NATIONAL WATER COMMISSION — Low flows report series i Macroinvertebrate responses to low-flow conditions in New South Wales rivers Bruce Chessman, Tim Haeusler and Andrew Brooks NSW Office of Water Low flows report series, June 2012 NATIONAL WATER COMMISSION — Low flows report series ii © 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 bookshop@nwc.gov.au. Online/print: ISBN: 978-1-921853-73-9 Published by the National Water Commission 95 Northbourne Avenue Canberra ACT 2600 Tel: 02 6102 6000 Email: enquiries@nwc.gov.au Date of publication: June 2012 An appropriate citation for this report is: Chessman B et al 2012, Macroinvertebrate responses to low-flow conditions in New South Wales rivers, 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 or the NSW State Government. NATIONAL WATER COMMISSION — Low flows report series iii Low flows report series 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 NSW Office of Water on behalf of the National Water Commission. NATIONAL WATER COMMISSION — Low flows report series iv Contents Executive summary Report context 1. Introduction 2. Methods 2.1. Macroinvertebrate data 2.2. Site selection 2.3. Hydrological analysis 2.4. Macroinvertebrate traits 2.5. Statistical analysis 3. Results 3.1. Canonical correspondence analysis 3.2. Principal components analysis 4. Discussion 5. Conclusion Shortened forms References viii ix 1 2 2 2 2 4 4 7 7 9 12 14 15 16 Tables Table 1: Seasonal and habitat distribution of macroinvertebrate samples..................................3 Table 2: Summary of canonical correlation analysis for the full set of riffle samples (n = 40) with hydrological metrics calculated over various antecedent periods. The correlation and loadings are for the first canonical variable. ................................................................................................................................7 Table 3: Summary of canonical correlation analysis for the reduced set of riffle samples (n = 36) with hydrological metrics calculated over various antecedent periods. The correlation and loadings are for the first canonical variable. ................................................................................................................................8 Table 4: Summary of canonical correlation analysis for the full set of edge samples (n = 289) with hydrological metrics calculated over various antecedent periods. The correlation and loadings are for the first canonical variable. ................................................................................................................................8 Table 5: Summary of canonical correlation analysis for the reduced set of edge samples (n = 273) with hydrological metrics calculated over various antecedent periods. The correlation and loadings are for the first canonical variable. ................................................................................................................................9 Table 6: Summary of least squares regression of invertebrate traits and low-flow conditions derived from the first principal component of PCA of the hydrological metrics calculated over various antecedent periods. The full set of riffle samples (n = 40) were used for one- to 12-month antecedent periods and a reduced set (n = 36) was used for the 24-month period. .........................................10 Table 7: Summary of least squares regression of invertebrate traits and low-flow conditions derived from the first principal component of PCA of the hydrological metrics calculated over various antecedent periods. The full set of edge samples (n = 289) were used for one- to 12-month antecedent periods and a subset (n = 273) was used for the 24-month period....................................10 Figures Figure S1: Context of reports produced for the Low Flow Ecological Response and Recovery Project. Each circle represents the location of individual case studies and the size of each circle represents the spatial extent of each case study. ................................................................................................................................... ix NATIONAL WATER COMMISSION — Low flows report series v Figure 1: Location of macroinvertebrate sampling sites with hydrological data. .........................3 Figure 2: Example relationships between invertebrate traits of aerophily, rheophily and thermophily and a gradient of low-flow conditions derived from the first principal component of a PCA of hydrological statistics. The low-flow gradient in this figure represents flow conditions for the preceding 12 months. ...............................................................................................................................11 NATIONAL WATER COMMISSION — Low flows report series vi Acknowledgements Macroinvertebrate and associated data were kindly provided by Greg Long and Alison Reardon (Murray-Darling Basin Authority), Sonia Claus, Jan Miller and Chris Rush (NSW Office of Environment and Heritage), Minal Khan (Queensland Department of Environment and Resource Management), Peter Goonan (South Australian Environment Protection Authority) and Lisa Singleton (Victorian Environment Protection Authority). Hydrographic data were obtained from the NSW Office of Water Hydsys database. We thank Simon Williams, Nick Marsh and Stephen Balcombe for comments on a draft version of this report. NATIONAL WATER COMMISSION — Low flows report series vii Executive summary This study is part of a larger project being undertaken by the National Water Commission to investigate the ecological response to and recovery from low flows in Australian rivers. The response of aquatic macroinvertebrates to low-flow conditions in New South Wales rivers was investigated using existing data collected as part of the Murray-Darling Basin Authority’s Sustainable Rivers Audit and the state’s Monitoring, Evaluation and Reporting program. The study examines flow regimes at aquatic macroinvertebrate sampling sites for up to 24 months before sampling, and relates the