Behavioral response of tilapia (Oreochromis niloticus) to acute ammonia stress monitored by computer vision

Behavioral response of tilapia ( Oreochromis niloticus ) to acute ammonia stress monitored by computer vision 1 3 first-author †E-mail:xujianyu@nbu.edu.cn; yinliu@ms.qdio.ac.cn †E-mail:xujianyu@nbu.edu.cn; yinliu@ms.qdio.ac.cn xujianyu@nbu.edu.cn 1 2 corresp ‡Corresponding Author ‡Corresponding Author †E-mail:xujianyu@nbu.edu.cn; yinliu@ms.qdio.ac.cn †E-mail:xujianyu@nbu.edu.cn; yinliu@ms.qdio.ac.cn xujianyu@nbu.edu.cn 1 Institute of Agricultural and Bio-Environmental Engineering, Zhejiang University, Hangzhou 310029, China Institute of Agricultural and Bio-Environmental Engineering, Zhejiang University, Hangzhou 310029, China Institute of Oceanology, Chinese Academy of Sciences, Qingdao 266071, China Institute of Oceanology, Chinese Academy of Sciences, Qingdao 266071, China Faculty of Information Science and Technology, Ningbo University, Ningbo 315211, China Faculty of Information Science and Technology, Ningbo University, Ningbo 315211, China The behavioral responses of a tilapia ( Oreochromis niloticus ) school to low (0.13 mg/L), moderate (0.79 mg/L) and high (2.65 mg/L) levels of unionized ammonia (UIA) concentration were monitored using a computer vision system. The swimming activity and geometrical parameters such as location of the gravity center and distribution of the fish school were calculated continuously. These behavioral parameters of tilapia school responded sensitively to moderate and high UIA concentration. Under high UIA concentration the fish activity showed a significant increase ( P < 0.05), exhibiting an avoidance reaction to high ammonia condition, and then decreased gradually. Under moderate and high UIA concentration the school’s vertical location had significantly large fluctuation ( P < 0.05) with the school moving up to the water surface then down to the bottom of the aquarium alternately and tending to crowd together. After several hours’ exposure to high UIA level, the school finally stayed at the aquarium bottom. These observations indicate that alterations in fish behavior under acute stress can provide important information useful in predicting the stress. Ammonia stress Tilapia Computer vision Aquaculture INTRODUCTION 1 Fish behavior under culture conditions holds important information for aquaculturist (McFarlane et al., 2004 2004 ; Kristiansen et al., 2004 2004 ). Most physiological, environmental changes and handling process can induce variations in fish behavior (Israeli and Kimmel, 1996 1996 ; Israeli-Weinstein and Kimmel, 1998 1998 ; Petrell and Ang, 2001 2001 ; Almazán-Rueda et al., 2004 2004 ). Methods of monitoring and quantifying the behavioral response have become potential alternatives for assessing stress, disease, water pollution and toxic material in water (Nogita et al., 1988 1988 ; Kane et al., 2004 2004 ). In intensive rearing systems, particularly recirculating aquaculture systems, ammonia concentrations may reach high levels. In water, total ammonia-nitrogen (TAN) exists in two forms: ionized form NH 4 + , and unionized form NH 3 (UIA) which is far more toxic than NH 4 + (Lawson, 1995 1995 ). Water pH levels have profound influence on the proportion of UIA, with an increase of one pH unit generally causing tenfold increase in the proportion of UIA (Albert, 1973 1973 ). Addition of baking soda to a system to increase its alkalinity can inadvertently increase UIA concentration resulting in catastrophic effect within a short time (Timmons et al., 2002 2002 ). Early detection of acute stress conditions can reduce production losses. Growth of tilapia, routinely measured in chronic ammonia toxicity experiments (Hargreaves and Kucuk, 2001 2001 ; El-Shafaia et al., 2004 2004 ) was not a sensitive indicator of toxicity. Behavioral parameters can be measured continuously in-situ, and are sensitive to responses by fish to acute stress condition (Beitinger, 1990 1990 ). The objective of this study was to monitor the behavioral alterations of a tilapias school in response to different UIA concentration using a computer vision system. MATERIALS AND METHODS 