats on the lower Ovens River floodplain and (in the second year only) on private land on the lower Goulburn – Campaspe catchments. These areas were also where much of the cross-sectional data was sourced (see chapter 3.2.1). Site selection was highly dependent on winter–spring hydrology (site connection, accessibility, etc.) as well as the predictions of the correlative models, especially in relation to species composition. During the first season of the experiment (2009–10), all sites were located on the lower Ovens River floodplain. The area experienced some minor flooding during early spring, filling floodplain habitats such as anabranches and billabongs which had been dry for several years. The majority of sites during the second year of the experiment (2010–011) were shifted to the lower Goulburn and Campaspe region given the major flooding of the Ovens river sites in spring and long range forecasts predicting further flooding in the region. The new sites were all on private land, where there was a much lower risk of reconnecting to adjoining habitats during late spring and summer compared to sites on the Ovens River. Several sites on the Ovens River used in the first season were used as reference sites in 2010–11 (see below). The population dynamics of Gambusia species involve a population increase, peak, decline, and low phase. Based on pest management theory for organisms with similar population synamics (e.g. rodents), we hypothesised that the best time to undertake control operations would be during the low phase or early in the increase phase (Ramsey and Wilson 2000). In the case of Eastern Gambusia this is in late winter – spring (spawning begins in early summer when water temperatures exceed 16 °C and day length exceeds 12–13 hours). Unfortunately, late winter and early spring is also when the probability of flooding is highest, which would make long-term removal difficult and risk confounding results of any removal that may precede reconnection to other aquatic habitats. Therefore, to minimise this risk, and still maintain removal exercises commencing during the most effective period, the field depletion experiment commenced in late October just prior to the onset of gambusia spawning, and following flooding and connectivity to adjoining aquatic habitats. Based on the hypotheses derived from Phase 1, the project steering committee agreed that, ideally, sites used in the field depletion experiment should be selected on the basis that they: • contain a high Eastern Gambusia density • contain at least two native species across all sites • are smallish in size (not more than 400 m2 in area) yet should still retain water throughout summer • have an accessible perimeter • have similar habitat (density snags and aquatic vegetation) • support few other alien species (particularly Redfin).

61 Native fish recovery following Eastern Gambusia removal

Finding sites that met all of these criteria was difficult, mainly because of flooding of the region in previous months. Nevertheless, the proposed site selection guidelines were followed as closely as possible using data from fish surveys in adjoining waterbodies in recent years (e.g. if large numbers of Eastern Gambusia occupied an adjoining waterbody, it was assumed that they had migrated into a newly inundated site) as well as random seine netting and staff experience of sampling in the region (e.g. sites that were known to retain water throughout the summer period). This resulted in the selection of 13 sites in both the 2009/10 and 2010/11 seasons (some sites were repeated in the 2010/2011 season). Sites were assigned to one of two treatments (repeated removal or single removal) or one of two control types (control or reference). The treatments and controls were as follows: • Repeated removal — 6 sites (3 sites in each season) × removal of Eastern Gambusia for a period of 5 days each month + monitoring from November until February (i.e. monthly removal). • Single removal — 4 sites (2 sites in each season) × removal of Eastern Gambusia for a period of 5 days during November only (release thereafter) + monitoring from November until February (i.e. a single spring removal). • Control (no removal) — 6 sites (3 sites in each season) × release of all Eastern Gambusia + monthly monitoring from November until February (i.e. no removal). • Reference — 4 sites (4 sites in each season) × monthly monitoring from November until February. Examples of the sites used in the study and the experimental timeline are presented in Figures 4.1.1 and 4.2.1 respectively. The conclusion of the removal experiment also coincided with the time when the cross-sectional study was undertaken (February – March), enabling a direct comparison.

62 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Figure 4.1.1 Examples of removal sites used in the field trial.

63 Native fish recovery following Eastern Gambusia removal

4.2.2 Removal of Eastern Gambusia Eastern Gambusia removal was carried out each month, from late October – early November until late February, at each of the repeated removal, single removal and control sites (Figure 4.2.1). Each removal was generally carried out over a 5-day period, but the first removal in each season was extended over a 10-day period because of the extra effort required to undertake the mark–recapture component (see below). Eastern Gambusia removal was carried out at both treatment and control sites, the difference between each treatment being whether Eastern Gambusia were permanently removed and euthanised, or released back into the site after capture. For example, the same trapping regime was carried out at all sites, along with an equivalent effort of seine netting (based on the number of shots per square meter of site), during each sampling trip. This standardised any influence the removal exercise may have had on the native fish population. Reference sites were not subject to any removal exercises.

Treatment

Repeated M,R, E M,R, E M,R, E M,R, E

Single M,R, E M,R M,R M,R

Control M,R M,R M,R M,R

Reference M M M M

Oct Nov Dec Jan Feb Mar

Figure 4.2.1 General timeline of the experiment for each of the two years, including exercises carried out for each of the treatments (repeated removal and single removal) and controls (control and reference). M = monitoring; R = removal; E = Eastern Gambusia permanently removed and euthanised.

At removal sites we attempted to reduce fish numbers to as close to zero as possible. We used targeted seine netting and a variety of trapping techniques to maximise the efficiency of removal of Eastern Gambusia while minimising the impact of the removal activities on sympatric native species. The specific techniques described below were based largely on work presented by Maynard et al. (2008) in an attempt to exploit Eastern Gambusia behaviour; specifically, its attraction to light and heat, and other microhabitat preferences (see chapter 2.3). Standard collapsible bait traps It was decided that standard collapsible bait traps would be the main passive removal technique used in the experiment because of the results obtained by Maynard et al. (2008). By applying a ranking system to compare catch data for four trap designs, they concluded that the standard collapsible bait trap containing a light source and set in a manner where the inception area (entrance) was near the water’s surface was the best design to employ in an

64 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Eastern Gambusia control program. This was because it has a high CPUE during winter and is selective for large breeding females. It is also readily available, cheap, and easy to deploy. Therefore, at each site and for each trip, up to ten 10 bait traps were deployed. Although all traps were set so that the inception area was near the surface of the water, several variations in setting were applied in the first season to enable an assessment of the best technique for maximising Eastern Gambusia collection using the standard collapsible bait traps. Specifically, for the first season of the field depletion experiment (2009/2010), the traps: • were set for a minimum of three daylight sets (approximately 0900–1700 h) and three night-time sets (approximately 1800–0800 h) for each sample trip • contained three illuminated traps (either solar lights which operate between dusk and dawn, or a 12-hour chemical light stick placed in the trap at dusk) and a minimum of three (maximum of seven) unilluminated traps during night-time sets for each sample trip (Figure 4.2.2a) • contained three traps baited with dry pet food for a minimum of three sets during daylight hours (approximately 0900–1700 h) and three night-time sets (approximately 1800–0800 h) for the first sample trip. At the completion of each set, the number of each species of fish captured was recorded. All fish were then released unless the site was a designated removal site, in which case all Eastern Gambusia were retained and euthanised. Therefore, in addition to the primary purpose of collecting Eastern Gambusia for the field removal experiment, the traps set in the first season provided data that was analysed to determine whether the addition of light or bait, or being set during the day or night, improved the catch rate of Eastern Gambusia. An assessment of trapping variables on the catch rates of native species was also undertaken because these traps are commonly employed for general fish surveys (e.g. in the Murray–Darling Basin Authority’s Sustainable Rivers Audit). The second season of trapping involved the same number of replicates, but the variations were refined in accordance with the results of the first season.

Fyke netting To complement the bait traps used during the first season, we set a fine-mesh, single-wing fyke net at each site for one daylight set (approximately 0900–1700 h) and one night-time set (approximately 1800–0800 h). As with the collapsible bait traps, the inception area was set at the water’s surface (Figure 4.2.2b). For the first removal in the first season, a 12-volt fluorescent light was positioned in the rear chamber of the net during night-time sets in an attempt to increase the attraction of Eastern Gambusia. This was discontinued after the first trip. For each set, catch data (CPUE) was standardised to the number of fish collected per hour to allow for the shorter soak time employed during the daylight set (approximately 8 hours compared to 14 hours soak time).

65 Native fish recovery following Eastern Gambusia removal

(a)

(b)

Figure 4.2.2 (a) Standard collapsible bait trap containing a solar light which automatically operated between dusk and dawn, and recharged throughout the day, and (b) single-wing fine-mesh fyke net. Note that both traps are set so the inception area is at the water surface.

66 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Targeted seine netting Because Eastern Gambusia is generally an open water species, swimming at or near the surface, seine netting is a very reliable method for their capture. However, this method can be difficult where there is complex habitat or the banks are steep, and it is potentially destructive to habitat such as aquatic vegetation and fragile organisms such as larval fish. To overcome these disadvantages, the lightweight seine nets that were used in the cross-sectional study (7 m long × 1.5 m drop, 1 mm mesh diameter, 500 mm square purse) were employed in a manner that specifically targeted Eastern Gambusia. This was achieved by exploiting two aspects of Eastern Gambusia behaviour: 1. Eastern Gambusia generally occur at or near the surface throughout spring and summer, making them relatively easy to locate (using polarised sunglasses). 2. Eastern Gambusia’s affinity to warmth and light (see literature review) makes the areas it occupies within specific sites very predictable, particularly during times when water temperatures are low. For example, during late afternoon the fish will be concentrated in in areas that receive the last remaining sunlight. Subsequently, seine netting was employed only when Eastern Gambusia were directly observed occupying an area, or to target an area where they were likely to be concentrated. Once fish or such an area was located, the seine net was deployed to encircle the fish or the designated area, and then retrieved (Figure 4.2.3a). At the completion of each targeted seine netting the numbers of Eastern Gambusia were recorded, along with the presence of any other native species (which were then released). The number of seine nettings was also recorded for each site to document the effort of the removal exercises. A single seine netting took anywhere from 30 seconds to 15 minutes, as it was highly dependent on the abundance of both Eastern Gambusia and native species. An attempt was also made to mimic Eastern Gambusia’s preferred areas using 100 watt, halogen globe spotlights powered by deep-cycle, heavy duty 12-volt batteries. These lights not only produced an immense amount of light, but also heat. During the first sample trip for each site, one light was erected in the area receiving the last sunlight. Each was positioned at a 45 ° angle, 30 cm above the water and was triggered approximately one hour before darkness (Figure 4.2.3b). Each light was left on for approximately two hours after dark, upon which time, a single seine netting circling the illuminated area was made, and the catch assessed. Although the technique concentrated large numbers of zooplankton and native species, not a single Eastern Gambusia was collected using this technique at any of the sites so it was not repeated after the first sampling in the first season.

Sweep-net electrofishing Although sweep-net electrofishing is used only as a general monitoring method, any Eastern Gambusia collected within removal sites using this technique were euthanised. (See Chapter 3 for a description of the technique.) The data generated from this method were analysed in the methods assessment.

67 Native fish recovery following Eastern Gambusia removal

(a)

(b)

Figure 4.2.3 (a) Adult Eastern Gambusia collected during a targeted seine netting, and (b) trial of artificial light and heat using 100 watt halogen spotlights to attract Eastern Gambusia.

68 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

4.2.3 Monitoring fish biota All sites were monitored for fish biota to test the hypotheses generated in Phase 1 of the project. This enabled us to assess Eastern Gambusia population trends (including the effectiveness of removal and subsequent population growth) as well as any response of native fish to the removal. During each trip (monthly from November to February), the fish biota of each site was assessed by a combination of:

• Catch data generated from the removal exercises over the first four days (see chapter 4.2.2) given the methodologies used for the Eastern Gambusia removal (bait trap, fyke nets and seine nets) are also efficient at capturing native species; and • Sweep net electrofishing (SNE; employed in the cross-sectional surveys) to enable a direct comparison with results obtained in phase 1 (see chapter 3.2.2).

A minimum of three 20-second SNE passes were made before removing any Eastern Gambusia, within complex habitats in the littoral zone (aquatic vegetation, snags/debris). As with removal methodologies, at the completion of each SNE pass all fish were identified and counted. For each site, a minimum of twenty individuals of each species were randomly selected and measured for total length (TL; nearest mm). This enabled an assessment of recruitment, and subsequent tracking of a cohort’s growth throughout the study period.

Additionally, samples of young-of-year Carp Gudgeon, Flat-headed Gudgeon, Australian Smelt and Murray–Darling Rainbowfish collected from some sites in January and February were euthanised in an overdose of anaesthetic (immersion for 10 mintes in 40 mg/L alfaxalone) and then preserved in 95% ethanol to enable further measures on condition to be made in the laboratory. Only fish collected in January and February were assessed given this was the period where young-of-year native species and high abundances of Eastern Gambusia coincided. All fish were measured for standard length (Ls - nearest 0.1mm) and weight (nearest 0.001g) after a minimum of 10 days to allow shrinkage associated with preservation to stabilise (Fey & Hare 2005). Individual fish were also examined for fin condition using the same assessment protocol as described in Chapter 3. Any Southern Pygmy-perch collected were released because of their rarity in the region, but photographs of juvenile and adult fish were taken for assessing fin damage (Figure 4.2.4).

Figure 4.2.4 Adult Southern Pygmy-perch. Photographs were taken of rarer species for assessment of caudal fin condition.

4.2.4 Mark-recapture assessment A mark-recapture component was undertaken at selected sites to compliment estimates of Eastern Gambusia removal efficiency and population growth derived from the monitoring

69 Native fish recovery following Eastern Gambusia removal data. In the first season this was undertaken at two of the sites during the first removal (in spring) and at another two sites in February. It was envisaged that this would occur at all sites, but the extremely low numbers of Eastern Gambusia i spring 2009 and the subsequent difficulty with their capture meant that only the two sites with the highest numbers of Eastern Gambusia present were used. In the second season the mark–recapture assessment was undertaken at three sites during the first trip of the season, but was not repeated in February as a result of flooding throughout the entire experimental period, which compromised any interpretation of population change (because of influences of immigrating and emigrating fish). For each of the sites where mark–recapture was used, all Eastern Gambusia collected on the first day were marked and released back into the site. Marking was by osmotic induction, which involves a chemical dye (calcein) that creates a fluorescent mark on bony structures of the fish, such as fin rays and otoliths (see Appendix 1 for details). Marked fish can be rapidly identified without being euthanised, using a handheld blue light. The procedure was favoured for this project as it can mark a large number of fish and minimise handling before release, compared to other techniques such as fin clipping and tagging. In this procedure a 5% saline solution was prepared by dissolving 100 g of commercially available natural salt in 2 L of water. A 0.5% solution of calcein was prepared by adding 10 g of powdered calcein (2,4-bis-[N,N0-fdicarbo methylg-aminomethyl] fluorescein) to 2 L of water. The consequential decrease in the pH of the solution was counteracted by gradually adding sodium hydroxide until a pH of 7.0 was obtained. The salt and dye solutions were aerated by bubbling air through the solutions during the marking procedures (Figure 4.2.5a). Up to 20 Eastern Gambusia were placed together in a 2 L plastic container having a mesh bottom and immersed in the salt solution for 3 minutes. After immersion in the salt solution the fish were rinsed in fresh water and then immersed in the calcein dye solution for 5 minutes. They were then placed in a 300 L bin of aerated fresh water and assessed for condition. When their condition was judged to be satisfactory they were released where they were collected. Removal of Eastern Gambusia at these sites commenced two days later to ensure that marked fish regained normal behaviour. Eastern Gambusia collected afterwards were exposed to a handheld blue light, which caused bony structures such as the spines, fins and head on any fish that had been subjected to the calcein dye to fluoresce, thus identifying it as an individual from the initial capture (Figure 4.2.5b). The overall number of marked fish collected during each visit gave an estimation of the population size of Eastern Gambusia, as well as an estimate of the proportion of the population removed in the experiment.

70 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

(a)

(a)

(b)

Figure 4.2.5 (a) Field set-up for Eastern Gambusia osmotic induction marking with calcein, with examples of marked fish. (b) Detection of marked Eastern Gambusia, with an example of a marked fish alongside an unmarked individual.

71 Native fish recovery following Eastern Gambusia removal

4.2.5 Analysis 4.2.5.1 Assessment of Eastern Gambusia removal methodologies Because of the large amount of data, we assessed the methodologies for the first season only, allowing us to refine the techniques for the second season of removal. To assess the catch rates of the different methods used (and subsequent ranking of each method as a removal method) we analysed the standardised catch data using an unbalanced Analysis of Variance (ANOVA), using sample trip and method as the factors and blocking for site. The response variable was Eastern Gambusia abundance, but species-specific abundances were also tested. This enabled a comparison between bait trapping, fyke netting, sweep net electrofishing and seine netting CPUEs. All trapping data was standardised to the number of fish per hour soak time, and targeted seine netting data was standardised to the number of fish per pass. The sweep net electrofishing data used for the monitoring was kept standard as the number of fish per 20 second pass. We considered a single targeted seine pass to take a similar amount of time as an electrofishing pass, however, processing times (often a large by-catch of native species) and observation time (time it would take to observe fish) is highly variable, and as a result, is explored in the Discussion (section 4.4). To assess the variables incorporated within each trapping regime, the standardised catch data for bait trap and fyke netting was analysed (separately) using an unbalanced Analysis of Variance (ANOVA), and using site, sample trip, and trap variable as the factors (along with their interactions with trip), blocking for site. The primary response variable tested for was Eastern Gambusia abundance, but species specific abundances were also tested. This enabled an assessment of catch data in relation to: • whether the trap was set during the day or night • whether the trap had a light or not • whether the trap was baited or unbaited • whether any of the above was influenced by trip (interactions). The analysis of methods did not consider the size of Eastern Gambusia collected. This was mainly because only mature fish were present during the first removal trip, but also because we felt that targeting for a specific size range of fish would be futile given the rapid growth rates and subsequent maturation the species exhibits. As removal success depends largely on the removal of fish prior to the onset of the spawning season, we predicted that maximising the effectiveness of removal strategies during the first trip would be of the greatest importance. Therefore, an assessment of Eastern Gambusia catch for each of the methods during the first trip was also undertaken by analysing the standardised catch data using an unbalanced ANOVA for the first trip, using method as the factor, blocking for site.

4.2.5.2 Population growth and impacts of Eastern Gambusia The results from the monitoring data provided information on population dynamics for several species and, importantly, how it relates to the abundance of Eastern Gambusia. Unfortunately the analysis was restricted to the first season of the experiment, since any information used to inform population growth was compromised by frequent flooding (and thus frequent immigration and emigration). There were also numerous missing values within the time series database (given the absence or omission of data caused b y drying or flooding of sites) as well as small numbers of other species collected in the experiment. The analysis was therefore restricted to assessments of Eastern Gambusia (Gh), Carp Gudgeon (Hsp), Australian Smelt (Rs) and Common Carp (Cc).

72 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

The data consisted of a time series of observations of the four species over four time periods from 13 sites between October 2009 and February 2010, with the sampling periods approximately 1.3 months apart. Each species was sampled using three methods (SNE, bait traps and seine nets), and the abundance of each species was expressed as the mean count averaged over all sampling methods. To monitor population change we used a non-linear state-space model (SSM) to fit a mathematical model of population growth to the observed counts to estimate the trajectory of each species through time. In addition to the population growth rate, the SSM also included parameters that governed interactions between species. The SSM considers that the observed data arise from two distinct processes: the system process , which models the underlying biological process of population growth, and the observation process , which takes into account the error associated with the sampling process (King et al. 2010). The underlying model we used for the system process to model population growth and species interactions was a discrete time Gompertz model (e.g. Dennis et al 2006). The system process equation is given by N = N exp( a + b ln N + E ) (equation 1) t t−1 t−1 t where Nt is the population abundance at time t, a and b are constants, and E is the process error or stochastic variation around the true biological process. E is assumed to be normally distributed with mean 0 and variance σ2. On a logarithmic scale, the Gompertz model is a linear autoregressive time series model of order 1. Thus X = X + a + bX + E (equation 2) t t−1 t−1 t where Xt = log( Nt), a represents the intrinsic rate of increase and b governs the strength of density dependence. When b = 0 the model corresponds to one with no density-dependence (i.e. density independent growth). However, the true (log) population abundances are not observed directly because of errors induced from the observation or sampling process. Hence the observed counts Yt are given by the observation process Y = X + U t t t where Yt are the observed (log) abundance data and Ut is the random variability that results from the sampling process assumed to also be normally distributed with mean 0 and variance τ2. Hence the full SSM model is specified as X = X + a + bX + E t t−1 t−1 t Y = X + U t t t E ~ N ,0 σ 2 t ( ) U ~ N ,0 τ 2 (equation 3) t ( ) We can also generalise equation 3 to multiple species to estimate population growth of each species as well as interactions between species. If there are p interacting species, a multivariate version of equation 3 is given by X = X + A + BX + E (equation 4) t t−1 t−1 t

Where Xt is a vector of size p of (log) abundances of each species, A is a vector of growth rates, B is a p × p matrix whose elements bij give the effect of the abundance of species j on the per-capita growth rate of species i, and E is a vector of process errors (Ives et al. 2003).

73 Native fish recovery following Eastern Gambusia removal

A useful addition to equation 4 occurs when there are other covariates that could be used to model population growth. This is easily accommodated in the model by adding a term for covariate effects: X = X + A + BX + CZ + E (equation 5) t t−1 t−1 t t

Where Zt is a vector of covariate values and C is a p × p matrix of the effects of covariate j on species i. For the present analysis, we had a single covariate describing whether Gambusia removal took place at a site or not. Estimation of equation 4 is usually undertaken using maximum likelihood via either the EM (expectation maximisation) or Kalman filter algorithm. Here we take a Bayesian approach and fit equation 4 using MCMC techniques constructed in OpenBUGS software. Initial attempts at fitting equation 5 ran into problems because of the short nature of each time series (four observations) and the large number of missing observations (30% of observations were missing in the Eastern Gambusia time series). As a result there was poor mixing of the chains, and the model failed to converge. To remedy this we fitted a simpler version of equation 5 that excluded all density-dependent terms, so that it assumed that each species was subject to exponential growth modified by potential interactions with other species. Hence, the density-dependent regulation terms in the model were set to 0. We also assumed that the only species interactions were the effects of Eastern Gambusia on native species, and that the treatment effect (Eastern Gambusia removal) applied only to Eastern Gambusia population growth. Parameter estimates for the fit of this simplified version of equation 5 were sampled from 4 chains following a burn-in of 10 000 samples, thereafter a further 40 000 samples were taken with a thinning rate of 10, leaving a sample of 4000 to give the posterior distribution of each parameter. Convergence was assessed from the Gelman/Rubin diagnostic, with a values less than 1.05 used as the criteria for convergence of the chains.