flow data to flow-relevant biological traits of the macroinvertebrates sampled. Three traits that have a logical conceptual connection to sensitivity or tolerance to low flow were considered: aerophily (preference for high versus low concentrations of dissolved oxygen), rheophily (preference for fast-flowing versus still waters) and thermophily (preference for high versus low temperatures). A combination of canonical correspondence analysis and principal components analysis was used to investigate relationships. The study results supported our a priori hypotheses that aquatic macroinvertebrate assemblages that had been exposed to severe flow reduction or cessation during the period before sampling would be dominated by taxa tolerant of low oxygen concentrations, low water velocities and high temperatures, whereas assemblages not exposed to very low flows would be dominated by taxa that favour aerated, fast-flowing and cool conditions. The relationships were strongest in riffle habitats that are characterised by faster flowing, well-oxygenated water and tend to be the first habitat type to be impacted by reduced river discharge. NATIONAL WATER COMMISSION — Low flows report series viii Report context This report is part of a larger series of reports produced for the National Water Commission’s Low Flow Ecological Response and Recovery Project (Figure S1). This report presents one of 11 hydro-ecological case studies. The purpose of the case studies is to test hypotheses that relate ecological process and function and biological traits to key hydrological measures that are affected by low flows. A summary of the findings in this report and the other case studies are contained in Synthesis of case studies quantifying ecological responses to low flows (Marsh et al. 2012). Guidance on ecological response and hydrological modelling for low-flow water planning Low-flow hydrological classification of Australia Review of literature quantifying ecological responses to low flows Early warning, compliance and diagnostic monitoring of ecological responses to low flows Synthesis of case studies quantifying ecological responses to low flows Figure S1: Context of reports produced for the Low Flow Ecological Response and Recovery Project. Each circle represents the location of individual case studies and the size of each circle represents the spatial extent of each case study. NATIONAL WATER COMMISSION — Low flows report series ix 1. Introduction This study is part of a national project being undertaken to investigate the ecological response to and recovery from low flows in Australian rivers. The project’s aim is to support water planning by developing a contemporary, Australia-wide understanding and quantification of thresholds of response to the onset of, and recovery from, single and successive low-flow events by individual species, biotic assemblages and ecosystem processes. As part of the project’s aim to develop a national understanding, each state or territory is assessing existing ecological and hydrological datasets to investigate the ecological response to low flows. The NSW Office of Water has been engaged by the National Water Commission to investigate the relationships between biota and low flows. This study uses existing aquatic macroinvertebrate data collected across News South Wales by state government agencies and partner organisations as part of the Murray-Darling Basin Authority’s Sustainable Rivers Audit (SRA) (Davies et al. 2010) and the Monitoring, Evaluation and Reporting (MER) program of New South Wales. The study examines flow regimes at aquatic macroinvertebrate sampling sites for up to 24 months before sampling, and considers the longer-term flow regime at each site. The study relates the flow data to flow-relevant biological traits of the macroinvertebrates sampled. The analysis was based on the trait composition of local assemblages rather than their taxonomic composition because taxa that share certain trait states are likely to be affected similarly by low flows regardless of their degree of phylogentic relatedness (Bonada et al. 2007; Brooks et al. 2011). In addition, an analysis of traits provides insights into the mechanisms by which flow reduction may impact on sensitive taxa. NATIONAL WATER COMMISSION — Low flows report series 1 2. Methods 2.1. Macroinvertebrate data A total of 3267 SRA and MER samples collected from 1405 sites across New South Wales and the Australian Capital Territory from spring 2004 to autumn 2010 were considered for possible inclusion in the analysis. Each sample was collected with a hand net (250 μm mesh) from either fast-flowing water (‘riffles’) or slow-flowing or still water (‘edges’). Most samples were collected by New South Wales agency staff according to the state’s sampling manual of the Australian River Assessment System (AusRivAS) (Turak et al. 2004), but some were taken by staff of interstate agencies according to the Queensland, South Australian and Victorian AusRivAS manuals. Collected invertebrates were generally identified only to family or higher taxonomic levels. The AusRivAS sampling and subsampling procedures are not quantitative (e.g. sampling and subsampling effort varies among samples and operators), thus the data were transformed to recorded presence or absence of each identified taxon per sample. 