Waterlines 75. Ecological limits of hydrologic ...

Ecological limits of hydrologic alteration: a test of the ELOHA framework in south-east Queensland AH Arthington, SJ Mackay, CS James, RJ Rolls, D Sternberg, A Barnes, SJ Capon Waterlines Report Series No 75, March 2012

Waterlines This paper is part of a series of works commissioned by the National Water Commission on key water issues. This work has been undertaken by the International WaterCentre in collaboration with the Australian Rivers Institute at Griffith University on behalf of the National Water Commission.

© Commonwealth of Australia 2012 This work is copyright. Apart from any use as permitted under the Copyright Act 1968, no part may be reproduced by

any process without prior written permission.

Requests and enquiries concerning reproduction and rights should be addressed to the

Communications Director, National Water Commission, 95 Northbourne Avenue, Canberra

ACT 2600 or email [email protected].

Online ISBN: 978-1-921853-57-9

Ecological limits of hydrologic alteration: a test of the ELOHA framework in south-east

Queensland, March 2012

Authors: AH Arthington, SJ Mackay, CS James, RJ Rolls, D Sternberg, A Barnes and SJ

Capon

Published by the National Water Commission

95 Northbourne Avenue

Canberra ACT 2600

Tel: 02 6102 6000

Email: [email protected]

Date of publication: March 2012

Cover design by: Angelink

Front cover image courtesy of David Sternberg

An appropriate citation for this report is:

Arthington AH, et al. 2012 Ecological limits of hydrologic alteration: a test of the ELOHA

framework in south-east Queensland, Waterlines report, National Water Commission,

Canberra.

Disclaimer This paper is presented by the National Water Commission for the purpose of informing discussion and does not necessarily reflect the views or opinions of the Commission.

Contents

Acknowledgements Executive Summary Project overview Objectives Study area Project methods Key findings Synthesis of key outcomes of the south-east Queensland ELOHA trial Conclusion 1. Introduction 1.1. Background 1.2. The ELOHA framework 1.3. Project objectives 1.4. Study area 2. Hydrologic classification 2.1. Introduction 2.2. Methods 2.3. Results 2.4. Discussion 3. Study sites—selection and environmental variation 3.1. Introduction 3.2. Methods 3.3. Results 3.4. Discussion 4. Riparian vegetation Introduction 4.1. Methods 4.2. Results 4.3. 4.4. Discussion Aquatic vegetation 5. Introduction 5.1. Methods 5.2. 5.3. Results Discussion 5.4. Fish 6. Introduction 6.1. 6.2. Methods Results 6.3. Discussion 6.4. Synthesis and key findings 7. 7.1. Hydrologic regimes of unregulated rivers in south-east Queensland 7.2. Hydrologic alteration in regulated river basins in south-east Queensland Ecological relationships with hydrologic gradients 7.3. 7.4. Limiting hydrologic variables Influence of other environmental variables 7.5. Geographic scope of the south-east Queensland ELOHA trial results 7.6. 8. Conclusions 8.1. Key outcomes of the South-east Queensland ELOHA trial 8.2. Key recommendations Bibliography

NATIONAL WATER COMMISSION — WATERLINES

11

xiv

1

1

3

5

6

14

16

18

18

20

21

22

27

27

28

32

55

57

57

58

66

69

70

70

72

77

85

88

88

89

97

108

110

110

113

118

136

141

141

143

147

152

159

163

164

164

166

170

iv

Tables

Table E1: Summary of hydrologic metrics correlated with the structure of riparian tree and shrub communities, aquatic vegetation assemblage composition and fish assemblage structure (based on catch per unit effort) ....................................... 11

Table 2.1: Continental-scale flow regime classes occurring in south-east

Queensland ...................................................................................................................... 27

Table 2.2: Hydrologic metrics used in the classification of south-east Queensland

flow regimes ..................................................................................................................... 30

Table 2.3: Pre- and post-dam periods of hydrologic record used for the range of

variability approach (RVA). Insufficient pre-dam data was available for Borumba Dam (Yabba Creek) and Maroon Dam (Burnett Creek) ................................... 32

Table 2.4: Membership of IQQM nodes in reference hydrologic classes as

determined by classification of reference flow metrics using model-based clustering. Continental flow class membership (Kennard et al. 2010a) is shown in brackets; ‘?’ indicates assumed continental flow class based on adjacent IQQM nodes....................................................................................................... 38

Table 2.5: Membership of stream gauges in historic hydrologic classes as

determined by classification of historic flow metrics using model-based clustering. Gauges directly influenced by dams and weirs shown in bold text. Continental flow class membership (Kennard et al. 2010a) shown in brackets; ‘?’ indicates assumed continental flow class membership based on adjacent gauges ............................................................................................................... 40

Table 2.6: Qualitative comparison of reference and historic flow class

membership of south-east Queensland study sites ......................................................... 43

Table 3.1: Pre-development and historic reference reaches for flow-regulated

reaches in the study area, as determined from the site selection principles and criteria. Each regulated reach has a geographically close and geographically distant reference site ................................................................................ 60

Table 3.2: Details of study sites. Sites 29–42 were added in 2009 to provide

coverage of reference and historic hydrologic classes poorly represented through other criteria ........................................................................................................ 62

Table 3.3: Landscape-scale environmental variables ............................................................. 65

Table 3.4: Primary categories used to describe catchment land uses based on

the Australian land use and management classification (version 6) ................................ 66

Table 4.1: Hydrologic metrics used to assess riparian vegetation patterns ............................ 73

Table 4.2: Riparian vegetation metrics .................................................................................... 75

Table 4.3: Regulated sites and corresponding reference sites for selected

reference hydrologic classes (RFC) and historic hydrologic classes (HFC) used to examine influence of hydrologic alteration on riparian vegetation patterns ............................................................................................................................. 76

Table 5.1: Environmental variables, including site-scale hydrologic metrics, used

in aquatic vegetation component ...................................................................................... 91

Table 5.2: Details of sites surveyed for aquatic vegetation ..................................................... 93

Table 5.3: Metrics representing aquatic vegetation. Density metrics are richness

metrics standardised by site area. SUB, ATE, ATI, ARP , ARF, TDA and TDR are functional groups described by Brock and Casanova (1997) and defined below. See science report for allocation of plant taxa to functional groups ............................................................................................................................... 96

Table 6.1: Hydrologic metrics included in multivariate analysis of fish

assemblage structure in relation to environmental factors in the study area ................. 114

Table 6.2: Metrics used to test fish species and assemblage-level differences

between hydrologic classes and between regulated and unregulated sites .................. 117

Table 6.3: Species, native/alien status (*alien), age, density per 10 m, frequency

of occurrence and proportion of total catch in streams of south-east Queensland over three sampling periods 2009–2010. Families and common names are given in Appendix 3 to the scientific report ................................................... 119

Table 6.4: Significance levels (p) and proportion of variance explained by the

reference hydrologic class x regulation interaction in both reference and


v

historic hydrologic classes. Pairwise tests are shown to indicate where significant differences between regulated and unregulated sites in each flow regime class were detected. ND indicates no significant difference (p>0.05). For univariate response variables, the mean percentage difference between unregulated and regulated (supplemented) sites is given to indicate level of change. VC = variance component ................................................................................ 128

Table 6.5: Summary of general least squares regression models describing

relationships between fish metrics and selected hydrologic metrics. D is the value of the log-likelihood ratio test comparing model fits (i.e. the null model compared to the alternate model). Significance for D is determined by 2 comparison with Χ with one degree of freedom (3.841). Significance: * 0.01 ≥ p > 0.001; **p ≤ 0.001 .................................................................................................. 131

Table 7.1: Summary of flow metrics correlated with the structure of riparian tree

and shrub communities, aquatic vegetation assemblage composition and fish assemblage structure (based on catch per unit effort) ............................................ 153

Figures Figure E1: The ELOHA framework ............................................................................................ 2

Figure E2: Study area and locations of field sites (reaches proximal to gauges

and/or IQQM nodes) ........................................................................................................... 3

Figure E3: ELOHA plot depicting the degree of change in non-migratory fish

species richness in relation to percentage change in the CV in mean daily discharge .......................................................................................................................... 13

Figure 1.1: The ELOHA framework ......................................................................................... 21

Figure 1.2: Location and principal river catchments of the study area. Plots show

maximum (filled circle) and minimum (open circle) mean daily temperatures and mean monthly rainfall (filled square) ......................................................................... 23

Figure 1.3: Hydrographs of conditions preceding and during field surveys. Study

period (June 2008 – August 2010) shown by arrows. Note differences in yaxis scales ........................................................................................................................ 25

Figure 2.1: Ordination (nMDS) of IQQM nodes based on reference hydrologic

metrics. (a) Distribution of IQQM nodes in ordination space as shown by reference class (numbers). Vectors show hydrologic metrics significantly correlated with the ordination space. (b) Distance of individual IQQM nodes to the centroids of each flow class (numbered) ................................................................ 34

Figure 2.2: Ordination (nMDS) of stream gauges based on historic hydrologic

metrics. (a) Distribution of stream gauges in ordination space as shown by historic flow class (numbers). Vectors show hydrologic metrics significantly correlated with the ordination space. (b) Distance of individual stream gauges to the centroids of each hydrologic class (numbered) ......................................... 37

Figure 2.3: Comparison of reference and historic hydrologic regimes using the

Gower metric. Higher values for the Gower metric indicate greater divergence of the historic hydrologic regime from the reference hydrologic regime (maximum Gower metric is 1 for total dissimilarity). Only IQQM nodes with a currently operating stream gauge are shown. Names in bold text indicate gauges where study sites were established ................................................ 46

Figure 2.4: Box and whisker plots of prediction error from the reference random

forest model used to allocate gauges to reference hydrologic classes. (a) Prediction error across historic hydrologic class. (b) Prediction error arranged by river catchment. Noosa and Pine–Caboolture catchments are not shown due to the low number of sites in these catchments ....................................... 47

Figure 2.5: Heat map showing the percentage change in hydrologic metrics

between reference and historic hydrologic regimes, expressed as (historic value – reference value) / reference value. MeanZeroDay is expressed as the difference between reference and historic values due to division by zero. Negative values indicate that the reference metric value is higher than the historic metric value and positive values indicate that the historic metric value is greater than the reference metric value (see legend). Yellow and light blue cells indicate a change of 10% or less .............................................................. 48


vi

Figure 2.6: Percentage change in flow metric values for metrics identified by clustvarsel as discriminating between historic hydrologic classes (see Sections 2.3.1 and 2.3.2.). MeanZeroDay is calculated as the difference between historic and reference values due to zero values. A positive difference indicates the historic value is greater than the reference value ...................... 49 Figure 2.7: Heat map showing the percentage change in hydrologic metrics between reference and historic hydrologic regimes, expressed as (historic value – reference value) / reference value. MeanZeroDay is expressed as the difference between reference and historic values due to division by zero. Negative values indicate that the reference metric value is higher than the historic metric value and positive values indicate that the historic metric value is greater than the reference metric value (see legend). Yellow and light blue cells indicate a change of 10% or less. The dendrogram groups gauges downstream of dams with similar flow regime characteristics and was calculated using the Gower metric and hierarchical agglomerative clustering .......................................................................................................................... 52 Figure 2.8: Hydrologic alteration values for flow metrics identified by comparison of pre-dam and post-dam flow regimes using the range of variability approach (RVA) (see Section 2.2.4). Yellow bars indicate the low RVA category (0–33 percentiles), green bars indicate the middle RVA category (34–66 percentiles) and red bars indicate the high RVA category (>66th percentile) ......................................................................................................................... 53 Figure 3.1: Locations of field sites (reaches proximal to gauges and/or IQQM nodes)............................................................................................................................... 61 Figure 3.2: Heat map showing the magnitude of Pearson’s correlation coefficients between landscape variables. Red and pink indicate positive correlations among variables while dark blue and light blue indicate negative correlations among variables. Landscape variable abbreviations are defined in Table 3.3 ....................................................................................................................... 67 Figure 3.3: Principal components analysis plots of landscape-scale environmental variables (left) and study sites (right). Principal components 1 and 2 explained 20.17% and 20.23% of the variance, respectively. Landscape variable abbreviations are defined in Table 3.3............................................. 67 Figure 4.1: Ordination non-metric multidimensional scaling (nMDS) of sites based on bankfull tree and shrub density data (a) Position of sites in ordination space (see Table 3.2 for site codes). Vectors show taxa significantly correlated with the ordination (see science report for species abbreviations) (b) Distance to group centroids for sites in each historic hydrologic class (c) Environmental variables significantly correlated with the ordination (see Table 4.1 for variable abbreviations) (d) Distance to group centroids for sites in each reference hydrologic class...................................................... 78 Figure 4.2 Plots of model fits for significant generalised least squares models showing relationships between variables describing variation in hydrologic metrics (CVDry and CV) and selected near-stream riparian vegetation metrics. Abbreviations for hydrologic variables and riparian metrics are given in Tables 4.1 and 4.2 respectively .......................................................................... 81 Figure 4.3: Plots of model fits for significant generalised least squares models showing significant linear or quadratic relationships between variables describing bankfull flow conditions (BFShear and BFDis) and selected riparian vegetation metrics. Metrics are bankfull metrics unless indicated otherwise (e.g. NS = near-stream metrics). Equation coefficients are given in Tables 6.11 and 6.12. Abbreviations for hydrologic variables and riparian metrics are given in Tables 4.1 and 4.2 respectively ....................................................... 82 Figure 4.4: Box and whisker plots of riparian metrics for individual historic hydrologic classes where Kruskal-Wallis tests showed significant differences in bankfull metric values between historic hydrologic classes. Multiple comparison tests (Tukey’s HSD) are also shown where applicable (Bonferroni-corrected significance for each test is α/10 = 0.005). See Table 4.2 for metric abbreviations .............................................................................................. 83


vii

Figure 4.5: Box and whisker plots of bankfull riparian metrics for individual reference hydrologic classes where Kruskal-Wallis tests showed significant differences in metric values between reference classes. Multiple comparison tests (Tukey’s HSD) (Bonferroni-corrected significance for each test is α/10 = 0.005) did not return any significant differences between classes for any of the metrics. See Table 4.2 for metric abbreviations ......................................................... 84 Figure 5.1: Ordination (nMDS in three dimensions) of sites based on species cover data (logx+1 transformed for x>0) (a) Position of sites in ordination space as represented by historic hydrologic class membership. Vectors show taxa significantly correlated with the ordination (p=0.01) (b) Environmental variables significantly correlated with the ordination (p≤0.01) (c) Distance to group centroids for sites in each historic hydrologic class. See Table 5.1 for environmental variable abbreviations and science report for species abbreviations.................................................................................................. 99 Figure 5.2: Box and whisker plots of in-stream aquatic vegetation metrics for individual historic hydrologic classes. The results of multiple comparison tests (Tukeys HSD) are shown where Kruskal-Wallis tests showed significant differences in vegetation metrics across historic hydrologic classes (Bonferroni-corrected significance p = 0.005). See Table 5.3 for description of aquatic vegetation metrics ....................................................................... 100 Figure 5.3: Plots of model fits for significant generalised least squares models describing relationships between vegetation metrics and selected hydrologic metrics ............................................................................................................................ 102 Figure 5.4: Box and whisker plots of (a) bankfull substrate stability for hydrologic regulated and unregulated sites, (b) median particle size for regulated and unregulated sites, and (c) total cover for regulated and unregulated sites .................... 103 Figure 5.5: Box and whisker plots of ranked Bray-Curtis dissimilarities for comparisons of vegetation composition between regulated and unregulated sites in (a) RFC 1 (n=28) and (b) RFC 5 (n=19) ............................................................ 104 Figure 5.6: Scatterplots showing relationships between the effect of hydrologic change versus the Gower metric (an indicator of overall hydrologic change) for sites downstream of dams in the study area. Two sites were surveyed downstream of each dam and each dam is represented by a pair of points (the average of hydrologic alteration effects for all samples). Site codes: Obi = Obi Obi Creek, Six = Six Mile Creek, Bur = Burnett Creek, Ybb = Yabba Creek, Rey = Reynolds Creek, Nrg = Nerang River ...................................................... 106 Figure 5.7: Scatterplots of change in total cover as predicted by partial least square models (effect of flow regulation) versus percentage change in individual hydrologic metrics, calculated as (historic–reference)/reference). Site codes: Obi = Obi Obi Creek, Six = Six Mile Creek, Bur = Burnett Creek, Ybb = Yabba Creek, Rey = Reynolds Creek, Nrg = Nerang River ................................ 107 Figure 6.1: Relationship between gradients of environmental variation and (a) presence–absence patterns (b) composition of fish assemblage structure at sample time 1 (July–August 2009) ................................................................................. 123 Figure 6.2: Relationship between gradients of environmental variation and (a) presence–absence patterns (b) composition of fish assemblage structure at sample time 2 (October–December 2009) ..................................................................... 124 Figure 6.3: Relationship between gradients of environmental variation and (a) presence–absence patterns (b) composition of fish assemblage structure at sample time 3 (April–May 2010)..................................................................................... 125

Figure 6.4: Plots indicating patterns in the mean densities of long-finned eel (Ang rei), Duboulay’s rainbowfish (Mel dub) and native species richness in relation to mean daily discharge, CV of daily flow and percentage of zeroflow days in the 4 years prior to sampling at each sample point. Clear circles indicate unregulated sites; solid circles indicate regulated sites .................................... 130 Figure 6.5: Plots of model fits for significant generalised least squares (GLS) models describing relationships between fish metrics and selected flow metrics. See Table 6.5 for GLS model summaries ........................................................ 132 Figure 6.6: Scatterplots testing the relationship between overall hydrologic alteration (Gower metric) and densities of long-finned eel (Ang rei),


viii

Duboulay’s rainbowfish (Mel dub) and native fish species richness in south east Queensland rivers. Clear circles indicate unregulated sites; solid circles indicate regulated (supplemented) sites. Probability values (P) indicate the significance of the relationship differing from zero (i.e. a flat line) Fish 2 sample data was standardised to 450 m reach surface area combining all habitat units sampled...................................................................................................... 133 Figure 6.7: Scatterplots showing relationships between the effect of hydrologic alteration versus the Gower metric (an indicator of flow regime change) for sites downstream of dams in the study area. Two sites were surveyed downstream of each dam and each dam is represented by a pair of points (the average of flow regulation effects for all samples). Site codes: Obi = Obi Obi Creek, Six = Six Mile Creek, Bur = Burnett Creek, Ybb = Yabba Creek, Rey = Reynolds Creek, Nrg = Nerang River .................................................................. 134 Figure 6.8: Scatterplots of change in total cover as predicted by partial least squares regression models (effect of hydrologic alteration) versus percentage change in individual hydrologic metrics, calculated as (historic– reference)/reference × 100)............................................................................................ 135 Figure 7.1: Graph of the response of non-migratory fish species richness to % change in the CV of mean daily discharge. Graph shows hypothetical acceptable levels of ecological change, and the stream sites that have unacceptable levels of ecological change. ..................................................................... 159


ix

Abbreviations, acronyms and definitions

ANOSIM

analysis of similarity

ARI

average recurrence interval

CPUE

catch per unit effort

CV

coefficient of variation

ELOHA

ecological limits of hydrologic alteration

GLS

generalised least squares

HFC

historic flow class

IQQM

integrated quantity quality model

Ml

megalitres (1 000 000 litres)

NWC

National Water Commission

NWI

National Water Initiative

PCA

principal components analysis

PERMANOVA

permutational multivariate analysis of variance

PLS

partial least squares projection to latent structures

RFC

reference flow class

Supplemented

In Queensland, water supply from releases of water stored in infrastructure. Equivalent to a regulated water supply (NWI). (from National Water Commission 2011, National Water Planning Report Card 2011, NWC Canberra)

Tukey’s HSD

Tukey’s Honestly Significant Difference (multiple comparison procedure)

Unsupplemented

In Queensland, water supply not involving releases of water stored in infrastructure. Equivalent to an unregulated water supply (NWI). (from National Water Commission 2011, National Water Planning Report Card 2011, NWC Canberra)


x

Acknowledgements

The project team would like to acknowledge and thank the National Water Commission (NWC) for funding and managing this ELOHA field trial. We sincerely thank the following staff at the NWC for their guidance and assistance throughout the project: Anthea Brecknell, Richard Davis, Ginni Glyde, Ralph Ogden, Bronwyn Ray and Emma Richter. The International Water Centre hosted the project and provided numerous contributions to its conduct, financial reporting and management, including development of a communications strategy and forward research plan. Our special thanks go to Mark Pascoe, Fiona Chandler, Sabine Stolle, and staff who assisted with the communications report and publication of the scientific report accompanying this project, entitled Hydro-ecological relationships and thresholds to inform environmental flow management and three attached literature reviews. Research on streams and rivers and their environmental flow requirements is profoundly dependent upon hydrologic records and modelling of flow regimes. We are most grateful to the Queensland Department of Environment and Resource Management for the supply of discharge data, and particularly thank Ray Maynard and Cia Musgrove for facilitating the provision of modelled and gauged discharge data. We also thank Water Accounting for providing modelled discharge data for Obi Obi Creek. James Udy and Sean Gibson are also thanked for provision of discharge data for water storages managed by Queensland Bulk Water Supply Authority (trading as Seqwater). The Lands Office (Department of Environment and Resource Management) provided access to historical photographs for the region. We also thank Erin Peterson, CSIRO, for access to her paper and advice on methods for the assessment of land-use pressures around streams. Several colleagues drew our attention to documents to support four literature reviews. The review of Cooper Creek fish research and management (Appendix 4 of the scientific report accompanying this project) was supported by the former Co-operative Research Centre (CRC) for Freshwater Ecology, Canberra and its successor the eWater CRC. We thank colleagues from the Australian Rivers Institute at Griffith University, the Bureau of Meteorology, the Queensland Department of Environment and Resource Management, and the Murray-Darling Basin Freshwater Research Centre (Northern Basin Laboratory) for field assistance, data on Cooper Creek discharge and valuable discussions. Gaining easy and safe access to stream and river sites is a vital aspect of conducting field research. The support and contributions of land owners who allowed access to their properties around south-east Queensland and provided responses to our land-use survey are gratefully acknowledged. We are indebted to landowners Bob Morrish (Springfield), Angus Emmott (Noonbah), Sandy Kidd (Mayfield), David Smith (Hammond Downs) and George Scott (Tanbar) for access to waterholes on their properties along Cooper Creek and for their hospitality and encouragement. We wish to thank members of the Australian Rivers Institute who assisted with the field and laboratory studies, data analysis, the preparation of milestone reports, financial reporting and general administration throughout the project. We particularly wish to thank Tim Howell, Iris Tsoi, Ben Stewart-Koster, Stephen Balcombe, Deslie Smith, Sally Mather, Alexa Apro and Shoni Pearce. We acknowledge and thank the individuals and management institutions who contributed to guidance of the project by their participation in the Steering Committee: Diana Wood and Jonathan Marshall (Queensland Department of Environment and Resource Management, Brisbane), David Rissik (formerly of Queensland Environmental Protection Agency, now Griffith University, Brisbane), Nick Marsh (formerly CSIRO, Brisbane, now Yorb Pty Ltd, Brisbane) and Mark Kennard (Australian Rivers Institute, Griffith University, Brisbane).


xi

A draft of this report was read by two external reviewers and one member of the project Steering Committee and we thank them sincerely for their comments and helpful suggestions. Griffith University conducts research in accordance with the National Statement on Ethical Conduct in Research Involving Humans. Surveys involving land owners were conducted under human research ethics approval GU Ref No: ENV/36/08/HREC. All fish research work was carried out under Queensland Fisheries Permit PRM00157K and Griffith University Animal Ethics Approval No. ENV/21/08/AEC.


xii

Executive summary

This Waterlines provides a summary of the key findings of the project Hydro-ecological relationships and thresholds to inform environmental flow management and river restoration. It is the first study in Australia to explore the scientific and management implications of the ELOHA (Ecological Limits of Hydrologic Alteration) framework for regional environmental flow assessment. The ELOHA trial was funded by the National Water Commission through the Raising National Water Standards program, hosted and managed by the International Water Centre and undertaken by the Australian Rivers Institute, Griffith University in Brisbane, Queensland. The ELOHA framework is a new approach to informing the regional development of environmental flow guidelines that explicitly takes into account spatial variation in flow regimes as well as the potential influence of other environmental variables such as climate and land-use. In its entirety, the ELOHA framework includes both a biophysical and a social module. Using south-east Queensland as a study region, this project aimed to test the four central concepts of the ELOHA framework’s biophysical module which can be summarised as: 1.

Rivers of a chosen region can be grouped into distinctive flow regime classes on the basis of ecologically relevant flow metrics, such as measures of magnitude, duration, timing, frequency and variability of flows

2.

Ecological characteristics of rivers within each flow regime class will be relatively similar compared to those of other classes. Therefore these flow regime classes represent distinct management units or groups of streams that can be managed in similar ways in terms of environmental flows

3.

Rivers within each flow regime class that are ‘regulated’ (or supplemented) in the same way by dams and other infrastructure will show similar ecological responses to flow regime change

4.

Increasing degrees of flow regime change will have increasing impacts on ecological response variables

In addition to testing the key concepts that underpin the ELOHA framework, this project analysed the full field database to provide new knowledge of relevance to the management of flow regimes and river ecosystems of south-east Queensland. This includes: i) information and guidelines on the relative influence of hydrology and other pressures, for example land-use, on river ecosystems in south-east Queensland and advice on how to manage particular combinations of flow alteration and other pressures so as to achieve healthier rivers ii) information about ecological responses to hydrologic alteration and quantitative relationships to inform the development of environmental flow requirements for the protection and/or restoration of selected ecological assets in rivers of different hydrologic character. Guided by the ELOHA framework, the project comprised three major components: i) an analysis of flow regimes of south-east Queensland river basins including an assessment of hydrologic alteration ii) a synthesis of existing knowledge of ecological responses to patterns of hydrologic variability and alteration across the study region


xiii

iii) a field research program to identify the impacts of hydrologic alteration within the study region on selected ecological assets - riparian vegetation, aquatic vegetation and fish. In doing so, the project aimed to identify linear relationships (or thresholds) of ecological response of these assets to hydrologic alteration. Responses of these assets to a limited suite of flow variables that collectively influence the condition or ecological ‘health’ of sites on each river system, were also determined.