4.2.5.3 Impacts of Eastern Gambusia on morphometric condition of juvenile native species To test the hypotheses based on Eastern Gambusia abundance influencing the condition of native species, the juveniles of common native species collected in December, January and February were assessed for both fin condition and body condition. As in the cross-sectional component, body condition was measured using the relative condition factor (Froese 2006). Length (TL) and weight (W) data of juvenile fish were used as an assessment of general body form of juvenile Australian Smelt, Carp Gudgeon and Flat- headed Gudgeon, using the equation b W = aL where W is the weight in grams, L is the TL in mm, and a and b are constants. Length–weight relationships could then be used to compare body condition using the relative condition factor, where

b Krel = W/ aL

General linear models were used to assess the influence of Eastern Gambusia abundance on the condition of juvenile Carp Gudgeon, Flat-headed Gudgeon and Australian Smelt. Site specific Eastern Gambusia abundance (derived from monitoring data) were log 10 transformed for normality and used as the fixed factor, with relative condition of juvenile fish used as the dependent variable for each of the species.

74 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Each of the individual juvenile fish subject to the body condition assessment was also assessed for fin damage. Using the same protocol as that in cross-sectional analysis, a categorical assessment was made based on the degree of damage to an individual’s caudal fin. The scoring system was as follows: 1 = no or minimal damage —no sign of damage to fin, or only minor splitting of fin rays. 2 = moderate damage — part of the fin was missing, but more than 50% remained. 3 = major damage — less than 50% of the caudal fin remained. Fin scores were subsequently converted to either 0 (scores of 1), or 1 (scores of 2 or 3), i.e. ‘no damage’ or ‘damaged’. Logistic regressions (modelling of binomial proportions) were used to assess the relationship between Eastern Gambusia abundance and the probability of juvenile fin damage for each of the three native species, using the log 10 Eastern Gambusia abundance as the fixed factor and fin condition score as the dependant variable. A separate species-specific analysis on relative condition was conducted for each of the years because of the possibility of morphological differences between populations specific to the two regions of the study (i.e. year one was conducted in the lower Ovens, but year two predominantly the lower Goulburn – Campaspe). 4.2.5.4 Mark-recapture component An estimate of the Eastern Gambusia population size was generated using the Chapman estimator (Seber 1982) closed population mark – recapture method for each of the sites where mark–recapture was used. This method was applied because of its widespread use and theoretical basis, as it assumes that: 1. the population is geographically closed 2. the population is demographically closed with no births or deaths 3. no marks are lost or missed 4. marking does not change fish behaviour or vulnerability to capture 5. marked fish mix at random with unmarked fish 6. all have equal probability of capture (Seber 1982; Hayes et al. 2007). The Chapman estimator for population size is given by

( n + 1)(n + 1) ^ 1 2 N = (m 2 + 1)

where n1 = number caught and marked during the first sampling period; n2 = number caught in second sampling period; and m2 = number of marked animals in second sampling period. After the second season of marking we suspected that the protocol might have caused an increased mortality of the species, which is not desirable for subsequent population measures derived from recapture data. A laboratory study to assess the marking protocol’s impact on the short-term mortality of Eastern Gambusia confirmed these concerns (see Appendix 1). Therefore, to improve estimates of population size and capture rates, the initial number of fish marked included only female fish ( n1 in the Chapman estimator equation). The revised n1 was also reduced by 25% and the equation was re-run, giving an estimated population range rather than a single value. This was based on the mortality figures generated in the laboratory trial (see Appendix 1). The population estimates were then used to generate removal efficiencies.

75 Native fish recovery following Eastern Gambusia removal

4.3 Results 4.3.1 Site parameters Season one Removal and subsequent monitoring commenced during the week of 26 October 2009. Monitoring of river levels and temperatures during this period confirmed the experiment commenced after the last connection to adjoining aquatic habitats as a result of flooding (Figure 4.3.1), and before Eastern Gambusia had initiated spawning. In general, sites were of similar size and had similar water quality and habitat complexity, all being within the largely natural river redgum floodplain of the Ovens River (Table 4.1). During the final trips in summer, several sites either dried completely, or had drawn down to extremely low levels, most likely as a result of lowered groundwater levels and/or disturbed substrates during the drought (Figure 4.3.2). In these circumstances, monitoring data that was collected during this period was excluded from the analysis. 132

130 Site connection height

128

126

Gauge height (m) height Gauge

124

Oct Nov Jan Jul Aug Sep Dec Feb Mar Figure 4.3.1 Ovens river height at Peechelba bridge between July 2009 and March 2010. The red dashed line represents the approximate river height at which experimental sites connect to adjoining river or anabranch habitats. Grey blocks represent the removal and monitoring periods of the first year of the field-depletion experiment.

Season two In contrast to the previous year, the 2010–11 season experienced major flooding before the trial and also several weeks after initial removals. Even sites that did not frequently connect directly to major water courses (and thus perceived to be ‘low risk’ of flooding) connected to adjoining treatments as a result of both localised flash flooding and widespread regional flooding. As a result our experimental treatments were compromised, with major fish immigration and emigration between sites post-treatment. Nevertheless, sites were still subject to the monthly monitoring protocol, providing information on recolonisation of Eastern Gambusia after control actions along with information on juvenile condition in relation to Eastern Gambusia abundance. Habitat parameters collected for each site at the commencement of the experiment are presented in Table 4.1. Being largely located on private land, the majority of sites (other than the Ovens river sites from the previous season) had some degree of disturbance including access by domestic stock and lower structural habitat complexity.

76 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

(a)

(b)

Figure 4.3.2 Example of site variability during the course of the project. (a) Site 2 in late October, and three months later at the end of January during the 2009–10 season. (b) Hillgrove’s site at the commencement of the experiment, and following localised flooding in the area that resulted in connection with the adjoining creek and control site.

77 Native fish recovery following Eastern Gambusia removal

Table 4.1. Treatment type, habitat and water quality parameters for each site, recorded at the commencement of trials in 2009–10 and 2010–11. Categories H = high; M = medium and; L = low. Note that all sites in 2010–11 were subject to frequent or ongoing connection to adjoining habitats after the trial had commenced. Aquatic Snag/Debris Conductivity Turbidity Surface Max. depth Site Treatment Size (m 2) Veg. (H,M,L) (H,M,L) pH (µS/cm) (NTU) DO (mg / L) temp. (°C) (m) 2009–10 1 Repeated 225 H M 6.48 49.8 2.9 3.55 29 1.8 2 Single 600 L M 6.01 45.8 8.5 2.3 26.6 3 19 Control 900 H M 5.94 38.8 10.9 3.73 32.4 2.5 20 Repeated 180 H M 6.26 15.9 20.6 2.4 23.3 2 21 Single 180 H M 6.24 18.2 33 2.24 23.3 1.3 22 Control 120 H M 6.17 62.2 9.4 1.8 22.5 2.3 8 Repeated 180 L M 6.89 47.5 13.1 2.32 26.2 1.5 9 Control 176 L M 6.76 464 25.3 2.63 24.8 1.8 ST Ref 625 L M 6.22 83 27 1.24 21 2 2010–11 Willow 1 Repeated 300 M L 7.2 131 36 8.55 20.5 1.5 Willow 2 Control 320 M L 7.2 142 104 8.5 21.6 1.5 Hillgrove’s 1 Single 400 M L 6.82 126 48 6.77 21 1.8 Hillgrove’s 2 Control 250 M L 6.39 126 19 5.08 19 1.3 Browns 1 Repeated 150 L M 6.5 164 8 6.74 15.2 2 Browns 2 Control 400 L M 6.5 164 8 6.74 15.2 2 Cornella Ck Single 1000 L M 7.08 450 24 6.48 18 3 Colbo wetland Repeated 1225 L M 7.37 169 141 6.8 21.5 1.2 1 Ref 244 M M 6.54 5.12 20.1 8.1 18.6 1.8 2 Ref 600 M M 7.78 43.2 40.9 7.57 15.2 3 19 Ref 900 M M 7.34 50 10.1 11.3 21.2 2.5 ST Ref 600 M M 6.9 53.5 11.2 10.48 19.5 2

78 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

4.3.2 Methodology assessment In the first season, 11 397 fish were collected from the eight sites (excluding reference sites) (Table 4.2). Eastern Gambusia was the most abundant species collected ( n = 5165), followed by Common Carp (n = 3039) and Carp Gudgeon ( n = 2872). The analysis of the standardised catch data indicated that the number of Eastern Gambusia collected was significantly influenced by both trip number ( p < 0.001), and the method of collection ( p < 0.001). Specifically, the targeted seine netting was by far the most effective method for collecting Eastern Gambusia when comparing standardised efforts across all trips (Figure 4.3.3a), and more importantly, for the first trip (Figure 4.3.3b), which only collected mature fish prior to spawning. Method type also had a significant influence on collection of each of the native species (all p < 0.001). The seine netting was also significantly more efficient at collecting the two most abundant native species, Carp Gudgeon and Australian Smelt, than the trapping methods (Figure 4.3.4). CPUE of southern pygmy-perch were highest for fyke nets. The different variables employed in the trapping methods (bait traps and fyke nets) were also assessed for differences in CPUE of Eastern Gambusia and native species. The analysis of the standardised catch data indicated that the number of Eastern Gambusia collected by both bait traps and fyke nets was significantly influenced by trip number ( p < 0.001 and p < 0.01 respectively), but not whether the trap was set during the day or night (both p > 0.05). Eastern Gambusia CPUE for bait traps was higher during the day than night (Figure 4.3.5a), but this difference was not statistically significant (p = 0.07). For traps set at night, there was no significant difference in Eastern Gambusia CPUE between traps that did and did not contain a light (both p > 0.05). There was also no significant difference between baited and unbaited traps during trip one (p > 0.05). The pattern in trapping variables and CPUE for native species was similar to those for Eastern Gambusia, in that the CPUE for Carp Gudgeon, Australian Smelt and Southern Pygmy-perch was not significantly influenced by whether traps were set during the day or night, contained a light or not, or were baited or unbaited (all p > 0.05).

Table 4.2. Raw abundances of each species collected from removal and control sites, for each trip using all removal methodologies during the field depletion experiment in season one. Species Trip 1 2 3 4 Total Eastern Gambusia 52 225 1718 3170 5165 Carp Gudgeon 196 228 1073 1375 2872 Flat-headed Gudgeon 1 3 1 0 5 Australian Smelt 16 51 122 9 198 Southern pygmy-perch 3 20 4 5 32 Common Carp 506 1444 786 303 3039 Goldfish 4 0 34 4 42 Oriental weatherloach 0 5 2 13 20 Redfin 1 22 1 0 24

Total 779 1998 3741 4879 11397

79 Native fish recovery following Eastern Gambusia removal

30 (a)

20

10

0

CPUE 1.2 (b)

0.8

0.4

0 BT Fyke Seine SNE Per hour Per shot Figure 4.3.3 Eastern Gambusia catch per unit effort (CPUE) predictions for the four methods used in the study for (a) all trips, and (b) the first trip only. Analysis based on catch per hour for a single bait trap (BT) and fyke net, and fish per pass for targeted seine net and sweep net electrofishing (SNE). Number of observations = 1698.

80 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

18 (a) Carp gudgeon

12

6

0

2.5 (b) Australian smelt 2 1.5

CPUE 1 0.5

0

0.025 (c)Southern pygmy perch 0.02

0.015

0.01

0.005 0 BT Fyke Seine SNE

Per hour Per shot

Figure 4.3.4 Catch per unit effort (CPUE) predictions for the four methods utilised in the study across all trips for (a) Carp Gudgeon, (b) Australian Smelt, and (c) Southern Pygmy- perch. Analysis based on catch per hour for a single bait trap (BT) and fyke net, and fish per shot for seine net and sweep net electrofishing (SNE). Number of observations = 1698.

81 Native fish recovery following Eastern Gambusia removal

0.3 (a) Bait traps

0.2

0.1

0 Day Night Light No light Bait No bait

CPUE 4 (b) Fyke netting

3

2

1

0

Day Night Light No light

Figure 4.3.5 Catch per unit effort (CPUE + SE; fish per hour) predictions for Eastern Gambusia for each of the trapping variables applied for (a) bait traps, and (b) fyke netting for all trips (number of observations = 33). Variables included in the analysis are whether traps were set during the day or night, contained a light or not (for night sets only; fyke nets during first trip only), and were baited or unbaited (bait traps during first trip only). p > 0.05 for all comparisons.

82 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Bait traps Fyke nets (a) Carp gudgeon

0.15 2.5

2 0.1 1.5 0.05 1 0.5

0 0 (b) Pygmy perch 0.005 0.04

0.004 0.03 0.003 0.02 0.002 CPUE 0.01 0.001 0 0 (c) Australian smelt 0.15 0.006

0.004 0.1

0.002 0.05

0 0 Day Night Light No light Bait Unbaited Day Night

Figure 4.3.6 Bait trap (left column) and fyke net (right column) catch per unit effort (CPUE+ SE; fish per hour) predictions for (a) Carp Gudgeon, (b) Southern Pygmy-perch, and (c) Australian Smelt for each of the trapping variables for all trips. Variables included in the analysis are whether traps were set during the day or night, contained a light or not, and were baited or unbaited (bait traps only). p > 0.05 for all comparisons.

83 Native fish recovery following Eastern Gambusia removal

4.3.3 Eastern Gambusia population parameters and effectiveness of removal Season 1 In general, all sites during the first season contained very low numbers of Eastern Gambusia at the onset of the trial (late October – early November). During the first trip the raw numbers of Eastern Gambusia collected at each site ranged from 0–21 individuals; at six of the eight sites, fewer than six were captured (Table 4.3). Mark–recapture data collected from the two sites that were subject to the protocol during the first trip also suggested small Eastern Gambusia population sizes at the start of the trial (Table 4.4), estimating a maximum population size of 25 and 41 individuals respectively (sites 1 and 19; Table 4.4). Despite the low numbers of Eastern Gambusia present at the onset of the trial (and subsequent low numbers of fish removed), the initial removals substantially reduced population sizes. Although limited, mark–recapture data suggest that, despite minimal effort, the physical removals over five to seven days collected a large proportion of the population (e.g. the initial number of fish removed from site 1 represented between 84 and 100% removal; Table 4.4). The most encouraging result was that Eastern Gambusia were not recorded in follow-up monitoring at two of the four sites where they were removed (Table 4.3). The success of removal exercises did rely on consecutive visits to sites within this early trip, both to detect and ultimately capture fish (Figure 4.3.7). Each of these site visits were, however, relatively brief (generally less than one hour). Unfortunately it was evident that, despite a similar effort, repeated removals during the following trips (which were also after the onset of spawning) did not result in complete eradication of Eastern Gambusia from the remaining removal sites. Nevertheless, the rate of increase of Eastern Gambusia population growth was far less within removal sites when compared to control sites. The value for the effect of the removal treatment was negative, indicating that the removal treatment resulted in a lower growth rate of Eastern Gambusia at removal sites, although the 95% CL just overlapped 0, indicating a small amount of uncertainty around this estimate (Table 5; Figure 4.3.8). Hence, the growth rate of Eastern Gambusia at control sites was estimated to be 1.19 while the growth rate at removal treatment sites was estimated to be 0.39 (i.e. 1.189 – 0.796). Populations of Eastern Gambusia at the onset of the removal experiment comprised entirely adult fish, particularly females (Figures 4.3.9, 4.3.10). Furthermore, female fish collected during the initial removal period were all in gravid condition. Consequently, the sites still containing Eastern Gambusia during the second trip all had an influx of the first cohorts for the season (Figure 4.3.10). From this point onwards, Eastern Gambusia population sizes rapidly increased (Figure 4.3.8) despite the extremely low abundances of Eastern Gambusia at the onset of the experiment or a removal efficiency greater than 80%. For example, mark– recapture at control site 19 gave population estimates of Eastern Gambusia ranging from 38 to 41 in November, but this increased to 19 030 to 25 365 at the end of February. Similarly, at control site 2, where only two Eastern Gambusia were collected at the start of the experiment, had an estimated population size (derived from the mark–recapture data) of between 6520 and 8646 four months later. Further detailed assessment of population trajectories are presented below in the monitoring results.

Season 2 Eastern Gambusia were much more abundant at the sites at the onset of the experiment during the second season (Table 4.3). Unlike the first season, raw numbers of fish collected in removals and mark–recaptures suggested (with the exception of two sites) populations of hundreds of fish.

84 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

As in season 1, mark–recaptures at the onset of the trial still suggested that a large proportion of the population were being collected (Table 4.4) and that the initial removal exercises substantially reduced population sizes. This was confirmed when Eastern Gambusia were not recorded during several site visits (which included observations and seine netting) over three consecutive weeks following the initial removal. Again, the success of removals relied on consecutive visits to sites during this early trip, although each visit was brief. For example, eight days of removal exercises undertaken at Colbinabbin wetland resulted in a gradual reduction amounting to 75% of the population, based on mark–recapture information (Figure 4.3.11). Unfortunately, all sites were completely inundated shortly afterwards, and large numbers of adult and juvenile fish were consequently reintroduced to the sites. Because of the constant flooding, which conrinued until late February, no further removals or monitoring of Eastern Gambusia population change could be conducted after the initial removals. Nevertheless, these results provided important information about Eastern Gambusia removal and reinvasion.

5 2009/10 4

3

2

1 n/a n/a n/a n/a 0

160 2010/11

No. Fish removed Fish No. 120

80

40

0 1 2 3 4 5 6 7 8 9 10

Day Figure 4.3.7 Mean (± SE) number of Eastern Gambusia extracted from removal sites each consecutive day of trip one during the 2009–10 and 2010–11 seasons. n/a indicates no removals were undertaken.

85 Native fish recovery following Eastern Gambusia removal

Table 4.3 Raw numbers of Eastern Gambusia collected from each of the removal and control sites using physical removal techniques during the first trip (spring) of each of the two years. The detection of Eastern Gambusia during follow-up assessments is also presented.

Fish detected in follow-up Year Site Treatment Total fish collected assessment 2009– 10 1 Repeated 21 Yes 19 Control 15 Yes 2 Single 2 No 8 Repeated 0 No 9 Control 2 Yes 20 Repeated 5 Yes 21 Single 3 No 22 Control 2 Yes

2010– 11 Colbo Repeated 555 Yes* Willow 1 Repeated 195 No* Willow 2 Control 186 Yes* Hillgroves 1 Single 269 No* Hillgroves 2 Control 23 Yes* Browns 1 Repeated 3 No* Browns 2 Control 1 Yes* Cornella Ck Single 220 Yes*

Table 4.4 Eastern Gambusia mark-recapture parameters collected during the start of the experiment (spring) during each of the two years (two sites in each year) and end of the experiment in the first year only (end of summer). Population estimates are presented as a range allowing for a maximum mortality of marked female fish of 30% (see Appendix 1).

Year Period Site Treatment No. fish Total fish No. Population Recapture / removal marked collected recap’s estimate efficiency (%)

2009– Spring 1 R. removal 11 21 9 21 – 25 82 – 100 10

Spring 19 Control 13 15 5 36 – 41 41 – 47

Summer 9 Control 0 2 0 < 10 n.a.

Summer 19 Control 1000 1038 40 19030 – 25366 4 – 6

Summer 9 Control 60 3 6520 – 8646 7 – 9

2010– Spring Colbo R. removal 25 555 14 731 – 963 58 – 76 11

Spring Willow 1 R. removal 35 195 11 444 – 587 32 – 44

86 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Table 4.5 Parameter estimates from the Bayesian SSM for Eastern Gambusia (Gh), Carp Gudgeon (Hsp), Australian Smelt and carp (Cc). a – intrinsic rate of increase of; b – The effect of Eastern Gambusia (Gh) abundance; c – effect of the removal treatment on the rate of increase of Gh ; τ – observation error; σ – process error.

Species

Gh Hsp Rs Cc

Parameter x lwr upr x lwr upr x lwr upr x lwr upr a - - 1.189 0.633 1.755 0.250 -0.042 0.547 0.063 -0.080 0.206 0.008 0.262 0.257 b - - -0.016 -0.316 0.290 -0.075 -0.222 0.078 0.061 0.332 0.200 c -0.796 -1.664 0.034

τ 0.510 0.029 1.038 0.153 0.005 0.438 0.127 0.008 0.259 0.346 0.067 0.549

σ 0.777 0.073 1.421 0.470 0.191 0.707 0.184 0.014 0.346 0.246 0.007 0.652

Control sites Treatment sites

Abundance Abundance

0 50 100 150 200 0 50 100 150 200

1 2 3 4 5 1 2 3 4 5

Time Time

Figure 4.3.8 Predicted trajectories for Eastern Gambusia abundance at control (no removal) and treatment (removal) sites from November to March (trips 1–5) predicted for 1000 simulated sites from the SSM model (includes predictions for one additional time period). Dashed lines indicate the 95% credible interval. Predictions exclude observation error.

87 Native fish recovery following Eastern Gambusia removal

(a)

(b )

Figure 4.3.9 (a) Female Eastern Gambusia collected during the first trip of the experiment (before the onset of spawning). Note that all fish are in gravid condition. (b) Eastern Gambusia collected in a single targeted seine shot at the end of the experiment.