2.2. Site selection The macroinvertebrate sampling locations were mapped in ArcGIS and compared with the locations of 791 active hydrometric (gauging) stations across New South Wales. Gauges located on storages, weirs and irrigation offtakes were excluded from consideration. Macroinvertebrate samples were considered potentially suitable for analysis if collected within approximately five river kilometres of a gauging station, with no major tributary inflow or anabranch outflow between the macroinvertebrate sampling location and the gauge. The record of each gauging station was assessed for its suitability to analyse flow history before each macroinvertebrate sampling event. For each gauge matched to at least one macroinvertebrate sample, a gap analysis in the River Analysis Package (RAP) (Marsh et al. 2003) determined whether there was an unbroken daily flow record for the 12 months preceding each sampling date. The flow record and associated macroinvertebrate sample or samples were excluded from the analysis if a gap in the record was greater than four consecutive days. Gaps smaller than this were filled with the linear fill option in the RAP. A total of 103 of the 1405 macroinvertebrate sampling locations had suitable associated hydrological data (Figure 1) and form the basis of the statistical analysis. Sampling of these sites between October 2004 and June 2010 resulted in 329 samples, collected from both riffle and edge habitats and primarily in spring and autumn (Table 1). 2.3. Hydrological analysis The RAP was used to calculate several metrics to describe the hydrology at each of the sampling locations. The full flow record for each gauge was used to calculate general statistics such as flow percentiles, daily mean and median discharge. Any gaps in the flow record were ignored for the calculation of these metrics, provided they did not occur within the 12 months preceding the sampling dates. The flow record at each sampling location was analysed for four time periods before each sampling occasion: one, three, six and 12 months. We chose these periods to determine whether invertebrate responses to antecedent low-flow conditions were rapid or gradual. The RAP was used to calculate general statistics such as mean and total flow for each of these NATIONAL WATER COMMISSION — Low flows report series 2 periods, as well as distribution statistics such as standard deviation, coefficient of variation, and the mean daily baseflow. The baseflow calculation in RAP employs the Lyn and Holick digital filter method to separate the component of the hydrograph attributable to recent runoff events from the more consistent low flows resulting from groundwater inflow (Marsh et al. 2003). A spell analysis for low-flow periods was also calculated for each of the antecedent periods. For the spell analysis, the low-flow threshold for each gauge was defined as 10 per cent of the mean for the full period of record. This figure was selected because sharp declines in wetted area and velocity are typically observed once flow falls below about 10 per cent of the mean for a particular river (Tennant 1976; Reinfelds et al. 2004). Thus, this figure represents a level below which considerable stress on the in-stream fauna can be expected as a result of loss of habitat. Sample Site Major River Catchment 0 90 180 360 Kilometres Figure 1: Location of macroinvertebrate sampling sites with hydrological data. Table 1: Seasonal and habitat distribution of macroinvertebrate samples. Habitat Edge Autumn Spring Summer Winter Total 128 119 20 22 289 Riffle 17 19 2 2 40 Total 145 138 22 24 329 NATIONAL WATER COMMISSION — Low flows report series 3 2.4. Macroinvertebrate traits Three traits that have a logical conceptual connection to sensitivity or tolerance to low flow were considered for this analysis: aerophily – preference for high versus low concentrations of dissolved oxygen (DO) rheophily – preference for fast-flowing versus still waters thermophily – preference for high versus low temperatures. We hypothesised that conditions of severe flow reduction or cessation of flow would be detrimental to aerophilous macroinvertebrates with high oxygen requirements, because decay of organic matter in sluggish or stagnant streams would lower DO concentrations, and reaeration of stream water from the atmosphere would be less effective without fast, turbulent flow. Similarly, we expected that low-flow and no-flow conditions would not favour rheophilous macroinvertebrates that prefer fast currents, because such habitats would be rare or absent under these conditions. Conversely, we hypothesised that low or zero flows would benefit thermophilic macroinvertebrates, which prefer warmer conditions, because daytime stream temperatures would be likely to rise under these conditions through greater solar heating of shallower waters. Levels of each trait applicable to each macroinvertebrate family were estimated on continuous scales according to the occurrence of the families in 8928 biomonitoring samples collected in New South Wales and the Australian Capital Territory between 1994 and 2010, plus associated environmental measurements (Chessman in review). The aerophily of each family was estimated from the average DO concentration associated with all samples in which the family was collected for which contemporaneous DO data were available, divided by the average DO concentration of all samples with DO data. The thermophily of each family was estimated in a similar manner from the average instantaneous water temperature associated with samples in which that family was detected. Because data on associated current velocity were not available for any of the samples, rheophily was estimated from the habitats from which samples were collected. Habitats were scored in order of increasing energy: 0 for pool edge-waters and other still waters, 1 for glides, 2 for runs, 3 for riffles, 4 for rapids and 5 for cascades and waterfalls. The average hydraulic score of samples in which each family was detected was calculated and divided by the average score of all hydraulically rated samples. 2.5. Statistical analysis 2.5.1. Preparation of data for analysis The statistical analysis related the trait signatures of macroinvertebrate samples from each of the edge and riffle habitats to hydrological metrics characterising the hydrological regime of each sampling site for various periods before sampling. The average aerophily, rheophily and thermophily of the families recorded in each macroinvertebrate sample were calculated, ignoring any taxa not identified to family level. A data screening procedure was used to select a small number of hydrological metrics from the large number generated by the RAP. Metrics were assessed according to their conceptual relevance to likely impacts of low flows on invertebrate assemblages, and by examining bivariate scatterplots of relationships of individual metrics to trait averages. Five metrics were selected as a result of this screening. NATIONAL WATER COMMISSION — Low flows report series 4 1. Minimum discharge: the minimum flow was considered relevant as it was likely to characterise the greatest hydrological stress to which the resident fauna were exposed during the antecedent period under consideration. 