The ‘wet–dry’ in the wet–dry tropics drives river ecosystem structure and processes in northern Australia


D. M. Warfe, Tropical Rivers and Coastal Knowledge, Research Institute for the Environment and Livelihoods, Charles Darwin University, Darwin, NT 0909, Australia. E-mail:


1. Northern Australia is characterised by a tropical wet–dry climate that regulates the distinctive character of river flow regimes across the region. There is marked hydrological seasonality, with most flow occurring over only a few months of the year during the wet season. Flow is also characterised by high variability between years, and in the degree of flow cessation, or intermittency, over the dry season.

2. At present, the relatively low human population density and demand for water in the region means that most rivers have largely unmodified flow regimes. These rivers therefore provide a good opportunity to understand the role of natural flow variability in river ecosystem structure and processes.

3. This review describes the major flow regime classes characterising northern Australian rivers, from perennial to seasonally intermittent to extremely intermittent, and how these regimes give rise to marked differences in the ecological character of these tropical rivers, particularly their floodplains.

4. We describe the key features of these flow regimes, namely the wet and dry seasons and the transitions between these seasons, and how they regulate the biophysical heterogeneity, primary productivity and movement of biota in Australia’s wet–dry tropical rivers.

5. We develop a conceptual model that predicts the likely hydrological and ecological consequences of future increases in water abstraction (e.g. for agriculture), and suggest how such impacts can be managed so that the distinctive ecological character of these rivers is maintained.


Natural flow variability is a key part of the physical template of river systems, shaping biophysical heterogeneity through space and time and leading to a greater range of habitats (Tockner, Malard & Ward, 2000; Ward & Tockner, 2001; Thoms & Parsons, 2002). Together, biophysical heterogeneity and temporal variation in precipitation, flow and flooding operate as selective pressures on biota and result in a broad range of life history traits (Bunn & Arthington, 2002; Lytle & Poff, 2004). Both mechanisms create greater functional diversity, allowing a greater range of biota to ‘share’ the same system and providing a greater scope for ecosystem processes, thereby sustaining high biodiversity and ecosystem resilience (Poff & Allan, 1995; Poff et al., 1997; Ward et al., 2001; Naiman et al., 2008; McCann & Rooney, 2009). However, if flow variability is extreme, as in extended ‘supra-seasonal’ droughts or extreme floods, it can have negative effects on habitat availability, biological populations and ecosystem processes (Lake, 2003; Naiman et al., 2008). Maintaining or restoring natural flow regimes and their natural variability is thus fundamental to the ecology of river systems (Poff et al., 1997; Puckridge et al., 1998; Naiman et al., 2008; Reich et al., 2010).

Tropical rivers (including their floodplains and estuaries) often have a characteristic and distinctive seasonal pattern of flow, where most discharge occurs during summer and results in marked wet and dry seasons (Lewis, 2008). Rivers in the Australian wet–dry tropics display this seasonality to a greater degree than many other tropical regions, with most flow occurring over a few months because of monsoonal rainfall, and are also characterised by high interannual variability in flows (Puckridge et al., 1998; Petheram, McMahon & Peel, 2008). These characteristics also vary across the region, reflecting the inherent variability of flow regimes across the Australian tropics (Kennard et al., 2010). Strong hydrological seasonality has been proposed as a key factor determining ecosystem structure and processes in rivers generally, in both tropical (Junk, Bayley & Sparks, 1989) and temperate regions (Tockner et al., 2000), and specifically in the Australian wet–dry tropics (Douglas, Bunn & Davies, 2005; Hamilton & Gehrke, 2005). Indeed, the structure and function of all rivers are regulated by the same fluvial processes, particularly with regard to hydrological connectivity, through longitudinal, lateral, vertical and temporal dimensions (Ward, 1989; Tockner et al., 2000). The pattern of these fluvial dynamics dictates when and where hydrological connections occur and thus defines the flow and flood regimes, and consequently the ecological character, of different river systems (Ward & Tockner, 2001; Boulton et al., 2008).

In contrast to the highly populated and/or hydrologically modified river catchments of some other tropical regions (McClain, 2008), Australian tropical rivers remain relatively intact. Northern-draining Australian catchments cover approximately 17% of the continent’s land area and generate over 60% of Australia’s surface-water run-off, but support <2% of the population (about one person per 2.5 km2) (ABS, 2006; Woinarski et al., 2007). Consequently, these rivers have undergone little development, drain relatively undisturbed landscapes and have a relatively unaltered hydrology (Douglas et al., 2005; Woinarski et al., 2007). These rivers therefore provide a good opportunity to understand the role of natural flow variability in determining the ecological structure and function of tropical rivers in general, and this improved understanding can inform the development of regional water resources into the future. If the same fluvial drivers operate in all river systems, and Australian tropical rivers can be seen to represent the relatively unimpacted end of a continuum of hydrological alteration (Ward & Tockner, 2001; Boulton et al., 2008), then understanding the role of flow variability on the ecological structure and function of these systems provides a potential model for river restoration and conservation efforts elsewhere.

This review describes the natural flow and flood regimes of rivers in Australia’s wet–dry tropics, compares them with rivers in other tropical regions and highlights the key features of Australia’s wet–dry flow regimes that result in their distinctive ecological structure and function. We also describe how the hydrological regime dictates periods of floodplain inundation (from protracted to episodic), the variability of which can also be a strong determinant of aquatic ecosystem structure and processes. Evidence for the importance of these flow features is predominantly based on recent research across the region, which has been synthesised in a series of workshops, and hence we draw on a considerable amount of hitherto unpublished research. We also draw on the literature from other tropical systems to develop hypotheses of flow–ecology relationships where research from northern Australia is lacking. Finally, we discuss how rivers in Australia’s wet–dry tropics may be affected by water resource development and the changing climate, suggesting approaches to support the planning, use and management of these relatively unaltered freshwater ecosystems.

Landscape, climate and hydrology

The wet–dry tropics in northern Australia encompasses a region of approximately 1.3 million square kilometres, stretching from Broome in the northwest of Western Australia, eastwards to the northeast Queensland coast (Woinarski et al., 2007; Fig. 1). The region excludes the wet tropics of eastern Australia, south of Cairns (Fig. 1), which experiences rainfall throughout the year. It is an ancient, highly weathered landscape, with soils largely leached of nutrients and a lack of geological rejuvenation. The dominant vegetation across the region is extensive grassy woodland savanna, although other vegetation such as Spinifex grasslands, heathlands and relict pockets of tropical rainforest are also present (Woinarski et al., 2007). While there are some ranges and escarpments in the Kimberley, Arnhem Land and Cape York Peninsula (Fig. 1), the region is generally of low topographical relief (under 400 m altitude). A diversity of freshwater and coastal marine ecosystems is important in the region, including rivers, mangroves, salt-marsh flats, floodplain wetland complexes, ephemeral waterbodies and upland groundwater-fed wetlands (Pusey & Kennard, 2009). Some of Australia’s largest and most diverse wetlands occur in northern Australia, in Kakadu National Park and Arafura Swamp (both east of Darwin, Northern Territory), and the Southern Gulf region where several large rivers merge during major floods and vast areas are inundated (Pusey & Kennard, 2009).

Figure 1.

 Region comprising the wet–dry tropics of northern Australia. Catchments drain into the Timor Sea, the Gulf of Carpentaria or the Coral Sea. Major townships are identified (black circles), as are catchments (labelled in white) mentioned in the text. The dotted lines are lines of latitude.

Darwin is the largest urban centre in the Australian wet–dry tropics with a population of about 80 000. The remaining population, of which 40% is indigenous, lives in remote towns and communities scattered across the region (Woinarski et al., 2007). The dominant land use in northern Australia is cattle grazing, which extends over about two-thirds of the total land area, followed by traditional indigenous land uses. Less than 2% of the land area is used for production forestry, cropping, horticulture or mining, all of which tend to be localised around permanent water sources (Woinarski et al., 2007). Given the relatively low demand for water, few rivers in northern Australia are impounded: there are 27 storages >0.2 GL in the region, compared with 467 around the rest of the country (Pusey & Kennard, 2009). Prescribed seasonal burning is practised throughout the region (Russell-Smith, Whitehead & Cooke, 2009), but cattle grazing of native savanna vegetation is potentially the major impact on catchment processes through increased erosion, sedimentation, nutrient export to aquatic ecosystems, and the trampling of riparian and floodplain vegetation (Brodie & Mitchell, 2005; Pusey & Kennard, 2009).

