Linköping University | Department of Physics, Chemistry and Biology Type of thesis, 60 hp | Educational Program: Physics, Chemistry and Biology Spring term 2020 | LITH-IFM-A-E-x-20/3808-SE
Cognitive bias and welfare of egg-laying
chicks: Impacts of commercial hatchery
procedures on cognition.
Tiphaine Palazon
Examiner, Per Jensen
Avdelning, institution
Division, Department
Department of Physics, Chemistry and Biology Linköping University Datum Date Språk Language Svenska/Swedish Engelska/English ________________ Rapporttyp Report category Licentiatavhandling Examensarbete C-uppsats D-uppsats Övrig rapport _____________ URL för elektronisk version
Titel
Title
Cognitive bias and welfare of egg-laying chicks: Impact of commercial hatchery procedures on cognition.
Författare
Author Tiphaine Palazon
Sammanfattning
Abstract
Egg-laying hens coming from commercial hatchery go through hatchery procedures considered as stressful and engaging prolonged stress response in adult chickens. The aim of our study was to evaluate the impact of commercial hatching procedure on the affective state of chicks, on their short- and long-term memory and on their need for social reinstatement. To assess the affective state of the chicks we used a cognitive bias protocol integrating the ecological response of a chick to the picture of another chick, to an owl and to an ambiguous cue mixing features of both the chick and the owl pictures. Short-term memory was evaluated by using a delayed matching-to-sample experiment (with 10, 30, 60 and 120 s delays), with conspecifics as sample stimuli. We assessed long-term memory with an arena containing multiple doors leading to conspecifics, in which a chick had to remember which door was open after a delay of one hour or three hours. Finally, we observed the need for social reinstatement through a sociality test arena allowing a chick to be more or less close to conspecifics. We found that chicks coming from commercial hatchery were in a depressive affective state compare to control group. Those chicks also showed higher need for social reinstatement and loss weight. No differences were found regarding short- and long-time working memory between the two groups, but the methods used during these experiments will be discussed. Studying how commercial procedures impact the cognition and more specifically the emotions and state of mind of chickens, is a necessary step forward into the understanding of farm animals’ welfare.
Nyckelord
Keyword
Depression, Early life stress, Emotion, Memory, Optimistic, Pessimistic, Prenatal stress, Welfare, White leghorn.
ISBN
ISRN: LITH-IFM-x-EX--20/3808--SE
_______________________________________________ __________________
Serietitel och serienummer ISSN
Title of series, numbering
Contents
1. Abstract ...1
2. Introduction ...2
3. Material and Methods...6
3.1 Ethical note ...6
3.2 Animals and experimental treatment ...6
3.3 Housing ...7 3.4 Cognitive bias ...8 3.5 Short-term memory ... 10 3.6 Long-term memory ... 11 3.7 Sociality test ... 13 3.8 Tonic immobility... 15 3.9 Weight: ... 16 3.10 Statistical analysis... 16 3.10.1 Cognitive bias ... 16 3.10.2 Short-term memory ... 17 3.10.3 Long-term memory ... 17 3.10.4 Sociality test ... 17 3.10.5 Tonic immobility ... 18 3.10.6 Weight... 18 4 Results ... 19 4.1 Cognitive bias ... 19
4.1.1 Performance of the chicks when exposed to the mirror compared to when exposed to the chick, ambiguous and owl pictures ... 19
4.1.2 Start latencies under the chick, ambiguous and owl picture ... 19
4.1.3 Goal latencies under the chick, ambiguous and owl picture ... 20
4.1.4 Order of exposition to the chick and owl pictures ... 21
4.2 Short-term memory ... 26 4.3 Long-term memory ... 27 4.4 Sociality test ... 29 4.5 Tonic immobility... 32 4.6 Weight ... 33 5. Discussion... 34 5.1 Cognitive bias ... 34 5.2 Short-term memory ... 38
5.3 Long-term memory ... 40
5.4 Sociality test ... 41
5.5 Tonic immobility... 42
5.6 Weight ... 42
6. Conclusion ... 44
7. Societal & ethical considerations ... 45
8. Acknowledgments ... 47
1 1. Abstract
Egg-laying hens coming from commercial hatchery go through hatchery procedures considered as stressful and engaging prolonged stress response in adult chickens. The aim of our study was to evaluate the impact of commercial hatching procedure on the affective state of chicks, on their short- and long-term memory and on their need for social reinstatement. To assess the affective state of the chicks we used a cognitive bias protocol integrating the ecological response of a chick to the picture of another chick, to an owl and to an ambiguous cue mixing features of both the chick and the owl pictures. Short-term memory was evaluated by using a delayed matching-to-sample experiment (with 10, 30, 60 and 120 s of delays), with conspecifics as sample stimuli. We assessed long-time memory with an arena containing multiple doors leading to conspecifics, in which a chick had to remember which door was open after a delay of one or three hours. Finally, we observed the need for social reinstatement through a sociality test arena allowing a chick to be more or less close to conspecifics. We found that chicks coming from commercial hatchery were in a depressive affective state compare to control group. Those chicks also showed higher need for social reinstatement and loss weight at day 1 after hatching. No differences were found regarding short- and long-time memory between the two groups, but the methods used during these experiments will be discussed. Studying how commercial procedures impact the cognition and more specifically the emotions and state of mind of chickens, is a necessary step forward into the understanding of farm animals’ welfare.
Keywords:
Depression, Early life stress, Emotion, Memory, Optimistic, Pessimistic, Prenatal stress, Welfare, White leghorn.