Key findings The key findings of this ELOHA trial of relevance to the management of flow regimes and river ecosystems of south-east Queensland can be synthesised into the following major points: 1. Unregulated and regulated flow regimes vary across south-east Queensland mainly with respect to discharge magnitude: A reference classification of ‘natural’ flow regimes in south-east Queensland has been developed in this project using modelled predevelopment flow data derived from an integrated quantity quality (IQQM) model. Six reference flow classes have been identified that vary primarily with respect to the magnitude of flows and, to a lesser degree, flow variability. All reference flow classes (RFCs) include localities from several catchments but low rainfall western and north western localities tend to group together in one reference flow class (RFC 4) while coastal, eastern sites typically fall into another reference flow class (RFC 5) (see Table 2.4). A historic flow regime classification has also been developed in this project using actual stream gauge data. Five historic flow classes (HFCs) have been identified which also vary predominantly in relation to the magnitude of flows and in how these flows have been changed by the presence of dams (see Table 2.6). 2. Flow regime alteration is widespread across south-east Queensland but the overall degree of change is relatively minor: All streams and rivers in the region exhibit some degree of flow regime alteration due to dams, weirs or land-use. The greatest change is generally apparent downstream of dams for Nerang River, Reynolds Creek, Yabba Creek, Lockyer Creek, Brisbane River and Burnett Creek. Some dams such as Six Mile Creek, have had a minor effect on the overall character of the downstream flow regime. Furthermore, high levels of flow regime alteration are also evident in streams without dams, for example, Running Creek, Mudgeeraba Creek and the South Pine River, possibly due to the impact of extensive land-use change associated with agriculture and urbanisation. The degree of overall flow regime alteration across the study area is relatively minor when expressed by a summary metric (Gower metric) of dissimilarity. The Gower metric is a statistical term calculated to express overall levels of flow regime alteration and has a potential range from 0, indicating no change, to 1, indicating complete change across all the hydrologic variables considered in the calculation. A maximum value of 0.25 was determined for the study region, which suggests a relatively low degree of overall hydrologic alteration based on the project’s analyses of 35 different flow metrics. In some cases, particular characteristics of flow regimes in the study area have changed markedly from reference conditions, especially those metrics describing the magnitude of flows, including the duration of periods of low flow (which increased), mean rates of rise and fall of flows (which also increased or became more variable), mean monthly discharge (which decreased) and annual minimum flows (which also decreased). Some


xiv

aspects of flow frequency, duration, variability and seasonal timing have also changed from reference conditions. 3. Every dam in south-east Queensland has altered downstream flow regimes in a different way: The assessment of hydrologic alteration conducted in this project indicates that every dam in the study region has altered downstream discharge patterns in a different way. These alterations depend on the characteristics of each dam’s flow class, location, storage capacity, water release strategies and downstream water extraction practices. Consequently, each dam in the study region has generated a unique downstream flow regime. This finding prevented confirmation of the ELOHA concept that rivers within each hydrologic class that are regulated in the same way by dams and other infrastructure, will show similar ecological responses to flow regime change. 4. Flow regime alteration due to dams and other factors has had significant impacts on riparian and aquatic vegetation and fish in south-east Queensland. These impacts vary between flow regime classes and downstream of particular dams: The results of this study suggest that riparian vegetation of regulated sites in reference flow class 5, for example, sites on the Nerang River below Hinze Dam, has shifted in structure in response to flow regime alteration, and is now similar to that of low discharge sites in the Mary and Logan river catchments. Strongly regulated sites also have significantly lower riparian species density and basal area (ground covered by the bases of trees) per hectare than unregulated sites across all flow classes in the study region. Furthermore, densities of reeds, rushes and sedges are higher at regulated sites, probably as a result of reductions in high in-channel discharges and flood flows that would normally dislodge these plants. Aquatic vegetation structure differed between regulated and unregulated sites within two reference flow regime classes: reference flow class 1, including Obi Obi Creek downstream of Baroon Pocket Dam, and reference flow class 5, which includes the Nerang River downstream of Hinze Dam. When the effects of local, within-site habitat variation are removed, total cover of submerged aquatic vegetation is higher in regulated sites than unregulated sites. Higher species richness of fish assemblages was found to be associated with regulated sites within the study region probably because low flows below dams were elevated compared to their normally low discharge during dry months. In one regulated site, Six Mile Creek, non-migratory fish species richness was almost double that of unregulated sites within the same historic flow class despite relatively slight overall flow regime change. Densities of Duboulay’s rainbowfish were also significantly higher in regulated sites compared with unregulated sites within reference flow class 2 while densities of Pacific blue-eye were significantly lower in regulated sites than in unregulated reference sites in reference flow class 1. 5. Ecologically important flow metrics in south-east Queensland rivers range across a suite of hydrologic variables encompassing the magnitude, frequency, duration, timing and variability/predictability of low to medium and high discharges: Statistically significant hydro-ecological relationships have been quantified in this project. Relationships have been established between flow and each of the ecological assets explored – riparian vegetation, aquatic vegetation and fish. Many of these relationships have not been quantified previously for the study region. Variability in flows during the dry season was identified as a particularly important influence on riparian vegetation structure, probably because riparian vegetation is more reliant on stream discharge during the dry season when there is less rainfall and moisture levels of bank soils are lower. Furthermore, variation in flows at this time of year may


xv

result in frequent spells during which the streams cease to flow, resulting in a dropping of the local riparian groundwater table. These processes influence which riparian species can persist during dry periods. Bankfull discharge and duration were significantly related to the diversity of near-stream vegetation. High flows that reach the bankfull height along a stream channel, influence soil moisture levels, the vigour of riparian vegetation and its capacity for seed production, seedling growth rates and plant survival. Flow characteristics relating to the discharge required to mobilise the median size of stream substrate particles were identified as most significant for aquatic vegetation, as these high flows influence the periodic removal of plants. This normal process in an unregulated river with periods of natural high flows mobilises stream substrates and prevents the build-up of dominant aquatic plant species, some of which may be alien species. A more diverse and patchy plant assemblage tends to develop on suitable substrates. Diverse plant assemblages offer greater diversity of habitat for invertebrates and fish. Fish assemblage structure was most influenced by the occurrence and duration of zero and low discharges, daily hydrologic variability, high discharges, for example, the number of floods greater than the median discharge, and seasonal patterns of mean monthly discharge. Duration of zero and low discharges and daily flow variability can influence fish habitat availability and fish survival. Some alien species are more tolerant of zero and low flows than native species. Seasonal patterns of monthly discharge and water temperature, drive seasonal patterns of fish breeding. Floods disturb stream substrates and aquatic vegetation and help to generate habitat diversity. They also provide hydrologic connectivity and movement pathways for fish throughout the channel network, allowing individuals at various life history stages, to access suitable habitat and food resources and to encounter reproductively active mates. 6. The ecological importance of discharge patterns is situational and ecological responses to flow regime alteration therefore depends on climate and other catchment variables: Considerable geological and climatic variation is present across the south-east Queensland study area and the effects of environmental factors other than flow, especially climate, were apparent for all of the ecological assets examined in this study. With respect to riparian vegetation, significant climatic, catchment and land-use characteristics were identified including the coldest month mean temperature, catchment relief and upstream geological characteristics, as well as the effects of dryland agriculture and intensive land-uses. For aquatic vegetation, mean substrate particle size emerged as particularly significant with depth, bankfull shear stress, water quality, turbidity, riparian canopy cover as well as dryland agriculture having a significant influence. Fish assemblage patterns were associated with gradients in climatic factors (rainfall and temperature), catchment geology, channel morphology, stream habitat structure and hydrology.

Testing of ELOHA concepts The key findings of this project with respect to testing the central concepts of the ELOHA framework can be summarised as follows:


xvi

1. Rivers of a chosen region can be grouped into distinctive flow regime classes on the basis of ecologically relevant flow metrics, such as measures of magnitude, duration, timing, frequency and variability of flows: Two flow regime classifications have been developed, a reference flow classification describing ‘natural’ conditions and a historic flow classification describing the actual conditions measured at gauging stations. Flow regime classes within each classification are distinguished mainly with respect to aspects of flow magnitude, and to a lesser extent, flow variability. Flow alterations caused by dams have created a new historical flow class (HFC1) with more variable seasonal flow peaks and elevated low flows in normally dry months. 2. Ecological characteristics of rivers within each flow regime class will be relatively similar compared to those of other classes. Therefore these flow regime classes represent distinct management units or groups of streams that can be managed in similar ways in terms of environmental flows: The project found mixed support for the concept that ecological characteristics of rivers within each flow regime class will be relatively similar compared to those of other classes. Fish assemblage structure displayed the greatest correlation with flow regime class, and a range of differences in fish assemblage diversity, density and composition were apparent between reference flow classes. Riparian vegetation structure was a relatively poor predictor of reference flow regime class although significant differences were evident amongst a range of metrics describing bankfull vegetation structure across the classes of both the reference and historic flow classifications. No consistent, statistically significant differences in aquatic vegetation were detected amongst historic flow regime classes, with the exception that amphibious plant species - those with some tolerance of exposure and dessication - were found to dominate sites in one historic flow regime class characterised by high flow variability. 3. Rivers within each flow regime class that are ‘regulated’ in the same way by dams and other infrastructure will show similar ecological responses to flow regime change: Mixed support was found for this concept. Since no two dams in the study area have produced the same types of hydrologic change, ecological effects also vary among sites below dams across the study region. Nevertheless, significant differences in riparian vegetation were apparent between regulated and unregulated sites in reference flow class 5 (containing the Nerang River downstream of Hinze Dam) and for aquatic vegetation within two reference flow classes: reference flow class 5 and flow class 1 (containing Obi Obi Creek downstream of Baroon Pocket Dam). 4. Fish assemblage structure also differed between regulated and unregulated sites within reference flow classes 1 and 2, particularly with respect to densities of Pacific blue-eye and Duboulay’s rainbowfish. Pacific blue-eye and Duboulay’s rainbowfish. Pacific blue-eye were significantly lower in regulated sites in reference flow class 1 and densities of Duboulay’s rainbowfish were significantly higher in regulated sites in reference flow class 2.


xvii

5. Increasing degrees of flow regime change will have increasing impacts on ecological response variables: This concept was tested firstly by relating ecological changes to the overall gradient of hydrologic alteration across the study area (as measured by the Gower dissimilarity metric). There were no significant ecological response gradients associated with the gradient of overall hydrologic alteration. This was due to a number of contributing factors but particularly the relatively gentle gradient of overall hydrologic alteration present across the study region, with a relatively low level of maximum change (0.25 on a scale of 0-1) and the presence of only a few strongly regulated sites. Other factors were that the overall hydrologic alteration gradient did not account for ecological differences among hydrologic classes spread over the gradient. Significant ecological responses to hydrologic alteration within each hydrologic class were established for the ecological assets studied in this project. However they did not form linear or threshold relationships apparently because each dam altered downstream hydrology and ecology in a different way. 6. In the second test of this ELOHA concept, overall gradients of change in individual hydrologic metrics were examined and several pronounced ecological responses to alteration of these metrics were discovered. These relationships can be presented as graphs (see Figure 7.1) that summarise all of the positive and negative ecological changes associated with positive and negative changes in individual hydrologic metrics. These graphs summarise the unique ecological responses to each type of hydrologic alteration downstream of individual dams and weirs. Overall, the findings from these tests support the ELOHA principle that it is necessary to classify the hydrologic regimes of a region and examine ecological responses to each type of hydrologic alteration within each flow class. The finding that each dam has altered hydrologic regimes in a different way and produced different ecological responses supports the third ELOHA concept as described above.

Recommendations For management 

Management and monitoring of flow regimes and river ecosystems across south-east Queensland should take into account regional hydrologic variation, for example, by incorporating the hydrologic classifications developed in this project. For example, monitoring programs might ensure adequate coverage of all flow classes, while assessment of monitoring information might be conducted within and across these classes to ensure monitoring objectives and outcomes are considered with respect to regional variation in hydrology.



Management and monitoring of flow regimes and river ecosystems across south-east Queensland should take landscape context, particularly climatic factors, into account. These factors are significant drivers of ecological patterns, both directly and indirectly, through their influence on patterns of stream discharge, particularly for riparian vegetation and fish assemblages.



Management and monitoring of flow regimes and river ecosystems across south-east Queensland should recognise the significance of local habitat characteristics for flowecology relationships, especially hydraulic characteristics (e.g. sheer stress, substrate particle composition) with respect to the structure of riparian and aquatic vegetation.


xviii



To maintain the health of aquatic ecosystems and minimise the potential for further ecological impacts of hydrologic alteration, management of flow regimes and catchments in south-east Queensland should endeavour to maintain the relatively low current levels of overall hydrologic alteration.



None of the rivers or streams included in this study can be considered to have ‘pristine’ discharge regimes since hydrologic alteration in south-east Queensland has been shown here to be geographically widespread.



Flow regimes in south-east Queensland should be managed with particular emphasis on characteristics identified as having ecological importance, including:





-

for riparian vegetation, the variability of flows during the dry season and bankfull discharge

-

for aquatic vegetation, flows that mobilise the median particle size of stream substrates and the frequency that such flow events occur; and

-

for fish, the occurrence and duration of periods of zero and low flow as well as daily hydrologic variability, the number of high discharge events, for exampl, floods greater than the median discharge, and seasonal patterns of mean monthly discharge.

Different management approaches are required for each dam within the study area since each dam has altered the downstream flow regime in a different way. Dam management should take into account the specific landscape, catchment, hydrologic and ecological characteristics associated with each dam and stream/river. Some specific recommendations for dam management include: -

consideration should be given to increasing the numbers of high in-channel flow events downstream of all dams in the study region to limit the encroachment of the active channel by aquatic vegetation and riparian reeds, rushes and sedges

-

flow management at Six Mile Creek Dam should aim to restore specific flow metrics (e.g. moving averages of the annual minimum 3-90 day flows and their duration) closer to reference conditions since decreases in low flows are associated with relatively significant ecological impacts downstream (e.g. for fish).

The flow-alteration-ecological response relationships established during this study can be presented as ELOHA graphs (see Figure 7.1). These demonstrate the effects of altered stream hydrology by revealing measured relationships between altered flow metrics and particular ecological responses compared to unregulated conditions for each study site. These graphs can be used to guide levels of user-defined ‘acceptable change’ in ecological metrics and their associated flow metrics at each site studied, and therefore can inform environmental flow management in each impounded stream/river.

For policy development 

Hydrologic alteration should be considered as a probable current and future risk to river ecosystem health in south-east Queensland based on the findings of this ELOHA trial. The ecological effects of hydrologic alteration across south-east Queensland are likely to continue, particularly for longer lived riparian vegetation and fish that may still be responding to past as well as present changes in hydrologic regime.


xix



Priority for revisions of environmental flow arrangements should be given to dams that have had relatively strong impacts on flow regimes in the region, since the greatest ecological impacts of hydrologic alteration generally occur in association with moderate to strong flow regulation downstream of dams, particularly on the Nerang River downstream of Hinze Dam.



Metrics describing the condition of riparian and aquatic vegetation, fish assemblages and fish species could be used as indicators of hydrologic alteration impacts in monitoring programs such as the Ecosystem Health Monitoring Program of south east Queensland. They provide strong signals of hydrologic alteration, catchment condition and climate variability across the study area, and may therefore be useful indicators of land-use and climate change as well as hydrologic alteration.



Restoration of flow regimes in south-east Queensland should be used to provide opportunities for further validation of the hydro-ecological relationships identified in this project.

For future research 

Further analyses of the ecological datasets produced by this project are recommended. Considerable knowledge with significant implications for management and policy could be extracted from the datasets developed in this ELOHA trial through further analyses, addressing different questions from those asked in this project. Some key analyses might include: -

identification of threatened or refugial habitats and biotic assemblages in rivers and streams and their riparian zones across south-east Queensland, and development of targets for restoration

-

development of population models for key riparian, aquatic plant and fish species to gain improved understanding of the importance of flow and other factors for life history stages, strategies and recruitment processes

-

modelling of climate change impacts under a range of scenarios for riparian and aquatic vegetation and fish assemblages of south-east Queensland

-

synthesis of the outcomes of further data analyses to support improvements to the ELOHA framework by suggesting ecological metrics that provide deeper insight into the ecological impacts of hydrologic alteration.



Further trials of the ELOHA framework are recommended across different types of aquatic ecosystems and along stronger gradients of hydrologic alteration than those present in the south-east Queensland study region.



Key requirements to conduct such a trial include the availability of a good hydrologic monitoring network and either existing hydrologic models of pre-regulation discharge patterns or project team members skilled in hydrologic modelling. Ecological components considered in future trials may differ from those examined here depending on skills, knowledge and values placed on ecological assets in the selected region. Considerable skills in statistical analyses of ecological datasets are essential as are strong collaborative relationships between researchers, managers and other stakeholders such as land owners and community groups.


xx

Project overview Objectives This project sought to test the key scientific concepts of the biophysical module of the ELOHA framework, summarised in Figure E1, in the south-east Queensland region. The central objective of the ELOHA framework is to inform the development of environmental flow guidelines for a particular region that explicitly take into account spatial variation in flow regimes as well as the potential influence of catchment and environmental variables other than flow (e.g. land use). This is in contrast to the river-by-river approach taken by most existing environmental flow methodologies. The ELOHA method was first proposed by Arthington et al. (2006) and has been developed into an environmental flow framework by Poff et al. (2010). The project did not consider the ‘social’ module illustrated in Figure E1, since the primary objective was to test the scientific basis for the approach within the ‘biophysical’ module specifically. In addition to testing the key concepts of the ELOHA framework, the major scientific objectives of the project were to: 

provide an analysis of the hydrologic regimes of unregulated river basins in south east Queensland



provide a quantitative assessment of how the flow regimes of regulated rivers in south-east Queensland have been altered by water infrastructure and the array of types and degrees of flow regulation (including analyses of flow metrics of relevance to ecological responses to flow alteration)



synthesise existing knowledge of ecological responses to flow regime alteration in selected rivers within the study area



design and conduct a field research program to identify how existing flow regime alterations in the study area have affected the structure and responses of river ecosystem habitats and selected biological components



identify thresholds (if any) or linear relationships of habitat and ecological response to flow regime alteration with an emphasis on responses of riparian vegetation, aquatic vegetation and fish



identify a limited suite of flow variables that together govern the condition or ‘health’ of each river system and thresholds of levels of ecological response to flow regime alteration for the whole suite of flow variables



assess the relative influence of flow regime alteration versus other pressures (e.g. land-use extent and type, riparian degradation, water quality impairment, presence of alien species) on habitat and ecological condition or ‘health’.

The key management objectives of the project were to: 

provide information and guidelines on the relative influence of flow and other pressures on river ecosystems and practical advice on how to manage particular combinations of flow alteration and the other pressures so as to achieve healthier rivers



provide flow-ecological response information and quantitative or rule-based relationships that will help to inform definition of the environmental flow requirements that will protect or restore selected ecological assets of economic and/or societal value in rivers of contrasting flow regime type



show how the findings of this study can be related to rivers and flow regime types beyond the geographic scope of this research project.


1

Figure E1: The ELOHA framework

Scientific process Step 1. Hydrologic foundation Baseline hydrographs Flow data and modelling

Geomorphic subclassification

Hydrologic classification

River type

Step 3. Flow alteration (for each analysis node)

Developed hydrographs

Monitoring

Step 2. River classification (for each analysis node)

Analysis of flow alteration

Measures of flow alteration

Step 4. Flow-ecology relationships Flow-ecology hypotheses for each river type

Ecological data for each analysis node

Flow alteration-ecological response relationships for each river type

Social process Implementation

Environmental flow standards

Acceptable ecological conditions

Societal values and management needs

Adaptive adjustments

Source: Poff et al. (2010)

The south-east Queensland trial of the ELOHA framework set out to test the major tenets of the framework and to develop quantitative relationships between flow variables and three biotic communities of streams in the region—riparian and aquatic vegetation and fish. The project objectives, field methods, analyses and results are described in the following sections of this document, together with a summary of implications for management of environmental flows and monitoring the health of rivers and streams in the study region.


2

Study area The ELOHA trial was conducted in the South Coast, Logan–Albert, Brisbane, Pine– Caboolture, Maroochy, Noosa and Mary river catchments of coastal south-east Queensland, Australia (Figure E2). The region is highly varied topographically with coastal lowlands in the east giving way to foothills and plateaus over 300 metres above sea level to the west, north and south. Figure E2: Study area and locations of field sites (reaches proximal to gauges and/or IQQM nodes)

Mary (Miva)

Teewah

Mary (Fishermans pocket)

Glastonbury Six Mile

# * Mary !( (Degun)

Amamoor

Maroochy Yabba

!(

Mary (Moy)

!(

!(

Obi Obi Eudlo

Stanley Emu

Caboolture

Legend

!( # *

Gauges (regulated) Proposed dams Gauges (unregulated) Coomera

# * Te v

io

t

!(

Reynolds

Nerang

!( 0 5 10

20 Kilometers

Logan Burnett


3

The climate of the region is subhumid and subtropical and is influenced by tropical and temperate weather patterns. Minimum temperature of the coldest month is between –3 and o 18 C, rain occurs in all months, and the maximum temperature of the warmest month o exceeds 22 C. Rainfall patterns are more variable than temperature with a distinct east-west rainfall gradient across the study area. Average annual rainfall ranges from 1400 mm on the coast to 800 mm in the western part of the study area. Most rainfall occurs from January to March, typically associated with thunderstorms. Streams and rivers of the region therefore tend to have late summer – early autumn discharge regimes, with periods of low discharge occurring from August to November.

The highest mean annual runoffs per unit of catchment occur in the Noosa (560.8 Ml.year 1 -2 -1 -2 .km ) and Maroochy catchments (782.8 Ml.year .km ) compared with the Brisbane (82 -1 -2 -1 -2 Ml.year .km ) and Mary (213 Ml.year .km ) catchments; the variation reflecting rainfall gradients. The storage capacity of dams and weirs in the study area is approximately 38% of the mean annual runoff, with the greatest volume of water held in storages in the Brisbane River catchment due to the presence of Wivenhoe and Somerset Dams. With the exception of the Noosa and South Coast catchments, the study area had been drought-declared before the commencement of field surveying (Queensland Government 2007). However, flooding occurred throughout the study period and the magnitude of the floods at some study sites was relatively large in comparison to floods in the five years preceding sampling. By the end of the study (September 2010), only a small area in the south-west of the study region remained drought-declared (Queensland Government 2010). Agriculture (grazing and cropping) is the dominant land use in the region and extensive urbanisation has also occurred, particularly along the coastal corridor including the lower Brisbane River, Maroochy and Gold Coast catchments. Approximately 20% of the study area is National Park or State Forest reserve, but widespread clearing of native vegetation has occurred in association with urbanisation and agriculture. The south-east Queensland Bioregion is characterised by high floral and faunal diversity. Pre-European vegetation of the region was dominated by rainforest, dry open sclerophyll forest in inland areas and wallum heathlands and Melaleuca swamps on coastal flats. Threats to biodiversity in the region include population expansion and associated land-use changes, weeds, feral animals and alien fish species.


4

Project methods

The project followed the major steps outlined in the science module of the ELOHA framework: 1.

development of a ‘hydrologic foundation’ of baseline and current hydrographs for stream and river segments throughout the study area based on hydrologic modelling and stream gauge data

2.

hydrologic (flow) classification of divider streams and rivers of the region into distinctive regime types that are expected to have different ecological characteristics

3.

an assessment of hydrologic alteration between current discharge patterns and baseline or reference regimes

4a. identification of relationships between hydrologic alteration and ecological responses within each hydrologic class based on syntheses of existing knowledge and field studies 4b. identification of a limited suite of hydrologic metrics that explain variation in ecological responses to particular types of hydrologic alteration in each hydrologic class. Steps 1 to 3 were addressed in this project through the collection and analyses of modelled and gauged streamflow data to produce hydrologic classifications and assessment of hydrologic alteration across the study region. Step 4 was approached by designing and undertaking a field research program, supported by literature reviews, and extensive analyses of the resulting datasets. Methods for these major components are summarised in the body of this Waterlines report and in the full scientific report for the study. Both reports can be accessed electronically on the National Water Commission website at www.nwc.gov.au.

Analysis of flow regimes Discharge data for 87 IQQM nodes and 72 stream gauges was obtained from the Queensland Department of Environment and Resource Management. Using this data, hydrologic regimes were described in terms of 35 hydrologic metrics characterising the five key facets of discharge regimes (magnitude, frequency, timing and duration of discharge events, and discharge variability/predictability), selected for their ecological relevance and ease of computation using available software. Two hydrologic classifications were then computed using multivariate clustering statistical techniques: 1. a reference hydrologic classification based on modelled, pre-European data (from IQQM nodes) representing more-or-less ‘natural’, hydrologic conditions 2. a historic hydrologic classification based on actual discharge data recorded by individual stream gauges and therefore influenced by changes to discharge patterns that may have occurred over time as a result of land-use changes, water resource development or unsupplemented extraction. A range of statistical analyses was conducted to identify which hydrologic metrics best discriminated between hydrologic classes as well as to assess the extent of alteration to discharge patterns in the study area. The Gower metric (a measure of dissimilarity between historic and reference hydrologic characteristics) was calculated as an index of overall hydrologic alteration across the study area. Changes of individual hydrologic metrics were also assessed.

Field research program A field research program was designed and conducted to identify relationships between hydrologic characteristics and selected ecological components of south-east Queensland


5

rivers, streams and riparian zones as well as the effects of hydrologic alteration on these ecosystems. Based on the prior knowledge of biotic responses to flow and the expertise of the project team, the ecological components selected for inclusion in this trial of the ELOHA framework were riparian vegetation, aquatic vegetation and fish. These ecological components perform vital ecological roles and can be regarded as ecological ‘assets’ due to their contribution to the overall ecological ‘health’ of streams and rivers. Forty-four individual study sites associated with stream gauges were selected for sampling along 20 streams/rivers that reflected the major hydrologic gradients in south-east Queensland (Figure E2, above). Riparian vegetation surveys were conducted once at all 44 sites, while aquatic vegetation and fish surveys were conducted several times at 40 sites. Two sites in Tinana Creek were too deep to conduct aquatic vegetation and fish surveys and two sites in Teviot Brook at Wyaralong were excluded from analysis when landowners denied access to one of the sites in the study reach. Field methods for each study component differed but followed established methodologies and focused primarily on describing the composition (i.e. species presence–absence) and structure (e.g. relative abundance of taxa) of biotic assemblages. Many ecological metrics (e.g. species richness, abundance of native and alien species, abundance of trait groups) were calculated for each component. Catchment and local environmental variables other than stream discharge were also measured in the field and from a variety of existing datasets (e.g. land-use patterns and climate records). Ecological datasets were analysed using a range of univariate and multivariate statistical techniques to assess ecological variation in each component in relation to hydrologic classes as well as other environmental factors. Particular focus was given to identifying hydrologic metrics with most influence on each ecological component, as well as detecting ecological responses to hydrologic alteration across the entire study area and within hydrologic classes. Specific analyses were conducted to examine ecological relationships with individual hydrologic metrics along gradients of hydrologic variability and hydrologic alteration.

Key findings Hydrology of unregulated rivers in south-east Queensland This project has produced a reference hydrologic classification for river basins in south-east Queensland using modelled pre-development flow data derived from an IQQM. This reference classification allows the characteristics and variability of ‘natural’ hydrologic regimes in the study area to be described. Six classes have been identified in the reference hydrologic classification (RFC), which separate along a gradient of discharge magnitude and, to a lesser degree, discharge variability. All reference hydrologic classes include localities from several catchments, but some geographic trends are also apparent. In particular, low-rainfall western and north-western localities tend to belong to RFC 4 (long periods of low flow and low discharge magnitude per unit catchment area) while coastal, eastern sites typically fall into RFC 5 (high discharge magnitude per unit catchment area). The reference hydrologic classification has some concordance with the Australian continental hydrologic classification of Kennard et al. (2010a), although direct comparison is difficult


6

because only 30.5% of IQQM nodes used in the reference hydrologic classification here have equivalent sites in the continental hydrologic classification. Most IQQM nodes used in the south-east Queensland hydrologic classification with analogues in the continental hydrologic classification belong to continental flow class 7 (intermittent-unpredictable). Two classes in the south-east Queensland reference hydrologic classification align broadly with continental perennial classes, three with intermittent and one with intermittent-unpredictable classes.

Hydrologic alteration in regulated river basins in south-east Queensland A second hydrologic classification was developed to examine changes in the hydrology of localities affected by dams, weirs, abstraction and land-use change, based on historical data recorded at stream gauges across the study area. This historic hydrologic classification (HFC) contains five classes, distinguished primarily by six hydrologic metrics describing discharge magnitude, the timing of high and low spells and discharge variability. As for reference flow classes, discharge magnitude is the main driver of spatial variation in historical (i.e. ‘actual’) flow regimes across the south-east Queensland study area. The reference and historic hydrologic classifications are somewhat comparable in structure. The main similarities are: 

HFC 2 is equivalent to RFC 4, both comprising localities in drier parts of the study area



HFC 5 is equivalent to RFC 5, both comprising localities in small, coastal catchments with high rainfall.

The main differences are: 

fewer flow classes in the historic hydrologic classification (n=5) compared with the reference hydrologic classification (n=6)



the emergence of HFC 1, a perennial flow class comprising gauges influenced by flow regime alteration and one unregulated creek (Teewah Creek) with a relatively high groundwater discharge component



the redistribution of IQQM nodes from RFC 1 mostly into HFC 3 and 4.

The geographic extent of hydrologic alteration across the south-east Queensland region is broad with all streams and rivers exhibiting some degree of hydrologic alteration due to dams, weirs or land use. The greatest change in discharge patterns has generally occurred downstream of dams—for example, in Nerang River, Reynolds Creek, Yabba Creek, Lockyer Creek, Brisbane River and Burnett Creek—but not all dams (e.g. Six Mile Creek Dam) have had a strong effect on overall hydrologic character downstream. Furthermore, some streams, for example, Running Creek, Mudgeeraba Creek and the South Pine River, exhibit high levels of hydrologic alteration in the absence of dams, possibly due to extensive land-use change for agriculture and urbanisation. The overall degree of hydrologic alteration across the study area is relatively minor when all hydrologic metrics are analysed together and expressed by the Gower metric of dissimilarity (i.e. maximum value of the Gower metric is 0.25 on a scale of 0–1). However, certain individual hydrologic metrics have changed markedly from reference conditions. The greatest changes have occurred in hydrologic metrics describing discharge magnitude, especially low-spell duration (increased), mean rates of rise and fall (increased), mean monthly discharge (decreased) and annual minima (decreased). Metrics describing flow frequency, duration, variability and timing also exhibit change from reference conditions.


7

The assessment of hydrologic alteration conducted in this project indicates that every dam in the study area has altered the downstream discharge patterns in a different way depending on the characteristics of each dam’s hydrologic class, location, storage capacity, water release strategies and downstream water abstraction practices. Consequently, each dam in the study area appears to have generated a unique hydrologic regime downstream.

Ecological relationships with reference and historic hydrologic classes The ELOHA framework predicts that ecological characteristics should vary across different hydrologic classes if stream discharge has strong influences on ecological systems. Additionally, the ELOHA framework predicts that regulated and unregulated sites within a particular reference hydrologic class (RFC) should differ ecologically if hydrologic alteration is a primary driver. A corollary of this is that regulated sites should also be more ecologically similar to unregulated sites within the same historic hydrologic class (HFC) if discharge has strong influences on ecological systems. The following section presents the findings for each ecological component.

Riparian vegetation

Sampling riparian vegetation (Photo: C James)

Significant differences were apparent among reference hydrologic classes for bankfull riparian vegetation but not near-stream riparian vegetation. Significant differences in 12 metrics describing bankfull riparian vegetation were detected among reference hydrologic classes and 9 metrics differed significantly between historic hydrologic classes. Significant differences were evident between regulated and unregulated sites within RFC 5 and HFC 2. Regulated sites on the Nerang River below Hinze Dam had particularly different


8

riparian tree and shrub assemblage structure from unregulated sites within the same hydrologic class and were also more similar to unregulated sites of the same historic flow class. These findings suggest that riparian vegetation below Hinze Dam has shifted in structure and is now similar to that of lower discharge sites in the Mary and Logan river catchments. Species density and the basal area of late successional riparian species also differed significantly between strongly regulated and unregulated sites across all flow classes. Higher densities of reeds, rushes and sedges than predicted by models were also detected at regulated sites, possibly as a result of reductions in high in-channel discharges and flood flows. It should be noted that a number of environmental variables other than discharge, particularly those relating to climate, also differed significantly between reference hydrologic classes and were identified as important influences on riparian vegetation composition and structure, confounding the ability to detect discharge effects in isolation.

