The implications of brief floodplain inundation for local and landscape-scale ecosystem function in an intermittent Australian river

Introduction

The natural flow regime of a river has been described as the ‘master variable’, where the magnitude, timing, predictability and duration of flow underpin ecological integrity across the riverine landscape (Poff et al. 1997). On the floodplain, this natural hydrological variability through space and time interacts with geomorphology, promoting habitat heterogeneity (Junk et al. 1989; Poff and Ward 1990; Poff et al. 1997; Amoros and Bornette 2002), where local environmental conditions drive ecosystem structure and functioning (Thorp et al. 2006; Warfe et al. 2011; Rolls et al. 2012). This is particularly evident in systems with discrete ‘wet’ and ‘dry’ seasons, such as those in the tropics, where seasonal extremes of water availability across the landscape influence ecological processes in line with the prevailing physical and chemical conditions (Junk et al. 1989; Lewis et al. 2000; Winemiller et al. 2014; Feitosa et al. 2019).

In the tropics, seasonal flooding during the wet season drives patterns of floodplain productivity, where terrestrial nutrient subsidies, warm water temperatures and high light availability stimulate primary productivity with flow-on effects to higher trophic levels (Junk et al. 1989; Lewis et al. 2000; Pettit et al. 2011, 2017). However, there is an assumption that only large flood events with long water-residence periods provide a significant contribution to both local and landscape-scale ecosystem functioning (Junk et al. 1989). Indeed, much of the relevant literature pertains to systems with vast floodplains, where inundation may persist for several months (Junk and Furch 1993; Sparks 1995; Thomaz et al. 2007; Pettit et al. 2011; Molinari et al. 2021; Mosepele et al. 2022). There is comparatively limited information regarding rivers that exhibit brief floodplain inundation, where floodplain waterbodies isolate quickly from the main river channel and reconnect only the following wet season. Many of these floodplain habitats are ephemeral or semi-permanent, and pool persistence is generally dependent on the timing, magnitude and spatial extent of wet-season flood flows.

Local environmental conditions drive ecosystem structure and functioning (Thorp et al. 2006; Warfe et al. 2011; Rolls et al. 2012). Where physical, chemical and biological characteristics differ among habitats, it is likely that ecological processes, such as predation and primary production, vary in strength also. The relative influence of these fundamental ecological processes can be inferred by assessing the strength of consumer regulation of food resources (top–down forces) relative to resource limitation (bottom–up forces). Top–down and bottom–up forces can influence food-web structure (Koning and McIntyre 2021), biotic community structure (das Candeias et al. 2022) and ecological functioning (Power 1990) in riverine ecosystems. In floodplain rivers of the neo-tropics, bottom–up forces predominate (Trevisan and Forsberg 2007; Rejas and Muylaert 2010; Frau et al. 2019), although top–down control can vary in strength through space and time as low-water conditions increase predator densities (Winemiller et al. 2006, 2014). The literature examining top–down and bottom–up processes across northern Australia show a similar pattern (Garcia et al. 2015; Turschwell et al. 2019), although much of this research pertains to the main river channel and often with a focus on benthic algae in recognition of the importance of this resource in food webs across the wet–dry tropics (Douglas et al. 2005; Jardine et al. 2012a; Pettit et al. 2017). However, in a recent review Burrows et al. (2022) concluded that other energy sources may provide a significant contribution to productivity, depending on the stage of an individual’s lifecycle, hydrological period, or habitat. Indeed, there is strong evidence that the phytoplankton resource pathway, from zooplankton, is important for many fish species found in northern and central Australia (Medeiros and Arthington 2008; Pusey et al. 2021; Pratt et al. 2023). Moreover, zooplankton are typically the main food item consumed by most fish species during the larval life-stage (Nunn et al. 2012). Zooplankton are often highly abundant in floodplain pools relative to the main river channel (Godfrey et al. 2021; Pratt et al. 2023) and thus may provide an important food resource in these habitats.

