Coping styles and learning in fish: developing behavioural tools for welfare-friendly aquaculture

Mesquita, Flavia de Oliveira (2011) Coping styles and learning in fish: developing behavioural tools for welfare-friendly aquaculture. PhD thesis, University of Glasgow.

Full text available as:
[thumbnail of 2011mesquitaphd.pdf] PDF
Download (3MB)
Printed Thesis Information:


- Given the known stressful effects of many husbandry practices in carp aquaculture and the desirability of improving the welfare of farmed fish, the main aim of the study described in this thesis was to explore the possibility of developing a low-stress sorting system for common carp, based on a conditioned response to a visual cue signalling the presence of food.
 An additional aim was to investigate possible effects of individual stress coping strategy, which necessitated recording the behaviour of and keeping track of known individuals over periods of weeks to months. Photographic images of scales patterns in common carp can be used reliably for individual identification over periods of months. These individual identifiers, together with dye marks, were deemed sufficient for the purposes of the programme of research described in this thesis (Chapter 2).
 In general, rate of emergence from shelter into a potentially-dangerous novel environment containing food (a commonly-used method for screening fish for risk-taking) proved to be a consistent individual trait in common carp, even when fish were tested in different, randomly composed groups of fish on different occasions (Chapter 2).
 Consistent individual differences were also found in frequency of inspection of an unfamiliar object and in ability to gain access to a restricted food source. However, individual differences in performance in these 3 tests (novel environment, novel object and food competition) were unrelated when carp were tested in unfamiliar groups (Chapter 3).
 An examination was carried out on 5 data sets in which morphometric data were collected from common carp or goldfish assigned to a risk-taking phenotype on the basis of a novel environment test. Statistical differences were found in only 2 of these studies; both on common carp, with risk-takers in better condition than risk-avoiders. These support the “growth-mortality trade off” model (Chapter 6).
 Common carp classified as risk-taking, risk-avoiding and intermediate (on the basis of a series of novel environment tests) were given a simple conditioning treatment in which the presence of food in one of two potential feeding compartments was signalled by one of two movable coloured lights. Patterns of settlement (emergence from shelter to explore the learning tank and time to feed) confirmed the original classification into risk-taking phenotype (Chapter 3).
 Over successive trials, the carp learned to forage fast and efficiently. 51.67% achieved this by using the coloured landmark; the remainder adopted a different strategy, swimming to one of the feeding compartments at random and switching immediately to the other compartment if no food was found. This was an efficient foraging strategy because of the close proximity of the two feeding chambers (Chapter 3).
 Once the criterion for learning had been reached, the fish that had learned to associate a particular visual cue were given a reversal learning test, in which food was associated with the previously un-rewarded colour. 83.33% of fish adjusted their behaviour (choosing LC/RS strategy), learning to swim to the previously un-rewarded colour within an average of 12 training sessions (Chapter 3).
 The colour red seems to be more efficient for training carp. In chapter 3, more fish learned to follow the red light compared to the yellow light and in chapter 4, fish trained with red light had a higher percentage of correct choices than fish trained with blue or green lights (Chapter 3 and 4).
 Some differences in behaviour between risk-taking categories were found during both the learning and the reversal learning phase. Risk-taking fish were faster to emerge and find food than risk-avoiders during the learning phase and tended to adopt the random switch strategy during the learning phase. Fish classified as risk-avoiders in terms tended to follow the cue (Chapter 3).
 Small groups comprising one risk-taking, one risk-avoiding and one intermediate carp (tentatively assigned on the basis of a series of novel environment tests) were exposed to a demand-feeding system in which pellets of food were delivered whenever a fish approached and/or touched a sensor identified by a coloured light of a specific colour (red, green or blue). 62% of the 18 groups (with a slight predominance of fish trained using a red light) tested were able to form this association and to feed efficiently under the demand regime. Within these groups, in general the individual that touched the sensor most gained most food. The behaviour of the groups that had failed to learn was unaffected by the addition of a trained “tutor” fish from one of the groups that had learned to touch the sensor for food (Chapter 4).
 For those groups that had learned to approach and touch the sensor, the fish were then exposed to three sensors located in different parts of the training tank signalled by different coloured lights, only one of which (that on which the fish had been trained) delivered food. The position of the sensors was changed between trials. In general, the fish tended to move towards and exploit the sensor signalled by the light colour on which they had originally been trained; this was particularly the case for fish trained on the red light. Carp classified as risk-avoiders made fewer correct choices early on in the three-light phase, but made predominantly correct choices in later trials (Chapter 4).
