Assessing effects of flow alteration on ... - Wiley Online Library

Freshwater Biology (2010) 55, 1780–1800

doi:10.1111/j.1365-2427.2010.02403.x

APPLIED ISSUES

Assessing effects of flow alteration on macroinvertebrate assemblages in Australian dryland rivers B R U C E C . C H E S S M A N * , H U G H A . J O N E S * , N I R V A N A K . S E A R L E †, I V O R O . G R O W N S ‡ A N D MARITA R. PEARSON§ *New South Wales Department of Environment, Climate Change and Water, Parramatta, NSW, Australia † Ecowise Environmental, Brisbane, Australia ‡ New South Wales Department of Environment, Climate Change and Water, Armidale, NSW, Australia § Western Catchment Management Authority, Dubbo, NSW, Australia

SU M M A R Y 1. Possible impacts of water-resource development on assemblages of freshwater macroinvertebrates were investigated in the upper Darling River and some of its tributaries in north-western New South Wales (Australia), an arid and semi-arid region of low relief where alteration of river flows has intensified through expansion of irrigated agriculture. 2. Study sites were grouped into four hydrological regimes resulting from impoundment, flow regulation, water abstraction and natural variation, namely (i) intermittent flow with relatively little hydrological alteration from water-resource development, (ii) intermittent flow with substantial alteration, (iii) near-perennial flow with substantial alteration but unimpounded and (iv) near-perennial flow with substantial alteration plus impoundment by weirs that stabilise water levels. 3. Macroinvertebrates were sampled with three methods (a quantitative cylinder sampler, handnet sampling and baited traps) in three periods with differing hydrology (recessional low flow in June 2003, high flow in March 2004 and increasing flow after drought in December 2004). 4. Taxonomic richness, assemblage composition and catch per unit effort of the crayfish Cherax destructor differed significantly among the site groups, but total macroinvertebrate density and the AUSRIVAS O ⁄ E (Australian River Assessment System observed-overexpected) index did not. The principal spatial differences were between the intermittent and near-perennial rivers, and apparent effects of water-resource development and impoundment were more subtle. Temporal differences in richness, abundance and composition were substantial and appeared to be related mainly to variations in discharge and temperature. 5. Current macroinvertebrate-based methods for assessing the ‘condition’ or ‘health’ of Australian dryland rivers are inadequate. Such assessments might be improved with (i) reference data that take adequate account of antecedent hydrological conditions, (ii) consideration of long-term taxonomic richness as well as richness on individual sampling occasions, (iii) evaluation of invertebrate population sizes, (iv) analysis of assemblage data by trait composition and (v) adoption of the genus as the default level of taxonomic resolution. Keywords: arid, macroinvertebrate, river, taxonomic resolution, water use

Correspondence: Bruce C. Chessman, New South Wales Department of Environment, Climate Change and Water, POB 3720 Parramatta, NSW 2124, Australia. E-mail: [email protected]

1780

2010 Blackwell Publishing Ltd

Flow alteration and macroinvertebrates in dryland rivers Introduction In many arid and semi-arid parts of the world, the conservation of plants and animals that depend on rivers and wetlands is jeopardized by water demands, especially for irrigated agriculture (Kingsford, 2000a; Lemly, Kingsford & Thompson, 2000; Hauer & Lorang, 2004; Nilsson et al., 2005; Kingsford, Lemly & Thompson, 2006; Poff et al., 2007). In Australia, the continent with the lowest average rainfall after Antarctica, water-resource development has profoundly changed the hydrological regimes of many dryland rivers and floodplains, particularly those in the Murray-Darling Basin (Walker, 1985; Kingsford, 2000b). Most of these rivers traverse flat landscapes where rainfall is erratic and periodically intense, partly in association with supra-seasonal cycles of the El Nin˜o-Southern Oscillation and Indian Ocean Dipole (Puckridge, Walker & Costelloe, 2000; Ummenhofer et al., 2009). Consequently, their natural flow regimes fluctuate irregularly between floods, when extensive floodplains are inundated, and droughts, when surface water contracts to disconnected waterholes in river channels and isolated floodplain lakes and ponds (Walker, Sheldon & Puckridge, 1995). Between these extremes are times when flow persists but is confined to the main river channels (Bunn et al., 2006). Water-resource development produces complex changes from this natural pattern (Thoms & Sheldon, 2000; Thoms & Parsons, 2003). The partial or total capture of high flows in dams and off-river storages limits the extent and duration of downstream flooding, and the abstraction of water from river channels reduces in-channel flows and water levels. Conversely, releases of water from storages, and irrigation return flows, can artificially augment in-channel flows during dry periods, producing ‘anti-droughts’ (McMahon & Finlayson, 2003). These diverse hydrological changes are likely to induce equally complex alterations in the quantity, quality and connectivity of habitats for aquatic species (Boulton, 1999; Sheldon & Thoms, 2006). Macroinvertebrates are a functionally significant and diverse component of the biota of dryland rivers, although such rivers generally have fewer invertebrate species than those in wetter regions (Boulton, 1999; Vinson & Hawkins, 2003; Boulton, Sheldon & Jenkins, 2006; Maltchik & Medeiros, 2006). The effects 2010 Blackwell Publishing Ltd, Freshwater Biology, 55, 1780–1800

1781

of water-resource development on macroinvertebrate assemblages are not well understood for Australian dryland rivers and wetlands, where most impact studies to date have focused on waterbirds (e.g. Kingsford & Johnson, 1998; Kingsford & Thomas, 2004). Better knowledge of human impacts on macroinvertebrate assemblages in Australian dryland rivers is needed not only for the conservation of invertebrate biodiversity but also because assessment of macroinvertebrate assemblages is widely used to infer the ‘condition’, ‘ecological condition’ or ‘health’ of these rivers (e.g. Choy, Thomson & Marshall, 2002; Halse et al., 2007; Norris et al., 2007). Such inferences are usually based on the Australian River Assessment System (AUSRIVAS; Simpson & Norris, 2000), a derivative of the British River Invertebrate Prediction and Classification System (RIVPACS; Clarke, Wright & Furse, 2003). AUSRIVAS assessments focus on an observed-over-expected (O ⁄ E) index (Clarke et al., 1996), calculated from the presence and absence of families and some higher taxa in samples from assessment sites and reference sites. However, this approach was developed primarily to evaluate impacts of chemical pollution in non-arid regions, and the interpretation of AUSRIVAS O ⁄ E scores is problematic for dryland rivers. Chessman, Thurtell & Royal (2006) found that AUSRIVAS O ⁄ E scores failed to discriminate between sites with different degrees of exposure to anthropogenic disturbance in the lowland, dryland portion of the Lachlan River system in the Murray-Darling Basin. Sheldon (2005) reported a wide range in AUSRIVAS O ⁄ E scores for sites in the dryland Georgina-Diamantina River system in the Lake Eyre basin, where intensive agriculture is absent, water-resource development is minimal, and broad-scale pastoralism is the only evident source of human impact. She suggested that the principal cause of score variation was changes in macroinvertebrate assemblages driven by natural hydrological fluctuations. Therefore, AUSRIVAS O ⁄ E scores for dryland rivers may be more indicative of antecedent weather than of human impact. Here, we report on a study of macroinvertebrate assemblages in the upper Darling River system and its northern tributaries in north-western New South Wales (NSW). This is a dryland region where frequent concerns have been raised about the ecological consequences of irrigated agriculture (Kingsford, Boulton & Puckridge, 1998; Kingsford, Thomas & Curtin, 2001;

1782

B. C. Chessman et al.

Ogden, Thoms & Levings, 2002). Because of uncertainty about which techniques of macroinvertebrate sampling and methods of data interpretation might best reveal impacts of water-resource development, we compared three sampling methods and analysed the richness, abundance and composition of macroinvertebrate assemblages at three levels of taxonomic resolution: species, genus and family. We also calculated the AUSRIVAS O ⁄ E index. Our objectives were (i) to obtain a better appreciation of the impacts of water-resource development on riverine macroinvertebrate assemblages in our study region and (ii) to compare the merits of different methods of macroinvertebrate sampling and sample processing for assessment of human impacts on Australian dryland rivers. We sampled in three periods with differing antecedent hydrological conditions in case impacts might be more obvious under some conditions than others. However, the assessment of temporal patterns was not a primary purpose of our study.