Typically, more than 90% of annual rainfall in the Australian wet–dry tropics occurs during the summer wet season from November to April and is generated by monsoonal lows and thunderstorms (Petheram et al., 2008). Rainfall intensity (i.e. volume/time) during these storms is among the highest recorded in the world (Jackson, 1988). Climate diagrams illustrate that while annual rainfall in Darwin is comparable with locations in other tropical countries (such as Stung Treng in Cambodia), it occurs as greater monthly rainfall over a shorter wet season, and there is comparatively less rainfall over the dry season months (Fig. 2). Darwin’s rainfall is also greater than average for Australia’s wet–dry tropics, which is better represented by Cloncurry, Queensland (Fig. 2).

Figure 2.

 Climate diagrams summarising average monthly rainfall and temperature in four tropical locations: Darwin, Australia (top left); Cloncurry, Australia (top right); Stung Treng, Cambodia (bottom left); and Ciudad Bolivar, Venezuela (bottom right). The solid line represents rainfall on the right axis, the dashed line represents temperature on the left axis, and the shaded area represents the periods when evaporation exceeds rainfall. Diagrams sourced from the Worldwide Bioclimatic Classification System, 1996–2009, S. Rivas-Martinez & S. Rivas-Saenz, Phytosociological Research Center, Spain (

Rainfall rapidly declines from a maximum of 2000 mm year−1 near Australia’s northern coast to 300–400 mm year−1 at the inland boundary of the region, averaging about 400 km south (Petheram et al., 2008; Cresswell et al., 2009). Rainfall in the wet–dry tropics is highly variable both spatially and from year to year: over the period 1930–2006, the range of annual flows was 2.5 times greater than the mean (Cresswell et al., 2009). High air temperatures generate high evapotranspiration rates throughout the year. These range from 700 to 900 mm year−1 in the dry season from 1100 to 1200 mm year−1 during the warm wet season, and result in an annual water deficit across the region except in the very wettest of years (Cresswell et al., 2009; Fig. 2).

A characteristic of Australia’s tropical wet–dry rivers is that their higher water temperatures coincide with higher flows over the wet season, e.g. between 28 and 32 °C in the Daly River, Northern Territory (Townsend, Webster & Schult, 2011), but even higher depending on timing and location. For example, water temperatures up to 35 °C have been measured in the Daly River during the early wet season (Townsend et al., 2011) and can reach 37 °C in floodplain waterholes of the Mitchell River, Queensland (Hamilton, 2010). The occurrence of warm water during the wet season is similar to some South American floodplain systems that are inundated during the warmer months, e.g. the Pantanal wetland of Brazil, but differs from elsewhere, e.g. the Okavango Delta in Botswana and Andean tributaries to the Amazon and Orinoco Rivers, where floodwaters coincide with cooler water (Montoya et al., 2006; Hamilton et al., 2007; Lewis, 2008; Hamilton, 2010). Higher water temperatures during the high-water phase can lead to greater primary productivity and microbial activity in tropical systems (Winemiller, 2004; Hamilton, 2010). For example, northern Australian floodplains that have protracted inundation periods display marked increases in macrophyte biomass during the wet season (Finlayson, 1991; Pettit et al., 2011). However, thermal optima are narrow for many tropical species and, as they are living close to their maximum tolerable limits, biotic diversity can decrease beyond 35 °C (Deutsch et al., 2008; Hamilton, 2010). This suggests that a modest increase in temperature could produce unexpected changes in biotic composition and ecosystem processes (Hamilton, 2010).

There are 60 major river catchments within the wet–dry tropics of northern Australia that drain into the Gulf of Carpentaria, the Timor Sea or the Coral Sea (Fig. 1). The majority of rivers in these divisions are relatively short and drain directly to the coast (Hamilton & Gehrke, 2005; CSIRO, 2009). The geomorphological character of the region’s rivers varies greatly but typically they are of low average gradient, have a low density of streams per unit catchment area, and have large lowland floodplains comprising up to a third of the total catchment area (Stein et al., 2009). For example, rivers draining into the Gulf of Carpentaria can have floodplains of 20 000 km2 that comprise over 35% of the total catchment (Pusey & Kennard, 2009). Floodplain wetlands cover 25% of the entire region (Pusey & Kennard, 2009) and represent the largest area of unmodified wetlands in Australia (Woinarski et al., 2007). The combined river and floodplain matrix across northern Australia represents one of the last free-flowing river networks in the world and is consequently a globally significant asset (ATRG, 2004; Blanch, 2008).

The topographical separation of catchments, especially in the coastal low-relief terrain bordering the Gulf of Carpentaria, is often limited and results in connections between rivers during peak inundation in the wet season. In contrast, rivers draining into the Timor Sea are generally deeply incised and well separated. Given that rainfall is greater nearer the coast and declines inland, a substantial proportion of run-off from northern rivers may originate in the lowlands; floodplain wetlands can thus be inundated by both overbank flows from upstream and local sources. A key aspect of floodplains in northern Australia is the variability in their period of inundation: some are flooded for months, whereas others are flooded only episodically for days to weeks.

Most northern Australian rivers show a distinct and predictable hydrologic seasonality reflecting the wet–dry climate: high flows occur during the wet season and low flows, often interrupted by lengthy periods of zero flow, occur during the dry season (Kennard et al., 2010). Rivers with intermittent flow are common across northern Australia, although the degree of intermittency varies depending on climate, latitude and underlying geology. The most extreme intermittent rivers, i.e. those with the longest periods of no flow, occur near the southern boundary of the region on the edge of the arid zone. Flow may cease for extended periods of time (over 200 days per year), with remaining water restricted to a series of disconnected in-channel and floodplain waterholes (Douglas et al., 2005; Kennard et al., 2010). Perennial rivers, such as the Daly River in the Northern Territory, are uncommon across tropical Australia and supplied mainly by groundwater during the dry season (Petheram et al., 2008).

An ecohydrological regionalisation of Australia’s rivers has recently been developed to support the generalisation of flow–ecology relationships and responses to flow alteration (Kennard et al., 2010). The majority of rivers in the wet–dry tropics are within three flow regime types (Table 1; Fig. 3): stable summer baseflows (perennial rivers), predictable summer highly intermittent (intermittent rivers, the most common type in this region) and extreme harsh summer intermittent (extremely intermittent rivers, with often ‘flashy’ and largely unpredictable flows). Along with the considerable spatial variation in flow regime (Kennard et al., 2010), and despite the highly seasonal nature of flow, there can also be great interannual and apparent decadal variability in wet season discharge related to cyclonic events and El Niño-Southern Oscillation (ENSO), Interdecadal Pacific Oscillation (IPO), and Indian Ocean Dipole (IOD) climatic variation (Hamilton & Gehrke, 2005; Shi et al., 2008; Hamilton, 2010; Kennard et al., 2010; Fig. 3).

Table 1.   Summary of selected flow characteristics representing ecologically relevant flow regime components (magnitude, frequency, duration, timing and rate of change) for each flow regime class describing rivers across Australia’s wet–dry tropics, adapted from Kennard et al. (2010) and Pusey et al., (2009). Relative differences rather than actual values are presented, and summer flows coincide with the wet season
Flow characteristicStable summer baseflowPredictable summer highly intermittentExtreme harsh summer intermittent
Flow permanencePerennialHighly intermittentExtremely intermittent
Seasonal timingSummerSummerSummer
Runoff magnitudeHighModerateLow
Variability and skewnessLowModerateHigh
Number zero flows daysLowModerateHigh
1st percentile flood
Rate of rise and fallLowModerateHigh
25th percentile flow
Figure 3.