In the Fitzroy River, located in the wet–dry tropical region of north-western Australia, floodplain inundation typically lasts for days to several weeks (Jardine et al. 2012a). This contrasts with other large rivers across northern Australia where water may remain on the floodplain for several months, for example, the East Alligator River, up to 6 months, the Daly River, up to 4 months, and the Mitchell river, up to 2 months (Pettit et al. 2011; Jardine et al. 2012a). Currently, there is limited knowledge of the physical, chemical and biological characteristics of floodplain pools across northern Australia in general, and no thorough characterisation of these habitats has been conducted in the Fitzroy River. Existing research regarding floodplain pools in the Fitzroy River has primarily focussed on fish (Jardine et al. 2012a; Beesley et al. 2020; Lear et al. 2023; Pratt et al. 2023) and macroinvertebrates (Beesley et al. 2023); thus, knowledge of their contribution to local and landscape-scale ecosystem functioning is incomplete. These habitats are under threat from water-resource development and information regarding the role these habitats play in the wider riverine landscape can help craft water planning policy to maintain ecological integrity across the catchment.

In this study, we characterise the physical, chemical and biological conditions of floodplain waterbodies and the main channel of the Fitzroy River. We also examine the spatial and temporal dynamics of top–down and bottom–up processes influencing a primary producer (phytoplankton) and a primary consumer (zooplankton). The Fitzroy River is an ideal system to conduct this study because floodplain inundation is typically brief, meaning floodplain pools isolate quickly from each other and the main channel, providing spatially distinct habitat types that diverge in their physical and chemical character over time. We hypothesise that (1) bottom–up processes will dominate in both main-channel and floodplain pools, as is typical in other tropical rivers, and that (2) top–down forces acting on zooplankton (fish predation) will be strong in floodplain pools, owing to the abundance of zooplanktivorous fish. Finally, we expect (3) the relative strength of top–down forces on the floodplain to increase over time as the dry season progresses and consumer densities increase. Understanding the dynamics of fundamental ecological processes at a local and landscape scale can assist managers when crafting water planning policy, protecting aspects of the natural flow regime that promote habitat heterogeneity and maintain ecosystem functioning across the catchment.

Materials and methods

Study site

The Fitzroy River is located in a savanna landscape in the wet–dry tropical region of north-western Australia (Fig. 1). The river is set within a catchment of 94,000 km², with a main channel ~700 km long and mean annual discharge of 6600 GL (Petheram et al. 2018a). The lower ~350 km of the river, downstream of Geikie Gorge (Fig. 1), is characterised by a broad flat floodplain and wide meandering main channel with occasional anabranches (Taylor 1999), contrasting with the narrow channel and marginal floodplain associated with steep-sided gorges typical in the upper reaches. The distinct wet (November–April) and dry (May–November) seasons are predictable (Bureau of Meteorology 2024) and marked by significant hydrological variability leading to extreme periods of water availability (both high and low) across the catchment. Inter-annual variability in flow is high and the river is classified as ‘wet-season highly intermittent’ by Kennard et al. (2010). Wet-season flows connect main channel pools and backfill floodplain distributary creeks, facilitating ecological exchange between main channel and floodplain habitats. Overbank inundation of the floodplain is typically brief (days to weeks), in contrast to many other large rivers in the wet–dry tropics where water may remain on the floodplain for up to several months (Jardine et al. 2012a). During the dry season, the main channel disaggregates into a series of elongated pools separated by shallow runs. Deep water creek systems on or adjacent to the floodplain, but separate from the main channel, hold water year round (Lear et al. 2023). Shallow-water (<2 m), semi-permanent floodplain pools formed in topographic depressions and distributary creeks create a mosaic across the landscape and contract under drying conditions. Flow in the river is mostly unmodified with the main regulatory feature a 3-m-high barrage at Camballin (Fig. 1). However, there is potential for water-resource development to support irrigated agriculture in the lower river catchment (Petheram et al. 2018b). If development proceeds, there is a possibility of diverting water from distributary creeks into off-stream storage facilities, an action that could decrease hydrological connectivity between the river and its floodplain (Karim et al. 2018).

Fig. 1.

The location of main-channel (n = 11) and floodplain sites (n = 14 pond-form and n = 6 channel-form pools) sampled in the Fitzroy River, Kimberley, Western Australia.

Results

In total, 16,497 individual zooplankton were counted, measured and identified from 20 floodplain pools and 10 main-channel pools. In total, 25,827 individual fishes comprising 17 species were caught from the 20 floodplain pools. From this total, 25,393 individuals belonged to the six species identified as zooplanktivores, which were used in analysis of top–down and bottom–up processes in floodplain habitats.