Methods Study area and sites The study area encompassed the upper Darling River and several of its northern tributaries (Fig. 1). This is a sparsely populated, arid and semi-arid area where rangeland grazing is the most extensive land use. The climate is hot with mean daily minimum temperatures of ca. 13 C and maxima of ca. 27 C. Annual

rainfall averages between 200 and 500 mm, increasing from west to east, but with high intra-annual and inter-annual variability. River flows are naturally highly varied (Thoms & Sheldon, 2000) and have been affected to varying degrees by recent expansion of irrigated agriculture, especially cotton growing (Arthington, 1996; Kingsford, 1999; Thoms & Sheldon, 2000; Porter, 2002; Thoms & Parsons, 2003). While there are no large on-river dams within the study area, river flows are affected by several such dams farther upstream, as well as by widespread water abstraction and capture of floodwaters in off-river storages. Large and small weirs are common, especially on the Barwon and Darling rivers, and because of gentle stream gradients a single weir can impound many kilometres of river. Seventeen study sites were selected close to flow gauging stations (Fig. 1; Table 1). These sites were chosen to encompass a range of hydrological alterations resulting from impoundment, flow regulation and water abstraction and were placed into four distinct groups. Group A comprised sites in the Paroo and Warrego River basins that have naturally intermittent flow and relatively little hydrological alteration from water-resource development. In the case of the Paroo, less than 1% of the long-term average discharge is available for diversion (CSIRO, 2007a), whereas for the Warrego up to 11% may be diverted (CSIRO, 2007b). Site group B comprised naturally intermittent streams with substantial flow alteration in the Moonie and Condamine-Culgoa River basins,

Fig. 1 Location of sampling sites on the Darling River and tributaries, New South Wales. Letters A–D represent site groups. 2010 Blackwell Publishing Ltd, Freshwater Biology, 55, 1780–1800

Flow alteration and macroinvertebrates in dryland rivers Table 1 List of the study sites and their flow frequencies and sampling periods. Site groups are: A, tributaries with lowlevel water abstraction; B, tributaries with high-level water abstraction; C, freely flowing sites on the Barwon-Darling River; D, weir-pool sites on the BarwonDarling River. Flow frequency is the proportion of days with recorded flow for the period 1 January 2000–31 December 2004. The sampling periods listed for each site are 1 = June 2003; 2 = March 2004; 3 = December 2004

1783

Site group

River

Location

Flow frequency

Sampling periods

A A A B B B B B B C C C D D D D D

Cuttaburra R Paroo R Warrego R bywash Bokhara R Culgoa R Culgoa R Moonie R Narran R Narran R Barwon R Barwon R Darling R Barwon R Barwon R Barwon R Darling R Darling R

Turra Willara Crossing Fords Bridge Bokhara Brenda Upstream of Collerina Gundablouie Narran Park New Angledool Walgett Geera Tilpa Brewarrina Collarenebri Mungindi Bourke Menindee

0.32 0.57 0.56 0.31 0.29 0.44 0.49 0.09 0.22 0.99 1.00 0.90 1.00 0.99 0.96 0.93 1.00

1, 2, 3 1, 2, 3 1, 2 1, 2, 3 1, 2, 3 1, 2, 3 1, 2 1, 2, 3 1, 2, 3 1, 2, 3 1, 3 1, 2, 3 1, 2, 3 1, 2, 3 1 1, 2, 3 1

the latter also known as the Condamine-Balonne. In the Moonie Basin, maximum surface water use under current entitlements is 34% of average water availability, although recent historical use is lower (CSIRO, 2008c), whereas in the Condamine-Culgoa about 53% of average available water is diverted for use (CSIRO, 2008b). Groups C and D consisted of sites on the Barwon and Darling rivers that have near-perennial flow and highly modified flow regimes, with an average of 39% of surface water diverted for use upstream of Bourke (CSIRO, 2008a). Sites in group C are freely flowing, i.e. not located within impounded reaches upstream of weirs, whereas those in group D are within weir pools. Sites in group D have similar discharge regimes to those in group C but water levels are stabilised by downstream weirs. The rivers sampled are meandering, alluvial streams with gentle gradients, fringed mostly by stands of river red gum (Eucalyptus camaldulensis Dehnh.), black box (Eucalyptus largiflorens F. Muell.) and coolibah (Eucalyptus coolabah Blakely & Jacobs). Their substrata are dominated by clay, silt and fine sand, with exposed bedrock in places. Submerged wood is common, and aquatic macrophytes occur patchily.

Sampling design and methods The sites were sampled in three periods: late June 2003 (austral winter), early March 2004 (autumn) and early December 2004 (summer). Five sites were not 2010 Blackwell Publishing Ltd, Freshwater Biology, 55, 1780–1800

sampled in every period (see Table 1) because of time constraints, access problems because of wet weather, and rivers’ being either dry or too flooded to sample. On each site visit the temperature, electrical conductivity (EC), dissolved oxygen concentration (DO) and pH of the water were measured with a Hydrolab Datasonde 4 multiprobe and Surveyor 4 data logger. Turbidity was measured with a Hach 2100P turbidimeter and alkalinity with Titrets hand-held titration kits. Temperature and DO were measured at the water surface and the deepest point that was accessible by wading (depth range 0.3–3.0 m). Other measurements were made at the surface only, and all measurements were made on site. Flow data from gauging stations near the study sites were obtained from the hydrographic data base of the NSW Department of Environment, Climate Change and Water. Three techniques were used to sample macroinvertebrates. Quantitative samples were taken with a metal cylinder 0.5 m high and 0.35 m in diameter. Five cylinder samples were taken per site per sampling period, spaced at 20-m intervals. The cylinder was inserted upright into the river bed in shallow water of the littoral zone, so that the bottom formed a seal with the bed and the top was not submerged. The substratum within the cylinder was disturbed to a depth of approximately 0.1 m, and the cylinder was emptied with a bailer. The contents were sieved with a 0.25-mm or 0.50-mm mesh, and the retained material was placed in a plastic jar together with at least twice its volume of 100% ethanol. In the laboratory,

1784


each sample was washed on a 0.50-mm mesh and distributed in a large white tray from which macroinvertebrates were picked with forceps under magnification. Picking stopped once two passes of the entire tray could be made without further specimens being found. Picked animals were preserved in 70% ethanol and identified to the lowest level possible with published keys and descriptions. Three handnet samples were taken at each site on each occasion: two from edge waters and one from submerged wood. The net had a 0.1-m2 opening and 0.25-mm mesh. Each edge-water sample comprised a series of adjacent sweeps along the substratum running at right angles into the bank from a distance of about 0.5 m, which continued until 10 m of bank had been sampled. Samples from submerged wood were collected for three minutes with sweeps along wood surfaces and the use of a paintbrush to move attached invertebrates from the surfaces into the net opening. All handnet samples were subsampled on site while macroinvertebrates were alive; the samples were placed in large white trays, and macroinvertebrates were picked out by eye with forceps and pipettes, the aim being to recover the maximum number of species within time limits. One edge-water sample was subsampled according to the NSW AUSRIVAS protocol (Turak, Waddell & Johnstone, 2004), so that valid AUSRIVAS O ⁄ E scores could be calculated. Under this protocol, picking must cease when 60 min. have elapsed, but can stop at any time beyond 40 min. if no new taxa have been found in the previous 10 min. The remaining edge-water sample and the wood sample were picked for 30 min. each. Picked animals were preserved in 70% ethanol and identified as described earlier. Collapsible funnel traps, measuring approximately 650 · 470 · 230 mm with two openings of 80 mm diameter and a stretched mesh of 25 mm, were used to target large, mobile macroinvertebrates (decapod Crustacea). Twenty traps were submerged at each site on each sampling occasion for approximately two hours (average 2 h 14 min., SD 13 min.), baited with one tablespoonful of tinned commercial dog food per trap in a pouch of nylon stocking material. Traps were placed within 10 m of the bank in an area set away from other sampling activity (to minimise disturbance) and spaced at 10-m intervals. Captured animals were identified to species and released.

