Introduction

Movement among habitats is essential for aquatic organisms to optimise growth and increase survival (Northcote 1978; Kahler et al. 2001; Dingle and Drake 2007). For many species, access to multiple habitats is required to complete their life cycle, with movement driven by a variety of abiotic (e.g. temperature, salinity, flow) and biotic (e.g. species interaction, predation, spawning) factors (Lucas and Baras 2000, 2008). Diadromous species utilise both marine and freshwater habitats to complete their lifecycle, and movement patterns can be specific to life-history stages, cohorts, or even between sexes (Lucas and Baras 2001, Brooks 2002, Dingle and Drake 2007). However, movement patterns are dynamic, with varying patterns often existing within a single species. Establishing these patterns for exploited species and defining baseline spatial ecology are essential for effective fishery assessment and management (Rosenfeld 2003; Cooke et al. 2012, 2016). This is particularly important given migratory species are under increasing global threat and the knowledge required to manage these species is often incomplete (Wilcove and Wikelski 2008; Lennox et al. 2019).

Many migratory fish species require access to both fresh and marine waters. Coastal river systems are highly complex and serve as important corridors for species that require both marine and freshwater habitats (McDowall 1988). This transition zone is highly dynamic, shifting in time and space due to the influence of tides, temperature, rainfall, and associated freshwater inflows (Pritchard 1967; McLusky and Elliott 2004). Diverse habitats exist within estuaries (e.g. wetlands, macrophyte beds) and provide critical nurseries and nutrient-rich foraging areas for various fish species (West and King 1996; Beck et al. 2001; Pihl et al. 2002; Barletta and Blaber 2007). A wide variety of fish species utilise estuaries, including marine migrants, freshwater migrants, and those that reside solely within the estuary (Elliott et al. 2007).

Robust information of the movement behaviours and associated environmental cues, of species within estuaries, is essential for the management of these habitats and the fisheries they support (Secor and Rooker 2005). Freshwater inflows are important for the productivity of estuaries as they transport nutrients and sediment downstream, enhance the productive estuarine ‘salt wedge’, and provide movement cues for some species (Loneragan 1999; Montagna et al. 2002; Heupel and Simpfendorfer 2008; Taylor et al. 2014; Amtstaetter et al. 2021). In the last 20 years, coastal environments throughout south-eastern Australia have experienced increased pressure from population growth and associated human activities in these zones (Roy et al. 2001). Human-influenced impacts to coastal areas, including pollution, urbanisation, and agricultural and industrial development, can change the natural behaviour of coastal rivers (Kennish 2002; Creighton et al. 2015; Brink et al. 2018). For example, anthropogenic changes in coastal river basins have resulted in the decrease of native species diversity and abundance in fish communities (Moraga et al. 2022) and a disconnect between marine and freshwater habitats (Potter et al. 2010; Lloyd et al. 2012).

Water resource development in coastal river catchments, including the regulation of freshwater inflows to estuaries, can have considerable impacts on coastal rivers and estuaries (Loneragan 1999; Alber 2002; Kennish 2002; James et al. 2013). Changes in freshwater flows affect estuarine fish populations both directly and indirectly (Whitfield 2005; Gillson 2011). In extreme cases, changes in inflow can restrict access to necessary habitats and impede an individual’s ability to complete their life cycle, resulting in them having to utilise sub-optimal habitats, abandon reproduction, or move away from the area altogether (Harris 1986). Furthermore, significant changes to environmental cues, such as temperature and flow that accompany large flow events, can cause migration disruptions (Harris 1983; Mallen-Cooper 1999; Walsh et al. 2012). Many Australian native fish species rely on large flow events for critical stages of their reproductive cycle, such as migration cues or for the transport of juveniles downstream (Harris 1986; Mallen-Cooper 2000). Freshwater inflows can indirectly alter estuarine ecosystems through changes to the hydrologic, salinity, and sediment regimes, which in turn can alter the physical habitats and presence and abundance of aquatic organisms (Kennish 2002). Over time, these impacts can cause population declines.

