Random and systematic sampling error when hooking fish to monitor skin fluke (Benedenia seriolae) and gill fluke (Zeuxapta seriolae) burden in Australian farmed yellowtail kingfish (Seriola lalandi)

Abstract

The Australian farmed yellowtail kingfish (Seriola lalandi, YTK) industry monitor skin fluke (Benedenia seriolae) and gill fluke (Zeuxapta seriolae) burden by pooling the fluke count of 10 hooked YTK. The random and systematic error of this sampling strategy was evaluated to assess potential impact on treatment decisions.

Fluke abundance (fluke count per fish) in a study cage (estimated 30,502 fish) was assessed five times using the current sampling protocol and its repeatability was estimated the repeatability coefficient (CR) and the coefficient of variation (CV). Individual body weight, fork length, fluke abundance, prevalence, intensity (fluke count per infested fish) and density (fluke count per Kg of fish) were compared between 100 hooked and 100 seined YTK (assumed representative of the entire population) to estimate potential selection bias.

Depending on the fluke species and age category, CR (expected difference in parasite count between 2 sampling iterations) ranged from 0.78 to 114 flukes per fish. Capturing YTK by hooking increased the selection of fish of a weight and length in the lowest 5th percentile of the cage (RR = 5.75, 95% CI: 2.06–16.03, P-value = 0.0001). These lower end YTK had on average an extra 31 juveniles and 6 adults Z. seriolae per Kg of fish and an extra 3 juvenile and 0.4 adult B. seriolae per Kg of fish, compared to the rest of the cage population (P-value < 0.05).

Hooking YTK on the edge of the study cage biases sampling towards the smallest and most heavily infested fish in the population, resulting in poor repeatability (more variability amongst sampled fish) and an overestimation of parasite burden in the population. In this particular commercial situation these finding supported that health management program, where the finding of an underestimation of parasite burden could provide a production impact on the study population. In instances where fish populations and parasite burdens are more homogenous, sampling error may be less severe. Sampling error when capturing fish from sea cage is difficult to predict. The amplitude and direction of this error should be investigated for a given cultured fish species across a range of parasite burden and fish profile scenarios.

Introduction

Ectoparasitic infestations represents substantial fish health and welfare challenges for sea cage aquaculture systems worldwide (Whittington et al., 2000; Ernst et al., 2002). Industry implications of such infestations include; direct stock loss, depressed fish growth, poor fish health and welfare, reduced value of market product, and costs associated with monitoring and treatment programmes (Sharp et al., 2003; Hutson et al., 2007; Ernst et al., 2002). In Port Lincoln, Australia, the yellowtail kingfish (Seriola lalandi, YTK) industry has suffered substantial production setbacks in recent years due to recurrent infestation of two monogenean ectoparasites; Benedenia seriolae (skin fluke, SF, Sub-class Monopisthocotylea) and gill fluke, Zeuxapta seriolae (gill fluke, GF, Sub-class Polyopisthocotylea) (Hutson, 2007). These two parasites have a direct lifecycle, with adult stages colonising and feeding on the fish and mature adult females releasing egg bundles that attach on cage infrastructure allowing for rapid re-infestation and amplification in sea cage systems where fish hosts are stocked in high density (Tubbs et al., 2005). Both flukes are specific to the Seriola genus and do not represent any concern for human consumption (Hayward, 2005).

SF actively feed on epithelial cells following attachment to skin surfaces (Whittington, 2005) which cause skin irritation and depression in feed intake of infested host which respond by rubbing against the cage net and any floating devices. Subsequently skin lesions can occur with erosions and progressing to ulceration and secondary bacterial infections, in severe cases (Whittington, 2005; Ernst et al., 2002). GF are sanguineous, attaching exclusively to the gill lamellae resulting with time in anaemia, jaundice and emaciation of the fish host (Grau et al., 2003; Chih-Hui et al., 2012). Destruction of gill epithelium and vascular damage at the attachment site induces focal gill inflammation and lamellar fusion (Montero, 2004). The duration of the flukes’ life cycle is temperature dependent and uncontrolled outbreaks commonly occur during summer months (Ernst et al., 2002). The increase in sea water temperature shortens the duration of fluke maturation, incubation period and increases egg hatchability (Tubbs et al., 2005).

The control of YTK flukes involves treating the sea cage population with a hydrogen peroxide bathe (Mansell et al., 2005). This process is costly, labour intensive, logistically complex and has narrow safety margins (Mansell et al., 2005; Williams et al., 2007). Hydrogen peroxide does not destroy fertilised fluke eggs (Sharp et al., 2004) and within few days to weeks (according to sea temperatures) a new generation of flukes hatches and reinfests the cage (Tubbs et al., 2005). Therefore, bathing strategy uses a second consecutive bath to timely kill the newly hatched juvenile flukes before they reach sexual maturity and release new eggs. The time lapse between bathing depends on water temperature and is dictated by the burden and age distribution of flukes in the cage. The monitoring of flukes’ burden in the cages is instrumental to optimise bathing schedule (Whittington, 2005). Poorly timed treatments may waste resources (too early) or impact productivity, fish health and welfare (too late). The accuracy of the fluke monitoring is paramount to properly time treatment.

Following commercial reality, monitoring of parasite burden in aquaculture should be fit-for-purpose, i.e. providing accurate and meaningful management information for the least resources (time, labour, and money) possible (Revie et al., 2007). In Australia, the industry fluke monitoring protocol involves capturing up to ten fish using hook-and-line from the edge of the sea cage. This method of fish capture is routinely used in other aquaculture industries to conveniently sample fish. However, hook-and-line is believed to bias the sample, especially when the fish population is not homogenous (Oidtmann et al., 2013). Fish cage populations are rarely homogenous in size and growth because of the hierarchical nature of fish interaction (dominant fish grow faster and bigger). Parasite burden is also expected to not be uniform especially at the early stage of the colonisation when not all the fish are infested (Heuch et al., 2011). It was expected that large and dominant fish are preferentially sampled using hook-and-line (lure-based method), and that also larger fish are healthier. In consequence, low parasitized fish would be over-represented in the sample and the parasite burden in the cage would be under-estimated. An under-estimation of fluke burden in YTK cage would delay treatment and potentially allow the next generation flukes to reach sexual maturity and release eggs in the environment before intervention. The knowledge of the presence and direction of a sampling error when using hook-and-line was deemed of primary importance by the Australian YTK farming industry to properly schedule fluke treatments.

The aim of this study was to evaluate the presence of random and systematic sampling error of the SF and GF burden monitoring in sea caged YTK. The objectives were to evaluate; firstly, the repeatability (precision) of the industry protocol and, secondly, the potential of hook-and-line sampling to bias the estimate of fluke burden. It was hypothesised that hook-and-line biases towards larger, less parasitized YTK and therefore underestimates fluke burden in the sea cage population.