Using underwater cameras to describe the reproductive behaviour of the endangered eastern freshwater cod Maccullochella ikei

Authors


G. L. Butler, NSW Department of Primary Industries, Grafton Aquaculture Centre, PMB 2, Grafton, NSW 2460, Australia; e-mail: Gavin.Butler@dpi.nsw.gov.au

Abstract

Abstract –  Underwater cameras were used to observe the breeding behaviour of the endangered eastern freshwater cod, Maccullochella ikei, over 3 years and across three areas in the Mann and Nymboida rivers, Australia. The annual breeding season for M. ikei was short and succinct, lasting only 8–10 weeks. Spawning commenced each year in the lowest altitude area during the first week of spring, and approximately 1 week later in the closest upstream area. Day-length is the primary spawning cue for M. ikei, but increasing water temperature may also be of importance. Nesting sites were located in slow-flowing pools, under cover such as large boulders and bedrock shelves, at depths of 0.9–4.0 m, and with one or two entrances only. The nesting site was vigorously cleaned by the male up to 1 week prior to spawning and was only entered by the female for spawning. Paternal care of eggs and larvae was undertaken for up to 24 days, after which larvae dispersed. Greater protection of breeding M. ikei must be a management priority to ensure long-term conservation of the species.

Introduction

The long-term conservation of a species is in part, dependent on a sound understanding of its reproductive behaviour. This is of particular importance when managing threatened species, as the enhancement of natural recruitment is often the most effective and genetically sound means of increasing the population size (Fleming & Gross 1993; Roberts et al. 1995; Garcia de Leaniz et al. 2007). One of the major issues in studying behaviour is the difficulties associated with minimising interaction between the observer and the study animal (Caine 1990; Marsh & Hanlon 2004). Within aquatic environments this can prove difficult, as many species are highly cryptic and are frequently found in habitats nonconducive to direct observation.

The production of fish in hatcheries for aquaculture grow-out and for stocking into rivers and impoundments, has produced an abundance of information relating to the reproduction of many of Australia’s larger, recreationally important freshwater fish species (Humphries et al. 1999). This includes all four freshwater cod species (Maccullochella), which have been successfully produced in hatcheries and earthen ponds under controlled conditions (Lake 1967; Rowland 1983, 1985, 1988; Ingram & Rimmer 1992; Simpson & Jackson 1996; Ingram et al. 2004). However, little is known of the breeding behaviour of Australia’s freshwater cod in the wild. Current fisheries management practices for Maccullochella spp. are based primarily on the assumption that wild and hatchery fish behave similarly.

The eastern freshwater cod, Maccullochella ikei Rowland 1985, is an endangered species endemic to the Richmond and Clarence River systems of eastern Australia. A major decline in distribution and abundance in the early part of the 20th century resulted in the wild population contracting to one sub-catchment of the Clarence River (Rowland 1993; Harris & Rowland 1996). Restocking programmes in the late 1980s and throughout the 1990s and early 2000s saw over 300,000 M. ikei fingerlings introduced into waters throughout the species’ original range (Rowland 1989; Faragher et al. 1993; Pollard & Wooden 2002). The success of these programmes remains unquantified, with survival appearing to be inconsistent and little or no evidence that stocked populations are reproducing (Faragher et al. 1993; Pollard & Wooden 2002). Other conservation actions have included a total ban on both the commercial and recreational targeting and retention of cod throughout the Clarence and Richmond River systems since 1982. However, recreational angling, including the use of artificial lures is popular throughout both systems and incidental catch and the illegal poaching of M. ikei still occurs (Lintermans et al. 2005).

Like most fish, M. ikei is iteroparous, reaching sexual maturity at 4–5 years of age (Rowland 1996), and living and reproducing beyond 15 years (Butler & Rowland 2008). Based on the assessment of gonads and a small number of natural spawnings in ponds, Rowland (1989) hypothesised that M. ikei spawn in late August and early September in the wild, when water temperatures rise from 16 to 18 °C. Rowland (1996) also suggested that M. ikei most likely deposits its eggs onto hard surfaces such as rocks or in log hollows, with some level of post-spawn care possibly offered by one or both parents. Hatching under controlled conditions typically commences after 8 days and is completed by 12 days, if water temperatures remain above 17 °C (Rowland 1985, 1996).

There has been limited research undertaken into the reproductive behaviour of wild M. ikei. Targeted sampling aimed at locating spawning sites and capturing larvae throughout the 1980s and 1990s was unsuccessful despite various attempts (I. Wooden, S. Rowland, unpublished data). The apparent poor reproductive success of stocked populations has further highlighted the need to better understand the species' breeding habits. The aims of this study were to; (1) describe the reproductive behaviour of M. ikei in the wild and (2) determine if parental care of eggs and larvae is disrupted by recreational angling. Given the threatened status of the species, nondestructive methods were employed throughout the study, with underwater cameras used as the primary sampling tool.

