MJ Milicich's research while affiliated with Griffith University and other places

In this chapter, a case study from marine ecology is presented in which the application of techniques from nonlinear time series analysis is shown to provide insight into the interplay between stochastic physical forcing and nonlinear biological response in a natural system. Specifically, the replenishment of a population of reef fishes is analyzed in detail, and important nonlinearities are demonstrated in the processes underlying variability in the supply of larval propagules to the reef. This information is used to guide the construction of a series of models which attempt to forecast larval supply from readily measured physical variables. The most successful models are those that account for the nonlinearities in the response of larvae to their physical environment. Such models provide better forecasts than can be achieved with conventional linear techniques and identify processes hidden to linear analysis. The importance of understanding the interplay between noise and nonlinearity in ecological systems is discussed.

The question of how far the larvae of marine organisms disperse is fundamental to an understanding of their population dynamics1, 2, 3, the management of exploited species4, 5 and the conservation of marine biodiversity6, 7. It is generally assumed that larvae disperse away from their natal population so that local populations operate as 'open' systems, driven by recruitment of larvae from other sub-populations8. However, this assumption has never been critically tested. Here we show for the first time that juveniles from a coral reef fish population can return to their natal reef. We marked otoliths (ear bones) of over 10 million developing embryos of the damselfish, Pomacentrus amboinensis, at Lizard Island (Great Barrier Reef). Subsequently, from an examination of 5,000 juveniles settling at the same location, we found 15 marked individuals. On the basis of an estimate of the proportion of embryos marked (0.5–2%), as many as 15–60% of juveniles may be returning to their natal population (self-recruitment). We challenge the assumption that long-distance dispersal is the norm for reef fish populations.

The lack of a clear relationship between spawning output and recruitment success continues to confound attempts to understand and manage temporally variable fish populations. This relationship for a common reef fish is shown to be obscured by nonlinear processes in operation during the larval phase. Nonlinear responses of larval fish to their noisy physical environment may offer a general explanation for the erratic, often episodic, replenishment of open marine populations.

Larval supply is crucial for the persistence of coral reef fish populations, yet little direct information exists about any aspect of this life history phase. Using light traps, this study directly measured larval supply of 27 coral reef fish taxa to Lizard Island (northern Great Barrier Reef) over 2 successive recruitment seasons. Descriptions of the magnitude and synchrony of replenishment to 3 different habitats (front-reef, back-reef and lagoon) within a single reef are the first of their kind and reveal several important facets of the larval supply process. Larval supply during the first season was much stronger into the front-reef side than into the back-reef habitat. Synchrony of daily replenishment between these 2 habitats was characteristically weak during this year. This result was not related in any simple way to low levels of absolute abundance in the back-reef habitat. In the following year, this dominance of the front-reef habitat was diminished. In many cases, this decline could be explained by a rise in the strength of the replenishment signal in the back-reef habitat. Correlated with this, there was a general increase in the strength of synchrony in larval supply between the 2 habitats. Again this trend could not be simply explained by any increase in absolute abundance in the back-reef habitat. It appears that there was a change in the nature of the larval supply process between these habitats over the 2 years. Although temporally variable, replenishment into either habitat showed consistent patterns across many taxa indicating that larval supply to a single reef is systematic rather than 'stochastic'. We suggest that investigation of the relationship between larvae and the local physical conditions surrounding each reef may reveal the reasons for much of this complexity in replenishment. Larval supply into the lagoon was highly variable among taxa. The greater association of the lagoon with the front-reef habitat was revealed by the cross-correlation coefficients of many taxa. In some cases this could be explained by a progressive depletion of larvae as they crossed the front-reef crest and entered the lagoon. However, larval abundance of 4 taxa in the lagoon was so low (seasonal sum < 10 for Pomacentrus coelestus, Lethrinidae, P. lepidogenys and Neopomacentrus spp.) that active behavioural choice of settling larvae seems to be involved.

Oceanographic processes have been shown to play a pivotal role in the control of recruitment variability in some commercial, northern temperate fish populations. This study investigates the link between inter-annual changes in larval supply and wind stress for unexploited reef fish populations at Lizard Island, northern Great Barrier Reef. For the first time, this relationship is analysed separately for front-reef and back-reef habitats and results for 3 major reef fish familes (Pomacentridae, Apogonidae, Blenniidae) reveal that the link between wind stress and larval supply appears to be habitat-dependent. In the back-reef habitat, there appears to be a positive correlation between interannual changes in the frequency of onshore winds and the level of larval supply. This phenomenon is associated with large-scale changes in weather systems, suggesting that the relationship may be predictable at a region-wide scale. Results for the front-reef habitat suggest that the effect of wind stress on inter-annual levels of larval supply is less important. Survival and advection of earlier larval stages may be more influential in determining fluctuations in the magnitude of replenishment to this habitat. Although not helpful in predicting inter-annual changes in larval supply, a local retention mechanism may keep larval supply into this habitat elevated above that of the back-reef side. Comparison of these results with a non-reef mixed-age group of schooling pelagic fishes suggests that these patterns are a unique feature of the replenishment process.