Physical and chemical characterisation

Physical and chemical conditions varied substantially among sites. The greatest differences were observed between the main-channel and pond-form floodplain pools, primarily being related to differences in water depth, habitat complexity, turbidity, dissolved nitrogen content and Chl-a biomass. These parameters were strongly loaded along PC1, which accounted for 45.83% of variation (Fig. 2). Main-channel pools tended to be deep with clear water and high habitat complexity, whereas pond-form floodplain pools were shallow and turbid with greater nutrient loading, Chl-a biomass and homogenous habitat. Channel-form floodplain pools often represented a mid-point between the extremes of pond-form floodplain pools and the main channel (Fig. 2). Water depth and habitat complexity tended to be greater in channel-form pools than pond-form pools, and turbidity and Chl-a were generally lower. However, channel-form floodplain pools were still typically shallower, more turbid and with higher Chl-a concentrations than in the main-channel pools (Table 1, Fig. 3, 4). Differences in habitat complexity between macro-habitat and meso-habitats were primarily driven by variable water depths and substrate types (Fig. 3). Physical and chemical conditions were substantially more variable in floodplain habitats than in the main channel, with pond-form floodplain pools displaying the greatest variability (Table 1, Fig. 3).

Fig. 2.

Principal component analysis and ordination bi-plot of physical, chemical and chlorophyll-a biomass data collected in pond-form (n = 14) and channel-form (n = 6) floodplain pools and main-channel pools (n = 6).

Table 1.Summary of physical and chemical data collected from pond-form (n = 14) and channel-form (n = 6) floodplain pools and main-channel pools (n = 6).

Physio-chemical parameter	Floodplain		Main channel
	Pond form	Channel form
Average water depth (m)	0.39 ± 0.19	0.74 ± 0.32	4.68 ± 3.02
Ammonia (μg L−1)	80.09 ± 187.01	16.19 ± 5.04	15.57 ± 4.56
Nitrate and nitrite (μg L−1)	17.96 ± 24.01	4.57 ± 1.22	4.35 ± 0.84
Total dissolved inorganic N (μg L−1)	98.04 ± 187.96	20.7 ± 6.08	19.91 ± 5.8
Phosphorus (μg L−1)	10.31 ± 12.71	3.98 ± 2.67	2.66 ± 0.35
Dissolved oxygen (mg L−1)	10.06 ± 4.20	8.07 ± 1.11	8.14 ± 0.85
pH	9.12 ± 1.09	8.72 ± 0.47	8.79 ± 0.10
Electrical conductivity (μS cm−1)	250.93 ± 135.00	217.31 ± 96.94	350.37 ± 30.40
Salinity (ppt)	0.12 ± 0.07	0.10 ± 0.05	0.16 ± 0.01
Turbidity (NTU)	160.44 ± 244.73	24.92 ± 25.65	2.43 ± 1.64
Habitat-complexity score	4 ± 2.25	7.33 ± 1.63	7.17 ± 0.41
Large woody debris (%)	0.29 ± 0.83	0.75 ± 0.61	0.32 ± 0.38
Small woody debris (%)	5.0 ± 13.09	2.92 ± 3.58	0.32 ± 0.38
Mud substrate (%)	89.93 ± 26.27	82.50 ± 17.82	22.17 ± 19.70
Sand substrate (%)	6.07 ± 13.37	11.33 ± 15.32	57.0 ± 17.61
Gravel substrate (%)	0.36 ± 0.63	2.83 ± 4.02	20.0 ± 22.80
Cobble substrate (%)	1.36 ± 4.80	3.33 ± 6.06	0.83 ± 2.04
Macrophytes (%)	26.0 ± 36.73	18.67 ± 30.54	4.33 ± 5.71

Values show means ± s.d.

Fig. 3.

Box plots of raw physical and chemical data collected in pond-form (n = 14) and channel-form (n = 6) floodplain pools and main-channel pools (n = 6). Bold line indicates the median value, the two hinges are the 25th (lower) and 75th (upper) quartiles, whiskers represent the inter-quartile range and black circles are outlying points.

Fig. 4.