 Example hydrographs of daily run-off (ML day−1 km2) for a typical stream gauge in each flow regime class (note different y-axis scales for each class). Variation in run-off is shown for three scales of temporal resolution: the long-term record, a year and a 3-week period encompassing the flow event with the highest peak magnitude. The number, name and upstream catchment area (km2) of each stream gauge are given in parentheses.

Key flow features underpinning ecosystem structure and processes

The characteristic hydrological seasonality has been proposed as a strong driver of biotic assemblages and ecological processes in rivers and floodplains of Australia’s wet–dry tropics (Douglas et al., 2005; Leigh & Sheldon, 2008). We hypothesise that four key features of the flow regime underpin the structure and function of these tropical Australian river systems: (i) peak wet season flows and their variability, (ii) the drawdown period of flows and flood residence times during the transition from the wet to the dry season, (iii) the low and disconnected flows during the dry season, and (iv) the initial flushing flows during the transition from the dry to the wet season (Fig. 4).

Figure 4.

 Sample hydrograph from the Daly River (Northern Territory), over 1 year from August 2005 to July 2006, illustrating key flow features of rivers across the wet–dry tropics of northern Australia.

Wet season

Peak flows in the wet season, and their variability from year to year, determine the structure of channels and floodplains, regulate primary production on floodplains and in riparian zones, and provide hydrological connectivity for transporting nutrients, sediments and organic matter. They also provide opportunities for the movement and recruitment of biota between reaches that are isolated during the dry season. The wet season is the time in the annual hydrograph when floodplains are reconnected with their rivers, an important phase given the relatively large proportion of catchment area occupied by floodplains across the region. There is a continuum in the inundation period, or flood residence time, of floodplains across the region: some floodplains, such as those in Kakadu National Park and the Daly River (Fig. 1), generally flood each year and some areas can be inundated for up to 6 months at a time (Pettit et al., 2011). Others, such as the Fitzroy River (Western Australia) and the Mitchell River (Queensland; Fig. 1), have floodplains that may only be inundated for days to weeks, even in years of high rainfall (Fig. 5). On floodplains that experience short periods of inundation lasting from days to weeks, there has been little development of aquatic plant communities adapted to long periods of inundation, and aquatic primary production is largely limited to permanent waterholes. Primary production of terrestrially-adapted plant communities can be initially suppressed and then increase following the recession of floodwaters when soils are recharged with nutrients and moisture. In these latter systems, there is only a relatively short period available for nutrient transfer and use, aquatic–terrestrial fluxes in food resources and the active movement of organisms across the floodplain (Douglas et al., 2005). These floodplains clearly have a different ecology to those inundated for long periods, but we are unable to find any literature on equivalent systems elsewhere that bears on the observations we report here. The ecology of these systems, particularly the ability of biota to take advantage of short inundation periods, represents a significant knowledge gap for northern Australia. Consequently, much of the ensuing discussion refers to the more studied systems in the Northern Territory that are characterised by relatively long inundation periods.

Figure 5.

 Percentage of floodplain area on the Daly (black) and Mitchell (grey) River floodplains, inundated by floodwaters during 2008 (D. P. Ward, unpubl. data).

Rapid pulses of wet season discharge can have high erosional power through the bedrock-constrained valleys, headwaters and steep gorges that are features of Australia’s northern escarpment country, resulting in little material entrapment in these reaches (Brodie & Mitchell, 2005). A large proportion of the sediment in Australian rivers, including those in the northern wet–dry tropics, is derived from channel and gully erosion (Prosser et al., 2001; Wasson et al., 2010). Although the escarpments are generally restricted to Arnhem Land in the Northern Territory and the Kimberley Plateau in Western Australia (Fig. 1), and most of the region lies below 400 m altitude, erosional forces can still be high in low-gradient, sand-bed rivers (Brooks et al., 2009). For example, since 1990, more frequent overbank flows have contributed to increased riverbank erosion and slumping, channel-widening and sedimentation in the Daly River, Northern Territory (Wasson et al., 2010). On floodplains that are inundated for long periods, rainfall and overbank flows redistribute and deposit sediments according to floodplain topography, and as water velocity dissipates over the extent of the floodplain, the inundation extent being related to flow magnitude (Steiger & Gurnell, 2002; Naiman, Décamps & McClain, 2005; Wasson et al., 2010). On floodplains with short inundation periods, floods are rapid and episodic, occurring as ‘sheet flow’ over an essentially terrestrial system, but can still exert significant erosional force and move a considerable amount of sediment (D. P. Ward, unpubl. data). Variability in annual rainfall and flood peaks, combined with landscape topography, can thus dictate the structural heterogeneity of both channels and floodplains, and consequently the presence and persistence of aquatic refugia in channels and floodplain waterholes through the dry season (Richards, Brasington & Hughes, 2002; Bunn et al., 2006).

Wet season peak flows can also structure riparian vegetation communities and produce distinct lateral zonation largely controlled by bank height, sediment deposition and fluvial disturbance (Pettit, Froend & Davies, 2001; Lamontagne et al., 2005; Petty & Douglas, 2010). Attenuation of floods and flood variability, as has occurred in the lower Ord River (Western Australia; Fig. 1) because of the dual impoundments of Lake Kununurra and Lake Argyle, can lead to reduced structural heterogeneity in the riparian vegetation community and narrowing of the riparian zone (Pettit et al., 2001). Wet season floods also restructure channels through the provision and transport of wood and other organic matter that contribute to substratum heterogeneity as well as habitat and food resources for instream biota (Pusey & Arthington, 2003). Recent research in the Daly River indicates that the annual fluvial disturbance of riparian zones can lead to high turnover of wood deposited in channels; up to 50% of instream wood can be translocated over a single wet season, suggesting that the instream habitat for biota can be highly dynamic from year to year with interannual flood variability (N.P. Pettit & D.M. Warfe, unpubl. data). Fish communities associated with wood patches in the Cinaruco River (Venezuela) reassemble after wet season floods in a non-random manner, according to specific habitat patches regardless of the variability of those patches (Arrington & Winemiller, 2006). Such fidelity to habitat patches remains to be investigated in northern Australian rivers, but early research suggests that instream wood plays a significant role in structuring fish assemblages (N.E. Pettit, D.M. Warfe, M.J. Kennard & B.J. Pusey, unpubl. data).

Wet season flows are characterised by low sediment loads and nutrient concentrations because of the highly weathered, ancient geological nature of Australia’s tropical soils (Moliere et al., 2004; Brodie & Mitchell, 2005), and rivers across Australia’s wet–dry tropics are predominantly heterotrophic and nutrient-limited (Webster et al., 2005; Ganf & Rea, 2007). In contrast, while typically still nutrient-limited (Burford et al., 2011), estuaries appear to be more autotrophic owing to the higher aquatic productivity of mangrove forests (Alongi, Clough & Robertson, 2005; Burford et al., 2008a). In Darwin Harbour, nutrients are predominantly provided by tidal rather than tributary inputs and primary production is dominated not only by mangroves but also by benthic algae and phytoplankton, albeit to a lesser degree (Burford et al., 2008a). In the central Gulf of Carpentaria, wet season nitrogen inputs via river flows do not appear to contribute much to primary productivity, which seems instead to be supported by cyanobacterial nitrogen fixation (Burford, Rothlisberg & Revill, 2009). Nevertheless, riverine nutrient inputs during the wet season, while low, could still potentially contribute to primary production closer to the coast (Burford et al., 2011).