Aquatic vegetation

Sampling aquatic vegetation across transects (l) and Myriophyllum growing in-stream (r) (Photo: S Mackay)

Broad patterns in aquatic vegetation across the study area were not consistently related to differences in reference and historic hydrologic classes. There was evidence, however, that hydrologic alteration has influenced aquatic vegetation patterns in the study area. Aquatic vegetation assemblage structure differed between regulated and unregulated sites within both RFC 1, including Obi Obi Creek downstream of Baroon Pocket Dam, and RFC 5, which includes the regulated Nerang River downstream of Hinze Dam. It should be noted, however, that hydrology is not the only factor that can result in differences in aquatic vegetation between regulated and unregulated sites within a hydrologic class and the differences observed here may be partially due to the coarser substrates of the Nerang River relative to reference sites in this reference hydrologic class that have sandy substrates.

Fish Fish assemblages differed significantly between reference hydrologic classes with respect to species richness, non-migratory fish species richness, total fish density, native fish density, alien fish density, non-migratory fish density and species density. Significant differences were also detected between regulated and unregulated sites within historic hydrologic classes for several fish assemblage metrics.


9

Higher species richness of fish assemblages was associated with regulated sites with Gower metric values ranging between 0.05 (Logan River) and 0.24 (Nerang River). Non-migratory species richness was almost twice as high in one regulated site (Six Mile Creek regulated by Six Mile Creek Dam) when compared with unregulated sites within that historic hydrologic class even though flow regime change has been relatively slight (0.052 on the multivariate Gower scale).

Sampling fish using multiple-pass backpack electro-fishing equipment, Mary River, Qld (Photos: S Mackay)

Densities of Duboulay’s rainbowfish (Melanotaenia duboulayi) were significantly higher in regulated sites when compared with unregulated sites within RFC 2, while densities of Pacific blue-eye (Pseudomugil signifer) were significantly lower in regulated sites than in unregulated reference sites in RFC 1, with Six Mile Creek again showing a marked response. The mean annual 1-day, 3-day, 7-day and 30-day minima are in the range 50–100% lower than at reference sites, and the low-spell duration is 100% higher below Six Mile Creek Dam. Such large reductions in low flow levels and the huge increase in duration of low flows could bring about the 77% reduction in densities of Pacific blue-eye at this site.

Limiting hydrologic variables This project sought to identify a limited suite of hydrologic variables that together govern the condition or health of each river system as well as threshold levels of ecological response to hydrologic alteration for the whole range of hydrologic metrics considered. Among the three ecological components studied, hydrologic metrics on their own explained relatively low proportions of observed variation in assemblage composition and structure across the study area: 14.08% for riparian tree and shrub assemblages and 4.1% for aquatic vegetation (based on species cover data). For fish, 8.97–20.34% of variation in assemblage structure was explained by short-term hydrologic variables (based on 4 years of antecedent hydrology), which were found to be most important in the first sampling period (July–August 2009). Long-term hydrologic variables (14 years of antecedent hydrology) explained 1.24– 9.43% and were also most important in the July–August 2009 sampling period.


10

Table E1 provides a summary of important hydrologic variables correlating significantly with the composition and structure of the three ecological components considered in this study. These variables have been identified by several types of analysis (ordination techniques, analysis of variance, analysis of similarity, Kruskal-Wallis test, regression). Statistically significant flow-ecology relationships have been developed between individual hydrologic and individual ecological metrics measured across study sites. Most of these quantitative relationships are novel for the south-east Queensland region. Findings from this aspect of the study illustrate that the important features of hydrologic regimes in south-east Queensland, with respect to riparian and aquatic vegetation and fish, range across a suite of hydrologic variables encompassing the magnitude, frequency, duration, timing and variability/predictability of low to medium and high discharges. These facets of flow regimes are known to influence the biota and ecological processes in many streams and rivers (Poff et al. 1997).

Ecological response to gradients of hydrologic alteration A major scientific objective of the field research program of this project was to identify any thresholds or linear relationships of habitat and ecological response to hydrologic alteration among the selected ecological components (riparian vegetation, aquatic vegetation and fish). The ELOHA framework predicts that increasing hydrologic change will result in an increasing degree of ecological change from the reference condition. Table E1: Summary of hydrologic metrics correlated with the structure of riparian tree and shrub communities, aquatic vegetation assemblage composition and fish assemblage structure (based on catch per unit effort) Ecological component

Important hydrologic metrics

Riparian vegetation

   

coefficient of variation in dry season flows (CVDry) mean bankfull shear stress (BFShear) mean bankfull discharge (BFDis) a number of other hydrology variables were also important, particularly coefficient of variation (CV) and mean duration of bankfull discharge (BFDur)

Aquatic vegetation



discharge required to mobilise the median particle size, Q_D50MOVE logarithm of the frequency of discharge events required to mobilise the median particle size (FD50MOVE) number of zero-flow days mean daily flow baseflow index mean daily baseflow minimum daily flow coefficient of variation daily flows (CVDaily) magnitude of 10th percentile flow number of high-flow spells (25th percentile exceedance flow) magnitude of 1-year average recurrence interval constancy of mean monthly flow predictability of mean monthly flow

 Fish

          

Q = discharge of sufficient volume and frequency to mobilise the mean particle size (D50) of stream substrates


11

Response to overall gradient of hydrologic alteration When the overall gradient of hydrologic alteration was considered, this prediction was not supported. No statistically significant gradients of increasing ecological response to the overall gradient of increasing hydrologic change (as measured by the Gower metric) were detected for any of the many ecological metrics considered. There were effects but they did not manifest as linear or threshold relationships. Total aquatic plant cover was significantly influenced by flow regulation and was correlated with the Gower metric; however, this effect decreased with increasing levels of flow regime change, possibly as a result of the variation in individual effects of dams on downstream flow regimes across the study area. Factors contributing to the lack of consistent and statistically significant ecological responses (either increases or decreases) to the overall gradient of hydrologic alteration across the study area are likely to include: 

the occurrence of six different flow regime classes with known ecological differences within the overall gradient of hydrologic change, therefore confounding any consistent ecological response



a relatively gentle gradient of hydrologic change across the study area with a low level of maximum change (i.e. 0.25 on a Gower dissimilarity scale of 0–1) and a low number of strongly regulated sites available for testing the effects of overall hydrologic change



variation across the study area in the way each dam has affected the downstream discharge regime resulting in a mixture of changes in individual hydrologic metrics being embedded in the overall measure of flow regime change



construction or alteration of dams in the study area having occurred 10–50 years ago so that ecological adjustments to altered hydrology may still be occurring, especially among longer-lived riparian species



the effect of other environmental gradients in the study area on biotic assemblages and their confounding influence on patterns of ecological responses to the gradient of hydrologic alteration.

These findings support the ELOHA principle that it is necessary to classify the hydrologic regimes of a region and examine ecological responses to hydrologic alteration within each different hydrologic class. The presence of relatively few strongly regulated sites within each hydrologic class prevented a full exploration of this requirement in south-east Queensland.

Response to individual gradients of hydrologic alteration Hydro-ecological relationships established during this study have been expressed in many statistical formats discussed above. Another way to present the effects of altered stream hydrology is an ELOHA plot (see Figure E3). These plots reveal the measured relationships between altered hydrologic metrics and ecological response compared to the expected condition for each study site. The ELOHA plot in Figure E3 reveals that individual hydrologic metrics may show an increase and/or a decrease relative to the values at reference sites, and the pattern of change may not form a simple gradient. Likewise, the ecological response to change in an individual hydrologic metric may be positive (an increase) or negative (a decrease).


12

Sites on Burnett Creek and Yabba Creek show that the coefficient of variation (CV) mean daily flow increased by 40–80% at some sites, and each change in CV is associated with different ecological responses below individual dams. This example highlights the fact that the same type of hydrologic change can have different ecological implications depending upon the individual dam and its effects on downstream characteristics. While the modelling process undertaken in this study has aimed to remove the effects of other environmental variables, there may be in-stream characteristics that have not been eliminated by modelling. Figure E3: ELOHA plot depicting the degree of change in non-migratory fish species richness in relation to percentage change in the CV in mean daily discharge

The individuality of ecological responses to different types of dam and water management is illustrated by this example. ELOHA plots can be used to guide levels of acceptable change in biological metrics and the associated hydrologic metrics. Sites that exceed the hypothetical ‘benchmark’ levels of ecological change suggested in Figure E3 can be delineated as ‘unacceptable level of ecological change from zero’ and earmarked for review with respect to the relevant flow metric. This process has been used in the benchmarking methodology (Brizga et al. 2002) based on expert opinion and ranking of ecological impacts. ELOHA provides a quantitative methodology to assess the ecological impacts of hydrologic alteration. Considered together with other types of hydro-ecological relationship and knowledge of significant differences between reference study sites and sites downstream from dams, ELOHA plots may help to guide environmental flow decisions.

Influence of other environmental variables The final scientific objective of this project was to assess the relative influence on habitat and ecological condition of flow regime alteration compared with other pressures such as land use (extent and type), riparian degradation, water quality deterioration and the presence of alien species. While stream discharge patterns are recognised as one of the principal influences on stream ecology, many other catchment characteristics that are not directly related to hydrology can also have important influences on stream habitat structure and ecological processes. A key question for this ELOHA field trial, and for environmental flow assessments


13

as a whole, is whether the influences of discharge patterns can be extricated from the influences of other ‘natural’ environmental gradients and gradients of anthropogenic disturbance such as land-use change. Here the influence of other environmental gradients is examined for each ecological component.

Riparian vegetation Climatic and catchment variables alone explained 16% variation in bankfull riparian vegetation structure with important variables including coldest month mean temperature, catchment relief, and upstream geological characteristics. Land-use variables alone explained 5.6% of variation in bankfull riparian vegetation structure, particularly production from dryland agriculture and plantations and intensive uses. Hydrologic metrics alone explained 14.08% of observed variation in bankfull riparian tree and shrub communities. Land management practices at a local scale can have a strong impact upon riparian communities and stream ecosystems through extremely localised activities (e.g. vegetation clearance, selective weed control, riparian replanting, localised riparian grazing and burning) that are unlikely to be reflected in broader-scale land-use data such as that used here. Hence, although distance-weighted land-use metrics were used in this study (c.f. Peterson et al. 2010) that place greater weighting within the analyses on land use close to the survey site, it is highly likely that the role of land use has been underestimated here.

Aquatic vegetation Catchment and land-use variables explained 23% variation in aquatic vegetation structure, with important variables including median particle size, riparian cover, production from irrigated agriculture and plantations, production from dryland agriculture and plantations, depth and bankfull shear stress. Median particle size of substrates is a particularly important determinant of aquatic vegetation structure. Hydrologic metrics alone explained 4.1% of observed variation in aquatic vegetation structure. Local within-site variables of water quality, turbidity, riparian canopy cover and Reynolds number explained a further 6.7% of variation in aquatic vegetation structure.

Fish Around 53–57% of spatial variation in fish assemblage composition (i.e. species presence– absence) was explained by climatic factors, geology and channel morphology as well as 4 year and 10-year historical discharge patterns. Fish distribution patterns are strongly associated with climatic gradients, particularly rainfall and reach-scale air temperature. Hydrologic metrics alone explained 5–6.5% of observed variation in fish assemblage structure in terms of species presence–absence and 1.24–20.34% of variation in terms of relative abundances. Hydrologic variables were most important in the first sampling period (July– August 2009).

Synthesis of key outcomes of the south-east Queensland ELOHA trial Central concepts of ELOHA framework The results of this trial of the ELOHA framework with respect to its central scientific concepts are broadly summarised below.


14

1. Rivers of a chosen region can be grouped into distinctive flow regime classes on the basis of ecologically relevant flow metrics, such as measures of magnitude, duration, timing, frequency and variability of flows Reference and historic hydrologic classifications have been developed in this project and distinctive flow regime classes for ‘natural’ and actual conditions identified that separate predominantly with respect to discharge magnitude characteristics.

2. Ecological characteristics of rivers within each flow regime class will be relatively similar compared to those of other classes. Therefore these flow regime classes represent distinct management units or groups of streams that can be managed in similar ways in terms of environmental flows Support for this concept of the ELOHA framework varied between the ecological components considered as follows: 

Fish assemblage structure exhibited the greatest correlation with hydrologic classes, with differences apparent among reference flow classes across a range of metrics describing diversity, density and composition.



Riparian vegetation assemblage structure was a comparatively poor predictor of reference hydrologic class. However, signficant differences were apparent across both reference and historic hydrologic classifications in many metrics describing bankfull riparian structure.



Consistent and statistically significant differences in aquatic vegetation assemblages among historic hydrologic classes were not evident, although amphibious plant species (e.g. species with some tolerance of desiccation and exposure) dominated sites in hydrologically variable HFC 2.

3. Rivers within each flow regime class that are ‘regulated’ in the same way by dams and other infrastructure will show similar ecological responses to flow regime change Support for this concept of the ELOHA framework was mixed depending on the ecological component (and metrics) and the hydrologic class considered. Furthermore, since no two dams in the study area have produced the same types of hydrologic change, ecological effects also vary among sites below dams. The main effects detected in support of this hypothesis were: 

Riparian tree and shrub assemblages differed significantly between regulated and unregulated sites within RFC 5 (including Nerang River downstream of Hinze Dam).



Aquatic vegetation assemblages differed significantly between regulated and unregulated sites within RFC 5 and RFC 1 (including Obi Obi Creek downstream of Baroon Pocket Dam).



Fish assemblage structure differed significantly between regulated and unregulated sites in RFCs 1 and 2. Densities of Pacific blue-eye were significantly lower in regulated sites than in unregulated reference sites in RFC 1, with Six Mile Creek again showing a marked response. Densities of Duboulay’s rainbowfish were significantly higher in regulated sites when compared with unregulated sites in RFC 2. Other species did not show either type of significant difference.

4. Increasing degrees of flow regime change will have increasing impacts on ecological response variables This concept was not confirmed in this trial of the ELOHA framework due to a number of contributing factors, including:


15



the occurrence of six different hydrologic classes with known ecological differences within the gradient of hydrologic alteration



a relatively gentle gradient of hydrologic alteration across the study area with a relatively low level of maximum change (i.e. 0.25 on a scale of 0–1) and a low number of strongly regulated sites available for testing the effects of overall hydrologic alteration



variation across the study area in the way each dam has affected the downstream discharge regime resulting in a mixture of changes in individual hydrologic metrics being embedded in the overall measure of hydrologic alteration



construction or alteration of dams in the study area having occurred 10–50 years ago so that ecological adjustments to altered discharge regimes may still be occurring, especially among longer-lived riparian species



the effect of other environmental gradients in the study area on biotic assemblages and their confounding influence on patterns of ecological responses to the gradient of hydrologic alteration.

These findings support the ELOHA principle that it is necessary to classify the hydrologic regimes of a region and examine ecological responses to hydrologic alteration within each different hydrologic class. The presence of relatively few strongly regulated sites within each hydrologic class prevented a full exploration of this requirement in south-east Queensland.

Additional findings from ELOHA trial The important features of hydrologic regimes in south-east Queensland, with respect to riparian and aquatic vegetation and fish, range across a suite of hydrologic variables encompassing the magnitude, frequency, duration, timing and variability/predictability of zero, low to medium and high discharges. Statistically significant flow-ecology relationships have been developed between individual hydrologic and individual ecological metrics measured across study sites. Most of these quantitative relationships are novel for the south-east Queensland region (see Table E1 above). A key question for this ELOHA field trial, and for environmental flow assessments as a whole, is whether the influences of discharge patterns can be extricated from the influences of other ‘natural’ environmental gradients and gradients of anthropogenic disturbance such as land use around study sites. In this study, hydrologic metrics alone explained 14.08% of observed variation in bankfull riparian tree and shrub communities and only 4.1% for aquatic vegetation structure. For fish, hydrologic metrics alone explained 5–6.5% of observed variation in assemblage structure (species presence–absence) and 1.24–20.34% of variation in terms of relative abundances of species. Other factors were found to influence ecological communities, including climatic and catchment characteristics (elevation, topography, geology, soils) and broad patterns of land use around study sites. The influence of localised activities (e.g. vegetation clearance, selective weed control, riparian replanting, localised riparian grazing and burning) has likely been underestimated for want of suitable data.

Conclusion The south-east Queensland trial of the ELOHA framework has revealed widespread hydrologic alteration across the study region and documented ecological responses to this alteration. Significant hydrologic variables driving patterns in riparian and aquatic vegetation and fish have been identified along with a range of other important climatic, landscape and


16

habitat factors. This information is directly relevant to the management of environmental flows and monitoring the health of rivers and streams in the study region. Although this trial could not fully test all of the scientific concepts of the ELOHA approach due to the particular characteristics of the study area, (especially the lack of a strong gradient of overall hydrologic alteration and the fact that every dam changes downstream hydrology in a different way), the framework did provide an objective and repeatable methodology for examining ecological relationships to patterns of discharge and responses to hydrologic alteration that take into account spatial variability in hydrologic regimes and other environmental characteristics. Further trials of the ELOHA framework are recommended, particularly those that encompass different types of aquatic ecosystem and stronger gradients of hydrologic variability and change than those present in the south-east Queensland study area. The key requirements to conduct such a trial include the availability of a good hydrologic monitoring network and either existing hydrologic models of pre-regulation discharge patterns or project team members skilled in hydrologic modelling. The ecological components considered in future trials may differ from those examined here and will depend on the skill, knowledge and values placed in ecological assets in the selected region. Considerable skills in statistical analyses of ecological datasets are necessary, since the interpretation of ecological patterns against hydrologic and other environmental variability can be a complex task. Finally, the success of this project was made possible by strong collaborative relationships between the research team, the management teams at the National Water Commission and International Water Centre, and a range of stakeholders.


17

1. Introduction

Environmental flow management must be underpinned by clear relationships between hydrology and ecological responses. These hydro-ecological relationships can be used to predict and communicate ecological responses to changes in hydrologic regime and thereby contribute both to the restoration of ecologically important elements of the natural regime in regulated rivers, as well as the prediction of consequences of future changes that may arise from new water resource developments or climate change. This Waterlines document reports on the project Hydro-ecological relationships and thresholds to inform environmental flow management and river restoration, funded by the National Water Commission, hosted and managed by the International Water Centre and undertaken by researchers at the Australian Rivers Institute, Griffith University. The project represents the first attempt in Australia to explore the scientific implications of using the ecological limits of hydrologic alteration (ELOHA) framework as a way of understanding how changes to flow regimes affect river ecosystems at a regional scale and how to interpret these findings with respect to environmental flows and water management. Using the ELOHA framework, this project sought to provide a literature synthesis of hydroecological relationships in rivers of coastal and inland Queensland and undertake field studies comparing unregulated and regulated rivers to identify thresholds of ecological responses to altered hydrologic regimes. The outcomes of the project are intended to inform environmental flow management in rivers with contrasting hydrologic regimes and human ‘footprints’ such as land use, levels of riparian degradation and water quality impairment, and alien species. This knowledge will contribute to the management of environmental flows in rivers that may be regulated in the future as well as to the restoration of rivers that have been or are still regulated. Information from the project may also contribute to strategies for adaptation to climate change. The field study component of this project assessed the responses of riparian vegetation, aquatic vegetation and fish to hydrologic variability and regime changes below dams in rivers of the region. These ecological components perform vital ecological roles and can be regarded as ecological assets by contributing to the overall ecological health of streams and rivers. Riparian vegetation, aquatic vegetation and fish are sensitive to hydrologic variability and change, and provide useful indicators of stream ecosystem health by responding to the stress of hydrologic change, land use and climate change.

1.1. Background Land-use change, river impoundment, surface water and groundwater abstraction, and artificial transfers within and between basins can all profoundly alter natural hydrologic regimes. Globally, the modification of river hydrology is so pervasive that the approximately 3 45 000 dams above 15 m high are capable of holding back more than 6500 km of water or about 15% of the total annual river runoff globally (Nilsson et al. 2005). Furthermore, increasing numbers of rivers are so deprived of water that they no longer reach the ocean, either permanently or for parts of the year (Postel and Richter 2003). A recent synthesis of threats to the world’s rivers identified the direct degradation and reduction of river and floodplain habitat resulting from impoundments and depletion of river discharge as the clearest threats to biodiversity, with 65% of global river discharge and aquatic habitat under moderate to high threat (Vörösmarty et al. 2010). Given escalating trends in species extinction, human population growth, water use, development pressures


18

and climate change, this new global synthesis predicts that freshwater systems, and societies dependent on them, will remain under threat well into the future. Australia’s freshwater crisis, according to this synthesis, is less serious than in many parts of the world, yet still demands new solutions to the challenges of sharing limited and spatially uneven water resources to achieve productive capacity, environmental and social objectives. The National Water Initiative has made a commitment to identifying overallocated water systems and restoring those systems to environmentally sustainable levels of extraction, with an emphasis on the quality of science underpinning water plans and clear articulation of the environmental outcomes sought. The 2011 Biennial Assessment report, The National Water Initiative – securing Australia’s water future produced by the National Water Commission, concludes that whilst there has been some improvement in planning for environmental objectives, overall there remain inadequacies in the transparency of plan objectives. The findings acknowledge the historic opportunity the development of the Murray-Darling Basin Plan provides to address overallocation, however it also states that jurisdictions remain reluctant to explicitly identify overallocated and overused systems and to fully implement measures to move them to sustainable levels of extraction.

Determining sustainable levels of water extraction What exactly is a sustainable level of water extraction? How much water does a river need, when and how often? Many scientists and water managers have provided answers to this question in the form of environmental flow recommendations for thousands of rivers and floodplain wetlands around the globe. Australian scientists have been at the forefront of recent developments in the field of environmental flow assessment, proposing what is now universally accepted as the way forward; that is, to recognise the importance of the entire hydrologic regime for the structure, functioning and productive capacity of rivers from source to terminus (Arthington et al. 1992, 2010; Walker et al. 1995; Puckridge et al. 1998; Kingsford 2006). With publication of the ‘natural flow regime paradigm’ as a template for river conservation and restoration (Poff et al. 1997), scientists and practitioners have increasingly recognised that the structure and functions of the riverine ecosystem, and many adaptations of its biota, are dictated by patterns of temporal variation in river flows (Lytle and Poff 2004). There is now broad general agreement among scientists and many water managers that to protect freshwater biodiversity and maintain the ecosystem services provided by rivers, natural hydrologic variability should be maintained. The rapid acceptance of the natural flow regime paradigm has been accompanied by an expectation that ecologists can easily provide specific environmental flow prescriptions for riverine ecosystems. Unfortunately, translating general hydro-ecological principles and knowledge into specific management rules for particular river basins and reaches remains a daunting challenge (Arthington et al. 2006). Of the 200 or so methods available, around 70% still focus almost entirely on the habitat requirements of a few species (usually fish of recreational or commercial value), with less than 10% attempting to consider the entire natural hydrologic regime and its ecological correlates (Tharme 2003). Australian river scientists have made major contributions to the development of this small minority of holistic ecosystem methods in collaboration with colleagues from other countries, particularly South Africa (Arthington et al. 2003, 2010; King et al. 2003; King and Brown 2010). From the early foundations of a source-to-terminus ecosystem approach, based upon the natural flow regime paradigm and concepts of river restoration, several frameworks have emerged—the building block methodology, flow restoration methodology, flow events method, FLOWS, the benchmarking methodology, DRIFT (downstream response to imposed flow transformation) and, most recently, the framework termed ELOHA—the subject of this Waterlines report.


19

1.2. The ELOHA framework

The central objective of the ELOHA framework is to develop environmental flow prescriptions for multiple rivers rather than taking a river-by-river approach as most methods do. Arthington et al. (2006) suggest a process that aims to quantify relationships between hydrologic alteration and ecological responses for different types of river systems classified according to their natural hydrologic characteristics (i.e. magnitude, timing, frequency, duration and variability/predictability). Rather than ranking the ecological condition of regulated river sites based on the severity of their alteration from a natural or reference condition (as in the benchmarking and DRIFT methodologies), this process involves determining quantitative relationships between hydrologic alteration and ecological responses via empirical measurements taken along gradients of hydrologic change. As well as improving the quantification of hydro-ecological relationships, Arthington et al. (2006) sought to study how streams of different hydrologic character (e.g. stable predictable rainforest streams versus highly variable arid zone streams, to use an extreme example) might differ in their responses to hydrologic alteration (or to restoration of a regulated hydrologic regime). In so doing, this approach could address the needs of water managers for transferable hydro-ecological relationships and environmental flow guidelines, rather than managing for the uniqueness of each river’s flow regime, the common practice. Within a region, the ecological characteristics of streams within a hydrologic class are expected to be relatively similar compared to the ecological characteristics between different classes; therefore, hydrologic classes may represent distinct ‘management units’ (Arthington et al. 2006). Consequently, by comparing ecological condition along gradients of hydrologic alteration, it may be possible to develop and calibrate ecologically relevant flow standards for each hydrologic class of river. The aim is to develop empirical flow response curves for each natural asset of interest (e.g. habitat, aquatic and riparian vegetation, invertebrates, fish, ecosystem process rates) and each ecologically relevant hydrologic variable defining the stream class (e.g. low discharge; the magnitude, timing and frequency of flood flows; and temporal variability). Attracted by these suggestions, a group of river scientists developed a fully fledged working environmental flow assessment framework now known as ELOHA (Figure 1.1). The ELOHA process consists of a biophysical and a social science module, with the major steps in the science module as follows: 1. hydrologic modelling—to build a ‘hydrologic foundation’ of baseline and current hydrographs for stream and river segments throughout the chosen study region 2. hydrologic classification—to divide streams and rivers of the region into a few distinctive flow regime types that are expected to have different ecological characteristics; these river types can be further subclassified according to important geomorphic features that define hydraulic habitat conditions for biota 3. determination of levels of deviation of current flows from baseline flow for a suitable length of flow records 4. development of hydrologic alteration – ecological response relationships for each river type based on synthesis of existing ecological literature, field studies across gradients of hydrologic alteration and expert knowledge; ideally, a parsimonious suite of hydrologic metrics will emerge that collectively depicts the major facets of the flow regime and explains much of the observed variation in ecological response to particular kinds of hydrologic alteration in each river flow class


20

5. interpretation of these hydro-ecological relationships and thresholds in a consensus context where stakeholders and decision makers explicitly evaluate acceptable risk as a balance between the perceived value of the ecological goals (and ecosystem services), the economic costs involved and the scientific uncertainties in functional relationships between ecological responses and flow alteration (Poff et al. 2010) 6. implementation of ELOHA studies in an adaptive management context, where the objective is to formalise ongoing collection of monitoring data, and targeted field sampling, to test and finetune the hypothesised flow alteration – ecological response relationships. Figure 1.1: The ELOHA framework

Scientific process Step 1. Hydrologic foundation

Step 2. River classification (for each analysis node)

Baseline hydrographs Flow data and modelling

River type

Step 3. Flow alteration (for each analysis node) Analysis of flow alteration

Developed hydrographs

Monitoring

Geomorphic subclassification

Hydrologic classification

Measures of flow alteration

Step 4. Flow-ecology relationships Flow-ecology hypotheses for each river type

Ecological data for each analysis node

Flow alteration-ecological response relationships for each river type

Social process Implementation

Acceptable ecological conditions

Environmental flow standards

Societal values and management needs

Adaptive adjustments

Source: Poff et al. (2010)

The south-east Queensland trial of the ELOHA framework set out to test the major tenets of the framework and to develop quantitative relationships between flow variables and three biotic communities of streams in the region—riparian and aquatic vegetation, and fish. The project objectives, field methods, analyses and results are described in the following sections of this document, followed by a summary of implications for management of environmental flows and monitoring the health of rivers and streams in the study region.

1.3. Project objectives The key scientific objectives of the project were to: 

provide an analysis of the hydrologic regimes of unregulated river basins in south east Queensland



provide a quantitative assessment of how the hydrologic regimes of regulated rivers in south-east Queensland have been altered by water infrastructure and the array of types and degrees of alteration (including analyses of hydrologic metrics of relevance to ecological responses to altered discharge patterns)


21



synthesise existing knowledge of ecological responses to flow regime alteration in selected rivers within the study area



design and conduct a field research program to identify how existing hydrologic alterations in the study area have affected the structure and responses of river ecosystem habitats and selected biological components



identify thresholds (if any) or linear relationships of habitat and ecological response to hydrologic alteration with an emphasis on responses of riparian vegetation, aquatic vegetation and fish



identify a limited suite of hydrologic variables that together govern the condition or ‘health’ of each river system and thresholds of levels of ecological response to hydrologic alteration for the whole suite of flow variables



assess the relative influence of hydrologic alteration versus other pressures (e.g. land-use extent and type, riparian degradation, water quality impairment, presence of alien species) on habitat and ecological condition or ‘health’.

The key management objectives of the project were to: 

provide information and guidelines on the relative influence of hydrologic alteration and other pressures on river ecosystems and practical advice on how to manage particular combinations of hydrologic alteration and the other pressures so as to achieve healthier rivers



provide hydrologic alteration – ecological response information and quantitative or rule-based relationships that will help to inform definition of the environmental flow requirements that will protect or restore selected ecological assets of economic and/or societal value in rivers of contrasting hydrologic type



show how the findings of this study can be related to rivers and flow regime types beyond the geographic scope of this research project.