Materials and methods

Study site

The Mann and Nymboida River system is a sub-catchment of the Clarence River system on the north coast of New South Wales, Australia (Fig. 1). Typical of Australia’s easterly draining streams and rivers, the system is comparatively short but steep, falling from 1350 m above sea level (m.a.s.l.) at the headwaters of the Mann River, to 60 m.a.s.l. at the confluence with the Clarence River, over a distance only slightly in excess of 200 km. The system is considered relatively pristine, with clear flowing waters and abundant in-stream cover for fish (Rowland 1993, 1996). The Mann and Nymboida Rivers are typified by long slow-flowing pools interrupted by equally long sections of broken, fast-flowing waters. Rainfall and flow are normally lowest during late winter and early spring, although spring storms do produce short and sharp rises in river height.

Figure 1.

 The Clarence River system showing the Mann and Nymboida rivers and mid-points of the three sample areas.

Three areas (referred to as Areas 1, 2 and 3) were sampled within the Mann and Nymboida River system. Each area was separated by approximately 40 km (Fig. 1), with all three considered homogenous for most environmental parameters. Areas were 5–8 km in length and included a mix of pools, glides, runs, riffles and rapids. Each area also contained a variety of in-stream habitats including woody debris, large and small boulders, undercut banks, as well as open sand and cobble areas. The only notable difference between the three areas was altitude, with Areas 1, 2 and 3 at 80, 160 and 220 m.a.s.l., respectively.

Underwater cameras

Underwater cameras were used to observe spawning sites and behaviour during the breeding seasons of 2003, 2004 and 2005. Sampling commenced in August each year and continued until the end of October. Cameras were either infra-red or white-light illuminated, measured 53.2 mm long and 35.5 mm in diameter, and weighed approximately 280 g (Sunkwang Electronics Co. Ltd, Bucheon, South Korea). Areas were sampled in rotation every 3–5 days, primarily using a mobile camera operated from a canoe. This involved lowering a camera from the canoe via an extendable aluminium pole and slowly moving downstream checking all habitat types up to depths of 4 m. Images were viewed using a 15 cm, 12 v LCD monitor, with a VCR used to record activity at nesting sites. Because of the clarity of the water (>4 m visibility), much of the shallower and featureless sections of the river were inspected directly from the boat without the aid of the camera. Scuba diving was used on a number of occasions to check areas where water depths exceeded 4 m.

Fixed camera arrays were placed at two nesting sites in 2003 and 2004, and at three sites in 2005. An array consisted of three or four cameras placed at the entrance or entrances of a nesting site, with cameras also placed inside the site. Generally, infrared cameras were used at the entrance of sites, while white-light cameras were used inside due to the relatively low light intensity. The behaviour of cod appeared to be largely unaffected by the light from either camera type. Nudging and pushing of the camera was observed at some sites after initial installation, but this generally ceased within several hours. Images were recorded from mid-August until late October in each year using a four channel digital multiplex recorder.

Radio-tags

As part of a concurrent movement and habitat study (G. Butler, unpublished data), the breeding behaviour of 18 radio-tagged cod was monitored over the 2005 breeding season using underwater cameras. Radio-tags were implanted within the peritoneal cavity of 15 male and three female adult cod. Sex of cod was determined during the implantation of tags using techniques described by Rowland (1983, 1988) for Murray cod Maccullochella peelii peelii (Mitchell). Tagged cod were released in March 2005, 7 months prior to the breeding season. Cod were located weekly leading up to the breeding season, with more frequent observations made during August, September and October.

Artificial spawning sites

Artificial habitats, similar to those used in the hatchery production of M. peelii peelii in ponds (Lake 1967; Rowland 1983; Cadwallader & Gooley 1985; Ingram et al. 2004), were used to collect eggs and breeding information on wild M. ikei. Three, weighted 200 l, plastic drums (length = 900mm, diameter = 600 mm) were placed in pools within each of the three areas in 2005. Drums were positioned a minimum of 2 weeks prior to the breeding season and were set horizontal to the river bottom at depths of 2–3 m. The drums were positioned throughout pools, generally in association with instream habitat such as large boulders and woody debris. Drums were open at one end only, with the opening at a minimum of 90° to flow. Insect mesh (3 mm) covered the lower half of the internal surface of the drum to allow the removal of eggs and to hold egg masses intact during transport (S. Thurstan, personal communication). Drums were checked weekly using cameras leading up to the breeding season and every 3–4 days over August, September and October.