 Groups of 3 carp that had reached a criterion for having learned to approach a light of each of the three colours (i.e. one red-trained, one blue-trained and one green-trained) were then placed at the centre of a large tank with three lights, one of each colour, in the corner and the light approached by each fish recorded. In general, the fish were significantly more likely to approach the colour of light on which they had been trained, even though this meant separating from their companions. This effect was stronger for fish trained on the red light and disappeared after several (unrewarded) trials (Chapter 4). This result suggests that it might be possible to apply spatial separation of individuals within groups of carp on the basis of a learned association between the delivery of food and a light cue of a specific colour.
 During the course of this programme of work, the opportunity arose through the COST STSM programme to examine risk-taking phenotype, physiological stress response and brain structure in common carp of the 4 families reared either at high densities, in tanks under intensive farming condition or in natural ponds. A disease outbreak compromised the aims of this study, but significant family effects were found among both pond- and tank-reared fish for length, weight and condition factor as well as for emergence time in a novel environment test and approach to a novel object, indicating a heritable component to the variation in these traits. There was no relationship at the family level between emergence time and tendency to approach a novel object (Chapter 5)
 Fish from families that, on average, were heavier and longer took a long time to emerge from shelter, while those from families that were smaller and in poorer condition took more risks in this set up. Tank-reared fish were much slower to emerge than were pond-reared fish, possibly because the latter were in poorer condition (Chapter 5).
 Plasma cortisol levels were markedly higher in pond-reared fish compared to tank-reared fish of the same family, presumably due to the stressful experience of both harvesting and disease. In contrast, plasma glucose levels were lower in pond-reared fish, presumably due to their poor nutritional status (Chapter 5).
 The relationship between an estimate of forebrain size and overall brain size was different in pond and tank reared fish, with most pond reared families having a larger forebrain area than tank reared fish (Chapter 5).
 Also during the course of this programme of work, two related studies were carried out in collaboration with colleagues in the Division of Ecology & Evolutionary Biology. Together with Hussein Jen-Jan, we explored some hidden costs of an aggressive, proactive life style by examining respiratory function in relation to coping strategy in common carp (chapter 6).
 Morphometric analysis of the fine structure of the gills was used to estimate respiratory area and histological analysis of sections through the gill filaments was used to measure the extent to which the secondary lamellae were obscured by epithelial cells. There was a significant relationship between risk-taking phenotype and both the size of the respiratory surface and the extent to which this is exposed as opposed to covered with epithelial cells. Risk-taking fish had larger and more exposed respiratory surfaces than did risk-avoiding fish, with fish with intermediate risk-taking phenotype having intermediate scores. These differences are interpreted as an adaptation to the known high resting metabolic rate of risk-taking fish (Chapter 6).
 Together with Priyadarshini, we look at social interactions and growth in relation to risk-taking phenotype in goldfish. Within the social groups, though most goldfish showed no aggressive behaviour, some of the fish attacked their companions at least once per minute of observation and some individuals showed as many as 8 attacks per minute. These levels are surprisingly high for what is usually seen as a non-aggressive species. In groups comprising 3 goldfish of each risk-taking category, the risk-avoiding fish showed relatively little aggression. Overall, fish that showed any aggression within social groups gained preferential access to a restricted food supply (Chapter 6).
 There were no differences in weight, length or condition between risk-taking and risk-avoiding goldfish at the point of initial screening, but by the end of the experiment the risk-avoiding fish held in groups with other risk-avoiders had gained less weight and had strikingly lower condition factors compared to the other categories of fish (i.e. all risk-avoiders and risk-takers held in mixed groups). It is suggested that some sort of social facilitation of fear keeps levels of stress high in groups composed entirely of risk-avoiding fish (Chapter 6).
 The implications of all these results are considered in a final general discussion (Chapter 7).

Item Type: Thesis (PhD)
Qualification Level: Doctoral
Keywords: Learning, coping styles, fish behaviour, carp
Subjects: S Agriculture > SH Aquaculture. Fisheries. Angling
Q Science > QL Zoology
Colleges/Schools: College of Medical Veterinary and Life Sciences
Supervisor's Name: Huntingford, Professor Felicity Ann
Date of Award: 2011
Depositing User: Ms FO Mesquita
Unique ID: glathesis:2011-2785
Copyright: Copyright of this thesis is held by the author.
Date Deposited: 19 Aug 2011
Last Modified: 10 Dec 2012 14:00

Actions (login required)

View Item View Item


Downloads per month over past year