In this study, we aimed to determine the movement behaviour of an abundant member of the Ariidae family (Neoarius graeffei) in a naturally and anthropogenically dynamic estuarine system. Neoarius graeffei, a riverine sea catfish, is the most abundant species in the Clarence River, Australia. The Ariidae family is widespread with various species occurring throughout Asia, India, Africa, South America, New Guinea, and northern Australia and is harvested for sustenance and as an economic resource (Kailola 1990). Their widespread global abundance is made possible by their flexible diet which includes aquatic (i.e. fish, macroinvertebrates) and terrestrial items (i.e. plants, insects, seeds; Pusey et al. 2004). At present, N. graeffei from the Clarence River are only commercially fished in small quantities, but other Ariidae species are popular and an important sustenance fish throughout Asia, South America, and India (Kailola 1990). Breeding occurs annually from early November to early December for N. graeffei and, similarly to other members of the Ariidae family, is followed by a paternal mouthbrooding period of 2–8 weeks (Rimmer and Merrick 1983; Pusey et al. 2004; Gomon and Bray 2021). Previous studies have reported contrasting movement patterns throughout the freshwater-marine interface, including anadromy (Quinn 1980) in eastern Australia and euryhaline tendencies in northern Australia (Oughton 2014). Defining movement behaviours in N. graeffei, a representative high-biomass species, in response to the natural variations in hydrology, will help predict potential impacts to abundant species arising from future climate change and river developments.

Methods

Study Site

The Clarence River Basin, located in northern New South Wales (NSW), Australia, is one of the only unregulated coastal river systems on the east coast of Australia, with the basin covering approximately 22,400 km2. The Clarence River itself is the longest of the basin’s rivers at approximately 250 km long and has an extensive estuarine reach of approximately 110 km that supports one of the most productive estuarine fisheries in NSW. During periods of low flow, the salt wedge extends over ~ 50 km upstream from the river mouth, whereas large flow events can result in freshwater pushing many kilometres into oceanic waters (Burgess and Woolmington 1981). The lower reaches of the Clarence River are subject to many anthropogenic stressors including draining of wetlands for agriculture (including sugar cane and cattle grazing), aquaculture, commercial fisheries, and residential development (Burgess and Woolmington 1981). Poor floodplain management has led to runoff of acidic and anoxic water, resulting in many large fish kills in the Clarence River over the past 50 years (Ryder 2013). There have been several proposals to construct a large dam and inter-catchment transfer system on the Clarence River. There are concerns that such a scheme may impact fish migrations and natural biotic cues, but there is little biological data on possible impacts.

During this study, the Northern Rivers area of NSW, including the Clarence River, experienced a series of high rainfall events which led to severe flooding. These flood events, a result of La Niña weather patterns on the east coast of Australia, resulted in irregular river discharge throughout parts of the study period (June 2021 to June 2022, specifically late February to June 2022), with a major (1 in 1000 years) flood event occurring in late February/early March 2022. These flow events also affected temperature and conductivity in the Clarence River in the latter half of the study.

Study Species

Neoarius graeffei is a benthic, medium to large (max standard length (SL) 600 mm) member of the Ariidae family of sea catfish, and is found throughout most coastal northern Australia and southern Papua New Guinea freshwater, estuarine, and marine environments (Rimmer 1985b; Gomon and Bray 2021). Studies in eastern Australia have noted anadromous movement tendencies (e.g. Ellway and Hegerl 1972; Quinn 1980), although this pattern was not noted in the Clarence River system (Rimmer 1985c). Otolith microchemistry studies in northern Australia suggest that N. graeffei does not exhibit well-defined migration patterns typical of diadromous species and are more euryhaline, typified by moving between fresh and saline habitats with no specific patterns or timing (Oughton 2014).

Study Design

Acoustic Array

Daily and seasonal movement patterns of N. graeffei were recorded throughout the Clarence River and associated tributaries between June 2021 and June 2022. An array of 32 acoustic receivers (Innovasea VR2W 69kHz) were deployed throughout the lower Clarence Basin (Fig. 1), covering approximately 164 km of waterway length. Receivers were distributed throughout the estuary and associated tributaries of the Clarence River, which included the Mann River, Coldstream River, Orara River, and Sportsmans Creek and covered the entire salinity gradient, from the lower estuary to freshwater (Fig. 1).

Fig. 1
figure 1

Overview of acoustic telemetry receivers (coloured circles) used to monitor the movements of Neoarius graeffei across the lower Clarence River Basin. Receivers are grouped based on their location within the estuary. Conductivity and temperature loggers (CT) and a WaterNSW river height station (RH) are also shown. Fish capture (CPL) and release (RL) locations are indicated on the map

Full size image

All receivers were suspended vertically, inverted within the water column, and attached to semi-permanent structures such as navigation buoys, private jetties, or overhanging trees. In locations where these structures were not present, receivers were attached to a submerged cement block and held upright (hydrophone up) with 1 m of cable and an attached polystyrene float. The block and float were secured to shore via submerged cable and attached to a sturdy structure, such as a large tree or rock. Receivers were downloaded every 6 months using VUE software (vs. 2.8.0; Innovasea, Amirix, Nova Scotia, Canada).