2 2. Introduction
Chickens are the most abundant domesticated animal with a world population of approximately 23 billionin 2018 (FAO). Egg-laying chicks come from breeder companies in which primary breeders are reared. Those chickens can produce up to 300 eggs per year per individual (Clauer, 2012). The eggs are collected daily and transported to a hatchery owned by the breeder company. In the hatchery, the eggs will be incubated and hatched. Once hatched, the chicks are transported back to the breeder company from where they will be sold to other farms and poultry industries. With strains bred to produce large number of eggs per year (Clauer, 2012), techniques of rearing and culling might put at risk the welfare of chickens. Noisy and dark incubators able to hold thousands of eggs, transportation through conveyor belt, manual sexing and vaccination, are known to contribute to an increase of stress in egg layers chicks (Hedlund et al., 2019) and would then be referred as early stress (occurring during the pre-natal and juvenile phase of an individual). Intensive productions with large number of individuals are considered as economically convenient but they are also known to provide poor welfare (Craig and Swanson, 1994; De Jong et al., 2019; Elon, 2015). Nowadays, the pressure we put on our environment and more especially how we impact wild animals and how we manage domestic animals for food production rises question among general public: as the trend of organic food increases, more and more people want to be sure that the animal they eat lived in good conditions (Harper and Makatouni, 2002; Duncan, 2005). It is considered that humans have the responsibility of providing welfare to the animals they are keeping captive for their use (Désiré et al., 2002). In that sense, animal welfare became not only an ethical concern but also a field of scientific research (Duncan, 2005), even though research is still at early stage (Lay et al., 2011).
According to Fraser (2008), there are different views and criteria of animal welfare, depending on how scientists approach it: basic health and functioning (e.g. freedom from pain), natural living (e.g. possibility to perform natural behaviour) and affective states (e.g. possibility to experience positive emotions and limited exposure to negative emotions (Désiré et al., 2002)).
However, intensive farming could impair animal welfare. According to Lay et al. (2011), housing is considered as a possible welfare issue: poor air quality, small cages, low quality of the litter and other parameters might lead to different types of diseases in hens (bronchitis, osteoporosis, footpad dermatitis etc) (Lay et al., 2011). The stress induced by those conditions
3
might lead to abnormal behaviours such as feather pecking and higher level of aggressivity toward other individuals (Keeling and Jensen, 2017). What about the affective state of those chickens? Farm animals are sentient being that are subjected to emotions (Désiré et al., 2002), however, few studies are dedicated to that topic. Regarding the affective state of an animal, or affective valence, it is said that this animal can be in a positive affective state or a negative affective state (Deakin et al., 2016). With this come the terms “pessimistic” and “optimistic”. An individual is regarded as pessimistic when it is subjected to an increase in expectation of negative events, while an optimistic individual will be defined as increasing its expectation in positive events (Mendl et al., 2009; Salmetoet al., 2010).
As deeply subjective, it is hard to assess affective state in animals; however, by testing cognitive bias, it is possible to measure aspect of emotions in animals (Harding et al., 2004). This method is based on the fact that information processed by humans and non-human animals can be biased by their affective states (Harding et al., 2004), and that emotions can affect judgments (Paul et al., 2005). According to Salmeto et al. (2011), exposure to stressors that impair the emotional state of an individual, might alter the judgment in those tasks. In that sense, judgment bias could be considered as a useful indicator of affective state and thus help in assessing welfare of animals (Deakin et al., 2016).
Typically, in order to assess cognitive bias in non-human animals, an individual needs to be trained to associate one cue with a positive event, while another cue is associated with a negative event. The animal is then presented with an ambiguous cue. It is expected that the animal in a pessimistic affective state will interpret an ambiguous cue as a negative cue while an animal in an optimistic affective state would interpret an ambiguous cue as a positive cue (Mendl et al., 2011). This has been supported by studies on rats, dogs, starlings, rhesus monkeys and humans (Mendl et al., 2009). For example, Harding et al. (2004), in the first published study using cognitive bias on animals, used an operant discrimination task with rats. The rats were trained to press a lever when exposed to a certain tonality to experience a positive event and trained to not press a lever when exposed to another tonality in order to avoid an unpleasant event. The animals were then exposed to an ambiguous cue (intermediate tonality) and their behaviours were observed. However, this required that rats first scored a correct answer more than 50% of the time for each tone, using three consecutive daily sessions of 30 minutes (Hardig et al., 2004). Similarly, rhesus macaques had to go through one training session per day, every day during seven day in order to be trained for the cognitive bias experiment (Bethel et al., 2012).
4
In order to analyse if the incubation and hatchery routine of egg-laying chicks in commercial conditions could have an impact on their affective states, we wanted to avoid any training of animals. Thus, we used the test suggested by Salmeto et al. (2010). In their study, Salmeto et al. (2010) used an arena in which cognitive bias was assessed by observing the latencies to cross as straight alley, influenced by different silhouettes: a chick silhouette considered as a positive cue, an owl silhouette considered as a negative cue, and ambiguous cues which contained various mix of the features of the chick and the owl morph. The chick and owl silhouettes are considered to have a predetermined valence for chickens and to be ecologically relevant stimuli. In that sense, no training was required. In their paper, Salmeto et al. (2010) managed to validate this method as the runway latencies of the control group to reach all pictures were gradually longer, as a function of the amount of aversive cues presented in the silhouette. Furthermore, they showed that chicks presenting anxiety-like state of mind had increased latencies toward aversive cues, compare to the control group, suggesting a more pessimistic-like judgment. Depressed-like chicks showed increased latencies toward aversive and appetitive cues compare to control group, reflecting a more pessimistic-like and a less optimistic-like behaviour. There are few clear definitions of animal depression in the literature. However, in this paper we refer to depression as a despairing emotional state, cognitive processes and depressive mood (Fraser and Morilak, 2005; Mc Kinney et al., 1969), which is expected to be linked to a decreased tendency in expecting and recalling positive events (Mendl et al., 2009).
It has already been shown that pre-natal stress could have an impact on the physiology and behaviour of the chicks (Henriksen et al., 2011). Additionally, Rodenburg et al. (2017) found that chicks hatched in similar conditions to commercial production do perform differently in cognitive bias compared to a control group hatched in conditions considered non-stressful. We would then predict that the chicks coming from the commercial hatchery would present a more pessimistic affective state than the control group.
