Understanding the growth and distribution of the prawns is critical for optimising the feed and harvest strategies. An inadequate understanding of prawn growth can lead to reduced financial gain, for example, crops are har-vested too early. The key to maintaining a good understanding of prawn growth is frequent sampling. However, the most commonly adopted sampling practice, the cast net approach, is unable to sample the prawns at a high frequency as it is expensive and laborious. An alternative approach is to sample prawns from feed trays that farm workers inspect each day. This will allow growth data collection at a high frequency (each day). But measuring prawns manually each day is a laborious task. In this article, we propose a new approach that utilises smart glasses, depth camera, computer vision and machine learning to detect prawn distribution and growth from feed trays. A smart headset was built to allow farmers to collect prawn data while performing daily feed tray checks. A computer vision + machine learning pipeline was developed and demonstrated to detect the growth trends of prawns in 4 prawn ponds over a growing season.

The objective of the paper is to explain how we used soft sensing based on machine learning models to estimate some water quality parameters in lined pond conditions from an indoor commercial shrimp ( Litopenaeus vannamei ) farm in Vietnam. Specific water quality parameters provide valuable insight into shrimp pond conditions which are critical for managerial decision making. Some parameters can be easy to measure using relatively inexpensive hand-held sensors submerged in the water and require minimal experience. Other parameters are far more expensive to measure because they require experienced labour, time consuming processes such as laboratory analyses of pond water samples, and ongoing materials costs. Soft sensing refers to the process of estimating a variable from other directly measured variables. In this case, estimating variables that are difficult or time consuming to measure (ammonia, settling solids and total suspended solids) from variables that are easy and quick to measure along with pond input data. The aim is to reduce the time, cost, and requirement for experienced labour to monitor key pond water quality parameters. The study summarises the machine learning models we adopted and the accuracies we achieved in estimating key water quality parameters using soft sensing for commercial, super-intensive indoor shrimp farming. We investigated different machine learning models to accurately estimate the target parameters. We investigated several different machine learning models for predicting the above target variables including Neural Network, long short-term memory Networks, Recurrent Neural Network, and Convolutional Neural Network etc. But these deep learning models did not produce good estimation results. This is most likely because these algorithms require huge volumes of data for effective training and the current data set is very small. Support Vector Regression was a good choice for modelling on small data sets. However, SVR models sometimes generate negative values that makes it unsuitable for estimation of WQ parameters. We used an ensemble tree-based modelling (Random Forest) approach that produced accurate as well as positive predictions hence making it suitable for these datasets. We conducted multiple validation process to understand the effectiveness of the machine learning models. We used a leave-out-one-pond cross validation approach where we left one pond for testing and used the remaining ponds within a trial for model training. These validations were performed within a single trial (called ‘within trial’). In another validation approach, we trained models on ponds from one or multiple trials and tested on ponds from a separate trial (called ‘cross trial’). Ammonia estimation results based on machine learning models indicate that more accurate estimations were achieved using the ‘within trial’ validation than the ‘cross trial’ validation. This variability of ammonia among ponds in initial trials lead to relatively worse ‘cross trial’ estimation performance. However, ‘cross trial’ validations at later stages provided the highest accuracy. This demonstrates that as protocols are managed more consistently, estimating ammonia with high accuracy could become very likely. For total suspended solids estimation, the predicted value provided a reliable enough estimate for pond managers to make informed decisions about the total suspended solids concentrations in the pond. There are some occasions where total suspended solids was underestimated. In occasions where this occurred, the estimation aligned itself with the actual values within the next few samples. Therefore, using the more frequently measured turbidity values to estimate total suspended solids might provide a more realistic indicator of the changes in pond conditions from day to day. Estimation of settling solids was highly inaccurate compared to total suspended solids and further investigation is needed on this front.