Environmental Variables

Conductivity, used as a proxy for salinity, was recorded at four locations within the lower Clarence River using a logger (HOBO U24-002-C conductivity logger; Onset Computer Corporation, Massachusetts, USA). Loggers were programmed to measure and record conductivity (µS cm−1) and temperature (°C) each hour (Fig. 1). To prevent biofouling, loggers were placed in nylon stockings within perforated PVC pipe to allow water passage and were cable-tied to the receiver. Loggers were downloaded at the conclusion of the study using the HOBOware Pro Software (v 3.7.21, Onset Computer Corporation, Massachusetts, USA). River height (m) was used as a proxy for river discharge (ML d−1) due to the variation in river width throughout the Clarence River and was recorded daily by WaterNSW at Grafton (204400 station; Fig. 1).

Fish Capture and Tagging Procedures

Fish were captured using 20–30 m, 70-mm stretched mesh size monofilament gill nets. All nets were set within 10 km of the uppermost limits of the estuary (Copmanhurst, NSW; 29.586° S 152.776° E; Fig. 1). Upon capture, fish were transported to the Grafton Fisheries Centre (29.612° S 152.956° E) in 100-L tanks and held in large, aerated 9000-L tanks for up to 3 days prior to surgery. All tanks were filled with static aerated river water with approximately 8 g L−1 salt (NaCl) added to minimise stress and reduce the risk of fungal infections (Selosse and Rowland 1990; Mifsud and Rowland 2008).

Acoustic tags (Innovasea V13; 36 mm in length) were surgically implanted into 30 adult N. graeffei on 4th June 2021, 14 males and 16 females. Each tag was programmed to transmit a unique pulse series at randomly spaced intervals between 60 and 120 s. Tagged fish measured 436 ± 50 mm (mean ± SE) total length (TL; range = 366 − 585 mm) and 1012 ± 470 g in weight (range = 481–2488 g; 0.5–2.1% tag to body weight ratio). All fish were anaesthetised, measured (total length and weight), and surgically implanted with acoustic tags as described in Carpenter-Bundhoo et al. (2020). Sex was determined by the size of the pelvic fins (see Rimmer 1985c). Once equilibrium was reached, all fish were released at one site near the mid-section of the array (RL; Fig. 1).

Data Analysis

All analyses were conducted in R (v.4.2.1; R Core Team 2022) and based on receiver locations. Preliminary data manipulation was conducted using the ‘actel’ package, a specialised package designed to analyse and filter acoustic telemetry data through a systematic conditional pipeline (Flávio and Baktoft 2021). Using the ‘explore’ function, multiple quality checks were conducted to eliminate potential flaws in the data, including false detections. To understand N. graeffei movement patterns, the retained data was then analysed using the ‘RSP’ package (Refined Shortest Paths; Niella et al. 2020), which allows for realistic animal trajectories to be estimated within water boundaries, using a correlated random walk least-cost path analysis. Fish movement trajectories were estimated using 24-h maximum intervals (i.e. if the fish was not detected for longer than 24 h, this would characterise the beginning of a new track), and interpolated locations were included at 100-m intervals (Fig. 2).

Fig. 2
figure 2

Daily average (filled circles) and +/- SD of a water temperature, b conductivity, and c water height throughout the Clarence River from 1 June 2021 to 31 May 2022