Judgment bias might not be the only cognitive aspect affected by commercial hatchery process. For example, an experiment made by Rodrick et al. (2006), showed that corticosterone injection at embryonic day 10 resulted in poor short-term memory, while injection at embryonic day 14 resulted in poor long-term memory. However, the type of stress (acute or chronic stress) and the moment of application during the embryonic development might results in different outcomes in the offspring and might be a component of maternal programming (Henriksen et al., 2011). Furthermore, not only judgment bias but also memory
5
was reported to be affected by the emotional state of an individual (Mendl et al., 2009; Paul et al., 2005).
As a general basic for our experiment, we used conspecifics as stimuli and cues in order to avoid training and increase motivation to perform the test, based on the high sociality of chickens (Salmeto et al., 2010). We considered that “short-term memory” could be defined as memories with fast temporal decay, whereas “long-term memory” would last for at least hours due to memory consolidation. Indeed, according to Cowan (2009), long-term memory does not demonstrate temporal decay or chunk capacity in contrast to the short-term memory. Additionally, we based ourselves on the characteristics of short-and long-term memory given by Roderick et al. (2006): memory tested 10 m after a single trial discriminative avoidance training refers to short-term memory, while memory tested 60 m after the experiment refers to long-term memory. In order, to detect possible impairment in the short-term memory of chicks coming from commercial hatchery, we used a delayed matching-to-sample experiment. Instead of using objects, we used conspecifics as sample-stimuli, as Regolin et al. (2005), showed that chicks were reacting better with an imprinted object than with food. To assess long-term memory, we used a circular arena presenting four doors allowing the tested chicks to join conspecifics. Only one door was opened during the experiment and the chicks had to remember where this open door was after some delays. If commercial hatchery process is considered as an early stress, then we could expect to see similar results as those seen by Rodrick et al. (2006), meaning that the hatchery chicks would be more pessimistic than control chicks, due to hatchery procedures.
Additionally, a higher motivation for social reinstatement in chickens could be link to a stress state in those individuals (Zidar et al., 2018). Thus, we used a sociality test arena in order to evaluate the stress state of the chicks coming from the commercial hatchery. Finally, duration of tonic immobility is often considered as a criterion to evaluate stress and fear level in individuals (Campo et al., 2007). In that sense we used two tonic immobility tests, one with prior stress (isolation) and one without, to have an indication on if the commercial hatchery increased stress and fearful answer in the chicks.
Being able to assess the affective states of animals, in this case in commercial conditions, could be a step forward in the improvement of their welfare (Mendel et al., 2011). Studying the impact of stressful conditions on the affective state of laying hens and on their memory
6
would not only be relevant at a big scale as hens are one of the most consumed produce but could also be relevant for other captive animals.
The aim of our study was to evaluate if and how commercial hatchery procedures can impact judgment bias, memory, and if those procedures induce stress in the chicks. We would predict that the chicks coming from the hatchery would present a more pessimistic affective state than the control group, an impaired memory and a stronger need to be close to conspecifics due to a stress state.
3. Material and Methods 3.1 Ethical note
All experimental protocols were approved by Linköping Council for Ethical Licensing of Animal Experiments, ethical permit no 1496-2018. Experiments were conducted in accordance with the approved guidelines.
3.2 Animals and experimental treatment
All chickens, both the control and experimental chicks, came from Gimramäs AB hatchery and were from the White Leghorn Hybrid Lohmann LSL strain (Gallus gallus domesticus). On the day of hatching of the control chicks, a commercial hatchery treated group of chicks was collected from the same commercial hatchery as control animals. After standard processing, they were brought to Linköping University and placed in pens in the same room as the control chicks. From this point, both control and hatchery processed chicks were treated in the same way. Behavioural measurements were done at different time periods as outlined below.
Only female chicks were used. Indeed, only females go through the entire commercial hatchery procedure while males are culled before getting vaccination. A total of 88 control chicks and 101 hatchery treated chicks were used during the time of the experiments.
Hatchery treated chickens (HC): At day 22 when taken out of the incubator, chicks were
collected from the Gimramäs AB hatchery. Those chicks went through the entire incubation, hatching and post-hatching procedures at the commercial hatchery. After fertilization, eggs were placed in large incubators (Temperature: Days zero to 18 started with 37.9 °C and ended with 37.1 °C (+/- 0.2). Days 19 to 22 started with 36.8 °C and ended with 36.4 °C (+/- 0.2). Humidity was 30 % (+/- 2 %)). At day 18, eggs were transferred from the incubators to the
7
hatchers until 22 days. Most of the chicks hatched on day 21 but were kept in the hatchers one more day according to commercial procedure. Once hatched, each trail of the incubator was placed on a conveyor belt on which the eggshells were removed by hand. Manual sex sorting was then processed on another conveyor belt by using wing inspection. Following this procedure, males were culled immediately while females were kept. Via a conveyor belt, the chicks were then transported to a vaccination station, which consisted in a needle attached to the table, in order to receive vaccination against Marek’s disease. Finally, chicks were separated and counted automatically before being dropped into transport boxes. Chicks were then transported to Linköping University which required 3.5 hours by car.
Control chickens (CC): 210 eggs were taken for the same commercial hatchery before
incubation. They were then placed into the Linköping University incubators. These eggs were kept and hatched in small incubators (Humidity=55 %, Temperature=37.8 °C, Rotation=on) and hatchers (Temperature=37.5 °C, Humidity=65 %, Rotation=off) at Linköping University, relatively silent compare to commercial hatchery incubators. The chicks were moved from the incubators to the hatchers at day 19. Most of the chicks hatched on day 21 but were kept in the hatchers one more day according to commercial procedures. The chicks were taken out of the hatchers, weighed and leg-ringed at the same time the experimental chicks were transported from the commercial hatchery. After hatching, the chicks were placed in rearing pens, in the same building as the incubators and hatchers to limit additional transport and were used as control animals.
Half of the control chicks went through a dysfunction of the incubators leading to a drop of the temperature for several hours at day 14 of incubation. Those chicks were not used for the first experiment (cognitive bias) but were part of the rest of the experiments. The chicks coming from the broken incubators did not significantly differ in their performances during the experiments expect for their weight at day 21, thus they were not taken into account in the analysis of the weight for this day.