1.4. Study area This ELOHA trial was conducted in the South Coast, Logan–Albert, Brisbane, Pine– Caboolture, Maroochy, Noosa and Mary river catchments of coastal south-east Queensland, Australia (Figure 1.2). The benefits of this study area to trial the ELOHA framework include that the area has been well investigated by staff of the Australian Rivers Institute, has a relatively dense stream gauging network and a variety of flow regime types. The water resource needs of the region have also been investigated (Moreton Basin Water Resource Plan; Logan–Albert Water Resource Plan; Mary Basin Water Resource Plan) and the results of this project will be of direct relevance to the management of these catchments and particularly to the monitoring and review of water resource plans over the next 5 to 10 years.

Topography The topography of the region is highly varied (Murphy et al. 1976; Beckman et al. 1987; Bridges et al. 1990; Malcolm et al. 1998). The main features are coastal lowlands of varying width, characterised by gently undulating terrain, often less than 30 m in elevation (Murphy et al. 1976; Young and Dillewaard 1999; Loi et al. 1998), that eventually give way to foothills and plateaus over 300 m above sea level to the west, north and south of the study region (Figure 1.2). The Brisbane and Mary river catchments comprise approximately 72% of the 2 total study area of 32 000 km .


22

Climate and hydrology The climate of the region is subhumid and subtropical and is influenced by tropical and temperate weather patterns (Bridges et al. 1990; Pusey et al. 2004). The climate is classified as ‘Cfa’ under the Koeppen-Geiger climate classification system; that is, the minimum o temperature of the coldest month is between –3 and 18 C, rain occurs in all months, and the o maximum temperature of the warmest month exceeds 22 C (Linacre and Hobbs 1977). Average maximum and minimum temperatures do not vary substantially throughout the study o area and differ by around 20 C at any given location. Rainfall patterns are more variable and a distinct east–west rainfall gradient exists across the study area, with average annual rainfall varying from 1400 mm on the coast to 800 mm in the western part of the study area (Bridges et al. 1990; Young and Dillewaard 1999). However, rainfall can be high on western ranges bordering the region. Most rainfall occurs from January to March and is often associated with thunderstorms (Bridges et al. 1990). Streams and rivers of the region generally have late summer/early autumn discharge regimes (Musgrove 2003), with periods of low discharge from August to November (Pusey et al. 2004). Figure 1.2: Location and principal river catchments of the study area. Plots show maximum (filled circle) and minimum (open circle) mean daily temperatures and mean monthly rainfall (filled square)

¯ 

25



 



 

15







   

 5 J

F

M

A

M

J



40



A

S

O

Somerset Dam   







15



 

N

D

 

 

5 F

M

A

M



J

J

  



40 20



0 A

S

O

N

D

Brisbane





   





 









 5 J

F

M

A

M

J

 J









80 

40



20

 A

O

N

D







10



 





  

J

A



250

200

150





100

0 F

0 10 20

M

A

M

40

J

60

S

Logan

50

5 J

B R IS B A N E

0 S





60

 

 

15

Pine

100



Southport                 

120





10

 

t



Beaudesert

Albe r



15

20

80

140

    

25

25

120

60



35

30

140

100

  



J

  



 



10

20

Pacific Ocean

20 0

J



30

G Y M P IE

160

   

25 20

80 60



35 30

140

100 





160

120





10

   



  20

Mary

180

Gympie

Noo sa

     

hy

30

Marooc

35

O

N

South Coast

Queensland

D

80 Kilometers

New South Wales


23

Temperate weather systems that produce winter rain in southern Australia may also produce significant rainfall in the study area from autumn to mid-winter (Bridges et al. 1990; Pusey et al. 2004). As the occurrence and intensity of summer and autumn–winter rainfall is irregular, discharge regimes of rivers and streams in the region are highly variable (Pusey et al. 2004). The incidence of flooding in summer and autumn in south-east Queensland is unpredictable and consequently the coefficient of variation of mean daily discharge is relatively high (Pusey et al. 2004). The Noosa and Maroochy catchments have higher mean annual runoff per unit of catchment -1 -2 -1 -2 area (560.8 and 782.8 Ml.year .km ) than the Brisbane (82 Ml.year .km ) and Mary (213 -1 -2 Ml.year .km ) catchments, reflecting rainfall gradients across the region. The volume of water held in storages varies considerably between catchments. The greatest volume of water is held in the Brisbane River catchment due to the presence of Wivenhoe and Somerset dams. The storage capacity of dams and weirs in the study area is approximately 38% of the mean annual runoff. With the exception of the Noosa and South Coast catchments, the study area had been drought-declared before the commencement of field surveying (Queensland Government 2007). However, flooding occurred throughout the study period and the magnitude of the floods at some study sites was relatively large in comparison to floods in the five years preceding sampling (Figure 1.3). By the end of the study (September 2010) only a small area in the south-west of the study region remained drought-declared (Queensland Government 2010).


24

Figure 1.3: Hydrographs of conditions preceding and during field surveys. Study period (June 2008 – August 2010) shown by arrows. Note differences in y-axis scales 250 200

Burnett Creek at Maroon Dam tailwater (HFC 1)

150 100 50 0 30000

Wide Bay Creek at Brooyar (HFC 2) 25000 20000 15000 10000

Mean daily discharge (Mlday-1)

5000 0 140000 120000

Mary River at Miva (HFC 3)

100000 80000 60000 40000 20000 0 12000

Amamoor Creek at Zachariah Lane (HFC 4) 10000 8000 6000 4000 2000 0 8000 7000

North Maroochy River at Eumundi (HFC 5)

6000 5000 4000 3000 2000 1000 0 2003

2004

2005

2006

2007

2008

2009

2010

2011


25

Land use and vegetation Land use throughout the region reflects topography, soils and distance from the coast. Agriculture is the dominant land use within the region, with approximately 40% of the total area used for grazing and 4% for cropping (QDPI 1993). Extensive urbanisation has also occurred within the region, particularly along the coastal corridor. South-east Queensland is the most intensely populated region of Queensland and has one of the highest rates of population increase in Australia with the population predicted to increase by approximately 40% from 2004 to 2026 (Queensland Government 2004). The lower Brisbane River, Maroochy and Gold Coast catchments are the river catchments most impacted by urbanisation. The extent of urbanisation and agriculture within the region has resulted in widespread clearing of native vegetation (Young and Dillewaard 1999). Approximately 20% of the study area is National Park or State Forest reserve (QDPI 1993) but remnant native vegetation is often located in steeper areas or on soils unsuitable for agriculture (Beckman et al. 1987). The study area is wholly contained within the south-east Queensland Bioregion. This bioregion is characterised by high floral and faunal diversity (Young and Dillewaard 1999). Threats to biodiversity in the region include population expansion and associated land-use changes, weeds, feral animals and alien fish species (Young and Dillewaard 1999; Kennard et al. 2005). Prior to European colonisation the vegetation of the region was dominated by rainforest (especially along the coast and adjacent hinterland at higher elevations), dry open sclerophyll forest characterised by Eucalyptus and Angophora spp. and wallum heathlands and Melaleuca swamps on coastal flats with sandy soils (Coaldrake 1961; Beckman 1967). The composition of the riparian vegetation varies with elevation and hydrology but Callistemon spp. (bottlebrush), Castanospermum australe (black bean), Syzygium floribundum (River Myrtle) and Casuarina spp (river oak) are common riparian tree species within freshwater reaches of the study area (Paton 1971; Arthington et al. 2000). Extensive weed invasion has occurred in some riparian zones impacted by agriculture and urban development (Arthington et al. 2000).


26

2. Hydrologic classification 2.1. Introduction The hydrologic (or flow) regime of a river or stream refers to its temporal or seasonal pattern of discharge (Gordon et al. 2005) and is typically characterised by five major attributes: 1. discharge magnitude 2. frequency of discharge events (i.e. floods and droughts) 3. timing of discharge events 4. duration of discharge events 5. rate of change in discharge (Poff et al. 1997). Each of these attributes can be described by numerous metrics, which can be calculated for a variety of time scales (e.g. days, months or years). Hydrologic classification uses a selected suite of ecologically meaningful hydrologic metrics to group streams and rivers into classes with similar hydrologic characteristics. This chapter presents classifications of natural and existing (historic) flow regimes for river catchments in south-east Queensland within the study area of this trial of the ELOHA framework. The hydrologic classifications presented here represent the first step in the ELOHA framework; that is, the building of the ‘hydrologic foundation’ to underpin analyses of hydroecologic relationships in rivers with natural flow regimes as well as in those with flow regimes altered by dams and weirs.

Existing hydrologic classifications Several hydrologic descriptions and classifications have been undertaken prior to this study that are relevant to rivers in south-east Queensland (Pusey et al. 1993; Mackay 2007). Most notable, however, is the first continental-scale classification of Australian flow regimes by Kennard et al. (2010a) which used 120 flow metrics calculated from daily discharge data for 830 minimally disturbed stream gauges to identify 12 flow regime classes across Australia. Four of these continental-scale classes occurred within south-east Queensland (Table 2.1), the most common of which were class 4 (perennial-unpredictable baseflow), mostly occurring in the Logan–Albert and other southern catchments, and class 7 (intermittent-unpredictable), which was spread throughout south-east Queensland. Table 2.1: Continental-scale flow regime classes occurring in south-east Queensland Flow class

Description

Characteristics

1

Perennial-stable baseflow

High magnitude runoff, high constancy of baseflow and relatively low seasonality in discharge

4

Perennial-unpredictable baseflow

Weak seasonal discharge pattern but with relatively variable daily and annual flows

7

Intermittent-unpredictable

Low constancy and predictability of flows, low number of zero-flow days per year (median approximately 150)

Source: Kennard et al. (2010a)


27

2.1.1.Aims The main objectives of this component of the study were to: 

identify pre-European development (reference) and gauged (historic) hydrologic classes in the study area based on the classification of metrics representing the five major attributes of the hydrologic regime



identify hydrologic metrics that best discriminate between the identified hydrologic classes



characterise changes to hydrologic regimes in the study area, including those that have resulted from the construction and operation of dams and weirs.

Two datasets were used for hydrologic classification: 

a reference dataset—modelled, pre-development flow data derived from an

integrated quantity quality model (IQQM, Simons et al. 1996)



a historic dataset—actual flow data recorded by stream gauges.

The historic dataset represents actual discharge recorded at individual stream gauges and will therefore have been influenced by changes to flows that may have occurred over time as a result of land-use changes, water resource development or unsupplemented extraction. In contrast, the reference dataset represents estimated pre-European flow conditions.

2.2. Methods The major steps undertaken to meet the objectives of the hydrologic classifications were: 

selection of appropriate sources of stream flow data from which to calculate

hydrologic metrics



selection and calculation of appropriate hydrologic metrics by which to characterise hydrologic regimes in the study area



statistical analyses of hydrologic metrics to identify flow classes, assess relationships between these classes and identify which hydrologic metrics best discriminated between flow classes



statistical analyses of hydrologic metrics and flow classes to assess the extent of alteration to flow regimes in the study area.

The methods used in each of these steps are described briefly here. Where more detail is required, the reader is referred to the full scientific report accompanying this Waterlines report.

2.2.1.Selection and preparation of discharge data Discharge data for a total of 87 IQQM nodes and 72 stream gauges were obtained from the Queensland Department of Environment and Resource Management (QDERM). Discharge data for one gauge, Obi Obi Creek at Kidaman (gauge number 138104a) was obtained from Water Quality Accounting (QDERM) as modelled gauged data from a calibration model for the Mary River catchment. Since the duration and continuity of hydrologic records varied between and within the datasets (e.g. due to differences in periods of stream gauge operation), selection of an appropriate length and time period for hydrologic records was required to calculate


28

comparable metrics.. Based on the available data, the period 1975–2000 was selected and a minimum flow record length of 15 years within this period was considered sufficient to calculate metrics representative of the long-term values (Kennard et al. 2010b). To avoid splitting the flood season across consecutive calendar years, hydrologic data were arranged by water year with the month with the lowest mean monthly discharge (October) used as the first month (Gordon et al. 2005). Flow data for sixty gauges fulfilled these criteria. The historic dataset was screened for missing data and gaps were filled in using statistical (i.e. multiple regression) or mathematical (i.e. linear interpolation) methods where appropriate. Where stream gauges had periods of missing records that could not be appropriately filled in, these gauges were excluded from the flow regime classification based on the historic flow dataset. The reference dataset was complete and no statistical manipulation was required.

2.2.2.Selection and calculation of hydrologic metrics Thirty-five hydrologic metrics (Table 2.2) were selected to characterise ecologically relevant facets of the hydrologic regime (i.e. magnitude, frequency, timing and duration of discharge events and rate of change in). These metrics can be calculated by the Indicators of Hydrologic Alteration (IHA) software package (Nature Conservancy 2007). To reduce the total number of metrics, those describing mean monthly discharge for February, April, June, October and December were excluded. Although the metrics calculated by the IHA software package describe key hydrologic characteristics, high flow conditions are not considered to be adequately represented (Olden and Poff 2003). Consequently, two additional high-flow metrics were included in this analysis: median of annual maximum flows and specific mean annual maximum flows. The magnitude of floods with average recurrence intervals (ARIs) of 1, 2 and 10 years were also included as these have potential relevance to the inundation frequency of riparian vegetation, for example. Finally, Colwell’s Indices (Colwell 1974) were included as indicators of flow predictability, constancy and seasonality. Hydrologic metrics were calculated for each of the two datasets (i.e. reference and historic) using the River Analysis Package (RAP, Marsh et al. 2003) and the IHA software package. Metrics relating to discharge magnitude were standardised using upstream catchment area to reduce their influence on hydrologic classifications. All metrics were statistically analysed (using principal components analysis) to identify any redundant metrics that could be excluded from the flow classification of each dataset. This process resulted in three flow metrics (JDMin, JDMax and CVDaily) being excluded from the flow classification of the historic dataset (see Table 2.2). All of the flow metrics were used in the classification of the reference dataset.


29

Table 2.2: Hydrologic metrics used in the classification of south-east Queensland flow regimes Flow regime

Metrics

Abbreviation

RFC

HFC

Mean daily flow (January, March, May, 1 July, September, November)

MDF_Jan, MDF_Mar, etc.

X

X

Annual minima, 1-day mean

MA1dayMin

X

X

Annual minima, 3-day means

MA3dayMin

X

X


MA7dayMin

X

X


MA30dayMin

X

X


MA90dayMin

X

X

Annual maxima, 1-day mean

MA1dayMax

X

X

Annual maxima, 3-day means

MA3dayMax

X

X


MA7dayMax

X

X


MA30dayMax

X

X


MA90dayMax

X

X

Baseflow index (ratio of baseflow to total flow)

BFI

X

X

Mean number of zero-flow days per year

MeanZeroDay

X

X

Magnitude of floods with average recurrence intervals of 1, 2, 10 years

ARI_1yr, ARI_2yr, ARI_10yr

X

X

Specific mean annual maximum 2 discharge

Sp_MeanAnnMax

X

X

Median of annual maximum discharge

MedAnnMax

X

X

Julian date of each annual 1-day maximum discharge

JDMax

X

Julian date of each annual 1-day minimum discharge

JDMin

X

Predictability of mean daily discharge

PREDICT

X

X

Constancy of mean daily discharge

CONSTAN

X

X

Seasonality of mean daily discharge (contingency/predictability)

SEASON

X

X

HSNum

X

X

LSNum

X

X

component Magnitude (23 metrics)

Timing (5 metrics)

3

Frequency

Number of high pulses in each year

and duration

Number of low pulses in each year

(4 metrics)

Duration of high pulses in each year

HSDur

X

X

Duration of low pulses in each year

LSDur

X

X

Rate and

Mean rate of discharge rise

RateRise

X

X

frequency of

Means rate of discharge fall

RateFall

X

X

change

CV of mean daily discharge

CVDaily

X

3

(3 metrics) 1 Mean monthly discharge metrics reduced from 12 to 6 to downweight the influence of magnitude metrics on hydrologic classifications. 2 Calculated as the mean annual maximum flow divided by catchment area. 3 Based on 25th (low-spell threshold) and 75th percentiles (high-spell threshold). Spell independence criteria: high spells = seven days between the peak of each spell; low spells = seven days between spells.


30

2.2.3.Classification Classification of the hydrologic metrics calculated for each dataset was undertaken via a multivariate clustering technique using the Mclust package of the R statistical software platform (Fraley and Raftery 2008; R Development Core Team 2010). This process identified clusters (i.e. flow classes) of stream gauges (historic dataset) or IQQM nodes (reference dataset) within which flow metrics were more similar to those of other gauges or nodes within the cluster than to those belonging to other clusters. Two flow regime classifications were computed; a reference hydrologic classification (based on the reference dataset) and a historic hydrologic classification (based on the historic dataset). Several analyses were also conducted to determine which flow metrics best discriminated between the classes identified in the reference and historic hydrologic classifications, including use of the clustvarsel package of R (Dean and Raftery 2009) as well as random forest models. Relationships between classes were examined using non-metric multidimensional scaling (nMDS) in the vegan package for R (Oksanen et al. 2010). Separate ordinations were calculated for each set of flow metrics (i.e. reference and historic) using the Gower metric (Gower 1971) to generate association matrices from the two datasets. Characterisation of hydrologic alteration. The extent of hydrologic alteration in the study area was examined by comparing the reference and historic flow regimes using the Gower metric (Gower 1971). As a measure of dissimilarity, the Gower metric essentially describes the distance, in n-dimensional space, between samples (in this case stream gauges or IQQM nodes) based on their flow metrics. The maximum Gower metric possible is one, representing total dissimilarity. Here, the Gower metric was calculated for the 48 IQQM nodes (and 35 flow metrics) in the reference dataset that had corresponding stream gauges in the historic dataset. Hydrologic change was further examined by assessing the allocation of individual stream gauges (from the historic dataset) to reference flow classes using a decision tree method (random forest model) that was calculated to determine the relative importance of flow metrics in discriminating between reference classes (see Section 2.2.3). It was assumed that if little or no hydrologic change had occurred, this model would allocate individual gauges to the same reference flow class as their corresponding IQQM nodes. Finally, changes in individual metrics between the reference and historic datasets were investigated using the range of variability approach (RVA) to compare flow records from two distinct time periods, such as pre- and post-dam periods (Richter et al. 1997). Gauged (historic) flow data was used for this analysis as the reference (IQQM) flow data would not have included flow regime changes prior to dam construction. Flow data for a two-year period (one year before and after the construction date of the dam) was excluded from analysis. There was sufficient pre-dam (gauged) flow data to perform this analysis for Baroon, Six Mile Creek, Moogerah and Hinze dams (Table 2.3) once data from the two-year period either side of dam construction was excluded. This analysis compares the frequency of values within categories between pre- and post-dam periods for each flow metric; a low-flow category (67th percentile), to produce a hydrologic alteration factor for each category and each flow metric. A positive hydrologic alteration factor indicates that the frequency of values in the category has increased in the post-dam period, while a negative value indicates a reduction in frequency (Nature Conservancy 2009).


31

Table 2.3: Pre- and post-dam periods of hydrologic record used for the range of variability approach (RVA). Insufficient pre-dam data was available for Borumba Dam (Yabba Creek) and Maroon Dam (Burnett Creek) Dam

Construction

Pre-dam record

Post-dam record

date Six Mile Creek Dam (Six Mile Creek)

1964

1947–1962

1966–2005

Baroon Pocket Dam (Obi Obi Creek)

1989

1979–1987

1991–1999

Moogerah Dam (Reynolds Creek)

1961

1931–1959

1980–2009

1

1956–1974

1991–2009

Hinze Dam (Nerang River) 1 2

1989

2

The dam was first constructed in 1976 (capacity 42 400 ML) but substantially upgraded in 1989. Pre-dam record taken to end in 1974. 65 days of missing record in this interval.

2.3. Results 2.3.1.Reference hydrologic classification Six reference hydrologic classes were identified (Table 2.4). Reference flow class (RFC) 1 comprised 26 nodes representing all river basins in the study area. Approximately one-third of these were from the Mary catchment, including both tributary nodes and nodes from the upper and middle regions of the main channel, while the Noosa and Maroochy catchments were each represented in this reference class by single nodes. The Brisbane and South Coast catchments were each represented by two nodes and the Logan–Albert catchment by all of the Albert River nodes, the Canungra Creek node and one upper Logan River node. All of the Pine–Caboolture catchment nodes were allocated to RFC 1. Reference flow class 2 comprised 17 nodes from the Mary, Brisbane and Logan–Albert catchments including several nodes that were directly influenced by flow regulation by dams, weirs and an interbasin transfer scheme. Reference flow class 3 included five nodes, three from the Logan–Albert catchment and the other two from the Mary and Noosa catchments. Reference flow class 4 comprised 17 nodes, all from the Mary and Brisbane catchments, including all of the main channel nodes of the Brisbane River. Reference flow class 5 consisted of 18 nodes from the Mary, Maroochy, Brisbane and Gold Coast catchments with seven of the nine Maroochy catchment nodes allocated to this class. Reference flow class 6 comprised five nodes from five different catchments with three of these having headwaters on the Maleny plateau and therefore potentially influenced by similar weather patterns. Six hydrologic metrics identified by the clustvarsel technique best discriminated between the six reference flow classes: MDF (mean daily flow)_Mar, MDF_Sep, MA (mean annual) 1dayMin, MA1dayMax, MedAnnMax and MeanZeroDays (see Table 2.2), all of which were associated with discharge magnitude, and discharge minima in particular. Ordination of the reference flow metrics (Figure 2.1) also revealed a prominent gradient relating to discharge magnitude, with RFCs 4 and 6 representing the low-flow extremes of this gradient. IQQM nodes in RFC 4 had relatively low discharge magnitude per unit of catchment area, while those in RFC 6 tended to have high discharge magnitude per unit of catchment area. Hydrologic variability, as indicated by CV (coefficient of variation) Daily, was also associated with this discharge magnitude gradient with RFC 6, as well as Teewah Creek (RFC 3) characterised by low flow variability (Figure 2.1). Reference classes 2 and 4 included nodes with a high mean number of zero-flow days per year (MeanZeroDays, Figure 2.1); approximately 20 and 60 days respectively. A second gradient, representing spell number and duration as well as baseflow, was also apparent (i.e. roughly vertical in the ordination space) but this was relatively minor compared with the discharge magnitude gradient (Figure 2.1). With the exception of RFC 5 all reference


32

flow classes were arranged perpendicular to this gradient. The outlier at the top of this gradient in the ordination space (Figure 2.1), Teewah Creek, is unusual in having a high baseflow index and low numbers of high- and low-flow spells, suggesting a relatively stable flow regime influenced by groundwater (Brizga et al. 2005). In contrast with the clustvarsel technique, the random forest model identified flow metrics describing discharge maxima as the most important in discriminating among the six reference hydrologic classes: these metrics are the moving averages of the 1, 3, 7, 30 and 90 day maximum discharges and the 2-year ARI (see Table 2.2).


33

0.6

Figure 2.1: Ordination (nMDS) of IQQM nodes based on reference hydrologic metrics. (a) Distribution of IQQM nodes in ordination space as shown by reference class (numbers). Vectors show hydrologic metrics significantly correlated with the ordination space. (b) Distance of individual IQQM nodes to the centroids of each flow class (numbered)

3

0.2

6

LSDur

5 12 3 1 5 11 1 1 44444422231 1 1 5 5 3 1 5 4 1 1 2 2 JDMin 4 42222 1 11 5 442 5 5 424 111 5 5 55 2 11 1 5 4422 5 CVDaily 1 2 5 5 MeanZeroDay MedAnnMax 5 SEASON LSNum HSNum

-0.4

-0.2

0.0

NMDS2

0.4

MA1dayMin MA3dayMin MA7dayMin MA30dayMin BFI MA90dayMin CONSTAN MDF_Sep HSDurPREDICT

-0.2

0.0

MDF_Jul MDF_May 6 MDF_Nov MDF_Jan MA90dayMax MA30dayMax MDF_Mar MA7dayMax MA3dayMax ARI_10yr ARI_2yr MA1dayMax Sp_MeanAnnMax ARI_1yr 6

6

MRateFall MRateRise 6

0.2

0.4

0.6

0.8

1.0

0.8

1.0

0.2

4

2

1 5

6

-0.2

0.0

3

-0.4

NMDS2

0.4

0.6

NMDS1

-0.2

0.0

0.2

0.4

0.6

NMDS1


34

Since the principal gradient in the reference classification relates to discharge magnitude, it is difficult to align the reference hydrologic classes with the two major continental-scale flow regime types (i.e. perennial and intermittent) identified by Kennard et al. (2010a). Reference flow class 3 can be defined as perennial as only one node (Christmas Creek) ceased to flow and three of the nodes in this class with corresponding stream gauges (Teewah Creek, Christmas Creek and Logan River at Round Mountain) were classified as perennial by Kennard et al. (2010a). Reference flow classes 1 and 6 are defined as rarely intermittent on the basis of the low values for MeanZeroDay (100%


48

-100

400

-100

200

100

50

0

-50

-100 138009 Tinana 142202 South Pine 138903 Tinana 138003 Glastonbury 138113 Kandanga 138107 Six Mile 138010 Wide Bay 146014 Back 138001 Mary 138102 Amamoor 138109 Mary 138111 Mary 141003 Petrie 145102 Albert 146010 Coomera 141001 S Maroochy 145101 Albert 141009 N Maroochy 141006 Mooloolah 145107 Canungra 143010 Emu 145012 Teviot 138104 Obi Obi 142001 Caboolture 141008 Eudlo 138002 Wide Bay 143007 Brisbane 143110 Bremer 138007 Mary 143107 Bremer 138110 Mary 146012 Currumbin 145003 Logan 145020 Logan 143210 Lockyer 143009 Brisbane 145008 Logan 145014 Logan 143303 Stanley 138004 Munna 146020 Mudgereeba 140002 Teewah 145010 Running 143001 Brisbane 143112 Reynolds 146002 Nerang 143035 Brisbane 145099 Burnett 138119 Yabba

Change in ARI_10yr (%)

40

143035 Brisbane 143001 Brisbane 145099 Burnett 141009 N Maroochy 145020 Logan 141006 Mooloolah 138102 Amamoor 146014 Back 138004 Munna 138109 Mary 140002 Teewah 143112 Reynolds 141001 S Maroochy 143110 Bremer 138119 Yabba 146020 Mudgereeba 145003 Logan 141003 Petrie 146010 Coomera 146012 Currumbin 145008 Logan 142001 Caboolture 145107 Canungra 143303 Stanley 143007 Brisbane 138113 Kandanga 138110 Mary 143010 Emu 138003 Glastonbury 141008 Eudlo 138903 Tinana 143107 Bremer 145101 Albert 143009 Brisbane 145012 Teviot 138104 Obi Obi 138009 Tinana 138111 Mary 145014 Logan 138107 Six Mile 138010 Wide Bay 142202 South Pine 138001 Mary 138007 Mary 145010 Running 138002 Wide Bay 145102 Albert 146002 Nerang 143210 Lockyer

Change in MA1dayMin

60

143035 Brisbane 143001 Brisbane 143112 Reynolds 145020 Logan 141006 Mooloolah 138109 Mary 140002 Teewah 138102 Amamoor 141009 N Maroochy 145008 Logan 145003 Logan 145099 Burnett 142001 Caboolture 138004 Munna 146014 Back 145107 Canungra 146010 Coomera 141003 Petrie 138110 Mary 143303 Stanley 143007 Brisbane 143009 Brisbane 146020 Mudgereeba 141008 Eudlo 138113 Kandanga 146012 Currumbin 138903 Tinana 145101 Albert 138111 Mary 138104 Obi Obi 143107 Bremer 138007 Mary 145014 Logan 145012 Teviot 138010 Wide Bay 143110 Bremer 138001 Mary 138003 Glastonbury 145010 Running 143010 Emu 145102 Albert 146002 Nerang 138107 Six Mile 138002 Wide Bay 138009 Tinana 142202 South Pine 141001 S Maroochy 138119 Yabba 143210 Lockyer

Change in MA30dayMin

Figure 2.6: Percentage change in flow metric values for metrics identified by clustvarsel as discriminating between historic hydrologic classes (see Sections 2.3.1 and 2.3.2.). MeanZeroDay is calculated as the difference between historic and reference values due to zero values. A positive difference indicates the historic value is greater than the reference value  

20

200 

0 

100 

0   

 



 





-20











-40

-60





   

-80 



 

800

700 

600

500

300 





300 

250

150 








49

200

-50

150

100

50

0

-50

138119 Yabba 143112 Reynolds 145099 Burnett 138003 Glastonbury 138010 Wide Bay 143035 Brisbane 145014 Logan 143107 Bremer 143110 Bremer 141001 S Maroochy 138107 Six Mile 145102 Albert 138001 Mary 138009 Tinana 142202 South Pine 138002 Wide Bay 145101 Albert 145008 Logan 143009 Brisbane 143007 Brisbane 146020 Mudgereeba 145012 Teviot 145020 Logan 145107 Canungra 146014 Back 140002 Teewah 138004 Munna 143303 Stanley 145003 Logan 146012 Currumbin 138110 Mary 138104 Obi Obi 138007 Mary 138109 Mary 141003 Petrie 138113 Kandanga 146010 Coomera 141009 N Maroochy 145010 Running 142001 Caboolture 138111 Mary 143001 Brisbane 141006 Mooloolah 138102 Amamoor 143010 Emu 141008 Eudlo 138903 Tinana 146002 Nerang 143210 Lockyer

Change in HSDur (%)