Full size image

To understand the size of the areas used by N. graeffei, dynamic Brownian Bridge Movement Models (dBBMM) were calculated independently for both male and females, using the RSP package (Niella et al. 2020). Unlike more traditional movement models, dBBMM incorporate temporal and behavioural characteristics of movement paths into estimation of home ranges (Kranstauber et al. 2012). Effects of environmental variables on movement patterns of N. graeffei were analysed using two metrics for movement as the response variables: (i) size of the areas of space use (in square metres) for the 50% contour level of dBBMM (i.e. equating to the areas used during half of each corresponding tracking days; Fig. 3) and (ii) mean daily distance in relation to the Clarence River mouth (in km), using Generalised Additive Models (GAM) and Generalised Additive Mixed Models (GAMMs) with the ‘mgcv’ package (Wood 2018). Contour levels were set at 50% as it represents an area of higher use (areas where fish spent most of their time.) vs. 75% or 90% which are lower use areas. Models were run independently for the size of space use for each sex and for the overlapping areas between sexes to assess intra-specific variation in space use. The response variables were log-transformed, and models used Gaussian error distributions. Variable selection was based on the Akaike information criteria (AIC), which estimates the quality of each model and the best fit for the data (Akaike 1987). There was no correlation between river height (m) and temperature (r(359) = 0.080, p = 0.129). However, river height and conductivity were significantly correlated (r(359) = − 0.378, p ≤ 0.001) meaning that the two variables were not combined in the same model, and river height was used as a proxy for conductivity. The total GAM formula used to model male or female 50% dBBMM or area overlapping between the two sexes as the response variables, in relation to environmental variables as a fixed effects, was

Fig. 3
figure 3

Space use dynamic Brownian Bridge Movement Model of 50% and 90% contours for male (top) and female (bottom) Neoarius graeffei on 2021–06-05 and 2021–06-06

Full size image
$$\mathrm{log}({\mathrm{Area}}_{i}) \sim \mathrm{ River\;height }+\mathrm{ Temperature}$$

To account for possible intra-individual variation in the movement patterns observed, the environmental models of distances to the river mouth also included transmitter ID as a random factor, since the resolution of this dataset was at the individual level. The total GAMM formula used for sex-specific models for distance from the river mouth (distance) was

$$\mathrm{log}\left(\mathrm{Distance}\right) \sim \mathrm{ River\;height }+\mathrm{ Temperature }+ \left(1\;|\mathrm{ \;transmitter\;ID}\right)$$

To avoid collinearity issues when including month simultaneously to environmental variables in the same models, we analysed the temporal variation on the above same movement metrics by running separate GAM and GAMM, but included month as the fixed effect with a cyclic-cubic regression spline. The temporal GAM formula for area used (i.e. sex-specific 50% dBBMM area and size of the overlapping areas as the response variables) was

$$\mathrm{log}\left({\mathrm{Area}}_{i}\right) \sim \mathrm{ Month}$$

The GAMM formula used for the temporal sex-specific models for distance from the river mouth was

$$\mathrm{log}\left(\mathrm{Distance}\right) \sim \mathrm{ Month }+ \left(1\;|\;\mathrm{ transmitter\;ID}\right)$$

Visual assessment of the environmental data highlighted five high-flow events during the study period (see Supplementary Information Fig. S5). To quantify the influence of these events on fish movements, we used a temporal 10-day window to separate each event into ‘previous’, ‘during’, and ‘after’ periods (see Supplementary Information Fig. S5). A Generalised Linear Mixed Model (GLMM) was used to investigate the possible effects that each high-flow event period might have had on area used (50% dBBMM) and distance to river mouth (km). These models tested the fixed effects of the categorical variable time period (i.e. previous, during, and after) and included a variable Flow ID (categorical = 1 to 5) as random factors, to determine if movement patterns varied in relation to each event. The total GLMM formula used for the sex-specific models (i.e. including 50% dBBMM area or distance from the river mouth as the response variables) was

$$\mathrm{log}\left(\mathrm{Response}\right) \sim \mathrm{ Period }+ \left(1\;|\;\mathrm{ Flow\;ID}\right)$$

Results

No immediate post-surgery mortality was observed, with all fish detected throughout the study period with an average detection of 347 days (min = 275, max = 349). Tagged fish did not leave the study as no fish were detected on the upstream, freshwater terminal receivers and only one individual (female, transmitter ID # A69-1602 54065) was detected at the most downstream receiver located at the river mouth (Fig. 1). The median distance travelled by individual fish over the course of the study was 883.7 km (min = 397.8, max = 2,619.6). Most fish remained primarily in the mid- and upper estuary throughout the study (see Supplementary Information Figs. S3 and S4).

Males and females varied in length (males = 361.5 ± 34.68 mm; females = 394.87 ± 45.92 mm) and weight (males = 0.834 ± 0.29 kg; females = 1.17 ± 0.55 kg) but no significant difference (t = 0.369, p = 0.715) was detected in cumulative distance travelled between sexes (males = 1003 ± 568 km; females = 1079 ± 567 km). In female N. graeffei, there was no correlation between distance travelled and fish length (t = − 1.285, p = 0.220) or weight (t = − 0.917, p = 0.375). Similarly, males displayed no correlation between distance travelled and fish length (t = − 1.048, p = 0.315) or weight (t = − 0.735, p = 0.476). Also, there was no correlation between distance travelled and tracking duration (t = 0.631, p = 0.533; see Supplementary Information Fig. S2).