88 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

30 2009/10 2010/11

20 n = 22 n = 206 Trip 1 10

0 30

20 Trip 2 n = 197 n = 263 10 0 30

Frequency (%) Frequency 20 n = 44 Trip 3 n = 545 10

0 30

20 Trip 4 n = 290 n = 339 10

0 6 10 14 18 22 26 30 34 38 42 6 10 14 18 22 26 30 34 38 42 TL (mm) TL (mm)

Figure 4.3.10 Length frequency histograms (% frequency) of Eastern Gambusia collected on each trip (October–November, December, January and February–March) for both the 2009–10 and 2010–11 seasons of the trial.

200 100

Cum. %Cum. removed 160 80

120 60

80 40

No. No. fishremoved 40 20

0 0 1 2 3 4 5 6 7 8

Day

Figure 4.3.11 Total raw number of fish extracted (bars) and the cumulative percentage of population removed (line) based on mark–recapture data from Colbinabbin wetland over eight days of removal exercises before the onset of spawning (trip 1, season 2). Note day one involved the setting of traps only, and therefore no fish were extracted on this day.

89 Native fish recovery following Eastern Gambusia removal

4.3.4 Fish community monitoring and condition indices

4.3.4.1 Fish community monitoring A total of 44 482 fish were collected during the field removal experiment (14 621 and 29 600 in 2009–10 and 2010–11 respectively), comprising seven native and five alien species (Table 4.6, Figure 4.3.12). There was considerable variation in species composition and abundances across sites, although Carp Gudgeon were present at all sites and both Eastern Gambusia and Goldfish were present at 19 of the 20 sites (Table 4.7). In the 2009–10 season, Eastern Gambusia were the dominant species, followed by Carp Gudgeon, Common Carp and Australian Smelt respectively. Very few individuals of other species were collected, so they could not be used in the population modelling exercise. In the 2010–11 season, Eastern Gambusia and Carp Gudgeon were again the dominant species, followed by Flat-headed Gudgeon, Murray–Darling Rainbowfish and Australian Smelt. Three Flat-headed Galaxias were collected following the frequent flooding in the second season of the experiment — the first time this species had been recorded in the study (including the cross-sectional study). An assessment of length frequency data gave some indication of the degree of recruitment into the populations. Of the native species, only Carp Gudgeon appeared to in late summer, with the arrival of early cohorts (i.e. fish under 10 mm TL: Figure 4.3.13). Australian Smelt, Flat-headed Gudgeon, Murray–Darling Rainbowfish and Southern Pygmy- perch, apparently spawned only in the early months of the trial, or just prior to the start of the trial (although data is limited for the last two species). This pattern was also evident for the alien species: Common Carp and Goldfish spawned in the months prior to the trial, or in the first month of the trial (Figure 4.3.14). Recruitment patterns of Eastern Gambusia were discussed in section 4.3.3. Given the high variability in species composition and abundance between sites, an assessment of population trajectories of the most common species collected in the first year of the experiment was undertaken using the SSM approach. The intrinsic rate of increase was positive for Eastern Gambusia, Carp Gudgeon and Australian Smelt, and slightly negative for Common Carp (Table 4.5). Eastern Gambusia had by far the greatest rate of population increase (using control site data), being almost five times greater than the highest native species, Carp Gudgeon (Figure 4.3.16). The populations of Australian Smelt and Common Carp were relatively stable because the project commenced after the major spawning period (as indicated by the length frequency data). Observed population trajectories for each site are shown in Figure 4.3.14, and predicted trajectories from the SSM are shown in Figure 4.3.15. Predicted simulated mean trajectories (+95% CL) over five time periods are shown in Figures 4.3.8, 4.3.16 and 4.3.17. The results suggest that Eastern Gambusia had a negative influence on the population growth of native species and Common Carp, so that removing Eastern Gambusia would result in slightly increased abundances of other fish. The values for the interaction coefficient b were all negative, indicating that the presence of Eastern Gambusia had a negative impact on the per capita growth rate of all three species, with the largest estimated impact (–0.75) occurring for Australian Smelt (Table 4.5, Figure 4.3.17). The predicted effects of different abundances of Eastern Gambusia on the population growth of Carp Gudgeon, Australian Smelt and Common Carp indicates an increasing negative impact with increasing abundance of Eastern Gambusia (Figure 4.3.17). It must be noted, however, that the predicted negative effects of Eastern Gambusia on population growth was extremely low, even when Eastern Gambusia abundances were extremely high, and especially for Carp Gudgeon. Again, the 95% CL for these estimates all overlapped 0, indicating a degree of uncertainty around these effects.

90 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Table 4.6 Total numbers of each species collected from all sites during each year of the removal trial. Month

Species Nov Dec Jan Feb Total

2009–10

Carp Gudgeon 197 233 1078 1851 3359

Flat-headed Gudgeon 1 3 1 40 45

Australian Smelt 16 51 122 151 340

Flat-headed galaxias 0 0 0 0 0

Murray–Darling rainbowfish 0 0 0 4 4

Southern pygmy-perch 3 20 4 5 32

Unspecked hardyhead 0 0 0 0 0

Carp 505 1444 787 341 3077

Oriental weatherloach 0 5 2 13 20

Redfin 1 22 1 1 25

Eastern Gambusia 43 243 1739 5694 7719

Goldfish 4 0 35 40 79

Total 766 2021 3734 8100 14700

2010–11

Carp Gudgeon 1711 569 1042 526 3848

Flat-headed Gudgeon 161 52 74 74 361

Australian Smelt 65 29 22 35 151

Flat-headed galaxias 1 1 1 0 3

Murray–Darling rainbowfish 85 63 30 27 205

Southern pygmy-perch 5 1 1 0 7

Unspecked hardyhead 0 0 5 0 5

Carp 3 49 75 8 135

Oriental weatherloach 11 0 23 0 34

Redfin 4 7 3 1 15

Eastern Gambusia 2202 4287 5344 13003 24836

Goldfish 24 19 111 28 182

Total 4248 5058 6620 13674 29782

91 Native fish recovery following Eastern Gambusia removal

Figure 4.3.12 Examples of species collected at sites during the field removal experiment. Left, top to bottom: native species Murray–Darling Rainbowfish, Carp Gudgeon, Southern Pygmy-perch and Flat-headed Galaxias; and alien species. (Right, top to bottom: Common Carp, Eastern Gambusia, Goldfish and Oriental Weatherloach).

92 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Table 4.7 Total numbers of each species collected from each site for all trips and during both seasons of the experiment. Note that the sampling effort was not equal at each site because of the effects of drought and flood.

Site

Br’ns Br’ns Corn Hill. Hill. PB PB PB Will. Will. Species 1 2 8 9 19 20 21 22 Rd 1 Rd 2 Colbo . Ck 1 2 17b 1b 2b ST Rd 1 Rd 2

Carp Gudgeon 132 763 45 404 2273 5 3 3 5 9 622 152 444 7 212 162 87 294 326 1259

Flat-headed Gudgeon 1 8 1 0 5 0 0 0 13 8 271 76 0 0 12 9 0 2 0 0

Australian Smelt 0 133 12 41 4 9 1 3 8 0 2 128 0 0 16 82 44 7 1 0

Flat-headed Galaxias 0 0 0 0 0 0 0 0 0 0 2 1 0 0 0 0 0 0 0 0

Murr.–Darl. Rainbowfish 0 0 0 0 0 0 0 0 0 0 209 0 0 0 0 0 0 0 0 0

Southern Pygmy-perch 0 0 0 0 34 0 0 0 0 0 0 4 0 0 0 0 0 0 0 0

Unspecked Hardyhead 0 0 0 0 5 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Common Carp 2306 291 57 84 32 126 57 128 52 8 0 0 1 0 17 0 21 32 0 0

Oriental Weatherloach 10 12 0 0 21 0 0 0 0 0 0 0 0 0 0 0 0 11 0 0

Redfin 3 1 0 0 0 5 3 12 0 1 2 12 0 0 0 1 0 0 0 0

Eastern Gambusia 1709 847 0 938 3237 372 3 1146 3167 522 2004 557 5320 1479 270 172 307 1810 7241 1454

Goldfish 24 26 9 10 41 5 0 2 13 3 3 8 10 1 1 2 33 19 20 31

Total 4185 2081 124 1477 5652 522 67 1294 3258 551 3115 938 5775 1487 528 428 492 2175 7588 2744

93 Native fish recovery following Eastern Gambusia removal

November December January February

12 n = 272 n = 450 n = 185 n = 319 8 Carp gudgeon 4

0

35 30 n = 39 25 n = 37 n = 24 n = 71 Australian 20 smelt 15 10 5 0

80

60 n = 72 Flat-headed n = 23 n = 30 n = 84 gudgeon 40 20

0 % frequency %

100 n = 30 80 n = 7 n = 30 n = 63 Murray River 60 rainbowfish 40 20 0

80 n = 4 60 n = 7 n = 4 n = 4 Southern Pygmy perch 40 20

0 <520 30 40 >50 <5 20 30 40 >50 <5 20 30 40 >50 <520 30 40 >50

TL (mm)

Figure 4.3.13 Length frequency histograms (% frequency) for the five most common native species collected for each trip (October–November, December, January, February–March) of the trial (both seasons combined).

94 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

November December January February

80

60 n = 183 n = 84 n = 64 n = 24 Carp 40

20

0

80 n = 3 % frequency % 60 n = 10 n = 97 n = 26 Goldfish 40

20 0 <5 20 30 40 >50 <520 30 40 >50 <5 20 30 40 >50 <5 20 30 40 >50

TL (mm)

Figure 4.3.14 Length frequency histograms (% frequency) of Common Carp and Goldfish collected for each trip (October–November, December, January, February–March) of the trial (both seasons combined).

95 Native fish recovery following Eastern Gambusia removal

Gh Hsp

Abundance Abundance

0 50 100 150 200 0 10 20 30 40 50

1 2 3 4 1 2 3 4

Time Time

Rs Cc

Abundance Abundance

0 5 10 15 20 0 5 10 15 20

1 2 3 4 1 2 3 4

Time Time

Figure 4.3.14 Observed abundance of Eastern Gambusia (Gh), Carp Gudgeon (Hsp), Australian Smelt (Rs) and Common Carp (Cc) from November to February (trips 1–4). Lines connecting points indicate data from the same site. Red lines indicate Eastern Gambusia removal sites.

96 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Gh Hsp

Abundance Abundance

0 50 100 150 200 0 10 20 30 40 50

1 2 3 4 1 2 3 4

Time Time

Rs Cc

Abundance Abundance

0 10 20 30 40 50 0 5 10 15 20

1 2 3 4 1 2 3 4

Time Time

Figure 4.3.15 Trajectories of Eastern Gambusia (Gh), Carp Gudgeon (Hsp), Australian Smelt (Rs) and carp (Cc) abundance from November to February (trips 1–4) predicted by the SSM (dashed lines and small solid circles) overlaid on the observed trajectories (solid lines and open circles).

97 Native fish recovery following Eastern Gambusia removal

Gh Hsp

Abundance Abundance

0 50 100 150 200 0 5 10 15 20

1 2 3 4 5 1 2 3 4 5

Time Time

Rs Cc

Abundance Abundance

0 1 2 3 4 5 0 1 2 3 4 5

1 2 3 4 5 1 2 3 4 5

Time Time

Figure 4.3.16 Mean trajectories from 1000 simulated sites for Eastern Gambusia (Gh), Carp Gudgeon (Hsp), Australian Smelt (Rs) and Common Carp (Cc) abundance from November to March (trips 1–5) predicted by the SSM including (includes predictions for one additional time period). Predictions include process error but exclude observation error. Dashed lines indicate the upper and lower 95% credible interval.

98 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Hsp

Gh = 0 Gh = 7 Gh = 148

Abundance Abundance Abundance 0 5 10 15 20 0 5 10 15 20 0 5 10 15 20

1 2 3 4 5 1 2 3 4 5 1 2 3 4 5

Time Time Time

Rs

Gh = 0 Gh = 7 Gh = 148

Abundance Abundance Abundance 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5

Time Time Time

Cc Gh = 0 Gh = 7 Gh = 148

Abundance Abundance Abundance

0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5

Time Time Time

Figure 4.3.17 Predicted effects of different abundances of Eastern Gambusia on the population growth of Carp Gudgeon (Hsp), Australian Smelt (Rs) and Common Carp (Cc) based on 1000 simulated sites. Dashed lines are the upper and lower 95% credible intervals. Predictions exclude observation error.

99 Native fish recovery following Eastern Gambusia removal

Condition indices of juvenile native species A total of 860 juvenile native species were collected for the body and fin condition assessment — predominantly Carp Gudgeon, Australian Smelt and Flat-headed Gudgeon ( n = 580, 170 and 83 respectively). There was a general trend that increasing Eastern Gambusia abundance resulted in an increased likelihood of fin damage and decreased body condition of juvenile fish, although the degree of this influence varied between species. The assessment of all available data (trips 2–4) indicated that, for all three species, the relative condition of juvenile fish decreased with increasing Eastern Gambusia abundance (Figure 4.3.18). While there is some uncertainty in these relationships for Australian Smelt and Flat-headed Gudgeon (both p > 0.05), the predictions for Carp Gudgeon are significant ( p < 0.001). The probability of fin damage increased with increasing Eastern Gambusia abundance for Australian Smelt ( p < 0.05) and Flat-headed Gudgeon but was negligible for Carp Gudgeon, although there was some uncertainty in the last two relationships (all p > 0.05). The numbers of juveniles of rarer species collected, such as Murray–Darling Rainbowfish, Flat-headed Galaxias and Southern Pygmy-perch, were insufficient for a statistical analysis of body and fin condition. All of these species were collected from sites containing high abundances of Eastern Gambusia. Although adults of these rarer species did not have any fin damage, 50% of juvenile Murray–Darling Rainbowfish and all juvenile Southern Pygmy- perch collected in January and February had fin damage (Figure 4.3.19). On the other hand, none of the juvenile Southern Pygmy-perch collected in December displayed fin damage. A detailed examination of individual juvenile Southern Pygmy-perch collected in December, January and February suggested that there was very little growth in this cohort after December (Figure 4.3.20), but as very few individuals were collected this suggestion should be treated with caution.

100 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Australian smelt Carp gudgeon Flat-headed gudgeon n = 170 n = 580 n = 83

1.0

0.81.0

0.8 0.6

0.6 Value 0.4

FINSCORE 0.4 Prob.damage fin 0.2 0.2 P < 0.05 P = 0.667 P = 0.129 0.00.0 0.00.0 0.5 0.5 1.0 1.0 1.5 1.5 2.0 2.5 0.02.0 0.5 1.0 1.51.5 2.02.0 2.5 0.0 0.5 1.0 1.51.5 2.02.0 2.5 LOGCPUELOGCPUE LOGCPUE LOGCPUE

) 0.9 3.0

rel 6

0.8 2.5 5 0.7 2.0 4 0.6 3 1.5 0.5 2 1.0 (K condition Relative 0.4 P = 0.391 1 P < 0.001 0.5 P = 0.465

0.0 0.4 0.8 1.2 1.6 0.50 1.00 1.50 2.00 0.6 0.8 1.01.2 1.4 1.6 1.8

Log_Gambusia abundance Figure 4.3.18 General linear model outputs (± 95% confidence intervals) examining (top) the probability of fin damage, and (bottom) relative condition (Krel ) of juvenile Australian Smelt, Carp Gudgeon and Flat-headed Gudgeon in relation to Eastern Gambusia abundance (log 10 abundance) at all sites from December to February.

101

Figure 4.3.19 Juvenile Southern Pygmy-perch collected in January 2010. Note the damage of the caudal fin.

70 60

50 40

30

20 10

(mm) TL perch Pygmy 0

Oct/Nov Dec Jan Feb

Month Figure 4.3.19 Individual Southern Pygmy-perch lengths (mm) collected in each month (trips 1– 4) from Site 19 during the first year of the study. Fish < 30 mm represent the cohort of juvenile fish.

Arthur Rylah Institute for Environmental Research 102 Native fish recovery following Eastern Gambusia removal

4.4 Discussion 4.4.1 Physical control of Eastern Gambusia One of the most encouraging outcomes of the study was demonstrating that, under certain situations, physical removal can be effective in reducing Eastern Gambusia abundance. During both years of the study, physical removal conducted over a 5–10 day period (before the onset of the Eastern Gambusia spawning season), resulted in major reductions in Eastern Gambusia abundance (generally over 40%), and even complete eradication at several sites. Furthermore, the modelling exercises indicated the removal substantially reduced the rate of population increase. Physical removal techniques have rarely been considered as a management tool for Eastern Gambusia, the focus instead being on unselective draining and chemical poisoning (Maynard et al. 2008). While the effectiveness of these two techniques for removing alien species cannot be questioned, there will always be regions or sites where these methods are not feasible. For example, where threatened species are present or the hydrology is or unfavourable, draining or chemical treatment may have a greater impact on the site’s biota than the impact of Eastern Gambusia themselves. Three recent studies have documented the potential of physical removal to reduce gambusia populations in the wild (Maynard et al. 2008; Kerezsy 2009; Brookhouse and Coughran 2010). Maynard et al. (2008) and Brookhouse and Coughran (2010) demonstrated the potential utility of trapping methods that exploit specific behavioural responses of Eastern Gambusia to maximise the efficiency as a removal tool and, just as importantly, minimise their impact on non-target biota. For example, Maynard et al. (2008) established that Eastern Gambusia exhibit a positive phototactic response (attraction to light) and positive thermotactic response to heat (attracted to warm water). The authors then exploited this behaviour by incorporating heat or light sources (or both) into a variety of trap designs, to ultimately develop a trap and setting methodology that maximises the capture of Eastern Gambusia while having a minimal effect on native fish. Similarly, Brookhouse and Coughran (2010) established that setting bait traps in such a manner that their inception area is at the surface, as well as positioned in unshaded areas during the day, vastly increases catch rates of Eastern Gambusia and minimises catch rates of native species. The success of the Eastern Gambusia reductions in our study was a result of a combination of repeated site visits on each trip and the effectiveness of the seine netting method, which employed lightweight fine mesh nets to specifically target Eastern Gambusia. As in the examples mentioned above, seine netting exploits aspects of Eastern Gambusia behaviour, in particular their occurrence at or near the surface throughout spring and summer and their attraction to warmth and sunlight. For example, Maglio and Rosen (1969) found that the the major concentration of Gambusia affinis roughly corresponded to the areas of a pond receiving the most direct sunlight. This makes the areas they occupy within specific sites very predictable, particularly during times when water temperatures are low so that they are relatively easy to locate using polarised sunglasses. This behaviour also reduces any limitations that site depth had on the removal technique, as areas that were originally perceived being too deep to seine were generally the non-preferred areas for Eastern Gambusia (particularly during spring). For example, during spring the adult fish were constantly trying to access the shallowest, brightest area of each site (presumably to initiate reproduction). These areas would be targeted with seine netting several times a day for a 5–10 day period. Ultimately this method involved considerably less effort (each site visit was less than one hour) and created far less site disturbance than traditional random netting of an entire site. Understanding and exploiting the target species behaviour is not new in pest management activities, including those designed for fish. The management of carp in Australia is a prime example, with the development and optimisation of exclusion screens for wetlands (e.g. Hillyard et

103 Native fish recovery following Eastern Gambusia removal al. 2010), separation cages on fishways (e.g. Stuart et al. 2006), and general targeted removal and exclusion (e.g. Stuart and Jones 2002), all being based on thoroughly understanding carp behaviour and applying it to the region in question (e.g. Stuart and Jones 2006). Recently Kerezsy (2009) also investigated physical control of Eastern Gambusia using manual dip-netting in Australian artesian springs. He reported some success with the method, suggesting that sustained physical removal using dip nets in small springs may be an effective means of reduction or control. It must be noted that both our study and that of Kerezsy (2009) relied on active targeted removal techniques rather than trapping. Maynard et al. (2008) observed a reduction in CPUE over time, but our study and that of Brookhouse and Coughran (2010) did not assess the overall efficiency or effectiveness of trapping techniques in reducing population sizes, so the efficiency of trapping techniques alone is still in question. Although we included the trapping methods used in these earlier studies (using key findings of light and net position) in our the methodology assessment, we demonstrated that targeted seine netting was by far the most effective physical method for collecting Eastern Gambusia across all trips, and more importantly, for the first trip. While it is difficult to compare the efforts across different methods, the difference in catch rates between the targeted seine netting and other methods was so great that an equivalent catch of Eastern Gambusia using any of the other methods would require a huge increase in replication to achieve the removal outcomes of a single targeted seine netting. For example, during the trial the bait trap equivalent to the catch rate of a single targeted seine netting was approximately 171 hours immersion time (e.g. 10 traps immersed for 17 hours). This difference was even greater when comparing methods during the first trip (over 1000 times difference between CPUEs). Maynard et al. (2008) also compared CPUEs of dip netting with a variety of traps. They found that dip netting had the overall highest CPUE, although they stated that it was hard to compare CPUEs of trapping (trap hour) and active netting. Furthermore, damage to the aquatic environment, labour costs, and size and sex selectivity must also be considered when comparing methods. Nevertheless, this suggests that the use of currently available trapping techniques (e.g. bait traps) alone would require extremely high trap replication, or would have be undertaken for a much longer period of time before the onset of spawning if it were to achieve similar removal results to a control strategy which also employed active removal techniques such as seine netting. The active physical removal techniques employed in both this study and that of Kerezsy (2009) did have limitations. While depth did not seem to influence removal efficiency because Eastern Gambusia were not using deeper areas when they were targeted, the amount of structural habitat such as dense macrophytes and large woody debris that was present at a site did make these techniques difficult to use. The targeted techniques described in both studies overcome the problems of habitat to some degree, although high densities of this habitat does reduce the efficiency of the technique (particularly if eradication is the desired outcome). Although the targeted netting is far less damaging to habitat and non-target fauna than chemical applications or drying (or conventional seine netting), it did collect more native species and appeared to damage aquatic vegetation more than trapping methods. Therefore trapping is still a useful method in an Eastern Gambusia control program, particularly in sites that are extraordinarily fragile or contain dense structural complexity. The variables tested within the trapping component of the study were used primarily to streamline removal during the trial, but the information disseminated from results is applicable to future control programs that use trapping for removal. The results of the trapping assessment indicated that the number of Eastern Gambusia collected from daytime sets was higher than numbers collected during night sets, although this difference was not statistically significant ( p = 0.07).