Both the control group and the hatchery group were fed for the first time after hatching at the same moment in the university with commercial chicken starter feed. HC were sham vaccinated as the same time as the CC were vaccinated at three weeks of age.
3.3 Housing
Chickens were kept in four identical pens. HC and CC were kept separately throughout the whole experiments (45 control chicks in pen one and 47 in pen three. 49 hatchery chicks in
8
pen two and 51 in pen four). The chicks were initially held in pens measuring 74x68x160 cm, and pen size was doubled after two weeks. The chickens were kept on saw dust and provided with ad libitum food and water. All of them had access to perches from 1 week of age.
3.4 Cognitive bias
Cognitive bias assessment was conducted at 4 to 6 days of age (HC, n=30; CC, n=30).
The arena we used was described by Salmeto et al. (2011) (Figure 1). The apparatus was 50x30x10 cm and covered by a transparent lid. It consisted in a straight alley maze adjacent to a holding arena. The straight alley maze contained a start box of 10x10x10 cm and a guillotine door. This was followed by a 30x10x10 cm straight pathway that led to the goal line. The goal line was 10 cm before the wall of the arena on which a 10x10x10 cm stimulus cue (picture of a chick, an owl or an ambiguous morph mixing both animals features) was placed (Figure 1). Next to this runway was the holding arena in which five chicks were held. The wall between the alley and the holding arena was opaque but the tested chick could see its conspecifics once it reached the morph (Figure 1).
Cognitive bias was assessed by observing the start latencies (the time the chick took to completely exit the start box) and the goal latencies (the time the chick took to reach the goal line) of chicks toward different goal images (Salmeto et al., 2011): a mirror (reflecting the individual that was tested), a silhouette of a chick, of an owl and silhouette of an ambiguous morph featuring 50 % of characteristics of chick (50c) and 50 % of the owl (50o). The chick silhouette is supposed to represent a positive stimulus followed by a short latency, while the owl is expected to be perceived as a predator, leading to a longer latency (Salmeto et al, 2011). Approach toward the ambiguous morph is expected to reflect cognitive bias of an individual. If an individual is optimistic, it should perceive the ambiguous morph as a potential conspecific and shows shorter latencies than if it perceives it as a predator (indicator of a pessimistic state of mind) (Salmeto et al, 2011). Chicks are motivated to approach the stimuli cues by the presence of other conspecifics next to the goal line but out of sight, unless they reach this line (Figure 1).
This method presents the advantage to be based on ecological answers from the individuals instead of training (Salmeto et al, 2011).
The morphs were made with the Morpheus Photo Morpher v3.01 Professional for Mac (Morpheus Software, LLC) as described by Salmeto et al (2011). After prior pilot study, we
9
decided to use realistic pictures of a chick and an owl instead of silhouettes (Figure 2). The ambiguous picture we used was a 50c:50o of chick and owl characteristics.
Figure 1: Diagram of the arena used for the cognitive bias experiment, showing the start box, the straight alley maze adjacent to the conspecifics arena and ending with the goal line facing the presentation of a mirror or a picture (chick, ambiguous, owl) as stimulus cues. From Salmeto et al. (2011).
a) b) c)
Figure 2: Pictures used during the cognitive bias experiment: a) chick; b) owl; c) ambiguous picture (50o/50c). All pictures measured 10x10cm.
During the experiment five chicks were placed in the holding arena. One of them was gently taken away from the arena and placed in the start box for 15 seconds. The door was then opened for the chick to have access the straight alley maze for maximum 5 minutes. First, the chick was exposed to a mirror and the latencies to exit the start box and to reach the goal line were recorded. If the chick did not reach the goal line within 300 seconds, this chick will not continue the rest of the experiment and another subject will replace it. Using the mirror would also help to habituate the chicks to the arena. Then, latencies to exit the start box and reach the first picture (Chick picture, CC n=10 HC n=20; Owl picture, CC n=20 HC n=10) were
10
recorded. If a chick did not reach the picture within 5mn, its latency was accounted for 300 seconds. The same group was exposed to the second picture after approximately one hour and a half. Finally, they were exposed to the ambiguous picture after the same delay.
3.5 Short-term memory
This experiment was conducted during 7 to 9 days of age (HC, n=20; CC, n=20).
This experiment’s purpose was to evaluate and compare short-term memory between the CC and HC group. One chick was presented with conspecifics either on the right side or on the left side of an arena. After a retention delay during which it could not see the conspecifics, the chick had to recall on which side it saw the conspecifics. This experiment based itself on other delayed matching-to-sample experiment and was assumed to measure the working memory of the chicks.
The arena was 90x60x40 cm. On one side of the box was placed a start box of 10x10x10 cm with a transparent guillotine door. 35 cm in front of the start box, two transparent screens of 30 cm were aligned. One partition of 25 cm was place between the two screens to help the observer recording the choice of the chicks. Another partition of 35 cm was place behind the screens to create two rooms in which conspecifics were held (Figure 3).
Three chicks were kept in each room with water and food. The tested chick was kept during one minute in the start box while two pieces of dark fabric were hiding the transparent screens and therefore the view of the conspecifics. One of the fabrics (either on the left or on the right) was taken off, enabling the chick to see its conspecifics behind one of the screens for 30 seconds. The fabric was placed back for 10, 30, 60 or 120 seconds depending on the moment of the experiment. After this delay, the door was opened, and the chick was free to explore the arena for two minutes maximum. During those two minutes, the first choice made by the chick was recorded: if the chick placed at least half of its body on the side that was previously without the fabric, a correct choice was recorded. If it placed at least half of its body on the other side, an incorrect choice was recorded. If the chick did not make any choice (cf: the chick stays in the start box or do not reach the first partition) a null choice was recorded. All chicks were exposed once to all delays.