Movements of tagged fish were related to temperature and river height (Fig. 2). Tagged individuals were detected in all four reaches in the study area: lower estuary, mid-estuary, upper estuary, and freshwater (Fig. 1), spanning the salinity gradient from fresh (0 µS cm−1) to saline (42.5 µS cm−1; Fig. 2). Almost all detections were recorded in the mid- (49.02%; 21–45 km upstream from the river mouth) and upper estuary (45.15%; 58–103 km upstream). Freshwater detections accounted for 5.54% of all detections (99–165 km upstream), whilst the lower estuary accounted for only 0.29% (4–22 km upstream).

Large-Scale Movement Patterns

Visual inspection of the data revealed two major synchronised movements (see Supplementary Information Figs. S3 and S4). Firstly, 97% of tagged fish moved downstream into the mid-estuary immediately upon release in June 2021 after tagging. Secondly, an upstream movement to the upper estuary and lower freshwater reaches (58–125 km upstream) was made by most fish (93%) in September 2021. Other individual behaviours were also observed, such as the evidence of homing behaviour, with nine individuals returning to the area of capture (approximately 100 km upstream) for extended periods of time ranging from 2 weeks to 1 month. Additionally, some individuals exhibited residence at a particular receiver. For example, one individual (transmitter ID# 54094, male) spent half of the study period (September 2021–February 2022) at a particular freshwater location approximately 140 km upstream before moving downstream.

Group-Specific Patterns of Area Use

Temperature proved to be a significant determinant of movement (area used) for males and females and the overlap between sexes, with river height only influencing males and females independently (Table 1). As temperature increased from 16 to 23 °C, both males and females used significantly smaller areas (Fig. 4a). Once ~ 23 °C was reached, males tended to then use larger areas whereas the areas used by females continued to decrease (Fig. 4a). As river height increased, both males and females used larger areas until river height exceeded 2–3 m, after when the area used became smaller for both males and females (Fig. 4b). The areas in which males and females overlapped significantly decreased as temperature increased above 16 °C (Fig. 4c).

Table 1 Generalised Additive Models of area size variation at the 50% dynamic Brownian Bridge Movement Model contours, including the effective degrees of freedom (edf), reference degrees of freedom (Ref.df), F-values (F), and p-values (p) for environmental trends. Only the respective significant model variables are included
Full size table
Fig. 4
figure 4

Generalised Additive Models of 50% area used for male (green) and female (purple) Neoarius graeffei, including the corresponding effects of a water temperature and b river height and the overlap in area used between males and females and c water temperature. Shaded areas and dashed lines represent the 95% confidence intervals and null effects, respectively. Positive standardised partial residuals indicate fish occupied larger areas, whereas negative residuals represent smaller areas occupied by fish

Full size image

There were significant relationships between the size of the areas used and month for both sexes (Table 2). Females used smaller areas in the warmer months (November–March) but increased the size of the area used throughout the cooler months (April–September), peaking in mid-June (Fig. 5). Males showed the opposite trend and used larger areas in the summer months with a peak between December and January, with the size of area used decreasing throughout winter (Fig. 5).

Table 2 Generalised Additive Models of temporal trends of area size variation at the 50% dynamic Brownian Bridge Movement Model contours including the effective degrees of freedom (edf), reference degrees of freedom (Ref.df), F-values (F), and p-values (p). Only the respective significant model variables are included
Full size table
Fig. 5
figure 5

Generalised Additive Models of 50% area used for male (green) and female (purple) Neoarius graeffei including the corresponding temporal effects of month. Shaded coloured areas and dashed lines represent the 95% confidence intervals and null effects, respectively. Positive standardised partial residuals indicate fish occupied larger areas, whereas negative residuals represent smaller areas occupied by fish. Dark grey vertical rectangle represents spawning as per Rimmer (1985c)

Full size image

Group-Specific Movements in Relation to the River Mouth

All variables were significant for both sexes when evaluating distance from the river mouth, including the random factor tag ID which highlights significant variation in movement patterns between individuals (Table 3). Both males and females moved significantly further upstream as temperature increased until 23 °C with females moving further upstream than males (Fig. 6a). Increased river height above 1 m was significantly related to fish movements downstream for both sexes (Fig. 6b).