104 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Furthermore, for traps set at night there was no significant difference in numbers of Eastern Gambusia collected in traps with and without a light, and with and without bait. This suggests that setting collapsible bait traps or fyke nets throughout the day without any additional light or bait attraction will collect as many Eastern Gambusia as setting these nets with lights or bait overnight. Because it is the most simplistic setting technique (both in time and investment) it is also the cheapest option for setting traps. Maynard et al. (2008) and Brookhouse and Coughran (2010) also assessed different trapping variables. Maynard et al. (2008) concluded that the use of collapsible bait traps set with the inception area at or near the surface was the best available option for Eastern Gambusia removal (which is why we used this method in the current study). While that study did not assess the catch rates of nets during the daytime, the authors concluded that having nets fitted with a portable light source improved catches. The reasons for the differences between that conclusion and our findings are unclear, but may be linked to the different climates where the studies were conducted (Tasmania vs the mid-MDB) or the different light sources used in the studies. Our study also recorded slightly higher numbers of Eastern Gambusia during the day, and the early trial using the floodlight (generating immense amounts of light and heat) attracted no Eastern Gambusia. We suggest that this is because Eastern Gambusia are largely inactive during the night, given that they feed during daylight hours and rely on sight to detect and attack prey (Swanson et al . 1996; McKay et al . 2001). Brookhouse and Coughran (2010) did not assess light attraction, but did conclude that nets set during daylight hours collected far greater numbers of fish if they were set in sunlight (as opposed to shade) and near the surface. The patterns in trapping variables of native species were similar to that of Eastern Gambusia, in that numbers of Carp Gudgeon, Australian Smelt and southern pygmy-perch collected were not influenced by whether traps were set during the day or night, contained a light or not and, were baited or unbaited. This information should be considered when designing survey regimes, potentially saving costs that would otherwise have been spent on setting traps throughout the night or investing in bait or light attraction sources. Furthermore, it supports the use of unbaited, daytime sets of collapsible bait traps used in the MDBA sustainable rivers audit (SRA) fish sampling program currently used throughout the MDB. Irrespective of methodology, the success of physical removal was strongly influenced by a number of other factors in the current study, most notably timing, hydrology (site connection) and size of the site. Although replication was limited, the results suggested that the success of removal undertaken before spawning had commenced largely governed the overall effectiveness of the control strategy, whether repeated removal was undertaken after spawning had commenced or not. This is because of the ability of Eastern Gambusia to rapidly colonise a site, as demonstrated by this study and others (e.g. Milton and Arthington 1983) coupled with the much more intensive effort required to physically remove post-larval and early juvenile Eastern Gambusia (e.g. Kerezsy 2009). In most cases this means that the window for physical removal of Eastern Gambusia is during winter and early spring, when daylight is less than 12 hours and water temperatures are below 16 °C (Pyke 2005). However, in many regions in Australia, including the northern regions of the MDB, this window for Eastern Gambusia removal is severely reduced, and in tropical areas even non-existent. To overcome this, Kerezsy (2009) suggested that removals be repeated over several sampling occasions to allow all size classes of fish to be targeted. This allows juvenile Eastern Gambusia to grow and then be subsequently targeted, thus overcoming the difficulties of removing early life-stages. This has implications for the design and, in particular, the methodology employed during control program. For example, if a strategy such as that proposed by Kereszy (2009) were employed, a window of just a few weeks exists between early juveniles and spawning

105 Native fish recovery following Eastern Gambusia removal adults. As a result, active removal techniques should be employed because a trapping program on its own is highly unlikely to achieve successful removal outcomes in this small window of time. Hydrology is another major factor governing the success, and more specifically, the duration of the success of Eastern Gambusia control exercises. The second season of the removal experiment provided a good example of how hydrology will largely govern the overall success of a control program. After what appeared to be very successful removal efforts at numerous sites, late season flooding and subsequent connection to adjoining habitats enabled both adult and early juvenile Eastern Gambusia to recolonise these sites approximately two weeks after the initial removal. Additionally, many sites in temperate regions, including those on the Ovens floodplain, generally reconnect at predictable periods (e.g. every winter–spring), which further restricts the ideal window for control actions. For this reason the removal exercises were conducted early enough to be before the onset of Eastern Gambusia removal, but late enough to minimise the risk of reconnection. Alternatively, measures could be taken to prevent reinvasion. For example, Kerezsy (2009) erected polypropelene barrier fencing across inflow points following Eastern Gambusia removal, to prevent reinvasion. However, the effectiveness of such measures is still uncertain. Current control strategies require an extensive knowledge of the distribution of Eastern Gambusia throughout a region and an equally detailed knowledge of the region’s hydrology, irrespective of methodology (e.g. Freeman 2007; Scurr 2007; Kerezsy 2009). Finally, the surface area of a site will also govern the effectiveness of control programs. Sites with a large surface area and greater numbers of preferred habitats that need to be repeatedly targeted require a far greater effort than small sites with a single target area. While eradication in the current study was achieved in sites up to 600 m2, Kerezsy (2009) found that physical removal could only be a successful strategy for complete eradication for sites of less than 3 m2. Of course this is a limitation for all control strategies, whether they be physical, chemical or drying of sites. Our study has highlighted the fact that that, unless removal exercises result in complete eradication, the suppression of Eastern Gambusia population size lasts little more than a single month. Lydeard and Belk (1993) considered two management strategies for Gambusia species: partial removal by seining to reduce population size, and complete removal by poisoning and subsequent reintroduction of native species. After conducting mesocosm experiments they found that the presence of Eastern Gambusia at both high and low densities had a negative effect on population growth of Least Killifish Heterandria formosa . The experiment involved an initial stocking of mesocosms with zero, low and high densities of Eastern Gambusia, together with the native species. Eastern Gambusia numbers in the low-density treatment increased to levels present in the high-density treatment in less than a month, finishing in higher abundances at the completion of the four-month trial. Therefore, if managers are aiming for longer-term benefits from Eastern Gambusia removal, they must either achieve complete eradication or undertake removal exercises at regular enough intervals so that the rate of removal is greater or equal to the rate of population growth. For Eastern Gambusia, the latter is extremely hard to achieve because of the remarkable ability of the species to recolonise a site. Typically this would involve repeated site visits to undertake control actions, and is thus one of the primary reasons why current management strategies focus on chemical and drying techniques, as they are the most likely techniques that can result in complete eradication of the species using fewer visits to a site. Even so, repeated treatments are still recommended in chemical control strategies (e.g. Elkington and Maley 2005). Willis and Ling (2000) noted the difficulties in using chemical treatment alone as a means of eradication, suggesting that a single treatment is unlikely to completely eradicate a small livebearing fish such as Western Gambusia. While our study and that of Kerezsy (2009) has demonstrated that eradication can be achieved by

106 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal physical removal, it is far less likely to achieve this result in much larger systems, or in warmer regions where the the ideal removal period is severely reduced or is non-existent. This also highlights the importance of repeated removal efforts, whatever method is used. This is not to say that physical removal should not be trialled in larger enclosed systems, particularly in those within temperate regions where the period of non-spawning activity is much longer and repeated site visits can be undertaken. Research should now be directed to new control techniques for Eastern Gambusia because of the current lack of successful and feasible control techniques. Trap designs have already been investigated in Tasmania (Maynard et al . 2008) and gene technology aimed at affecting the reproductive capabilities of Eastern Gambusia has been trialled (Fairfax et al . 2007). Rigorous investigation into possible biological controls needs to continue (or expand or commence) if Eastern Gambusia control or eradication is to be successful. It is an encouraging sign that scientists and managers are now considering alternative removal exercises beyond chemical applications and drying. 4.4.2 Eastern Gambusia as an invasive species: colonisation and population dynamics During the course of the trial, Eastern Gambusia displayed an astonishing capacity to rapidly colonise habitats. By combining mark–recapture data with a population growth modelling component, the trial revealed that just a few individuals of this alien species can colonise an area over the course of just several months. At the onset of both years of the experiment (October – November), Eastern Gambusia were found in very low abundances, often just a few individuals. Where these fish persisted (i.e. in control sites) populations rapidly established, reaching population sizes in their thousands in three to four months. This supports the general view that a single gravid female can colonise a site (Pyke 2005; Rowe et al. 2008). The state-space modeling approach used in this study demonstrated the alarming intrinsic rate of increase. This is not surprising given the species generally has several broods in a single breeding season, a gestation period usually around 22–25 days and maturation at around 4–6 weeks of age (Milton and Arthington 1983; Pyke 2005; Rowe et al . 2008). While not accounting for factors such as predation and resource availability, Maglio and Rosen (1969) calculated that 10 adult females could produce a population of 5 million individuals in six months, such is the reproductive potential of the species. Although not following this extreme trajectory, our study provides field- based evidence that astonishing population increases are still achieved even with resource limitations and predation. The modelling approach also indicated that the intrinsic rate of increase of Eastern Gambusia was far higher than even the most common native species in the region. This emphasises the species’ potential to out-compete native species, particularly if we also consider that Eastern Gambusia is an opportunistic or generalist omnivore (Arthington and Marshall 1999; McKay et al . 2001; Maynard et al . 2008) and has an extreme tolerance of poor water quality (Arthington, et al. 1983; Kennard et al . 2005; King and Warburton, 2007). Wide environmental tolerances, short generation times, rapid growth, broad diet, early sexual maturity and a high reproductive capacity have all been identified as being important attributes of successful invasive species (Koehn 2007). These attributes make Eastern Gambusia particularly suited to establishing populations after major disturbances of native fish populations, whether they are natural or human-induced. For example, ihe first year of our trial the sites had only recently been refilled following several years of extreme drought. Similar abundances of adult Eastern Gambusia and native species had recolonised these habitats, yet it was Eastern Gambusia that quickly asserted dominance. This pattern is also evident in many urban systems that have suffered extreme habitat alteration. For example, habitat alteration and subsequent water pollution have contributed to the decline of native fishes and

107 Native fish recovery following Eastern Gambusia removal subsequent establishment of Eastern Gambusia in urban Brisbane waterways (Arthington et al . 1983). 4.4.3 Fish community response to Eastern Gambusia removal The ultimate aim of the field removal experiment was to assess the fish community response (in particular native species) following the reduction in Eastern Gambusia, and subsequently, test the assumptions generated in phase 1 of the experiment. Whilst the native fish response component of the experiment was limited to a reduced number of sites and species (as a result of extremes in environmental factors, namely drought followed by flood), the results gave some indication of the responses we might expect to see by native small bodied fish communities in the Mid-Murray region following reductions in Eastern Gambusia abundances. These results also showed some alignment with the predictions of the cross-sectional component (phase 1 of the study). The assessment of overall fish community responses to Eastern Gambusia removal (population modeling) were limited to the first year of the experiment and could only focus on two common native species and one alien species. Nevertheless, the overall results suggest Eastern Gambusia had a slight negative impact on the short term intrinsic rate of population increase for Australian Smelt, Carp Gudgeon and Common Carp, with this negative impact increasing with increasing abundances of Eastern Gambusia. Additionally, the assessment of condition indices also suggested that increasing Eastern Gambusia abundance will increase the likelihood of fin damage for Australian Smelt and Flat-headed Gudgeon, and decrease the morphometric condition of all three native species. Whilst there is still some uncertainty surrounding these predictions, the combined results suggest that Eastern Gambusia removal resulted in an improvement in fish condition (both increased and decreased fin damage) which is likely to have resulted in increased population sizes for these species. These results provide field-based support for numerous predictions based on theoretical considerations and aquarium trials regarding the likely impacts of Eastern Gambusia in the natural environment. For example, Macdonald and Tonkin (2008) identified significant ecological niche overlaps (habitat and/or dietary) between Eastern Gambusia and many small- bodied native species within the MDB (including those investigated in this study). Additionally, aquarium experiments investigating behavioural interactions between Eastern Gambusia and several small Australian native fishes demonstrated that interference competition and aggression towards Pacific Blue-eye Pseudomugil signifier , Duboulay’s Rainbowfish , ornate rainbow fish Rhadinocentrus ornatus and Firetail Gudgeon Hypseleotris galii increased with relative densities of Eastern Gambusia (Knight 1999; Breen 2000; Conte 2001). All of these studies concluded that the establishment of Eastern Gambusia is likely to have contributed to reductions in natural populations of these species. Although our modelling predictions indicated that Eastern Gambusia removal would result in a positive response in condition and population size of the three co-occurring fish species, the degree of improvement, as indicated by changes in population trajectory and condition, is likely to be minimal, at least in the short term. Despite the uncertainty in the modelled predictions of population growth and condition indices, the predictions for all of the species indicated that the degree of severity of these negative effects was extremely low. Australian Smelt displayed the largest estimated negative impact on population growth and condition indices in response to increasing Eastern Gambusia abundance, most likely because of major trophic niche overlaps in all life stages (see Macdonald and Tonkin 2008). Carp Gudgeon displayed a significant reduction in morphometric condition with increasing Eastern Gambusia abundance, but only a slight negative change in population growth. These results support the modelled predictions of the cross-sectional analysis, in which Eastern Gambusia was predicted to have very little impact on the occupancy and population size of the common native species in floodplain wetlands (the greatest negative impact was predicted to be for Australian Smelt). As suggested in the chapter 3, it may be that the

108 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal minimal impact of Eastern Gambusia on these species is a major factor contributing towards them remaining largely widespread and abundant throughout the MDB. There is therefore uncertainty as to why the impact of Eastern Gambusia on these species is minimal, despite what appears to be major ecological niche overlaps. The Carp Gudgeon species complex displayed a significant decline in condition in relation to increasing gambusia abundance, but overall population change was shown to be minimal. The species is still widespread throughout the Basin (Lintermans 2007), despite major overlaps in dietary and habitat niches with Eastern Gambusia. Stoffels and Humphries (2003) reported all size classes of Carp Gudgeon and Eastern Gambusia showed significant partitioning of food resources, despite small Carp Gudgeons exhibiting high spatial overlap with Eastern Gambusia within surface habitats coupled with similar diets. This may suggest that while similar, the trophic niche of smaller Carp Gudgeon is broad enough to accommodate Eastern Gambusia. In addition to the Carp Gudgeon species complex, Flat-headed Gudgeon and Australian Smelt are all still relatively widespread throughout the MDB and all also occupy a broad range of habitats (King et al . 2007). Ling (2004) reported that many of New Zealand’s small native fish occupy relatively broad niches and that some degree of niche contraction caused by Eastern Gambusia may be accommodated without severe consequences. Similarly, Ayala et al . (2007) observed that the impacts of gambusia on Least Chub Iotichthys phlegethontis may also be mitigated by changes in the seasonal and daily habitat use of these two species. During the summer months, Least Chub exploit cooler, deeper habitats that are free of Western Gambusia, or moved to shallow, warmer habitats at night when Western Gambusia were less active. The availability of such habitats, even though suboptimal, may have acted as a buffer, reducing the magnitude of impacts and promoting the coexistence of the two species. The current study was conducted in largely intact floodplain wetlands, which provides frequent connection to adjoining habitats, as well as a variety of microhabitat niches (e.g. a variety of structural complexities in the form of aquatic vegetation and woody debris). The minor impact of Eastern Gambusia we observed on these common species may be a reflection of the complexity of microhabitats in natural systems, allowing these generalist native fishes to shift niches to those not utilised by Eastern Gambusia and thereby potentially reducing resource overlap and competitive interactions, and lessening the impacts of high Eastern Gambusia densities. A recent study of rivers in the Lake Eyre Basin suggested that the naturally variable hydrological regimes and native dominant fish assemblages of the area afford some resistance to the establishment and proliferation of alien fish (Costelloe et al. 2010). The negative impacts of Eastern Gambusia on native fish populations is likely to be far greater for fish populations occupying highly degraded sites. Highly degraded sites typically exhibit a reduction in connectivity and habitat uniformity resulting in limited opportunity for recolonisation or niche contraction. Morgan et al . (2004) found that lentic habitats that contained Eastern Gambusia, but provided cover such as aquatic vegetation and snags, contained more native fish than areas without Eastern Gambusia where cover was lacking. This suggests that the minor impacts of Eastern Gambusia on the common generalist native species that we have reported in this study are likely to be exacerbated in more degraded sites. Pyke (2008) suggested that reducing any negative impacts of Gambusia species on native species can be achieved not only by a reducing their abundance but by reducing the impacts per individual. From a management perspective, minimising the impact of Eastern Gambusia on common generalist species may be achieved by habitat restoration, such as the introduction of woody debris, flow restoration and the rehabilitation of riparian vegetation (e.g. Kennard et al. 2005). Further assessments of more isolated or degraded sites, involving longer-term monitoring of the fish community following Eastern Gambusia removal, is required to confirm this.

109 Native fish recovery following Eastern Gambusia removal

The patterns reported thus far in the study have focused on the minimal impact on the common generalist species of the MDB, suggesting that the ability of these species to shift or contract their trophic niche dampens the negative impacts of Eastern Gambusia. This also suggests that the negative impacts of Eastern Gambusia are likely to be far greater on species which do not display such a broad trophic niche. The low numbers of Murray–Darling rainbowfish, southern pygmy- perch and flat-headed galaxias collected during the experiment did not allow the population modeling to be undertaken, however, the limited information collected on fin condition for southern pygmy-perch and Murray–Darling rainbowfish suggested that these species were much more prone to fin damage than the generalist species. The extremely low numbers of these species collected during the study, coupled with the dominance of Eastern Gambusia across sites may in itself be an indicator of potentially greater negative impacts of Eastern Gambusia on these more specialist species. A large proportion of these more specialist wetland and confined waters species have undergone major reductions in range and abundance across the MDB. For example, Southern Pygmy-perch, Southern Purple-spotted Gudgeon and Olive Perchlet are now classified as threatened within all or certain areas of the MDB (Lintermans 2007). While habitat loss and degradation are likely to be the major cause of the declines in such species, numerous studies have suggested that the dominance of Eastern Gambusia within much of these species’ preferred habitat is likely to have accelerated the declines (e.g. Arthington 1983; Lloyd and Walker 1986; Lintermans 2007). For example, during surveys in the lower Murray River, Lloyd and Walker (1986) found that Southern Pygmy-perch was absent from sites of suitable habitat but with an abundance of Eastern Gambusia, suggesting that the interaction between these species could be responsible. Similarly, recent surveys of the Wimmera River basin during the Victorian Sustainable Rivers Audit showed that there were few or no Southern Pygmy-perch in areas where Eastern Gambusia were collected (SRA 2007), although this may be the result of factors such as habitat degradation. Collectively, the results of this and earlier studies suggest that, in the short term, management and control of Eastern Gambusia should focus on sites containing species with a narrow trophic niche, or on sites containing highly uniform habitats. Of course, this may also include sites which connect to these sites (thus potentially act as a source of Eastern Gambusia invasion), or where these species could be re-introduced. Further field based studies assessing the response of these rarer specialist species following Eastern Gambusia removal is required to ascertain these predictions. The results of the trial also indicated a slight negative impact on another co-existing alien species, Common Carp. This therefore suggests that removal of Eastern Gambusia may have unexpected benefits to other alien species. One question that is often not considered in pest species management programs is, ‘What negative effects will a reduction or complete removal of the target alien species have on ecosystem function?’ There is other evidence that successful eradications of alien species can have unexpected and undesired impacts on native species and ecosystems, particularly in areas which have accommodated the alien species for long periods of time and where they are an established species in the food chain (e.g. Murphy et al . 1998). Maezono and Miyashita (2004) suggested two ways in which such undesired impacts may occur. First, the removal of an alien species can enhance secondary establishment, or increase the impact of other alien species. Secondly, negative impacts to native biota may occur if the alien species performs functions similar to those of native species that are no longer in the system. An example of the first mechanism was reported by Maezono and Miyashita (2004), who investigated the removal of introduced Micropterus salmoides on native communities in farm ponds. While removal of this alien species did result in an increase in native fish and shrimp, there was also a substantial increase in alien crayfish, which resulted in a substantial reduction in macrophytes and associated rare odonate species. Zavaleta et al . (2001) suggest eradication of the alien prey species only can also cause problems by forcing the alien predator to switch to native

110 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal prey. Indeed, such alien predator and prey interactions may be relevant for areas containing both Eastern Gambusia and Redfin. McNeil (2004) suggested that Eastern Gambusia may be subject to selective predation by Redfin in billabong habitats. This alien species predator–prey interaction was hypothesised as responsible for the dominance of small native species in habitats containing Redfin, and the dominance of Eastern Gambusia in habitats without Redfin. This suggests that the removal of Redfin alone may result in an increase in Eastern Gambusia dominance, or that the removal of Eastern Gambusia alone may result in increased predation pressure on the existing native fish community. 4.4.4 Conclusions While the extremes in environmental variables have limited the analysis and conclusions that could be drawn from the field removal trial, the experiment still provided important information on Eastern Gambusia removal, population dynamics and native fish responses following such intervention, as follows. • Under certain conditions, physical removal can be used as a management tool to achieve major reductions in Eastern Gambusia populations, but the degree of success depends on a thorough consideration of various aspects of a sites hydrology, climate, habitat and size. • Eastern Gambusia has a remarkable capacity to rapidly recolonise habitats: just a few individuals rapidly established population sizes in their thousands in three to four months. The intrinsic rate of increase of Eastern Gambusia populations was far higher than even the most common native species in the region. This emphasises the species’ ability to establish populations rapidly and out-compete native species. • Most importantly, reductions in Eastern Gambusia abundance benefit small-bodied native fish populations. The negative impacts on the more common generalist species within these intact floodplain wetlands were shown to be relatively minimal. However, negative impacts might be far greater on species with narrow trophic niches. (A large proportion of such species have already suffered major reductions in range and abundance.) This suggests that, in the short term, management and control of Eastern Gambusia should focus on sites containing these species, or on sites containing highly uniform habitats. Further field studies to assess the response of these rarer specialist species, and longer- term monitoring of fish communities within more degraded sites following Eastern Gambusia removal, are required to test these predictions. • The removal of Eastern Gambusia may have unexpected benefits to other alien species, such as Common Carp, so site-specific ecosystem function in the absence of Eastern Gambusia must also be considered before undertaking a removal program.