11
Figure 3: a) Picture of the arena used for the delayed matching-to-sample experiment, including the dark curtains preventing the chick subject to see the conspecifics when they are not presented during the sample stimuli presentation. b): Diagram of the arena used during the delayed matching to sample experiment, with the same orientation as picture a). The start box holds the tested chick before realising it while the conspecifics area holds the stimuli (three chicks in each conspecifics area).
3.6 Long-term memory
This experiment occurred at 13 and 16 days of age (HC, n=30; CC, n=30).
For the purpose of this experiment, we referred to long-term memory as the retention for at least one hour, of an action previously learned.
The goal of this experiment was to assess if the early stress of commercial hatchery processing might affect the long-term memory of the chicks, by comparing the ability of the CC and HC to remember which entrance would lead them to their conspecifics one hour and three hours after initial training. We will refer to this entrance as “the correct entrance” for the rest of this study.
12
To do so, a circular arena of 62 cm diameter and 35 cm high was build, with four openings of 20 cm facing each other. All openings had removable guillotine doors made of mesh. On each side of all doors, colour cues of 10x25 cm were applied, the colours being black, white, dark grey and light grey with pattern on it. After each door, a covered alley of 30 cm was build followed by a companion holding pen. Those pens measured 38x20x25 cm and had a removable mesh door (Figure 5). Four spotlights were placed above each companion pen. During the experiment two chicks were placed in each pen, with water and food, and one chick was tested. To test the long-term memory of the chicks, one of the four doors was open, enabling the chicks to get closer to the companion pen, while the other doors were closed by a mesh door. The same door was open throughout the experiment for one chick, but for each chick the opened door was different.
For each chick, a series of five consecutive trials was made (training session). During those trials, the chick was placed in the middle of the arena with the opened door behind him. The chick then had maximum five minutes to explore and to find the correct entrance to reach the companion box. If the chick entered the alley and reached the companion box, it was allowed to stay there for one minute (food was placed near the companion box to encourage the chick to stay there during the 1mn break) before being gently placed back in the middle of the arena with the same initial position and continued with the four other initial tests. If the chick did not find the correct entrance within those five minutes, it was directly put back in the middle of the arena, the opened door behind it and continued with the four other initial tests.
Once the five tests were done on all chicks, a delay of one or three hours was applied before doing the same experiment with only one trial (test session).
13
Figure 5: Long-term memory arena’s scheme from above. Red line: doors. a) Dark visual cues, b) Light grey with pattern visual cue, c) White visual cues, d) Dark grey visual cue. On this example all doors were closed except the door on top, with the dark visual cue, which was open. The conspecifics areas held two individuals each and were preceded by a straight alley linking them to the circular arena in which a chick is tested.
3.7 Sociality test
The chicks were subjected to the sociality test at 10 and 11 days of age (HC, n=18; CC, n=18).
The experiment aimed to evaluate the stress state of an individual. Indeed, it is expected that, in an arena offering the possibility to be close to conspecifics or to explore, stressed individual might show more motivation to spend more time next to their conspecifics (Zidar et al., 2018). In that sense we would expect the HC group to spend more time near the conspecifics area (for social reinstatement) while the CC group might explore more the arena and spend less time near their conspecifics (Zidar et al., 2018).
The arena was based on the description of the apparatus made by Wirén and Jensen (2011) and Zidar et al. (2018). The arena consisted in two circles. The outer circle was 160 cm of diameter. A non-fully close second circle of 80 cm diameter was delimited by four partitions
14
facing each other and placed between the first and the second circle, creating this inner circle. In the middle of this second circle was place a companion box of 30 cm diameter made of mesh and covered by a cone to avoid the tested chicks to jump on this holding arena. This box held two chicks that were used as social cues. The wall of the arena was 60 cm high (Figure 4).
Three areas, invisible to the chicks but used by the observers were established. The “social area” was located between the companion pen and the partitions. The “neutral area” was located between the outer circle and the space left between each partition: the chick is not in the social area but can still look at the companion pen. The “behind the screen area” was located behind the partitions: the chick is out of the social area and cannot see its companions anymore. A chick was considered to be in a particular zone when at least 50 % of its body, including the head was located in this zone (Zidar et al., 2018).
During the experiment, two individuals were held in the companion pen while one chick was tested. No food was provided in the companion box to avoid food motivation behaviours from the tested chick. First the three chicks were kept in the companion arena for 5 mn of habituation. At the end of this delay, one chick was gently removed from this pen and was placed in the neutral area, close to the wall of the arena and to a partition but with the possibility to see the companion box, its head directed to the companions arena. All chicks were placed on this same start position. The chick was then free to explore the arena for maximum ten minutes. The experiment was recorded by camera and the start latency (the moment the chick started to move after it was placed on the start position), the time and the frequencies spent in the different areas were measured by the experimenter. Additionally, time spent moving and not moving was recorded (Table 1).
15
Figure 4: Sociality test arena with Soc=Social area (the chick is close to conspecifics); Neu=neutral area (the chicks is further away from conspecifics but can still see them); Exp=explorative zone (the chicks cannot see its conspecifics) and the mesh arena for the conspecifics holding pen in the centre of the arena (Wirén and Jensen, 2018).
Table 1: Ethogram of the behaviours of the chicks recorded in the sociality test.
Moving
Walking: making more than 3 consecutive steps
Running Flying
Not Moving
Pecking
Walking less than 3 consecutive steps Being still for more than 3 seconds
3.8 Tonic immobility
Two tests of tonic immobility were conducted on 48 chicks, one at day 11 and the second one at day 12. One of them consisted on a regular tonic immobility (TI, CC n=29, HC=29): the chicks were placed on their back on a cradle, and a light pressure of the hand of the experimenter was apply on the chest of the chick for 10 sec. Time of first vocalisation, first head movement and rightening was recorded as well as frequency of vocalisations.
The other test was performed as explained above but was preceded by a stressor (TI-S, CC n=29, HC n=29). The stressor consisted in social isolation of 10 minutes in a box containing only sawdust but with enough space for the chicks to move.