Table 3 Generalised Additive Mixed Models of distance from river mouth including the effective degrees of freedom (edf), reference degrees of freedom (Ref.df), F-values (F), and p-values (p). Only the respective significant model variables are included
Full size table
Fig. 6
figure 6

Generalised Additive Mixed Models of distance from the river mouth for male (green) and female (purple) Neoarius graeffei, including the significant effects of a water temperature and b river height. Shaded areas and dashed lines represent the 95% confidence intervals and null effects, respectively. Positive standardised partial residuals indicate fish were further from river mouth, whereas negative in residuals represents fish were closer to river mouth

Full size image

There was a significant relationship between month and distance to the river mouth for both species (Table 4). Both males and females inhabited environments closer to the river mouth during cooler months (May–August) and moved significantly further upstream in warmer months (November–February), with females moving further upstream than males (Fig. 7). Tag ID was again significant suggesting variation in movement patterns among individuals.

Table 4 Generalised Additive Mixed Model of distance from river mouth including the effective degrees of freedom (edf), reference degrees of freedom (Ref.df), F-values (F), and p-values (p) with * denoting the intercept. Only the respective significant model variables are included
Full size table
Fig. 7
figure 7

Generalised Additive Mixed Models of distance from the river mouth for male (green) and female (purple) N. graeffei, including the corresponding temporal effects of month. Increased standardised partial residuals indicate fish were further from river mouth, whereas a decrease in residuals represents fish were closer to river mouth. Dark grey vertical rectangle represents spawning as per Rimmer (1985c)

Full size image

Fine-Scale Movement in Relation to River Flow

When assessing the relationship between high-flow events and area used, both sexes (particularly females) used significantly larger areas during the 10-day periods of peaked high-flow events, with the size of the areas used during the after periods being statistically similar to the previous periods (Fig. 8a; Table 5). Furthermore, the magnitude of flow (Flow ID) did not significantly influence the size of the area used (Table 5).

Fig. 8
figure 8

Generalised Linear Mixed Models of spatial activity during high-flow events for male (green) and female (purple) Neoarius graeffei, including a area used (50%) and b distance from the river mouth (m). Horizontal dashed lines and error bars represent the null effect and standard errors, respectively. Positive residuals indicate an increase in a area used and b distance from the river mouth (further from river mouth), whereas negative residual represent a decrease in a area used and b distance from the river mouth (closer to river mouth)

Full size image
Table 5 Generalised Linear Mixed Model of area size variation at the 50% area used in relation to time during high-flow events, including the corresponding estimate (Est), standard error (SE), t-values (t), and p-values (p) with * denoting the intercept. Only the respective significant model variables are included
Full size table

The relationship between high-flow events and distance to the river mouth was also significant for both males and females, with both sexes (particularly males) moving closer to the river mouth during the peaks of the high-flow events (Fig. 8b; Table 6). Whilst females returned to similar regions during post-high-flow periods, males remained further downstream (Fig. 8b; Table 6). Flow magnitude was not significant for males or females, indicating that the fish moved similarly in relation to the distance from the river mouth between the different flow events.

Table 6 Generalised Linear Mixed Models of distance from the river mouth in relation to time during high-flow events, including the corresponding estimate (Est), standard error (SE), t-values (t), and p-values (p) with * denoting the intercept. Only the respective significant model variables are included
Full size table

Discussion

This study defined the movement patterns of an abundant species, N. graeffei in a coastal river system, at the southern extent of its reported range. Neoarius graeffei remained within the river throughout the calendar year, traversed the entire salinity gradient, and predominantly resided in the mid- and upper estuary of the river system. Although some distinctive movement patterns were evident, typical migration patterns across the marine-freshwater interface, such as diadromy, were not present as most of the animals were resident and did not exit the system. Elliott et al. (2007) proposed categorisation of estuarine fishes that builds upon the classical diadromy model and further classify fish that utilise estuarine habitats. Results from our study would suggest that N. graeffei could be best considered ‘estuarine migrants’ as they can complete their entire lifecycle within the estuary but did utilise freshwater habitats (Elliott et al. 2007). Other studies from the northern extent of the species range have found some similarities, with N. graeffei moving across salinity gradients for purposes other than reproduction (Oughton 2014; Pusey et al. 2017). This movement pattern is exhibited in other Ariidae family members including the Guri sea catfish Genidens genidens in South America (Silva Junior et al. 2013). The apparent lack of requirement for varying habitats across the marine-freshwater interface is likely associated with the species parental care strategy. Well-defined diadromous migration patterns are seen in freshwater species such as pinkeye mullet Trachystoma petardi and Australian bass Percolates novemaculeata, with both species needing to spawn in saline environments to ensure the development and hatching success of their non-adhesive, free floating eggs (Van Der Wal 1985). Parental care in fishes likely evolved in freshwater to avoid competition and predation seen in estuarine nurseries (Oppenheimer 1970; Kailola 1990). As paternal mouthbrooders, N. graeffei are not confined to specific habitats required by diadromous broadcast spawning species or those species that nest, and they are able to actively disperse to optimal habitats following changes to conditions or water quality (Rimmer and Merrick 1983).