Protection of spawning sites

The reaction of a nest-guarding male cod to artificial lures was tested at one post-hatch nesting site in Area 2 in 2004. A hard-bodied lure with hooks removed was cast from a stationary canoe anchored approximately 10 m from the nest entrance. The lure was thrown 5–10 m past the nest and retrieved slowly ensuring it remained approximately 0.5–1 m from the river bottom. The lure was retrieved a minimum of five times within three zones, approximately 5–10, 2–5 and 0–2 m from the entrance of the nest. The reaction of the cod was viewed using four stationary cameras, and rated as either; (1) no reaction, (2) agitated but remained within nest or, (3) left the nest and followed or attacked lure.

Habitat and environmental parameters

The following meso and microhabitat parameters were recorded at each nesting site: channel unit type (pool, glide, rapid, riffle, run, cascade, waterfall), location within the channel unit, distance from shore, cover type, substrate, depth, water velocity at the entrance of nests, number of access points and light inside the nest (Table 1). The number of access points and relative light intensity were estimated visually, either directly from the surface or by camera. The remaining parameters were measured using a variety of methods as outlined in Table 1.

Table 1.   Habitat and environmental parameters recorded for nesting sites and areas, respectively, and associated collection methods.
ParametersMethod
Macrohabitat
 Channel morphologyPool, glide, rapid, riffle, run, cascade, waterfall (Armantrout 1998)
 Longitudinal location in pool Top, middle or bottom
 Distance from bankBushnell Yardage Pro Sport 450 (to nearest metre)
Microhabitat
 Cover typeOpen water, bedrock, large boulder (>500 mm), small boulder (200–500 mm), macrophytes, woody debris, undercut bank and artificial (Simonson et al. 1994)
 SubstrateBedrock, boulder (>200 mm), small boulder (200–500 mm), cobbles (60–199 mm), pebbles (15–60 mm), gravel (2–15 mm), sand (0.06–1 mm) and mud/silt (<0.059 mm) (Simonson et al. 1994)
 VelocityGeneral Oceanics Mechanical Flowmeter (range = 6 cm·s−1–100 cm·s−1)
 DepthGarmin Fishfinder 80 (to nearest 10 cm)
 Access pointsBy count
 LightDark = parent, eggs or larvae not visible without provision of additional light
Low = movement detectable but not clear
Light = parent, eggs or larvae clearly visible without provision of additional light
Environmental parameters
 Water temperatureGemini Tinytalk TK-0014 (range = −40 °C to +85 °C) recorded every 2 h
 FlowAverage daily river height based on event-based logger scanning at intervals of 15 min (source: NSW Department of Natural Resources)
 Moon phaseNew moon, first quarter, full moon and last quarter (source: Geoscience Australia)
 Barometric pressureRecorded every 3 h (source: Bureau of Meteorology Australia)
 Day-lengthBased on civil twilight to the nearest minute (source: Geoscience Australia)

Water quality, weather and lunar data were collected for each of the three areas. In 2003, water temperature data were collected within Area 1 using a data-logger, with data for Area 2 sourced from a nearby river gauging station. Water temperature was recorded within each of the three areas in 2004 and 2005 using individual data loggers (Gemini Tinytalk TK-0014, Hastings Data-Loggers, Port Macquarie, NSW, Australia). Loggers were placed midstream at depths of 2–3 m, with temperature recorded every 2 h to the nearest 0.1 °C. Data on flow, moon-phase, day-length and barometric pressure were all sourced externally (Table 1). All data were area-specific where possible, with only barometric pressure and moon-phase derived from a centralised point at the nearest available weather station at Grafton (S 29.612; E 152.956) (Fig. 1).

Data analysis

Binomial (spawn, no spawn) logistic regression (BLR) modelling (MLwiN 2.02; Rashbash et al. 2005) was used to compare the day of first spawning to flow, water temperature, barometric pressure and day-length within areas as well as across years. For the analysis, the day of first spawning was considered as the first day when eggs were observed within an area, with a one-day buffer used in the model to allow for rotation in sampling times. Flow was considered as the peak-flow (height in metres) at each site for each day. Mean, minimum and maximum daily temperatures were tested separately to determine which if any of the parameters were the best fit for the model. Barometric pressure was considered as the daily mean, and day-length as the total daylight civil twilight minutes for each day. The Wald statistic was calculated from the modelled partial coefficients and standard errors, and significance was tested using a Z-normal distribution. Wald statistics greater than or equal to 1.96 were considered to be two tail significant (< 0.05).