The rapidly expanding prawn farming industry is vital in providing high-quality protein for the increasing global population. The key to consistently high yield is the efficient and effective management of pond water, maintaining the optimal growth environment. Our consultations with seven Australian commercial prawn farms found that the lack of access to essential real-time water information and trends has hindered farmers from making confident and timely decisions while in the field, which contributes considerably to either prolonged suboptimal pond conditions or increases in operational costs. This paper describes two aquaculture pond water management tasks that can benefit from wearable technology and discusses general concerns about using smart eyewear in harsh outdoor farming environments. A Pondside Visualisation System was created to allow farmers to access visualised historical and real-time water data during field operations using Google Glass Enterprise Edition 2. The application uses a new data visualisation style that is optimised for small near-eye displays. A validation study was conducted with seven industrial practitioners (including a supervisor and six farm technicians) on a commercial farm with real data from the recent growing season, where each technician made 100 pond management decisions using our system. Objective measurements of task completion time and task accuracy indicated the farmers achieved an accuracy of 86.4% and 89.2% for two management tasks with at least 41.0% less time compared to existing desktop-based practice. A structured expert review confirmed the usability of our system and discussed methods to mitigate issues discovered during the study. We also discussed lessons learned from the project.

We present a decision support system for managing water quality in prawn ponds. The system uses various sources of data and deep learning models in a novel way to provide 24-h forecasting and anomaly detection of water quality parameters. It provides prawn farmers with tools to proactively avoid a poor growing environment, thereby optimising growth and reducing the risk of losing stock. This is a major shift for farmers who are forced to manage ponds by reactively correcting poor water quality conditions. To our knowledge, we are the first to apply Transformer as an anomaly detection model, and the first to apply anomaly detection in general to this aquaculture problem. Our technical contributions include adapting ForecastNet for multivariate data and adapting Transformer and the Attention model to incorporate weather forecast data into their decoders. We attain an average mean absolute percentage error of 12% for dissolved oxygen forecasts and we demonstrate two anomaly detection case studies. The system is successfully running in its second year of deployment on a commercial prawn farm.

In super-intensive shrimp culture, water supplements are often used to adjust and stabilize water quality parameters, microbiological and environmental conditions, as well as suppress undesirable microorganisms. In addition, some water supplements are frequently utilized to boost the diatom population in water. These microorganisms are well known to enhance the shrimp growth performance, and present inhibitory effect against pathogenic vibrio. The present study evaluated the effect of sodium metasilicate (SM) supplementation in the culture water on the diatom population, water quality, zootechnical performance and economics of indoor commercial-scale biofloc-based culture. Litopenaeus vannamei (PL10, 4.1 +/- 1.5 mg initial weight) were stocked in eight 500 m(-2) commercial ponds with (BFT-SM) and without SM supplementation (BFT as a control), with four replicates per treatment. After 92 days (single phase, direct stock), the average values of silica (SiO2) were 8.2 mg L-1 in BFT-SM and 2.5 mg L-1 in BFT (control) (P < 0.05). Microscopical analysis indicated higher diatom dominance in BFT-SM, corroborated by higher chlorophyll-A concentrations. The shrimp biomass and yield were significantly higher in BFT-SM with 3.0 kg m(-3) and 36.2 ton ha cycle(-1) compared to BFT with 2.8 kg m(-3) and 33.5 ton ha cycle(-1), respectively. The BFT-SM treatment produced a 9.75% higher net profit margin than the BFT treatment and a 10.55% increase in revenue. Our study demonstrated positive impacts of SM supplementation on diatom population, resulting in enhanced shrimp performance and economic returns in indoor commercial-scale biofloc-based L. vannamei culture.