Wet season hydrology appears to be a main driver of productivity on floodplains, and, consequently, rates of primary production are variable across the riverine landscape (Davies, Bunn & Hamilton, 2008). Floodplains with extended periods of inundation can support substantial aquatic primary production during the wet season (Pettit et al., 2011). These floodplain dynamics are consistent with the Flood Pulse Concept, which proposes that seasonal inundation and subsequent drainage are the primary drivers of ecological processes in large floodplain rivers (Junk et al., 1989; Winemiller, 1996; Tockner et al., 2000). Estimates of carbon production on the Magela Creek floodplain in Kakadu National Park (east of Darwin, Northern Territory) show that primary production shifts from being predominantly algal-based and restricted to refugial waterholes during the dry season to extensive macrophyte production during the wet season (Pettit et al., 2011). However, despite dominating wet season primary production and biomass on the floodplain, macrophytes do not appear to contribute directly to the aquatic food web. Rather, they provide a large surface area for the attachment of epiphytic algae, which support most of the secondary production on the floodplain (Davies et al., 2008; Pettit et al., 2011). Similar observations have been made on the Orinoco floodplain in South America (Hamilton, Lewis & Sippel, 1992; Lewis et al., 2001). There is also evidence that algae produced on the floodplain during the wet season, even on floodplains with short inundation periods, may subsidise food webs in upstream reaches via the movement of fish consumers (T.D. Jardine, B.J. Pusey, S.K. Hamilton, N.E. Pettit & S.E. Bunn, unpubl. data). Such subsidies may be important for upstream food webs, as primary production in rivers during the wet season is generally low because of the scouring effect of high flows (Townsend & Padovan, 2005).

Extensive macrophyte growth can also provide important habitat for floodplain fauna. Magpie geese (Anseranas semipalmata Latham) are widespread and abundant across tropical Australian floodplains, particularly in Kakadu National Park (Bayliss & Yeomans, 1990; Morton, Brennan & Armstrong, 1990). In years of early monsoonal rainfall, there is subsequent high growth of wild rice (Oryza spp.) and water chestnut [Eleocharis dulcis (Burm.f.) Trin. ex Hensch], the latter providing nesting sites during the wet season and abundant food for adults and fledglings during the drawdown phase, and the former providing food for newly hatched goslings (Bayliss & Yeomans, 1990; P. Bayliss, unpubl. data). Hence, there is a general positive relationship between wet season rainfall, peak floods and both nesting success and dry season survival of magpie geese, particularly in years of early rainfall (Whitehead & Saalfeld, 2000). Magpie goose populations also appear to exhibit decadal trends that are tightly coupled with decadal trends in rainfall and river flows. On average, wetter years are followed by drier years over an approximate 20-year period that is mirrored in magpie goose numbers across the ‘Top End’ of the Northern Territory (Bayliss, Bartolo & van Dam, 2008).

Peak flows towards the end of the wet season often extend the persistence of aquatic habitats and are positively correlated with the abundance of plotosid and ariid catfish (Madsen & Shine, 2000). The abundance of catfish is in turn positively correlated with the body condition and yearling abundance of their main predator, the aquatic filesnake (Acrochordus arafurae McDowell) (Madsen & Shine, 2000). Conversely, the extended inundation period afforded by late season flooding reduces the available habitat, and thus the abundance, of the dusky rat (Rattus colletti Thomas), which results in poorer body condition and reduced reproduction in its main predator, the water python (Liasis fuscus Peters) (Madsen et al., 2006). These unusually strong relationships illustrate that variability in both the magnitude and timing of annual peak flows can have far-reaching effects on the composition of tropical floodplain communities, well into the following dry season. These relationships are unlikely to be as strong or prevalent on floodplains of short inundation periods.

An important consequence of wet season flows is that they provide both lateral and longitudinal connectivity throughout the entire drainage system and the opportunity for aquatic invertebrates, fish and reptiles to move between reaches to spawn (Douglas et al., 2005). Preliminary evidence from the Daly River catchment suggests that the emergence of aquatic insects peaks during the wet season, as does the input of terrestrial arthropods into streams (E. A. Garcia & M. M. Douglas, Charles Darwin University, unpubl. data). The same pattern has been observed in tropical rivers in Hong Kong, where the lateral flux of aquatic and terrestrial insects across the riparian zone also peaks during the wet season and provides an important link between aquatic and terrestrial food webs (Chan, Zhang & Dudgeon, 2007, 2008). The emergence of mature macroinvertebrates during the wet season has been suggested to be an evolutionary response to flood-induced mortality of large larvae, as well as providing available habitat for new recruits (Dudgeon, 2000; Jacobsen et al., 2008).

Of the 90 fish species recorded from freshwaters of the Daly River, one-third moves between freshwater and estuarine reaches to spawn, and one-third area vagrant estuarine species, such as bull sharks (Carcharhinus leucas Müller & Henle), and can be found hundreds of kilometres upstream (B.J. Pusey & M.J. Kennard, unpubl. data). Many of the remaining species move between different freshwater reaches during the wet season for spawning. Many of these fish species are widespread across northern Australia but, because most rivers are intermittent, their movements generally only occur during wet season flows when river, floodplain and estuarine reaches are connected. In the Magela Creek system (northeast of the Daly River in Kakadu National Park), sooty grunter (Hephaestus fuliginosus Macleay) move downstream from escarpment refugia during the wet season to spawn (Bishop, Pidgeon & Walden, 1995). Plotosid catfish move upstream into tributaries (Pusey, Kennard & Arthington, 2004), and juveniles of a number of species such as freshwater sole (Leptachirus triramis Randall) and barramundi (Lates calcarifer Bloch) also use wet season flows to move upstream, potentially escaping predation pressure in more open downstream reaches (Pusey et al., 2004; Staunton-Smith et al., 2004). Like barramundi, the giant freshwater prawn, Macrobrachium rosenbergii (De Man), is catadromous (Short, 2004) and local anecdotal evidence and observations from the Daly River catchment indicate that the juveniles move upstream en masse during the late wet season, constituting a ‘sushi train’ along the littoral margins of the main channel (B. J. Pusey, N. E. Pettit & D. M. Warfe, pers. obs.).

Peak flows also have effects beyond the freshwater reaches of tropical Australian rivers and floodplains, clearly illustrating the importance of longitudinal connectivity through these systems. Analysis of commercial penaeid prawn and finfish fisheries in northern coastal waters suggests that fisheries catches are good in years of high wet season inflows (Loneragan & Bunn, 1999; Robins et al., 2005). The commercial catch of both banana prawns (Penaeus merguiensis Fabricius) and barramundi (Lates calcarifer) has been shown to be positively related to years of high freshwater inflows (Vance, Staples & Kerr, 1985; Bayliss et al., 2008; Balston, 2009). Furthermore, the recruitment of king threadfin salmon (Polydactylus macrochir Günther) and the recruitment and growth of barramundi are also positively related to wet season peak flows, which likely increase hydrological connectivity to estuarine habitats and therefore provide greater access to estuarine nursery areas (Staunton-Smith et al., 2004; Robins et al., 2006; Halliday et al., 2008).

Wet to dry season transition

The transition from the wet to the dry season is the period when rainfall ceases and flows steadily decrease, where floodwaters recede on long-inundation floodplains, and intermittent rivers start to become hydrologically disconnected. This is also a key time ecologically: aquatic plant biomass is at its peak on river floodplains that have been inundated for some time, aquatic biota respond to receding waters by moving into refugial reaches to wait out the coming dry season, and waterbirds congregate in large numbers as aquatic resources become more concentrated in diminishing aquatic habitats.

Floodplains with long inundation periods are characterised by extensive macrophyte growth and diversity (Davies et al., 2008; Pettit et al., 2011); biomass generally peaks in the late wet season when floodwaters begin to recede (Finlayson, 1991). A similar pattern has been observed on floodplains with short inundation periods, but macrophyte growth in these systems is largely restricted to floodplain waterholes and terrestrial grasses, such as Dicanthium spp., dominate the floodplain instead (N. E. Pettit, pers. obs.). The biomass of attached algae, namely epiphytic diatoms, is also greatest in the late wet season, before declining as the water recedes and causes the macrophytes on which they grow to senesce and the available aquatic habitat to contract (Pettit et al., 2011). Aquatic macroinvertebrates associated with macrophytes on the floodplain reflect this pattern of plant production, peaking in abundance during the late wet season and transition into the dry season (Marchant, 1982; Outridge, 1988; Douglas & O’Connor, 2003). A conceptual model, developed from data from South American and African floodplains, suggests there is a peak in microbial activity on the floodplain during the late wet season as senescent macrophyte material is consumed (Winemiller, 1996). Microbial dynamics on northern Australian floodplains represent a major knowledge gap for northern Australia, but it appears that fire, rather than microbial activity, is a major consumer of plant material during the wet to dry season transition (Pettit et al., 2011).