143210 Lockyer 138119 Yabba 138002 Wide Bay 142202 South Pine 145012 Teviot 143110 Bremer 143010 Emu 143107 Bremer 145102 Albert 138107 Six Mile 143009 Brisbane 138111 Mary 138104 Obi Obi 146002 Nerang 138010 Wide Bay 138110 Mary 138001 Mary 143035 Brisbane 138903 Tinana 146020 Mudgereeba 138009 Tinana 145008 Logan 138007 Mary 145014 Logan 143001 Brisbane 140002 Teewah 138109 Mary 145010 Running 145020 Logan 145107 Canungra 145101 Albert 146012 Currumbin 141003 Petrie 143303 Stanley 146010 Coomera 138113 Kandanga 141006 Mooloolah 145003 Logan 145099 Burnett 138004 Munna 141008 Eudlo 143007 Brisbane 146014 Back 138102 Amamoor 141009 N Maroochy 142001 Caboolture 138003 Glastonbury 143112 Reynolds 141001 S Maroochy

-100

143035 Brisbane 143001 Brisbane 145020 Logan 143112 Reynolds 146002 Nerang 143210 Lockyer 143110 Bremer 138002 Wide Bay 143010 Emu 145012 Teviot 142202 South Pine 138010 Wide Bay 138119 Yabba 143107 Bremer 143009 Brisbane 141008 Eudlo 141001 S Maroochy 143007 Brisbane 138009 Tinana 138003 Glastonbury 146020 Mudgereeba 138109 Mary 138113 Kandanga 145099 Burnett 146014 Back 138903 Tinana 141009 N Maroochy 140002 Teewah 138102 Amamoor 145003 Logan 138004 Munna 145008 Logan 146010 Coomera 141006 Mooloolah 138110 Mary 145014 Logan 141003 Petrie 142001 Caboolture 146012 Currumbin 138104 Obi Obi 145010 Running 145107 Canungra 138107 Six Mile 143303 Stanley 138007 Mary 138111 Mary 145101 Albert 138001 Mary 145102 Albert

Change in CONSTAN (%)

Change in MeanZeroDay

Figure 2.6 (continued) 150 

100 

150 

50 

0 

50 

  











0 







 













-50 





 







250 

 

100



300

250  

200






50

50

0

-50

-150 138119 Yabba 143112 Reynolds 145099 Burnett 140002 Teewah 143001 Brisbane 143035 Brisbane 143303 Stanley 145020 Logan 141006 Mooloolah 141008 Eudlo 138104 Obi Obi 145107 Canungra 145008 Logan 143107 Bremer 145003 Logan 146012 Currumbin 138004 Munna 146014 Back 142001 Caboolture 138110 Mary 138109 Mary 141009 N Maroochy 145101 Albert 141003 Petrie 138002 Wide Bay 145010 Running 138007 Mary 138102 Amamoor 146010 Coomera 143110 Bremer 138903 Tinana 146020 Mudgereeba 145012 Teviot 138107 Six Mile 138111 Mary 138113 Kandanga 145014 Logan 138010 Wide Bay 143009 Brisbane 143007 Brisbane 138001 Mary 145102 Albert 141001 S Maroochy 138003 Glastonbury 143010 Emu 142202 South Pine 138009 Tinana 146002 Nerang 143210 Lockyer

Change in CVDaily (%)

Figure 2.6 (continued) 100

         

-100

 


51

Figure 2.7: Heat map showing the percentage change in hydrologic metrics between reference and historic hydrologic regimes, expressed as (historic value – reference value) / reference value. MeanZeroDay is expressed as the difference between reference and historic values due to division by zero. Negative values indicate that the reference metric value is higher than the historic metric value and positive values indicate that the historic metric value is greater than the reference metric value (see legend). Yellow and light blue cells indicate a change of 10% or less. The dendrogram groups gauges downstream of dams with similar flow regime characteristics and was calculated using the Gower metric and hierarchical agglomerative clustering

143035 Brisbane

143001 Brisbane

143112 Reynolds

145099 Burnett

138119 Yabba

146002 Nerang

138107 Six Mile

138104 Obi Obi

JDMax JDMin SEASON CONSTAN PREDICT HSDur LSDur HSNum LSNum CVDaily MRateFall MRateRise Sp_MeanAnnMax MeanZeroDay BFI MA90dayMax MA30dayMax MA7dayMax MA3dayMax MA1dayMax MA90dayMin MA30dayMin MA7dayMin MA3dayMin MA1dayMin MedAnnMax ARI_10yr ARI_2yr ARI_1yr MDF_Nov MDF_Sep MDF_Jul MDF_May MDF_Mar MDF_Jan

Percentage change –100% to –50% –50% to –10% –10% to 0% 0% to 10% 10% to 50% 50% to 100% >100%


52

Figure 2.8: Hydrologic alteration values for flow metrics identified by comparison of pre-dam and post-dam flow regimes using the range of variability approach (RVA) (see Section 2.2.4). Yellow bars indicate the low RVA category (0–33 percentiles), green bars indicate the middle RVA category (34–66 percentiles) and red bars indicate the high RVA category (>66th percentile)


53

Figure 2.8 (continued)


54

2.4. Discussion

The flow classifications undertaken in this project for the south-east Queensland study area identified six reference hydrologic classes and five historic hydrologic classes. The main determinant of spatial flow regime patterns in the study area is discharge magnitude. The reference and historic flow classifications are similar in that they have several comparable hydrologic classes, including a hydrologic class in each comprising mostly Maroochy catchment sites (RFC 5 and HFC 5) and a class in each comprising sites in the drier part of the study area including the Brisbane River and tributaries upstream of Wivenhoe Dam, Munna Creek and Wide Bay Creek. The major differences between the reference and historic flow classifications are the loss of two reference flow classes (RFCs 3 and 6), the redistribution of RFC 1 nodes mostly into HFCs 3 and 4, and the creation of a perennial historic flow class comprising gauges influenced by flow regime alteration and one unregulated creek (Teewah Creek) with a relatively high groundwater component to discharge (HFC 1). Human influence on the surface water resources of the south-east Queensland study area is spatially extensive and all major river catchments in the region have at least one dam. The greatest flow regime changes have mostly occurred downstream of dams, although the presence of dams does not necessarily imply extensive flow regime change (e.g. Six Mile Creek has changed relatively little). In some cases relatively significant flow regime change has occurred at gauges that are not directly influenced by dams or weirs; for example, Running Creek, Mudgeeraba Creek and the South Pine River. This indicates that unsupplemented extraction, land-use change and other undetermined factors are likely to have played a significant role in altering the hydrologic regimes of many waterways in the study area. While the maximum levels of overall hydrologic alteration below dams in the study area appear to be relatively low (0.25 on the Gower dissimilarity scale of 0–1), ecological changes may still have occurred as a result of significant changes in particular flow metrics downstream of some of the dams. Even a 20% change in some hydrologic metrics may have a significant impact on certain ecological indicators and processes. The hydrologic metrics that have changed most markedly in the study area include low-spell duration (LSDur), which has increased from reference condition for most stream gauges. This change could represent the effects of dams on downstream flows, or levels of water extraction from impounded and regulated rivers, or increasingly dry conditions over the study period, or all three processes. Rates of rise and fall have also increased substantially when compared to reference values, indicating greater hydrologic variability captured in the gauged flow records. In contrast, the moving averages of the annual 3-day to 90-day minima have declined in value relative to reference conditions, as have mean monthly discharge values. Again, these decreases could reflect the effects of dams on downstream flows, or levels of water extraction from regulated rivers, or increasingly dry conditions over the study period, or all three processes. The results of the analyses presented here indicate that each dam in the study area has altered the flow regime of its host river in a different way. Consequently, there is no replication of each type of hydrologic regime change as required for a complete test of ELOHA predictions. Instead this study has documented a range of different changes that have occurred according to the characteristics of the dam, water release strategies and downstream water abstraction practices. If every individual dam has a different effect on the overall hydrologic regime downstream, ecological impacts might also be expected to differ among regulated sites (an ELOHA principle). Similar ecological responses might still become


55

evident, however, if certain flow metrics have a particularly powerful influence on biological systems. The extent of hydrologic change downstream of dams in the study area is highly dependent upon the height of the dam wall, the storage volume and the management strategy for the dam. Therefore, making generalisations about management strategies may be difficult since each dam appears to have generated a unique hydrologic regime downstream. If every dam has different ecological effects, then it follows that ecological restoration by providing environmental flows will be likely to take a different form. However, this form will depend also on the particular ecological changes downstream, and the overall ecological goals associated with an environmental water allocation.


56

3. Study sites—selection and

environmental variation

3.1. Introduction While stream discharge and hydrologic patterns are recognised as one of the principal orchestrators of stream ecology (Poff et al. 1997; Bunn and Arthington 2002), many other environmental variables and catchment land-use patterns are also important drivers of stream ecological processes. Understanding the influences of hydrology on ecological responses given the inherent underlying variability in environmental and land-use activities across the study region presented significant challenges to the ELOHA trial in south-east Queensland. To develop robust flow-ecology relationships for natural and regulated streams, as the ELOHA framework seeks to do, the possible influence of these other environmental variables must be teased out from the direct influences of discharge patterns.

Environmental variation in the study area There is considerable natural environmental variation across the study area, with complex geology (Ellis 1968; Murphy et al. 1976; Whitaker and Green 1980; Bridges et al. 1990) and associated soils. Distinct topographic regions occur in the region with coastal plains, river floodplains and estuaries in the east, and foothills and mountains with plateaus over 300 metres above sea level to the west, north and south. Overall, the climate is subtropical and dominated by summer rainfall with warm summers and mild winters but the region sits adjacent to the temperate/subtropical transitional zone. A strong rainfall gradient also exists across the study area, with rainfall declining in a westerly (inland) direction across the study area (Bridges et al. 1990). In addition to the natural environmental variation present in the study area, land use and land management practices are also diverse. More than two-thirds of the native vegetation of the south-east Queensland region has been cleared since human settlement began (Catterall and Kingston 1993) and land uses include urban and industrial areas; forestry in native and plantation forests; national parks; and dryland and irrigated production of sugar, dairy, beef, grain, fruit and vegetables. Land uses are not evenly distributed across the catchments with more intensive land uses, for example horticulture and urbanisation, tending to occur in river valleys and on floodplains. Furthermore, localised land management activities such as riparian clearing, riparian grazing, weed control, vegetation replanting and burning may also be undertaken within riparian zones and these can have a direct impact on riparian vegetation communities and stream ecosystems more generally. Anthropogenic activities such as agriculture, urbanisation and water management are also likely to co-vary with natural environmental variation as many natural factors often determine the suitability of sites for such anthropogenic activities (Allan 2004). A key question for the ELOHA trial, and for every environmental flow methodology, is whether the influences of hydrology can be extricated from the influences of landscape-scale environmental variability and anthropogenic disturbances. This separation of influences is necessary to develop generalised flow-ecology response models for regions and for the distinctive river hydrologic classes fundamental to the ELOHA framework (see Section 1.2). This chapter describes the selection of field sites for the fieldwork component of this ELOHA trial as well as an investigation of environmental variation among these sites. In particular, landscape and land-use variables of potential relevance to ecological responses were


57

examined to see how they might co-vary with each other as well as across reference and historic hydrologic classes and in response to flow regime alteration.


select field sites based on the results of the hydrologic classification (Chapter 2) that were suitable for surveying each of the focal biological components (i.e riparian vegetation, aquatic vegetation and fish) in relation to the project’s aims



identify environmental variables of relevance to the project’s aims for which data were available



characterise spatial patterns in environmental variables across the study area, particularly in relation to the hydrologic classifications and hydrologic alteration



identify how landscape and land-use variables that are potentially relevant to

ecological responses co-vary with each other and in relation to the hydrologic

classifications and hydrologic alteration.

3.2. Methods 3.2.1.Selection of study reaches and sites Potential study reaches, that is, topographically homogeneous sections of streams or rivers upstream and downstream of an individual stream gauge, were initially selected based on the following criteria: 

proximity to a currently operating stream gauge so that discharge data was available for the field survey period



accessibility of sites for fieldwork



presence of suitable riparian vegetation (i.e. limited modification, clearing, burning and/or grazing of riparian zone as determined by observation and discussion with landholders)



location of tributaries near gauges



workplace health and safety issues.

Reaches on the main channel of the Brisbane River were not considered since the Brisbane River was the focus of a recent environmental flow study (Arthington et al. 2000). Study reaches were selected downstream of the other major dams in the study area where discharge data from stream gauges was available and access for fieldwork was possible and safe.

Selection of regulated reaches and reference reaches From the potential study reaches identified according to the above criteria, a subset of regulated reaches, that is, reaches close to a stream gauge or IQQM node located downstream of a dam or weir, were identified. Only regulated reaches for which considerable hydrologic alteration was evident (Chapter 2) and for which appropriate non-regulated reference reaches could be found were included (Table 3.1). Non-regulated reference reaches, (i.e. reaches close to a stream gauge or IQQM node not subject to significant hydrologic alteration) were selected for several categories of comparisons:


58



unregulated references of pre-development condition (from IQQM data—

geographically close)



unregulated references of pre-development condition (from IQQM data—not

geographically close; i.e. a different catchment)



unregulated references of developed condition (from stream gauge data—

geographically close)



unregulated references of developed condition (from stream gauge data—not geographically close; i.e. a different catchment)



a replicate regulated reach (from pre-development and developed condition).

Additional criteria for the selection of study reaches were: 

the reference reach(es) was in the same hydrologic class within the classification as the regulated reach. However, where an appropriate reference condition could not be established for a particular flow class due to limited data availability (modelled and gauged) or availability of potential stream reaches within each hydrologic class, reference reaches were selected from the next closest class determined from the reference classification



if the closest reference reach was upstream of a dam (i.e. Burnett Creek and the South Maroochy River), and therefore involved an upstream–downstream comparison, an additional nearby reference reach that was independent of the regulated system was selected



where a reach was used for a number of comparisons (e.g. Amamoor Creek and Burnett Creek upstream of Maroon Dam), an additional replicate reference reach (e.g. Glastonbury Creek and Teviot Brook) was selected to provide greater confidence in the reference condition



an additional reach was selected near the location of the partially completed Wyaralong Dam on Teviot Brook and appropriate references were also selected as ‘before’ baseline data in the event of the construction of these dams. Wyaralong Dam has since been completed.


59

Table 3.1: Pre-development and historic reference reaches for flow-regulated reaches in the study area, as determined from the site selection principles and criteria. Each regulated reach has a geographically close and geographically distant reference site Dam

1

RFC

HFC

Pre-development

Pre-development

Historic

Historic

reference

reference

reference

reference

(geographically

(geographically

(geographically

(geographically

close)

distant)

close)

distant)

Yabba Creek

2

2

Glastonbury Creek

Logan River at Rathdowney

Amamoor Creek 1 (HFC 4)

Teviot Brook at The Overflow

Six Mile Creek

1

4

Amamoor Creek

Coomera River

Amamoor Creek

Coomera River

Obi Obi Creek

1

3

Amamoor Creek

Coomera River

Mary River at Moy Pocket

Burnett Creek upstream of Maroon Dam

Burnett Creek

2

1


Glastonbury Creek

Teviot Brook at Croftby

Glastonbury Creek

Reynolds Creek

2

1


Glastonbury Creek

Teviot Brook at Croftby

Glastonbury Creek

Nerang River

5

3

Currumbin Creek

Stanley River

Burnett Creek upstream of Maroon Dam

Mary River at Moy Pocket

An appropriate historic reference (geographically close) could not be found in HFC 2 for Yabba Creek.


60

Selection of study sites within reaches Within each suitable study reach, two field sites were selected. Sites were located at least two kilometres apart to maximise independence between sites while maintaining reasonable proximity to the nearest stream gauge. Grazed sites were unavoidable in some reaches due to the widespread extent of grazing in some catchments and lack of livestock exclusion from the riparian zone (e.g. Burnett Creek and Teviot Brook). Selection of sites was also limited by the presence of sufficient water depth for sampling aquatic vegetation and fish (i.e. maximum depth of 1–1.5 m). Final sites selected for the study are shown in Figures 3.1 and 3.2 and listed in Table 3.2. Figure 3.1: Locations of field sites (reaches proximal to gauges and/or IQQM nodes)

Mary (Miva)

Teewah

Mary (Fishermans pocket)

Glastonbury Six Mile

# * Mary !( (Degun)

Amamoor

Maroochy Yabba

!(

Mary (Moy)

!(

!(

Obi Obi Eudlo

Stanley Emu

Caboolture

Legend

!( # *

Gauges (regulated) Proposed dams Gauges (unregulated) Coomera

# * Te v

io

t

!(

Reynolds

Nerang

!( 0 5 10

20 Kilometers

Logan Burnett


61

Table 3.2: Details of study sites. Sites 29–42 were added in 2009 to provide coverage of reference and historic hydrologic classes poorly represented through other criteria Site number and name

Latitude

Longitude

Reference

Historic flow

flow class

class

1. Stanley River at Cove Road

-26.9197

152.7725

5

5

2. Burnett Creek downstream of gauge 145018a

-28.2163

152.6138

No class

3

3. Burnett Creek upstream of gauge 145018a

-28.2274

152.6040

No class

3

4. Nerang River at Grand Manor Golf Course

-28.0223

153.3026

5

3

5. Coomera River at Coomera Scouts Hall

-28.0468

153.1899

1

4

6. Nerang River at Weber Court near Chantrill Avenue

-28.0060

153.3139

5

3

-28.1621

152.5583

No class

3

-28.1562

152.5718

No class

3

-26.3461

152.6560

1

4

10. Yabba Creek at Stirling Crossing

-26.4901

152.6275

2

2

11. Yabba Creek at No. 8 Crossing

-26.4983

152.5916

2

2

12. Obi Obi Creek downstream of number 2 crossing

-26.6340

152.7837

1

3

13. Obi Obi Creek upstream of number 2 crossing

-26.6397

152.7901

1

3

14. Mary River downstream of Walker Road

-26.5117

152.7461

1

3

15. Six Mile Creek at Old Noosa Road

-26.3297

152.8092

1

4

16. Six Mile Creek at Grahams Road

-26.3420

152.8641

1

4

17. Glastonbury Creek at Greendale Road Crossing

-26.1835

152.5276

2

3

18. Eudlo Creek at gauge site

-26.6625

153.0181

5

5

19. Eudlo Creek upstream of Bruce Highway

-26.6860

152.9964

5

5

20. Reynolds Creek at Yarramalong campground

-28.0116

152.5565

2

1

21. Reynolds Creek downstream of Purdons Bridge

-28.0007

152.5699

2

1

22. Amamoor Creek at Zachariah Lane

-26.3669

152.6223

1

4

23. Glastonbury Creek 2 km from Mary River confluence

-26.1544

152.5528

2

3

7. Teviot Brook near Brennan Road

1

8. Teviot Brook at Croftby 9. Amamoor Creek at Harrys Creek Road

2


62

Site number and name

Latitude

Longitude

Reference

Historic flow

flow class

class

24. Mary River at Moy Pocket north of quarry

-26.5257

152.7395

1

3

25. Coomera River at Tucker Lane

-28.0561

153.1786

1

4

26. Stanley River at gauge site

-26.8392

152.8403

5

5

27. Burnett Creek 2 km downstream of Maroon Dam

-28.1759

152.6722

2

1

28. Burnett Creek at Splityard Creek Road

-28.1659

152.6814

2

1

29. Currumbin Creek at Currumbin Valley Primary School

-28.2075

153.3953

5

5

30. Currumbin Creek at Fordyce Court

-28.1908

153.4167

5

5

31. Wide Bay Creek downstream of gauge 138002c

-26.0019

152.4286

4

2

32. Wide Bay Creek upstream of gauge 138002c

-26.0047

152.4072

4

2

33. Munna Creek at gauge 138004b

-25.9042

152.3489

4

2

34. Munna Creek downstream of gauge 138004b

-25.9014

152.3500

4

2

35. North Maroochy River at Eumundi

-26.4697

152.9544

5

5

36. North Maroochy River at North Arm – Yandina Creek Road

-26.5231

152.9600

5

5

37. Mary River at Bauple–Woolooga Road

-25.8861

152.4864

3

3

38. Mary River at Orphants Road

-25.9533

152.4956

3

3

39. Tinana Creek at gauge site

-25.8200

152.7222

2

3

40. Tinana Creek at upstream of gauge

-25.8356

152.7229

2

3

41. Logan River at Running Creek Road

-28.2128

152.8739

2

1

42. Logan River at upstream Tilleys Bridge

-28.2233

152.8583

2

1

43. Teviot Brook at Wyaralong a

-27.8969

152.9014

2

2

44. Teviot Brook at Wyaralong b

-27.9061

152.8583

2

2


63

3.2.2.Selection of environmental variables A wide range of landscape-scale environmental variables was selected for analysis in this study based on their ecological relevance, resilience to short-term effects of hydrologic alteration and human activity, and relative independence from each other (Table 3.3). A recently derived national dataset (Stein et al. 2009) was used to source a number of these variables at a reach-scale relating to climate, topography and geology. Other data sources included the Bureau of Meteorology (rainfall), the south-east Queensland region geoscience dataset (QDNRM 2002) and digital 1:100 000 scale geology maps for the region. Several other variables concerning topography and channel morphology were measured in the field using a dumpy and staff and compass (Table 3.3). In addition to the landscape-scale environmental variables listed in Table 3.3, percentages of primary land-use classes for each catchment were calculated from the Queensland land-use mapping program (QLUMP) dataset (Witte et al. 2006). Land-use classes were based on the Australian land use and management classification version 6 (BRS 2002, Table 3.4) and -1 metrics were calculated using an inverse-distance weighting (d+1) metric following Peterson et al. (2010), which gives greater weighting to land uses close to the stream. Land use based on the QLUMP data provides information at a relatively coarse scale (smallest mapped feature is one hectare and minimum width for linear features is 50 metres). Land management practices at a local scale can have a strong impact upon riparian communities and stream ecosystems through extremely localised activities (e.g. selective weed control, riparian replanting, localised riparian grazing and burning) that are unlikely to be reflected in the broader-scale land-use datasets available to this study. Landholder surveys were undertaken in order to ascertain the extent of local management activities specific to our riparian surveyed zones.

3.2.3.Statistical analyses A range of statistical analyses was undertaken to explore relationships between environmental variables and patterns in relation to the selected study sites (see Table 3.2) and hydrologic classifications (see Chapter 2). Analyses included principal components analysis (PCA), to investigate patterns in landscape variables across sites, and the KruskalWallis test, to identify significant differences in environmental variables between reference and historic hydrologic classes. Relationships among landscape-scale environmental variables and land use were explored using Pearson’s correlation coefficient and scatter plots. Potential correlations between land use and hydrologic alteration were explored by comparing changes in the Gower dissimilarity metric (see Section 2.2.4) as well as individual flow metrics, to the proportion of land-use classes in catchments.


64

Table 3.3: Landscape-scale environmental variables Variable

Unit

Abbreviation

Source

Latitude

degrees

Declat

1

Longitude

degrees

Declong

1

Site elevation

m.a.s.l

Elevat

1

Distance to source

km

DtoS

1

Distance to mouth

km

DtoM

1

Channel aspect – north

degrees

C_Asp_N

1

Channel aspect – east

degrees

C_Asp_E

1

Bank aspect – north

degrees

B_Asp_N

1

Bank aspect – east

degrees

B_Asp_E

1

-1

S_Grad

2

-1

B_Slope

1

CAT_AREA

1

Topography and morphology

Stream gradient

m.m

Stream bank gradient

m.m

Catchment area

km

Elongation ratio

no units

CAT_ELON

1

Relief ratio

no units

CAT_RELI

1

Reach valley confinement

%

V_Conf

2

Annual mean rainfall

mm

A_Rainfall

3

Annual mean temperature

o

C

A_Temp

2

Hottest month mean temperature

o

C

HMA_Temp

2

Coldest month mean temperature

o

C

CMA_Temp

2

2

Climate variables

Substrate characteristics % mafic

%

Mafic

% felsic

%

Felsic

4

% sedimentary (% siliclastic and undifferentiated)

%

Sed-Silic

4

% sedimentary (% carbonates)

%

Sed-Carb

4

% mixed sedimentary and igneous

%

Mixed

4

% unconsolidated rocks (alluvium, colluvium etc)

%

Unc_Catch

4

Unconsolidated material for reach

%

Unc_Reach

2

Source data: 1. Measured in this study 2. Stein et al. (2009) 3. Bureau of Meteorology (2009) 4. Queensland Department of Natural Resources and Mines (2002)


65

Table 3.4: Primary categories used to describe catchment land uses based on the Australian land use and management classification (version 6) Primary land-use class

Acronym

Subcategories

Production from relatively natural environments

PNE

Grazing natural vegetation, production forestry

Production from dryland agriculture and plantations

PDA

Cropping, grazing modified pastures, horticulture, plantation forestry

Production from irrigated agriculture and plantations

PIA

Irrigated cropping, irrigated modified pastures, irrigated horticulture, irrigated plantation forestry

Conservation and natural environments

CAN

Conservation areas (national parks, nature reserves), other protected resources

Intensive uses

IU

Residential, industrial, transport and utilities, intensive horticulture, intensive animal production

3.3. Results 3.3.1.Patterns in landscape-scale environmental variables Weak correlations were found between the majority of landscape-scale environmental variables (Figure 3.2). Where stronger correlations were apparent, these tended to occur between variables within the same broad category; that is, climate, topography or geology. The PCA of the landscape-scale environmental factors revealed three significant principal components, together explaining 51% of the variance in the dataset (Figure 3.3). The first principal component (PC1) explained 20.17% of the variation and was positively associated with one climatic variable (HMA_Temp = hottest month mean temperature), two topographic variables (CAT_AREA = catchment area, DtoS = distance to source) and two geological variables (Sed_Carb = sedimentary rocks (carbonates) and Mixed sediments) and negatively associated with one climatic variable (A_Rainfall = mean annual rainfall) and one topographic variable (CAT_RELI = catchment relief). The second principal component (PC2) explained 20.23% of the variance and was positively associated with two climatic variables (A_Temp = mean annual temperature and CMA_Temp = coldest mean monthly temperature) and one geological variable (Unc_Catch = unconsolidated sediments) and negatively associated with three topographic variables (CAT_RELI, S_Slope = steam bank gradient and Elevat = elevation). PC3 explained a further 10.6% of the variance and was positively associated with one geological variable (Igneous) and negatively associated with the geological variable Sed_Silic (siliclastic and undifferentiated sediments). Additional principal components explained less than 10% of the total variance.


66

Figure 3.2: Heat map showing the magnitude of Pearson’s correlation coefficients between landscape variables. Red and pink indicate positive correlations among variables while dark blue and light blue indicate negative correlations among variables. Landscape variable abbreviations are defined in Table 3.3

Figure 3.3: Principal components analysis plots of landscape-scale environmental variables (left) and study sites (right). Principal components 1 and 2 explained 20.17% and 20.23% of the variance, respectively. Landscape variable abbreviations are defined in Table 3.3

0.2

CMA_Temp

36 1918 35

A_Temp

37

4039

38

0.1

1

Unc_Catch

16

C_Asp_N

A_rainfall

32 31

46 1 15

30 29 0

23 24 149 17 44 43 22 10

PC2

41 42 20 21 28

-1

0.0

26 25 13 5 12

-0.1

Cat_Reli V_Conf S_slope DtoM 2

Elevat -0.2

-0.1

0.0

PC1

11

8 7 27

-2 -0.2

PC2

Unc_Reach Sed_carb Cat_Area DtoS C_Asp_E B_Asp_N Mixed B_Asp_E Igenous Cat_Elon B_slope Sed_silic HMA_Temp

34 33

3 0.1

0.2

-2

-1

0

1

2

PC1


67

Broad geographic and climatic trends across the study area, as indicated by PCA results (Figure 3.3), included: 

higher hottest month mean temperature (HMA_Temp) and catchment area (CAT_AREA) in many of the most northerly sites (e.g. Mary River at Miva, Yabba Creek, Wide Bay Creek and Munna Creek)



higher mean annual temperture (A_Temp) and coldest month mean temperature (CMA_Temp) in more northerly sites (North Maroochy River, Eudlo Creek, Tinnana Creek and Mary River at Miva)



higher mean annual rainfall (A_Rainfall) and catchment relief ratio (CAT_RELI) in coastal streams (e.g. Currumbin Creek, Eudlo Creek and North Maroochy River)



higher elevation in sites in the south-west of the study region (e.g. Burnett Creek).

Differences between hydrologic classes Significant differences in landscape-scale environmental variables were also detected (via the Kruskal-Wallis test) between hydrologic classes within each flow classification. These included several relating to topography: 

significantly higher distance to source (DtoS) in historic hydrologic class (HFC) 2 compared with HFCs 3 and 5



smaller catchment area in HFC 5 compared with HFCs 1 and 2



greater catchment area in reference hydrologic class (RFC) 4 compared with RFC 5



differences in channel aspect (northerly) betweeen RFCs 2 and 4.

No differences in elevation were detected among reference or historic hydrologic classes. Climatic landscape-scale environmental variables also exhibited several significant differences between hydrologic classes, including: 

higher annual rainfall in HFC 5 compared with HFC 2



higher annual rainfall in RFC 5 compared with RFCs 2 and 4



lower hottest month mean temperatures in RFC 5 compared with RFCs 2 and 4



lower coldest month mean temperatures in RFC 2 compared with RFC 5.

Pair-wise comparisons revealed only a single significant difference between hydrologic classes among the geological variables: 

lower proportion of unconsolidated materials in HFC 4 compared with HFC 5.