2. Number of low-flow spells: this metric was considered relevant because of the likelihood that repeated excursions of flow into the defined low-flow range (i.e. flows below 10 per cent of long-term mean flow) during the antecedent period could result in cumulative stress on the fauna. 3. Length of the longest low-flow spell: this metric was also considered a useful measure of low-flow stress, since it is possible that some invertebrates might tolerate brief periods of low flow but would be unable to cope with extended low flow. 4. Total duration of low-flow spells: this metric was considered a further potentially useful predictor of cumulative stress imposed by low-flow conditions. 5. Mean daily baseflow: this metric was considered a useful summary measure of the persistence of flow during the entire antecedent period, excluding high-flow events. Before the analysis, the variables minimum discharge and mean daily baseflow were standardised to account for differences in stream size among sites by dividing each value by the long-term average discharge for the applicable site. The resulting standardised variables were highly positively skewed and were therefore transformed to log(100x+1), which largely removed the skew. The other hydrological variables were not highly skewed and did not require transformation. 2.5.2. Canonical correspondence analysis The relationships between the traits and hydrological metrics were analysed with canonical correlation analysis (CCA). This method was chosen because a high degree of correlation existed among the three traits and among the five hydrological metrics, which could have created problems of multi-colinearity in many statistical methods. CCA deals with intercorrelations among both dependent variables (in this case, traits) and independent variables (hydrological metrics) by creating linear combinations of both sets of variables (called canonical variables) that are maximally correlated with each other. Thus the first pair of canonical variables (CVx1 and CVy1) has the relationship: CVx1 = ax1 + bx2 … + kxn CVy1 = my1 + ny2 … + wyn where y1 … yn are dependent variables, x1 … xn are independent variables, and a, b, … w are coefficients with values that result in the maximum possible correlation between CV x1 and CVy1. After the first pair of canonical variables is formed, subsequent pairs are created from the residual variance; that is, the variance remaining after the variance accounted for by previous pairs has been removed from the original variables. CCA was performed separately for each antecedent flow period, ranging from one to 12 months. Analyses were also done separately for riffles and edges because of the well-known differences in biota and hydraulics between the two habitats. Because the analyses showed strong relationships with the maximum antecedent period, an additional analysis was done with hydrological metrics calculated over an antecedent period of 24 months. This required that a few samples were excluded because of gaps in the flow record over the longer period. The analyses for the shorter antecedent periods were repeated for this reduced set of samples to permit comparisons of all antecedent periods, including 24 months, for the same set of sites. NATIONAL WATER COMMISSION — Low flows report series 5 2.5.3. Principal components analysis We also analysed the relationship between individual traits and a gradient of low-flow conditions. A principal components analysis (PCA) was performed on the hydrological metrics to obtain a single gradient of low flow. The relationship between each trait and the first principal component axis for each antecedent period was tested using least squares linear regression. The PCA and subsequent regression analyses were performed separately for each antecedent flow period, ranging from one to 24 months. Analyses were also done separately for riffles and edges. NATIONAL WATER COMMISSION — Low flows report series 6 3. Results 3.1. Canonical correspondence analysis For the CCA, results for only the first pair of canonical variables are reported for each analysis. Subsequent pairs were not informative as they generally had low correlation coefficients and few strong loadings for (i.e. correlations with) the original dependent and independent variables. For the riffle analysis including the full set of samples, all CCAs were statistically significant except for the analysis with a one-month antecedent flow period, which was marginally nonsignificant (P = 0.075) (Table 3). The R2 values and first canonical correlation coefficient were highest for the analysis with an antecedent period of 12 months, though not greatly different from the values for periods of three and six months. Results for the riffle analysis including the reduced set of samples were generally similar to those for the full set for antecedent periods of one to 12 months (Table 4). For the additional analysis with an antecedent period of 24 months, the R2 values and first canonical correlation coefficient were slightly lower than for the 12-month analysis. For the edge analysis including the full set of samples, R2 values and values of first canonical correlation coefficient were much lower than for the riffle analysis, but statistical significance was greater because the number of edge samples much exceeded the number of riffle samples (Table 5). The strength of the relationships showed little variation according to the length of the antecedent flow period considered. Results were similar for the analysis of the reduced set of samples that included analysis for a 24-month antecedent flow period. Table 2: Summary of canonical correlation analysis for the full set of riffle samples (n = 40) with hydrological metrics calculated over various antecedent periods. The correlation and loadings are for the first canonical variable. 