The transition between the wet and the dry season appears to be a key time for the large-scale movements of organisms. Unlike spawning movements during the wet season, movements during this transition period are more likely to be associated with finding refuge as aquatic habitats begin to disconnect and contract. For example, saltwater crocodiles (Crocodylus porosus Schneider) that have moved onto the floodplains during the wet season return to channel reaches as floodwaters recede (Jenkins & Forbes, 1985). Melanotaeniid rainbowfish and ambassids, among other species, moved from floodplain waterholes to upstream refugial areas towards the end of the wet season in the Magela Creek system in Kakadu National Park (Bishop et al., 1995). Recent research on fish movement in tributaries of the Daly River shows large numbers of melanotaeniid rainbowfish and plotosid catfish moving downstream during the transition from the wet to the dry season, particularly in intermittent rivers that were in the process of becoming disconnected (D.M. Warfe & N.E. Pettit, unpubl. data). Fish assemblage structure in the remaining waterholes of rainforest streams in Queensland has been shown to be influenced by the magnitude of the preceding wet season, which in effect ‘sets up’ the assemblage that will persist through the dry season (Perna & Pearson, 2008). Evidence from streams in Australia’s wet tropics indicates that, while macroinvertebrate structure is not affected by season (Cheshire, Boyero & Pearson, 2005), the rate of macroinvertebrate colonisation peaks during the late wet season flows as early instars disperse to suitable habitats (Benson & Pearson, 1987). We are analysing data on benthic macroinvertebrate assemblages to determine whether these patterns hold in Australia’s wet–dry tropical rivers.

There are also increased aggregations of waterbirds on the floodplain during the transition from the wet to the dry season, as floodwaters contract to waterholes and aquatic resources become more concentrated. Darters, cormorants, pelicans and grebes that feed on aquatic invertebrates and fish become more abundant around waterholes (Franklin, 2008), and assemblages can shift as water depths decrease (Chatto, 2000). Magpie geese move to areas of high resource availability during this period (Traill, Bradshaw & Brook, 2010), occurring in their highest densities where a range of macrophyte species required for nesting and feeding occur (Bayliss & Yeomans, 1990), and can represent the largest proportion of waterbird biomass on the floodplain (Pettit et al., 2011).

Dry season

The dry season is a period of limited resources as aquatic habitats become disconnected and contract across most of the region. Isolated waterholes on floodplains and in intermittent rivers become critical for sustaining aquatic biota and play an important refugial role during the dry season (Bunn et al., 2006).

There appears to be considerable variation in water quality and primary production between isolated waterholes, both on the floodplain and along river channels (Butler, 2008). Some waterholes are naturally turbid, and the dominant primary production supporting their aquatic food webs is the narrow band of benthic algae in the littoral zone (Bunn, Davies & Winning, 2003) or potentially the phytoplankton in the water column (Robertson et al., 1999). Other waterholes can support a very high biomass of macrophytes and benthic algae and have clear, deep water (Finlayson, 1991; Butler, 2008; Davies et al., 2008). In the latter case, it appears that the epiphytic algae attached to these macrophytes support the aquatic food web (Hamilton et al., 1992; Douglas et al., 2005; Pettit et al., 2011). However, on floodplains with only short inundation periods, terrestrial animals such as wallabies, horses and cattle have been observed consuming macrophytes when terrestrial vegetation becomes scarce at the end of the dry season (S. K. Hamilton, pers. obs.).

Reflecting the isolated nature of refugial waterholes, the species richness of dry season fish assemblages in intermittent rivers and reaches in the Daly catchment tends to be lower than in perennial reaches (B.J. Pusey & M.J. Kennard, unpubl. data). Macroinvertebrate communities differ between intermittent and perennial reaches, with assemblages comprising more lentic and lotic taxa, respectively (Humphrey, Hanley & Camilleri, 2008; Leigh & Sheldon, 2009). Trophic diversity within aquatic food webs can narrow (D.M. Warfe, N.E. Pettit, E.A. Garcia & M.M. Douglas, unpubl. data), and the diets of resident fish can also narrow and become poorer in quality (Balcombe et al., 2005) as consumers are forced to become more dependent on local food resources (T.D. Jardine, D.M. Warfe & N.E. Pettit, unpubl. data). Narrowing of fish diets owing to limited resource availability during the dry season also appears to be a common feature in Central and South American rivers (Winemiller, Agostinho & Caramaschi, 2008). In perennial rainforest rivers in Queensland, feeding links between fish and their food sources do not vary greatly between the wet and the dry season, supporting the hypothesis that intermittency can lead to more limited resources and more striking seasonal changes in food webs (Rayner et al., 2010).

As the dry season progresses, the available habitat contracts, increasing the potential for predation and competition, as has been shown in both Neotropical floodplain rivers (Winemiller, 1996; Rodriguez & Lewis, 1997) and Australian arid-zone rivers (Arthington et al., 2005). In shallow waterholes (<3 m), habitat reduction can be accompanied by a deterioration in water quality that can contribute to fish kills when flow resumes (Townsend, Boland & Wrigley, 1992). It is possible that fish mortality may fuel algal and bacterial growth towards the end of the dry season, as has been shown in Australian arid-zone rivers (Burford et al., 2008b). Furthermore, late dry season flowering of riparian species such as Melaleuca leucadendra L. (Pettit, 2000) can attract insects and flying foxes (Pteropus alecto Temminck) (Vardon et al., 2001), potentially contributing riparian subsidies to depleted waterholes and supporting juveniles of species that breed at the end of the dry season, for example barred grunter (Amniataba percoides Günther) (Pusey et al., 2004; N. E. Pettit & D. M. Warfe, pers. obs.).

In the few perennial rivers within northern Australia, dry season baseflows are maintained by groundwater discharge and their typically oligotrophic nature (e.g. Townsend & Padovan, 2005) reflects the low nutrient concentrations in the supporting aquifers. For example, in the Daly River, as groundwater contributes a larger proportion of flow over the course of the dry season, nitrate concentration and turbidity decrease, and the potential euphotic zone exceeds the river depth (Townsend & Padovan, 2005; Townsend et al., 2011). The clear water and low nutrient concentrations common to these perennial rivers can make them vulnerable to algal blooms as a result of nutrient enrichment (Ganf & Rea, 2007). In the Daly River and its major tributaries, photosynthesis increases over the dry season because of the accumulation of primary producer biomass rather than higher incident radiation (Webster et al., 2005; Townsend et al., 2011). While photosynthesis has been shown to be light-limited in these generally open-canopy rivers, the accumulation of primary producer biomass is probably nutrient-limited within the hydraulic and other physical constraints of the river (Webster et al., 2005; Townsend et al., 2008). Phytoplankton is likely to briefly contribute a significant proportion to primary production in the early dry season, before the subsequent growth and accumulation of benthic algae and submerged macrophytes dominate primary production (Townsend & Padovan, 2005; Townsend et al., 2011). As the dry season progresses, producer biomass generally increases before being scoured out during the first flows of the early wet season (Townsend & Padovan, 2005), a pattern also apparent in the low-nutrient wet–dry rivers of the Neotropics (Cotner et al., 2006).

While bottom-up controls probably limit primary producer biomass, in common with the Neotropics (Cotner et al., 2006; Roelke et al., 2006), there is evidence that benthic algae in Australia’s wet–dry tropics are also partly regulated by grazing pressure from macroconsumers such as freshwater prawns and catfish, particularly during the dry season when there is no disturbance from floods (Douglas et al., 2005; M.M. Douglas, unpubl. data). Such top-down control of algal resources has been observed in other tropical regions, e.g. Costa Rican streams (Pringle & Hamazaki, 1997), Andean streams (Flecker & Taylor, 2004) and the Cinaruco River in Venezuela (Winemiller et al., 2006), suggesting that it is a characteristic pattern of tropical streams and rivers (Douglas et al., 2005). There are suggestions that the meiofauna may also regulate benthic algae in tropical sand-bed rivers (e.g. Winemiller et al., 2006), but their importance in northern Australian rivers remains to be investigated (Humphreys, 2008).