The rapid growth of prawn farming on an international scale will play an important role in meeting the protein requirements of an expanding global population. Efficient management of the commercial ponds for healthy production of prawns is the key mantra of success in this industry. It is a necessity to maintain the water quality parameters in these ponds within specific ranges to create an ideal environment of optimal growth of healthy prawns. The current practice of water quality data collection and their usage for decision making on most farms is not efficient and does not take full advantage of the latest technologies. The research presented in this paper aimed at addressing this problem by systematic investigation and development of an integrated framework where (i) modern sensors were investigated for their suitability and deployed for continuous monitoring of the water quality variables in prawn ponds; (ii) novel machine learning models were investigated based on collected data and deployed to accurately forecast pond status over next 24 h. This provides farmers insight into upcoming situations and take necessary measures to avoid catastrophic situations; and (iii) augmented reality-based visualisation methods were investigated for improved data capture process and efficient decision making through real-time interactive interfaces. The paper presents the integrated framework as well as the details of sensing, machine learning, and augmented reality components. We found that (i) YSI EXO2 Multi-Sonde is the best sensor for continuous monitoring of prawn ponds; (ii) ForecastNet (our developed machine learning model) provides best forecasting results with symmetric mean absolute percentage error of 6.1 %, 9.6 %, and 8.5 % for dissolved oxygen, pH, and temperature; and (iii) augmented reality-based interactive interface achieves accuracy as high as 89.2 % for management decisions with at least 41 % less time. The experience of the project as presented in this paper can act as a guide for researchers as well as prawn farmers to take advantage of latest sensors, machine learning algorithms and augmented reality tools.

Water quality (WQ) is a key factor that affects harvest outcome from freshwater ponds. Irregular or aperiodic variations of different WQ variables can influence the growth, survival, and yield of aquatic livestock in the ponds. In this research, WQ and harvest data collected from an Australian prawn farm over a whole grow-out season is used to investigate how the variations of WQ influence the harvest outcome of prawns from the ponds. We present a set of approaches based on machine learning to: (i) understand the effect of five WQ variables in differentiating high and low performing ponds (in terms for harvest performance); and (ii) identify how the variations in these WQ variables over the grow-out season contributed to final harvest outcome (growth and yield). To develop the ponds classification approach, we apply eight different machine learning classifiers: neural networks, support vector machine, k-nearest neighbours, logistic regression, gaussian naive bayes, decision tree, random forest, and AdaBoost. To identify the driving factors (in terms of variations of WQ) that affect growth and yield of aquatic livestock in ponds, we apply three feature selection methods: mutual information, correlation-based feature selection, and ReliefF. Results demonstrate that dissolved oxygen, salinity, and temperature are the three WQ variables that have the greatest influence on overall harvest performance of the ponds. Changes in dissolved oxygen and salinity in the last quarter of the grow-out season, and variations of temperature immediately after stocking contributed the most to differentiate the high and low performing ponds. Crown Copyright (c) 2021 Published by Elsevier Ltd on behalf of IAgrE. All rights reserved.

A number of variables can affect the harvest yield in prawn ponds including dissolved oxygen, ammonia, pH, nitrite, and so on. A set of industry standards are there to maintain these variables within specific ranges for maintaining ideal growing environments for the prawns. However recent harvest results in a prominent prawn farm in South East Asia have shown different performance across ponds even after maintaining these variables within the industry standard ranges. An experiment was conducted recently to collect data on different influence variables (mentioned above) by measuring them at different times over the whole prawn growing season. We have conducted a set of analytical experiments on this data set using machine learning methods to answer three questions: (1) What level of predictive power do the influence variables have i.e. how well they can differentiate between good and bad performing ponds, (2) What is the relative importance of influence variables in predicting pond performance, and (3) How the perceived variables influence the harvest metrics. The paper presents a set of machine learning based analytical approaches undertaken to answer these questions.

The contribution of this study is a novel approach to introduce mean reversion in multi-step-ahead forecasts of state-space models. This approach is demonstrated in a prawn pond water quality forecasting application. The mean reversion constrains forecasts by gradually drawing them to an average of previously observed dynamics. This corrects deviations in forecasts caused by irregularities such as chaotic, non-linear, and stochastic trends. The key features of the approach include (1) it enforces mean reversion, (2) it provides a means to model both short and long-term dynamics, (3) it is able to apply mean reversion to select structural state-space components, and (4) it is simple to implement. Our mean reversion approach is demonstrated on various state-space models and compared with several time-series models on a prawn pond water quality dataset. Results show that mean reversion reduces long-term forecast errors by over 60% to produce the most accurate models in the comparison.