1 Fish and experimental system 2 Tilapias ( Oreochromis niloticus ), (75.2±12.1) g (mean weight± SD ) were assigned into experimental aquaria with 15 fish per aquarium (50 fish/m 3 ) to acclimatize them for 3 weeks prior to the experiment. Water quality parameters were: temperature (23 to 27 °C), oxygen level (6.5 to 7.5 mg/L), TAN (below 1.0 mg/L) and pH (6.5 to 7.3). The photoperiod was 12 h light and 12 h dark. The fish were fed a commercial diet at 3% bodyweight/day. Glass aquarium (115 cm×60 cm×70 cm, 58 cm water depth) with a transparent side was divided into two compartments by a vertical partition wall. The bigger one (90 cm long) was for rearing and the smaller one (25 cm long) for filtering, aeration and heating. The water in the small compartments flowed to the rearing compartments by means of an aquarium pump at the rate of approximately 7 L/min. Water in the rearing compartment returned by gravity to the small compartment for filtering and heating through a hole in the partition wall. Water was replenished with heated, dechlorinated tap water at a flow rate of 15 L/h. Computer vision system 2 Fish images were acquired by color video cameras (Honeywell, GC-655N) connected to a Pentium IV computer equipped with a frame grabber (Microview, MVPCI-700). The light intensity at the water surface was about 300 lx illuminated by overhead fluorescent lamps. Each frame was digitized with a resolution of 768 pixels×576 pixels. Frames were captured using the modified ATVC version 5.0 software package (All_Time Imaging System Corp. Hangzhou, China). The experimental facilities were separated to prevent disturbances to fish. The hardware and coordinate system is shown in Fig. 1 1 , in the X direction between 0 and 500 pixels, in the Z direction between the bottom of the compartment (0 pixels) and the water surface (489 pixels). Behavioral parameters calculations 2 Quantified fish school behavioral parameters included swimming activity and distribution of the school. Swimming activity was calculated using the method of Xu et al.( 2006 2006 ). In short, frame sequences captured at special time interval were subtracted in pairs after image segmentation and extraction. By labelling components caused by fish movement in difference frame, the projected area caused by the movement of every fish in the capture interval was calculated; this projected area was divided by the projected area of every fish in the later frame to get body length moving distance of each fish, and further obtained the body length (BL) speed. The average speed of all fish can respond well to the fish school activity. The distribution parameters were calculated using the method of Israeli and Kimmel ( 1996 1996 ) with the coordinates of the gravity center of the fish school CX , CZ indicating the mean location of the school in Z and X directions, and the spatial standard deviations of fish coordinates in X and Z directions SDX , SDZ indicating average dimension or density of the school. These indices were measured in pixels. Experimental procedures 2 All fish were fasted the day prior to experiments. During the experiment water replenishment was halted and water temperature was maintained at (24.9±0.3) °C. The experiment began at 8:30 am. Ammonia stress was induced by adding ammonium chloride to each aquarium to achieve TAN concentrations of 2.0, 12.0, 40.0 mg/L. At the beginning the pH was maintained at 6.9±0.1 for about 120 min to act as a reference and corresponding UIA concentrations of 0.008, 0.048, 0.159 mg/L respectively. Sodium hydroxide solution was added gradually into the water to increase the pH to 8.1±0.1 during a period of about 20 min, accordingly UIA concentrations increased to 0.13 (low), 0.79 (moderate), 2.65 mg/L (high) level. pH was measured by a pH meter (Hanna Instruments, HI9024); TAN concentration was measured by Nesslerization; UIA was calculated as a function of TAN, temperature, and pH (Albert, 1973 1973 ). UIA remained at the high level for two days. The ammonia and pH level declined slowly due to some ammonia evaporation and nitrification. The water parameters were measured and adjusted by adding ammonium chloride