1 month 3 months 6 months 12 months R2 0.494405 0.668095 0.666695 0.699076 P 0.074606 0.000926 0.000972 0.000289 Canonical correlation 0.662490 0.721544 0.698583 0.737324 Aerophily 0.963160 -0.855161 -0.804215 0.851099 Rheophily 0.990991 -0.923541 -0.961875 0.979526 Thermophily -0.850558 0.892701 0.737281 -0.776689 Minimum discharge 0.901596 0.727399 -0.767455 -0.758168 Number of low-flow spells -0.828480 -0.378084 0.441923 0.369613 Length of the longest low-flow spell -0.799949 -0.977211 0.872529 0.923550 Total duration of low-flow spells -0.873585 -0.902993 0.900177 0.902297 Mean daily baseflow 0.930447 0.683078 -0.952488 -0.899817 Loadings for dependent variables: Loadings for independent variables: NATIONAL WATER COMMISSION — Low flows report series 7 Table 3: Summary of canonical correlation analysis for the reduced set of riffle samples (n = 36) with hydrological metrics calculated over various antecedent periods. The correlation and loadings are for the first canonical variable. 1 month 3 months 6 months 12 months 24 months R2 0.476161 0.690818 0.724976 0.752287 0.720548 P 0.185162 0.001980 0.000598 0.000197 0.000706 Canonical correlation 0.638020 0.734494 0.713110 0.758536 0.748183 Aerophily 0.954474 -0.811449 -0.945562 -0.903783 0.881082 Rheophily 0.993554 -0.888271 -0.989604 -0.987626 0.995955 Thermophily -0.839299 0.877186 0.862013 0.833560 -0.758029 Minimum discharge 0.909009 0.696785 -0.905937 0.824217 -0.692462 Number of low-flow spells -0.820431 -0.287950 0.532724 -0.355764 0.431928 Length of the longest low-flow spell -0.830412 -0.985057 0.811971 -0.887776 0.879578 Total duration of lowflow spells -0.909498 -0.880176 0.870674 -0.874552 0.876606 Mean daily baseflow 0.932894 0.623955 -0.873532 0.877540 -0.893573 Loadings for dependent variables: Loadings for independent variables: Table 4: Summary of canonical correlation analysis for the full set of edge samples (n = 289) with hydrological metrics calculated over various antecedent periods. The correlation and loadings are for the first canonical variable. 1 month 3 months 6 months 12 months R2 0.234082 0.219761 0.208149 0.242797 P 0.000000 0.000000 0.000000 0.000000 Canonical correlation 0.461693 0.428750 0.427388 0.459200 Aerophily 0.885208 0.922630 0.926448 0.885508 Rheophily 0.572350 0.652876 0.675874 0.582183 Thermophily -0.909880 -0.936482 -0.950059 -0.919992 Minimum discharge -0.894570 -0.944712 -0.963885 0.939346 Number of low-flow spells 0.668224 0.324554 0.297190 -0.157640 Length of the longest low-flow spell 0.775575 0.813728 0.734371 -0.769803 Total duration of low-flow spells 0.789473 0.801203 0.831521 -0.904530 Mean daily baseflow -0.928194 -0.735519 -0.764754 0.881112 Loadings for dependent variables: Loadings for independent variables: NATIONAL WATER COMMISSION — Low flows report series 8 Table 5: Summary of canonical correlation analysis for the reduced set of edge samples (n = 273) with hydrological metrics calculated over various antecedent periods. The correlation and loadings are for the first canonical variable. 1 month 3 months 6 months 12 months 24 months R2 0.251106 0.225719 0.227034 0.255047 0.275948 P 0.000000 0.000000 0.000000 0.000000 0.000000 Canonical correlation 0.459028 0.431135 0.454099 0.478620 0.493079 Aerophily 0.877343 -0.916273 0.916579 0.873191 0.881573 Rheophily 0.579601 -0.616429 0.629034 0.551564 0.544291 Thermophily -0.919671 0.896523 -0.913397 -0.892860 -0.862958 Minimum discharge -0.883056 -0.935813 0.954021 0.923033 0.888585 Number of low-flow spells 0.681297 0.319102 -0.263049 -0.117507 -0.201847 Length of the longest low-flow spell 0.774602 0.850391 -0.759068 -0.791070 -0.748848 Total duration of lowflow spells 0.780598 0.821824 -0.832076 -0.906723 -0.923730 Mean daily baseflow -0.918626 -0.760594 0.750702 0.890125 0.937617 Loadings for dependent variables: Loadings for independent variables: 3.2. Principal components analysis We found significant relationships between all traits and low-flow conditions derived from the first principal component of the PCA of the hydrological statistics (tables 6 and 7). These relationships were also significant for all antecedent periods and both riffle and edge habitats (tables 6 and 7). In general, invertebrates that favour cool, fast-flowing, well-oxygenated water dominated assemblages where low-flow conditions were not prevalent (Figure 2). Conversely, invertebrates tolerant of high water temperatures, slow flow and reduced oxygen were common in assemblages exposed to very-low-flow conditions (Figure 2). NATIONAL WATER COMMISSION — Low flows report series 9 Table 6: Summary of least squares regression of invertebrate traits and low-flow conditions derived from the first principal component of PCA of the hydrological metrics calculated over various antecedent periods. The full set of riffle samples (n = 40) were used for one- to 12month antecedent periods and a reduced set (n = 36) was used for the 24-month period. 1 month 3 months 6 months 12 months 24 months 82.8 73.7 71.2 71.1 68.6 Linear regression y = 1.03 + 0.01x y = 1.03 + 0.01x y = 1.03 + 0.01x y = 1.03 + 0.01x y = 1.03 + 0.01x R2 0.38 0.36 0.35 0.42 0.46 P <0.01 <0.01 <0.01 <0.01 <0.01 Linear regression y = 1.60 + 0.1x y = 1.60 + 0.11x y = 1.60 + 0.11x y = 1.60 + 0.12x y = 1.60 + 0.13x R2 0.38 0.40 0.43 0.49 0.50 P <0.01 <0.01 <0.01 <0.01 <0.01 Linear regression y = 0.98 0.01x y = 0.98 0.01x y = 0.98 0.01x y = 0.98 0.01x y = 0.98 0.01x R2 0.31 0.33 0.33 0.36 0.37 P <0.01 <0.01 <0.01 <0.01 <0.01 % variation explained by PC1 Aerophily Rheophily Thermophily Table 7: Summary of least squares regression of invertebrate traits and low-flow conditions derived from the first principal component of PCA of the hydrological metrics calculated over various antecedent periods. The full set of edge samples (n = 289) were used for one- to 12month antecedent periods and a subset (n = 273) was used for the 24-month period. 