Statistical comparisons were also made of the temperature and day-length from the 7 days preceding first spawning, with both year and area treated as separate samples. One-way analysis of variance and Tukey’s honestly significant difference (HSD) test were used to test for differences in day-lengths between Area 1 and 2. As the water temperature data were not normally distributed, a Kruskall–Wallis test was used to determine if there were overall differences across sites and years. Post hoc pairwise comparisons were undertaken using Mann–Whitney U-tests, with the results compared against a Bonferonni adjusted P-value to determine significant differences. Breeding behaviour, larval behaviour and nesting site choice were based on data collected from spawning sites observed across all years and areas. Where applicable, data from artificial breeding habitats were also included in the analysis. For the purposes of the study, the breeding season was defined as the period between the first observed selection of nesting sites by males, through until the last day when larvae where observed within a nest.

Results

Spawning events

Thirty three spawning events were observed over the three breeding seasons (Fig. 2). This included 14 events within Area 1, including seven of the radio-tagged male cod monitored over the 2005 breeding season. Within Area 2, 19 events were observed over the 3 years, including three within two of the artificial habitats in 2005. No spawning events were observed in the artificial habitats placed in Area 1. No nesting site or spawning activity was observed within Area 3 in any year.

Figure 2.

 Spawning events of Maccullochella ikei, (a) barometric pressure, (b) moon-phase and flow and (c) temperature over the period 2003–2005 for Area 1 (bars and lines = inline image), Area 2 (bars and lines = inline image) and Area 3 (lines only = inline image). Length of breeding period (inline image) in each year is shown in (c).

The annual breeding season for M. ikei in Area 1 and 2 extended across a relatively short period of only 8–10 weeks (Fig. 2c). Initial spawning took place in the first week of spring (September) within Area 1 in all years, and 7–10 days later at Area 2 in all years (Fig. 2). A second series of spawning events ensued approximately 3 weeks later within both areas in years 1 and 3 but not in year 2 (Fig. 2). These events tended to be fewer in number and were related primarily to the reuse of sites. The multiple use of spawning sites was observed both within years (= 3) and between years (= 5). It is not known if the same males and females involved in the first round of spawning participated in the second round, however, none of the radio-tagged cod monitored over the 2005 breeding season spawned more than once.

Spawning cues

There was a significant interaction between the first day of spawning and when the variables ‘day-length’ and ‘site’ were combined (BLR: = 0.04). However, day-length and site when considered alone, and water temperature, barometric pressure, flow and moon-phase were not significantly related to the day of first spawning (BLR: > 0.05). Spawning occurred across a wide range of barometric pressures and on rising and falling trends (Fig. 2a). Flow tended to be relatively stable across years and sites, with the majority of spawning events occurring on declining flows (Fig. 2b). Similarly, first spawning occurred across a variety of moon-phase types, from new to full moon.

There were significant differences in the day-length for the 7 days preceding the first spawning events observed within Area 1 and Area 2 (anova, > 0.05). Tukey’s HSD test revealed that the differences were primarily site related, with no significant difference between any years within Area 2, and only 2004 grouping separately within Area 1. Average (±SE) day-length on the day of first spawning for all 3 years combined for Area 1 was 12.38 ± 0.04 h and for Area 2, 12.61 ± 0.03 h.

Overall, water temperature among years and sites was significantly different in the 7 days prior to initial spawning (Kruskall–Wallis, P > 0.05). Pairwise comparisons revealed that those samples not significantly different (P < 0.05) (= 6) tended to be randomly distributed across both years and sites. The average temperature on the day of first spawning for all years combined within Area 1 was 18.2 ± 0.6 °C, and in Area 2 was 17.8 ± 0.2 °C. The average minimum and maximum temperatures for the same days were 17.5 ± 0.7 °C and 19.1 ± 0.5 °C for Area 1, and 16.4 ± 0.4 °C and 19.3 ± 0.5 °C for Area 2.

Water temperature rose on average 0.24 ± 0.12 °C·day−1 in Area 1, and 0.25 ± 0.13 °C·day−1 in Area 2 in the 7 days prior to spawning for all years combined. Similarly, temperatures increased consistently across the entire breeding period (20th August to the 31st October in 2004 and 2005) rising on average by 0.14 ± 0.06 °C·day−1 in Area 1, and by 0.12 ± 0.08 °C·day−1 in Area 2. The average daily temperature difference between Areas 1 and 2 for the same periods was 0.77 ± 0.16 °C, and between Areas 2 and 3, 1.34 ± 0.54 °C, with the higher temperatures recorded in the downstream areas in both cases.