The digestibility of a suite of raw materials was determined when fed to black tiger shrimp (Penaeus monodon) in a series of three experiments. A total of 29 commercial and research raw materials were evaluated using the diet replacement digestibility method. Each of the reference and test diets were fed to tanks of shrimp for one-week prior to commencing faecal collection. The collected faecal samples were kept separate from any feed residue through using a discrete feeding period, after which uneaten feed was removed before a separate faecal collecting period. The same reference diet and soy protein concentrate diet were used across each of the three experiments and demonstrated consistent digestibility using this method. Most raw materials demonstrated some utility for use in diets for shrimp, with digestible protein or energy values >0.800. However, there were some raw materials (e.G. camelina meal) that provided very little nutritive value for shrimp. This study presents data on the digestibility and digestible nutrient content of a wide variety of raw materials, providing a clear basis for progressing to formulating shrimp diets on a digestible protein and energy basis, thereby optimising dietary formulation, maximising ingredient utilisation and reducing impacts of uneaten feed.

Feed management strategies that maximize shrimp growth and optimize feed utilization are critical to the cost-effectiveness of production. In this study, juvenile Penaeus monodon (similar to 3g) were cultured for 6weeks in a laboratory-based clear-water tank system. The experiment design was a three-way factorial with two diets (Diet A - standard industry formulation or Diet B - the same diet with 10% microbial biomass), two feed frequencies (twice or six times daily) and three rations (60%, 80% and 100% of satiation). The results demonstrated clear growth benefits of feeding more than two times per day and feed efficiency benefits of a restricted ration. There was also a significant interaction between frequency and ration, which demonstrated that growth improved using six feeds compared with two feeds as ration amount decreased. The effects of frequency and ration were consistent for both diets; however, the addition of a microbial biomass provided significant growth improvements across all treatments. These outcomes define the gains produced by the combined effect of frequency and ration and suggest a compromise between feed utilization and feeding effort for adoption in feed management strategies.

20 Feed management strategies that maximise shrimp growth and optimise feed utilisation are 21 critical to the cost effectiveness of production.. In this study, juvenile shrimp (~3 g) were 22 cultured for six weeks in a laboratory based clear-water tank system. The experiment design 23 was a three way factorial with two diets (Diet A – standard industry formulation, or Diet B – 24 the same diet with 10% microbial biomass), two feed frequencies (twice or six times daily) 25 and three rations (60%, 80% and 100% of satiation).. The results demonstrated clear growth 26 benefits of feeding more than 2 times per day and feed efficiency benefits of a restricted 27 ration. There was also a significant interaction between frequency and ration, which 28 demonstrated that growth improved using 6 feeds compared with 2 feeds as ration amount 29 decreased. The effects of frequency and ration were consistent for both diets; however, the 30 addition of a microbial biomass provided significant growth improvements across all 31 treatments. These outcomes define the gains produced by the combined effect of frequency 32 and ration, and suggest a compromise between feed utilization and feeding effort for adoption 33 in feed management strategies. 34 35

During spermatogenesis, giant tiger shrimp (Penaeus monodon) from Queensland, eastern Australia had a high proportion of testicular spermatids that appeared hollow' because their nuclei were not visible with the haematoxylin and eosin stain. When examined by transmission electron microscopy, the nuclei of hollow spermatids contained highly decondensed chromatin, with large areas missing fibrillar chromatin. Together with hollow spermatids, testicular pale enlarged (PE) spermatids with weakly staining and marginated chromatin were observed. Degenerate-eosinophilic-clumped (DEC) spermatids that appeared as aggregated clumps were also present in testes tubules. Among 171 sub-adult and adult P.monodon examined from several origins, 43% displayed evidence of hollow spermatids in the testes, 33% displayed PE spermatids and 15% displayed DEC spermatids. These abnormal sperm were also found at lower prevalence in the vas deferens and spermatophore. We propose Hollow Sperm Syndrome (HSS)' to describe this abnormal sperm condition as these morphological aberrations have yet to be described in penaeid shrimp. No specific cause of HSS was confirmed by examining either tank or pond cultured shrimp exposed to various stocking densities, temperatures, salinities, dietary and seasonal factors. Compared with wild broodstock, HSS occurred at higher prevalence and severity among sub-adults originating from farms, research ponds and tanks. Further studies are required to establish what physiological, hormonal or metabolic processes may cause HSS and whether it compromises the fertility of male P.monodon.