and sodium hydroxide solution twice daily. The first monitoring period (120 min) was recorded for reference when pH was 6.9; the second (300 min) began at 10 min after pH increased to 8.1. During the monitoring period 11 frames were captured once every 2 min. The fish swimming speeds at 10 intervals between 11 frames were averaged and the result AV was taken to represent the activity of the school. The geometrical parameters of the school CX , CZ , SDX and SDZ were the average value calculated from the 11 frames. All trials were conducted in triplicate. F -test was used to test the heterogeneity of variance before and after the UIA treatment. Analysis of variance (ANOVA) was used for determining the temporal effects under the same UIA level. To test for the difference among means under different UIA level at the same time period, Tukey’s test was used. RESULTS AND DISCUSSION 1 AV , CX , CZ , SDX and SDZ under normoxic condition (three periods of an hour respectively in morning, noon, afternoon) had no significant changes. The typical variations of school activity AV , geometrical parameters of school CZ and SDZ at three UIA concentration levels are shown in Fig. 2 2 . Under low UIA concentration level the fish usually occupied the whole tank and were observed to swim actively and randomly in all directions; the activity and distribution parameters had no large fluctuation. The moderate UIA level condition elicited readily observable changes in fish school location and density in vertical direction; CZ and SDZ had significantly large fluctuation ( P < 0.05). After a period of exposure, SDZ decreased with the fish being more crowed together between times. High UIA level condition elicited a marked increase in school activity at the first hour ( P < 0.05) exhibiting an avoidance reaction (trying to flee from high UIA condition). The vertical distributions became wavering, CZ and SDZ had obvious larger fluctuation ( P < 0.05), and later showed concurrent decrease when the swimming activity decreased. The fish gradually dived to the bottom of the aquarium, and stayed there, finally 36% died. CZ and SDZ had more sensitive responses to UIA level changes than swimming activity, frequency distribution of CZ at different location and times under three UIA levels are shown in Fig. 3 3 . The locations of CZ were divided into three ranges: low (0~100 pixels), moderate (100~200 pixels) and high (200~300 pixels). Under normal and low UIA levels CZ were all in the moderate and high location ranges; under moderate UIA level CZ had significant increase ( P < 0.05) in low location range, time had no significant effect on the frequency distributions of CZ ; under high UIA level more CZ were distributed in the low location range, time had significant effect on the frequency distributions of CZ , and in the 5th hour almost all CZ were in low location range. The results of SDZ under different levels and time periods were the same as those of CZ . CX and SDX had no obvious characteristic change during the whole course. Under high UIA level exposure, darkened skin color of fish was observed in 2 h. At the second day, under low and moderate UIA level the behavior had no significant difference with that of the first day, while under high UIA level the fish school had significant small distribution ( P < 0.05) and the location was also significantly low ( P < 0.05). CONCLUSION 1 The behavioral parameters responded sensitively to the increase in UIA level. The fish schools under normal conditions swim actively and distribute evenly. The abrupt increase in swimming activity, acute fluctuation of school: staying at the bottom or ascending to the water surface, obvious decrease in distribution and darkening in body color are typical stress signals of fish to high UIA concentration. These methods of monitoring behavioral parameters have the potential to be developed for detecting stress early in aquaculture systems and therefore decrease production losses. References A Albert . Selective Toxicity , 1973 . : London Chapman and Hall , . Selective Toxicity P Almazán-Rueda , JW Schrama , JAJ Verreth . Behavioural responses under different feeding methods and light regimes of the African catfish ( Clarias gariepinus ) juveniles . Aquaculture , 2004 . 