COMPARATIVE GENOMIC ANALYSIS OF HYPERVIRULENT AEROMONAS HYDROPHILA STRAINS FROM STRIPED CATFISH (Pangasianodon hypophthalmus) IN VIETNAM

Thao P. H. Ngo¹, Huong T. T. Vu¹, Thao T. T. Le¹, Hieu C. N. Bui¹, Mark R. Liles² and Channarong Rodkhum³

¹ Aquacultural Biotechnology Division, Biotechnology Center of Ho Chi Minh city, Vietnam

² Department of Biological Sciences, Auburn University, Alabama, United States of America

³ Center of Excellence in Fish Infectious Diseases (CE FID), Department of Veterinary Microbiology, Faculty of Veterinary Science, Chulalongkorn University, Bangkok, Thailand

*Coressponding author: nhpthao.snn@tphcm.gov.vn

Motile Aeromonad septicemia (MAS) outbreaks caused by hypervirulent Aeromonas hydrophila (vAh) strains belonging to sequence type (ST) 251, have been reported and investigated widely in China and United States (US). In this study, for the first time, two genomes of vAh strains isolated from striped catfish in Vietnam have been identified and characterized. These two vAh strains caused catfish mortalities of 51.7 - 86.7% post immersion challenge with 2 x 10⁸ CFU mL^-1 or intraperitoneal injection challenge with 1 x 10³ CFU fish^-1, while no mortalities were observed in catfish challenged with regular A. hydrophila (rAh) strains (p< 0.05). These challenge trials indicate that vAh virulence is impacted by different degrees of penetration into the fish host and suggest that disease outbreaks in production ponds are affected by other pathogens or factors that contribute to vAh invasiveness. Comparative genomic analyses of vAh and rAh strains from Vietnam, together with reference A. hydrophila strains from China and US, were performed to gain an insight into the pathogenicity of the epidemic A. hydrophila strains collected from various locations worldwide. Besides the myo-inositol utilization and mobile genetic elements identified by previous studies, this work found that attachment and motility mediated by pili /fimbriae are also key factors underlying the hypervirulent nature of vAh strains. The vAh heterogeneity was observed on the core genome phylogenetic tree and the O-antigen gene clusters, where Vietnamese vAh strains showed closer genetic relationship with vAh strains isolated from carp species in China compared to vAh strains isolated from catfish in the US. The persistent occurrence of ST251 strains has been responsible for MAS outbreaks occurring frequently in Vietnamese catfish farms since 2013. This study provides important information about the spread of vAh in farmed catfish in Vietnam, its virulence in fish and the highly clonal nature of vAh strains identified in multiple countries.

Keywords: Aeromonas hydrophila, hypervirulence, comparative genomics, striped catfish, phylogenetic tree.

EXPERIMENTAL ANALYSIS EVALUATING A NEW INTEGRATED PERIPHYTON SYSTEM FOR INTENSIVE Macrobrachium rosenbergii AQUACULTURE

David Marioni^{1, *}, Mhd Ikhwanuddin²

¹Institute of Tropical Aquaculture and Fisheries, University Malaysia Terengganu 21300 Kuala Terengganu

² Institute of Tropical Aquaculture and Fisheries, University Malaysia Terengganu 21300 Kuala Terengganu

^*Corresponding author: d_marioni@yahoo.ca

A series of experiments was designed to compare operation in tank-based freshwater aquaculture of an integrated periphyton system (IPS) in tanks as treatment, with a RAS system as control (RAS). Two aquaculture systems were evaluated on growth performance and subsequent survival rate of mixed sex Macrobrachium rosenbergii, comparing Feed Conversion Efficiency (FCE), Body Weight (BW, g), Survival Rate (SR, %), and Average Daily Gain (ADG, g d^-1). Population was 500 in the first trial (T₅₀₀, and C₅₀₀) of 30 days and 1,500 in the second trial (T₁₅₀₀, and C₁₅₀₀) of 40 days. Densities were considered based on plan area of tank, available resting area, and on periphyton biofilm surface area. An applied feed schedule managed feed, alkalinity adjustment, and C:N ratio through addition of molasses. IPS culture volumes were clear water systems, where periphyton biofilm provided ecosystem functions. There were three replications for each treatment, excepting the first run where a single RAS system was compared with triplicate of IPS. The initial weight of individual PL was estimated at 0.05 gm. The mean final BW for IPS₅₀₀, RAS₅₀₀ IPS₁₅₀₀, and RAS₁₅₀₀ were 1.23±0.42 g, 0.8 g single tank, 0.94±0.28 g, and 0.84±0.13 g, respectively. The mean final SR for T₅₀₀, C₅₀₀ T₁₅₀₀, and C₁₅₀₀ were 50±16 %, 65 % single tank, 69±9 %, and 74±4 %, respectively. BW and BW multiplied by SR (BWxSR) were examined statistically by Welch ANOVA with Games-Howell post hoc testing, to evaluate performance and productivity. IPS₅₀₀ and RAS₅₀₀ results for BW and BWxSR were significantly different and demonstrated higher values for each IPS₅₀₀ tank over RAS₅₀₀.  IPS₁₅₀₀ and RAS₁₅₀₀ did not differ significantly as groups for BW or BWxSR. However, Games-Howell revealed the equivalence of IPS and RAS for pairs of tanks, without significant difference, which demonstrated equivalence of the two systems, and it was concluded that IPS can be considered equivalent to RAS in this application. The water quality parameters for both systems were within the suitable range throughout the trials. This study establishes a promising new technological approach with advantages of biofloc technology in terms of no external effluent treatment, zero water discharge, and potential for improved FCR, for freshwater prawn culture, which demonstrated equivalence to a RAS system approach in M. rosenbergii culture. Improvement in intensive aquaculture systems is an efficient way of overcoming water and land scarcity and decreasing environmental impact from aquaculture.