Box plot showing (a) log-transformed chlorophyll-a biomass and (b) log-transformed zooplankton biomass in pond-form (n = 14) and channel-form (n = 6) floodplain pools and main-channel pools (n = 10). Bold line indicates the median value, the two hinges are the 25th (lower) and 75th (upper) quartiles, whiskers represent the inter-quartile range and black circles are outlying points.

Ecological processes

Top–down and bottom–up processes varied in strength spatially between macro-habitats (i.e. main-channel and floodplain), but no changes were detected over time. In floodplain pools, bottom–up processes controlled phytoplankton and zooplankton biomass. Chl-a was strongly positively influenced by dissolved inorganic nitrogen (Table 2, Model i; Fig. 5) and zooplankton biomass was positively influenced by Chl-a concentration (Table 2, Model ii; Fig. 6). No top–down effects were observed in either floodplain model, with both zooplankton biomass and zooplanktivorous fish abundances having a negligible effect on Chl-a biomass and zooplankton biomass respectively (Table 2). Contrary to the strong bottom–up processes observed on the floodplain, in the main channel we found no evidence of bottom–up or top–down effects and none of the covariates were good predictors of Chl-a or zooplankton biomass (Table 2, Models iii and iv). We found no evidence of temporal variation in top–down and bottom–up processes. On the floodplain, the strength of top–down and bottom–up forces remained consistent between the early dry season and late dry season (Table 2). Additionally, we observed no change in the strength of top–down and bottom–up forces between floodplain meso-habitats, i.e. pond-form pools or channel-form pools (Table 2).

Table 2.Parameter estimates from analysis of top–down and bottom–up forces acting on (i) phytoplankton in floodplain pools (n = 20), (ii) zooplankton in floodplain pools (n = 20), (iii) phytoplankton in main-channel pools (n = 10) and (iv) bottom–up forces acting on zooplankton in main-channel pools (n = 10).

Model	Variable	Effect size	s.d.	LCI	UCI	w
(i) Phytoplankton in floodplain pools	Dissolved inorganic nitrogen A	0.724	0.135	0.416	0.954	1.00
	Phosphorus	0.097	0.135	−0.037	0.408	0.52
	Zooplankton biomass	0.064	0.134	−0.183	0.395	0.48
	Meso-habitat	0.061	0.194	−0.336	0.545	0.36
	Season	−0.268	0.328	−0.991	0.073	0.59
(ii) Zooplankton in floodplain pools	Chlorophyll-a biomass A	1.418	0.422	0.613	2.237	0.99
	Zooplanktivorous fish CPUE	0.193	0.355	−0.255	1.080	0.45
	Meso-habitat	−0.019	0.373	−1.090	0.740	0.39
	Season	−0.619	0.630	−1.714	0.273	0.67
(iii) Phytoplankton in main-channel pools	Dissolved inorganic nitrogen	0.013	0.058	−0.069	0.200	0.20
	Phosphorus	−0.009	0.056	−0.183	0.069	0.17
	Zooplankton biomass	0.037	0.086	−0.027	0.274	0.30
(iv) Zooplankton in main-channel pools	Chlorophyll-a biomass	0.161	0.381	−0.624	0.946	0.50

Credible intervals that do not overlap zero indicate significance at α = 0.05. w is an indication of the support for the inclusion of a given variable in the best predictive model, with 0 = low support and 1 = strong support. s.d., standard deviation; LCI, 95% lower credible interval; UCI, 95% upper credible interval; w, inclusion probability.

^A Signifies parameter inclusion in the best predictive model and CIs that do not overlap zero. Meso-habitat refers to the effect of pool type (pond form, n = 14; or channel form, n = 6), channel form is the default in Models i and ii.

Fig. 5.

Model-predicted chlorophyll-a biomass v. dissolved inorganic nitrogen concentration in floodplain pools (n = 20). Area of shading surrounding the central line represents 95% Bayesian credible intervals. Circles and triangles are transformed data representing pond-form (n = 14) and channel-form (n = 6) pools respectively.

Fig. 6.

Model-predicted zooplankton biomass v. chlorophyll-a biomass in floodplain pools (n = 20). Area of shading surrounding the central line represents 95% Bayesian credible intervals. Circles and triangles are transformed data representing pond-form (n = 14) and channel-form (n = 6) pools respectively.