The aims of this study were to identify genes involved in reproduction and lipid pathway metabolism in Penaeus monodon and correlate their expression with reproductive performance. Samples of the hepatopancreas and ovaries were obtained from a previous study of the reproductive performance of wild and domesticated P. monodon broodstock. Total mRNA from the domesticated broodstock was used to create two next generation sequencing cDNA libraries enabling the identification of 11 orthologs of key genes in reproductive and nutritional metabolic pathways in P. monodon. These were identified from the library of de novo assembled contigs, including the description of 6 newly identified genes. Quantitative RT-PCR of these genes in the hepatopancreas prior to spawning showed that the domesticated mature females significantly showed higher expression of the Pm Elovl4, Pm COX and Pm SUMO genes. The ovaries of domesticated females had a significantly decreased expression of the Pm Elovl4 genes. In the ovaries of newly spawned females, a significant correlation was observed between hepatosomatic index and the expression of Pm FABP and also between total lipid content and the expression of Pm CYP4. Although not significant, the highest levels of correlation were found between relative fecundity and Pm CRP and Pm CYP4 expression, and between hatching rate and Pm Nvd and Pm RXR expression. This study reports the discovery of genes involved in lipid synthesis, steroid biosynthesis and reproduction in P. monodon. These results indicate that genes encoding enzymes involved in lipid metabolism pathways might be potential biomarkers to assess reproductive performance.

A factorial experiment was conducted with black tiger shrimp (Penaeus monodon) juveniles to determine the effects of varying protein inclusion in the diet and also varying inclusion of a microbial biomass on growth, feed and nutrient utilization when fed in indoor laboratory conditions. The growth performance of the shrimp improved with increasing diet protein level. However, in the absence of the added microbial biomass, this growth performance plateaued at the 480g/kg protein level. The addition of the microbial biomass improved growth at each inclusion level of both protein and microbial biomass. No plateau in growth was observed with the addition of the microbial biomass. Improvements in feed conversion were seen with increasing dietary protein levels and also the inclusion of the microbial biomass. Examination of the feed intake of each treatment supports that there was a combined effect of an increase in feed intake and improvements in feed conversion that contributed to the improvements in growth performance with the use of the microbial biomass, but the increases in dietary protein level largely influenced growth through improvements in feed conversion.

A series of experiments were conducted with black tiger shrimp (Penaeus monodon) juveniles to firstly determine the effects of reducing fishmeal inclusion in a diet and then to evaluate the potential for a microbial bioactive to support complete replacement of both fishmeal and fish oil in feeds when fed under clear-water and green-water conditions. The isoproteic and isoenergetic replacement of fishmeal resulted in a consistent decline in growth performance indicating that at every decrease in fishmeal below an inclusion level of 45% there was a decline in performance. In a subsequent trial undertaken in a clear-water tank system diets devoid of both fishmeal and fish oil fed to shrimp were demonstrated to produce poorer performance than a fishmeal and fish oil reference diet. However the addition of a microbial bioactive to the diet resulted in not only a compensation for the replacement of these ingredients but also additional growth. Replication of the clear-water trial in a green-water tank system not only produced similar results, but also showed that the green-water system largely compensated for the performance lost through replacement of fishmeal and fish oil. However it was also shown that the use of the microbial bioactive in the diets still resulted in improved growth performance of shrimp. This study has effectively demonstrated a viable strategy for not only a complete replacement of all fishery products in shrimp diets, but also an improved performance strategy.