111 Native fish recovery following Eastern Gambusia removal

5 Phase 3: Cost-effectiveness and logistics of Eastern Gambusia removal 5.1 Introduction The effective management of alien fish relies largely on prevention and containment, with the chances of success usually being directly proportional to the extent of dispersal (Raynor and Creese 2006; Gozlan et al. 2010). Consequently, for established, widespread alien fish species the options for management are restricted mainly to mitigation schemes that compensate for the presence of the species (Gozlan et al. 2010). Eastern Gambusia is now well established throughout the Murray–Darling Basin (MDBC 2008). If we also consider its ability to rapidly colonise an area (the results of phase 2 of this project being a prime example) and the absence of any effective large-scale control, it is clear that it is not feasible at present to undertake large-scale Eastern Gambusia control or eradication across the MDB. This does not, however, preclude management actions for this species. The NSW control plan for Common Carp reported that there is anecdotal evidence of localised efforts to reduce carp numbers helping to keep numbers under control, especially when repeated over a period of time (NSW 2010), despite this species being well established throughout NSW. Although direct ecological benefits were the ultimate aim of these efforts, the plan also notes these actions have important social benefits, particularly in community education, which in itself is an important management tool for pest fish, e.g. preventing new incursions through deliberate introductions, illegal use as live bait, and accidental translocation via gear (NSW 2010). Even at a more local scale there is very little information available on mitigating the impacts of Eastern Gambusia; the few documented cases of control focus predominantly on chemical techniques and drying of habitats (see McKay et al. 2001). While managers await the development of possible wide-ranging solutions such as large-scale harvesting techniques and daughterless technology, there is an urgent need for control options at a local scale, where total eradication using chemical treatments is undesirable because of the presence of threatened species or fragile ecosystems. Consequently, minimising the ecological impact of Eastern Gambusia at a local scale may be an important conservation strategy, particularly in areas containing threatened fish species, until new control measures are developed. The key objective of an alien species removal program is to reverse the negative impacts the species has had on environmental and socio-economic values. An assessment of these values in relation to invasive alien species is critical for sound environmental management and policy development, but such assessments are rare, particularly at an economic level. In Australia for example, economic assessments of freshwater fish invasions to our knowledge total one, for Common Carp (McLeod 2004; Rowe et al. 2008). In the case of Eastern Gambusia control, the benefits of controlling or eradicating populations do not have a quantifiable monetary value (primarily because ecological benefits accrue in systems for which the economic values are unknown), so they cannot be contrasted directly with the costs of control (e.g. research, sampling time and effort, removal). This makes justifying management and environmental policy for the species difficult because a true cost–benefit analysis cannot be undertaken. Nevertheless, the results of the study have provided important information on the physical removal of Eastern Gambusia that can be used to identify strategies that maximise the level of improvement to the native fish community for a fixed budget (budget maximisation), or minimise the cost of achieving a defined removal outcome (cost minimisation) (e.g. Choquenot et al. 2004; Koehn and MacKenzie 2004).

112 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

5.2 Cost-effectiveness and logistics of Eastern Gambusia removal using physical control programs Chapter 4 provided a thorough assessment of the methodology and ecological responses associated with the physical control of Eastern Gambusia, incorporating the results of the current project, other studies and theoretical information. We concluded that physical control programs can be used to substantially reduce Eastern Gambusia populations and ultimately provide benefits to native fish communities, and provide an alternative to conventional non-specific methods such as chemical treatments and drying of sites. The physical methodology used in this study involved a combination of targeted trapping and seine netting. Aside from the advantages of these methods in targeting Eastern Gambusia while having minimal impact on native fauna, these methods are also relatively inexpensive in terms of equipment costs (approximate AUD$90 and AUD$5 for a light- weight seine net and bait trap respectively) and are easy to deploy and maintain (e.g. Brookhouse and Coughran 2010). After brief training and supervision by an experienced zoologist (particularly in fish identification), both techniques could be used easily by community groups such as Landcare and schools. Although the these techniques are low-cost and require little training, the costs associated with the amount of effort, and their effectiveness in achieving the primary objective of substantially reducing Eastern Gambusia abundances to benefit native fauna, is highly variable between sites, and depends on a number of factors. As discussed in chapter 4, the factors influencing overall costs and the effectiveness of a physical control program at a specific site can be generalised to four simple variables relating to a site’s hydrological connectivity, ecological value, habitat complexity and size. Considering these factors in a methodical manner would enable managers to prioritise sites for the physical control of Eastern Gambusia in a way that maximises the ecological benefits per dollar invested, or that minimises the costs to achieve a defined ecological benefit (e.g. eradication from a certain number of sites). Subsequently, we have developed a simple decision support tool for managers considering investing in a physical control program (Figure 5.1). By considering a few basic questions regarding the primary factors we have discussed, the tool enables managers to rate individual sites (from ‘very low’ to ‘high’) according to the ecological benefits per dollar invested. This then would enable managers to decide whether to employ a physical control program and to prioritise the sites that the program should target.

113

Very Low All sites Permanently connected Permanently sites facilitating constant connected immigration of pest fish Does the site facilitate permanent immigration?

High frequency Low Isolated sites Low ecological value of connection Isolated sites of low ecological Is the site of high ecological value How frequently does the site connect to value and frequent connection (species / habitat)? adjoining waterbodies? to adjoining habitats facilitates frequent immigration of pest Low frequency fish of connection

High structural Medium High ecological High frequency of complexity Isolated sites of high ecological value but frequent connection value connection to adjoining habitats. Value of investment increases with How frequently does the site connect How much structural habitat reduced structural habitat to adjoining waterbodies? does the site contain? Low structural complexity complexity and surface area

Low frequency of High structural Large surface High area Isolated sites of high ecological connection complexity Benefits per $ invested $ per Benefits value and infrequent connection to adjoining How much structural habitat What is the size of the site? does the site contain? habitats. Value of investment Small surface increases with reduced area structural habitat complexity and surface area due to a reduction in required effort; increased ability to undertake Low structural Large surface active netting methods and area complexity increased negative interaction between pest and native What is the size of the site? species due to a reduction in habitat niche partitioning. Small surface area

Figure 5.1 Decision support tool for prioritising sites for physical control or Eastern Gambusia on the basis of maximising the ecological benefits per dollar invested. Note that this does not factor in the socio-economic benefits of control programs.

Arthur Rylah Institute for Environmental Research 114

Irrespective of methodology, a fundamental knowledge of a site’s hydrology is required before large amounts of time and money are invested in Eastern Gambusia removal (e.g. Kerezsy 2009) — in particular, is a site isolated, or is it permanently connected to adjoining habitats so that a continuing immigration or emigration of fish is possible? Clearly, if the latter is the case (for example, if the site is a creek that connects directly to a river), any ecological benefits of Eastern Gambusia removal will be short-lived (or non-existent) if fish simply recolonise the site immediately after (or during) the removal program (particularly given the re-population rate demonstrated for the species). For more isolated sites, consideration must also be given to how often they are isolated. Many sites reconnect to adjoining waterbodies every year (e.g. many wetlands on the lower Ovens River, Victoria), but others may only connect during a one-in-ten- year event. This implies that, for a similar investment, one site may get up to 10 years of ecological benefits compared to less than 12 months at another site. Prioritising sites based on their frequency of connection seems obvious, but the ecological value of each sites must also be considered. Managers must question what ecological benefits may arise from investing in an Eastern Gambusia removal program at a site which is not associated with high ecological values (e.g. contains threatened fish species or frequently connects to areas which do), compared to investing in those connecting to or containing threatened species or communities. For example, if one site supports only Carp Gudgeon (which is widespread and, based on this study, not drastically impacted by Eastern Gambusia abundance) and another site contains fish such as pygmy-perch or glassfish that are more likely to be more heavily impacted, then the ecological benefits achieved at the latter site will far outweigh those achieved at the former, irrespective of differences between frequency of connection. For example, the basis behind employing Eastern Gambusia control in a range of Australian artesian springs was the presence of two critically endangered fish species: Red-finned Blue-eye Scaturiginichthys vermeilipinnis and Edgbaston Goby Chlamydogobius squamigenus (Kerezsy 2009). Brookhouse and Coughran (2010) suggested that trapping Eastern Gambusia could have particular use in short-term applications, such as greatly reducing the abundance of fish in preparation for the spring and summer spawning periods of rare native fishes and amphibians. In the Murray–Darling Basin there are several small-bodied fish species occupying small isolated sites that are now severely reduced in range and abundance including pygmy-perch (e.g. Nannoperca australis ), galaxids ( Galaxias rostratus ), Murray Hardyhead ( Craterocephalus fluviatilis ), Southern Purple-spotted Gudgeon ( Morgurnda adspersa ) and Olive Perchlet ( Ambassis agassizii ). Managers need to be aware of such species, as well as other rare aquatic fauna such as some amphibian species (e.g. Komak and Crossland 2000), if planning Eastern Gambusia control. Chapter 4 highlighted that the structural habitat complexity of a site will largely govern the effectiveness of a removal strategy, in that sites containing large amounts of this habitat (such as snags), will severely reduce the effectiveness of the most efficient physical removal method (targeted active netting such as seine and dip-netting). This implies that the control program will require a far greater effort, because of the reduced efficiency of targeted active methods and an increase in the reliance on trapping techniques. Of course, sites containing greater amounts of structural habitat may also be in less need of Eastern Gambusia control because there is likely to be fewer aggressive interactions and more ecological niche separation with native species, although this depends on the native species occupying the site. So for a fixed investment at sites with similar ecological value and frequency of connection, the success of the removal program (in terms of the proportion of fish removed along with the benefits to native species) will decrease as the structural habitat complexity of a site increases. Depth does not influence removal efficiency (as Eastern Gambusia is unlikely to be using deeper water when they are targeted), but the surface area of a site does govern the effectiveness of

Arthur Rylah Institute for Environmental Research 115 Native fish recovery following Eastern Gambusia removal control programs. The larger a site’s surface area, the greater number of preferred habitats that need to be repeatedly targeted. Thus, larger sites require a far greater effort than small sites with a single target area. This is a limitation for all control strategies, whether they are physical, chemical or drying of sites. While the decision support tool considers the primary factors governing the effectiveness of a physical control program for Eastern Gambusia, there may be other factors specific to jurisdictions which enable sites to be prioritised further. These factors range from detailed biological information to legislation. For example, chapter 4 highlighted the importance of undertaking removal before the onset of Eastern Gambusia spawning. Because spawning times of Eastern Gambusia depend on the regional climate, the ideal window for removal in a particular area needs to be considered when developing a physical control program (e.g. regions with a short removal window may need a short, intensive program). Additionally, ecological knowledge of a region’s native biota should also be considered. For example, removal activities should ideally be undertaken outside the peak spawning periods of threatened native fish species present at a site. Although not relevant specifically to this project, legislation in regard to the use of gear and animal ethics may also affect how a control program is undertaken. For example, we recommend the use of lightweight seine nets as an active removal technique, but their use in inland waterways is restricted to nets less than 6 m long and restricted for use in nine inland waterways. Realistically, this makes the technique illegal for the general public to use. If managers intend to involve the community in a removal program they must therefore consider relevant fisheries and animal ethics legislation beforehand. Finally, this decision support tool relates to the logistics and cost-effectiveness of the physical control of Eastern Gambusia, not other methods. If sites are characterised as being likely to experience very low or low ecological benefits for a given investment, alternative mitigation activities should still be investigated. For example, a site may be of low priority for physical removal due to the high frequency of connection compared to other sites however, there may be potential value in habitat restoration activities such as resnagging and replanting riparian and/or aquatic vegetation (see chapter 4). Of course, these activities may also be used in conjunction with physical removal programs. The decision support tool and associated information on the physical removal of Eastern Gambusia will also been presented in a ‘guidelines for managers’ brochure that will be available to CMAs and Landcare groups (see appendix 2).

5.3 Social benefits of Eastern Gambusia removal: participation and education The major objectives of the current project focused on ecological issues with respect to Eastern Gambusia impacts and control, but the social implications of alien fish management cannot be ignored. While the difficulties of quantifying purely economic measures associated with Eastern Gambusia impacts (and therefore benefits of control actions) have been discussed, the social benefits of removal programs are even more difficult to quantify but are a vital component of any alien fish management program. Community education is a vital component of pest fish management for a number of reasons (Koehn 2007). As is the case with Eastern Gambusia, people are the principal means by which alien fish are spread (Wells 2007), yet there is a substantial lack of community education and awareness of the threats posed by alien fish (Koehn et al. 2007). During the course of this project, frequent discussions with landholders and, on one occasion, a demonstration and information

116 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal session presented to a local school group (Figure 5.2) revealed an alarming lack of awareness in regard to Eastern Gambusia. For example, Eastern Gambusia were being ‘rescued’ by members of the community from drying water bodies and re-released back into rivers and wetlands because they were thought to be small-bodied native fish (Fern Hames pers. comm.). If a control program can also raise community awareness of the issue then there is less potential for humans to continue to spread this alien species through such measures as deliberate introductions and illegal use as bait. The NSW control plan for carp has also reported that localised carp removal can have important social benefits, particularly at an educational level, which in itself is an important management too for controlling pest fish (NSW 2010). Finally, Koehn and MacKenzie (2004) point out that funding for on-ground natural resource management activities in Australia is largely directed through regional Natural Resource Management (NRM) bodies, which determine NRM priorities for their region (funding is allocated according to those priorities). These groups also coordinate and support other community-based activities such as Landcare and integrated catchment management. Therefore funding and support for on-ground alien species management can be accessed or enhanced if regional bodies are provided with sufficient information about the significance of the alien species threat and how the community can assist (Koehn and MacKenzie 2004). A key aspect of an alien fish management plan should therefore include a community awareness and education component. This project included a variety of ways of disseminating information on the Eastern Gambusia problem and the objectives of the project (see appendix 2). Communication activities and materials produced through the MDBA’s Native Fish Strategy have also made excellent contributions in this area, including workshop reports (e.g. Ansell and Jackson 2007), materials such as the Aliens in the Basin! brochure and Alien Fish in the Murray–Darling Basin book, events associated with Demonstration Reaches and Native Fish Awareness Week, and many activities delivered by the Native Fish Strategy Community Stakeholder Taskforce and local coordinators. Such activities and materials are clearly important and should continue. A range of other educational tools and resources have also been introduced which are beginning to build community education around alien fish. These include resources such as the on-line River Rescue scenarios developed by the University of Canberra, and the Sustaining River Life kit established by WaterWatch, RiverSmart and the MDBA. In the case of Eastern Gambusia, one area we wish to highlight is the potential for community involvement in control programs. As we have discussed, the methodology used for the physical removal of Eastern Gambusia could easily be used by community groups such as Landcare and schools. Aside from any direct ecological benefits that may arise from these activities, involving the community in removal activities will provide them with a sense of ownership of the pest fish issue and an improved understanding of the complexities of managing alien fish species (Koehn and MacKenzie 2004). As the Victorian Coordinator for the MDBA Native Fish Strategy and Community Engagement has pointed out at community field days, ‘informing individuals that removing just one female Eastern Gambusia will prevent tens of thousands of others in a few months time, is an extremely powerful message to convey’. So even if community involvement in removal activities does not directly result in immediate ecological benefits, these exercises will still result in substantial social benefits, particularly in light of the evidently insufficient education on the threats of Eastern Gambusia.

117 Native fish recovery following Eastern Gambusia removal

Figure 5.2 Demonstration and information on Eastern Gambusia and wetland fish communities presented to Colbinabbin primary school. (Photo: Glenace Avard)

5.4 Evaluating the benefits to native fish and cost-effectiveness of controlling other potentially harmful alien species across the MDB: applicability of the processes used in the current study

Because the control of numerous alien species (not just Eastern Gambusia) within the MDB is one of the key driving actions of the Native Fish Strategy, it was relevant to consider whether the processes undertaken in this study could be applied to evaluate the ecological benefits of controlling other alien fish across the MDB. In this study we took a holistic approach to investigating the feasibility of Eastern Gambusia control that would result in measurable improvements to native fish communities. We did this by integrating surveys and quantitative experimental work in natural billabong systems throughout the MDB and collating this information with cost-effectiveness approaches, ultimately enabling managers to use resources in a manner which maximises the potential ecological benefits of control strategies for a given investment (i.e. the development of the decision support tool). The cost-effectiveness assessment, and ultimately the decision support tool (section 5.3), was derived from an assessment of the factors influencing the removal efficiency of Eastern Gambusia — in turn derived from this study and results of research into control strategies, e.g. Maynard et al. (2008) — and most importantly, the likely response of native fish following these reductions. While the decision support tool is specifically for the physical removal of Eastern Gambusia

118 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal removal, the processes it was derived from have major relevance for controlling other established alien species across the basin, particularly those which also have limited socio-economic value. These processes have been generalised to form a template for the required approach to maximise the benefits to native fish arising from an alien fish removal program (Figure 5.3). The basic premise of this template is the development of a thorough understanding of the ecological impacts of the alien species (and therefore the response following control measures) as well as the factors influencing the removal activities themselves. This in turn will enable maximum benefits to be achieved for native fish. Understanding the ecological impacts of an alien species is an essential component of best practice vertebrate pest management, which is based on the concept of managing impacts rather than numbers (Braysher 1993; Koehn and MacKenzie 2004). This together with the use of pest management principles (Braysher 1993; Bomford and Tilzey 1997) and adoption of integrated approaches will give the best results by ultimately providing a basis for benefit:cost analyses of management options (Choquenot et al. 2004; Koehn and MacKenzie 2004). Unfortunately, our understanding of the impacts of most alien species is still rudimentary, even for direct impacts such as predation, let alone impacts at the community or ecosystem levels (Townsend 2003). As a result, a thorough understanding of ecological impacts and therefore sound predictions of ecological responses to removal is rarely utilised in the management of alien fish in Australia. For example, despite millions of dollars being spent annually on carp control throughout the MDB, the effects of carp on native biota remains largely speculative (NSW 2010) and is limited to impacts on water quality (King et al. 1997; Koehn 2007). As a result, current cost- effectiveness assessments and subsequent control measures for the species generally focus on maximising the number of fish removed, as opposed to minimising the negative impacts of the species on native biota. While the latter is generally assumed to be a product of the first (and this is often the case), the ultimate goal of management for alien species is not a reduction in numbers per se but a reduction in the impacts caused by each species (Lodge and Shrader-Frechette 2003; Koehn 2007; Zavaletta 2010). Indeed, the current study has highlighted that, for established alien fish species, an understanding of the response of native species to a control program under different scenarios will enable the use of resources to maximise these ecological benefits, assuming that ecological benefits are the primary objective of an alien fish removal program. There is still the question of whether there is a need to definitively determine the specific impacts of an alien species, or whether managers should simply apply the precautionary principle (based on a reasonable suspicion of impacts) and commence management actions to control the threat (Lintermans et al. 2007). The management of new incursions of an alien pest species is a case in point: the high priority of preventing the spread and establishment of new populations (Koehn and MacKenzie 2004) tends to override the lengthy processes involved in developing the predictive capabilities (theoretical, proof-of-concept and cost-effectiveness stages in Figure 5.3), relying on a rapid response to maximise the number of fish removed (e.g. Ayres and Clunie 2010). This is a sensible approach for new incursions of alien fish species when they are generally still confined in their distribution, but (as we have highlighted) undertaking large-scale control or eradication of established alien species across the entire MDB is not feasible at present. For these species, options for management are largely restricted to localised mitigation schemes that compensate for the presence of the species (Gozlan et al. 2010). Therefore, for established species in the MDB such as Common Carp, Eastern Gambusia, Redfin and Goldfish, management should be based on an integrated and strategic approach that addresses the range of factors that threaten the health of native fish populations and the general health of the MDB (Braysher 2007). Collectively, this indicates a through understanding of the ecological impacts (and therefore sound predictions of the responses following control measures) of specific alien species is essential to maximise the

119 Native fish recovery following Eastern Gambusia removal ecological benefits (such as improvements in native fish fauna) of investment in the control of established alien species within the MDB. Again, this discussion and the template developed for prioritising control actions for established alien fish is based purely on achieving ecological benefits. It has not considered socio-economic factors, which are much more relevant for angling species such as Redfin and trouts. While this assessment is outside the scope of the study, we suggest the simplistic process presented in the study aimed purely at assessing the ecological benefits of alien fish control has relevance for other established alien fish species with limited socio-economic value, such as Goldfish and Oriental Weatherloach (for which the biological impacts on native biota is also largely unknown). Thus, we suggest that a similar approach to that undertaken in the present study would prove valuable.

Thorough understanding of the life-history and behavioural traits of the alien species and co-occurring native fish species

Identify the species / communities which Development of methodologies to are most impacted or at risk maximise the efficiency of removal

Undertake a field based proof of concept approach testing predictions

Predictive capabilities on ecological Identify factors influencing removal response measures

Cost-effectiveness assessment and development of site prioritisation tool

Maximise the ecological benefits of alien fish control program

Figure 5.3 Template for the processes used in the current study to maximise the benefits to native fish arising from an alien fish removal program.