Movement Patterns

Despite the lack of a definitive migration pattern, such as diadromy, there were distinctive movement patterns evident by N. graeffei during our study. Both location within the river system (distance from the river mouth) and area used were influenced by time of year, water temperature and river height, and the responses were often sex-specific. For example, temporal patterns were characterised by males and females inhabiting environments closer to the river mouth in late autumn–winter (May–August) and moving upstream into less saline habitats in the spring (September). Previous studies in the Clarence River have noted the absence of feeding and use of visceral fat stores in female N. graeffei during the late stages of gonad development (September–November) as well as an increase in N. graeffei stomach fullness from May to June, followed by a decrease from July–December (Rimmer 1985b). Winter occupancy of habitats closer to the river mouth could therefore be related to fish benefiting from greater abundances of food within estuarine environments, helping females with egg production, and ensuring males increase fat reserves for the upcoming period of parental care.

Movement towards upstream habitats, specifically the upper estuary and freshwater reaches, commenced in September by most tagged individuals. Upstream movement was also reported for N. graeffei in the Northern Territory (Australia) by Jardine et al. (2012), although these movements occurred at end of the tropical wet season (April–May). Upstream movements of N. graeffei coincided with the beginning of seasonal gonadosomatic changes for both males and females, which begins in September and peaks in November (Rimmer 1985c). This upstream movement could suggest that N. graeffei move further into the upper limits of the estuary prior to spawning in the Clarence River. This hypothesis is supported by the theory of parental care originating in freshwater due to the ability to seek the best habitats and conditions for larval development as well as lower rates of competition and predation (Oppenheimer 1970). Occupying upstream freshwater habitats as opposed to more estuarine areas further minimises intra-specific competition and predation risk upon adults and young. The presence of mouth brooding males was noted during the study in the upper estuary at the extent of the tidal limit. Preference for less saline environments was noted in the hardhead sea catfish Ariopsis felis during incubation of embryos due to their inability to tolerate elevated salinities (Harvey 1972). Adults inhabiting more upstream habitats were also noted through diet preferences. Differences in diet with size were reported from N. graeffei in the Northern Territory, with adults foraging in more upstream environments compared to juveniles that showed more estuarine/nearshore marine isotopic signatures (Pusey et al. 2020). There is anecdotal evidence of adult N. graeffei in freshwater habitats below the Clarence Gorge, preying on small-bodied fish such as migrating striped gudgeon Gobiomorphus australis and empire gudgeon Hypseleotris compressa (G. Butler, pers. com). Carbon isotopic studies from the Flinders River, in the Northern Territory, suggest juvenile N. graeffei inhabit floodplain habitats (Pusey et al. 2020). Young of the year could utilise summer/early autumn high-flow events typical of the Clarence River system to aid their transport downstream to these resource rich-estuarine and nearshore marine habitats. Whilst our study did not tag juvenile N. graeffei, future work with otolith microchemistry may aid in determining the potential estuarine/marine residency of juveniles in the Clarence River.bobo ka dreii

Environmental Variables

River height and temperature influenced the movements of N. graeffei during this current study. Research from Queensland, Australia, that were aimed at assessing the effectiveness of vertical slow fishways, noted that river flow and increasing temperature were factors that influenced the upstream movements of N. graeffei through fishways (Stuart and Berghuis 2002). Both males and females moved upstream in response to increases in water temperature. In the current study, females tended to use smaller areas as the temperature increased, whereas males used smaller areas until temperatures reached > 23 °C when the area used then increased in size. Rimmer (1985b) noted that females decreased their feeding activity prior to spawning which begins in October and coincides with the increases in water temperatures in the Clarence River. The increases in area used by males with warmer temperatures could be linked to spawning behaviour. Studies from the Clarence River noted that N. graeffei began to spawn when temperatures reached > 26 °C in November and December (Rimmer 1985c). After spawning, males incubate fertilised eggs in their buccal cavity for 6–8 weeks, during which they do not feed (Rimmer 1985a). An increase in area used with temperatures over 23 °C could be related to increased foraging after oral incubation is complete.