Data analysis The long-term flow frequency for each site was calculated as the proportion of days with non-zero flow in the 5-year period from 1 January 2000 to 31 December 2004. Differences in flow frequency among the site groups were tested by one-way analysis of variance (A N O V A ). A N O V A was also used to test for statistically significant differences among site groups and sampling periods in water-quality variables and summary measures of macroinvertebrate assemblages (numbers of individuals and taxa) for the 15 sites that were sampled more than once. Group was treated as a fixed factor and sampling period as a repeated measure, allowing for possible statistical non-independence of observations made at the same site at different times. A log-linear A N O V A with Poisson error was also applied to the number of taxa per site for data combined from all three sampling periods. All A N O V A s of numbers of taxa were performed at the species level and with data amalgamated to the genus and family levels. For each level, only taxa identified to that level were included in the analysis; for example, for the species level only specimens that could be assigned to a formally described species were included. Restricted maximum likelihood estimation, which adjusts for unequal sample sizes among treatments, was used to test hypotheses (Verbeke & Molenberghs, 2000). Pairwise differences between means were investigated with Tukey’s HSD with Kramer’s adjustment for unequal sample sizes (Kramer, 1956; Neter et al., 1996). Residual diagnostics were examined to assess the adequacy of the A N O V A models, and variables were transformed as required to either logarithms or square roots. Permutational multivariate analysis of variance (M A N O V A ) (Anderson, 2001; McArdle & Anderson, 2001), as implemented by the Vegan package (Oksanen et al., 2008), was used to test for significant compositional differences in macroinvertebrate assemblages among site groups and sampling periods. Constrained analysis of principal coordinates (CAP) was used to examine and display differences in assemblage composition among the site groups for each period. CAP is an ordination method that enables non-Euclidean dissimilarity measures to be used to test a priori hypotheses of differences among explanatory variables (Anderson & Willis, 2003). 2010 Blackwell Publishing Ltd, Freshwater Biology, 55, 1780–1800

Flow alteration and macroinvertebrates in dryland rivers Where the explanatory variables are a design matrix of factors, CAP is a generalised discriminant analysis based on distances (Anderson & Robinson, 2003). The test statistic used to examine the null hypothesis of no differences among site groups was the sum of canonical correlations (Anderson & Robinson, 2003). Permutational M A N O V A and CAP analyses were performed with R version 2.7.1 (R Development Core Team, 2007), and for both analyses, counts of macroinvertebrate taxa were transformed to fourth roots and the Bray-Curtis dissimilarity measure was used. Correlations of transformed counts of individual taxa with the first two discriminant axes were superimposed on CAP plots to identify taxa that were associated with particular site groups (Anderson & Willis, 2003). Values of the O ⁄ E index were calculated with AUSRIVAS software, which generates probabilities of detection of individual macroinvertebrate families (and some non-family taxa) when a prescribed sampling and subsampling protocol is applied to a particular site and mesohabitat (e.g. a riffle) in a particular Australian state or region and calendar season (Simpson & Norris, 2000). These predictions are derived from macroinvertebrate data collected at reference sites, mainly in the mid-1990s. Site- and habitat-specific environmental data contributed by the user determine which reference data are used in a particular case, and the weightings they are given. The O ⁄ E index is calculated by dividing E, the sum of predicted probabilities of taxa with probabilities above a specified threshold (usually 50%) into O, the number of these higher-probability taxa that were actually recorded (Clarke et al., 1996). The only AUSRIVAS model applicable to our study area requires combined data from two edge-water samples, one taken in spring (defined as 15 September-15 December; Turak et al., 2004) and one in autumn (15 March-15 June). Data were therefore combined from the subsamples collected with the AUSRIVAS method in June and December, and also from the subsamples obtained with this method in March and December; this allowed two O ⁄ E values to be derived for each site that was sampled in all three periods. Sampling dates were slightly outside the defined AUSRIVAS sampling windows in June (2– 10 days late) and March (5–13 days early), but it seemed unlikely that such small time discrepancies would have an appreciable influence on O ⁄ E values. 2010 Blackwell Publishing Ltd, Freshwater Biology, 55, 1780–1800

1785

Macroinvertebrate data were amalgamated to the taxonomic levels required for AUSRIVAS in NSW, and the two-season western-NSW edge-water model was run via the AUSRIVAS Internet site (http:// ausrivas.canberra.edu.au/ – accessed 31 August 2007). O ⁄ E scores were calculated with a threshold probability of 50%. A repeated-measures ANOVA was applied to AUSRIVAS O ⁄ E50 values in the same manner as for other variables, except with two seasonal combinations instead of three sampling periods as the trials factor.

Results Physical and chemical environment Hydrological conditions during 2003 and 2004 were highly varied both among the study sites and over time, as is typical for these dryland rivers (Fig. 2). Flows were generally low and declining during the first sampling period in June 2003, whereas the second sampling period in March 2004 was amid very high flows extending from January to April. The third sampling period in December 2004 was at the end of a long dry spell when flows were increasing rapidly at many sites. The long-term flow frequency varied significantly among site groups (F3,13 = 41.4, P < 0.001); post hoc Tukey tests showed that frequency was not significantly different (P > 0.05) between groups A and B or between groups C and D, but all other pairwise comparisons were significant (P < 0.001). Flow at all sites in the Barwon and Darling rivers (groups C and D) was continuous or nearly so (frequency 0.9–1.0), whereas flow in the other rivers (groups A and B) was intermittent (frequency 0.1–0.6), regardless of the level of waterresource development (Table 1). Most of the measured water-quality variables had a wide range of values across the entire study (Table 2). Repeated-measures A N O V A revealed significant differences among site groups in only EC (log-transformed; F3,10 = 7.3; P < 0.01) and turbidity (log-transformed; F3,10 = 7.5; P < 0.01). Differences among sampling periods were significant for temperature (F2,15 = 164.5 and 398.0 at the surface and bottom, respectively; P < 0.001 in both cases), pH (F2,15 = 12.3; P < 0.001) and alkalinity (log-transformed; F2,15 = 7.5; P < 0.01). Interactions between site group and sampling period were not significant for any variable.


1786 Maximum discharge (ML day–1)

80 000

60 000

40 000

20 000

01-Jan-2003

01-Apr-2003

01-Jul-2003

01-Oct-2003

01-Jan-2004

01-Apr-2004

01-Jul-2004

01-Oct-2004

01-Jan-2003

01-Apr-2003

01-Jul-2003

01-Oct-2003

01-Jan-2004

01-Apr-2004

01-Jul-2004

01-Oct-2004

0

Median discharge (ML day–1)

8000

6000

4000

2000

0

Minimum discharge (ML day–1)

200

150

100

50

01-Oct-2004

01-Jul-2004

01-Apr-2004

01-Jan-2004

01-Oct-2003

01-Jul-2003

01-Apr-2003

01-Jan-2003

0

Date

Fig. 2 Maximum (top), median (middle) and minimum (bottom) discharge of the 17 sampling sites on each day from 1 January 2003 to 31 December 2004. Vertical white bars indicate sampling periods.

Table 2 Median values and ranges (in parentheses) of environmental variables for sites in each group across all sampling periods. See Table 1 for group descriptions Variable

Site group A

Site group B

Site group C

Site group D

Temperature (surface) (C) Temperature (deep) (C) Electrical conductivity (lS cm)1) Turbidity (NTU) Dissolved oxygen (surface) (mg L)1) Dissolved oxygen (deep) (mg L)1) pH Alkalinity (mg L)1 CaCO3)

18 17 108 485 5.4 5.4 7.0 38

23 23 143 581 6.1 4.4 6.9 70

25 25 328 171 7.6 6.0 7.3 100

21 19 332 80 6.5 5.3 7.3 80

(9–28) (9–28) (74–180) (88–709) (3.7–7.8) (4.2–8.3) (6.4–7.2) (25–77)

(10–29) (9–29) (75–438) (264–3680) (3.4–8.8) (0.1–8.7) (6.7–8.0) (25–160)

(11–27) (11–27) (150–421) (71–774) (5.6–8.3) (5.1–8.9) (6.6–8.3) (50–110)

(11–31) (12–27) (102–2030) (20–1430) (3.9–10.2) (2.8–7.6) (6.9–8.9) (27–160)

2010 Blackwell Publishing Ltd, Freshwater Biology, 55, 1780–1800

Flow alteration and macroinvertebrates in dryland rivers Tukey’s HSD comparisons of site-group means showed that EC was significantly lower in group A than in group B, which had significantly lower EC than groups C and D. Turbidity was significantly lower in groups C and D than in group A, which had significantly lower turbidity than group B. pH was significantly higher in December 2004 than in the previous two sampling periods, and alkalinity gradually decreased over time. Surface and near-bed water temperatures were significantly lower in June 2003 than in the other two sampling periods, which did not differ significantly from each other.

1787

tion of genus-level data by CAP separated the site groups in two dimensions, apart from overlap of groups A and B in June 2003 (Fig. 5). Correlation vectors of genus abundances with the first two discriminant axes showed strong associations of certain genera with particular site groups in particular sampling periods. However, these relationships were inconsistent among periods, except for repeated affinities of the prawn Macrobrachium and its isopod ectoparasite Tachaea for site groups C and D (Fig. 5). Most common genera were well represented in more than one site group (Table 4), and the same was true at the species and family levels (data not shown).