Perennial rivers also provide important habitats within the dry season landscape for flow-dependent aquatic species. Juveniles of numerous fish species, such as sooty grunter (Hephaestus fuliginosus) and Butler’s grunter (Syncomistes butleri Vari), are predominantly restricted to riffle areas where they can escape predation pressure from larger fish, but also where there is more food (Pusey et al., 2004). Perennial reaches support distinctive macroinvertebrate communities consisting of groups such as baetid mayflies, hydropsychid caddisflies and hydrophilid beetles that prefer flowing water (Humphrey et al., 2008; D.M. Warfe & N.E. Pettit, unpubl. data). Together with microalgae, these macroinvertebrates are a source of benthic food that supports higher trophic levels (Douglas et al., 2005; Townsend & Padovan, 2009).

Other stream channel features can also become important as they become exposed or accessible during the dry season. Periods of low flow expose channel sediments that can become important sites for the germination and establishment of riparian seedlings (Pettit et al., 2001). The pig-nosed turtle (Carettochelys insculpta Ramsay) is restricted to perennial rivers, such as the Daly River, which are deep enough to allow extensive movement (up to 14 km) by females as they search for suitable egg-laying sites (Georges et al., 2003). The females target fluvially reworked sand banks in the inside of bends, at tributary mouths or behind large boulders and wood aggregations, and lay their eggs at the end of the dry season after water temperatures have begun to rise (Georges et al., 2003).

Dry to wet season transition

The major feature of the transition from the dry to the wet season is the occurrence of storm run-off events and the hydrological reconnection of isolated stream reaches. The first flow events of the wet season are often of poor water quality as they flush organic material and suspended particulates through the system and into the estuary, sometimes creating hypoxic conditions and causing fish kills in rivers and floodplain waterholes (Townsend et al., 1992; Townsend & Edwards, 2003; Butler, 2008). Tropical waters generally have a higher oxygen demand, and much lower oxygen saturation concentrations, than temperate waters for a given organic loading, and thus they are potentially vulnerable to anthropogenic organic loading (Lewis, 2008). Surface run-off events during the transition from the dry to the wet season can also bring a pulse of organic material, nutrients and sediments from the surrounding catchment (Schult et al., 2007); consequently, the impacts of fire and cattle grazing on aquatic systems are likely to be most concentrated at this time (Townsend & Douglas, 2000). Similarly, mobilisation of trace metals and reduced substances from floodplain soils may cause fish kills (Hart & McKelvie, 1986). This is especially likely in areas where acid sulphate soils occur on former marine sediments or as a result of past mining activities, e.g. in the Finniss River, south of Darwin (Taylor, 2007).

Run-off events during the early wet season were an important factor regulating mechanical breakdown, and consequent biotic breakdown, of leaf litter in an Amazonian floodplain stream (Rueda-Delgado, Wantzen & Tolosa, 2006). In general, decomposition rates of litter in tropical streams are strongly influenced by water temperature, turbidity, pH, salinity, dissolved organic carbon, nutrients and oxygen (Wantzen et al., 2008). Early evidence from streams in the Daly River catchment also suggests that flow regime, leaf species and the physical character of streams play a stronger role in leaf litter breakdown than microbial activity (T. Davies, N.E. Pettit & P.F. Grierson, unpubl. data). However, knowledge of microbial processes in Australian wet–dry tropical rivers is scant.

The reconnection of isolated refugia within rivers is an important time in the annual hydrograph for the movement of biota, as flows commencing after the initial poor-quality flush provide an opportunity for fauna to move into more favourable reaches and throughout the river system. Reaches in rivers and on floodplains that act as refugia during the dry season represent important sources of recolonising biota (Outridge, 1988; Perna & Pearson, 2008). Macroinvertebrates can rapidly colonise re-wetted reaches from upstream perennial reaches and floodplain waterholes via the dispersal of aerial adult stages, as well as from the hyporheic zone where microcrustaceans, oligochaetes and dipterans appear to aestivate over the dry season (Paltridge et al., 1997). The early wet season is a key time for saltwater crocodiles, Crocodylus porosus, to disperse throughout the river system and into floodplain waterholes (Jenkins & Forbes, 1985); their nesting effort is positively correlated with high rainfall and relatively cool weather over this period (Webb, 1991). Juvenile barramundi move from estuarine habitats to upstream floodplains and tributaries (Pusey et al., 2004). Males can spend between 3 and 5 years in freshwater riverine habitats before returning to estuaries for spawning, and there is also evidence for considerable movement throughout riverine habitats during this time (Griffin, 1987; Pusey et al., 2004). Recent research on fish movement in tributaries of the Daly River showed that rainbowfish (Melanotaenia australis Castelnau) and hardyheads (Craterocephalus stercusmuscarum Günther) moved upstream after flushing flows and were markedly more abundant where flow had ceased during the dry season rather than at perennially flowing sites (D.M. Warfe & N.E. Pettit, unpubl. data).

Consequences of altered flow regimes

The major implications of the distinct seasonal flow regime in Australia’s wet–dry tropics are that the biota within these systems must be able to take advantage of short periods of resource-rich conditions during the wet season, and be able to withstand relatively long periods of resource-poor conditions during the dry season (McMahon & Finlayson, 2003; Lytle & Poff, 2004; Douglas et al., 2005). The fact that these processes occur annually, and that the river and floodplain biota demonstrate seasonally triggered traits such as fish spawning, reptile nesting and vegetation fruiting, suggests that these ecosystems are effectively ‘locked’ into the strong seasonal pattern and support an aquatic biota that is adapted to cope with, and benefit from, the marked hydrological seasonality (Junk et al., 1989; Bishop & Forbes, 1991; Davis et al., 2010; Pettit et al., 2011). Recent research has found spatially concordant distributions of vegetation, fish and invertebrate assemblages across the region, which are primarily driven by environmental gradients, flow regime being a common structuring driver of all three assemblage types (D.M. Warfe & N.E. Pettit, unpubl. data). The corollary to this is that alterations and disruptions to the natural flow regime are likely to have far-reaching consequences for aquatic biodiversity and ecosystem processes in Australia’s wet–dry tropics (Bunn & Arthington, 2002).

Importantly for Australia’s tropical rivers, their flow regimes (in most cases) are relatively unmodified: <3% of the total water extracted in Australia occurs in the north (approximately 2000 GL), about 10–20% of which is taken by groundwater pumping (ABS, 2006; Cresswell et al., 2009). There are 27 impoundments with a capacity >0.2 GL across northern Australia – compared with 467 in the rest of the country – most of which are in Queensland and serve for urban water supply and irrigation (Pusey & Kennard, 2009). The largest impoundment in the region, the Ord River Dam (Lake Argyle) in the Kimberley region of Western Australia, clearly illustrates the impacts of such storages on wet-dry tropical flow regimes. The lower Ord River was once a highly variable intermittent system. It is now a perennial system with water being released during the dry season months to meet irrigation and hydroelectric demand, and all but the largest floods are attenuated by the dam (Doupé & Pettit, 2002). This has resulted in sedimentation and narrowing of the river channel below the dam (Cluett, 2005), and riparian vegetation assemblages becoming narrower and more homogenised (Pettit et al., 2001). It has also led to reductions in banana prawn populations because of lower salinity levels from perennial freshwater inflows (Kenyon et al., 2004), and lower primary production in the estuary (Burford et al., 2011).