Discussion

On the floodplain, bottom–up processes drove water-column productivity to a degree not observed in the main channel, providing a valuable resource pathway supporting consumer populations locally and catchment wide. In floodplain pools, high nutrient concentrations increased phytoplankton biomass, providing an abundant food resource, which boosted zooplankton populations. The strength of bottom–up forces on the floodplain was consistent in the early and late dry season for primary producers and primary consumers alike. No significant top–down effects were detected in either meso-habitat. Floodplain pools showed considerable physical, chemical and biological heterogeneity, which contrasted with the relatively uniform main channel. Floodplain pools were shallow and more turbid with higher nutrient loads than the main-channel, which was deep, clear and nutrient-limited. Phytoplankton and zooplankton biomass were considerably greater and more variable in floodplain pools than in the main-channel, where both were largely absent. In the face of water-resource development, maintaining aspects of the natural flow regime that promote habitat heterogeneity in the lower catchment is crucial for preserving local and landscape-scale ecosystem functioning.

Hydrological variability through space and time promotes habitat heterogeneity (Junk et al. 1989; Poff and Ward 1990; Poff et al. 1997; Amoros and Bornette 2002). The physical and chemical characteristics of main-channel and floodplain habitats in the Fitzroy River were markedly different. For instance, main-channel sites tended to be deep with clear water and high habitat complexity, whereas pond-form floodplain pools were shallow and turbid with greater nutrient loading and homogenous habitat. Pond-form pools typically displayed the greatest heterogeneity, which was in stark contrast to the main-channel, which displayed little variation in physical and chemical conditions among pools. Channel-form floodplain pools often represented a mid-point between the extremes of the main-channel and pond-form pools. Biotic assemblages displayed a pattern concurrent with the different physical and chemical conditions observed across macro- and meso-habitats. Phytoplankton and zooplankton biomasses were greater and more variable in pond-form floodplain pools compared to channel-form floodplain pools and substantially greater than in the main channel.

Local physical and chemical conditions, along with biological communities, influence ecosystem structure and function (Thorp et al. 2006; Warfe et al. 2011; Rolls et al. 2012). In this study, we have shown that the substantial differences in physical and chemical conditions and biological assemblages between floodplain and main-channel habitats in the Fitzroy River drive differences in the strength of top–down and bottom–up processes. Bottom–up forces in floodplain pools were greater than those observed in the main channel, where statistically significant bottom–up or top–down effects were not detected. Bottom–up effects on the floodplain were driven by nutrient availability, which increased phytoplankton biomass, boosting zooplankton populations. Elevated nutrient concentrations in floodplain habitats have been observed elsewhere in the wet–dry tropics and have been shown to stimulate water-column Chl-a concentrations (Pettit et al. 2012; Molinari et al. 2021). There are likely to be several mechanisms that contribute to increased nutrient loads in shallow floodplain pools, including mobilisation of latent terrestrial energy sources during flooding (Baldwin and Mitchell 2000; Balzer et al. 2023), wind-derived re-suspension of nutrient-rich sediments (Søndergaard et al. 1992) and substrate disturbance or waste inputs from large herbivore activity, e.g. cattle (Dixon et al. 2011; Pettit et al. 2012). Deep water and dense riparian vegetation at main-channel sites are likely to buffer against wind-driven sediment re-suspension, and annual wet-season flows scour sandy benthic substrate, transporting nutrients downstream to the estuary (Townsend and Padovan 2005). Steep banks probably restrict cattle access to the river along large parts of the main-channel, limiting direct nutrient inputs to local ‘hotspots’ of cattle activity. Although field observations of pugging and cattle presence suggested increased activity at some floodplain pools v. the main channel, cattle activity was not quantified as part of this study.