Cylinder samples A total of 2842 macroinvertebrate specimens was retrieved with the quantitative cylinder method. The number of individuals per site per sampling occasion varied from 2 to 537 for five samples combined, or 4 to 1074 m)2. Nearly, all specimens (>99%) could be identified to family level with published keys, and 93% to genus level, but only 50% to species level. The identifiable taxa at the three levels comprised 38 families, 71 genera and 56 species. Repeated-measures A N O V A of data from individual sampling periods showed that neither the number of individuals (equivalent to density; log-transformed) nor the number of species, genera or families per site (square-root-transformed) differed significantly among site groups; however, all differed among periods (F2,19 = 4.4, 5.4, 8.2 and 7.4; P < 0.05, 0.05, 0.01 and 0.01, respectively). Interaction terms were not significant. Tukey’s HSD tests revealed that both number of individuals and taxon richness (at all levels) were highest in June 2003 and lowest in March 2004 (illustrated for individuals and genera in Fig. 3). A N O V A of composite data from all three periods showed that the total number of genera per site differed significantly among site groups (likelihood ratio test:v23 = 18.7, P < 0.001) but not the number of species or families. On average, composite genus richness was lower at sites in group C than at sites in the other groups (Fig. 4). Permutational M A N O V A of cylinder-sample data revealed significant differences among site groups and sampling periods and for their interactions at the family and genus levels of taxonomic resolution (Table 3). At the species level, significant differences were detected only among sampling periods. Ordina 2010 Blackwell Publishing Ltd, Freshwater Biology, 55, 1780–1800

Handnet samples The subsamples of the handnet samples contained a total of 13 798 macroinvertebrate specimens, of which 90% were identified to family level, 83% to genus level and 43% to species level. There were 87 identifiable families, 143 genera and 125 species. The common taxa in handnet subsamples from submerged wood were nearly always also frequent in the handnet subsamples from edge waters, and only a few taxa, all rare, were collected solely in the wood subsamples. Repeated-measures A N O V A of data from individual sampling periods (composited for the three subsamples collected per site in each period) showed that the number of taxa per site (square-root transformed) differed significantly among site groups only at genus level (F3,11 = 3.9; P < 0.05). The numbers of individuals, species, genera and families per site (square-roottransformed) all differed significantly among sampling periods (F2,19 = 32.1, 36.2, 12.5 and 11.5, respectively; P < 0.001 in all cases). The interaction term was significant only for the number of species (F6,19 = 3.4; P < 0.05). Tukey’s HSD test showed that the number of genera was significantly lower in site group C than in group D, which in turn had significantly fewer genera than groups A and B, which did not differ significantly from each other. The temporal patterns in numbers of species, genera and families were similar to those found for the cylinder samples, with the fewest collected in March 2004 (illustrated for genera in Fig. 3). No significant pairwise differences in the number of individuals were detected with Tukey’s procedure, but the temporal pattern was the same as for taxonomic richness (Fig. 3).


1788 50

50

Handnet samples - June 2003

Cylinder samples - June 2003 40

40

30

30

20

20

10

10

0

0 1

10

100

1000

50

1

100

1000

50

Cylinder samples - March 2004

No. of genera

10

Handnet samples - March 2004

40

40

30

30

20

20

10

10

0

0 1

10

100

1000

50

1

10

100

1000

50

Cylinder samples - December 2004

Handnet samples - December 2004

40

40

30

30

20

20

10

10

0

0 1

10

100

1000

1

10

No. of individuals

ANOVA of composite data for all three sampling periods showed that the total number of taxa per site differed significantly among site groups at the species (likelihood ratio test:v23 = 12.9, P < 0.01), genus (v23 = 9.5, P < 0.05) and family (v23 = 10.9, P < 0.05) levels. On average, sites in groups A and B yielded more taxa than those in groups C and D (Fig. 4). Permutational M A N O V A of handnet data showed significant differences among site groups and sam-

100

1000

Fig. 3 Relationships between the number of individual macroinvertebrates and number of macroinvertebrate genera in cylinder and handnet samples pooled by site for each sampling period. Group membership of sites is shown as follows: n = A, m = B, • = C and ¤ = D.

pling periods at all three taxonomic levels, but with an interaction between these factors at the species level (Table 3). Ordination of genus-level data by CAP separated the site groups in two dimensions, except for overlap of groups A and B in December 2004 (Fig. 5). As for the cylinder samples, the common genera seldom showed consistent fidelity to particular site groups (Fig. 5; Table 5). This was also the case at the species and family levels (data not shown). 2010 Blackwell Publishing Ltd, Freshwater Biology, 55, 1780–1800

Flow alteration and macroinvertebrates in dryland rivers

1789

Trap samples

75 Cylinder samples

The catch in the funnel traps ranged from 0 to 139 specimens per site per sampling period. Three macroinvertebrate species were captured: the crab Holthuisana (Austrothelphusa) transversa (von Martens) (Parathelphusidae), the crayfish Cherax destructor Clark (Parastacidae) and the prawn Macrobrachium australiense Holthuis (Palaemonidae). The catch of C. destructor varied significantly among site groups (F3,11 = 4.8; P < 0.05) but not the total catch or that of M. australiense (all variables log-transformed). C. destructor catches were typically higher for site groups A and B than for groups C and D (Fig. 6). H.(A.) transversa was caught with this method at only three sites in one period each, and therefore the catch of this species was not amenable to A N O V A . The catch of C. destructor and M. australiense and the total catch varied significantly among sampling periods (F2,19 = 7.1, 22.4 and 20.3; P < 0.01, 0.001 and 0.001, respectively), and Tukey’s HSD test showed that significantly fewer individuals were trapped in June 2003 than in the other two periods (Fig. 6). Interaction terms were not significant. Trap data were not subjected to M A N O V A or CAP because of the small number of species captured by this method.

50

No. of genera per site

25

0 A

B

C

D

75 Handnet samples

50

25

AUSRIVAS scores Repeated-measures A N O V A showed that AUSRIVAS O ⁄ E50 values did not differ significantly among site groups or combinations of sampling periods. The median value across all sites and sampling periods was 0.91 (range was from 0.61 to 1.07), close to the value of 1.0 expected under equivalence to reference data.

0 A

B

C

D

Site group

Fig. 4 Numbers of macroinvertebrate genera in cylinder and handnet samples pooled by site for all sampling periods combined. Means and 95% confidence intervals are shown for those sites in each site group that were sampled in all three periods.

Table 3 Pseudo-F statistics from permutational M A N O V A of macroinvertebrate assemblages sampled with two methods and analysed at three taxonomic levels. Statistically significant values are in bold. ns, not significant; *0.01 < P < 0.05; **0.001 < P < 0.01; ***P < 0.001. The amounts of variation in the data not explained by the models are also shown Sampling method

Taxonomic level

Pseudo-F3,8 (site group)

Pseudo-F2,24 c(sampling period)

Pseudo-F6,24 (interaction)

Unexplained variation (%)

Cylinder Cylinder Cylinder Handnet Handnet Handnet

Species Genus Family Species Genus Family

1.4ns 2.0** 2.1* 2.0*** 2.1** 2.4***

2.4** 4.1*** 5.3*** 8.9*** 7.1*** 7.5***

1.1ns 1.4* 1.6* 1.4* 1.2ns 1.3ns

58 49 45 41 46 43


1790


Fig. 5 Ordination of sites by constrained analysis of principal coordinates of genuslevel data from cylinder and handnet samples in each sampling period. Group membership of sites is shown as follows: n = A, m = B, • = C and ¤ = D. Asterisks indicate centroids and ellipses or lines indicate approximate 95% confidence limits for each group. The test statistic tr(Qm’HQm) is the sum of the canonical correlations: ns, not significant; *0.01 < P < 0.05; **0.001 < P < 0.01; ***P < 0.001. Vectors indicate correlations of individual genera, abbreviated as follows: Abl, Ablabesmyia; Agr, Agraptocorixa; Ani, Anisops; Aqu, Aquarius; Bra, Branchiura; Car, Caridina; Cer, Ceriodaphnia; Che, Cherax; Chi, Chironomus; Cla, Cladotanytarsus; Clo, Cloeon; Cri, Cricotopus; Cry, Cryptochironomus; Dic, Dicrotendipes; Ecn, Ecnomus; Eni, Enithares; Eno, Enochrus; Ere, Eretes; Grp, Graptoleberis; Har, Harnischia; Hol, Holthuisana; Kie, Kiefferulus; Lar, Larsia; Lcc, Laccotrephes; Lio, Liodessus; Mac, Macrobrachium; Mes, Mesovelia; Mne, Micronecta; Oec, Oecetis; Ort, Orthetrum; Pol, Polypedilum; Pro, Procladius; Ptr, Paratrichocladius; Tac, Tachaea; Tan, Tanytarsus; Tas, Tasmanocoenis; Ten, Tenagogerris; Trp, Triplectides [Correction added on 14 June 2010, after first online publication: Panels in Figure 5 have been rearranged, with the cylinder samples in the left column, the handnet samples in the right column].