End-of-season counts offish recruits (young-of-the-year) on one nearshore, three mid-shelf and one outer shelf reef on the central Great Barrier Reef (GBR) over 2 years were compared to estimates of settlement based on 4 six-weekly censuses of the same sites in each year. The end-of-season recruitment surveys yielded relative patterns of recruitment between years and among reefs that were remarkably close to those determined from the six-weekly counts. For 23 out of 24 species examined, there was a significant correlation between the net total gain of recruits, determined from the six-weekly counts, and the number of recruits censused at the end of the season. We conclude that end-of-year recruitment surveys are a robust technique for determining relative patterns of early post-settlement distributions among reefs and between years in the central GBR. Most of the species censused, primarily pomacentrids and labrids, have relatively low post-settlement mortality rates. This, and annual recruitment patterns on the GBR which tend to be the result of a few very discrete pulses of settlement, often coherent over large distances, are likely to be the key to the success of the technique.

Spatial and temporal abundance patterns of pelagic-stage goatfish (Family Mullidae) across the northern Great Barrier Reef were examined (1986-1991). From these data inferences were made on the behaviour and schooling dynamics of this largely unstudied life-stage. Sampling was conducted using both a plankton-mesh purse-seine deployed around small aggregation rafts (I x 1 m), and automated light traps. A comparison of the two techniques showed no difference in the size composition of fishes caught. Neither the total number of mullids caught at a sampling station nor their minimum size was related to the current speed as measured by drogues (0 and 5 m depth). A comparison of the species composition of raft and light trap samples with that of fishes settling onto the reef below the rafts showed that, of the 11 species known to occur in these waters, only five were caught by the two devices. Size frequencies of catches around rafts suggested that at times multiple schools were present within a 90 m sampling area (i.e. one site). Schools were composed of a number of mullid species, each having a broad size range [e.g. Upeneus moluccensis (Bleeker), 20-56 mm Standard length (SL)]. The largest individuals were competent to settle. Samples from light traps deployed off the reef around Lizard Island for 280 days over the summers (1986/87 to 1988/89) found that late pelagic-stage mullids occurred close to the reef in irregular pulses. Pelagic mullids collected in purse-seine samples were consistently common in inter-reefal waters. A cross-shelf sampling programme involving six stations stretching from the inshore Turtle Group, past mid-shelf Lizard Island to 3 km seaward of an Outer Barrier reef (Carter Reef) was conducted. Catch composition and abundance was found to vary among locations over 3-monthly samples. Upeneus tragula (Richardson) and U. moluccensis dominated the catch at all but the Outer Barrier station, where the genus Parupeneus was most abundant. Findings stress the complex interaction between behaviour, ontogeny and abundance patterns, and the potential for late pelagic-stage fishes to influence their recruitment patterns.

We examined the relative influence of planktonic processes and larval production on the larval supply and recruitment of the damselfish Pomacentrus amboinensis during 2 summers at Lizard Island, northern Great Barrier Reef. Larval production was quantified by monitoring the nest sites of males in windward, lagoon and leeward habitats. Larval supply was estimated using light traps deployed in nearshore waters adjacent to the 3 habitats. Recruitment patterns were back-calculated from the otoliths of newly settled fish collected from small artificial patch reefs constructed in each of the 3 habitats. Temporal patterns of spawning, larval supply and recruitment were correlated when the spawning pattern was lagged by a period equivalent to the incubation time of eggs and the average planktonic life. This coupling occurred despite a change in the pattern of spawning between summers from occasional, large episodes to frequent, smaller pulses spread throughout the summer. Once regular cycles were removed from the data sets by ARIMA (integrated auto-regressive moving-average) modelling, correlations between patterns declined or became non-significant. This suggests that reproduction has an important influence on the timing of recruitment, but that the magnitude of these events is largely determined within the plankton.

This study investigated the hypothesis that recruitment variations of coral reef fish directly reflect variations in larval supply. At Lizard Island, northern Great Barrier Reef, larval supply and recruitment of 3 species of damselfish (Pomacentrus amboinensis, P. nagasakiensis and Dischistodus perspicillatus) were estimated from light trap and patch reef collections respectively, over 2 consecutive recruitment seasons, in each of 3 habitats. Comparison of these patterns pooled over species and seasons revealed a good overall correlation between larval supply and early recruitment levels, suggesting that pre-settlement distributions may be the major determinant of early recruitment patterns despite mediating influences from factors such as habitat selection and post-settlement mortality. This result also supports the use of broad recruitment surveys to estimate year-class strength and to hindcast patterns in larval supply. When analysed at the individual species/season level, the abundance of recruits on patch reefs and the abundance of larvae in light trap collections were also significantly correlated, with the exception of one species (P. amboinensis) during one season (1989/90). When the level of comparison was further narrowed to analyse correlations between pre- and post-settlement fish of individual species in each season, within each habitat, only 7 significant relationships were detected out of a possible 18 habitat/species combinations. The power of these tests was relatively high (mean = 0.846; SE = 0.03). These relationships provide some evidence for habitat selection suggesting that P. amboinensis preferred the leeward habitat, while D. perspicillatus preferred the windward habitat.