3.3.2.Patterns in land uses The dominant land-use category in the study area was ‘production from relatively natural environments’ (see Table 3.4), accounting for 58% of the total land uses, with ‘conservation and natural environments’ comprising a further 26% (total of 84%). The remaining land uses (intensive uses, production from dryland agriculture and plantations, and production from irrigated agriculture and plantations) accounted for 10%, 5% and 1% of the site catchment areas, respectively.


68

Land uses were found to be relatively uniform across both reference and historic hydrologic classes. Only the land-use class ‘production from relatively natural environments’ showed significant differences, which were between HFCs 1 and 2. Correlations between hydrologic alteration and the proportion of different land uses within catchments were found to be very weak. The strongest correlations identified included a minor relationship between the Gower metric of hydrologic alteration and the proportion of land in the ‘conservation and natural environments’ land-use class. A weak negative relationship between the change in 1-year and 10-year average recurrence interval discharges and the proportion of intensive land use in catchments was also detected. Landholder surveys were returned for 21 out of the 44 sites surveyed. The returned responses indicated considerable variability in the nature of the grazing practices within the riparian zone (i.e. continuous, rotational or opportunistic), the stocking rate (number of cattle per hectare) and grazing history. Furthermore, in a number of instances the owners had only relatively recent knowledge of land management. Historical grazing regimes may still be affecting the extant vegetation structure even where grazing has ceased. Given the above issues, producing meaningful metrics from the landholder surveys reflecting local land management practices presents a considerable challenge. As such, only the coarser-scale land-use metrics based on the QLUMP data were used in further analyses.

3.4. Discussion Forty-four sites along 22 reaches reflecting the major hydrologic gradients in south-east Queensland were selected for the field research program of this ELOHA trial (Table 3.2). These included 12 sites on six regulated reaches, each with four reference sites representing geographically close and distant reference hydrologic conditions and geographically close and distant historic hydrologic conditions (Table 3.1). Although stream discharge patterns are recognised as one of the principal drivers of stream ecology (Poff et al. 1997; Bunn and Arthington 2002), many other catchment characteristics— for example, catchment size, shape, geology and topography—can also be important drivers of stream ecological processes. To develop robust flow-ecology relationships for natural and regulated streams, as the ELOHA framework seeks to do, the possible influence of these other environmental variables must be teased out from the direct influences of hydrology. The broad patterns of environmental variation identified here across the selected study sites represent the broad geographic and climatic trends found in the study region. In summary, these patterns include higher mean annual temperatures, hottest month mean temperatures and coldest month mean temperatures as well as greater catchment areas among the most northerly sites. Higher catchment relief ratios and mean annual rainfall are typical of coastal streams, while sites in the south-west of the study region are associated with the highest elevations. Environmental variables also exhibited moderate variation in relation to reference and historic hydrologic classes (see Section 3.3.1). Given that catchment size, shape and topography are important influences upon stream number, size, water yield and hydrograph shape (Gordon et al. 2005), such differences in morphology and catchment variables among hydrologic classes are to be expected. It is in fact surprising that differences were not detected among a greater number of variables. Many natural variables including those describing catchment topography, geology, soils and climate are likely to be major drivers of hydrologic regimes (Poff et al. 1997; Kennard et al. 2010a) and it might be expected, therefore, that these would vary in a predictable manner in relation to natural (i.e. reference) hydrologic regimes.


69

4. Riparian vegetation 4.1. Introduction Evidence for the influence of hydrology and various facets of the hydrologic regime (timing, magnitude, duration, rate of change, predictability and variability) on riparian vegetation patterns are prevalent within the international literature. Hydrology has been shown to influence many key processes that occur during the life cycle of a riparian plant from seed dispersal (Merritt and Wohl 2002; Chambert and James 2008), to germination and seedling growth (Scott et al. 1997) and adult plant growth and productivity (Burke et al. 1999; Anderson and Mitsch 2008). The timing of key plant life history events such as flowering, seed release and germination of some riparian species may even coincide with specific hydrologic events (Blom et al. 1990; Pettit and Froend 2001). Complex relationships between riparian vegetation and flow may also include the mediation by flow of other disturbances to vegetation such as herbivory or granivory (Andersen and Cooper 2000; Elderd and Doak 2006), as well as the effects of competitive interactions between plant species themselves (Busch and Smith 1995). Riparian vegetation patterns are also likely to reflect a complex mix of underlying local and landscape variables and their interactions. Published analyses of riparian vegetation distribution patterns highlight the importance of broad-scale predictors such as climate, geology and soils in determining the distribution of riparian vegetation at a landscape scale (e.g. Tabacchi et al. 1996; Dixon et al. 2002; Sarr and Hibbs 2007). At a local scale, riparian vegetation often exhibits ‘zones’ that vary along gradients of distance and elevation from the stream edge. These lateral distribution patterns reflect the relative tolerances of riparian plant species to physical disturbances (e.g. shear stresses associated with stream hydraulic conditions) and chemical stresses (e.g. anoxia and chemical toxicities associated with soil waterlogging). The ability of plant species to acquire or intercept resources such as moisture (Lite and Stromberg 2005), light (Hall and Harcombe 1998; Battaglia and Sharitz 2006) and nutrients (Kotowski et al. 2006) also contributes to patterns of species distribution within the riparian zone. This chapter presents a test of the ELOHA methodology applied to riparian vegetation of the south-east Queensland study region. Hypotheses for testing were generated from a literature review of riparian vegetation and hydrologic relationships as well as riparian vegetation responses to hydrologic alteration in south-east Queensland (available in the full scientific report accompanying this Waterlines report). This chapter presents these hypotheses along with the field research program and data analyses designed to test these. The results are discussed in terms of the ELOHA trial and the management of riparian vegetation in south east Queensland.


identify the impact of hydrologic alteration on the structure, dynamics and productivity of riparian vegetation in the study area



identify thresholds or relationships between the structure and ecological responses of riparian vegetation and hydrologic alteration



identify hydrologic variables influencing the condition of riparian vegetation and thresholds of ecological response to hydrologic alteration for the whole suite of hydrologic variables


70



assess the relative influence of hydrologic alteration versus other environmental variables on riparian vegetation condition.

4.1.2.Hypotheses Hypothesis 1: The structure and composition of riparian assemblages in the south-east Queensland region will be influenced by streamflow. Stream discharge patterns are a major control on the distribution, abundance and diversity of plants on stream and river banks (Merritt et al. 2010). Here it is suggested that riparian plant distributions, abundances, diversity and variability of streams within the study area will be largely governed by hydrologic regimes. While there is significant evidence to suggest links between stream discharge and riparian vegetation for other regions of Australia and internationally, this link has not previously been made for riparian vegetation of south-east Queensland. The hypothesis includes several subhypotheses: a. flood and high hydrologic disturbances have a major influence on the composition and structure of riparian vegetation b. baseflow and low flows have a major influence on the composition and structure of riparian vegetation c. variability in stream flows will drive variability in the structure, composition and productivity of riparian vegetation. Hypothesis 2: The structure and composition of riparian vegetation assemblages in the south-east Queensland region will be influenced by interactions between flow variables, and other environmental variables. This hypothesis predicts that the character of riparian vegetation at a local scale will be strongly influenced by landscape-scale environmental variables (see Chapter 3) and that the influence of hydrologic variables will depend on this regional setting. Under this hypothesis it is predicted that the influence of flow will be greatest for near-stream vegetation as other disturbances and environmental influences are likely to become more important with increasing distance and elevation from the stream edge. Hypothesis 3: Riparian vegetation structure in the south-east Queensland region will differ across the reference and historic flow classes. This hypothesis addresses the basic premise of the ELOHA framework that different flow regime classes will have different riparian vegetation because stream flows are the major control on the distribution, abundance and diversity of plants on stream and river banks. In particular, riparian vegetation within a given reference hydrologic class should differ between regulated sites and unregulated sites according to this hypothesis. Conversely, for a given historic hydrologic class, regulated sites might be expected to have similar riparian vegetation structure to unregulated sites. Furthermore, if the influence of discharge is greatest for near-stream vegetation as suggested under hypothesis 2, near-stream tree and shrub assemblages should vary between hydrologic classes to a greater degree than bankfull vegetation. Hypothesis 4: Changes in flow regimes will alter the distribution, abundance and diversity of plants on stream and river banks.


71

This hypothesis focuses on responses of riparian vegetation to hydrologic alteration. A review of the evidence for the impacts of hydrologic regime alteration on riparian vegetation suggests a number of potential effects, depending on the nature of the hydrologic regime alteration, for example: a. changes to the density of near-stream vegetation in response to hydrologic alteration b. reduction in the regeneration of native species in bankfull vegetation where high flows and flood disturbance are reduced c. reduction in the proportion of species that are characteristic of early successional stages where flood disturbance is reduced d. reductions in species diversity where hydrologic variability is reduced e. increased proportions of exotic species in response to hydrologic alteration. The ELOHA framework assumes that increasing levels of hydrologic alteration (i.e. from reference condition) will be associated with an increasing degree of ecological change (Poff et al. 2010). The capacity of this project to test this hypothesis was limited by the degree of hydrologic alteration across the study area. Hydrologic regimes downstream of dams in south-east Queensland have undergone varying degrees of change from baseline (reference) condition (Chapter 2), hence there is scope only to examine changes in riparian vegetation over a relatively subtle gradient of overall hydrologic change.

4.2. Methods 4.2.1.Study sites, hydrologic metrics and environmental variables Study sites All 44 study sites listed in Table 3.2 were used for sampling riparian vegetation in this ELOHA trial. Hydrologic metrics (Chapter 2) and other environmental variables (Chapter 3) considered in this component of the study were selected based on their likely relevance to riparian vegetation.

Hydrologic metrics Hydrologic metrics were selected for use in this component of the study that represented gradients of water availability and fluvial disturbance to riparian habitats. A number of studies have also illustrated a link between riparian vegetation and depth to groundwater (Merritt et al. 2010). Here, it was assumed that surface flows would be analogous with the local riparian watertable, therefore hydrologic metrics describing average flow and low-flow conditions were used to provide proxies for riparian watertable dynamics (Table 4.1). Variables describing mean wet season (November–April) and dry season (May–October) flows and their variation were also calculated. To assess impacts of flood disturbance on riparian vegetation, flooding and high-flow statistics describing the frequency, duration and intensity of bankfull and high-spell flow conditions were used (Table 4.1). Bankfull discharge estimates were calculated from crosssectional surveys conducted at field sites. Since the riparian vegetation attributes to be considered were at community and population levels, relatively long temporal scales were used to calculate relevant hydrologic metrics; that is, over a 20-year period immediately before sampling (with the exceptions of sites on Nerang


72

River, Reynolds Creek and Burnett Creek downstream of Maroon Dam where only shorter discharge records were available). Table 4.1: Hydrologic metrics used to assess riparian vegetation patterns Category

Parameter

Abbreviation

Measures of average and

Median daily flow

MEDDaily

low-flow conditions

Median annual flow

MEDAnnual

Mean wet season flow (November to April)

MDFWet

Mean dry season flow (May to October)

MDFDry

Mean daily baseflow

MDBF

Baseflow index

BFI

Low-spell discharge (75th percentile)

LSDis

Mean duration of low-flow spells (75th percentile)

LSDur

Number of low-flow spells (75th percentile)

LSNum

Measures of flood

High-spell discharge (25th percentile)

HSDis

disturbance

Mean duration of high spells (25th percentile)

HSDur

Mean number of high spells (25th percentile)

HSNum

Bankfull discharge

BFDis

Mean duration of bankfull flow

BFDur

Mean number of bankfull flow events

BFNum

Shear stress at bankfull flow

BFShear

Measures of flow

CV of mean daily flow

CVDaily

variability

CV of annual flow

CVAnnual

CV of wet season flows (November to April)

CVWet

CV of dry season flows (May to October)

CVDry

Reference flow class

REFClass

Historic flow class

HISClass

Flow classes

Other environmental variables All of the landscape-scale environmental and land-use variables discussed in Chapter 3 (see Tables 3.3 and 3.4) were considered in the analyses of riparian vegetation patterns. In addition, proportions of clays and sands were determined from soil samples collected during field surveys.

4.2.2.Riparian vegetation field surveys Riparian vegetation field surveys were undertaken from 2008 to 2010. Initially, 28 sites were sampled from August to October 2008 with surveys of the remaining 16 sites completed in 2009 and 2010. Time constraints precluded completion of the entire survey within a single field season and splitting the sampling over two years rather than sampling across seasons within a year was deemed a more appropriate strategy for the riparian vegetation to avoid confounding potential seasonal effects. At each site, riparian vegetation was surveyed within three 50-metre-long transects randomly located perpendicular to a 100-metre stream section. To accommodate variation in vegetation


73

densities, channel forms and adjacent land uses across study sites, sampling areas varied 2 2 2 between 260 m and 1013 m with the sampling area for most sites being greater than 400 m as recommended by Walker and Hopkins (1984). All transects were conducted on the same side of the river to ensure any land-use impacts were similar. All trees, shrubs, ferns, reeds, rushes and sedges ≥50 cm in height were recorded in a 5 metre-wide belt for each transect. Variables recorded included the species, distance from the water’s edge, canopy height, trunk diameter at breast height, the presence and composition of vines (both exotic and native) and a measure of plant health ranging from 0 (dead) to 4 (healthy with >75% canopy cover and little or no evidence of disease or insect damage). Densities of reeds, rushes (including mat-rush, Lomandra spp.) and sedges were also estimated.

4.2.3.Statistical analyses Data obtained from the field surveys was combined at a site level and standardised by sampling area. A subset of the data, defined as vegetation within 5 metres of the waterline, was defined as ‘near-stream vegetation’, while the total assemblage present was designated ‘bankfull vegetation’. In addition to vegetation assemblage data, a range of riparian vegetation metrics were also calculated (Table 4.2). Hypotheses 1 and 2 (effects of flow and other environmental variables on riparian vegetation patterns) were tested using a range of multivariate ordination techniques; for example, nonmetric multidimensional scaling and partial canonical correspondence analysis. Sites with hydrologic regimes strongly impacted by hydrologic alteration (see Chapter 2) were omitted from these analyses since riparian vegetation may still be undergoing changes in structure and composition in response to hydrologic regime alteration. Regression techniques (e.g. regression random forests and generalised least squares regression) were also used to investigate the influence of hydrologic metrics and other environmental variables on riparian vegetation patterns. Analyses were conducted for vegetation assemblage data as well as for the suite of calculated vegetation metrics (Table 4.2). Hypothesis 3, that riparian vegetation will vary among hydrologic classes, was tested using a range of techniques including multivariate analysis of similarity (ANOSIM), random forests, Kruskal-Wallis tests and Tukey's HSD (honestly significant difference) tests. Hypothesis 4, that hydrologic alteration will influence riparian vegetation patterns, was also tested using ANOSIM to examine tree and shrub assemblage composition between regulated and unregulated sites regardless of hydrologic class as well as within selected reference and historic hydrologic classes (Table 4.3). Regression techniques were also used to examine relationships between hydrologic alteration and riparian vegetation patterns. The Gower dissimilarity metric (see Chapter 2) was used as a measure of the overall degree of hydrologic alteration in this analysis.


74

Table 4.2: Riparian vegetation metrics Type

Metric abbreviation

Definition

Total

RICH

Tree and shrub species richness

D_SPECIES

Density of trees and shrubs (number of species per m )

D_ALL

Density of all trees and shrubs per hectare

D_REGEN

Regeneration density per hectare

BA_ALL

Total basal area per hectare

D_EXOTIC

Exotic density per hectare

D_NATIVE

Native density per hectare

D_REGEN_NATIVE

Native regeneration density per hectare

D_REGEN_EXOTIC

Exotic regeneration density per hectare

%_EXOTIC

% exotic taxa

%_NATIVE

% native taxa

BA_EXOTIC

Exotic basal area per hectare

D_SHRUB

Shrub density per hectare

D_TREE

Tree density per hectare

D_LOMAND

Rush, reed and sedge densities per hectare

BA_SHRUB

Shrub basal area per hectare

BA_TREE

Tree basal area per hectare

D_EARLY

Early (E, EM, EML) density per hectare

D_INTER

Intermediate (EM, M, ML, EML) density per hectare

D_LATE

Late (ML, L, EML) density per hectare

BA_EARLY

Early (E, EM, EML) basal area per hectare

BA_INTER

Intermediate (EM, M, ML, EML) basal area per hectare

BA_LATE

Late (ML, L, EML) basal area per hectare

Origin

Growth form

Succession

2


75

Table 4.3: Regulated sites and corresponding reference sites for selected reference hydrologic classes (RFC) and historic hydrologic classes (HFC) used to examine influence of hydrologic alteration on riparian vegetation patterns Regulated

RFC

HFC

(supplemented)

Unregulated pre-development

Unregulated historic reference

reference

sites Obi Obi Creek downstream of Baroon Pocket Dam

1

3

Coomera River (5 and 25) Amamoor Creek (9 and 22) Mary River (14 and 24)

Burnett Creek (2 and 3) Teviot Brook (7 and 8) Mary River (14 and 24) Mary River (37 and 38) Glastonbury Creek (17 and 23) Tinana Creek (39 and 40)

Six Mile Creek downstream of Six mile Creek Dam

1

4

Coomera River (5 and 25) Amamoor Creek (9 and 22) Mary River (14 and 24)

Coomera River (5 and 25) Amamoor Creek (9 and 22)

Yabba Creek downstream of Borumba Dam (10 and 11)

2

2

Logan River (41 and 42) Glastonbury Creek (17 and 23) Tinana Creek (39 and 40) Teviot Brook (43 and 44)

Teviot Brook (43 and 44) Wide Bay Creek (31 and 32) Munna Creek (33 and 34)

Burnett Creek downstream of Maroon Dam (27 and 28)

2

1


Logan River (41 and 42)

Reynolds Creek downstream of Moogera Dam (20 and 21)

2

1


Logan River (41 and 42)

Nerang River downstream of Hinze Dam (4 and 6)

5

3

Currumbin Creek (29 and 30) Eudlo Creek (18 and 19) North Maroochy (35 and 36)

Burnett Creek (2 and 3) Teviot Brook (7 and 8) Mary River (14 and 24) Mary River (37 and 38) Glastonbury Creek (17 and 23) Tinana Creek (39 and 40)


76

4.3. Results 4.3.1.Riparian vegetation of south-east Queensland Totals of 191 tree and shrub species and 43 vine species were identified (see the accompanying science report for a complete list). The most diverse sites were those on Currumbin Creek, Amamoor Creek, Yabba Creek and Stanley River, which each had more than 40 tree and shrub species. The four most abundant native species were Ficus coronata (sandpaper fig), Castanospermum australe (black bean), Cryptocarya triplinervis (three veined laurel) and Syzygium floribundum (weeping lilly pilly). Exotic taxa comprised 23% of all individuals recorded with the most abundant exotic species including Celtis sinensis (Chinese elm), Lantana camara (lantana), Leucaena leucocephala (leucaena), Cinnamomum camphora (camphor laurel) and Ligustrum lucidum (broad-leaved privet). Densities of trees and shrubs ranged from just under 1000 per hectare (Burnett Creek site 27) to over 21 500 (Teviot Brook); the extremely high density at the latter site due to a large number of Celtis sinensis recruits. Proportions of trees belonging to early, intermediate and late successional stages also varied considerably across the sites with early successional species comprising around 24% of all individuals recorded and intermediate and late successional stage species comprising 42% and 33% respectively. Ordinations of the surveyed vegetation assemblage data (e.g. Figure 4.1) distinguished a number of broad riparian vegetation types. Drier, inland sites were typified by a relatively small suite of species including Grevillia robusta, Casuarina cunninghamiana, Melaleuca viminalis (Callistemon viminalis), Melaleuca bracteata and the exotic Celtis sinensis, while rainforest sites, particularly those of coastal creeks and to the north of the study region in the Mary River catchment, were typified by a diverse assemblage of rainforest species including several not generally considered obligate riparian species. Near-stream vegetation comprised a more limited suite of riparian species associated with rainforest vegetation types (both dry and wet rainforests) including Syzygium floribundum and Ficus coronata. However, even near-stream communities included some species that are typical of most rainforest types and not restricted to riparian zones (e.g. Guioa semiglauca and Cleistanthus cunninghamii).


77

5

0.0

2

42

8

-0.5

7 Cas_cun

1.0 12 1

31 33 38 17

41 44

32

35 Cle_cun Gui_sem Fic_cor 26 36

22

9

4 2

0.5

-1.0

NMDS1

5

2

42

8 7

31

23 34 37 2414 25 33

A_RAIN 29

22

IU

19 CMA_TEMP UNC_CATCH 16 15 A_TEMP PDA BFDur BFNum 18 39

44

-1.0

32 CVDry HSDur 43

NMDS1

30

26 36

1

1.0

35

12

9

38 17

41

CLAYS 13

0.5

4

5

1

3

2

-1.0

0.5

CAT_RELI ELEVFELSIC

0.0

0.5

3

-0.5

(d)

0.0

(c)

0.0

1.0 1.0

0.0

-0.5

-0.5

NMDS2

1.0

-1.0

-0.5

5

3 1

-1.5

Cin_oli Jag_pse 19 39 Ela_obo 16 Glo_fer15 Hym_fla End_dis 18 43 Cry_bid Rho_psi Aca_spp Alp_exc 40 Euc_spp Lop_sua

-1.5

-1.0

Cal_vim

23 34 37 2414 25

0.5

0.5

Gre_rob Mel_bra

(b)

0.0

Neo_dea Slo_aus30 Lig_luc End_pubArc_spp Cin_cam Cry_obo 29 13

-0.5

Mal_phi 3

-1.0

(a)

NMDS2

1.0

Figure 4.1: Ordination non-metric multidimensional scaling (nMDS) of sites based on bankfull tree and shrub density data (a) Position of sites in ordination space (see Table 3.2 for site codes). Vectors show taxa significantly correlated with the ordination (see science report for species abbreviations) (b) Distance to group centroids for sites in each historic hydrologic class (c) Environmental variables significantly correlated with the ordination (see Table 4.1 for variable abbreviations) (d) Distance to group centroids for sites in each reference hydrologic class

-1.5

-1.5

40

-1.0

-0.5

0.0

0.5

1.0

-1.0

-0.5

0.0

0.5


1.0

78

4.3.2.Influence of hydrologic metrics and other environmental variables on riparian vegetation patterns A large number of hydrologic metrics and other environmental variables correlated significantly with the ordinations of the tree and shrub assemblage data (e.g. Figure 4.1). Important hydrologic metrics differed depending upon whether bankfull vegetation or only near-stream vegetation was analysed. Hydrologic metrics characterising high-flow and flood conditions tended to correlate more strongly with ordinations of bankfull vegetation than those for near-stream vegetation. Bankfull shear stress and bankfull discharge, however, only correlated significantly with the ordination of near-stream vegetation. Measures of average flow conditions (median annual flow), mean wet season flow and mean dry season flow also only correlated with the near-stream vegetation ordination. Climate variables correlated relatively strongly with both the bankfull and the near-stream vegetation ordinations. In contrast, the percentage of unconsolidated material in the catchment was the only substrate variable that strongly correlated with the ordinations. The proportion of catchment land use in ‘production from relatively natural environments’, ‘production from dryland agriculture and plantations’ and ‘intensive uses’ also correlated with both the bankfull and the near-stream vegetation ordinations. In contrast to the analyses conducted for the tree and shrub assemblage data (see Figure 4.1), only a limited suite of predictor variables correlated significantly with the ordination of riparian vegetation metrics (see Table 4.2). Variables associated with climate were again dominant—coldest month mean temperature and mean annual rainfall for bankfull vegetation and coldest month mean temperature and mean annual temperature for near-stream vegetation. Statistical analyses (i.e. partial constrained correspondence analysis) suggested that the hydrologic metrics considered explained approximately 14% of the variation in the surveyed riparian tree and shrub assemblages, excluding the influence of other environmental variables and land uses. Environmental variables and land uses independently explained 16.3% and 5.3% respectively. For near-stream vegetation only, however, hydrologic metrics independently explained only 9.5% of variation while other environmental variables independently contributed to 19.7% and land uses to 7.0% of the variation across the study area. When the influence of hydrologic metrics were considered in isolation, coefficient of variation in dry season flows and the number of low-flow spells were both identified as useful predictors of both bankfull and near-stream vegetation assemblages. Statistical models (i.e. random forest models) of the bankfull riparian vegetation metrics (see Table 4.2) indicated that hydrologic metrics describing variability (e.g. CVDry) were particularly important but these were relatively unimportant for near-stream riparian vegetation metrics. Variables describing bankfull flow conditions (e.g. BFShear and BFDis) and average discharge conditions (e.g. MEDDaily and MEDAnnual) were also relatively important predictors in models of bankfull riparian vegetation metrics, but variables describing high-flow conditions (i.e. HSDur, HSDis and HSNum) were relatively unimportant as were those describing low flows and baseflows (e.g. LSDis, LSDur, LSNUm an MDBF). Variables describing bankfull flow conditions (particularly BFShear and BFDis) were also relatively important in the models of near-stream riparian vegetation metrics. Climate variables, especially the coldest month mean temperature, were also important predictor variables in the models of bankfull riparian vegetation metrics. Analysis of the models describing both bankfull and near-stream riparian vegetation metrics revealed the presence of several potential thresholds in relationships with selected hydrologic


79

metrics. In particular, after averaging the effects of all other predictor variables, most riparian vegetation metrics decreased substantially when levels of variation in dry season flow (CVDry) exceeded 0.9. Densities of rushes, reeds and sedges (D_LOMAND) and native regeneration (D_REGEN_NATIVE) as well as the basal area of late successional stage species (BA_LATE) also exhibited possible thresholds when average and low-flow discharges measured between –1 and 0 on a standardised scale, beyond which these metrics increased considerably. Analysis of relationships between selected hydrologic metrics and riparian vegetation metrics detected numerous significant relationships among which variation of dry season flows (CVDry) was identified as particularly significant. Analyses indicated that as CVDry increased, values for many riparian vegetation metrics (RICH, D_SPECIES, D_NATIVE, D_LATE, BA_LATE and D_REGEN_NATIVE) decreased linearly (Figure 4.2). A significant (quadratic) relationship was also found between species richness (RICH) and the coefficient of variation in annual flows (CV) with lowest values of species richness found at intermediate values of CV (Figure 4.2). Among the hydrologic metrics describing bankfull flow conditions, quadratic relationships for species richness (RICH) and tree and shrub density (D_SPECIES) of nearstream vegetation as well as the density of intermediate successional species (D_INTER) in the bankfull vegetation were detected with maximum values in all of these vegetation metrics occurring at intermediate values of bankfull shear stress (BFShear, Figure 4.3). A negative linear relationship was also found between bankfull discharge (BFDis) and these same three riparian vegetation metrics (Figure 4.3).


80

D_SPECIES

0.02

10

1.0

1.2

1.4

1.6

0.06

0.10

50 40 20

30

RICH

30 10

20

RICH

40

50

Figure 4.2 Plots of model fits for significant generalised least squares models showing relationships between variables describing variation in hydrologic metrics (CVDry and CV) and selected near-stream riparian vegetation metrics. Abbreviations for hydrologic variables and riparian metrics are given in Tables 4.1 and 4.2 respectively

4

6

8

12

1.0

CV

1.2

1.4

1.6

CVDry

150

9

50

100

BA_LATE

8 7 6

0

4

1.0

1.2

1.4

1.6

1.0

1.2

1.4

CVDry

1.6

1.0

1.2

1.4

1.6

CVDry

5

6

7

8

9

CVDry log(D_REGEN_NATIVE + 1)

5

log(D_LATE + 1)

8.5 7.5 6.5

log(D_NATIVE + 1)

9.5

10

CVDry

10

1.0

1.2

1.4

1.6

CVDry


81

50000

100000

150000

D_SPECIES

0.02

0

50

50000

100000

BFDis

0

50000

150000

log(D_INTER + 1)

0.20 0.15 0.10

0

50

100 BFShear

100000

150000

BFDis

0.05

log(D_SPECIES+1) (NS)

0.20 0.15 0.10

0

150

BFShear

0.05

log(D_SPECIES+1) (NS)

BFDis

100

150

7.0 7.5 8.0 8.5 9.0 9.5

0

0.06

0.10

20 15 5

10

RICH (NS)

10 5

RICH (NS)

15

20

Figure 4.3: Plots of model fits for significant generalised least squares models showing significant linear or quadratic relationships between variables describing bankfull flow conditions (BFShear and BFDis) and selected riparian vegetation metrics. Metrics are bankfull metrics unless indicated otherwise (e.g. NS = near-stream metrics). Equation coefficients are given in Tables 6.11 and 6.12. Abbreviations for hydrologic variables and riparian metrics are given in Tables 4.1 and 4.2 respectively

0

50

100

150

BFShear

4.3.3.Differences in riparian vegetation patterns across hydrologic classes Significant differences in bankfull riparian vegetation assemblages were detected across both reference and historic hydrologic classes. For the near-stream riparian vegetation, however, significant differences were only detected between historic hydrologic classes. Analyses indicated that the composition of tree and shrub communities was a poor indicator of hydrologic class. Although 9 riparian vegetation metrics were found to vary significantly between historic hydrologic classes and 12 across the reference hydrologic classes (Figures 4.4 and 4.5), pair-wise comparisons only indicated one significant difference—significantly higher species richness in HFC 5 compared with HFC 1. Similar analyses of near-stream riparian vegetation revealed even weaker differences between hydrologic classes.