Our finding that bottom–up forces were dominant is consistent with studies of other tropical floodplain systems globally (Trevisan and Forsberg 2007; Rejas and Muylaert 2010; Frau et al. 2019; Toruan et al. 2022). In the Australian wet–dry tropics, few studies have examined phytoplankton on the floodplain and none have simultaneously assessed the influence of top–down and bottom–up processes. However, pertinent research from the Cloncurry (Faggotter et al. 2013) and Mitchell (Pettit et al. 2012) rivers confirmed that nitrogen is a key driver of phytoplankton production on northern Australian floodplains. Resource limitation of phytoplankton has also been observed in the main channel of several rivers in the wet–dry tropics (Townsend and Edwards 2003; Webster et al. 2005; Townsend et al. 2012; Faggotter et al. 2013); however, the present study found no direct evidence for bottom–up control of phytoplankton in the Fitzroy River main channel. Light availability and high flow velocity can also constrain phytoplankton growth (Burford et al. 2012), although these factors are likely to have had little effect because water conditions were clear, shading from riparian vegetation was minimal and limited to channel margins and flow velocity was low. It is probable that phytoplankton production was constrained by nutrient availability; however, this relationship could not be demonstrated because of the low and uniform dissolved inorganic nitrogen, phosphorus and Chl-a concentrations in the main-channel. We found no evidence of consumer regulation of phytoplankton or zooplankton in floodplain or main-channel habitats, a finding consistent with a study of the Cloncurry River, Queensland (Faggotter et al. 2013). Across South America and temperate North America, consumer regulation of zooplankton by fish and phytoplankton by zooplankton has been observed (Schaus et al. 2002; Frau 2022). However, for significant top–down effects to occur, the intensity of predation needs to be sufficiently high, which even in aquatic systems is likely to be limited in time and space (Polis and Strong 1996). The impact of top–down forces in freshwater habitats can also be influenced indirectly by consumer effects on nutrient cycling. For example, fish and zooplankton, by excretory processes, can supply nitrogen and phosphorus at rates capable of supporting the nutrient demands of primary producers (Vanni 2002). Indeed, this indirect bottom–up process can override any top–down effect of consumers on phytoplankton or zooplankton biomass (Rejas et al. 2005).

The patterns and strength of top–down and bottom–up forces in floodplain habitats in the Fitzroy River were similar in the early and late dry season. Early in the dry season post-flooding, bottom–up forces are typically strong owing to increased availability of nutrients, which boosts primary production (Bunn et al. 2006; Pettit et al. 2012). Other studies have demonstrated that consumer regulation of resources can vary in strength through time as a function of seasonal hydrological variability, where low-water conditions during the dry season facilitate an increase in top–down forces as a result of increased consumer density (Winemiller et al. 2014; Frau et al. 2015). In central Australia, fish biomass in isolated waterholes has been shown to decrease significantly as the dry season progresses (Arthington et al. 2005; Burford et al. 2008), potentially reducing the intensity of top–down forces on zooplankton. Moreover, the release of nutrients into waterholes due to consumer mortality can replenish the available nutrient pool, which otherwise declines over time, stimulating primary production and maintaining the strength of bottom–up forces (Burford et al. 2008). Fish mortality during the late dry season may therefore directly reduce the strength of top–down forces at a time when low-water conditions would usually facilitate an increase in predation, and indirectly increase the strength of bottom–up forces during a period of diminishing nutrient availability for primary producers. This process may stabilise top–down and bottom–up processes over the dry season and contribute to the absence of a seasonal pattern observed on the floodplain in the Fitzroy River. However, repeat sampling of pools throughout the dry season is required to verify this.

In freshwater ecosystems around the world, the phytoplankton resource pathway provides an important contribution to consumer biomass (Schaus et al. 2002; Frau 2022). However, in the wet–dry tropics and dryland systems of northern and central Australia, benthic algae is considered to be the primary resource pathway sustaining both main-channel and floodplain heterotroph populations (Bunn et al. 2003; Douglas et al. 2005; Pettit et al. 2017). Nonetheless, there is mounting evidence that the phytoplankton pathway, from zooplankton, may provide an important contribution to consumer diets as well as influence broader ecosystem functioning (Medeiros and Arthington 2008; Pusey et al. 2021; Pratt et al. 2023). Indeed, Pratt et al. (2023) demonstrated that zooplankton biomass was the primary mechanism driving the growth rate of young-of-year bony bream in the Fitzroy River (WA), where high zooplankton biomass was exclusive to the floodplain. Faster growth in floodplain habitats led to an increase in modelled survival, suggesting that the floodplain has the potential to provide a disproportionally large contribution of this ecologically important species to the riverine meta-population. Similar benefits may apply to other fish species given that zooplankton are the primary food resource of most larval fishes at the onset of exogenous feeding, regardless of adult dietary guild (Barlow et al. 1993; Nunn et al. 2012). Zooplankton also contribute to the diet of many waterbirds, which are capable of transporting energy across catchments (Kingsford et al. 2010). Thus, the floodplain and the abundant zooplankton, which are exclusive to this habitat, are not just important for local food webs but may also contribute to main-channel and landscape-scale ecosystem functioning.