16
Half of the chicks were first tested with TI and then with the TI-S, and vice versa for the second half.
All tests were performed by the same person and recorded by camera.
3.9 Weight:
All chicks were weighedat day 1, 4, 7, 21 and 35 of age with a precision of 0.01g.
3.10 Statistical analysis
Statistical analyses were conducted via IBM SPSS Statistics 26 license.
All P-value between0.1 and 0.05 were considered as tendencies, and P-values below 0.05 were considered as significant. We used mean and standard error in our analysis.
Data that were not fitting a normal distribution required to use non-parametric statistic tests. When mixed-design ANOVA was used, non-normal data were transformed according to the log function and then fit a normal distribution. T-tests were applied on normal data.
3.10.1 Cognitive bias
When data distributions were sufficiently close to normal according to visual inspection, we used a T-test for pairwise comparison in order to test the effect of the pictures (chicks, ambiguous and owl pictures) compare to the effect of the mirror. This was done for each bird from each treatment (control and hatchery group) using raw data (in seconds).
Apart from the T-test pairwise comparison on the effects of the pictures and of the mirror, the latencies recorded during the experiment were standardized with the mirror representing the baseline (=1) in order to include individual performances in the analysis. The results were then no longer be expressed in seconds but as the relative proportion of increased latencies compare to the mirror latencies. In order to take into account the non-normality and the repeated measures of the standardized data, a generalized linear mixed model was used to compare the latencies of the treatments (control group and hatchery group) according to the different pictures. Pictures were considered as repeated measures, the treatments as fixed effects, the ID of each individual as random effects and the latencies (start latencies and goal latencies) were considered as fixed targets (dependant variable). Treatments and pictures were then used as main effect builders and as a two-way effect builder in order to see the interaction between those two factors. Normal distribution with identity link function was used. Indeed, the curve of our distribution was shifted to zero. Furthermore, we tested the
17
model fit using different distributions and the identity link, associated with the normal distribution was the one best fitting the data. Pairwise contrast type was chosen.
3.10.2 Short-term memory
The output for this experiment was counted data. In that sense, we used non-parametrical test. A Chi-square test was used to compare the number of correct, incorrect, and null choices between the control group and the hatchery group after the different delays (10 s, 30 s, 60 s, 120 s). In order to see the evolution in the number of correct, incorrect and null choices within each treatment (control and hatchery group) across delays, we use McNemar’stest.
3.10.3 Long-term memory
The data were log-transformed to fit a normal distribution, allowing the use of a mixed design ANOVA on the data from the training session, in order to compare the performance of the control group and of the hatchery group. Treatments (control and hatchery group) were used as in between subject while trials (trial 1, 2, 3, 4, 5) were used as within between subjects. An independent t-test was performed on the log-transformed data to compare the performance of the chicks between the control group and the hatchery group one hour after the training session and three hours after that session (test session). Additionally, a paired sample t-test was used to compare the latencies of the first trial with the latencies found one hour and three hours after the training session within each group.
Finally, an independent t-test was used the compare the differences between the latencies to find the correct entrance one hour and three hours after the training session within each treatment (control and hatchery group).
3.10.4 Sociality test
A Mann-Whitney test was used to compare the latencies to exit the start area between the control group and the hatchery group to counter the non-normal distribution of the data. The data of time spent in each area were not normally distributed. Thus, similarly to the “long-term memory” test, we used the log function to transform the data before using a mixed design ANOVA with the area type (social area, neutral area, behind-screen area) as the within-subject factor and the treatments (control and hatchery group) as the between-subject factor.
18
Finally, because the data on the time spent moving or not moving were normally distributed, an independent T-test was performed to compare the time spent moving and not moving between the two treatments through the experiment.
3.10.5 Tonic immobility
In order to compare the time taken by the control group and the hatchery group to right themselves, to make their first head movement, to produce their first beep, as well as the number of beeps and the number of beeps per second, we used a Mann-Whitney test.
Then in order to compare the difference of performance between the tonic immobility and the tonic immobility prior to stress within the control group and the hatchery group, we used a Wilcoxon sign-rank test.
3.10.6 Weight
An independent T-test was performed to compare the weight of the control group and the hatchery group at day one, four, seven, fifteen, twenty-five and thirty-one.
19 4 Results
4.1 Cognitive bias
4.1.1 Performance of the chicks when exposed to the mirror compared to when exposed to the chick, ambiguous and owl pictures
We compared the start latencies and the goal latencies, in seconds, between the mirror and the other pictures, for the control group (CC) and the hatchery group (HC) (Table 2). Regarding the goal latencies of both groups, there was a significant difference between the mirror and the three pictures, including the chick picture (Table 2). Finally, the control group and the hatchery group did not differ in their latencies to exit the start box (CC: M=17.93, SEM=6.832; HC: M=20.80, SEM=7.963; t58=-0.273, P=0.786) and to reach the goal line (CC:
M=42.47, SEM=11.534; HC: M=32.53, SEM=9.289; t58=0.671, P=0.505) under the mirror.
Table 2: Results of the Paired sample T-test comparing the start and goal latencies (s) made under the mirror with the three pictures (chick, ambiguous, owl) for the control group (CC) and the hatchery group (HC). Statistics values such as absolute mean values (M), Standard Error Mean (SEM), t and P values are shown.
4.1.2 Start latencies under the chick, ambiguous and owl picture
Regarding the start latencies, there was a significant effect of the pictures on the start latencies where the hatchery chicks exited the start box faster with the chick image as a goal than with the owl morph (t29=-2.398, P=0.023), where both groups showed tendency to exit the start box
faster under the chick image than under the ambiguous image (CC t29=-1.819, P=0.079; HC
20
the ambiguous picture than with the owl picture (CC t29=-1.822, P=0.079; HC t29=-1.705,
P=0.099) (Figure 6a, Table 3). There was no difference between the hatchery group and the control group (Figure 6a). Indeed, there was no effect of the treatments on the start latencies (Table 3). However, the hatchery group tended to be slower than the control group to exit the start box under the owl picture (t232=-1.947, P=0.053).