120 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

6 Conclusion This project has taken a forward step into researching the feasibility of controlling Eastern Gambusia populations to achieve measurable improvements to native fish communities. By integrating surveys and quantitative experimental work in natural billabong systems throughout the MDB, the project provided important information on Eastern Gambusia removal, population dynamics, and the responses of native fish following such intervention. Understanding the ecological impacts of an alien species is an essential component of pest management, based on the concept of managing impacts rather than simply reducing numbers (Braysher 1993; Koehn and MacKenzie 2004). The negative impacts of Eastern Gambusia on the more common generalist species in these intact floodplain wetlands were minimal, but negative impacts might be far greater on species that have narrow trophic niches. There was also some indication that Eastern Gambusia may have unexpected impacts on other alien fish species. While successful eradication of an established alien species such as Eastern Gambusia using current methodologies is not likely to be feasible in larger open systems, minimising the impact of Eastern Gambusia on native fish may be an important management strategy until new control strategies such as harvesting techniques and daughterless technology are developed. Despite Eastern Gambusia displaying an astonishing capacity to rapidly colonise habitats (with an intrinsic rate of increase far higher than even the most common native species in the region), we found that physical removal of Eastern Gambusia, when conducted under certain conditions, can achieve major reductions in Eastern Gambusia populations. 6.1 Management / research recommendations • Although an assessment of control methods was not a major objective of this study, we found that physical removal of Eastern Gambusia, when conducted under certain conditions, can be used as an effective management tool to achieve major reductions in Eastern Gambusia populations. However, the degree of success requires thorough consideration of various aspects of a sites hydrology, climate, habitat and size. • Reductions of Eastern Gambusia will result in improvements to native fish populations, but management and control actions should focus on sites containing species that have narrow trophic niches, or on sites containing highly uniform habitats, to maximise the ecological benefits of Eastern Gambusia reductions. A decision support tool has been developed for managers to assess the feasibility and prioritise sites for Eastern Gambusia control actions. • Further field studies to assess the response of rarer specialist native species and conduct longer-term monitoring of fish communities across different habitat conditions following Eastern Gambusia removal are required to refine these predictions. • Given the unexpected benefits to other alien fish species, ecosystem function in the absence of Eastern Gambusia must also be considered before implementing a removal program. • We found an alarming lack of community awareness about Eastern Gambusia. We strongly recommend that community awareness and education exercises be included in any Eastern Gambusia management program. Involving the community in removal activities is a possibility because of the methods used, which could easily be utilised by community groups such as Landcare and schools. Even if these activities do not directly result in immediate ecological benefits, they will result in substantial social benefits, particularly in light of the lack of education on the threats from this alien fish species.

121 Native fish recovery following Eastern Gambusia removal

• We suggest that the simple process used in the study could also prove valuable if applied to other established alien fish species, particularly those with limited socio-economic value.

122 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

7 References Allen, G. R. (1996a) Family Chandidae: Glassfishes, chanda perch. In Freshwater Fishes of South- Eastern Australia. (McDowall, R. M., Ed.), pp. 146-149. Reed Books, Sydney. Allen, G. R. (1996b) Family Melanotaeniidae: Rainbowfishes. In Freshwater Fishes of South- eastern Australia. (McDowall, R. M., Ed.), pp. 134-140. Reed Books, Sydney. Amundsen, P-A., Staldvik, F. J., Reshetnikov, Y. S., Kashulin, N., Lukin, A., Bøhn, T., Sandlund, O. T. and Popova, O. A. (1999) Invasion of vendace Coregonus albula in a subarctic watercourse. Biological Conservation 88, 405-413. Angeler, D. G., Sanchez-Carrillo, S., Rodrigo, M. A., Alvarez-Cobelas, M. and Rojo, C. (2007) Does seston size structure reflect fish-mediated effects on water quality in a degraded semiarid wetland? Environmental Monitoring and Assessment 125, 9-17. Ansell, D. and Jackson, P. (Eds). (2007) Emerging Issues in Alien Fish Management in the Murray–Darling Basin: Statement, recommendations and supporting papers . Proceedings of a workshop held in Brisbane QLD, 30-31 May 2006. Murray–Darling Basin Commission, Canberra. Arthington, A. H. and Lloyd L. N. (1989) Introduced Poeciliids in Australia and New Zealand. In: Ecology and evolution of livebearing fishes (Poeciliidae). (Meffe, G. K. and Snelson, F. F. Jnr., Eds.), pp. 333-348, Prentice-Hall, Inc., New Jersey. Arthington, A. H. and Marshall, C. J. (1999) Diet of the exotic , Gambusia holbrooki , in an Australian lake and potential for competition with indigenous fish species. Asian Fisheries Science 12, 1-16. Arthington, A. H. and McKenzie, F. (1997) Review of impacts of displaced/introduced fauna associated with inland waters, Australia: State of the Environment Technical Paper Series (Inland Waters), Department of the Environment, Canberra. Arthington, A. H., Hamlet, S. and Bluhdorn, D. R. (1990) The role of habitat disturbance in the establishment of introduced warm-water fishes in Australia. In: Introduced and translocated fishes and their ecological effect. (Pollard, D. A., Ed.), pp. 61-66. Australian Government Publishing Service, Canberra. Arthington, A. H., Milton, D. A. and McKay, R. J. (1983) Effects of urban development and habitat alteration on the distribution and abundance of native and exotic freshwater fish in the Brisbane region, Queensland. Australian Journal of Ecology 8, 87-101. Arumugam, P. T. and Geddes, M. C. (1987) Feeding and growth of golden perch larvae and fry (Macquaria ambigua Richardson). Transactions of the Royal Society of South Australia 111, 59- 65. Ayala, J. R., Rader, R. B., Belk, M. C. and Schaalje, G. B. (2007) Ground-truthing the impact of invasive species: spatiotemporal overlap between native least chub and introduced western mosquitofish. Biological Invasions 9, 857-869. Ayres, R. and Clunie, P. (2010) Management of freshwater fish incursions: a review. Arthur Rylah Institute for Environmental Research. Department of Sustainability and Environment, Heidelberg, Victoria Backhouse, G.N., and Frusher, D.J. (1980) The crimson-spotted rainbowfish, Melanotaenia fluviatilis (Castelnau 1878). Victorian Naturalist 97(July/Aug): 144-148.

123 Native fish recovery following Eastern Gambusia removal

Balcome, S. and Humphries, P. (2006) Diet of the western Carp Gudgeon ( Hypseleotris klunzingeri Ogilby) in an Australian floodplain lake: the role of water level stability. Journal of Fish Biology 68, 1484-1493. Balcome, S., Bunn, S. E., Mckenzie-Smith, F. J. and Davies, P. M. (2005) Variability of fish diets between dry and flood periods in an arid zone floodplain river. Journal of Fish Biology 67, 1552- 1567. Barlow, C. G. and Bock, K. (1981) Fish in farm dams: final report to the Reserve Bank’s Rural Credits Development Fund. NSW Fisheries Research Institute, Cronulla, Australia. Beesley, L.S., and Prince, J. (2010) Fish community structure in an intermittent river: the importance of environmental stability, landscape factors and within-pool habitat descriptors. Marine and Freshwater Research 61, 605-614. Beatty, S.J. 2006. The diet and trophic positions of translocated, sympatric populations of Cherax destructor and Cherax cainii in the Hut River, Western Australia: evidence of resource overlap. Marine and Freshwater Research 57, 825-835. Bomford, M. and Hart, Q. (2002) Non-indigenous vertebrates in Australia. In: Biological Invasions - Economic and Environmental Costs of Alien Plant, Animal and Microbe Species. (Pimental, D. Ed.), CRC Press, New York. Boulton, A. J. and Brock, M. A. (1999) Australian freshwater wetlands: processes and management. Gleneagles Publishing, Glen Osmond. Bomford, M., and Tilzey, R. (1997) Pest management principles for European carp. Controlling carp: exploring the options for Australia, Albury, pp. 9-22. Braysher, M. (1993) Managing vertebrate pests: principles and strategies. Bureau of Resource Sciences, Australian Government Publishing Service, Canberra. Braysher, M. (2007) Management of pest fish. Emerging Issues in Alien Fish Management in the Murray–Darling Basin: Statement, recommendation and supporting papers. Proceedings of a workshop held in Brisbane QLD, 30-31 May 2006, Brisbane, pp. 33-39. Breen, A. (2000) Density dependent interference competition between the exotic poeciliid Gambusia holbrooki (Girard, 1859) and the Australian native melanotaeniid Rhadinocentrus ornatus (Regan, 1914). Unpublished Honours thesis, Southern Cross University, Lismore. Brookhouse, N., and Coughran, J. (2010) Exploring the potential for an ecology-specific, physical control method of the exotic pest Mosquitofish, Gambusia holbrooki. Ecological Management and Restoration 11, 226-228. Brown, P., Sivakumaran, K.P., Stoessel, D., and Giles, A. (2005) Population biology of carp (Cyprinus carpio L.) in the mid-Murray River and Barmah Forest wetlands, Australia. Marine and Freshwater Research 56, 1151-1164. Cadwallader, P. L. and Backhouse, G. N. (1983) A guide to the freshwater fish of Victoria. Ministry for Conservation, Melbourne. Cadwallader, P. L. and Douglas, J. (1986) Changing food habits of Macquarie Perch, Macquaria australasica Cuvier (Pisces: Percichthyidae), during the initial filling phase of Lake Dartmouth, Victoria. Australian Journal of Marine and Freshwater Research 37, 647-657. Caiola, N. and de Sostoa, A. (2005) Possible reasons for the decline of two native toothcarps in the Iberian Peninsula: evidence of competition with the introduced . Journal of Applied Ichthyology 21, 358-363.

124 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Capinha, C. and Anastácio, P. (2011) Assessing the environmental requirements of invaders using ensembles of distribution models. Diversity and Distributions 17, 13-224. Chapman, P. and Warburton, K. (2006) Postflood movements and population connectivity in gambusia. Ecology of Freshwater Fish 15, 357-365. Choquenot, D., Nicol, S. J. and Koehn, J. D. (2004) Bioeconomic modelling in the development of invasive fish policy. New Zealand Journal of Marine and Freshwater Research 38, 419-428. Clavero, M. and García-Berthou, E. (2005) Invasive species are a leading cause of animal extinctions. Trends in Ecology and Evolution 20, 110. Clunie, P. and Koehn, J. (2001) Freshwater Catfish: A Resource Document. Final report to Murray–Darling Basin Commission, Canberra. Prepared by the Arthur Rylah Institute for Environmental Research, Department of Sustainability and Environment, Heidelberg. Conte, S. (2001) An investigation of density-dependant interference competition between the exotic poeciliid Gambusia holbrooki (Girard, 1859) and the Australian native fish Hypseleotris galii (Ogilby, 1903). Unpublished Honours thesis, Southern Cross University, Lismore. Copp, G. H., Bianco, P. G., Bogutskaya, N. G., Eros, T., Falka, I., Ferreira, M. T., Fox, M. G., Freyhof, J., Gozlan, R. E., Grabowska, J., Kovac, V., Moreno-Amich, R., Naseka, A. M., Penaz, M., Povz, M., Przybylski, M., Robillard, M., Russell, I. C., Stakenas, S., Sumer, S., Vila-Gispert, A. and Wiesner, C. (2005) To be, or not to be, a non-native freshwater fish? Journal of Applied Ichthyology 21, 242-262. Costelloe, J.F., Reid, J.R.W., Pritchard, J.C., Puckridge, J.T., Bailey, V.E., and Hudson, P.J. (2010) Are alien fish disadvantaged by extremely variable flow regimes in arid-zone rivers? Marine and Freshwater Research 61, 857-863. Courtenay, W. R. J. and Meffe, G. K. (1989) Small fishes in strange places: a review of introduced poeciliids. In: Ecology and evolution of livebearing fishes (Poeciliidae). (Meffe, G. K. and Snelson, F. F. Jnr., Eds.), pp. 319-332. Prentice-Hall, Inc., New Jersey. Crivelli, A.J. (1981) The biology of the Common Carp, Cyprinus carpio L. in the Camargue, southern France. Journal of Fish Biology 18, 271-290. Crook, D.A., Reich, P., Bond, N.R., McMaster, D., Koehn, J.D. and Lake, P.S. (2010) Using biological information to support proactive strategies for managing freshwater fish during drought. Marine and Freshwater Research 61, 379-387. Dennis, B., Ponciano, J.M., Lele, S.R., Taper, M.L., Staples, D.F. (2006) Estimating density dependence, process noise, and observation error. Ecological Monographs, 76 (3), pp. 323-341.

De Silva, S. S., Nguyen, T. T. T., Abery, N. W. and Amarasinghe, U. S. (2006) An evaluation of the role and impacts of alien finfish in Asian inland aquaculture. Aquaculture Research 37, 1-17. Elith, J., Leathwick, J.R. and Hastie, T. (2008) A working guide to boosted regression trees. Journal of Animal Ecology 77, 802-813. Elith, J. and Leathwick, J.R. (2009) Species distribution models: ecological explanation and prediction across space and time. Annual Review of Ecology, Evolution, and Systematics 40, 677- 697. Elkington, S. (2004) Eradication of Gambusia (Gambusia affinis) and koi carp (Cyprinus carpio) from the Nelson/Marlborough Conservancy. Department of Conservation Te Papa Atewhai, Nelson.

125 Native fish recovery following Eastern Gambusia removal

Elkington, S. and Maley, R. (2005) Eradication of Gambusia ( Gambusia affinis ), koi carp (Cyprinus carpio ), and rudd ( Scardinius erythrophthalmus ) from the Nelson/Marlborough Conservancy. Department of Conservation Te Papa Atewhai, Nelson. Fairfax, A., Fensham, R., Wager, R., Brooks, S., Webb, A. and Unmack, P. (2007) Recovery of the red-finned blue-eye: an endangered fish from springs of the Great Artesian Basin. Wildlife Research 34, 156-166. Faragher, R. A. and Lintermans, M. (1997) Alien fish species from the New South Wales River Survey. In: Fish and Rivers in Stress: The NSW Rivers Survey. (Harris, J. H. and Gehrke, P. C. Eds.), pp. 201-223. NSW Fisheries Office of Conservation and the CRC for Freshwater Ecology. Cronulla. Fausch, K. D. (2007) Introduction, establishment and effects of non-native salmonids: considering the risk of invasion in the United Kingdom. Journal of Fish Biology 71, 1-32. Fey, D.P., and Hare, J.A. (2005) Length correction for larval and early-juvenile Atlantic (Brevoortia tyrannus) after preservation in alcohol. Fishery Bulletin 103, 725-727. Fletcher, A. R. (1986) Effects of introduced fish in Australia. In Limnology in Australia (De Deckker, P. and Williams, W. D., eds.). Melbourne and the Netherlands: CSIRO/Dr W. Junk Publishers. Freeman, R. (2007) Tamar Estuary, Gambusia Eradication. Internal Report, Inland Fisheries Service, Hobart. Froese, R. (2006) Cube law, condition factor and weight-length relationships: history, meta- analysis and recommendations. Journal of applied ichthyology 22, 241-253. Froese, R. and Pauly, D. Eds. (2007) Fishbase Version 07/2008. http://www.fishbase.org/search.php Galat, D. L. and Robertson, B. (1992) Response of endangered Poeciliopsis occidentalis sonoriensis in the Rio Yaqui drainage, Arizona, to introduced Gambusia affinis . Environmental Biology of Fishes 33, 249-264. Gall, G. A. (1987) Inbreeding. In Population Genetics and Fisheries Management (Ryman N. and Utter, F. Eds.), pp. 47-87. University of Washington Press, Seattle. García-Berthou, E., C. Alcaraz, Q. Pou-Rovira, L. Zamora, G. Coenders, and C. Feo (2005) Introduction pathways and establishment rates of invasive aquatic species in Europe. Canadian Journal of Fisheries and Aquatic Sciences 62, 453-463. Geddes, M. C. and Puckridge, J. T. (1989) Survival and growth of larval and juvenile native fish. The importance of the floodplain. In: Proceedings of the Native Fish Management Workshop (Lawrence, B. Ed.), pp. 101-115. Murray Darling Basin Commission, Canberra. Gee, J.H. and Gee, P.A. (1991) Reactions of gobioid fishes to hypoxia: buoyancy control and aquatic surface respiration. Copeia 1, 17-28. Gehrke, P. C. (1992) Diel abundance, migration and feeding of fish larvae (Eleotridae) in a floodplain billabong. Journal of Fish Biology 40, 695-707. Gherardi, F. (2007) Biological invasions in inland waters: an overview. In: Biological invaders in inland waters: profiles, distribution and threats (Gherardi, F. Ed.), pp. 3-25. Springer, Dordrecht, The Netherlands.

126 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Gillespie, G. and Hero, J. M. (1999) Potential impact of introduced fish and fish translocations on Australian amphibians. In: Declines and Disappearances of Australian Frogs. (Campbell, A. Ed.), pp. 131-144. Environment Australia, Canberra. Gozlan, R. E. (2008) Introduction of non-native freshwater fish: is it all bad? Fish and Fisheries 9, 106-115. Gozlan, R. E., St-Hilaire, S., Feist, S. W., Martin, P. and Kent, M. L. (2005) Disease threat to European fish. Nature 435, 1046. Gozlan, R.E., Britton, J.R., Cowx, I., and Copp, G.H. (2010) Current knowledge on non-native freshwater fish introductions. Journal of Fish Biology 76, 751-786. Hargrave, C. W. (2006) A test of three alternative pathways for consumer regulation of primary productivity. Oecologia, 149, 223-232. Harris, J. H. and Gehrke, P. C. (1994) Modelling the relationship between streamflow and population recruitment to manage freshwater fisheries. Agricultural Systems and Information Technology 6, 28-30. Hayes, D.B., Ferreri, C.P., and Taylor, W.W. (1996) Active fish capture methods. In Fisheries Techniques. Edited by B.R.a.W. Murphy, D.W. American Fisheries Society, Bethesda, Marysland, USA. pp. pp 193-220. Hillyard, K.A., Smith, B.B., Conallin, A.J., and Gillanders, B.M. (2010) Optimising exclusion screens to control exotic carp in an Australian lowland river. 61, 418-429. Howe, E., Howe, C., R., L. and Burchett, M. (1997) Impact of the introduced poeciliid Gambusia holbrooki (Girard, 1859) on the growth and reproduction of Pseudomugil signifer (Kner, 1865) in Australia. Marine and Freshwater Research 48, 425-434. Humphries, P., Serafini, L. and King, A. J. (2002) River regulation and fish larvae: variations through space and time. Freshwater Biology 47, 1307-1331. Hurlbert, S. H., Zedler, J. and Fairbanks, D. (1972) Ecosystem alteration by Mosquitofish (Gambusia affinis ) predation. Science 175, 639-641. Irons, K. S., Sass, G. G., McClelland, M. A. and Stafford, J. D. (2007) Reduced condition factor of two native fish species coincident with invasion of non-native Asian carps in the Illinois River, U.S.A. Is this evidence for competition and reduced fitness? Journal of Fish Biology 71, 258-273. Ivantsoff, W. and Aarn (1999) Detection of predation on Australian native fishes by Gambusia holbrooki . Marine and Freshwater Research 50, 467-478. Ives, A.R., Dennis, B., Cottingham, K.L., Carpenter, S.R. (2003) Estimating community stability and ecological interactions from time-series data. Ecological Monographs, 73, 301-330. Jackson, D.A., Peres-Neto, P.R. and Olden, J.D. (2001) What controls who is where in freshwater fish communities - the roles of biotic, abiotic, and spatial factors. Canadian Journal of Fisheries and Aquatic Sciences 58, 157-170. Jackson, J. E., Raadik, T. A., Lintermans, M. and Hammer, M. (2004) Alien salmonids in Australia: impediments to effective impact management, and future directions. New Zealand Journal of Marine and Freshwater Research 38, 447-455. Jones, M., and Stuart, I.G. (2004) Impact of flow regulation structures on fish in the Barmah- Millewa forest. Final report to Barmah-Millewa Forum, Arthur Rylah Institute for Environmental Research, Heidelberg.

127 Native fish recovery following Eastern Gambusia removal

Karolak, S. (2006) Alien Fish in the Murray–Darling Basin. MDBC publication No. 03/06. Murray–Darling Basin Commission, Canberra. Kelleway, J., Mazumder, D., Wilson, G.G., Saintilan, N., Knowles, L., Iles, J. and Kobayashi, T. (2010) Trophic structure of benthic resources and consumers varies across a regulated floodplain wetland. Marine and Freshwater Research 61, 430-440. Keller, K. and Brown, C. (2008) Behavioural interactions between the introduced plague minnow Gambusia holbrooki and the vulnerable native Australian Rhadinocentrus ornatus , under experimental conditions. Journal of Fish Biology 73, 1714-1729. Kennard, M. J., Arthington, A. H., Pusey, B. J. and Harch, B. D. (2005) Are alien fish a reliable indicator of river health? Freshwater Biology 50, 174-193. Kerezsy, A. (2009) Gambusia control in spring wetlands.Final report South Australian Arid Lands Natural Resources Management Board. Kilsby, N.N. and Walker, K.F. (in press) Behaviour of two small pelagic and demersal fish species in diverse hydraulic environments. River Research and Applications. Kincaid, H. L. (1976) Inbreeding depression in rainbow trout. Transactions of the American Fisheries Society 105, 273-280. King, A. J. (2005) Ontogenetic dietary shifts of fishes in an Australian floodplain river. Marine and Freshwater Research 56, 215-225. King, A. J., Robertson, A. I. and Healey, M. R. (1997) Experimental manipulations of the biomass of introduced carp ( Cyprinus carpio ) in billabongs. I. Impacts on water-column properties. Marine and Freshwater Research 48, 435-443. King, A.J., and Crook, D.A. (2002) Evaluation of a sweep net electrofishing method for the collection of small fish and shrimp in lotic freshwater environments. Hydrobiologia 472, 223-233. King, A.J., Tonkin, Z. and Mahoney, J. (2007) Assessing the effectiveness of environmental flows on fish recruitment in Barmah-Millewa Forest. Final report to Murray–Darling Basin Commission, Canberra. Project No. BMF 2004.09. Prepared by the Arthur Rylah Institute for Environmental Research, Department of Sustainability and Environment, Heidelberg. King, A.J., Ward, K.A., O'Connor, P., Green, D., Tonkin, Z., and Mahoney, J. (2010) Adaptive management of an environmental watering event to enhance native fish spawning and recruitment. Freshwater Biology 55, 17-31. King, S. and Warburton, K. (2007) The environmental preferences of three species of Australian freshwater fish in relation to the effects of riparian degradation. Environmental Biology of Fishes 78, 307-316. Knight, J. (1999) Density dependent interference competition between the Australian native fish Pseudomugil signifer (Kner, 1865) and the introduced Poeciliid Gambusia holbrooki (Girard, 1859). Integrated project prepared as partial fulfilment of the requirements of the B. App. Sc. (Fisheries and Aquaculture Management). Southern Cross University, Lismore. Koehn, J. (2007) Managing pest fish in the Murray–Darling Basin: lessons learnt from previous invaders. Emerging Issues in Alien Fish Management in the Murray–Darling Basin: Statement, recommendation and supporting papers. Proceedings of a workshop held in Brisbane QLD, 30-31 May 2006, Brisbane, pp. 54-57. Koehn, J. D. and MacKenzie, R. F. (2004) Priority management actions for alien freshwater fish species in Australia. New Zealand Journal of Marine and Freshwater Research 38, 457-472.