This study is one of few in northern Australia to extensively characterise floodplain wetlands, providing an important contribution to baseline data of the physical, chemical and biological conditions in these habitats. With the exception of Pettit et al. (2012), the majority of studies have focussed on the main channel and only occasionally included a floodplain pool. Pertinent information from this body of research has been synthesised and compared with the present study to establish the transferability of our results. The physical and chemical conditions of pools sampled were consistent with those that have been reported in other studies of the Fitzroy River (Kay et al. 1999; Fellman et al. 2011; Burrows et al. 2020; Gleiss et al. 2021) and elsewhere in the wet–dry tropics (Dixon et al. 2011; Pettit et al. 2012; Townsend et al. 2012; Faggotter et al. 2013). Main-channel Chl-a concentration was in line with results reported by Gleiss et al. (2021) in two dry-season refuge pools in the Fitzroy River main stem, and is consistent with what has been reported in the main channel of other river systems across northern Australia (Webster et al. 2005; Rayner et al. 2009; Townsend et al. 2012; Jardine et al. 2015). Elevated concentrations of water-column Chl-a observed in floodplain pools compared with the main channel were similar to those reported in several other floodplain rivers in the wet–dry tropics (Pettit et al. 2012; Molinari et al. 2021) and elsewhere in Australia (Ning and Nielsen 2011). Zooplankton biomass is consistent with that reported by Pratt et al. (2023) in both main-channel and floodplain sites in the Fitzroy River. Zooplankton densities are in the same range as that reported by Faggotter et al. (2013) for main-channel and floodplain pools in the Edith River, Northern Territory, and for ephemeral floodplain wetlands in south-eastern Australia (individuals per litre) (James et al. 2008), but were considerably higher in both main-channel and floodplain habitats than that reported in the slack waters of the Mulgrave River, tropical Queensland (Godfrey et al. 2021). Overall, the results from the present study are consistent with what has been reported elsewhere in the literature, suggesting the characterisation of floodplain pools presented herein may have high transferability to other systems across northern Australia.

In this study, we have demonstrated that floodplain pools fill a unique role in the wider riverine ecosystem, even when floodplain inundation is brief, a finding that has implications for local and landscape-scale ecosystem functioning. This information can inform the management of the Fitzroy River as well as other systems throughout northern Australia and elsewhere. In the Fitzroy River, the natural flow regime is relatively unmodified compared with that of other large rivers globally. However, along with several other rivers in the region, there is interest in developing its water resources to support irrigated agriculture (Petheram et al. 2018b), a process which will likely alter the natural flow regime. Indeed, modelling suggests that development has the potential to reduce hydrological connectivity between the river and its floodplain, decreasing the area, depth and duration of seasonal floodplain inundation (Karim et al. 2018). Floodplain habitats in the Fitzroy River display substantial physical, chemical and biological heterogeneity, and support water-column productivity at a level well above that which is observed in the main channel. Pond-form floodplain pools displayed the highest nutrient concentrations and supported the greatest phytoplankton and zooplankton biomass. For this subhabitat type to fill, larger overbank flood events that inundate the floodplain proper are required. Channel-form floodplain habitats are likely to back-fill more frequently and during periods of lower main-channel discharge than pond-form pools. Water managers should account for the different inundation requirements needed to maintain both floodplain meso-habitats when designing water planning policy. Floodplain habitats are not just important for local biological communities, they also contribute to landscape-scale ecosystem functioning. Floodplain contributions to the main channel have been observed in other river systems globally (Tockner et al. 1999; Jardine et al. 2012b; Furst et al. 2014; Sturrock et al. 2022). Thus, any modification of the natural flow regime that might negatively affect floodplain habitats has the potential to alter ecosystem functioning at a local and landscape scale. It is therefore vital that water planning policy for the lower Fitzroy River catchment protects the aspects of the natural flow regime that promote habitat heterogeneity, to preserve ecosystem functioning in both floodplain and main-channel habitats.