1 month 3 months 6 months 12 months 24 months 79.0 71.2 66.0 65.5 64.7 Linear regression y = 0.98 + 0.003x y = 0.98 + 0.003x y = 0.98 + 0.003x y = 0.98 + 0.004x y = 0.98 + 0.004x R2 0.14 0.12 0.14 0.16 0.19 P <0.01 <0.01 <0.01 <0.01 <0.01 Linear regression y = 0.71 + 0.02x y = 0.71 + 0.02x y = 0.71 + 0.03x y = 0.71 + 0.03x y = 0.71 + 0.03x R2 0.05 0.05 0.07 0.08 0.08 P <0.01 <0.01 <0.01 <0.01 <0.01 Linear regression y = 1.02 0.005x y = 1.02 0.005x y = 1.02 0.005x y = 1.02 0.006x y = 1.02 0.006x R2 0.15 0.11 0.13 0.17 0.17 P <0.01 <0.01 <0.01 <0.01 <0.01 % variation explained by PC1 Aerophily Rheophily Thermophily NATIONAL WATER COMMISSION — Low flows report series 10 preference for high D.O. conditions Riffles Edges 1.08 1.08 y = 1.03 + 0.01x R2 = 0.42 P <0.01 1.04 Aerophily y = 0.98 + 0.004x R2 = 0.16 P <0.01 1.06 1.04 1.02 1.00 1.00 0.98 0.96 0.96 0.94 tolerance of low D.O conditions Rheophily preference for fastflowing waters preference for still waters 0.92 -5 -4 -3 -2 -1 0 1 2 3 2.40 2.00 -2 -1 0 1 2 3 4 5 1.20 1.20 0.80 0.80 0.40 0.40 -4 -3 -1 0 1 2 3 4 5 -1 0 1 2 3 4 5 y = 0.71 + 0.03x R2 = 0.08 P <0.01 2.00 1.60 -5 -3 2.40 y = 1.60 + 0.12x R2 = 0.49 P <0.01 1.60 tolerance of 1.08 high temperatures -2 -1 0 1 2 3 4 -4 -3 -2 1.08 y = 0.98 - 0.01x R2 = 0.36 P <0.01 1.04 1.04 Thermophily 0.92 4 -4 1.00 1.00 0.96 y = 1.02 - 0.01x R2 = 0.17 P <0.01 0.96 0.92 0.92 preference for low temperatures -5 -4 more extreme low flow conditions -3 -2 -1 0 1 Low flow gradient (PC1) 2 3 0.88 4 -4 less prevalent low flow conditions -3 more extreme low flow conditions -2 Low flow gradient (PC1) less prevalent low flow conditions Figure 2: Example relationships between invertebrate traits of aerophily, rheophily and thermophily and a gradient of low-flow conditions derived from the first principal component of a PCA of hydrological statistics. The low-flow gradient in this figure represents flow conditions for the preceding 12 months. NATIONAL WATER COMMISSION — Low flows report series 11 4. Discussion The three biological traits that we considered – aerophily, rheophily and thermophily – were highly correlated with one another. Thus invertebrates that appear to favour high oxygen levels also tend to favour fast-flowing habitats and cooler waters. This association is to be expected because there is a characteristic longitudinal gradient in river systems throughout the world from cool, turbulent and well-aerated headwater streams to warm, sluggish and less oxygenated lowland rivers (Vannote et al. 1980). Since particular combinations of environmental temperature, flow velocity and aeration are common, adaptation to these environments will favour specific trait combinations. Different invertebrates will have characteristic suites of interrelated traits according to the portions of the abiotic continuum that they inhabit and are adapted to. The study results supported our a priori hypotheses that invertebrate assemblages that had been exposed to severe flow reduction or cessation during the period before sampling would be dominated by taxa tolerant of low oxygen concentrations, low water velocities and high temperatures. Conversely, assemblages not exposed to very low flows would be dominated by taxa that favour aerated, fast-flowing and cool conditions. One way in which this association could arise is through shifts in the composition of assemblages at the same site over time in response to temporal changes in flow regime, with invertebrates that are adversely affected by the changes dying or emigrating, and those that are suited to the changes multiplying or colonising the site. It could also arise through long-term biogeographic processes whereby some streams frequently suffer low-flow stress, and therefore have developed persistent assemblages that are adapted to cope with that stress, whereas other streams rarely have low-flow stress and therefore support persistent stress-intolerant assemblages. Our analysis does not explicitly differentiate the role of these temporal and spatial processes, but some clues are provided by the observation that the strength of association between the low-flow regime and the trait signature of the invertebrate fauna varied with the length of the antecedent period over which flow was considered. If the flow/trait association was due mainly to temporally stable faunal differences between streams with different long-term flow regimes, we would expect a stronger association with longer antecedent periods – because the calculation of flow metrics over longer periods would better characterise the long-term flow regime of a site. Alternatively, if the flow/trait association was mainly due to short-term assemblage adjustments to recent flow conditions, we would expect the association to be stronger for analysis based on shorter antecedent periods. The flow/trait association that we demonstrated was generally strongest when calculated for the longer antecedent flow periods (12 and 24 months), and weakest for very short antecedent periods. This suggests that assemblage adjustment to the flow regime of a stream is a relatively long-term phenomenon, at least in relation to the typical life cycle length of aquatic invertebrates, rather than being dominated by rapid faunal tracking of fluctuations in flow regimes. Similarly, Finn et al. (2009) found that changes in macroinvertebrate assemblages within an individual stream in New South Wales were correlated more strongly with the number of low-flow events over the long term (12 months) than with the number over shorter terms (less than three months). It would be useful to assess the temporal pattern of within-site adjustment of invertebrate faunas to flow regimes more widely, by monitoring changes in flow and trait signatures at individual sites over time. Data from the routine stream macroinvertebrate monitoring program of New South Wales are not well suited to this task because sites are sampled only once or at intervals of about two years. However, other datasets may be available