Nesting sites

Habitat parameters were recorded for 23 nesting sites, including variables from two artificial habitats used in the 2005 breeding season. A number of nests were also established but not used when the occupant failed to secure a mate. These sites were not used in any analysis. Nesting sites covered an area approximately 1 m in diameter, with the size and shape of the nest largely dictated by the geomorphic make-up of the immediate area. Nesting sites were predominately positioned in slow-flowing pools, with only one site located outside of a pool in a pool/run backwater within Area 1. No sites were located in glides, riffles or rapids. There was no apparent preference for longitudinal (top = 9; middle = 7; bottom = 7) or latitudinal positioning (distance from shore = 0–30 m) of sites, with nests found throughout all areas of pools.

All nesting sites were located under cover, with no sites found in open water. Cover included large boulders (= 10), bedrock shelves (= 7), undercut root-ball masses of the aquatic grass Potamophila parviflora (N = 4) and artificial habitats (N = 2). No nesting sites were located under small boulders or undercut banks, or within macrophyte beds or woody debris hollows, despite the relative availability of these habitat types in all areas. Eggs in all cases were deposited on clean hard substrates, which included exposed bedrock (= 11), cobbles (= 6) and large boulders (= 4).

Average depth of nesting sites was 1.7 ± 0.2 m with a range of 0.9–4.0 m. Flow directly over deposited eggs was below measurable levels, with flow at the entrance to sites also negligible (<0.06 m·s−1). Generally, the light inside the site was low to dark, with the eggs visible at most sites only with the aid of artificial lighting. The majority of eggs were deposited towards the rear of the nest, with only small numbers of eggs in most cases laid at the outer exterior of sites. All nesting sites had only one or two clearly definable entrances, with the majority of openings facing either 90° (= 16) or 180° (= 4) to flow.

Breeding behaviour

The breeding behaviour of M. ikei was similar at all sites. Initial selection of nesting sites was undertaken exclusively by males and took place 1–3 days prior to spawning. Sites were occupied by a solitary individual, with nests generally well spaced (>5 m). Prespawning competition for sites between males was observed on a number of occasions, with the conflict often leaving visible wounds on the head and body of combatants. After establishing apparent ownership of the site, the winning male would undertake a period of vigorous cleaning to remove loose sediment and sand from the benthos. This involved vigorous pulsating of the body to loosen particles, with the pectoral fins then used to sweep the suspended material away. This behaviour was undertaken continuously in the days leading up to spawning, but increased in intensity as spawning approached.

No females were observed at or near prospective nesting sites until the day of spawning. However, mate selection appeared to be driven primarily by the female, with the male remaining at or near the nest site leading up to spawning. It is uncertain if the male in someway attracted the female to the site, but no obvious courting display was witnessed. In all cases where males and females were observed together they appeared to be approximately the same size (approximate range 400–600 mm), with females distinguishable by their distended abdomens.

There was a consistent prespawning ritual at all sites. The female would enter with the male in close attendance and undertake what appeared to be an inspection of the site. This involved the female rubbing its underbody vigorously on the substrate. The male would then nudge the female away from the bottom and the pair would swim nose to tail while touching and at times semi-entwining. After 10–20 s of this activity, the female would rapidly leave the site. The male would initially leave with the female, but would quickly return within 5–15 s and resume vigorous cleaning of the site. This sequence of events was repeated every 15–20 min with the female returning up to 4–5 times before spawning. It is unknown if the female was inspecting other potential nesting sites during these absences.

The release and fertilisation of eggs was not observed, but immediate postspawn activities were monitored at a number of sites (= 6) over the 3 years. Eggs were laid in a single laminate and were dispersed evenly across the nest site, covering an area in excess of 0.5 m2 in most cases. Females left the nesting site following spawning and did not return. The care of eggs and larvae lasted for periods up to 24 days and was undertaken exclusively by the male. This was validated by observations of radio-tagged cod, with seven tagged males nesting and undertaking parental care. Paternal care initially involved the protection and cleaning of eggs, using similar techniques as employed during nest cleaning. However, greater care was taken to ensure eggs were not disturbed, with fanning taking place 1–2 cm above the substrate. Fanning of eggs was undertaken continuously until hatching commenced.

Hatching commenced 7–8 days after spawning and was completed 2–3 days later. The commencement of hatching was difficult to determine precisely as larvae were small and relatively translucent. Therefore, the change in colour of the egg mass (i.e., from white to brown) was used as an indication of the onset of hatching. Larvae remained on the substrate during the first 5–6 days posthatch. By day-10 posthatch, larvae were observed swimming up to 1 m from the substrate, and moving and returning distances up to 2 m from the nest. The male remained in close attendance throughout this period and continued to defend the site vigorously. At 12–14 days posthatch all larvae left the site en masse, with the male in most cases also abandoning the site.