4.1.3 Goal latencies under the chick, ambiguous and owl picture
The hatchery group was overall slower than the control group in reaching the goal line (Figure 6b, Table 3). Indeed, there was an effect of the pictures on the goal latencies with both groups reaching the goal line faster under the chick picture than under the owl picture (CC t29=-3.606, P=0.001; HC t29=-3.196, P=0.003), with the control group having shortest goal
latencies under the chick picture than the ambiguous picture (CC t29=-3.013, P=0.005), and a
tendency for shortest latencies under the ambiguous picture than the owl picture (CC t29
=-1.992, P=0.056). Finally, there was a tendency for the hatchery group to be slower than the control group in reaching the goal line under the chick picture (t232=-1.726, P=0.086), along
with a significant treatments effect and a tendency for an interaction treatments*pictures (Table 3).
Figure 6: Standardized mean latencies to (a) exit the start box and to (b) reach the goal line for the control group and the hatchery group when presenting to a mirror, to the chick picture, to the ambiguous picture and to the owl picture (groups mean ± SEM are shown).
21
Table 3: Generalized linear mixed model comparing the start latencies and the goal latencies between the control group and the hatchery group under the mirror, the chick, the ambiguous and the owl pictures. Statistics values such as F values, degrees of freedom (df1 and df2) and P values are presented. The tests investigated the effects of the treatments, of the pictures and of the interaction between both factors (Treatments*Pictures). The corrected model represents the sums of squares that can be attributed to the set of all the between-subjects, except the intercept. A significant “Corrected model” shows the accuracy of the choice made upon the target distribution and the link with the linear model.
Latencies Start latencies Goal latencies
Statistics values F df1 df2 P F df1 df2 P
Corrected model 4.109 7 232 0.000 11.350 7 232 0.000
Treatments (Control group, Hatchery group)
0.002 1 232 0.964 6.741 1 232 0.010
Pictures (Mirror, Chick, Ambiguous, Owl pictures)
7.284 3 232 0.000 24.119 3 232 0.000
Treatments*Pictures 2.304 3 232 0.078 2.285 3 232 0.080
4.1.4 Order of exposition to the chick and owl pictures
The order of the pictures the control group was exposed to did not impact the start latencies nor the goal latencies (Figure 7a and 7b). Indeed, there was no main effect of the order of the cues and of the pictures, as well as no interaction order*pictures on the start latencies (Table 4). Regarding the goal latencies, there was no main effect of the order of the cues, but a main effect of the pictures along with a significant interaction Order*pictures (Table 4).
Within the hatchery group, the order of the cues it was exposed to, tended to impact the start latencies (Figure 8b, Table 5). The hatchery group tended to be slower to exit the start box under the owl picture when first exposed to the chick picture (t84=1.716, P=0.090).
Additionally, there was a strong tendency for an effect of the order of the cues and the pictures but no significant interaction between the two factors on the start latencies (Table 5, Figure 8a). The goal latencies were not impacted by the order of presentation of the cues (Figure 8b, Table 5). There was no main effect of the order of the cues nor of the pictures, but there was a tendency for an interaction Order*Picture (Figure 8b, Table 5).
22
Furthermore, when both groups were first exposed to the chick picture, the hatchery group tended to be slower to exit the start box under the owl picture than the control group (t112
=-1.733, P=0.086) (Figure 9a). There was a significant pictures effect but no significant treatments effect nor a significant interaction between the two factors on the start latencies (Figure 9a, Table 6). Regarding the goal latencies, there was no significant main effect of the treatments and a tendency for a significant interaction between the treatments and the pictures (Table 6). However, there was a significant main effect of the pictures (Figure 9b, Table 6). When both groups were first exposed to the owl picture, there was no difference between the control and the hatchery group (Figure 10a). There was no significant effect of the treatments, a tendency for a significant effect of the pictures but no significant interaction of both factors on the start latencies (Table 7). However, on the goal latencies, there was a significant effect of the treatments and of the pictures and a tendency for a significant interaction of both factors (Figure 10b, Table 7).
Figure 7: Standardized mean latencies to (a) exit the start box and to (b) reach the goal line for the control group when presenting to the chick picture, to the ambiguous picture and to the owl picture (groups mean ± SEM are shown). “Exposed to chick first” refers to the chicks that saw the chick picture as the first picture. “Exposed to owl first” refers to the chicks that saw the owl picture as the first picture.
23
Table 4: Generalized linear mixed model on the start and goal latencies of the control group after being exposed to the chick picture first or to the owl picture first. Statistics values such as F values, degrees of freedom (df1 and df2) and P values are presented. The tests investigated the effects of the order of the presentation of the cues (“Cues”), of the pictures and of the interaction between both factors (Cues*Pictures). The corrected model represents the sums of squares that can be attributed to the set of all the between-subjects, except the intercept. A significant “corrected model” shows the accuracy of the choice made upon the target distribution and the link with the linear model.
Latencies Start latencies Goal latencies
Statistics values F df1 df2 P F df1 df2 P
Corrected model 0.971 5 84 0.441 5.300 5 84 0.000
Order (Exposed to chick or owl first) 0.065 1 84 0.800 0.072 1 84 0.790 Pictures (Chick, Ambiguous, Owl
pictures)
1.813 2 84 0.169 8.037 2 84 0.001
Order*Pictures 0.038 2 84 0.963 4.115 2 84 0.020
Figure 8: Standardized mean latencies to (a) exit the start box and to (b) reach the goal line for the hatchery group when presenting to the chick picture, to the ambiguous picture and to the owl picture (groups mean ± SEM are shown). “Exposed to chick first” refers to the chicks that saw the chick picture as the first picture. “Exposed to owl first” refers to the chicks that saw the owl picture as the first picture.