Discussion Taxonomic richness With combined data from all three sampling methods, 139 species were recorded in this study, a similar total to the 109 species reported by Davis, Harrington & Friend (1993) from arid-zone streams in Central Australia. However, only about half of the specimens

we collected could be identified to species level. Marshall, Steward & Harch (2006b), who sampled macroinvertebrates from the Queensland part of the Condamine-Culgoa River system upstream of our study area, reported a similar proportion (54%), even with identification by a suite of specialist taxonomists. Taxonomic resolution is limited mainly because keys for some higher taxa do not extend to species level, or 2010 Blackwell Publishing Ltd, Freshwater Biology, 55, 1780–1800

Flow alteration and macroinvertebrates in dryland rivers

1791

Table 4 Average density (No.m)2) of the 30 most common genera (in descending order of overall abundance) recorded from cylinder samples in each site group in each sampling period (1 = June 2003; 2 = March 2004; 3 = December 2004). Blank, 0; , 0–2; , 2–10; , 10–50; , >50

do so only with characters confined to mature specimens or one sex. If this taxonomic impediment could be removed, the species tally for these dryland rivers would be much higher. So while these rivers may not be as rich in macroinvertebrate biodiversity as those in wetter parts of Australia, neither are they severely impoverished. Overall taxonomic richness per site, calculated with data composited from all three sampling periods, differed significantly among the site groups at the genus level for cylinder sampling and at the species, genus and family levels for handnet sampling (Table 6). However, when richness per site was calculated with data from individual sampling periods, it varied significantly among the groups only for handnet samples at the genus level. Sheldon & Thoms 2010 Blackwell Publishing Ltd, Freshwater Biology, 55, 1780–1800

(2006) observed little difference in taxonomic richness (a range of 54–67 taxa) among four Australian dryland rivers with differences in water regimes related to both natural factors and water-resource development, each sampled in a single period. However, they speculated that differences in richness might be better revealed by multi-period sampling because such sampling can capture the replacement of species through time. Our results confirm this prediction and suggest that sampling on only a single occasion can paint a misleading picture of the comparative biodiversity of dryland rivers with different hydrological regimes. Composite richness was highest in the intermittent rivers and lowest in free-flowing reaches of the nearperennial Barwon and Darling rivers. Intermittent

1792


Table 5 Average number of individuals per site of the 30 most common genera (in descending order of overall abundance) recorded from handnet subsamples in each sampling period (1 = June 2003; 2 = March 2004; 3 = December 2004). Blank, 0; , 0–2; , 2–10; ,10–50; , >50

rivers may support more taxa over time because they have a greater temporal diversity of environmental conditions, fluctuating between periods of flow when longitudinal connectivity is high and periods without flow when they fragment into separated waterholes. These waterholes may follow individualistic trajectories of community change driven by variability of within-waterhole environments and chance colonisation and extinction (Sheldon, Boulton & Puckridge, 2002, 2003). Conversely, free-flowing, near-perennial rivers are more constant environments that fragment only rarely. The relatively high long-term richness in the weir pools of the near-perennial rivers may relate to temporal environmental diversity caused by alternation between lotic conditions at times of high flow and essentially lentic conditions at times of low flow.

Variation in abundance and assemblage composition The density of macroinvertebrates in the quantitative cylinder samples and the number of individuals in the non-quantitative handnet subsamples did not differ significantly among the four hydrological regimes. However, the semi-quantitative trapping method, which was designed to target large-bodied Crustacea, produced significant differences in catch per unit effort among site groups for the crayfish Cherax destructor. This species had a higher catch rate in the intermittent rivers than in the perennial ones and is well adapted to temporary waters by its ability to survive in burrows when surface water evaporates (Johnston & Robson, 2009). Conversely, no significant difference was found in the catch per unit effort of the prawn Macrobrachium australiense, which is poorly 2010 Blackwell Publishing Ltd, Freshwater Biology, 55, 1780–1800

Flow alteration and macroinvertebrates in dryland rivers 5

5

June 2003

June 2003

4

4

3

3

2

2

1

1

0

0

–1

–1 A

B

C

D

A

5

B

C

D

5

March 2004 Loge (No. of M. australiense + 1)

March 2004 Loge (No. of C. destructor + 1)

1793

4

3

2

1

0

–1

4

3

2

1

0

–1 A

B

C

D

A

5

B

C

D

5

December 2004

December 2004

4

4

3

3

2

2

1

1

0

0

–1

–1 A

B

C

D

A

B

C

D

Site group

Fig. 6 Catch of Cherax destructor and Macrobrachium australiense in trap samples pooled by site for each sampling period. Means and 95% confidence intervals are shown for those sites in each site group that were sampled in the applicable period.

adapted to temporary waters, lacking a desiccationresistant life-history stage and not constructing burrows (Cook, Bunn & Hughes, 2002). However, this species has strong dispersal abilities (e.g. Lee & Fielder, 1979) and appears to readily recolonise intermittent streams after surviving dry periods in refugial pools (Carini, Hughes & Bunn, 2006). The crab 2010 Blackwell Publishing Ltd, Freshwater Biology, 55, 1780–1800

Holthuisana (Austrothelphusa) transversa, which was captured only in intermittent rivers, is also a burrowing species capable of aestivation (Greenaway, 1980). Invertebrate numbers in cylinder samples and handnet subsamples were both low in the middle sampling period in March 2004. The March catches might have been reduced by the dilution of invertebrate

1794


Table 6 Summary of significant differences (4) found in statistical tests comparing macroinvertebrate variables among site groups and sampling periods. TNP, test not possible Difference among site groups

Sampling method

Data form

Variable

Cylinder Cylinder Cylinder Cylinder Cylinder Cylinder Cylinder Cylinder Cylinder Cylinder Hand net Hand net Hand net Hand net Hand net Hand net Hand net Hand net Hand net Hand net Hand net Funnel trap Funnel trap Funnel trap

Individual sampling periods Individual sampling periods Individual sampling periods Individual sampling periods Individual sampling periods Individual sampling periods Individual sampling periods Three-period composite Three-period composite Three-period composite Individual sampling periods Individual sampling periods Individual sampling periods Individual sampling periods Individual sampling periods Individual sampling periods Individual sampling periods Three-period composite Three-period composite Three-period composite Two-period composite Individual sampling periods Individual sampling periods Individual sampling periods

Density Species richness Genus richness Family richness Species composition Genus composition Family composition Species richness Genus richness Family richness Number of individuals Species richness Genera richness Family richness Species composition Genus composition Family composition Species richness Genera richness Family richness AUSRIVAS O ⁄ E score Catch of Cherax destructor Catch of Macrobrachium australiense Total catch

populations in the high flows that occurred before and during this sampling period. In contrast, catches by passive trap sampling were lowest in June 2003, probably because of low water temperatures at this time. For example, Cherax destructor is virtually inactive at water temperatures below 16 C (Withnall, 2000). The suite of macroinvertebrates that was most frequently collected in the study rivers comprised mainly crustaceans (cladocerans; corallanid isopods; atyid shrimps; palaemonid prawns; parastacid crayfish) and insects (mostly coleopterans, especially dytiscids and hydraenids; dipterans, especially chironomids; ephemeropterans, especially baetids and caenids; hempiterans, especially corixids, gerrids, notonectids and veliids; odonates, especially coenagrionids; and trichopterans, especially leptocerids). A closely similar dominant fauna has been reported in other studies of dryland rivers in the Murray-Darling Basin (e.g. Sheldon & Walker, 1998; Sheldon & Thoms, 2006). The composition of macroinvertebrate assemblages differed significantly among the site groups at all taxonomic levels, for both cylinder and handnet sampling, with the exception of the species level for

4 4 4

4 4 4 4 4 4 4 4

Difference among sampling periods 4 4 4 4 4 4 4 TNP TNP TNP 4 4 4 4 4 4 4 TNP TNP TNP 4 4 4

cylinder sampling (Table 6). However, individual taxa seldom showed temporally consistent affinity for particular site groups. Thus, most taxa seem to lack a strong fidelity to particular long-term hydrological regimes, but rather may opportunistically exploit favourable short-term hydraulic conditions. Differences between the dominant taxa recorded from cylinder and handnet sampling in the same site groups and sampling periods can probably be accounted for by differences in habitats sampled by the two methods, as well as the selective subsampling that was applied to the handnet samples. The aim of selective subsampling was to recover as many taxa as possible within time constraints, and therefore it probably under-represented the dominance of the most abundant taxa. In addition, selective subsampling of live animals by eye in the field can exert a bias toward larger-bodied taxa (Nichols & Norris, 2006).