in which the same site was sampled at much shorter intervals over an extended period. It might also be useful to consider relationships of invertebrate trait signatures to antecedent flows calculated within particular seasons, as Extence et al. (1999) NATIONAL WATER COMMISSION — Low flows report series 12 did when relating an invertebrate index based on velocity preferences to antecedent flow regimes for rivers in the United Kingdom. Flow/trait relationships in this study were much stronger for the riffle fauna than for the edge fauna, suggesting that the biota in riffles are most at risk from extended periods of low flow. This is probably because the abiotic environment in riffles is more strongly affected by changes in flow than that of pools. In riffles, flow reduction will sharply reduce both velocity and wetted area, whereas in pools velocities are typically low even during periods of moderately high flow. In addition, low flows will only slightly lower water levels in pools, unless flow stops altogether and pools begin to dry out through evaporation or seepage. Moreover, the effects of low flow may be variable in pools depending on the characteristics of a particular pool and stream (Boulton & Lake 2008; Brooks et al. 2011). For example, pools with large amounts of organic matter may suffer deoxygenation during low flow, whereas other pools may contain low quantities of organic matter and maintain high oxygen levels when flow is low. There was no evidence of a threshold response to increasing severity of low flow, as relationships between traits and flow were essentially linear (Figure 2). This implies that any increase in the severity of low flows as a result of water abstraction (‘artificial drought’) will impact on the macroinvertebrate fauna, especially in fast-flowing habitats such as riffles. However, there was considerable variability in trait averages for any given flow regime, suggesting that these traits also respond to other environmental factors. In Denmark and the United Kingdom, Dunbar et al. (2010) found that an invertebrate index based on velocity preferences related to both antecedent flow regime and the physical structure of stream channels. In further development of our approach, it could be useful to incorporate other independent variables (e.g. stream morphology and temperature regime) in statistical models to strengthen the ability to predict invertebrate responses to particular changes in flow regime. Long-term studies with frequent sampling at individual sites would also be useful to assess the degree and pace of recovery after low-flow and zero-flow events. This would help formulate recommendations for management of water abstraction because greater impact might be tolerated if the impact is only transient. Previous studies have related flow regimes to a variety of macroinvertebrate traits other than the ones that we considered (e.g. lifespan, fecundity, voltinism; Bonada et al. 2007; Arscott et al. 2010). Our ability to test other traits (such as those listed above) was constrained by the scarcity of published lists of the traits of Australian freshwater invertebrates (though see Schäfer et al. 2011). Trait-based approaches would benefit from the compilation of a comprehensive trait database for Australian freshwater invertebrate taxa, using consistent trait definitions and measurement scales or categories. Such a database could probably not be compiled from published information alone, and would require substantial input from specialists in different invertebrate groups. NATIONAL WATER COMMISSION — Low flows report series 13 5. Conclusion This study investigated the response of macroinvertebrates to low-flow conditions in New South Wales using existing datasets for both macroinvertebrates and hydrology. The macroinvertebrate traits of aerophily, rheophily and thermophily were significantly related to the antecedent flow conditions, with the strongest relationships evident for flow conditions of 12 months before each of the samples being collected. The results support our a priori hypotheses that invertebrate assemblages that had been exposed to severe flow reduction or cessation during the period before sampling would be dominated by taxa tolerant of low oxygen concentrations, low water velocities and high temperatures, whereas assemblages not exposed to very low flows would be dominated by taxa that favour aerated, fast-flowing and cool conditions. The results also showed that these relationships were strongest in riffle habitats that are characterised by faster flowing, well-oxygenated water and tend to be the first habitat type to be impacted by reduced flow. Although the study found significant relationships, the use of existing datasets was less than ideal. Approximately only 10 per cent of samples collected under routine monitoring and evaluation programs were able to be used in the study due to the lack of appropriate hydrological data. It is also noted that very little site-specific information on channel morphology is collected during routine sample programs, making it very difficult to establish how flow conditions translate into the hydraulic conditions being experienced by biota in various habitats. We recommended that future efforts be directed to locating routine biological monitoring sites near hydrological monitoring sites where possible, and that greater effort be directed to collecting channel morphology or hydraulic information at the time of sampling. These minor refinements to routine MER programs at a statewide level would maximise their potential usefulness in evaluating water allocation management actions and policies. NATIONAL WATER COMMISSION — Low flows report series 14 Shortened forms ACT Australian Capital Territory AusRivAS Australian River Assessment System CCA canonical correlation analysis DO dissolved oxygen MER Monitoring, Evaluation and Reporting NSW New South Wales PCA principal components analysis RAP River Analysis Package SRA Sustainable Rivers Audit NATIONAL WATER COMMISSION — Low flows report series 15 References Arscott