128 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Koehn, J. D. and O'Connor, W. G. (1990) Biological Information for Management of Native Freshwater Fish in Victoria. Victorian Government Printing Office, Melbourne. Komak, S. and Crossland, M. R. (2000) An assessment of the introduced mosquitofish ( Gambusia affinis holbrooki ) as a predator of eggs, hatchlings and tadpoles of native and non-native anurans. Wildlife Research 27, 185-189. Koster, W. M. (1997) A study of the interactions between dwarf galaxias ( pusilla ), southern pygmy-perch ( Nannoperca australis ) and eastern gambusia ( Gambusia holbrooki ). B.Sc. Honours thesis, Deakin University, Clayton. Kuiter, R. H., Humphries, P. and Arthington, A. H. (1996) Family Nannoperca: Pygmy-perches. In: Freshwater Fishes of South-Eastern Australia (McDowall, R. M., Ed.), pp. 168-175. Reed Books, Sydney. Laha, M. and Mattingly, H. T. (2007) Ex situ evaluation of impacts of invasive mosquitofish on the imperilled Barrens topminnow. Environmental Biology of Fishes 78, 1-11. Lake, J. S. (1967) Rearing experiments with five species of Australian freshwater fishes. II. Morphogenesis and ontogeny. Australian Journal of Marine and Freshwater Research 18, 155-173. Larson, H. K. and Hoese, D. F. (1996) Gudgeons. In Freshwater Fishes of South-Eastern Australia (McDowall, R. M., ed.), pp. 200-228. Reed Books, Sydney. Leathwick, J.R., Elith, J., Francis, M.P., Hastie, T. and Taylor, P. (2006) Variation in demersal fish species richness in the oceans surrounding New Zealand: an analysis using boosted regression trees. Marine Ecology Progress Series 321, 267-281. Leathwick, J.R., Elith, J., Chadderton, L., Rowe, D. and Hastie, T. (2008) Dispersal, disturbance, and the contrasting biogeographies of New Zealand’s diadromous and nondiadromous fish species. Journal of Biogeography 35, 1481-1497. Leprieur, F., Beauchard, O., Blanchet, S., Oberdorff, T. and Brosse, S. (2008) Fish Invasions in the World’s River Systems: When Natural Processes Are Blurred by Human Activities. PLoS Biology 6, 0404-0410. Ling, N. (2004) Gambusia in New Zealand: really bad or just misunderstood? New Zealand Journal of Marine and Freshwater Research 38, 473-480. Lintermans, M. (2004) Human-assisted dispersal of alien freshwater fish in Australia. New Zealand Journal of Marine and Freshwater Research 38, 481-501. Lintermans, M. (2007) Fishes of the Murray–Darling Basin: an introductory guide: Murray– Darling Basin Commission, Canberra. Llewellyn, L. C. (1973) Spawning, development and temperature tolerance of the spangled perch Madigania unicolor (Günther) from inland waters in Australia. Australian Journal of Marine and Freshwater Research 24, 73-94. Llewellyn, L. C. (1974) Spawning, development and distribution of the southern pygmy-perch Nannoperca australis australis Günther from inland waters in Eastern Australia. Australian Journal of Marine and Freshwater Research 25, 121-129. Lloyd, L. N. (1987) Ecology and distribution of the small native fish of the lower River Murray, South Australia and their interactions with exotic mosquitofish, Gambusia affinis holbrooki (Girard). MSc thesis, University of Adelaide, Adelaide.

129 Native fish recovery following Eastern Gambusia removal

Lloyd, L. N. (1990) Ecological interactions of Gambusia holbrooki with Australian native fishes. In Introduced and translocated fishes and their ecological effect (Pollard, D. A., Ed.), pp. 94-97. Australian Government Publishing Service, Canberra. Lloyd, L. N., Arthington, A. H. and Milton, D. A. (1986) The mosquitofish – a valuable control agent or a pest? In: The Ecology of Exotic Animals and Plants; some Australian Case Histories (Kitching, R. L. Ed.), John Wiley and Sons, Brisbane. Lloyd, L. and Walker, K. F. (1986) Distribution and conservation status of small freshwater fish in the River Murray, South Australia. Transactions of the Royal Society of South Australia 110, 49- 57. Lodge, D. M. and Schrader-Frechette, K. (2003) Nonindigenous species: ecological explanation, environmental ethics, and public policy. Conservation Biology 17, 31-37. Lowe, S. J., Browne, M. and Boudjelas, S. (2000) 100 of the world’s worst invasive alien species. IUCN/SSC Invasive Species Specialist Group (ISSG), Auckland, New Zealand. Lukies, S. (2004) Diet and foraging interactions between the native crimsonspotted rainbowfish (Melanotaenia fluviatilis ) and the introduced mosquitofish ( Gambusia holbrooki ) in the Broken River of northeast Victoria. Unpublished Thesis, Department of Environmental Management and Ecology. La Trobe University, Wodonga. Lunn, D.J., Thomas, A., Best, N., Spiegelhalter, D. (2000) WinBUGS - A Bayesian modelling framework: Concepts, structure, and extensibility. Statistics and Computing 10, 325-337. Lydeard, C. and Belk, M. C. (1993) Management of indigenous fish species impacted by introduced mosquitofish: An experimental approach. The Southwestern Naturalist 38, 370-373. Lyon, J., Stuart, I., Ramsey, D., and O'Mahoney, J. (2010) The effect of water level on lateral movements of fish between river and off-channel habitats and implications for management. Marine and Freshwater Research 61, 271-278. Macdonald, J.I. and Crook, D.A. (2006) Contribution of the Barmah-Millewa floodplain lakes to carp populations in the mid-Murray River, 2005⁄06. Report to the Murray–Darling Basin Commission. Arthur Rylah Institute for Environmental Research, Heidelberg, Victoria. Macdonald, J., and Tonkin, Z. (2008) A review of the impacts of Eastern Gambusia on native fishes of the Murray–Darling Basin. Arthur Rylah Institute for Environmental Research. Department of Sustainability and Environment. Murray–Darling Basin Authority Publication No. 38/09 Maezono, Y. and Miyashita, T. (2004). Impact of exotic fish removal on native communities in farm ponds. Ecological Research 19, 263-267. Maglio, V. J. and Rosen, D. E. (1969) Changing preferences for substrate colour by reproductively active mosquito fish Gambusia affinis (Baird and Girard) (Poeciliidae: Atherinoformes). American Museum Novitates 2397, 1-39. Magoulick, D.D. and Kobza, R.M. (2003) The role of refugia for fishes during drought: a review and synthesis. Freshwater Biology 48, 1186-1198. Marchetti, M. P., Moyle, P. B. and Levine, R. (2004) Invasive species profiling? Exploring the characteristics of non-native fishes across invasion stages in California. Freshwater Biology 49, 646-661.

130 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Margaritora, F. G., Ferrara, O. and Vagaggini, D. (2001) Predatory impact of the mosquitofish (Gambusia holbrooki Girard) on zooplanktonic populations in a pond at Tenuta di Castelporziano (Rome, Central Italy). Journal of Limnology 60, 189-193. Mathews, W.J. and Marsh-Mathews, E. (2003) Effects of drought on fish across axes of space, time and ecological complexity. Freshwater Biology 48, 1232-1253. Maynard, D., Gaston, T. and Henderson, A. (2008) Assessment of removal techniques for Gambusia holbrooki , to assist in the eradication of an introduced pest from ecologically sensitive environments. Milestone report submitted to the Tamar Region Natural Resource Management Strategy Reference Group Inc. by the Australian Maritime College, University of Tasmania, Hobart. McDowall, R. M. (1996) Family Poeciliidae: Livebearers. In Freshwater Fishes of South-Eastern Australia (McDowall, R. M., ed.), pp. 116-122. Reed Books, Sydney. McDowall, R. M. and Fulton, W. (1996) Galaxiids. In Freshwater Fishes of South-Eastern Australia (McDowall, R. M., ed.), pp. 52-77. Reed Books, Sydney. McKay, R. J. (1984) Introductions of exotic fishes in Australia. In Distribution, Biology and Management of Exotic Fishes (Courtenay, W. R. and Stauffer, J. R., eds.). John Hopkins University Press, Baltimore. McKay, S., Clunie, P., Gillespie, G., Raadik, T., Saddlier, S., O'Brien, T., Ryan, T. and Aland, G. (2001) Predation by Gambusia holbrooki : A review of the literature. Arthur Rylah Institute for Environmental Research, Department of Natural Resources and Environment, Heidelberg. McKinnon, L.J. (1997) Monitoring of Fish Aspects of the Flooding of Barmah Forest. Final report to Murray–Darling Basin Commission. Marine and Freshwater Resources Institute, Queenscliff, Australia. McLeod, R. (2004) Counting the cost: Impact of Invasive Animals in Australia 2004. Cooperative Research Centre for Pest Animal Control, Canberra. McMaster, D., and Bond, N. (2008) A field and experimental study on the tolerances of fish to Eucalyptus camaldulensis leachate and low dissolved oxygen concentrations. Marine and Freshwater Research 59, 177-185. McNeil, D. G. (2004) Ecophysiology and behavior of Ovens River floodplain fish: Hypoxia tolerance and the role of the physicochemical environment in structuring Australian billabong fish communities. Unpublished PhD Thesis. Cooperative Research Centre for Freshwater Ecology, LaTrobe University, Wodonga. McNeil, D. G. and Closs, G. P. (2007) Behavioural responses of a south-east Australian floodplain fish community to gradual hypoxia. Freshwater Biology 52, 412-420. MDBC (2004) Native Fish Strategy for the Murray–Darling Basin 2003-2013. Murray–Darling Basin Commission Publication No. 25/04. Canberra. MDBC (2008) Murray–Darling Basin Rivers: Ecosystem Health Check, 2004-2007. A summary report based on the independent Sustainable Rivers Audit Group’s SRA Report 1: A report on the ecological report of rivers in the Murray–Darling Basin, 2004-2007, submitted to the Murray– Darling Basin Ministerial Council in May 2008. Meffe, G. K. and Snelson, F. F. Jnr. (1989) An ecological overview of poeciliid fishes. In Ecology and evolution of livebearing fishes (Poecilidae) (Meffe G. K. and Snelson, F. F. Jnr. Eds.). pp. 13- 31, Prentice-Hall, Inc., New Jersey.

131 Native fish recovery following Eastern Gambusia removal

Meffe, G. K., Hendrickson, D. A. and Minckley, W. L. (1983) Factors resulting in decline of the endangered Sonoran topminnow Poeciliopsis occidentalis (: Poeciliidae) in the United States. Biological Conservation 25, 135-159. Merrick, J. R. and Schmida, G. E. (1984) Australian freshwater fishes: biology and management. J. Merrick, North Ryde, New South Wales. Mills, M. D., Radar, R. B. and Belk, M. C. (2004) Complex interactions between native and invasive fish: the simultaneous effects of multiple negative interactions. Oecologia 141, 713-721. Milton, D. A. and Arthington, A. H. (1983) Reproductive biology of Gambusia affinis holbrooki Baird and Girard, Xiphophorus helleri (Gunther) and X. maculatus (Heckel) (Pisces; Poeciliidae) in Queensland, Australia. Journal of Fish Biology 23, 23-41. Morgan, D. L., Gill, H. S., Maddern, M. G. and Beatty, S. J. (2004) Distribution and impacts of introduced freshwater fishes in Western Australia. New Zealand Journal of Marine and Freshwater Research 38, 511-523. Morris, S.A., Pollard, D.A., Gehrke, P.C. and Pogonoski, J.J. (2001) Threatened and potentially threatened freshwater fishes of coastal New South Wales and the Murray–Darling Basin. NSW Fisheries Final Report Series No. 33. Morrongiello, J., Elith, J. and Crook, D. (2006) Impacts of drought on fish in Victorian rivers and streams. Arthur Rylah Institute for Environmental Research, Heidelberg, Victoria. Morrongiello, J.R., Bond, N.R., Crook, D.A. and Wong, B.B.M. (in press). Eucalyptus leachate inhibits reproduction in a freshwater fish. Freshwater Biology. Moyle, P. B. and Light, T. (1996) Biological invasions of fresh water: Empirical rules and assembly theory. Biological Conservation 78, 149-161. Murphy, E. C., Kay Clapperton, B., Bradfield, P. M. F. and Speed, H. J. (1998) Effects of rat- poisoning operations on abundance and diet of mustelids in New Zealand podocarp forests. New Zealand Journal of Zoology 25, 315-328. Myers, G. S. (1965) Gambusia , the fish destroyer. Australian Zoologist 13, 102. NSW (2003) New South Wales threat abatement plan. Predation by Gambusia holbrooki - the plague minnow. NSW, National Parks and Wildlife Service. Hurstville, NSW. NSW (2010) NSW control plan for the noxious fish carp Cyprinus carpio . Aquatic Biosecurity & Risk Managment Unit, Industry and Investment. Orange, NSW. Parenti, L. R. and Rauchenberger, M. (1989) Systematic overview of the Poeciliines. In Ecology and Evolution of Livebearing Fishes (Poeciliidae) (Meffe, G. K. and Snelson, F. F. Eds.), pp. 3-31. Englewood Cliffs, Prentice Hall, New Jersey. Pen, L. J. and Potter, I. C. (1991) Reproduction, growth and diet of Gambusia holbrooki (Girard) in a temperate Australian river. Aquatic Conservation: Marine and Freshwater Ecosystems 1, 159- 172. Pen, L. J., Potter, I. C. and Calver, M. C. (1993) Comparisons of the food niches of three native and two introduced fish species in an Australian river. Environmental Biology of Fishes 36, 167- 182. Peterson, D. P. and Fausch, K. D. (2003) Upstream movement nonnative brook trout ( Salvelinus fontinalis ) promotes invasion of native cutthroat trout ( Oncorhynchus clarki ) habitat. Canadian Journal of Fisheries and Aquatic Sciences 60, 1502-1516.

132 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Pimental, D., Lach, L., Zuniga, R. and Morrison, D. (2000) Environmental and economic costs of nonindigenous species in the United States. BioScience 50, 53-65. Pimentel, D., Zuniga, R. and Morrison, D. (2005) Update on the environmental and economic costs associated with alien-invasive species in the United States, Ecological Economics 52, 273- 288. Pitman, K. S. and Tinkler, P. (2007) Recovery planning survey for dwarf galaxias ( Galaxias pusilla ) in the West Gippsland region, Victoria. Arthur Rylah Institute for Environmental Research, Department of Sustainability and Environment, Heidelberg. Poff, N.E., and Allan, J.D. (1995) Functional organisation of stream fish assemblages in relation to hydrological variability. Ecology 76: 606-627. Potter, I. C. (1996a) Family Geotriidae: pouched lampreys. In: Freshwater Fishes of South-Eastern Australia (McDowall, R. M., Ed.), pp. 36-38. Reed Books, Sydney. Potter, I. C. (1996b) Family Mordaciidae: shortheaded lampreys. In: Freshwater Fishes of South- Eastern Australia (McDowall, R. M., Ed.), pp. 32-35. Reed Books. Sydney. Potter, I. C., Hilliard, R. W., Bradley, J. S. and McKay, R. J. (1986) The influence of environmental variables on the density of larval lampreys in different seasons. Oecologia 70, 433- 440. Puckridge, J. T. and Walker, K. F. (1990) Reproductive biology and larval development of a gizzard shad, Nematalosa erebi (Gunther) (Dorosomatinae: Teleostei), in the River Murray, South Australia. Australian Journal of Marine and Freshwater Research 41, 695-712. Pyke, G. H. (2005) A review of the biology of Gambusia affinis and G. holbrooki . Reviews in Fish Biology and Fisheries 15, 339-365. Pyke, G. H. (2008) plague minnow or mosquito fish? A review of the biology and impacts of introduced Gambusia species. Annual Review of Ecology and Systematics 39, 171-191. Ramsey, D.S.L., and Wilson, J.C. (2000) Towards ecologically based baiting strategies for rodents in agricultural systems. International Biodeterioration & Biodegradation 45, 183-197. Raynor, T.S., and Creese, R.G. (2006) A review of rotenone use for the control of non-indigenous fish in Australian fresh waters, and an attempted eradication of the noxious fish, Phalloceros caudimaculatus . New Zealand Journal of Marine and Freshwater Research 40, 477-486. Rehwinkel, R., and Sharpe, C. (2009) Gunbower Forest Fish Monitoring Surveys. Report prepared for North Central Catchment Management Authority, January, 2009. Richardson, A., Meredith, S., Conallin, A., and Sharpe, C. (2005) Status of the Gunbower Island fish community, June 2005 - Including recommendations for future monitoring. MDFRC Consultancy Report for the North Central Catchment Management Authority. Murray–Darling Freshwater Research Centre, Wodonga. Rincón, P. A., Correas, A. M., Morcillo, F., Risueño, P. and Lobón-Cerviá, J. (2002) Interaction between the introduced eastern mosquitofish and two autochthonous Spanish toothcarps. Journal of Fish Biology 61, 1560-1585. Roberts, J., Chick, A., Oswald, L. and Thompson, P. (1995) Effect of carp, Cyprinus carpio L. an exotic benthivorous fish, on aquatic plants and water quality in experimental ponds. Marine and Freshwater Research 46, 1171-1180.

133 Native fish recovery following Eastern Gambusia removal

Rowe, D. K., Moore, A., Giorgetti, A., Maclean, C., Grace, P., Wadhwa, S. and Cooke, J. (2008) Review of the impacts of gambusia, redfin perch, tench, roach, yellowfin goby and streaked goby in Australia. Draft report prepared for the Australian Government Department of the Environment, Water, Heritage and the Arts, Canberra. Rowe, D. K., Smith, J. P. and Baker, C. (2007) Agonistic interactions between Gambusia affinis and Galaxias maculatus : implications for whitebait fisheries in New Zealand rivers. Journal of Applied Ichthyology 23, 668–674. Rowland, S. J. (1992) Diet and feeding of Murray Cod ( Maccullochella peelii ) larvae. Proceedings of the Linnean Society of New South Wales 113, 193-201. Saccheri, I., Kuussaari, M., Kankare, M., Vikman, P., Fortelius W. and Hanski, I. (1998) Inbreeding and extinction in a butterfly metapopulation. Nature 392, 491-493. Sala, O. E., Chapin, F. S. III., Armesto, J. J., Berlow, E., Bloomfield, J., Dirzo, R., Huber- Sanwald, E., Huenneke, L. F., Jackson, R. B., Kinzig, A., Leemans, R., Lodge, D. M., Mooney, H. A., Oesterheld, M., Poff, N. L., Sykes, M. T., Walker, B. H., Walker, M. and Wall, D. H. (2000) Global biodiversity scenarios for the year 2100. Science 287, 1770-1774. Scurr, G. (2007) Gambusia control update. Tasmanian Conservationist 311, 10-11. Sivakumaran, K.P., Brown, P. Stoessel, D. and Giles, A. (2003) Maturation and reproductive biology of female wild carp, Cyprinus carpio , in Victoria, Australia. Environmental Biology of Fishes 68, 321-332. Smith, B.B. and Walker K.F. (2004) Reproduction of Common Carp in South Australia, shown by young-of-the-year samples, gonadosomatic index and the histological staging of ovaries. Transactions of the Royal Society of South Australia 128, 249-257. SRA (2007) Victorian Sustainable Rivers Audit Fish data summary. Arthur Rylah Institute for Environmental Research, Department of Sustainability and Environment, Heidelberg. Stoffels, R. J. and Humphries, P. (2003) Ontogenetic variation in the diurnal food and habitat associations of an endemic and an exotic fish in floodplain ponds: consequences for niche partitioning. Environmental Biology of Fishes 66, 293-305. Stuart, I. G. and Jones, M. (2002) Ecology and management of Common Carp in the Barmah- Millewa Forest. Final report to Agriculture, Fisheries and Forestry Australia. Arthur Rylah Institute for Environmental Research. Department of Sustainability and Environment, Heidelberg. Stuart, I.G., and Jones, M. (2006) Movement of Common Carp, Cyprinus carpio , in a regulated lowland Australian river: implications for management. Fisheries Management and Ecology 2006, 231-219. Stuart, I.G., Williams, A., McKenzie, J., and Holt, T. (2006) Managing a migratory pest species: a selective trap for Common Carp. North American Journal of Fisheries Management 26, 888-893. Swanson, C., Cech, J. J. J. and Piedrahita, R. H. (1996) Mosquitofish: Biology, Culture, and Use in Mosquito Control. California, U.S.A.: Mosquito and Vector Control Association of California; The University of California Mosquito Research Program. California. Tonkin, Z. and Rourke, M. (2008) Barmah-Millewa fish condition monitoring - 2008 annual summary and refuge habitat report submitted to the Murray–Darling Basin Commission. Arthur Rylah Institute for Environmental Research (Department of Sustainability and Environment) and NSW Department of Primary Industries.