Effects of anthropogenic flow alteration Although macroinvertebrate assemblages varied among the four site groups, differences between 2010 Blackwell Publishing Ltd, Freshwater Biology, 55, 1780–1800

Flow alteration and macroinvertebrates in dryland rivers groups A and B on the intermittently flowing rivers were few, despite group B rivers having considerably more water-resource development. There were no significant overall differences in richness, density or catch per unit effort between these two groups, and compositional differences between them were not always evident in ordination space. We also measured few physical and chemical differences between these two groups. Both electrical conductivity and turbidity were significantly higher in sites of group B than in those of group A, but flow frequency did not differ significantly between the two. In the absence of major abiotic differences, the lack of pronounced differences in macroinvertebrate assemblages between the two groups is not surprising. In addition, macroinvertebrate assemblages in dryland rivers can be quite resistant to hydrological changes brought about by water withdrawal, perhaps because they are well adapted to natural hydrological variation (Miller, Wooster & Li, 2007). However, impacts can be much more severe if hydrological change is coupled with pronounced shifts in water chemistry (Lind, Robson & Mitchell, 2006).

Assessment methods for macroinvertebrates in dryland rivers The AUSRIVAS O ⁄ E50 score failed to distinguish rivers with different degrees and types of hydrological alteration, generally rating sites in all groups as close to reference condition. AUSRIVAS predicts the probability of capture of each macroinvertebrate taxon at a particular place and time via extrapolation from reference sites (Halse et al., 2007). However, the few dryland rivers that remain with little waterresource development are biased to the more arid areas, development having concentrated initially on rivers with more reliable flow and then spread to progressively more intermittent ones. For example, all rivers in our study region with naturally nearperennial flow have substantial water-resource development in their catchments. Consequently, it has been necessary to accept sites with anthropogenically altered hydrology as AUSRIVAS reference sites to include examples of a broad range of river types (e.g. Turak et al., 1999). This compromise imposes an obvious limitation on the ability of AUSRIVAS to detect impacts of hydrological alteration. 2010 Blackwell Publishing Ltd, Freshwater Biology, 55, 1780–1800

1795

In addition, the AUSRIVAS model that we used employs only slope and geographic position (latitude, longitude and altitude) to match assessment sites with reference sites, and thus does not use hydrological variables. It is therefore uncertain how well this model matches assessment and reference sites in terms of the natural variation in hydrological regimes that can have a major bearing on the composition of macroinvertebrate assemblages in dryland rivers (Chakona et al., 2008; Argyroudi et al., 2009). Slope and geographic position may serve to some extent as surrogates for spatial variation in natural hydrology; for example, slope may be correlated with flow regime because headwater streams generally have steeper slopes than high-order streams, and the increase in rainfall from west to east across western NSW may result in some correlation between longitude and natural flow permanence. However, since slope and geographic position do not change between drier and wetter periods, the AUSRIVAS model has no means to vary predictions according to antecedent hydrology prior to sampling. Given all of these limitations, it is not surprising that the AUSRIVAS O ⁄ E50 index did not differentiate our site groups. An alternative way to generate reference macroinvertebrate data is to estimate likely assemblages in the absence of water-resource development via modelling that is not based simply on extrapolation from reference sites. Hydrological models have been developed that can simulate flow regimes in the absence of flow-regulating structures and major abstraction (e.g. Thoms & Parsons, 2003), and if predictive relationships could be developed between flow regimes and macroinvertebrate assemblages, these models might be extended to simulate expected macroinvertebrate faunas in the absence of major development. Chessman & Royal (2004) used data on the environmental tolerances and preferences of macroinvertebrate families to predict their occurrence in rivers with a given annual temperature range, substratum type and flow regime, but characterised flow only as either perennial or intermittent. For arid and semi-arid areas, where most rivers are naturally non-perennial, a more refined consideration of hydrological tolerances and preferences is likely to be required, for example, taking account of resistance to desiccation and capacity for dispersal to recolonise re-wetted sites (e.g. Bonada, Rieradevall & Prat, 2007b).

1796


Modelling of reference data also needs to take account of the temporal variability of macroinvertebrate assemblages. With the exception of AUSRIVAS O ⁄ E, every one of the statistical tests that we applied showed significant differences in macroinvertebrate assemblages among sampling periods, whereas only about half showed significant differences among site groups (Table 6). This dominance of temporal variability over spatial patterns is a common feature of dryland rivers (e.g. Marshall et al., 2006a; Bogan & Lytle, 2007) and requires that reference data are adjusted to take account of antecedent conditions, so that spatial signals related to human impact are not swamped by temporal noise. RIVPACS and AUSRIVAS consider temporal variability by creating predictive models for particular calendar seasons, but in dryland rivers much of the temporal faunal variation is associated with irregular, aseasonal cycles of flood and drought. Sheldon (2005) has suggested that reference data should be adjusted according to the hydrological conditions that precede macroinvertebrate sampling, and this idea warrants testing with long time series of data from a range of dryland rivers with varying levels and types of human disturbance. Our non-quantitative handnet sampling led to many more significant differences among site groups than our quantitative cylinder method (Table 6). Metzeling et al. (2003) likewise found that quantitative methods were less effective than rapid bioassessment methods in detecting a range of human impacts on macroinvertebrate assemblages in wetter parts of south-eastern Australia. The sensitivity of the cylinder method might have been limited by the comparatively low number of organisms it collected, and the performance of this technique could perhaps be improved with a greater sampling effort. However, this would be expensive because the picking of all macroinvertebrate specimens from associated sediment, detritus and macrophyte fragments collected with this technique was time consuming. A possible option is to take a large number of samples, composite them, and sort random subsamples of the composite until a sufficient number of organisms has been picked to provide representative data. However, Metzeling et al. (2003) found that quantitative sampling did not perform well even though in their study it collected substantially more individuals than handnet sampling.

It is possible that estimates of total population numbers would reveal impacts of water-resource development more effectively than measures of density, because development might reduce the total area of wetted habitats without greatly changing the density of macroinvertebrates in the habitat that remains. However, the estimation of total numbers would involve a mammoth effort to measure the area of each type of habitat and density within that habitat, multiplying the two to obtain total numbers. Such an exercise would be beyond the resources of routine monitoring but should be feasible as a research project. Analysis of macroinvertebrate assemblage data by trait composition rather than taxonomic composition might also be useful for assessment of impacts of altered hydrology on Australian dryland rivers. In the Northern Hemisphere, trait composition has been found to vary with differences in hydrology, both among rivers (Bonada, Dole´dec & Statzner, 2007a; Bonada et al., 2007b; Horrigan & Baird, 2008) and between wet and dry periods within rivers (Beˆche, McElravy & Resh, 2006; Beˆche & Resh, 2007). A potential constraint on the use of traits in Australian studies is that many Australian freshwater invertebrate species are endemic, and their traits are not well documented. However, many Australian genera and families also occur in the Northern Hemisphere, and it is possible that traits are sufficiently conservative within these higher taxa that compilations of their traits in North America or Europe (e.g. Poff et al., 2006; Statzner, Bonada & Dole´dec, 2007) could be used in Australia. A possible advantage of trait-based analysis is that it might provide insight into the mechanisms by which hydrological alteration impacts on invertebrate assemblages, for example by highlighting the traits that make particular taxa vulnerable to such alteration. We found appreciably more statistically significant differences among site groups with assemblage data at genus level (five) than at species level (two) or family level (three) (Table 6). The capacity of specieslevel data to differentiate among site groups was probably constrained by inability to identify a large proportion of specimens to that level, whereas the capacity of family-level data might have been limited by variation in environmental tolerances and preferences among taxa within a family (Haase & Nolte, 2008). Marshall et al. (2006b) suggested that familylevel identification is cost-effective for Australian 2010 Blackwell Publishing Ltd, Freshwater Biology, 55, 1780–1800

Flow alteration and macroinvertebrates in dryland rivers dryland rivers, if coupled with information on catch per unit effort, but their assessment was based on capacity to differentiate individual sites rather than hydrological regimes or levels of anthropogenic stress. Moreover, a major influence on their finding was an estimate that genus-level identification is five times as expensive as family-level identification. We believe that this differential may be overstated, because in our experience it can be substantially lower in laboratories where genus-level identification is routinely undertaken and supported with good identification aids such as reference photographs. We suggest that under current taxonomic constraints, the genus level would be an appropriate default for macroinvertebrate-based impact assessments of Australian dryland rivers, with other taxonomic levels being substituted if they are more appropriate or costeffective in meeting the objectives of a particular study.