82

5000

0.02

10

5

1

4

5

1

3

4

5

3

4

5

D_REGEN_NATIVE

0 1000

3000

D_SHRUB 4

Historic flow class

1

5

1

2

3

4

Historic flow class

2

3

4

5

Historic flow class

5000

14000 10000 6000

3

5

50 2

Historic flow class

2000

2

4

0 1

Historic flow class

1

3

150 BA_LATE

60 40 0 2

2

Historic flow class

20

BA_EARLY

6000 0 2000

D_LATE

1

D_TREE

3

Historic flow class

10000

Historic flow class

2

100

4

5

10000

3

6000

2

0 2000

1

10000 15000

D_ALL

0.10 0.06

D_SPECIES

30 20

RICH

40

50

Figure 4.4: Box and whisker plots of riparian metrics for individual historic hydrologic classes where Kruskal-Wallis tests showed significant differences in bankfull metric values between historic hydrologic classes. Multiple comparison tests (Tukey’s HSD) are also shown where applicable (Bonferroni-corrected significance for each test is α/10 = 0.005). See Table 4.2 for metric abbreviations

1

2

3

4

5

Historic flow class


83

5000

0.02

10

3

4

5

1

3

4

5

1


3

4

5

4000

D_INTER 3

4

5

1

5

3

4


5

5

3000

D_SHRUB 2

3

4

5

1

2

3

4

5

0.5

6000

BA_SHRUB

2.0

10000


0.0

0 2000

D_REGEN_NATIVE

12000 8000 4000

2

4

0 1000 1


0

1

3

5000

14000 10000 2000 4

2


6000

D_TREE

10000 D_LATE

6000

3


D_REGEN

2


0 2000

2

5

0 1


1

4

8000

120 0 2

3


20 40 60 80

BA_EXOTIC

10000 15000 5000

1

2

12000


2

1.5

2

1.0

1

D_NATIVE

10000 15000

D_ALL

0.10 0.06

D_SPECIES

30 20

RICH

40

50

Figure 4.5: Box and whisker plots of bankfull riparian metrics for individual reference hydrologic classes where Kruskal-Wallis tests showed significant differences in metric values between reference classes. Multiple comparison tests (Tukey’s HSD) (Bonferroni-corrected significance for each test is α/10 = 0.005) did not return any significant differences between classes for any of the metrics. See Table 4.2 for metric abbreviations

1

2

3

4


5

1

2

3

4

5



84

4.3.4.Relationships between riparian vegetation patterns and hydrologic alteration When all sites were considered, no significant differences were detected between regulated and unregulated sites in either the tree and shrub assemblage data or riparian vegetation metrics, for bankfull or near-stream vegetation. Significant differences were detected, however, for two bankfull riparian vegetation metrics, D_SPECIES and BA_LATE, when sites strongly impacted by hydrologic alteration only were compared. Furthermore, where comparisons between regulated and unregulated sites could be made within specific hydrologic classes (see Table 4.3), significant differences were apparent in the bankfull tree and shrub assemblage data in RFC 5 and HFC 2. A significant effect of flow regulation was detected for two riparian vegetation metrics, including the density of reeds, rushes and sedges (D_LOMAND) for which higher densities were found in all regulated sites than were predicted from regression models for unregulated sites. Species density (D_SPECIES) was also lower in strongly regulated sites. No evidence was found to support the hypothesis that increasing hydrologic alteration will result in predictable patterns of increasing biotic change, as proposed in the ELOHA framework.

4.4. Discussion The riparian vegetation of south-east Queensland is diverse, with 191 tree and shrub species and 43 vine species identified from 44 sites. The most diverse sites were those on Currumbin Creek, Amamoor Creek, Yabba Creek and Stanley River, which each had more than 40 tree and shrub species. Exotic taxa comprised 23% of all individuals recorded. Densities of trees and shrubs ranged from just under 1000 per hectare (Burnett Creek site 27) to over 21 500 (Teviot Brook); the extremely high density at the latter site due to a large number of Celtis sinensis recruits. Broad riparian vegetation types were evident in dry inland sites and rainforest sites. This database is the most comprehensive available for the riparian vegetation of south-east Queensland and has utility for several purposes discussed below. The ELOHA framework is underpinned by several concepts and this component of the study tested several hypotheses as a means of validating the ELOHA framework with respect to riparian vegetation in south-east Queensland. Each hypothesis is discussed below with respect to the results presented in Section 4.3.

Hypothesis 1 The ELOHA method is underpinned by the concept that hydrology is a key determinate of the ecological community (Arthington et al. 2006). Hydrologic metrics explained approximately 14% of the variation in the surveyed riparian tree and shrub assemblages across the study region, excluding the influence of other environmental variables and land uses. Environmental variables and land uses independently explained 16.3% and 5.3% respectively. With respect to the proposed subhypotheses, evidence was found that flood and high-flow disturbance are a major control on the composition and structure of riparian vegetation with numerous high-flow and flood metrics identified as important influences on bankfull riparian vegetation structure. In contrast, only limited evidence was found that baseflows, low flows and average flows are a major control on the composition and structure of riparian vegetation. Hydrologic metrics representing average discharge conditions were found to be moderately


85

important predictors of riparian vegetation composition and structure, but only a very limited suite of low and baseflow variables were found to be significant predictors. Evidence was also found to support the hypothesis that discharge variability will drive variability in riparian vegetation. In particular, variation in the mean dry season flow (CVDry) was identified as the single most important hydrologic metric driving riparian vegetation assemblage patterns as well as riparian vegetation metrics, many of which exhibited clear negative linear relationships with CVDry (see Figure 4.2). Variation in dry season discharge may be critical as riparian vegetation is likely to be more reliant on stream flows during the dry season, when rainfall is lower, than during the wet season. Furthermore, variation in discharge during this part of the year may result in frequent periods in which streams cease to flow, resulting in a dropping of the local riparian groundwater table. An important caveat with respect to this hypothesis is the coarse-scale measures of land use in the study area. Land management practices at a local scale can have a strong impact upon riparian communities and stream ecosystems through extremely localised activities (e.g. vegetation clearance, selective weed control, riparian replanting, localised riparian grazing and burning) that are unlikely to be reflected in broader-scale land-use data such as that used here. Numerous studies have illustrated the impact of grazing and trampling by stock within the riparian zone on riparian vegetation community structure (e.g. Jansen and Robertson 2001). Hence, although distance-weighted land-use metrics were used in this study (c.f. Peterson et al. 2010) that place greater weighting within the analyses on land use close to the survey site, it is highly likely that the role of land use has been underestimated here.

Hypothesis 2 Interactions between hydrologic variables and other environmental variables appear to be relatively unimportant in structuring riparian vegetation in the study region. Other environmental influences, however, independently explained the highest proportion of variation in both bankfull and near-stream vegetation with climatic gradients found to be particularly influential, especially those relating to mean annual rainfall and temperature and the mean temperature of the coldest month. Local and catchment topography appear to be relatively unimportant, although catchment relief ratio is likely to be significant. Catchment relief ratio can influence the shape of stream hydrographs (Gordon et al. 2005), therefore this characteristic may reflect an underlying link between riparian vegetation and stream hydrology. Substrate type also appears to have some importance in structuring riparian vegetation, with the proportion of unconsolidated catchment and the proportion of Felsic igneous geology within the catchment identified here as significant variables. Land use independently explained only a small proportion of the variation in riparian vegetation structure, but the proportion of land use under dryland agriculture and plantations was the most important predictor of riparian tree and shrub species richness and density. As per hypothesis 1, it should be noted that the measure of land use in this analysis may underestimate both the impact of land management within the riparian zone and the interaction between land use and river hydrology.

Hypothesis 3 There was mixed evidence to support the hypothesis that riparian vegetation in the study region differed across the reference and historic hydrologic classes. Differences in riparian vegetation assemblages were more distinct between historic hydrologic classes than between reference hydrologic classes but tree and shrub composition data were poor indicators of hydrologic class overall. Patterns were even weaker when only near-stream riparian


86

vegetation was considered. Given the importance of other environmental variables (i.e. climate, geology, topography and soils) in controlling hydrologic regimes, this result is not surprising. The analyses conducted here do not reveal whether hydrologic regimes directly control the composition and structure of riparian vegetation or whether they simply correlate with other overarching environmental factors that also differ among hydrologic classes. In particular, it should be noted that the climate variables coldest monthly average temperature and mean annual rainfall both differed significantly across hydrologic classes (see Chapter 3) as well as being important predictor variables of riparian vegetation.

Hypothesis 4 Support for the hypothesis that changes in hydrologic regimes will alter the distribution, abundance and diversity of plants on stream and river banks was mixed. No evidence was found of reduced regeneration of native species in the bankfull channel where high flows and flood disturbance have been reduced, nor that the proportion of early successional species has been reduced where flood disturbances have been reduced. Furthermore, the hypothesis that the proportion of exotic species will increase with hydrologic regime change could not be sufficiently tested. Some evidence was found, however, to support the hypothesis that there have been changes in near-stream vegetation densities where hydrologic regimes have been altered. In particular, an effect of hydrologic regulation was detected for the density of reeds, rushes and sedges—species that tend to occur in greatest abundance near stream edges. There is substantial evidence from national and international studies to suggest that reductions in high in-channel and flood flows will result in the encroachment of vegetation into the main channel and a subsequent reduction in active channel width. With respect to the study region, the Mary River Water Resource Plan (Brizga et al. 2004) suggests that riparian vegetation thickening may have taken place downstream of Cedar Pocket Dam on Deep Creek, while encroachment of riparian vegetation into the main channel has been reported on the Nerang River below Hinze Dam (Brizga et al. 2006a), on Reynolds Creek below Moogerah Dam (Brizga et al. 2006b) and on the Brisbane River below Wivenhoe Dam (McCosker 2000). The results of this ELOHA trial also suggest that large herbaceous vegetation groups are denser in regulated streams. Finally, evidence was mixed regarding the hypothesis that riparian vegetation species diversity has been reduced in regulated sites. Overall, species diversity was significantly lower in strongly regulated sites compared to unregulated sites. However, in contrast to the subhypothesis that reductions in species diversity will occur where hydrologic variability is reduced, higher species richness values were actually associated with low rather than high variation in dry season flows.


87

5. Aquatic vegetation 5.1. Introduction The hydrologic regime is known to be just one of several environmental factors that control the distribution and abundance of aquatic vegetation in rivers and streams (Biggs 1996; Carr et al. 1997). Consequently, the direct influence of the hydrologic regime on aquatic vegetation may be relatively less important than other environmental factors that may influence aquatic vegetation at local scales (e.g. riparian shading and substrate composition), or at broader spatial scales (such as climate zones and site position in catchment). Mackay (2007) summarised the key environmental factors that control the distribution and abundance of submerged vegetation in south-east Queensland in a simple conceptual model. This model, based on the work of Biggs (1996) and Riis and Biggs (2001), is a twodimensional habitat template consisting of disturbance and resource availability axes. The disturbance axis is represented by a combination of hydraulic and hydrologic factors and the resource axis is represented by light availability (riparian canopy cover and turbidity) and alkalinity. This model implies that hydrology (e.g. as the coefficient of variation of mean daily discharge) will interact with other key environmental factors to control the distribution and species composition of aquatic vegetation assemblages. This chapter tests the four ELOHA concepts with respect to aquatic vegetation of streams in south-east Queensland. Hypotheses were generated from a literature review of aquatic vegetation and hydrologic relationships as well as aquatic vegetation responses to hydrologic alteration in south-east Queensland (available in the full scientific report accompanying this Waterlines report). This chapter presents these hypotheses along with the field research program and data analyses designed to test them. The results are discussed in terms of the ELOHA trial and the management of aquatic vegetation in south-east Queensland.


identify the impact of hydrologic alteration on the structure and dynamics of aquatic vegetation in the study area



identify thresholds or relationships between the structure and ecological responses of aquatic vegetation and hydrologic alteration



identify hydrologic variables influencing the condition of aquatic vegetation and thresholds of ecological response to hydrologic alteration for the whole suite of hydrologic variables



assess the relative influence of hydrologic alteration versus other environmental variables on aquatic vegetation condition.

5.1.2.Hypotheses Hypothesis 1: Streams with similar flow regime characteristics should be more similar in terms of aquatic vegetation assemblage composition and structure than streams with different flow regime characteristics. This hypothesis relates to a key premise of the ELOHA framework. If this premise holds true in south-east Queensland, then differences in aquatic vegetation should be evident between hydrologic classes. Since aquatic plants are expected to respond to recent short-term


88

antecedent hydrologic regime characteristics and short-term hydrologic events, this hypothesis will be tested with respect to the historic hydrologic regime classification only (see Chapter 2). Hypothesis 2: Aquatic vegetation abundance will vary in relation to discharge magnitude, flood frequency and flow variability. The hypothesis includes several subhypotheses, identified from the literature review (see Appendix 2 of the science report): a. aquatic vegetation abundance will vary inversely with discharge magnitude b. aquatic vegetation abundance will vary inversely with flood frequency c.

aquatic vegetation abundance will be positively correlated with discharge variability.

Hypothesis 3: Aquatic vegetation abundance will be higher in regulated sites than in unregulated sites if flow regulation results in reduced flow variability or reduced frequency of substrate mobilisation. Since flood frequency and magnitude will be reduced downstream of dams, hydrologic alteration may result in changes in substrate stability which in turn may result in a reduced frequency of substrate mobilisation and hence increased abundance of aquatic plants. Hypothesis 4: Increasing levels of flow regime alteration from baseline conditions will result in increasing degrees of change in aquatic vegetation assemblages. The ELOHA framework assumes that increasing levels of hydrologic alteration (from reference condition) will be associated with an increasing degree of ecological change (Poff et al. 2010). Hydrologic regimes downstream of dams in south-east Queensland have undergone varying degrees of change from baseline condition (Chapter 2), hence there is scope to examine changes in aquatic vegetation assemblages over a gradient of hydrologic regime change.

5.2. Methods 5.2.1.Study sites, hydrologic metrics and environmental variables Study sites Forty sites were used for sampling aquatic vegetation in this ELOHA trial (see Chapter 3). Sites in Tinana Creek were too deep to conduct aquatic vegetation surveys and sites in Teviot Brook at Wyaralong were excluded from analysis when landowners denied access to one of the sites in the study reach.

Hydrologic metrics Hydrologic metrics were selected based on an extensive literature review (see science report) to represent those likely to have a significant influence on aquatic vegetation. These metrics describe flood frequency, spell frequency and duration, time between spells, and discharge variability at the site scale (Table 5.1).


89

Other environmental variables Additional environmental variables were considered at three spatial scales: within-site, sitescale and catchment-scale (Table 5.1). Within-site variables mostly included parameters describing the hydraulic environment (i.e. the forces exerted on the streambed by flowing water) and water quality and were measured in the field during vegetation surveys. Catchment-scale variables describing topography, geology, climate and land use, as described in Chapter 3, were also included.

5.2.2.Aquatic vegetation field surveys Four aquatic vegetation field surveys were undertaken between June 2008 and September 2010, but the number of sites surveyed at each time varied due, for instance, to flooding (Table 5.2). At each site, aquatic vegetation was surveyed within five random transects 2 across the stream. The cover of aquatic plant species was then measured within four 1 m quadrats on each transect for in-stream vegetation, defined as vegetation occurring within the 2 wetted perimeter of each transect, and a further two 1 m quadrats for bank vegetation, defined as vegetation rooted on the stream bank within 1 m of the stream edge. The study area was influenced by drought before the commencement of sampling in 2008 (Figure 1.3). With the exception of the Noosa and South Coast catchments, the study area had been drought-declared before the commencement of field surveying in June 2007 (Queensland Government 2007). Flooding occurred throughout the study period and the magnitude of the floods at some sites was relatively large in comparison to floods that had occurred in the five years preceding sampling (Figure 1.3). By the end of the final aquatic vegetation survey (September 2010) only a small area in the south-west of the study region remained drought-declared (Queensland Government 2010). Consequently, the aquatic vegetation survey period was preceded by drought but included periods of above-average rainfall (e.g. Chandler 2009). The year 2010 was the wettest year on record for Queensland (Bureau of Meteorology 2011).


90

Table 5.1: Environmental variables, including site-scale hydrologic metrics, used in aquatic vegetation component Spatial scale

Parameter

Unit

Abbreviation

Catchment and land use

Latitude

degrees

DECLAT

Longitude

degrees

DECLONG

Site

2

Catchment area upstream of site

km

Site elevation

m

ELEV

Site distance to source

km

DISTS

CATAREA

Site distance to mouth

km

Bank slope

m.m

B_SLOPE

DISTM

Catchment elongation ratio

no unit

CAT_ELON

-1

Catchment relief ratio

no unit

CAT_RELI

Reach valley confinement

%

V_Conf

% Felsic geology

%

FELSIC

% Mafic geology

%

MAFIC

% Sedimentary rock (siliclastic and undifferentiated)

%

SED_SILIC

% Sedimentary rock (carbonates)

%

SED_CARB

% Mixed sedimentary and igneous rock

%

MIXED

% Unconsolidated rock (alluvium, colluviums etc.)

%

UNC_CATCH

% Unconsolidated material for reach

%

UNC_REACH

Production from relatively natural environments

%

PNE

Production from dryland agriculture and plantations

%

PDA

Production from irrigated agriculture and plantations

%

PIA

Conservation and natural environments

%

CAN

Intensive uses

%

IU

Annual mean temperature

o

C

A_TEMP

Coldest month mean temperature

o

C

COLD_TEMP

Hottest month mean temperature

o

C

HOT_TEMP

Annual mean rainfall

mm

A_RAINFALL -1

Discharge required to mobilise D50

Ml.day

Q_D50

Percentage of days prior to sampling where discharge was above the threshold required to mobilise the D50

%

FD50MOVE

Number of days since discharge required to mobilise D50 occurred

no unit

DAYS_Q_D50

Number of high spells based on discharge required to mobilise D50

no unit

D50_HSNum

Mean duration of high spells (Q_D50 threshold)

days

D50_HSDur

Number of high spells (75th percentile) Mean duration of high spells (75th percentile)

HSNum days

Number of low spells (25th percentile) Mean duration of low spells (25th percentile)

HSDur LSNum

days

LSDur

Coefficient of variation of mean daily discharge

%

CVDaily


no unit

REF_CLASS

Historic flow class

no unit

HIS_CLASS

Median particle size

cm

D50

Shear stress

Nm

2

Critical shear stress

Nm

2

SHEAR CRITSHEAR


91

Spatial scale

Parameter

Unit

Abbreviation

Substrate stability

no unit

SUBSTAB

-2

Bankfull shear stress

Nm

Bankfull substrate stability

no unit

pH Turbidity

NTU

TURB

Bankfull depth

M

BFDEPTH

Bankfull width

M

BFWIDTH

Ratio of bankfull width:bankfull depth

no unit

BFWIDTH_BFDEP

-1

BEDSLOPE

-1

m.m

Waterslope

m.m

WATERSLOPE

Width

m

WIDTH

Depth

m

Water velocity Reynolds number

1

BFSUBSTAB PH

Bedslope Within-site

BFSHEAR

ms 1

DEPTH -1

VELOC

no unit

REYNOLD

Froude number

no unit

FROUDE

Riparian canopy cover

%

RipCov

Reynolds Number (Re) is the ratio of inertial forces to viscous forces and describes whether flow is laminar (smooth) or turbulent. It is calculated from the equation Re = VL/v where V is velocity (ms-1), L is length (m) and v is kinematic viscosity (m2s-1). Mean depth was used as the length measure for calculating Reynolds number (Gordon et al. 2005). Reynolds number described whether flow is smooth (Reynolds number 2000).


92

Table 5.2: Details of sites surveyed for aquatic vegetation Site number and name

Reference

Historic

Site length

flow class

flow class

(m)

5

5

80

Pool-Run

2. Burnett Creek downstream of gauge 145018a

no class

3

80

3. Burnett Creek upstream of gauge 145018a

no class

3

4. Nerang River at Grand Manor Golf Course

5

3

5. Coomera River at Coomera Scouts Hall

1

1. Stanley River at Cove Road

6. Nerang River at Weber Court

Habitat type

Distance to

No. of

gauge (km)

surveys

15.0

3

Riffle

4.4

4

100

Riffle-Pool

0.7

4

90

Pool-Run

3.3

3

4

100

Riffle

4.4

4

5

3

100

Pool

1.0

4

no class

3

100

Riffle

4.8

3

no class

3

100

Pool

1.7

4

1

4

100

1

Pool

7.1

4

10. Yabba Creek at Stirling Crossing

2

2

100

Riffle-Run

8.9

4

11. Yabba Creek at No. 8 Crossing

2

2

100

Pool

1.3

4

12. Obi Obi Creek downstream of number 2 crossing

1

3

45

Pool

2.0

3

13. Obi Obi Creek upstream of number 2 crossing

1

3

100

Riffle

4.1

3

14. Mary River downstream of Walker Road Bridge

1

3

55

Pool

3.1

3

15. Six Mile Creek at Old Noosa Road

1

4

50

Pool

0.7

3

16. Six Mile Creek at Grahams Road

1

4

40

Pool

7.5

3

17. Glastonbury Creek at Greendale Road Crossing

2

3

100

Pool

11.4

4

18. Eudlo Creek at gauge site

5

5

40

Pool

0.03

4

19. Eudlo Creek upstream of Bruce Highway

5

5

55

Run-Pool

4.5

4

20. Reynolds Creek at Yarramalong campground

2

1

100

Riffle

2.5

4

21. Reynolds Creek downstream of Purdons Bridge

2

1

50

Pool-Riffle

4.7

4

22. Amamoor Creek at Zachariah Lane

1

4

100

Riffle-Run

0.2

4

23. Glastonbury Creek 2 km from Mary River confluence

2

3

100

Run

6.2

4

24. Mary River at Moy Pocket (north of quarry)

1

3

100

Pool-Run

0.9

3

7. Teviot Brook near Brennan Road

1

8. Teviot Brook at Croftby 9. Amamoor Creek at Harrys Creek Road

2


93

Site number and name

Reference

Historic

Site length

flow class

flow class

(m)

25. Coomera River at Tucker Lane

1

4

100

26. Stanley River at gauge 143303a

5

5

27. Burnett Creek 2 km downstream of Maroon Dam

2

28. Burnett Creek at Splityard Creek Road

2

29. Currumbin Creek at Currumbin Valley Primary School

Habitat type

Distance to

No. of

gauge (km)

surveys

Riffle-Pool

8.0

4

100

Pool-Run

0.1

3

1

100

Riffle

2.0

4

1

50

Pool

3.5

4

5

5

100

Riffle-Pool

6.5

3

30. Currumbin Creek at Fordyce Court

5

5

100

Riffle-Pool

2.7

3

31. Wide Bay Creek downstream of gauge 138002c

4

2

100

Riffle-Run

1.8

2

32. Wide Bay Creek upstream of gauge 138002c

4

2

60

Pool

0.5

2

33. Munna Creek at gauge 138004b

4

2

100

Riffle

0.02

2

34. Munna Creek downstream of gauge 138004b

4

2

90

Pool

1.2

2

35. North Maroochy River at Eumundi

5

5

60

Pool

4.0

4

36. North Maroochy River at North Arm – Yandina Creek Road

5

5

100

Run

4.2

4

37. Mary River at Bauple–Woolooga Road

3

3

80

Riffle-Run

14.8

2

38. Mary River at Orphants Road

3

3

90

Riffle-Run

10.0

2

41. Logan River at Running Creek Road

2

1

100

Riffle-Run

0.8

3

42. Logan River at upstream Tilleys Bridge

2

1

65

Pool

3.3

3

1 This site was dry for the first survey.

2 Site length reduced to 30 m after flooding prior to the fourth survey made much of the site too deep to sample.


94

5.2.3.Statistical analyses Environmental variation across the study area was initially explored to determine whether individual historic hydrologic classes had unique environmental characteristics that could influence the interpretation of the role of recent gauged, rather than modelled, discharge on aquatic vegetation patterns. Environmental patterns were explored using non-metric multidimensional scaling (nMDS) and analysis of similarity (ANOSIM) was used to examine differences between historic hydrologic classes. The Kruskal-Wallis test was also used to test whether channel morphology, indicated by the ratio of bankfull width to bankfull depth, differed significantly among hydrologic regime classes as such geomorphic differences may influence biotic patterns (Poff et al. 2010). Aquatic vegetation data obtained from the field surveys was combined at a site level, standardised by sampling area and used to calculate a suite of aquatic vegetation metrics (Table 5.3). These included the allocation of species to functional plant groups as described by Brock and Casanova (1997). Aquatic vegetation metrics were calculated at a site scale and for in-stream vegetation data only. Hypothesis 1 (that aquatic vegetation would vary among historic hydrologic classes) was assessed using both the aquatic vegetation data and the aquatic vegetation metrics. Multivariate techniques (nMDS and ANOSIM) were used to analyse assemblage data, while the Kruskal-Wallis test was used to assess the metrics data. A decision tree method (random forest models) was also used to assess how well variation in aquatic vegetation assemblages corresponded to historic hydrologic classes. The influence of hydrologic metrics and other environmental variables on vegetation patterns was explored using a range of statistical techniques including partial constrained correspondence analysis and generalised least squares regression. The latter analysis was used specifically to test hypothesis 2. Hypotheses 3 and 4, concerning differences between regulated and unregulated sites and the effects of hydrologic alteration, were tested using ANOSIM and partial least squares projection to latent structures modelling, with the Gower dissimilarity metric used as a measure of the degree of hydrologic alteration.


95

Table 5.3: Metrics representing aquatic vegetation. Density metrics are richness metrics standardised by site area. SUB, ATE, ATI, ARP , ARF, TDA and TDR are functional groups described by Brock and Casanova (1997) and defined below. See science report for allocation of plant taxa to functional groups Type

Metric

Definition

Richness and density

TOTRICH

Species richness

TOTDENS

Species density (TOTRICH standardised by site area)

EMRICH

Number of emergent taxa

EMDENS

Density of emergent taxa (EMRICH standardised by site area)

FARICH

Number of floating (attached) taxa

FADENS

Density of floating (attached) taxa (FARICH standardised by site area)

TOTCOV

% In-stream vegetation cover

SUBCOV

% Submerged vegetation cover

FACOV

% Attached floating vegetation cover

EMCOV

% Emergent vegetation cover

NATIVE

Number of native taxa

NATIVEDENS

Density of native taxa

ALIEN

Number of alien taxa

ALIENDENS

Density of alien taxa

SUB

Number of submerged taxa

SUBDENS

Density of SUB taxa

ATE

Number of amphibious fluctuation-tolerators (emergent)

ATEDENS

Density of ATE taxa

ATL

Number of amphibious fluctuation-tolerators (low growing)

ATLDENS

Density of ATL taxa

ARP

Number of amphibious fluctuation-responders (morphologically plastic)

ARPDENS

Density of ARP taxa

ARF

Number of amphibious fluctuation-responders (floating, stranded)

ARFDENS

Density of ARF taxa

TDA

Number of terrestrial taxa associated with damp places

TDADENS

Density of TDA taxa

TDR

Number of terrestrial taxa associated with dry places

TDRDENS

Density of TDR

Cover

Status

Functional group


96

5.3. Results 5.3.1.Environmental variation across historic hydrologic classes Significant variation in environmental variables was detected across the historic hydrologic classes, largely due to the location of these within the study region (see Chapter 3). Sites in historic flow classes (HFCs) 5 and 1 were identified as being particularly distinct from each other. Sites in HFC 5 tended to be closer to the coast and occurred in areas with higher rainfall, riparian canopy cover and turbidity while most sites in HFC 1 were located in the southern part of the study region. Significant differences in channel morphology, indicated by the ratio of bankfull width to bankfull depth, were also apparent between historic hydrologic classes with sites in HFC 4 having significantly lower values than those for HFCs 2 and 3.

5.3.2.Aquatic vegetation of south-east Queensland Seventy-four plant taxa were recorded during the aquatic vegetation surveys (see science report for a full list), the most common of which were the submerged species Potamogeton crispus and Myriophyllum sp., mosses and the emergent species Lomandra spp., Carex spp., Hydrocotyle spp. and Persicaria decipiens. Attached floating vegetation was rare as were taxa that could be classified as amphibious fluctuation-tolerators (low growing) and terrestrial taxa associated with dry places (after Brock and Casanova 1997). Submerged taxa and amphibious fluctuation-tolerators (emergent) were the most common functional groups (Brock and Casanova 1997). Alien taxa comprised 27% of in-stream taxa recorded, the most common of which were watercress (Rorippa nasturtium-aquaticum), mistflower (Ageratina riparia) and Cyperus eragrostis. Eighteen taxa were recorded at a single site only and most taxa were recorded at less than 20% of sites. Species richness ranged from zero to 18 taxa per site with in-stream vegetation absent from three of the forty sites surveyed, all of which were from HFC 5: Six Mile Creek (site 15), Eudlo Creek (site 19) and the Stanley River (site 26). Total in-stream cover was generally less than 20% (median 10%) but high (>40%) at some sites (e.g. site 13 Obi Obi Creek, Burnett Creek downstream of Maroon Dam and Wide Bay Creek). Emergent vegetation cover tended to be higher than submerged vegetation cover. Most vegetation metrics did not vary appreciably over time. Total richness and emergent species richness were highest during the third survey (early 2010) and both decreased by the fourth survey (late 2010). Total cover peaked in survey 2 (late 2009) and then decreased over the remaining two surveys. In contrast, emergent species cover peaked in survey 3 (early 2010) and submerged cover peaked at the second survey (late 2009), decreased in the third survey (post flooding) but increased slightly at the fourth survey. The most variable aquatic vegetation metrics over the sampling period were those describing plant cover (i.e. TOTCOV, EMCOV and SUBCOV).