The egg and nauplii production parameters of a single stock of domesticated Penaeus monodon was evaluated in four discrete generations. In generations 1, 2, and 5 (G1, G2, G5) the stock was maintained under similar husbandry and dietary conditions within controlled temperature tank systems from two to 11months of age. In generation 8 (G8), the stock was reared in low-density ponds from two to five months of age and then transferred to tank systems for rearing through to 11months of age as per the earlier generations. At 11months of age, females were unilaterally eye-stalk ablated and their reproductive performance evaluated. Reproductive performance was evaluated for first spawnings obtained within 20days post-ablation and measures included: the number of eggs and nauplii per spawning and per gram of female weight per spawning; percentage of spawnings hatching, and; percentage of nauplii hatched. Numbers of eggs (000s) per spawning significantly increased over successive generations (mean±standard error) from 121±15 in G1 through to 380±56 by G8. Numbers of nauplii (000s) per spawning significantly increased beyond G2, increasing from 18±5 in G2 through to 161±23 by G5. These results demonstrate improvements in egg and nauplii production of the domesticated P. monodon stocks over successive generations.

Sibling harvest age Black Tiger shrimp triploids and diploids of both sexes were reared to reproductive maturity, crossed with wild caught females and males, conditioned for spawning and a comprehensive reproductive performance trial was undertaken. Ovarian development, spawning frequency, fecundity, hatch rate, gonad morphology, male reproductive tracts and thelycum impregnation rates of the wild female x triploid male cross were assessed. After ablation, ovarian development and cycling between wild G(0) diploid and G(1) diploids was not significantly different, whereas G(1) triploids failed to show any signs of ovarian development and cycling, thus resulting in no G(1) triploid female spawnings. There were 10G(0) diploid female x G(0) diploid male first-spawnings and 9G(0) diploid female x G(1) diploid male first-spawnings, all of which produced viable nauplii. In comparison, there were 7G(0) diploid female x G(1) triploid male first-spawnings, none of which produced viable nauplii. The 26 wild G(0) diploid female spawnings had more eggs than the 1G(1) diploid female spawning. Gonad morphology and male reproductive tract assessments showed impaired reproductive development in triploid gonadal tissues of both sexes (compared with sibling diploids and wild shrimp) to a point where complete maturation had not occurred. The thelycum of 16 wild G(0) diploid females crossed with G(1) triploid males had no visible spermatophore present, suggesting that G(1) triploid males are incapable of developing viable spermatophores and mating with females. This study demonstrates that the triploid females and males are incapable of producing viable gametes and are thus reproductively sterile.

Improving the seedstock production of domesticated broodstock remains a high priority for the black tiger shrimp, Penaeus monodon, farming industry in Australia and other shrimp farming regions. In this study, wild-caught (W) broodstock were reciprocally crossed with eighth generation (domesticated-selected) broodstock (G8) and the reproductive performance was evaluated to identify the key parameters and gender influences currently constraining seedstock production from domesticated broodstock. Nauplii production was significantly lower in G8 than W broodstock and differences in nauplii production were found to be more influenced by the female than the male origin of the broodstock. The main constraints on nauplii production in G8 females were inferior egg hatch rates and egg production. The percentage of eggs that hatched was 44% lower in G8 females (24.3%) than W females (43.7%). The total number of eggs per spawning was 25% lower in G8 females (413,000) than W females (552,000) and the relative number of eggs spawned per gram female body weight was 37% lower in G8 females (2476) than W females (3909). In addition, a significantly higher percentage of the W females (70.8±5.7%) matured to ovary stage 4 than G8 females (2.1±2.1%). But there was no difference in the percentage of females that developed stage 3 ovaries or the percentage of females that spawned. There was no difference in the total number of sperm per spermatophore or sperm quality (based on acrosome reaction) between W and G8 males, which may explain the lack of any male effect on fertilization. Our results suggest that female broodstock quality may often explain the low nauplii output of domesticated broodstock.

The Crustacean Society Summer Meeting (TCSSM) - 10th Colloquium Crustacea Decapoda Mediterranea (CCDM), 3-7 June 2012, Athens, Greece