Acknowledgments We are grateful to the former NSW Department of Primary Industries for macroinvertebrate collecting permits, Peter Bussell and Alan Hassett for information on site access, Sean Dwyer, Adrian Fisher, Geoff Horn, Warren Martin and Terry Mazzer for help with fieldwork, Ecowise and Water’s Edge Consulting for identification of macroinvertebrate specimens, and Eren Turak & Edwina Mesley for information on AUSRIVAS predictor variables. The manuscript was improved through comments from anonymous referees.

References Anderson M.J. (2001) A new method for non-parametric multivariate analysis of variance. Austral Ecology, 26, 32–46. Anderson M.J. & Robinson J. (2003) Generalized discriminant analysis based on distances. Australian & New Zealand Journal of Statistics, 45, 301–318. Anderson M.J. & Willis T.J. (2003) Canonical analysis of principal coordinates: a useful method of constrained ordination for ecology. Ecology, 84, 511–525. Argyroudi A., Chatzinikolaou Y., Poirazidis K. & Lazaridou M. (2009) Do intermittent and ephemeral Mediterranean rivers belong to the same river type? Aquatic Ecology, 43, 465–476. 2010 Blackwell Publishing Ltd, Freshwater Biology, 55, 1780–1800

1797

Arthington A.H. (1996) The effects of agricultural land use and cotton production on tributaries of the Darling River, Australia. Geo Journal, 40, 115–125. Beˆche L.A. & Resh V.H. (2007) Biological traits of benthic macroinvertebrates in California Mediterraneanclimate streams: long-term annual variability and trait diversity patterns. Fundamental and Applied Limnology, 169, 1–23. Beˆche L.A., McElravy E.P. & Resh V.H. (2006) Long-term seasonal variation in the biological traits of benthicmacroinvertebrates in two Mediterranean-climate streams in California, U.S.A. Freshwater Biology, 51, 56–75. Bogan M.T. & Lytle D.A. (2007) Seasonal flow variation allows ‘time-sharing’ by disparate aquatic insect communities in montane desert streams. Freshwater Biology, 52, 290–304. Bonada N., Dole´dec S. & Statzner B. (2007a) Taxonomic and biological trait differences of stream macroinvertebrate communities between Mediterranean and temperate regions: implications for future climate scenarios. Global Change Biology, 13, 1658– 1671. Bonada N., Rieradevall M. & Prat N. (2007b) Macroinvertebrate community structure and biological traits related to flow permanence in a Mediterranean river network. Hydrobiologia, 589, 91–106. Boulton A.J. (1999) Why variable flows are needed for invertebrates of semi-arid rivers. In: A Free-Flowing River: The Ecology of the Paroo River (Ed. R.T. Kingsford), pp. 113–128. National Parks and Wildlife Service, Sydney, N.S.W., Australia. Boulton A.J., Sheldon F. & Jenkins K.M. (2006) Natural disturbance and aquatic invertebrates in desert rivers. In: Ecology of Desert Rivers (Ed. R.T. Kingsford), pp. 133–153. Cambridge University Press, Cambridge. Bunn S.E., Thoms M.C., Hamilton S.K. & Capon S.J. (2006) Flow variability in dryland rivers: boom, bust and the bits in between. River Research and Applications, 22, 179–186. Carini G., Hughes J.M. & Bunn S.E. (2006) The role of waterholes as ‘refugia’ in sustaining genetic diversity and variation of two freshwater species in dryland river systems (Western Queensland, Australia). Freshwater Biology, 51, 1434–1446. Chakona A., Phiri C., Magadza C.H.D. & Brendonck L. (2008) The influence of habitat structure and flow permanence on macroinvertebrate assemblages in temporary rivers of northwestern Zimbabwe. Hydrobiologia, 607, 199–209. Chessman B.C. & Royal M.J. (2004) Bioassessment without reference sites: use of environmental filters

1798


to predict natural assemblages of river macroinvertebrates. Journal of the North American Benthological Society, 23, 599–615. Chessman B.C., Thurtell L.A. & Royal M.J. (2006) Bioassessment in a harsh environment: a comparison of macroinvertebrate assemblages at reference and assessment sites in an Australian inland river system. Environmental Monitoring and Assessment, 119, 303–330. Choy S.C., Thomson C.B. & Marshall J.C. (2002) Ecological condition of central Australian arid-zone rivers. Water Science and Technology, 45, 225–232. Clarke R.T., Furse M.T., Wright J.F. & Moss D. (1996) Derivation of a biological quality index for river sites: comparison of the observed with the expected fauna. Journal of Applied Statistics, 23, 311–332. Clarke R.T., Wright J.F. & Furse M.T. (2003) RIVPACS models for predicting the expected macroinvertebrate fauna and assessing the ecological quality of rivers. Ecological Modelling, 160, 219–233. Cook B.D., Bunn S.E. & Hughes J.M. (2002) Genetic structure and dispersal of Macrobrachium australiense (Decapoda: Palaemonidae) in western Queensland, Australia. Freshwater Biology, 47, 2098–2112. CSIRO (2007a) Water Availability in the Paroo. A Report to the Australian Government from the CSIRO MurrayDarling Basin Sustainable Yields Project. Commonwealth Scientific and Industrial Research Organisation, Canberra. CSIRO (2007b) Water Availability in the Warrego. A Report to the Australian Government from the CSIRO MurrayDarling Basin Sustainable Yields Project. Commonwealth Scientific and Industrial Research Organisation, Canberra. CSIRO (2008a) Water Availability in the Barwon-Darling. A Report to the Australian Government from the CSIRO Murray-Darling Basin Sustainable Yields Project. Commonwealth Scientific and Industrial Research Organisation, Canberra. CSIRO (2008b) Water Availability in the Condamine-Balonne. A Report to the Australian Government from the CSIRO Murray-Darling Basin Sustainable Yields Project. Commonwealth Scientific and Industrial Research Organisation, Canberra. CSIRO (2008c) Water Availability in the Moonie. A Report to the Australian Government from the CSIRO MurrayDarling Basin Sustainable Yields Project. Commonwealth Scientific and Industrial Research Organisation, Canberra. Davis J.A., Harrington S.A. & Friend J.A. (1993) Invertebrate communities of relict streams in the arid zone: the George Gill Range, Central Australia. Australian Journal of Marine and Freshwater Research, 44, 483–505.

Greenaway P. (1980) Water balance and urine production in the Australian arid-zone crab Holthuisana transversa. Journal of Experimental Biology, 87, 237–246. Haase R. & Nolte U. (2008) The invertebrate species index (ISI) for streams in southeast Queensland, Australia. Ecological Indicators, 8, 599–613. Halse S.A., Scanlon M.D., Cocking J.S., Smith M.J. & Kay W.R. (2007) Factors affecting river health and its assessment over broad geographic ranges: the Western Australian experience. Environmental Monitoring and Assessment, 134, 161–175. Hauer F.R. & Lorang M.S. (2004) River regulation, decline of ecological resources, and potential for restoration in a semi-arid lands river in the western USA. Aquatic Sciences, 66, 388–401. Horrigan N. & Baird D.J. (2008) Trait patterns of aquatic insects across gradients of flow-related factors: a multivariate analysis of Canadian national data. Canadian Journal of Fisheries and Aquatic Sciences, 65, 670–680. Johnston K. & Robson B.J. (2009) Habitat use by five sympatric Australian freshwater crayfish species (Parastacidae). Freshwater Biology, 54, 1629–1641. Kingsford R.T. (1999) Managing the water of the Border Rivers in Australia: irrigation, Government and the wetland environment. Wetlands Ecology and Management, 7, 25–35. Kingsford R.T. (2000a) Protecting rivers in arid regions or pumping them dry? Hydrobiologia, 427, 1–11. Kingsford R.T. (2000b) Ecological impacts of dams, water diversions and river management on floodplain wetlands in Australia. Austral Ecology, 25, 109–127. Kingsford R.T. & Johnson W. (1998) Impact of water diversions on colonially-nesting waterbirds in the Macquarie Marshes of arid Australia. Colonial Waterbirds, 21, 159–170. Kingsford R.T. & Thomas R.F. (2004) Destruction of wetlands and waterbird populations by dams and irrigation on the Murrumbidgee River in arid Australia. Environmental Management, 34, 383–396. Kingsford R.T., Boulton A.J. & Puckridge J.T. (1998) Challenges in managing dryland rivers crossing political boundaries: lessons from Cooper Creek and the Paroo River, central Australia. Aquatic Conservation: Marine and Freshwater Ecosystems, 8, 361–378. Kingsford R.T., Thomas R.F. & Curtin A.L. (2001) Conservation of wetlands in the Paroo and Warrego River catchments in arid Australia. Pacific Conservation Biology, 7, 21–33. Kingsford R.T., Lemly A.D. & Thompson J.R. (2006) Impacts of dams, river management and diversions on desert rivers. In: Ecology of Desert Rivers (Ed. R.T. 2010 Blackwell Publishing Ltd, Freshwater Biology, 55, 1780–1800