Keywords: Integrated periphyton system (IPS), Recirculating Aquaculture System (RAS), Average Daily Gain, Food Conversion Efficiency

IMAGE ANALYSIS-BASED MACHINE LEARNING TO CLASSIFY AVERAGE WEIGHT OF RED TILAPIA (Oreochromis niloticus LINN.)

Roongparit Jongjaraunsuk¹, Pimlapat Suwannasing¹, Ruangvit Yoonpundh¹, Wara Taparhudee^1*

¹Department of Aquaculture, Faculty of Fisheries, Kasetsart University, Bangkok, Thailand

^*Corresponding author: ffiwrt@ku.ac.th  

Weighing fish is laborious, time-consuming and causes fish stress. To mitigate this problem, image processing techniques and machine learning were applied. In this study, the segmented images of the upper body of fish near the water surface in which the fins, tail, and background were removed (5 fish per image) and were compared with non-segmented images (original images) using machine learning techniques for weight estimation of red tilapia. The images used consisted of 3 groups of fish with an average weight of 300–500, 501-700, and 701–900 g, with 48 images in each group. Decision Tree, K-nearest Neighbor, Naïve Bayes, Support Vector Machine, and Deep Learning models were applied to predict the average weight of the fish in images. The results showed that using the original images had better average accuracy, precision, and recall results than using the segmented images, and there was a statistical difference (P < 0.05). The averages of accuracy, precision, and recall of the original images were 80.97±3.35, 81.23±3.41, and 82.17±5.26 percent, respectively.  While the results of the segmented images were 59.85±3.45, 61.03±4.49, and 59.58±3.45 percent, respectively. In the case of the model used, the deep learning model was found to provide the highest average accuracy because it is a highly effective model to deal with complex data sets. The results of the study found that the use of original images in combination with the deep learning model could be used to estimate red tilapia weight and obtained acceptable results.

Keywords: Weight estimation, Machine learning, Image classification, Red tilapia

MORTALITY RATE OF RED TILAPIA (Oreochromis niloticus Linn.) FINGERLINGS PREDICTION USING MACHINE LEARNING TECHNIQUES

Putra Ali Syahbana Matondang¹, Wara Taparhudee^2*, Ruangvit Yoonpundh², Roongparit Jongjaraunsuk²

¹ Fishery Science and Technology, Faculty of Fisheries, Kasetsart University, Bangkok, 10900, Thailand

² Department of Aquaculture, Faculty of Fisheries, Kasetsart University, Bangkok, 10900, Thailand

^* Corresponding author: ffiswrt@ku.ac.th

Aquaculture has been gradually influenced by the advancement of automation and intelligent technologies. Combined with high-performance computers, namely machine learning technology that can mine information in data, thus providing solutions for smart aquaculture into a new era. This study aims to determine the best predictive model of machine learning techniques in the form of naïve bayes, deep learning, decision tree, random forest and support vector machine in reducing the mortality rate of red tilapia (Oreochromis niloticus Linn.) fingerlings raised with a recirculating aquaculture system. The study began by collecting data on water quality parameters and mortality rate (fish day^-1). The water quality parameters consisted of temperature (^°C), pH, dissolved oxygen (mg L^-1), total ammonia nitrogen (mg L^-1), nitrite-nitrogen (mg L^-1), transparency (cm), and alkalinity (mg L^-1). The results of this study showed random forest has the best performance model in predicting the mortality rate of red tilapia fingerlings with an accuracy of 86.00% ± 11.40%, precision of 100.00% ± 0.00%, and recall of 82.30% ± 13.40%. Through this study, a random forest model can be applied to support the decisions and actions of fish farmers to increase the productivity of red tilapia fingerlings and prevent water quality factors that can lead to mass mortality in aquaculture systems.