DB, Larned S, Scarsbrook MR & Lambert P 2010, ‘Aquatic invertebrate community structure along an intermittence gradient: Selwyn River, New Zealand’, Journal of the North American Benthological Society 29, 530–545. Bonada N, Rieradevall M & Prat N 2007, ‘Macroinvertebrate community structure and biological traits related to flow permanence in a Mediterranean river network’, Hydrobiologia 589, 91–106. Boulton AJ & Lake PS 2008, ‘Effects of drought on stream insects and its ecological consequences’, in Lancaster J & Briers R eds. Aquatic insects: challenges to populations, CABI Publishing, Wallingford, pp 81–102. Boulton AJ & Lake PS 2008, ‘Effects of drought on stream insects and its ecological consequences’, in Lancaster J & Briers R eds. Aquatic insects: challenges to populations, CABI Publishing, Wallingford, pp 81–102. Chessman BC (in review), Biological traits predict shifts in geographic ranges of freshwater invertebrates during climatic warming and drying. Davies PE, Harris JH, Hillman TJ & Walker KF 2010, ‘The Sustainable Rivers Audit: assessing river ecosystem health in the Murray-Darling Basin, Australia’, Marine and Freshwater Research 61: 764–777. Dunbar MJ, Pedersen ML, Cadman D, Extence C, Waddingham J, Chadd R & Larsen SE 2010, ‘River discharge and local-scale physical habitat influence macroinvertebrate LIFE scores’, Freshwater Biology 55, 226–242. Extence CA, Balbi DM & Chadd RP 1999, ‘River flow indexing using British benthic macroinvertebrates: a framework for setting hydroecological objectives’, Regulated Rivers: Research and Management 15, 543–574. Finn MA, Boulton AJ & Chessman BC 2009, ‘Ecological responses to artificial drought in two Australian rivers with differing water extraction’, Fundamental and Applied Limnology/Archiv für Hydrobiologie 175: 231–248. Marsh N, Sheldon F & Rolls R 2012, Synthesis of case studies quantifying ecological responses to low flows, National Water Commission, Canberra Marsh NA, Stewardson MJ & Kennard MJ 2003, River Analysis Package, Cooperative Research Centre for Catchment Hydrology, Monash University, Melbourne. Reinfelds I, Haeusler T, Brooks AJ & Williams S 2004, ‘Refinement of the wetted perimeter breakpoint method for setting cease-to-pump limits or minimum environmental flows’, River Research and Applications 20, 671–685. Schäfer RB, Kefford BJ, Metzeling L, Liess M, Burgert S, Marchant R, Pettigrove V, Goonan P & Nugegoda D 2011, ‘A trait database of stream invertebrates for the ecological risk assessment of single and combined effects of salinity and pesticides in South-East Australia’, Science of the Total Environment 409: 2055–2063. Tennant DL 1976, ‘Instream flow regimens for fish, wildlife, recreation and related environmental resources’, Fisheries 1, 6–10. Turak E, Waddell N & Johnstone G 2004, New South Wales (NSW) Australian River Assessment System (AUSRIVAS) sampling and processing manual, Department of Environment and Conservation, Sydney. Vannote RL, Minshall GW, Cummins KW, Sedell JR & Cushing CE 1980, ‘The river continuum concept’, Canadian Journal of Fisheries and Aquatic Sciences 37, 130–137. NATIONAL WATER COMMISSION — Low flows report series 16 Reports in the low flow series Balcombe SR & Sternberg D 2012, Fish responses to low flows in dryland rivers of western Queensland, National Water Commission, Canberra. Barma Water Resources & Sinclair Knight Merz 2012, Low-flow hydrological monitoring and modelling needs, report by for the National Water Commission, Canberra. Barmah D & Varley I 2012a, Hydrologic modelling practices for estimating low flows – stocktake, review and case studies, National Water Commission, Canberra Barmah D & Varley I 2012b, Hydrologic modelling practices for estimating low flows – guidelines, National Water Commission, Canberra Bond N 2012, Fish responses to low-flows in lowland streams: a summary of findings from the Granite Creeks system, Victoria, National Water Commission, Canberra. Bond N, Thomson J & Reich P 2012, Macroinvertebrate responses to antecedent flow, longterm flow regime characteristics and landscape context in Victorian rivers, National Water Commission, Canberra. Chessman B et al 2012, Macroinvertebrate responses to low-flow conditions in New South Wales rivers, National Water Commission, Canberra. Deane D 2012, Macroinvertebrate and fish responses to low flows in South Australian rivers, National Water Commission, Canberra. Dostine PL & Humphrey CL 2012, Macroinvertebrate responses to reduced baseflow in a stream in the monsoonal tropics of northern Australia, National Water Commission, Canberra. Hardie, SA et al 2012, Macroinvertebrate and water quality responses to low flows in Tasmanian rivers, National Water Commission, Canberra. Kitsios A et al 2012, Fish and invertebrate responses to dry season and antecedent flow in south-west Western Australian streams, National Water Commission, Canberra. Leigh, C 2012, Macroinvertebrate responses to dry season and antecedent flow in highly seasonal streams and rivers of the wet-dry tropics, Northern Territory, National Water Commission, Canberra. Mackay S et al; 2012, Low-flow hydrological classification of Australia, National Water Commission, Canberra. Marsh N et al 2012, Synthesis of case studies quantifying ecological responses to low flows, National Water Commission, Canberra. Marsh N et al 2012, Guidance on ecological responses and hydrological modelling for lowflow water planning, National Water Commission, Canberra. Rolls R et al 2012, Review of literature quantifying ecological responses to low flows, National Water Commission, Canberra. Rolls R et al 2012, Macroinvertebrate responses to prolonged low flow in sub-tropical Australia, National Water Commission, Canberra. Sheldon F et al 2012, Early warning, compliance and diagnostic monitoring of ecological responses to low flows, National Water Commission, Canberra. Smythe-McGuiness Y et al 2012, Macroinvertebrate responses to altered low-flow hydrology in Queensland rivers, National Water Commission, Canberra. NATIONAL WATER COMMISSION — Low flows report series 17