Flow alteration and macroinvertebrates in dryland rivers Kingsford), pp. 203–247. Cambridge University Press, Cambridge. Kramer C.Y. (1956) Extension of multiple range tests to group means with unequal numbers of replications. Biometrics, 12, 307–310. Lee C.L. & Fielder D.R. (1979) A mass migration of the freshwater prawn, Macrobrachium australiense Holthuis, 1950 (Decapoda, Palaemonidae). Crustaceana, 37, 219– 222. Lemly A.D., Kingsford R.T. & Thompson J.R. (2000) Irrigated agriculture and wildlife conservation: conflict on a global scale. Environmental Management, 25, 485– 512. Lind P.R., Robson B.J. & Mitchell B.D. (2006) The influence of reduced flow during a drought on patterns of variation in macroinvertebrate assemblages across a spatial hierarchy in two lowland rivers. Freshwater Biology, 51, 2282–2295. Maltchik L. & Medeiros E.S.F. (2006) Conservation importance of semi-arid streams in north-eastern Brazil: implications of hydrological disturbance and species diversity. Aquatic Conservation: Marine and Freshwater Ecosystems, 16, 665–677. Marshall J.C., Sheldon F., Thoms M. & Choy S. (2006a) The macroinvertebrate fauna of an Australian dryland river: spatial and temporal patterns and environmental relationships. Marine and Freshwater Research, 57, 61–74. Marshall J.C., Steward A.L. & Harch B.D. (2006b) Taxonomic resolution and quantification of freshwater macroinvertebrate samples from an Australian dryland river: the benefits and costs of using species abundance data. Hydrobiologia, 572, 171– 194. McArdle B.H. & Anderson M.J. (2001) Fitting multivariate models to community data: a comment on distance-based redundancy analysis. Ecology, 82, 290–297. McMahon T.A. & Finlayson B.L. (2003) Droughts and anti-droughts: the low flow hydrology of Australian rivers. Freshwater Biology, 48, 1147–1160. Metzeling L., Chessman B., Hardwick R. & Wong V. (2003) Rapid assessment of rivers using macroinvertebrates: the role of experience, and comparisons with quantitative methods. Hydrobiologia, 510, 39–52. Miller S.W., Wooster D. & Li J. (2007) Resistance and resilience of macroinvertebrates to irrigation water withdrawals. Freshwater Biology, 52, 2494–2510. Neter J., Kutner M.H., Nachtscheim C.J. & Wasserman W. (1996) Applied Linear Statistical Models. Irwin, Chicago. Nichols S.J. & Norris R.H. (2006) River condition assessment may depend on the sub-sampling method: field live-sort versus laboratory sub-sampling of 2010 Blackwell Publishing Ltd, Freshwater Biology, 55, 1780–1800

1799

invertebrates for bioassessment. Hydrobiologia, 572, 195–213. Nilsson C., Reidy C.A., Dynesius M. & Revenga C. (2005) Fragmentation and flow regulation of the world’s large river systems. Science, 308, 405–408. Norris R.H., Linke S., Prosser I., Young W.J., Liston P., Bauer N., Sloane N., Dyer F. & Thoms M. (2007) Verybroad-scale assessment of human impacts on river condition. Freshwater Biology, 52, 959–976. Ogden R.W., Thoms M.C. & Levings P.L. (2002) Nutrient limitation of plant growth on the floodplain of the Narran River, Australia: growth experiments and a pilot soil survey. Hydrobiologia, 489, 277–285. Oksanen J., Kindt R., Legendre P., O’Hara B., Simpson G.L., Henry M., Stevens H. & Wagner H. (2008) Vegan – Community Ecology Package. R package version 1.13-1. http://vegan.r-forge.r-project.org/. Poff N.L., Olden J.D., Vieira N.K.M., Finn D.S., Simmons M.P. & Kondratieff B.C. (2006) Functional trait niches of North American lotic insects: traits-based ecological applications in light of phylogenetic relationships. Journal of the North American Benthological Society, 25, 730–755. Poff N.L., Olden J.D., Merritt D.M. & Pepin D.M. (2007) Homogenization of regional river dynamics by dams and global biodiversity implications. Proceedings of the National Academy of Sciences, 104, 5732–5737. Porter J.W. (2002) Impacts of water management options on flows in the Condamine River in southern Queensland. Water Science and Technology 45, 233–240. Puckridge J.T., Walker K.F. & Costelloe J.F. (2000) Hydrological persistence and the ecology of dryland rivers. Regulated Rivers: Research and Management, 16, 385–402. R Development Core Team (2007) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. Sheldon F. (2005) Incorporating natural variability into the assessment of ecological health in Australian dryland rivers. Hydrobiologia, 552, 45–56. Sheldon F. & Thoms M.C. (2006) Relationships between flow variability and macroinvertebrate assemblage composition: data from four Australian dryland rivers. River Research and Applications, 22, 219–238. Sheldon F. & Walker K.F. (1998) Spatial distribution of littoral invertebrates in the lower Murray-Darling River system, Australia. Marine and Freshwater Research, 49, 171–182. Sheldon F., Boulton A.J. & Puckridge J.T. (2002) Conservation value of variable connectivity: aquatic invertebrate assemblages of channel and floodplain habitats of a central Australian arid-zone river, Cooper Creek. Biological Conservation, 103, 13–31.

1800


Sheldon F., Boulton A.J. & Puckridge J.T. (2003) Variable hydrological connection structures aquatic invertebrate composition in dryland rivers: data from Cooper Creek and Diamantina River. Records of the South Australian Museum Monograph Series, 7, 119–130. Simpson J.C. & Norris R.H. (2000) Biological assessment of river quality: development of AUSRIVAS models and outputs. In: Assessing the Biological Quality of Fresh Waters: RIVPACS and Other Technique (Eds J.F. Wright, D.W. Sutcliffe & M.T. Furse), pp. 125–142. Freshwater Biological Association, Ambleside, U.K. Statzner B., Bonada N. & Dole´dec S. (2007) Conservation of taxonomic and biological trait diversity of European stream macroinvertebrate communities: a case for a collective public database. Biodiversity and Conservation, 16, 3609–3632. Thoms M.C. & Parsons M. (2003) Identifying spatial and temporal patterns in the hydrological character of the Condamine-Balonne River, Australia, using multivariate statistics. River Research and Applications, 19, 443–457. Thoms M.C. & Sheldon F. (2000) Water resource development and hydrological change in a large dryland river: the Barwon-Darling River, Australia. Journal of Hydrology, 228, 10–21. Turak E., Flack L.K., Norris R.H., Simpson J. & Waddell N. (1999) Assessment of river condition at a large

spatial scale using predictive models. Freshwater Biology, 41, 283–298. 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, N.S.W., Australia. Ummenhofer C.C., England M.H., McIntosh P.C., Meyers G.A., Pook M.J., Risbey J.S., Gupta A.S. & Taschetto A.S. (2009) What causes southeast Australia’s worst droughts? Geophysical Research Letters, 36, L04706. Verbeke G. & Molenberghs G. (2000) Linear Mixed Models for Longitudinal Data. Springer, New York. Vinson M.R. & Hawkins C.P. (2003) Broad-scale geographic patterns in local stream insect genera richness. Ecography, 26, 751–767. Walker K.F. (1985) A review of the ecological effects of river regulation in Australia. Hydrobiologia, 125, 111– 129. Walker K.F., Sheldon F. & Puckridge J.T. (1995) A perspective on dryland river ecosystems. Regulated Rivers: Research & Management, 11, 85–104. Withnall F. (2000) Biology of yabbies (Cherax destructor). Aquaculture Notes. Department of Natural Resources and Environment, Victoria, Australia. (Manuscript accepted 27 January 2010)