Water planning has occurred in a few catchments across the region but, on the whole, rivers are unregulated and water is allocated on a licence-by-licence basis by state government agencies. The Northern Territory Government has a policy of capping water allocations to retain 80% of natural flows in rivers and aquifers in the wet–dry region of its jurisdiction. The policy recognises that unimpeded flow regimes provide a range of ecosystem goods and services to the region. For example, the productivity of commercial fisheries in the tropics, worth AU$220 million annually (Robins et al., 2005), and the recreational fishing industry, worth up to AU$26 million to the economy of the Northern Territory (NTG, 2009), rely on naturally flowing rivers. Similarly, saltwater crocodile harvesting and magpie goose hunting are worth AU$40 million and AU$2 million per annum (respectively) to the Northern Territory economy and are heavily reliant on floodplains retaining their natural inundation regimes (Delaney, Fukuda & Saalfeld, 2009; Leach, Delaney & Fukuda, 2009). These intact freshwater systems sustain nature-based tourism, one of northern Australia’s major industries valued at AU$1.5 billion annually (Woinarski et al., 2007; Clark et al., 2009). Furthermore, freshwater ecosystems are culturally and socio-economically important to local indigenous communities (Jackson, Storrs & Morrison, 2005); current research suggests that the economic value of the use of freshwater flora and fauna in some communities may comprise up to 20% of the median household income (M. Finn, unpubl. data).

Given the sparse human population, low-intensity land use and relatively low consumptive demand for water, Australia’s tropical rivers and estuaries are, hydrologically, the least impacted in the country (NLWRA, 2002; Stein, Stein & Nix, 2002) and represent some of the most unaltered systems in the world (Woinarski et al., 2007). They also represent a potentially great water resource, and consequently, there is persistent interest in developing Australia’s northern rivers to help support the country’s food production, particularly in the wake of extensive droughts in the Murray-Darling catchment such as the ‘Millenium Drought’ over 2000–08 (sensuLake, Likens & Ryder, 2010). However, recent research on sustainable yields across the region indicates that the water balance is effectively ‘closed’ (Cresswell et al., 2009). Despite large wet season rainfall, the landscape is annually water-limited, and the low topographical relief, lack of suitable dam sites and high evapotranspiration rates do not support the viability of water storages (Petheram et al., 2008; CSIRO, 2009).

Owing to the above constraints, large-scale development of Australia’s wet–dry tropical rivers seems unlikely at this time, yet there is still interest in smaller-scale development. Surface and groundwater abstraction, the construction of small off-stream storages and barriers such as tributary impoundments and road crossings may produce substantial and cumulative alterations in natural flow regimes. The ecological consequences of such alterations relate to hydrological connectivity and material transport, the availability of habitat, and opportunities for the movement and recruitment of aquatic biota, and could be extensive given the dependence of these ecosystems on existing flow regimes. Based on our review of how flow regimes and their key features structure riverine, floodplain and estuarine ecosystems in the Australian wet–dry tropics, we have summarised the predicted hydrological and ecological responses to small-scale, but cumulative, water development in the region (Fig. 6).

Figure 6.

 Conceptual ecological model based on the information reviewed in this paper, illustrating the predicted hydrological and ecological responses of water extraction for each of the key flow features comprising flow regimes in the wet–dry tropics of Australia. Arrows within boxes indicate an increase or decrease in that particular component. Boxes are defined by their shape in the lower right corner.

A major consequence is likely to be reduced dry season baseflows in perennial rivers (Fig. 6). A reduction in baseflow has the potential to reduce the availability of critical flow-sensitive habitats (e.g. riffles) for turtles, juvenile fish and benthic biota, consequently reducing aquatic primary production and faunal recruitment, and negatively impacting the food resources for larger species such as freshwater crocodiles, barramundi and black bream (Douglas et al., 2005; Webster et al., 2005; Pusey & Kennard, 2009; Townsend & Padovan, 2009; Chan et al., 2010; Fig. 6). In extreme circumstances, reductions in baseflow during the dry season could disconnect reaches and shift the flow regime class from perennial to intermittent (Pusey & Kennard, 2009); such shifts between flow regime classes are likely to be accompanied by major ecological change (Poff et al., 2010). Similarly, water extraction in intermittent rivers can increase the period of hydrological disconnection between refugial reaches to more extreme intermittency, hampering the movement and migration of aquatic biota during seasonal transitions (Fig. 6) and potentially increasing the risk of localised disturbance to disconnected populations and genetic ‘bottlenecks’ (Pusey & Kennard, 2009).

Water extraction is likely to affect surface and groundwater interactions, potentially decreasing groundwater levels and affecting riparian vegetation communities reliant on groundwater (O’Grady et al., 2006; Pusey & Kennard, 2009; Fig. 6). Groundwater can be important in maintaining the persistence of freshwater refugia over the dry season, so extraction may reduce the size and persistence of these refugia, thereby reducing the amount of already-limited aquatic habitat available across the landscape towards the end of the dry season (Pusey & Kennard, 2009; Fig. 6). Also, groundwater extraction may disturb the subterranean and groundwater ecosystems that are thought to play important filtering and water purification roles (Humphreys, 2008; Pusey & Kennard, 2009).

There is also the potential for water development, particularly the construction of instream storages and barriers, to reduce flood peaks during the wet season and therefore reduce the extent and duration of floodplain inundation, as well as altering the timing of flood peaks (Fig. 6). Alterations to flood dynamics can disrupt cues for spawning, nesting or hatching for many species such as barramundi (Bayliss et al., 2008), magpie geese (Delaney et al., 2009) and pig-nosed turtles (Georges et al., 2003), respectively, negatively affecting population recruitment. Reductions in floodplain inundation can also limit the opportunities for biota to move on and off the floodplain during the transition phase to the dry season, potentially impacting the ability of fauna to reach freshwater refugia (Pusey & Kennard, 2009; Fig. 6).

Climate change predictions are imprecise for northern Australia, but there is an increased likelihood of increased temperatures and evapotranspiration, and also of extreme storm, cyclone and drought events (CSIRO and BoM, 2007; Cresswell et al., 2009). Modelling flows under climate change scenarios suggests that river levels are likely to be lower more often, but even more so under water development scenarios (McJannett et al., 2009). While this is likely to reduce the availability of critical habitats for aquatic biota, specific ecological responses are difficult to quantify because of limited knowledge around environmental flow thresholds. Given the low altitude of most of the wetlands and floodplains throughout tropical Australia, particularly in the Northern Territory, one of the major projected impacts is saltwater intrusion attributed to rising sea levels and storm surge, and consequent losses of coastal freshwater habitats, biodiversity and wetland-dependent populations of waterbirds (Traill et al., 2009; Hamilton, 2010). Altered rainfall and storm events are likely to alter flood regimes, which dictate channel structure (Wasson et al., 2010) and floodplain inundation events that determine floodplain structure, extent and seasonality (Hamilton, 2010). Redistributions of both habitats and biota may also occur as populations expand or contract at the limits of their climatic range, and those species with restricted distributions or limited dispersal capacity are particularly at risk (Woinarski et al., 2007). The consequences of such species redistributions are unknown, but are likely to cascade to other biota through food web interactions (Pusey & Kennard, 2009), potentially leading to increased species invasions, increased disease and the formation of novel communities (Traill et al., 2009).

There are major knowledge gaps around how the effects of climate change might interact with increasing water use, and also how these effects will interact with land-use impacts such as land clearing and cattle grazing, weeds and feral animals, as well as cropping and nutrient inputs. In particular, more information is required on environmental flow requirements, hydrological connectivity across riverine landscapes, mechanisms for tolerating low- or no-flow periods, and potential flow thresholds to predict more effectively the consequences of water use on these ecosystems. This information is also necessary for identifying key species and processes for monitoring, and to assist with the identification of ecosystems or areas of high conservation value. Hamilton & Gehrke (2005) outlined the key knowledge requirements for Australia’s tropical rivers: the sustainable availability of water for human use; hydrological, biogeochemical and ecological linkages at landscape scales; an understanding and valuation of ecosystem processes and services; and managing climate change. We note that, despite the advances in knowledge of Australia’s northern rivers achieved over the past 5 years and documented here, the knowledge gaps identified by Hamilton & Gehrke (2005) are still relevant.