134 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Tonkin, Z., King, A.J., and Mahoney, J. (2008) Effects of flooding on recruitment and dispersal of the Southern Pygmy-perch (Nannoperca australis) at a Murray River floodplain wetland. Ecological Management and Restoration 9, 196-201. Tonkin, Z., Humphries, P. and Pridmore, P. A. (2006) Ontogeny of feeding in two native and one alien fish species from the Murray–Darling Basin, Australia. Environmental Biology of Fishes 76, 303-315. Townsend, C.R. (2003) Individual, population, community and ecosystem consequences of a fish invader in New Zealand streams. Conservation Biology 17, 38-47. Treadwell, S. and Hardwick, R. (2004) Review of Habitat Associations of Native Fish in the Murray–Darling Basin. Murray–Darling Basin Commission project R2105. Sinclair Knight Merz, Malvern. Unmack, P. J. (2001) Biography of Australian freshwater fishes. Journal of Biogeography 28, 1053-1089. Vitousek, P. M., Mooney, H. A., Lubchenco, J. and Melillo, J. M. (1997) Human domination of earth's ecosystems. Science 277, 494-499. Warburton, K., Retif, S. and Hume, D. (1998) Generalists as sequential specialists: diets and prey switching in juvenile silver perch. Environmental Biology of Fishes 51, 445-454. Webb, C. and Joss, J. (1997) Does predation by the fish Gambusia holbrooki (Atheriniformes: Poeciliidae) contribute to declining populations. Australian Zoologist 30, 316-324. Wedderburn, S. D., Walker, K. F. and Zampatti, B. P. (2007) Habitat separation of Craterocephalus (Artherinidae) species and populations in off-channel areas of the lower River Murray, Australia. Ecology of Freshwater Fish 16, 442-449. Weigel, D. E., Peterson, J. T. and Spruell, P. (2002) A model using phenotypic characteristics to detect introgressive hybridisation in wild Westslope cutthroat trout and rainbow trout. Transactions of the American Fisheries Society 131, 389-403. Wells, A. (2007) Community understanding and attitudes to alien fish. Emerging Issues in Alien Fish Management in the Murray–Darling Basin: Statement, recommendation and supporting papers. Proceedings of a workshop held in Brisbane QLD, 30-31 May 2006, Brisbane, pp. 69-72. West, P., Brown, A. and Hall, K. (2007) Review of alien fish monitoring techniques, indicators and protocols: implications for national monitoring of Australia’s inland river systems. Invasive Animals Cooperative Research Centre, prepared for the National Land and Water Resources Audit, Canberra. Williamson, M. (1996) Biological Invasions. Chapman & Hall, London. Willis, K. and Ling, N. (2000) Sensitivities of mosquitofish and black mudfish to a piscicide: could rotenone be used to control mosquitofish in New Zealand wetlands? New Zealand Journal of Zoology 27, 85-91. Wilson, G. (2006) Impact of invasive exotic fishes on wetland ecosystems in the Murray–Darling Basin. In: Native fish and wetlands in the Murray–Darling Basin: Action Plan, knowledge gaps and supporting papers. (Phillip, B. Ed.). Murray–Darling Basin Commission, Canberra. Winemiller, K.O., Tarim, S., Shoreman, D., and Cotner, J.B. (2000) Fish assemblage structure in relation to environmental variation among Brazos river oxbow lakes. Transactions of the American Fisheries Society 129, 451-468.

135 Native fish recovery following Eastern Gambusia removal

Zavaleta, E. S., Hobbs, R. J. and Mooney, H. A. (2001) Viewing invasive species removal in a whole-ecosystem context. Trends in Ecology and Evolution 16, 454-459.

136 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Appendices Appendix 1: Eastern Gambusia marking trial

An assessment of short term mortality of Eastern Gambusia Gambusia holbrooki after being subject to an osmotic induction marking technique Abstract The use of chemical markers has become a common fish tagging technique for mark-recapture studies, particularly in situations that involve either marking large numbers of small individuals and/or where long-term marks are required. A recent project utilised an osmotic induction marking protocol using calcein (adopted from Crook et al. (2009) to mark Eastern Gambusia Gambusia holbrooki in the wild. However, preliminary observations suggested that the marking protocol may have been causing increased mortality, which was not desirable for subsequent population measures derived from recapture data. This study was therefore designed to assess whether the marking protocol has any impact on short term mortality and growth.

The experiment indicated a significant increase in rates of short-term mortality of adult Eastern Gambusia following the osmotic induction marking protocol. This mortality stabilised 3 – 6 days following marking, with mortality of marked fish approximately 25% higher for female fish and 50% higher for male fish, than fish that were not subject to the marking procedure. There was no influence of the marking protocol on overall growth of male or female fish. The results of the study have demonstrated that the osmotic induction marking protocol will increase short term mortality of Eastern Gambusia, and that any mark-recapture data derived from field based marking using this protocol must account for these mortality rates to improve estimates of population size and capture rates. This study highlights the need for developing species specific marking protocols before being applied to field or hatchery based studies.

Introduction Mark-recapture studies provide vital information to key aspects of fisheries and ecological research including population, mortality and methodology assessments. The accuracy of these assessments rely on sound marking techniques which ideally ensure marks (or tags) are retained by individuals for the duration of the required detection period, do not increase individual mortality (or if they do, the rate of mortality must be quantifiable) and do not have any significant effect on the normal biological functioning or behaviour of the fish (either negative or positive). For this reason, chemical markers such as calcein and oxytetracycline have become preferable for marking large numbers of fish to traditional tagging methods (such as floy tags, pit tags etc), given that (i) they can be applied to fish of all ages and developmental stages; (ii) they are not visible under normal sunlight (so should not influence behaviour or predation) and; (iii) they take less time to apply to large numbers of fish and so are more cost effective (e.g. Leips et al. 2001; Crook et al. 2009).

A recent project examining Eastern Gambusia Gambusia holbrooki populations in small wetlands has utilised a mark-recapture method to assess the level of fish removal, as well as provide an

137 Native fish recovery following Eastern Gambusia removal estimate of the abundance of Eastern Gambusia in selected sites. The study utilised an osmotic induction marking protocol developed for use in fish hatcheries producing golden perch Macquaria ambigua (see Crook et al. 2009), to mark large numbers of gambusia . The technique uses the fluorescent dye, calcein, which binds to calcified tissue such as otoliths, scales and the skeleton of the fish (see Mohler 2003; Crook et al. 2009). Marked fish fluoresce in particular zones on their body when exposed to blue or ultraviolet light allowing external detection in a field environment without sacrificing the individual (e.g. Bashey 2004; Crook et al 2009). The osmotic induction technique has the advantage of marking large numbers of individuals in a very short time, and thus is ideal for use on small bodied species in the field.

The detection period of the mark-recapture component in the study was undertaken for a maximum of 14 days post-marking (generally less than 10 days). Given this technique has been shown to produce detectable external marks for months or even years (see Negus and Tureson 2004; Bashey 2004) any influence on mark retention over a period of days is likely to be minimal. This was confirmed by all individual fish recaptured displaying clear external marking, particularly on ventrally located bony structures which are not as subject to photodegredation as dorsal areas. Recapture rates however, were slightly lower than expected and several recaptured individuals displayed what appeared to be a fungal growth on their caudal fin. This caused concern that the technique may be influencing mortality in this species, which would have serious implications for reliably estimating of removal efficiency and population size.

This study was used to assess any short-term influence on growth and mortality of adult Eastern Gambusia after they have been subject to the osmotic induction marking protocol. This will ultimately improve the accuracy of the field based mark-recapture population and removal assessments which use this technique to mark adult fish.

Methods Approximately 100 adult Eastern Gambusia (> 24 mm total length) were collected from a farm dam in central Victoria during October 2010 using fine-mesh (1mm) seine netting and transported to an aquarium facility. The following day, half of the fish received a calcein mark via the osmotic induction method as described by Crook et al (2009). The technique involves a 5% salt solution being prepared by dissolving 100 g of commercially available natural salt in 2L of water. A 0.5% solution of calcein is prepared by adding 10 g of calcein powder (2,4-bis-[N,N0-fdicarbo methylg- aminomethyl] fluorescein) to 2L of water. At this concentration, the calcein causes a decrease in the pH of the solution, and is inturn adjusted back to 7.0 by gradually adding sodium hydroxide (NaOH) to the solution. Both solutions were aerated by bubbling air into the solutions during the marking procedures. Fifty Eastern Gambusia were placed in a mesh bottom vessel and immersed in the salt solution for three minutes. After immersion in the salt solution, fish were rinsed in fresh water for 5 seconds, and then immersed in the calcein dye solution for five minutes. Fish were then placed into a 20L aerated container and assessed for signs of stress. Individual fish from both were then assessed for gender and measured for total length (TL; nearest mm) and randomly assigned to one of four 10L bucket replicates at a rate of six females and three male fish per replicate. Additional control groups consisting of fish that were not exposed to the marking procedure were also assessed for gender, measured for TL and assigned to at the same abundance and sex ratio as the treatment tanks. This resulted in a total of eight 10 L vessels (four treatment and four controls), each of which contained six female fish and three male fish. Each vessel had a flow through water

138 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal system (ambient water temperature ranging 17 – 22 °C) and aeration supplied for the duration of the experiment. The experiment was run for a period of 15 days, where a daily routine occurred of feeding of blood worms, mortality check and general aquaria maintenance. Any dead fish were recorded and removed from the experiment. At the completion of the 15 th day, all surviving individuals were measured, assessed for an external mark with a handheld blue light, then euthanized with anaesthetic solution (Alfaxalone – 40 mg/L for 10 minutes), and then preserved in 95% ethanol.

Mortality data from the experiment at day zero, three, six, nine, 12 and 15, was converted to percentage of overall mortality and compared between treatment and controls using a repeated measures Analysis of variance (ANOVA) with treatment, sex, time and their interaction as factors. Total growth (mm) for all surviving individuals was calculated by using the mean initial and final standard length (log e[mean SL final ] - log e[mean SL initial]). A two-way ANOVA was used to test whether total growth differed due to treatment (marked or unmarked), sex and their interaction.

Results At the completion of the experiment, fish mortality was found to be at a relatively high rate in both treatment (marked) and control (unmarked) vessels. Total mortality for control and treatments combined was 46 % for males and 44% for females. Some fish that were subject to the marking procedure developed a fungus-type growth on their caudal fin, which rapidly developed over the posterior of their body, followed by mortality (within two days). As a result, mortality rates were significantly higher in fish that had been subject to the marking procedure than those that were not marked (Figure 1; treatment, F = 21.85; p < 0.001). Mortality rates in marked fish were higher in males than female fish, however, the differences were not significant (F = 4.07; p = 0.067).

Not surprisingly mortality rates were significantly influenced by the duration of the experiment (time; F = 17.93; p < 0.001), where mortality was highest across all treatments at the end of the experiment (Figure 1). Mortality rates of female fish in control and treatments also increased during the final days of the experiment. Of most relevance was the interaction between duration and treatment, with male and female fish having significantly higher mortalities 3-6 days following marking (Figure 1; F = 6.23; p = 0.001). This mortality stabilised after this time, with mortality of marked fish approximately 25% higher for female fish and 50% higher for male fish than fish that were not subject to the marking procedure. There was, however, much higher variability in male mortality (most likely due to the smaller sample size).

There was little variation in overall growth of surviving fish over the 15-day period. Overall, growth was not significantly influenced by gender, whether it was marked or unmarked, or their interaction (Figure 2; all p > 0.05). At the completion of the experiment, all fish in marked treatments displayed clear fluorescent marking on scales and bony surfaces when viewed under a handheld ultraviolet light (Figure 3).

139 Native fish recovery following Eastern Gambusia removal

100 Male

80

60

40

20

0

100 Female 80

Cumulativemortality % 60

40

20

0 0 3 6 9 12 15 Day

Figure 1. Mean ± SE cumulative percent mortality of fish subject to osmotic induction marking (treatment; Pink line) and unmarked fish (Control; black line) for both adult males (top; n = 12 marked and unmarked fish) and females (bottom; n = 24 marked and unmarked fish) for every third day of the experiment.

140 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

MALE FEMALE 8

6 4

2

(mm) Growth 0 Unmarked Marked Unmarked Marked

Figure 2. Growth in total length (Mean ± SE) of female and male fish over the 15-day period from unmarked (control) and marked (treatment) replicates.

Figure 3. Photograph of adult female gambusia 15 days post-marking displaying clear fluorescent marking under ultraviolet light (viewed through an amber lense).

141 Native fish recovery following Eastern Gambusia removal

Discussion The results of the experiment indicate a significant increase in mortality of adult Eastern Gambusia following the osmotic induction marking procedure. This mortality stabilised 3 – 6 days following marking after which time mortality of marked fish was approximately 25% higher for female fish and 50% higher for male fish than fish that were not subject to the marking procedure. The slightly higher mortality of marked male fish compared to marked female fish may be due to their smaller body size, although results should be treated with some caution given the small sample size. Leips et al. (2001) and Bashey (2004) reported calcein marking of other poeciliids ( Heterandria formosa and Poecilia reticulate ) having no significant effect on growth or mortality after conducting long and short term experiments (9 weeks and 14 days respectively). Both of these studies, however, did not use the osmotic induction technique for marking (i.e. no salt bath before exposure to a concentrated chemical). Crook et al. (2009) presented a protocol using the osmotic induction technique with calcein, which produced no significant increase in mortality (reduction in growth) when marking golden perch fingerlings. At fingerling stage, this species of percichthyid are much larger, and perhaps more robust, when compared to adult Eastern Gambusia, particularly male Eastern Gambusia. Therefore, larger fish may be more tolerant of the marking technique than smaller individuals. Mohler (2003), who developed the osmotic induction technique, reported no significant difference in mortality between Salmo salar that had been marked with calcein through osmotic induction and control fish, which had only been exposed to the salt bath. Mohler (2003) reported that after 47 days fish in both control and marked treatments had mortalities between 11-28%, which is similar to mortality levels of female Eastern Gambusia in the present study. Furthermore, a 5% salt solution was also used in his study. This may suggest that the salt bath is the process in the technique that may influence mortality, not the calcein itself. Further experimentation varying dosage rates and holding times used in the osmotic induction marking protocol are required to further test this. Similar to other studies (Negus and Tureson (2004) and Bashey (2004)), marking did not influence the short term growth of surviving fish in this study. Crook et al. (2009) reported significantly higher growth rates for calcein marked fish than unmarked fish during their study on golden perch fingerlings, suggesting the salt exposure may have had a prophylactic effect given it is commonly used in hatcheries for disease and parasite treatment. All surviving fish showed clear fluorescent marks on hard surfaces such as scales and fins, fifteen days after their initial marking. Whilst assessing mark duration was not an objective of this trial, these results are supported by other similar studies which have shown external mark longevity in numerous other species to greatly exceed the 15-days used in this study (e.g. Negus and Tureson 2004; Stubbing and Moss 2007; Crook et al. 2009). However, we suggest future trials of mark longevity are conducted outdoors (subject to sunlight and ambient temperatures) for Eastern Gambusia if external marks are required to be detected in the field for extended periods (> 15 days). Eastern Gambusia recaptured in the field have shown substantial photodegredation of marks as little as 7-days after marking (particularly dorsal surfaces; pers. obs). This is not unexpected given that the detectability of marked fish may decrease rapidly in fish that are more exposed to sunlight and higher temperatures (Leips et al. (2001) and Bashey (2004)), coupled with Eastern Gambusia’s tendency to occupy warm shallow / surface water (Karolak 2006). The results of the trial have demonstrated that that the osmotic induction marking protocol adopted from Crook et al. (2009), will increase short term mortality of Eastern Gambusia. Therefore, any mark-recapture data derived from field based marking that uses this marking procedure must account for the mortality rates presented in this study to improve estimates of population size and

142 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal capture rates. Whilst the protocol as it stands is indeed useful, the trial highlights a need for further experimental trials of varied exposure rates to develop species specific marking protocols. The trial has also demonstrated that mortality rates may also differ between genders and that this should be considered when developing such marking protocols. This reinforces the recommendation by Leips et al. (2001) whereby these techniques should be tested for adverse or beneficial effects on the species in question.

References Bashey, F. (2004). A comparison of the suitability of Alizarin Red S and Calcein for inducing a nonlethally detectable mark in juvenile guppies. Transactions of the American Fisheries Society 133, 1516-23.

Crook, D., A., O’Mahony, D.,J., Sanger, A., C., Munro, A., R., Gillanders, B., M., and Thurstan., S. (2009) Development and Evaluation of Methods for Osmotic Induction Marking of Golden Perch, Macquaria ambigua with Calcein and Alizarin Red S. Northern American Journal of Fisheries Management 29: 279- 287.

Karolak, S. (2006) Alien Fish in the Murray–Darling Basin. MDBC publication No. 03/06, Murray–Darling Basin Commission, Canberra.

Leips, J., Baril, C.T., Rodd, F.H., Reznick, D.N., Bashey, F., Visser, G.J. & Travis, J. (2001) The suitability of calcein to mark poeciliid fish and a new method of detection. Transactions of the American Fisheries Society, 130, 501-507.

Mohler, J.W. (2003) Producing Fluorescent Marks on Atlantic Salmon Fin Rays and Scales with Calcein via Osmotic Induction. North American Journal of Fisheries Management, 23, 1108-1113.

Negus, M.T. and Tureson, F.T. (2004). Retention and nonlethal detection of Calcein marks in rainbow and chinook salmon. North American Journal of Fisheries Management 24, 741-47.

Stubbing, D. N., and Moss, R. D. (2007). Success of calcein marking via osmotic induction in brown trout fry, Salmo trutta . Fisheries Management and Ecology 14:231–233.

143 Native fish recovery following Eastern Gambusia removal

Appendix 2: Project communications

Publications DSE (2008) “Recovery of Wetland Native Fish Communities Following Removal of Eastern Gambusia”. Arthur Rylah Institute for Environmental Research Website Article. http://www.dse.vic.gov.au/arthur-rylah-institute/research-themes/invasive-species Macdonald, J., Tonkin, Z., Ramsey, D., Kaus, A., King, A. K. and Crook, D.A. (In Press) Do invasive eastern gambusia (Gambusia holbrooki) shape wetland fish assemblage structure in south- eastern Australia? Marine and Freshwater Research Tonkin, Z., Macdonald, J., Ramsey, D., Kaus, A., Crook, D.A. and King, A.K. (In prep) Fish community responses to reductions of an alien species. Applied Ecology Tonkin, Z., Kaus, A., Ramsey, D., Macdonald, J., Crook, D., Hames, F., and King, A. (2011) A field based assessment of native fish responses following Eastern Gambusia removal: putting theory into practice. Conference proceedings from the MDBA Gambusia Forum, Melbourne , 1-2 June 2011. Tonkin, Z., Kaus, A., Ramsey, D., Macdonald, J., Crook, D., Hames, F., and King, A. (2010) Assessing the recovery of wetland native fish communities following removal of the introduced Eastern Gambusia, Gambusia holbrooki. Conference proceedings from the 2010 Native Fish Forum, National Museum of Australia, Canberra 15 – 16 September 2010. Macdonald, J., and Tonkin, Z. (2009) Eastern Gambusia and native fish communities in wetlands – assessing native fish recovery following removal of an alien invader. Article in the Australian Society for Fish Biology Newsletter, 2009. Macdonald, J., and Tonkin, Z. (2008) A review of the impacts of Eastern Gambusia on native fishes of the Murray–Darling Basin. Arthur Rylah Institute for Environmental Research. Department of Sustainability and Environment. Murray–Darling Basin Authority Publication No. 38/09 Macdonald, J., Tonkin, Z., Ramsey, J. and Jin, C. (2008) Native fish recovery following alien species removal. Conference proceedings from the Native Fish Forum, Canberra 9 – 10 September 2008. Tonkin, Z. and Macdonald, J (2008) “Alien invaders in our billabongs and wetlands”. DSE Inform Article. Department of Sustainability and Environmental, August 2008.

Presentations Tonkin, Z., Kaus, A., Ramsey, D., Macdonald, J., Crook, D., Hames, F., and King, A. (in prep) A field based assessment of native fish responses following Eastern Gambusia removal: putting theory into practice. Spoken presentation at the ASFB conference, Townsville, 22-23 July 2011. Tonkin, Z., Kaus, A., Ramsey, D., Macdonald, J., Crook, D., Hames, F., and King, A. (2011) A field based assessment of native fish responses following Eastern Gambusia removal: putting theory into practice. Spoken presentation by Z. Tonkin at the MDBA Gambusia Forum, Melbourne, 1-2 June 2011. Tonkin, Z., Kaus, A., Ramsey, D., Macdonald, J., Crook, D., Hames, F., and King, A. (2010) Assessing the recovery of wetland native fish communities following removal of the introduced

144 Arthur Rylah Institute for Environmental Research Native fish recovery following Eastern Gambusia removal

Eastern Gambusia, Gambusia holbrooki. Spoken presentation by Z. Tonkin, Native Fish Forum 2010 National Museum of Australia, Canberra 15 – 16 September 2010. Macdonald, J., Tonkin, Z., Ramsey, J. and Jin, C. (2008) Native fish recovery following alien species removal. Spoken presentation by J. Macdonald, Native Fish Forum, Canberra 9 – 10 September 2008.

Other communications activities (see below) DSE Media release, 31 st May 2011 Project fact sheet Removal guidelines for managers Waranga news, 28 th October 2010 DSE inform June 15 2011

145 Native fish recovery following Eastern Gambusia removal

ISSN 1835-3827 (print) ISSN 1835-3835 (online) ISBN 978-1-74287-410-4 (print)

146 Arthur Rylah Institute for Environmental Research