231 ( 1-4 ): 347 - 359 . . Behavioural responses under different feeding methods and light regimes of the African catfish (Clarias gariepinus) juveniles TL Beitinger . Behavioral reactions for the assessment of stress in fishes . J Great Lakes Res , 1990 . 16 ( 2 ): 495 - 528 . . Behavioral reactions for the assessment of stress in fishes SA El-Shafaia , FA El-Goharya , FA Nasra , NP van der Steenb , HJ Gijzenb . Chronic ammonia toxicity to duckweed-fed tilapia ( Oreochromis niloticus ) . Aquaculture , 2004 . 232 ( 1 ): 117 - 127 . . Chronic ammonia toxicity to duckweed-fed tilapia (Oreochromis niloticus) JA Hargreaves , S Kucuk . Effects of diel unionized ammonia fluctuation on juvenile hybrid striped bass, channel catfish, and blue tilapia . Aquaculture , 2001 . 195 ( 1-2 ): 163 - 181 . . Effects of diel unionized ammonia fluctuation on juvenile hybrid striped bass, channel catfish, and blue tilapia D Israeli , E Kimmel . Monitoring the behavior of hypoxia-stressed Carassius auratus using computer vision . Aquacult Eng , 1996 . 15 ( 6 ): 423 - 440 . . Monitoring the behavior of hypoxia-stressed Carassius auratus using computer vision D Israeli-Weinstein , E Kimmel . Behavioral response of carp ( Cyprinus carpio ) to ammonia stress . Aquaculture , 1998 . 165 ( 1 ): 81 - 93 . . Behavioral response of carp (Cyprinus carpio) to ammonia stress AS Kane , JD Salierno , GT Gipson , TCA Molteno , C Hunter . A video-based movement analysis system to quantify behavioral stress responses of fish . Water Res , 2004 . 38 ( 18 ): 3993 - 4001 . . A video-based movement analysis system to quantify behavioral stress responses of fish TS Kristiansen , A Ferno , JC Holm , L Privitera , S Bakke , JE Fosseidengen . Swimming behaviour as an indicator of low growth rate and impaired welfare in Atlantic halibut ( Hippoglossus hippoglossus L.) reared at three stocking densities . Aquaculture , 2004 . 230 ( 1-4 ): 137 - 151 . . Swimming behaviour as an indicator of low growth rate and impaired welfare in Atlantic halibut (Hippoglossus hippoglossus L.) reared at three stocking densities TB Lawson . Fundamentals of Aquacultural Engineering , 1995 . : Boston Kluwer Academic Publishers , . Fundamentals of Aquacultural Engineering WJ McFarlane , KF Cubitt , H Williams , D Rowsell , R Moccia , R Gosine , RS McKinley . Can feeding status and stress level be assessed by analyzing patterns of muscle activity in free swimming rainbow trout ( Oncorhynchus mykiss Walbaum)? . Aquaculture , 2004 . 239 ( 4 ): 467 - 484 . . Can feeding status and stress level be assessed by analyzing patterns of muscle activity in free swimming rainbow trout (Oncorhynchus mykiss Walbaum)? S Nogita , K Baba , H Yahagi , 等 . Acute Toxicant Warning System Based on a Fish Movement Analysis by Use of AI Concept. Artificial Intelligence for Industrial Applications . 1988 . IEEE AI '88, Proceedings of the International Workshop . Hitachi, Japan 273 - 276 . . Acute Toxicant Warning System Based on a Fish Movement Analysis by Use of AI Concept. Artificial Intelligence for Industrial Applications RJ Petrell , KP Ang . Effects of pellet contrast and light intensity on salmonid feeding behaviours . Aquacult Eng , 2001 . 25 ( 3 ): 175 - 186 . . Effects of pellet contrast and light intensity on salmonid feeding behaviours MB Timmons , JM Ebeling , FW Wheaton , 等 . Recirculating Aquaculture Systems , 2002 . 2nd Ed : LLC Ithaca, NY Northeastern Regional Aquaculture Center, Cayuga Aquaculture Ventures , . Recirculating Aquaculture Systems JY Xu , Y Liu , SR Cui , XW Miao . Quantifying the swimming activity of multi-fish using computer vision . Trans CSAE , 2006 . (in press) . Quantifying the swimming activity of multi-fish using computer vision 10.1631/jzus.2005.B0812 Project (Nos. 2001AA620104 and 2003AA603140) supported by the Hi-Tech Research and Development Program (863) of China 2005-03-30 2005-05-29 2005-07-29 2022-01-05 Copyright © 2005, Journal of Zhejiang University Science 2005 Journal of Zhejiang University-SCIENCE B (Biomedicine and Biotechnology) 08 6B