Keywords: Data mining, Machine learning, Red tilapia, Mortality rate, Water quality

POTENTIAL OF CULTURE RED NILE TILAPIA (Oreochromis niloticus) MONOSEX ON BIOFLOC SYSTEM WITH DIFFERENT DOSE FOR SPECIFIC GROWTH RATE, SURVIVAL RATE, FEED CONVERSION RATIO, AND FEED EFFICIENCY

Tantyo Ari Adi Wiguno¹, Nina Nurmalia Dewi^2*, Agustono², Agung Luthfi Fauzan³

¹Program Study of Aquaculture, Faculty of Fisheries and Marine, Universitas Airlangga, Surabaya 60115 Indonesia

²Department of Aquaculture, Faculty of Fisheries and Marine, Universitas Airlangga, Surabaya 60115 Indonesia

³Omah Koi Farm Indonesia, Banyuwangi, East Java, Indonesia

*Corresponding author: ninanurmaliadewi@fpk.unair.ac.id

Red nile tilapia are manageable to breed and adaptive to fluctuation. Intensive culture with a high population requires a high feed supply. This crisis can be solved using a biofloc system. The floc-bacteria utilize every waste to become extra food. The existence of male and female fish in one pond impacts low production. The male fish known has the fastest growth. The method used to produce male tilapia is masculinization as the sex reversal method. This study aims to compare two different treatment on the specific growth rate, survival rate, feed conversion ratio, and feed efficiency on monosex red nile tilapia in biofloc system. monosex red nile tilapia on biofloc system used three treatments belonging to control (non-probiotic), P1 (84 ml/m³), and P2 (96 ml/m³). The result shows that the survival growth rate has the best percentage on P2 (1.34% ± 0.35). The independent sample t-test only had a significant difference between control and P2. The survival rate indicates the best rate for control (98%) and P2 (98%). The feed conversion ratio shows the best result on P2 (2.38 ± 0.37), while feed efficiency shows the best percentage on P2 (41.97% ± 0.66). The independent sample t-test for both parameters didn't show a significant difference in all comparisons (p>0.05).

Keywords: tilapia, biofloc, monosex

DEVELOPING SIMPLE METHODS BY USING BOTH ABSORBANCE AND NEPHELOMETRIC 90⁰ SCATTERED- LIGHT FOR DETERMINING THE MICROALGAL BIOMASS IN AQUACULTURE IN VIETNAM

Thao Duc Mai^*1, Hung Quoc Pham¹ , and Kim Jye Lee - Chang²

¹ Institute of Aquaculture, Nha Trang University; thaomd@ntu.edu.vn, hungpq@ntu.edu.vn

² CSIRO Oceans & Atmosphere; Kim.Leechang@csiro.au

              * Correspondence:  thaomd@ntu.edu.vn 

The use of live microalgae as feed plays an important role in aquaculture in Vietnam. Microalgal cells enumeration is usually carried out daily to determine cell density by the traditional method of cell counting using a Haemacytometer. This method is laborious and often produces inaccurate results because it is subjected to user’s knowledge to avoid bias and misuse. Microalgal species including Chaetoceros muelleri and Thalassiosira weissflogii were used as studied subjects due to its rising popularity in aquaculture in Vietnam. In this study, we compared two methods to determine microalgal density by using both absorbance and nephelometric properties. The goals of the new method development are to minimize random error and improve efficiency of cell density determination. A UV-visible spectrophotometer (Biochrom Libra S50 UV/Vis) with a wavelength of 550-880nm was to be used to determine the maximum absorbance. Alternatively, a Sper Scientific Turbidity Meter (model 860040, Taiwan) with a wavelength of 850 nm was used in determination of nephelometric 90⁰ scattered light. The traditional method of cell count by using a Haemacytometer was used as the control and for providing comparison. The predicting models of C. muelleri density determination by using maximum absorbance was  (R²=0.99), and by using the nephelometric method was  (R²=0.97). Similarly, predicting model to determine the cell density of T. weissflogii were  (R²=0.99) and  (R²=0.974), respectively. The study demonstrated that both optical density and nephelometric 90⁰ scattered light can be used to determine the biomass of selected microalgae effectively and accurately.

Keywords: Absorbance, cell density determination, microalgal, nephelometric method