Protection of spawning sites

Male cod were not observed feeding whilst undertaking care of eggs and larvae. Ventures from the nest appeared to be restricted to the patrol and defence of an area of less than 5 m in diameter. Paternal defence involved initially yawning or jawing toward any intruder, whilst maintaining position over or near the nest. When aggressive displays proved ineffective, the male attempted to nudge and push the intruder away from the site. If this proved unsuccessful, biting and chasing was initiated, with the male rapidly returning to the nest after ensuring the predator had been displaced.

Potential predators were observed near or within nesting sites at various times during the study. These included freshwater turtles, small fish species including crimson-spotted rainbowfish Melanotaenia duboulayi (Castelnau) and Hypseleotris spp. Gill, as well as larger fish such as long-finned eels Anguilla reinhardtii (Steindachner) (<1000 mm) and Australian bass Macquaria novemaculeata (Steindachner) (<400 mm). On the three occasions where nests were abandoned prior to hatch, the majority of eggs were consumed by predators within a number of hours. No abandonment of nests was observed if hatching was successful (>10 days).

The reaction to artificial lures was similar to that observed towards predators. For the initial passes of the lure at 5–10 m from the nest entrance, the cod became agitated but remained inside the nest. For passes within the 2–5 m zone, the cod left the nest three of the five times, following the lure on at least one occasion to the canoe. Five passes were then made from either side of the nest site in the 0–2 m area directly in front of the nest. The cod became extremely agitated during each pass and was observed leaving the nest five of the 10 times. Again, the cod followed the lure to the canoe on at least one occasion, with strikes on the lure also felt on a number of occasions. In all cases the cod returned to the site after 1–2 min.

Breeding model

A summary of the proposed breeding model for M. ikei is presented in Fig. 3. The complex breeding behaviour involves site selection and preparation by males, mate selection, spawning, paternal protection of eggs and larvae, the rapid dispersal of free-swimming larvae and the abandonment of the nest. The breeding behaviour of cod is divided into three distinct phases of (1) prespawning activities, (2) spawning and care of eggs and (3) the care of larvae. In total, the breeding cycle lasts between 21–27 days (Fig. 3). Provision is also made in the model to allow for possible polygamous behaviour by males and females.

Figure 3.

 Suggested breeding model for Maccullochella ikei.

Discussion

The short and well-defined breeding season described for M. ikei in this study is similar to other Australian freshwater cod taxa. Rowland (1983, 1998) examined M. peelii peelii in both ponds and in the wild and reported the species has a relatively short breeding season of 4–5 weeks. Other studies have suggested a 2–3 month breeding period for M. peelii peelii, based primarily on the capture of dispersing larvae in the southern tributaries of the Murray–Darling River system (Humphries et al. 2002; Gilligan & Schiller 2003; Humphries 2005; Koehn & Harrington 2006). Maccullochella macquariensis (Cuvier) is hypothesised to have an even shorter breeding season of less than a month in some areas of its distribution (Koehn & Harrington 2006). It therefore becomes critical that optimal environmental conditions are experienced over this period to ensure the survival of larvae (Bye 1984; Humphries et al. 1999). Within the Mann and Nymboida rivers, the breeding season of M. ikei coincides with the change in season from winter to spring, typified by increasing day-length, rising water temperatures and relatively low-flow conditions.

Maccullochella ikei bred annually despite variations in environmental conditions such as flow, temperature, barometric pressure and moon-phase. This is consistent with the other taxa of Maccullochella, which also breed annually throughout their respective ranges (Rowland 1983, 1985, 1998; Simpson & Jackson 1996; Simpson & Mapleston 2002; Humphries 2005; Koehn & Harrington 2006). Lake (1967) suggested a rise in water level is required to induce spawning in a number of Australian fish species, including Maccullochella spp. However, while this appears not to be the case with M. ikei, floodplain inundation may still ultimately dictate successful recruitment, but through increased habitat and food availability rather than as a direct spawning stimulus (Lake 1967; Rowland 1983, 1992, 1998; Harris & Gehrke 1991).

The significant relationship between the day of first spawning and day-length indicates that day-length may be the primary spawning cue for M. ikei and not temperature as previously thought (Harris & Rowland 1996; Rowland 1996). However, temperature may still be a contributing factor but over an extended period rather than as a specific spawning cue. Gametogenesis is a prolonged process in most fishes, triggered by external factors such as seasonal changes in water temperatures (Scott 1979; Bye 1984). Rowland (1998) reported that gonadal development in the closely related M. peelii peelii followed a clear, extended sequence of 5–6 months from early winter to spring. The consistent spawning lag between Area 1 and Area 2, and the temperature gradient found between areas resulting from differences in altitude, suggests a possible connection between temperature and gonadal development for M. ikei. This could therefore extend the breeding season in higher altitude areas of the Mann and Nymboida rivers beyond that suggested for the lower reaches of the system. Similarly, breeding of M. ikei in other sub-catchments of the Clarence and other systems such as the Richmond River may also occur at different times.