Increased river height resulted in both male and female N. graeffei occupying smaller habitats closer to the river mouth. Analysis of event-based movement responses indicated that short-term changes in the size of the area used and distribution in the estuary also occurred during high-flow events. For example, both male and female N. graeffei moved closer to the river mouth and across larger areas during the 10-day periods of high flow, followed by a return to pre-event behaviours in the following 10-day periods. Fish movement plots fish (see Supplementary Information) supported this idea that N. graeffei use large flow events to move opportunistically but often return to the same reach in which they inhabited pre-flood. This behaviour was seen in some giant kokopu Galaxias argenteus in New Zealand, who move with high-flow events but return as flows subside (David and Closs 2002). Spawning-related movement is well-studied in Australian bass, who are known to use large winter flows to migrate downstream to estuarine habitats but only during spawning season when gonads are mature, regardless of high-flow events (Harding et al. 2017).

Management Implications

Changes to the physico-chemical conditions in the estuary as a result of river regulation would have the largest impact on estuarine residents such as N. graeffei. Estuaries provide important nursery habitats for many freshwater and marine fish species whilst providing food sources and sheltering and spawning habitat for species that reside in them throughout their lifespan. Altering a river’s natural flow regime through river regulation can reduce recruitment of all species that use estuaries, including N. graeffei, through habitat destruction and modified salinity gradients (Whitfield 2005). Impacts of flow modification on seagrass habitats, which young-of-year N. graeffei have been reported to use (Pusey et al. 2020), may affect the recruitment of N. graeffei and other estuarine species. Furthermore, although large-flow events cause short-term declines in abundance of fish species, they also provide a natural ‘reset’ of habitats and functions within the estuary (Whitfield 2005). Freshwater inflows in a coastal river cause dynamic shifts within estuarine habitats with large flow events pushing the salt wedge further downstream and increasing the transitionary zone within the estuary. Ensuring the continued presence of large freshwater inflows and the expansion of these brackish estuarine habitats decreases intra-specific competition among estuarine residents such as N. graeffei due to an increase in resources. Furthermore, coastal rivers will also experience changes due to climate change that will also affect estuarine residents such as N. graeffei. Estuaries will likely experiences changes associated with freshwater inputs such as changes in salinity gradients, dissolved oxygen, and suspended sediments with estuarine residents heavily impacted by salinity changes spatially as well as physiologically (Gillanders et al. 2011). Together, the cumulative impacts from climate change and potential river regulation throughout coastal rivers will be felt particularly on estuarine species such as N.graeffei.

Defining the spatial ecology of an abundant species, such as N. graeffei, provides a snapshot of potential ecological responses to anthropogenic stressors that may be experienced by other estuarine species. For abundant species more generally, small declines in abundance may have effects on broader ecosystem processes (Ellison et al. 2005; Gaston 2008, 2010). As a species contributing substantial biomass in coastal river systems throughout its range, any potential changes to N. graeffei populations arising due to anthropogenic impacts may well have broader consequences for the entire estuarine ecosystem.

Conclusion

Our findings emphasise the reliance of N. graeffei on natural cues that structure their distribution throughout the landscape and the links with essential life history processes. Changes to the natural river cycle, particularly the regulation of large flow events, which frequently occur in the summer and autumn throughout the Clarence River, are likely to have direct impacts on the movements of this species, as well as the quality and quantity of available estuarine habitat. Used by a wide range of fish species, estuaries are multifaceted habitats with the ability to support their own residents as well as support those species who utilise the areas as stopovers or seasonally. Estuarine residents, such as the blackchin tilapia Sarotherodon melanotheron in Africa, and other Ariidae family members (e.g. Madamango sea catfish Cathorops spixii) in north-east Brazil, are important species in sustenance fisheries (Elliott et al. 2007; Barletta et al. 2010). Knowledge of life history traits for species that have a large biomass within estuaries can be used as an indicator in environmental quality and future changes (Barletta and Costa 2009). This is especially important due to the nature of coastal rivers and estuaries and the multitude of anthropogenic stressors that they currently face.