Principles for water management in northern Australia

There have been numerous papers outlining key principles of providing environmental flows to preserve the world’s freshwater resources (Poff et al., 1997; Bunn & Arthington, 2002; Pinay, Clement & Naiman, 2002; Arthington et al., 2006), and many more describing methods for implementing them (Cottingham, Thoms & Quinn, 2002; Arthington & Pusey, 2003; Hughes & Rood, 2003; Tharme, 2003; Acreman & Dunbar, 2004). Ultimately, these methods all rely on a conceptual understanding of the relationships between flow variation and ecological responses that can be used to define environmental flow allocations, but in many cases, this knowledge remains scant (Naiman et al., 2008). Historically, many environmental flow studies have focussed on meeting the needs of specific taxa or populations, the assumption being that maintaining assets of high environmental ‘value’ will maintain ecological health. This often assumes direct and linear relationships between patterns of species distribution and ecological processes (Anderson et al., 2006). Our understanding of the links between ecological pattern and process is limited at best; nonlinear relationships are common and patterns are often a product of processes occurring at multiple and interacting scales (Walker, Sheldon & Puckridge, 1995; Harding et al., 1998; Bendix & Hupp, 2000; Bunn & Davies, 2000; Tockner et al., 2000). It would seem reasonable to focus the goals of environmental flows towards maintaining ecosystem resilience and integrity; the biotic processes leading to high biodiversity should then follow (Ward, Tockner & Schiemer, 1999; Bunn & Davies, 2000; Ward et al., 2001).

A recent advance in environmental flows methodologies is the ‘Ecological Limits of Hydrologic Alteration’, or ELOHA framework (Arthington et al., 2006; Poff et al., 2010), which incorporates the essential scientific requirements for setting environmental flows. It has been developed in acknowledgement of the fact that the increasing rate of water development is outstripping the ability to assess environmental flow requirements and that better links between science and management are needed. The ELOHA framework develops flow–ecology response relationships within flow regime types, which can then be empirically tested in an adaptive manner. It thus lends itself to application at a regional scale and across a range of river types, and can also address different levels and types of flow alteration such as regulated versus unregulated rivers (Naiman et al., 2008).

In regions of limited data on flow–ecology relationships, the ELOHA framework maximises the ability to extract useful information from across the region, and we propose it as the most suitable framework to derive environmental flows for rivers in Australia’s wet–dry tropics. An ecohydrological classification, incorporating the first and second steps in the ELOHA framework, has already been developed for Australian rivers and identifies three ecologically relevant flow classes in the wet-dry tropics, their range of variability, and their extent of hydrological development (Kennard et al., 2010). The third step in the ELOHA framework requires the deviation between current and natural flows to be calculated for each flow class (Poff et al., 2010). While the availability of gauged data across northern Australia is problematic because of sparse spatial coverage and inadequate time series, low water use and predominantly unimpeded flow regimes mean that most rivers have an essentially natural flow regime with low hydrological disturbance (Stein et al., 2002). The present paper characterises the ecological structure and function of those tropical flow classes and is therefore a step towards developing the flow–ecology relationships specific to each flow class, the final step in the ELOHA framework.

While the ELOHA framework, along with most methods and case studies of environmental flows, recommends retaining the natural flow regime as far as possible to maintain naturally functioning aquatic ecosystems (Poff et al., 1997), there are also other management approaches that could be used in combination to guide water resource management in the region. One approach is to designate ‘representative’ rivers for protection. We do not recommend an approach favouring the ‘top 20%’ of rivers, for example, as this requires a difficult decision on which rivers should be considered in the ‘top 20%’ and on what basis, particularly given the often high levels of uncertainty surrounding the choice of decision-support tools (Borchers, 2005; Burgman, Lindenmayer & Elith, 2005). Rather, we advocate using existing ecohydrological classification systems (e.g. Stein et al., 2009; Kennard et al., 2010) to ensure that rivers representing each hydrological and geomorphic ‘class’ are included in any conservation management scheme (Fitzsimons & Robertson, 2005).

Another approach is to recognise the variability of flow regimes in the wet–dry tropics and the range of biotic responses it engenders, and retain a mosaic of hydrological units or habitats in the lateral, longitudinal, vertical and temporal dimensions over which flow variability operates. For example, floodplains could be managed to sustain a range of waterhole types, from permanent and deep through to intermittent but inundated annually, across a diversity of systems with varying flow and flood regimes. Such a multi-scalar approach explicitly incorporates the role of hydrological connectivity and its role in mediating the transfer of materials and aquatic organisms between hydrological ‘elements’ (Pringle, 2001). It would also support a shifting habitat mosaic that underpins aquatic and terrestrial biodiversity, food web stability and ecosystem processes (McCann & Rooney, 2009; Larned et al., 2010), and thus the distinctive ecological character of river ecosystems in Australia’s wet–dry tropics.


Australian tropical rivers are driven by fluvial dynamics that are distinct from rivers elsewhere and result from a particularly strong wet–dry climatic seasonality with high interannual variability. Many of these river systems have intermittent flow and limited permanent freshwater refugia during the dry season, in spite of extensive flooding during wet season flows. This review has described the critical features of flow regimes that maintain ecosystem processes in rivers of Australia’s wet–dry tropics. Peak wet season flows are responsible for connecting river channels, floodplains and estuaries across their spatial extent and play an important role in material transfer and regulating biotic production and recruitment. Transition periods between the wet and the dry seasons, and vice versa, are key times for biota to move between different parts of the riverscape for spawning and nesting, or finding more favourable habitats and easing pressure on resources. And the dry season is a time where resources can become scarce and aquatic refugia become critical for sustaining aquatic biota through to the following wet season. This marked seasonal cycle has led to the development of life history strategies and ecosystem processes that allow the biota to cope with, and benefit from, such ‘boom and bust’ conditions similar to those observed in more arid-zone rivers (Bunn et al., 2006).

Hydrological seasonality describes flow variability on an annual scale, but variability at larger temporal scales is equally important, regulating the degree of ‘boom’ and ‘bust’ from year to year. For example, the 2008–09 wet season saw massive flooding in the southern Gulf rivers in Queensland and fairly average flooding across the rest of the region during the mid-wet season, whereas in the 2009–10 wet season, most rainfall was delayed to almost the beginning of the dry season and numerous floodplains experienced only limited inundation, if any at all (e.g. the Daly and Fitzroy rivers, respectively). In contrast, the 2010–11 wet season has seen near-record flooding in these same rivers. Such interannual variability generates spatial heterogeneity that can favour different biota from year to year (e.g. Madsen & Shine, 2000). It thereby sustains a wider range of biota over the long term, which in turn supports the ecosystem stability and resilience required to persist through ‘extreme’ conditions (McCann, 2000; McCann & Rooney, 2009). However, to manage variability for ecosystem resilience, it is critical to understand the range of variability experienced within each flow class in the wet–dry tropics so that the full range of natural variability is experienced, but not exceeded, and ecological thresholds are not crossed, potentially altering these ecosystems irrevocably.

Rivers in Australia’s wet–dry tropics are notable for their relatively unaltered hydrological nature and undeveloped catchments, so the fluvial dynamics we observe are largely natural and this sets them apart as a model for understanding flow–ecology relationships. The conceptual model summarises the key role of seasonal hydrology and hypothesises the hydrological responses and ecological consequences of water development in Australian wet–dry tropical rivers. Given that, on a global scale, these rivers are some of the few remaining naturally functioning examples, their ecological and socio-economic value should not be considered just in terms of local communities, but also in the context of national and international settings. This review has outlined some of the potential ecological consequences of water development in the region and suggested approaches to managing these rivers sustainably. There is an ecohydrological classification of tropical Australian rivers, and the development of flow–ecology relationships is underway, as is a burgeoning understanding of how these relationships might vary with flow type and hydrological variability. Thus, there is an excellent opportunity to implement the ELOHA framework across the wet–dry tropics in northern Australia, to test the hypotheses that underpin our conceptual model (before major water resource development proceeds), and to adaptively manage the region’s water resources into the future.


The authors thank two anonymous reviewers and Prof. Alan Hildrew for their considered suggestions that greatly improved the manuscript. This paper stems from research on tropical rivers, floodplains and estuaries conducted by the authors as part of the Tropical Rivers and Coastal Knowledge (TRaCK) research hub. TRaCK received major funding for its research through the Australian Government’s Commonwealth Environment Research Facilities initiative, the Australian Government’s Raising National Water Standards Programme, the Fisheries Research and Development Corporation and the Queensland Government’s Smart State Innovation Fund.