The breeding behaviour of M. ikei is consistent with that of a nesting, guarding spawner (Balon 1975, 1984). Unlike many nest guarders (e.g., various members of the family Cichlidae and the black bass Micropterus spp. Lacepède), M. ikei is a large and formidable species that grows to over 1 m in length and as observed in this study, is able to defend its nesting site against equally large intruders. However, typical of this strategy, M. ikei produces relatively low numbers of eggs at approximately 5000 kg−1 (S. Rowland, unpublished data) and directs the majority of reproductive effort towards postspawn care of eggs and larvae. As with most nest-guarding species, reproductive success is largely dependent on the ability of the respective parent to undertake and complete all aspects of the breeding cycle. With iteroparous species this often relates to the number of times an individual has reproduced, particularly where complex reproductive strategies are employed (Lavery & Lavery 1995; Daunt et al. 1999; Woodard & Murphy 1999). The failure of some males in this study to secure sites and mates, and the early abandonment of three nests, may have therefore been a reflection of the relative inexperience of individual parents.

With most nest guarders, competition between males for potential spawning sites can often result in larger, more aggressive or bourgeois males securing premium sites. This can facilitate alternative tactics such as sneaker or satellite behaviour by lesser or parasitic males in an attempt to ensure reproduction (Gross 1982; Taborsky 1994Taborsky 2001). A similar but less obvious tactic potentially employed by M. ikei in this study could be termed vicinity nest-building, where a parasitic male establishes a nest in a lower quality site near a bourgeois male in the hope of securing an attracted, but uncommitted female. The behaviour of males within this study suggests that this may be a strategy employed by M. ikei, with a number of smaller males establishing nest sites and successfully spawning in close proximity to larger, though ultimately unsuccessful males. The ambiguous prespawning behaviour of female M. ikei further enhances the potential success of such a strategy.

The specific nature of the nesting sites selected by M. ikei suggests that the availability of sites is fundamental for breeding success, and must therefore be a primary consideration when assessing areas for conservation actions such as restocking. In practice this may prove difficult, as many parts of both the Richmond and Clarence rivers have been subjected to significant siltation since European settlement (Luckie 1998; Dawson 2002). This has not only had the effect of covering potential nesting sites, but increased levels of free sediment may also make the establishment and maintenance of sites more difficult. The successful use of two of the artificial spawning habitats in this study offers a potential strategy for population remediation.

While the lack of suitable habitat may in part explain the apparent poor success of reproduction in restocked populations, the complex nature of M. ikei’s breeding cycle may also be a contributing factor. There is a growing amount of literature extolling the importance of social learning in the behaviour of fish. Previously, many of the life-skills considered as instinctive or inherited traits such as foraging, predator avoidance and reproductive behaviour are now considered to have a significant learned aspect (Jonsson 1997; Brown & Laland 2001, 2003). Given that many of the areas restocked with hatchery-reared fingerlings were devoid of cod (Pollard & Wooden 2002), it may be possible that the complex parenting skills essential for M. ikei to successfully reproduce are dependent in some part on learned behaviour. Extensive research involving mainly salmonids has revealed reduced reproductive fitness in the majority of cultured fish released into the wild (Fleming & Gross 1993; Fleming et al. 1996; Jonsson 1997). In many populations, cultured fish have been found to contribute very little to the gene pool, despite in some cases 30 years of releases (Heggenes et al. 2006). If similar issues are occurring within restocked populations of M. ikei, management strategies such as the planting of experienced parents to act as surrogate trainers, or the translocation of wild adults as an alternative to stocking hatchery-reared fish, may be required to ensure the success of future remediation programmes.