5.3.3.Differences in aquatic vegetation patterns across historic hydrologic classes Three broad aquatic vegetation types were identified from ordinations of the vegetation assemblage data (e.g. Figure 5.1). The first was characterised by Lomandra spp. and occurred in sites with high riparian canopy cover and Reynolds number and low water conductivity. The second assemblage type was dominated by bryophytes, Persicaria spp.,


97

Commelina spp. and Ageratina riparia (mistflower) and occurred at sites that tended to be more southerly, at higher elevations with coarse substrates and in catchments with largely natural environments. The final aquatic vegetation assemblage type was dominated by a mix of submerged (e.g. Hydrilla verticillata, Potamogeton crispus, Vallisneria nana) and amphibious species (e.g. Hydrocotyle spp., Myriophyllum spp. Ludwigia peploides subsp. montevidensis) and was associated with higher water conductivities, low riparian canopy cover and low Reynolds number. Ordinations indicated that the three broadly differing assemblage types were not distributed with respect to historic hydrologic classes (Figure 5.1). However, significant differences in composition between historic hydrologic classes were identified by ANOSIM, with all pairwise comparisons exhibiting significant differences with the exception of those between HFCs 1 and 3, 2 and 3, and 3 and 4. Statistical models revealed species presence to be a poor predictor of historic hydrologic class while models calculated using species cover (rather than just species presence) were relatively better at accurately allocating sites to HFC 1 (mostly sites influenced by hydrologic alteration) based on their aquatic vegetation. Prediction errors in both types of statistical model (i.e. species presence and species cover) were lowest for HFC 3 and highest for HFC 4, which have broadly similar hydrologic regimes (Chapter 2), and no significant differences in aquatic vegetation apparent. Since most aquatic plant taxa either occurred within too few samples in an individual HFC or occurred in a relatively high proportion of samples across several HFCs, no statistically significant indicator taxon for any of the HFCs was identified. Several taxa were, however, very common in a single historic hydrologic class including Ageratina riparia (mistflower) and Commelina spp. in HFC 1. Marsilea sp. only occurred in HFC 2, although at a relatively low frequency and, although Vallisneria nana occurred at high frequencies across several HFCs, it was especially common in samples from HFC 2. Using the Kruskal-Wallis test, seven aquatic vegetation metrics were identified that varied significantly between historic hydrologic classes—six of these related to differences between HFCs 1 and 5. In general, median values for individual vegetation metrics were highest in HFC 1 and lowest in HFC 5. Exceptions to this trend were the density of amphibious fluctuation-responders (morphologically plastic) and amphibious fluctuation-responders (floating, stranded) and submerged vegetation cover. Densities of amphibious taxa (belonging to the ARF and ARP functional plant groups) were highest in HFCs 2 and 3 respectively, while submerged vegetation cover was highest in HFC 2 (Figure 5.2).


98

Figure 5.1: Ordination (nMDS in three dimensions) of sites based on species cover data (logx+1 transformed for x>0) (a) Position of sites in ordination space as represented by historic hydrologic class membership. Vectors show taxa significantly correlated with the ordination (p=0.01) (b) Environmental variables significantly correlated with the ordination (p≤0.01) (c) Distance to group centroids for sites in each historic hydrologic class. See Table 5.1 for environmental variable abbreviations and science report for species abbreviations

NMDS2

Bryo 3 3 Ager.ripa 3 Carex Pers.deci Commelin 3 1 1 3 Cype.erag 1Rori.nast Pers.lapa Cardamine Persi.spp 11 3 3 313 11 Juncus 4 Lomandra 111 1 4 Pota.ochr 23 5 5 3 2 14 1 5 34 35 1 3 3 5 3 5 4 2 4 4 42 22 1 5 3 Pota.java 3 4 5 2 1 4 5 4 2 4 23 2 2 3 3 13 5 23 3 3 1 4 3 33 34 34 2 3 Hydr.vert 1 5 Ludw.pepl 3 3 Pota.cris 3 3 24 3 Eger.dens 33 3 33 1 2 Myrio_sp 3 3 Vall.nana

-1.0

-0.5

0.0

0.5

1.0

(a)

5 -1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

NMDS1

(b) 1.0

3

3

3 1 ELEV 1 3 SED_SILIC 1CAT_RELIEF DECLAT VAL_SLOPE RipCover 11 BFWID_BFDEP 3 3 313 11 4 111D50 PNE 1 4 23 5 5 3 2 14 1 LSNum 5 34 35 1 3 3 5 3 5 4 4 4 42 22 1 2 5 3 3 5 12 4 4 5 4 2 4 PDA 23 2 2 31 5 Depth 23 3 HYD_RAD 3 1 3 3 3 UNC_CATCH 34 4 2 4 1 3CATAREA 3 3 3 33 5 Q_D50MOVE 3 3 24 3 Width DISTS 33 PIA BFDEPTH 3 3 1 2 3 A_TEMP 3 3 COLD_TEMP 5

NMDS2

-1.0

-0.5

0.0

0.5

3

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

0.5

1.0

1.5

2.0

NMDS1

0.0

1

-0.5

4 5 23

-1.0

NMDS2

0.5

1.0

(c)

-1.5

-1.0

-0.5

0.0 NMDS1


99

3

4

5

3

4

5

1

2

3

4

5

0.008 ARPDENS

0.0010

5

ALIENDENS

1

2

3

4

20

1v5

3

4

5

4

5

1

2

3

4

5

1

2

3

4

5

2v5, 3v5

0

0

0

2

Historic flow class

0.004

5

5

20

40

EMCOV

60

1-3v5

1

3

40

4

SUBCOV

3

10

2

2

0.000

0.000

0.010

NATIVEDENS

0.020

0.0000

ARFDENS

0.010

ATEDENS

0.000 0.006 0.003

TDADENS

0.000

1

TOTCOV

1v5

1

0.000 0.002 0.004

2

2

0.004

1

1

0.002

5

30

4

20

3

10

2

0.0020

1

1-3v5

0.000

SUBDENS

0.010 0.020

1v5

0.000

0.015

EMDENS

1v5, 3v5

0.000

TOTDENS

0.030

Figure 5.2: Box and whisker plots of in-stream aquatic vegetation metrics for individual historic hydrologic classes. The results of multiple comparison tests (Tukeys HSD) are shown where Kruskal-Wallis tests showed significant differences in vegetation metrics across historic hydrologic classes (Bonferroni-corrected significance p = 0.005). See Table 5.3 for description of aquatic vegetation metrics

1

2

3

4

Historic flow class

5

1

2

3

4

Historic flow class


100

5

5.3.4.Influence of hydrologic and other environmental variables on aquatic vegetation patterns The environmental variables, including hydrologic metrics examined here (see Table 5.1), explained 42.9% in aquatic vegetation assemblage data (partial constrained correspondence analysis). Considered independently, hydrologic metrics explained 4.1% of the variation, within-site environmental variables 6.7% and catchment-scale variables (i.e. climate and land use) 23% of the total variation. The unexplained variation may be due to limitations of the environmental variables used here in explaining aquatic vegetation patterns at the assemblage scale (i.e. potentially important environmental variables were not included). Stochastic processes, for example, chance colonisation events, are also likely to be important. The environmental variables D50, RipCov, PIA, PDA, UNC_CATCH and BFSHEAR (see Table 5.1) were the most important variables describing patterns in aquatic vegetation metrics. In particular, D50 (median particle size) was found to be the single most important environmental variable in the majority of statistical models (i.e. random forest models) calculated. Hydrologic metrics were found to be relatively unimportant in the regression models calculated. However, the most important hydrologic metrics identified included DAYS_Q_D50 for submerged vegetation cover and D50_HSDur for total in-stream vegetation cover. Overall, hydrologic metrics had the greatest influence on submerged vegetation cover. Relationships between selected aquatic vegetation metrics and hydrologic metrics were also modelled using generalised least squares regression (e.g. Figure 5.3). The results of this modelling support hypothesis 2a (vegetation cover will be proportional to discharge magnitude) and hypothesis 2b (aquatic vegetation cover will be inversely proportional to flood frequency). In the analysis of hypothesis 2b, FD50MOVE (the frequency of occurrence of the discharge required to mobilise the median particle size in the 12 months prior to sampling) was a better measure of flood frequency than HSNum (number of high spells). There was no support for hypothesis 2c that aquatic vegetation abundance will be positively correlated with discharge variability, measured as the coefficient of variation of mean daily discharge (CVDaily).


101

60 40 0

20

TOTCOV

80

Figure 5.3: Plots of model fits for significant generalised least squares models describing relationships between vegetation metrics and selected hydrologic metrics

0

2

4

6

8

10

10 15 20 25 30 35 0

5

TOTCOV

log(Q_D50MOVE)

-1

0

1

2

3

4

3

4

10 0

5

EMCOV

15

20

log(FD50MOVE)

-1

0

1

2

log(FD50MOVE)

5.3.5.Relationships between aquatic vegetation patterns and hydrologic alteration Hypothesis 3 predicted that aquatic vegetation cover would be higher in regulated sites compared with unregulated sites, assuming substrate stability in regulated sites was higher


102

than unregulated sites. Median particle size did not differ between regulated and unregulated sites. However, bankfull substrate stability (calculated as the ratio of bankfull shear stress to the critical shear stress required to mobilise the median particle size; i.e. BF_SUBSTAB) was significantly higher in unregulated sites. Since the median value for BF_SUBSTAB in regulated sites was greater than 1, however, it is evident that the median particle size is still being mobilised at bankfull discharge in regulated sites (Figure 5.4). Figure 5.4: Box and whisker plots of (a) bankfull substrate stability for hydrologic regulated and unregulated sites, (b) median particle size for regulated and unregulated sites, and (c) total cover for regulated and unregulated sites

(c)

20

Total Cover (%)

5 4

0

1

2

10

Median particle size (cm)

3

5 4 3 2 1

Bankfull substrate stability

30

6

6

40

7

(b)

7

(a)

Regulated

Unregulated

Regulated

Unregulated

Regulated

Unregulated


103

Aquatic vegetation composition differed between regulated and unregulated sites in RFCs 1 and 5 (Figure 5.5). Reference flow class 1 includes Obi Obi Creek at Kidaman, subsequently regulated by Baroon Pocket Dam and RFC 5 includes the Nerang River at Glenhurst, subsequently regulated by Hinze Dam. Greater dissimilarity in vegetation composition in unregulated sites compared with regulated sites within RFCs 1 and 5 was also apparent (Figure 5.5). Figure 5.5: Box and whisker plots of ranked Bray-Curtis dissimilarities for comparisons of vegetation composition between regulated and unregulated sites in (a) RFC 1 (n=28) and (b) RFC 5 (n=19)

(b)

0.8 0.4

0.6

Bray-Curtis dissimilarity

0.6 0.4 0.0

0.2

Bray-Curtis dissimilarity

0.8

1.0

1.0

(a)

Regulated

Unregulated

Regulated

Unregulated

The effect of hydrologic alteration on five selected aquatic vegetation metrics (TOTDENS, TOTCOV, SUBCOV, EMCOV and NATIVE) was determined using partial least squares modelling. In general, measured values for aquatic vegetation metrics were lower than those predicted by the calculated models at regulated sites. There was, however, an effect of hydrologic alteration on total in-stream vegetation cover (TOTCOV), although this response was mixed, with higher values than predicted by the partial least squares models at three sites and lower values than predicted at nine sites. Four sites had values for TOTCOV close to that predicted. The effect of hydrologic alteration for these five aquatic vegetation metrics was also plotted against the Gower metric as a measure of the overall degree of hydrologic alteration (Figure 5.6). A significant negative relationship was detected for TOTCOV. In general, however, there was little difference in the effect of hydrologic alteration on aquatic vegetation downstream of dams even where hydrologic changes were high (i.e. high Gower metric values). The greatest differences in aquatic vegetation occurred at sites downstream of Baroon Pocket Dam (Obi Obi Creek) and Six Mile Creek Dam (Six Mile Creek) where hydrologic changes have been relatively minor (Figure 5.6). The effect of hydrologic change was also found to vary between sites within a given reach.


104

Finally, the effect of hydrologic alteration for TOTCOV was plotted against the percentage change from reference to historic values for selected hydrologic metrics (Figure 5.7). The selected flow metrics had changed downstream of dams in the study area or discriminated between historic hydrologic classes (see Chapter 3). There were no significant correlations between the effect of hydrologic alteration on TOTCOV and the change in hydrologic metric values from reference to historic (Spearman correlation coefficients, p>0.05). These plots show that the change in TOTCOV at sites in the same reach is not always consistent (e.g. change in 10-year average recurrence interval for Six Mile and Yabba creeks), suggesting site-specific factors not related to hydrology within each reach may have greater influence on TOTCOV than hydrologic alteration.


105

Obi

Nrg

Obi

Bur

40

50

Figure 5.6: Scatterplots showing relationships between the effect of hydrologic change versus the Gower metric (an indicator of overall hydrologic change) for sites downstream of dams in the study area. Two sites were surveyed downstream of each dam and each dam is represented by a pair of points (the average of hydrologic alteration effects for all samples). Site codes: Obi = Obi Obi Creek, Six = Six Mile Creek, Bur = Burnett Creek, Ybb = Yabba Creek, Rey = Reynolds Creek, Nrg = Nerang River

Nrg

-20

0

-50

Six

20

TOTCOV

Rey Ybb Rey

Bur

-100

TOTDENS

0

Ybb

Six

0.00

0.05

0.10

0.15

0.20

0.25

0.00

0.10

0.15

0.20

0.25

Degree of flow change (Gower metric)

-500

-200

0

0

500

1500

SUBCOV

400 200

EMCOV

600

2500

800


0.05

0.00

0.05

0.10

0.15

0.20

0.25

0.05

0.10

0.15

0.20

0.25


50 0 -50 -100

NATIVE

100

150

200


0.00

0.00

0.05

0.10

0.15

0.20

0.25



106

0

Ybb Bur

Rey Nrg

Six

Ybb

-20

0

20

40

60

80

60 40 20

Rey Nrg

0

Bur

-40 -20

% change in vegetaton cover

60 40

Obi Obi

20

Six

-40 -20


Figure 5.7: Scatterplots of change in total cover as predicted by partial least square models (effect of flow regulation) versus percentage change in individual hydrologic metrics, calculated as (historic–reference)/reference). Site codes: Obi = Obi Obi Creek, Six = Six Mile Creek, Bur = Burnett Creek, Ybb = Yabba Creek, Rey = Reynolds Creek, Nrg = Nerang River

100

-100

-250

-200

-150

-100

-50

0

60 40 20

-300

-250

% change in mean no. of zero flow days

-50

0

50

60 40 20

-100

-50

0

50

100

60 40 20 0 -40 -20


60 40


20

50

-100

% change in September mean daily flow

0

0

-150

0

50

-40 -20

-50

-200

-40 -20


60 40 20


0

0

% change in flow constancy

-100

100

% change in high spell duration

-40 -20

-50

50

0

50

% change in low spell duration

-100

0

-40 -20


60 40 20 0

-300

-50

% change in CV of mean daily discharge

-40 -20


% change in 10 year ARI

-600

-500

-400

-300

-200

-100

0

% change in timing of annual maximum flow


107

5.4. Discussion

The aquatic vegetation of south-east Queensland is diverse, with 74 plant taxa identified from 40 sites, the most common plants being the submerged species. Floating-attached vegetation was rare as were taxa that could be classified as amphibious or growing in dry places. Species richness ranged from zero to 18 taxa per site with in-stream vegetation absent from three of the 40 sites surveyed, all in HFC 5 (Six Mile Creek, Eudlo Creek and the Stanley River). Alien taxa comprised 27% of in-stream taxa recorded. This database is the most comprehensive available for the aquatic vegetation of south-east Queensland and has utility for several purposes discussed below. The ELOHA framework is underpinned by several concepts and this component of the study tested several hypotheses as a means of validating ELOHA predictions with respect to aquatic vegetation in south-east Queensland. Each of these hypotheses is discussed below with respect to the results presented in Section 5.3.

Hypothesis 1 There was only limited evidence to suggest that streams with similar hydrologic characteristics were more similar in terms of aquatic vegetation than streams with different hydrologic characteristics. Although both aquatic vegetation assemblage composition and vegetation metrics varied significantly across historic hydrologic classes, it was unlikely that hydrologic regime was the primary driver of these patterns. Patterns in aquatic vegetation metrics were more apparent across historic hydrologic classes than patterns among vegetation assemblage composition but were generally similar to the gradient in channel morphology (bankfull width:bankfull depth) and latitudinal gradients that were detected across historic hydrologic classes (see Section 5.3.1 and Chapter 3).

Hypothesis 2 Hypothesis 2a, that aquatic vegetation cover would vary inversely with discharge magnitude, is supported. Aquatic vegetation cover appears to be higher in streams where the probability of occurrence of the discharge required to mobilise the median bed particle size is low. Hypothesis 2b, that aquatic vegetation abundance would vary inversely with flood frequency, is also supported. Aquatic vegetation abundance varied inversely with flood frequency. A measure of flood frequency related to mobilisation of the median particle size (FD50MOVE) was a better measure of flood frequency than the number of high spells (HSNum). No evidence was found to support hypothesis 2c that aquatic vegetation abundance would be positively correlated with discharge variability. Although regression modelling indicated a trend for aquatic vegetation abundance to increase with the coefficient of variation in mean daily discharge, this relationship was not found to be statistically significant.

Hypothesis 3 This hypothesis predicted that aquatic vegetation abundance would be higher in regulated sites than in unregulated sites, if hydrologic alteration resulted in increased discharge stability or reduced frequency of substrate mobilisation. However, this hypothesis could not be tested, since the bankfull substrate stability of flow-regulated sites, while lower than unregulated sites, was still high enough to ensure substrate mobilisation and hence periodic biomass removal. Overall, the effect of flow regulation in the study area was to cause a reduction in


108

total aquatic vegetation cover. Differences between reference hydrologic classes 1 and 5 were particularly apparent.

Hypothesis 4 The hypothesis that increasing levels of hydrologic alteration from baseline condition will produce increasing degrees of change in aquatic vegetation assemblages was not supported. While the magnitude of flow regime change was found to be correlated with the magnitude of change in total aquatic vegetation cover, this relationship was negative; that is, greater hydrologic alteration was actually associated with small change in total vegetation cover.


109

6. Fish 6.1. Introduction Evidence for the influence of hydrology and various facets of the hydrologic regime (i.e. timing, magnitude, duration, rate of change, predictability and variability) on fish assemblage patterns and species is prevalent within the international literature (e.g. Poff et al. 1997; Lytle and Poff 2004; Kennard et al. 2007). In Australian rivers, hydrology has been shown to influence fish assemblage diversity and composition, movement, life history processes, recruitment and productivity (e.g. Arthington et al. 2010; King et al. 2010). The timing of key life history events, such as migration and spawning, may also coincide with specific hydrologic events, and discharge is the major determinant of patterns of longitudinal and lateral connectivity among riverine and estuarine habitats. Complex relationships with discharge may also include the effects on fish assemblages of alien species of fish and plants (Bunn and Arthington 2002). Although hydrologic patterns are an important determinant of the health of rivers, hydrology cannot be considered in isolation. Fish assemblage patterns are likely to reflect a complex mix of underlying local and landscape drivers and their interactions. Published analyses of fish assemblage and species distribution patterns highlight the importance of broad-scale features such as climate, geology, channel structure, habitat and water quality in providing controls on the distribution of fish at a broader landscape scale (e.g. Kennard et al. 2007; Stewart-Koster et al. 2007). This chapter documents the fish component of the ELOHA field trial in south-east Queensland. The objectives of the study and the major hypotheses tested during the field trial are presented, followed by field, laboratory and statistical methods. The results of statistical analyses are presented and interpreted in relation to the major concepts of the ELOHA framework and the hypotheses tested during the field trial. Four main themes are presented and interpreted: relationships between catchment and in-stream environmental variables and fish assemblage structure; the importance of hydrologic regime as an influence on fish assemblages and species; differences in fish response variables (‘indicators’) between reference hydrologic classes and between regulated (supplemented) and unregulated sites; and fish responses to hydrologic alteration variability and hydrologic alteration gradients. The utility and relevance of the field results as guides to water management in south-east Queensland are discussed together with their implications for the utility of the ELOHA framework for assessing environmental flow requirements.


110


identify hydrologic and other environmental variables that may influence the structure of fish assemblages in the study area



identify the effects of hydrologic variability and the impact of hydrologic alteration on measures or ‘indicators’ of fish assemblage structure and the abundance of individual species



identify thresholds or linear relationships between indicators of the structure of fish assemblages and the abundance of individual species and the overall gradient of hydrologic alteration in the study area



identify thresholds or linear relationships between indicators of the structure of fish assemblages and the abundance of individual species and the gradients of alteration of individual hydrologic metrics.

6.1.2.Hypotheses Hypothesis 1: The structure and composition of fish assemblages in the south-east Queensland region will be influenced by interactions between flow history, catchment characteristics, in-stream habitat factors and anthropogenic disturbances. While the history of stream discharge patterns is recognised as one of the principal influences on stream ecology (Poff et al. 1997; Bunn and Arthington 2002), many other catchment characteristics are also important drivers of stream ecological processes and biotic assemblages. Understanding the influences of hydrology on ecological responses given the underlying variability in environmental and anthropogenic activities across the study region presents the first significant challenge of the ELOHA field trial in south-east Queensland. Previous research in this region has demonstrated the influence of catchment characteristics, in-stream habitat and hydrologic history (and their interactions) on fish assemblage structure in the Mary and Albert rivers (Kennard et al. 2007; Stewart-Koster et al. 2007). A test of this hypothesis across a wider range of catchments will either validate previous findings or present differences that relate to the wider geographic scope and greater environmental variability of the catchments included in the present study. Following Kennard et al. (2007), it is expected that the distribution of fish (i.e. presence– absence patterns) will be driven largely by landscape variables and long-term patterns of river discharge, whereas patterns in fish assemblage composition (relative abundance of species) will be more strongly influenced by in-stream habitat and short-term hydrologic history. Hypothesis 2: The structure and composition of fish assemblages in the south-east Queensland region will differ across reference and historic flow regime classes and between regulated and unregulated sites over time. This hypothesis aims to test basic concepts of the ELOHA framework; that is, effects of hydrologic alteration on fish assemblages will be apparent but may vary between reference and historic hydrologic classes. If selections of reference sites (based on modelled predevelopment flow data) for comparison with regulated (supplemented) sites within each reference flow class are sound, then differences can be expected between regulated (supplemented) and unregulated fish assemblages within each flow class that are due to hydrologic characteristics alone, not other environmental factors. Furthermore, the fish


111

response among hydrologic classes may vary due to the particular characteristics of the modelled reference hydrologic regime and the type and degree of hydrologic alteration. In contrast, if hydrology alone is the major driver of fish assemblage structure, differential responses to hydrologic alteration can be expected across the gauged (historic) hydrologic classes but similar responses are expected within these classes. According to the ELOHA concept, streams that are regulated (supplemented) in similar ways should show similar ecological responses, and they should also be more similar ecologically to unregulated streams that fall into the same historic flow classes. Hypothesis 3: Fish population and assemblage indicators will vary predictably along gradients of flow variability and flow regime alteration. This hypothesis predicts that some indicators of fish assemblage structure (e.g. species richness, relative abundance of alien species, abundance of fish in particular guilds) will show responses to flow variability in the study area, and to alterations in particular hydrologic characteristics from their original state before dams were constructed.


112

6.2. Methods 6.2.1.Study sites, hydrologic metrics and environmental variables Study sites Forty sites were used for sampling fish in this ELOHA trial (see Chapter 3). Two sites on Teviot Brook surveyed in the riparian component had to be omitted from analysis of aquatic vegetation when landowners denied access to the study team.

Hydrologic metrics Hydrologic metrics used in this component of the study (Table 6.1) were selected for their known correlation with patterns of fish assemblage composition in the study area (e.g. Kennard et al. 2007; Stewart-Koster et al. 2007). These hydrologic metrics distinguish geographic patterns in river discharge regimes throughout south-east Queensland, broadly reflecting patterns in the magnitude, timing, frequency, variability and rate of change in river discharge (after Poff et al. 1997). Hydrologic metrics were calculated for each study reach based on two time periods, chosen to reflect differences in the mean longevity of fish species in south-east Queensland. Fish in the study region can be grouped into species that are small bodied and short lived: ≤~4 years (e.g. Australian smelt, Retropinna semoni), and larger bodied and longer lived: >4 years (e.g. Australian bass, Macquaria novemaculeata). Hydrologic metrics representing conditions during these two historical flow regime periods (4 years and 15 years) were calculated for each fish survey date to test for the influence of antecedent hydrologic conditions on patterns in fish assemblage composition.


113

Table 6.1: Hydrologic metrics included in multivariate analysis of fish assemblage structure in relation to environmental factors in the study area Flow metric

Abbreviation

Definition

4 yr

15 yr

Mean daily flow

MDF

Mean of all daily flows

Y

Y

Median daily flow

Med

Median of all daily flows

Y

Y

Minimum daily flow

Min

Minimum of all daily flows

Y

Y

Maximum daily flow

Max

Maximum of all daily flows

Y

Y

Q90

P 10

Discharge exceeded 90% of the time

Y

Y

Q10

P 90

Discharge exceeded 10% of the time

Y

Y

Coefficient of variation of daily flow

CV

Standard deviation divided by the mean for daily flows

Y

Y

Number of floods > median daily flow

HSNum

Number of floods greater than median daily flow

Y

Y

Mean rate of rise

MRateRise

Mean difference in daily flow during rising flow events

Y

Y

Mean rate of fall

MRateFall

Mean difference in daily flow during falling flow events

Y

Y

Mean magnitude of rise

MMagRise

Mean magnitude of rise for all flow events

Y

Y

Mean magnitude of fall

MMagFall

Mean magnitude of fall for all flow events

Y

Y

Number of zero-flow days per year

Under0.1

Number of zero-flow days per year

Y

Y

Baseflow index

BFI

Baseflow index (ratio of baseflow to total flow)

Y

Y

Mean daily baseflow

MDBF

Total baseflow component divided by number of days of record

Y

Y

Mean daily flow in January

MDFJanuary

Indicators that distinguish the reference (IQQM) flow classes

Y

Y

Mean daily flow in September

MDFSeptember

―

Y

Y

Mean daily flow in November

MDFNovember

―

Y

Y

ARI 1 year

PS1YrARI

―

Y

ARI 2 years

PS2YrARI

―

Y


114

ARI 10 years

PS10YrARI

―

1 day maximum flow

MA 1daysMaxMean

―

Y

Y

7 day maximum flow

MA 7daysMaxMean

―

Y

Y

Mean annual 1 day minimum

MA 1daysMinMean

―

Y

Y


MA 3daysMinMean

―

Y

Y


MA 7daysMinMean

―

Y

Y

Predictability of mean monthly flow

P_MDFM

―

Y

Constancy of mean monthly flow

C_MDFM

―

Y

Y

ARI = average recurrence interval; IQQM = integrated quantity quality model


115

Other environmental variables All of the landscape-scale environmental and land-use variables discussed in Chapter 3 (see Tables 3.3 and 3.4) were considered in the analyses of fish habitat and assemblage patterns. Local in-stream habitat variables describing flow velocity, depth, width, substrate, vegetation and bank condition were also recorded during field surveys. At random habitat sample points -1 in each study site, reach width (m), depth (m), flow velocity (ms ), substrate composition (estimated as a percentage of mud, sand, fine gravel, coarse gravel, cobble and bedrock) and presence of macrophytes, leaf litter, overhanging/ submerged/emergent vegetation, root mass, undercut bank, large woody debris (>15 cm diameter), small woody debris (75% frequency of occurrence

Density of long-finned eel

AngRei

Density of Duboulay’s rainbowfish

MelDub

Density of Australian smelt

RetSem

Density of freshwater catfish

TanTan

Density of western carp gudgeon

HypKlu

Density of firetail gudgeon

HypGal

Density of Pacific blue-eye

PseSig

Density of gambusia

GamHol

Density of Midgley’s carp gudgeon

HypSp1

Density of striped gudgeon

GobAus

Number of alien species at each sampling site

AlienRichness

Number of species at each sampling site

SPR

Number of native species at each sampling site

NativeRich

Number of non-migratory species at each sampling site

NonMigSPR

Proportion of all species that are native

PropNatRich

Density of all alien species at each site

AlienDensity

Density of all species at each sampling site

TotDensity

Density of all native species at each sampling site

NatDensity

Density of non-migratory species at each sampling site

NonMigDensity

Density of individuals that are native

PropNatDens

Ratio of the number of species:total individuals in a sample

SPDensity

Fish assemblage using presence–absence data, Sorensen index

Comp-PA

Fish assemblage including abundance data

Comp-CPUE

Fish assemblage excluding taxa present in 75% of all samples; freshwater catfish (Tandanus tandanus), western carp gudgeon (Hypseleotris klunzingeri) and fire-tailed gudgeon (Hypseleotris galii) were collected in >50% of all samples; and Pacific blue-eye (Pseudomugil signifer), mosquitofish (Gambusia holbrooki), Midgley’s carp gudgeon (Hypseleotris sp.) and striped gudgeon (Gobiomorphus australis) were collected in >25% of all samples (Table 6.3). These ten species comprised 84.9% of the total fish in electrofishing collections. The remaining 25 species were infrequently sampled (