The multiple use of spawning sites within seasons may be a reflection of a low availability of suitable sites. However, it may also be an evolutionary strategy to ensure recruitment success and ultimately the perpetuity of an individual’s genes. A recent study of the breeding behaviour of M. peelii peelii found that under controlled conditions in ponds, females may spawn multiple times in one season (Rourke 2007). Similarly, Maccullochella peelii mariensis Rowland is also thought to spawn multiple times in the one season in ponds (Simpson & Jackson 1996; Pusey et al. 2004). Spawning multiple times with an extended interval between events, in effect increases the likelihood of offspring being exposed to optimal environmental conditions for survival and growth (Moyle & Cech 2000; Knight et al. 2007). This becomes particularly important given the extreme seasonal variability often experienced within many Australian river systems, including the Mann and Nymboida rivers. Similarly, a polygamous reproductive strategy also increases the likelihood of mating with a physically or genetically superior partner (Jennions & Petrie 2000; Rubenstein 2007). Although no radio-tagged cod was observed spawning more than once in this study, multiple spawning events undertaken by female cod may have not been detected given their relatively brief involvement in the breeding cycle. Additionally, because of their high level of parental investment, male cod in many cases may be incapable of breeding effectively more than once in a season.

It has been hypothesised that M. peelii peelii and M. peelii mariensis larvae both drift as part of their early life-history strategy (Humphries et al. 2002; Gilligan & Schiller 2003; Koehn & Harrington 2005). Given the strong swimming ability of larvae observed within nests in this study, it is unlikely that M. ikei larvae undertake a true passive drift phase to disperse. The likely success of such a strategy is further reduced by the low-flow conditions experienced in pools where nesting sites were located. It is therefore suggested that any movements made by M. ikei larvae after leaving the nesting site should be considered an active dispersal, rather than the actions of a true drift-and-settle strategist.

This study has revealed the potential vulnerability of breeding male M. ikei to angling. Similar behaviour has been reported in a number of other nest-guarding species, including many of the black bass, Micropterus spp. Generally, studies of angling effects on black bass have shown that following catch-and-release, nesting males have a diminished capacity to effectively parent, with many abandoning nests completely (Cooke et al. 2002; Suski et al. 2003; Suski & Philipp 2004). This has resulted in long-term declines in many bass populations, attributed to reduced overall recruitment (Kubacki et al. 2002; Suski et al. 2002). Similarly, the breeding success of other nest-guarding species such as the bluegill Lepomis macrochirus Rafinesque (Beard et al. 1997) and the European pikeperch Sander lucioperca (Linnaeus) (Lappalainen et al. 2003), are also thought to be detrimentally affected by angling. Based on the findings of this study, complete angling closures are now implemented for August, September and October each year to protect remnant populations of M. ikei in the Mann and Nymboida River system.

The breeding model suggested for M. ikei in this study is most likely applicable to Australia’s other freshwater cod species. Various authors have described similar behaviour by M. peelii peelii under controlled conditions in ponds (Rowland 1983; Cadwallader & Gooley 1985; Ingram et al. 2004), while male M. peelii mariensis have been observed guarding their eggs and larvae for periods up to 15–18 days in ponds (Simpson & Jackson 1996; Pusey et al. 2004). Similarly, M. macquariensis is also thought to form pairs and spawn annually, most probably within the protection of hollow logs or among rocks (Cadwallader 1979; Douglas et al. 1994). Given the high likelihood that reproductive behaviour is at least similar in all four Maccullochella taxa, the protection of breeding individuals should be considered a management priority for all of Australia’s freshwater cods, with appropriate temporal and spatial closures implemented for those taxa currently not protected.

This study has provided the first detailed description of reproductive behaviour in Maccullochella spp. in the wild. Maccullochella ikei has a relatively protracted and complex breeding cycle involving site selection, courting, spawning, vigorous and continuous fanning of eggs and the aggressive defence of eggs and larvae from intruders. The information presented will contribute not only in the long-term conservation of remnant populations, but also in the re-establishment of the species throughout its original range. While little research has been undertaken into the significance of social learning for Australia's freshwater fishes, future stock enhancement programmes should consider its potential importance, particularly when stocking areas where remnant adults no longer occur. Finally, while uncertainty exists over the reproductive strategies of other Maccullochella spp., the model and recommendations proposed in this study provides a basis for the conservation and management of all Australian freshwater cod taxa until such times as further research is undertaken.

Acknowledgements

This study is from PhD research undertaken by Gavin Butler and was supported by Southern Cross University, the NSW Department of Primary Industries, the Australian Research Council, the NSW Recreational Freshwater Fishing Trust and Rous Water. All sampling was undertaken in accordance with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes (Permit No: 04/12). We would particularly like to thank Ralph Jahrling, Robert Predo, Mark Nixon and James Knight for their help with field work; Gary and Eva Ellis, Phillip and Melvina Dick, Alistair Maple and Bernhard McAllister for allowing us access to their respective sections of the river and Steve Thurstan and Peter Boyd for their guidance on the aquaculture of cod. Thanks also to Dr James Knight, Dr David Crook and the anonymous referees for their comments and help in the preparation of the final draft.

Ancillary