24
Table 5: Generalized linear mixed model on the start and goal latencies of the hatchery group after being exposed to the chick picture first or to the owl picture first. Statistics values such as F values, degrees of freedom (df1 and df2) and P values are presented. The tests investigated the effects of the order of the presentation of the cues (“Cues”), of the pictures and of the interaction between both factors (Cues*Pictures). The corrected model represents the sums of squares that can be attributed to the set of all the between-subjects, except the intercept. A significant “Corrected model” shows the accuracy of the choice made upon the target distribution and the link with the linear model.
Latencies Start latencies Goal latencies
Statistics values F df1 df2 P F df1 df2 P
Corrected model 2.781 5 84 0.023 2.131 5 84 0.018
Order (Exposed to chick or owl first) 3.933 1 84 0.051 0.109 1 84 0.818 Pictures (Chick, Ambiguous, Owl
pictures)
2.741 2 84 0.070 2.272 2 84 0.017
Order*Pictures 1.776 2 84 0.176 2.721 2 84 0.103
Figure 9: Standardized mean latencies to (a) exit the start box and to (b) reach the goal line for the control group and the hatchery group when presenting to a mirror, to the chick picture, to the ambiguous picture and to the owl picture, after being exposed to the chick cue first (groups mean ± SEM are shown).
25
Table 6: Generalized linear mixed model on the start latencies and the goal latencies of the control group and the hatchery group when the chicks are exposed to the chick cue first. Statistics values such as F values, degrees of freedom (df1 and df2) and P values are presented. The tests investigated the effects of the treatments, of the pictures and of the interaction between both factors (Treatments*Pictures). The corrected model represents the sums of squares that can be attributed to the set of all the between-subjects, except the intercept. A significant “Corrected model” shows the accuracy of the choice made upon the target distribution and the link with the linear model.
Figure 10: Standardized mean latencies to (a) exit the start box and to (b) reach the goal line for the control group and the hatchery group when presenting to a mirror, to the chick picture, to the ambiguous picture and to the owl picture, after being exposed to the owl cue first (groups mean ± SEM are shown).
Latencies Start latencies Goal latencies
Statistics values F df1 df2 P F df1 df2 P
Corrected model 3.895 7 110 0.001 4.547 7 110 0.000
Treatments (Control group,
Hatchery group)
0.682 1 110 0.411 0.546 1 110 0.461
Pictures (Mirror, Chick,
Ambiguous, Owl pictures)
6.890 3 110 0.000 6.178 3 110 0.001
Treatments*Pictures 2.116 3 110 0.102 2.403 3 110 0.071
26
Table 7: Generalized linear mixed model on the start latencies and the goal latencies of the control group and the hatchery group when the chicks are exposed to the owl cue first. Statistics values such as F values, degrees of freedom (df1 and df2) and P values are presented. The tests investigated the effects of the treatments, of the pictures and of the interaction between both factors (Treatments*Pictures). The corrected model represents the sums of squares that can be attributed to the set of all the between-subjects, except the intercept. A significant “Corrected model” shows the accuracy of the choice made upon the target distribution and the link with the linear model.
Latencies Start latencies Goal latencies
Statistics values F df1 df2 P F df1 df2 P
Corrected model 2.080 7 112 0.051 6.362 7 112 0.000
Treatments (Control group, Hatchery group)
1.917 1 112 0.169 5.978 1 112 0.016
Pictures (Chick, Ambiguous, Owl pictures)
2.413 3 112 0.070 14.362 3 112 0.000
Treatments*Pictures 0.920 3 112 0.434 2.192 3 112 0.093
4.2 Short-term memory
The control group did not make a significant different number of correct choices across delays (Figure 11a). However, the hatchery group significantly made more correct choices at delay 30 sec than after the delay of 60 s (χ2=4, df=1,P=0.039) as well as at delay 120s compare to delay of 60sec (χ2=5.143, df=1,P=0.016) (Figure 11a). Regarding the null choices, the control group significantly made fewer null choice at delay 120s than at delay 30s (χ2=4.167, df=1,P=0.031) (Figure 11b). No significant differences were found within the hatchery group for the null choices (Figure 11b). Finally, there was no significant difference in the number of incorrect choices made across the different delays within each group (Figure 11c).
The hatchery group significantly made more correct choices than the control group at delay 30s (χ2=5.584, df=1, P=0.018). There were no significant differences between the two groups at delay 10sec (χ2=0.114, df=1, P=0.74), 60s (χ2=0, df=1, P=1) and 120s (χ2=0.92, df=1, P=0.34) (Figure 11a). Similarly, the hatchery group made significantly fewer null choices than the control group (χ2=5.01, df=1, P=0.025). No significant differences were found between the groups at delay 10sec (χ2=0.10, df=1, P=0.75), delay 60s (χ2=0, df=1, P=1) and at delay 120s (χ2=0.96, df=1, P=0.33) (Figure 11b). Finally, the control group and the hatchery
27
group made the same number of incorrect choices at all delays expect at delay 10s (χ2=0,625, df=1, P=0.43) (Figure 11c).
Figure 11: Number of correct (a), null (b) and incorrect (c) choices made by the control group (CC) and the hatchery group (HC) after a delay of 10s, 30s, 60s and 120s (groups mean ± SEM are shown).
4.3 Long-term memory
The chicks went through a training session made of five tests (Figure 15). There was no difference between the control group and the hatchery group for any of the trials. However, within the hatchery group, the chicks were slower to find the correct entrance during the first trial than compare to the second trial (Z=-2.733, P=0.006), the third trial (Z=-3.097, P=0,002), the fourth trial (Z=-3.735, P<0.001) and the fifth trial (Z=-3.291, P=0.001) (Figure 15). The hatchery group was also slower to reach the correct entrance during the second trial than compare to the fourth trial (Z=-1.957, P=0.05), as well as during the fourth trial compare to
28
the fifth trial (Z=-2.072, P=0.038) (Figure 15). Additionally, the control group was significantly slower in finding the correct entrance during the first trial than during the second trial (Z=-1.972, P=0.049), the fourth trial (Z=-3.035, P=0.002) and the fifth trial (Z=-2.857, P=0.004). The chicks were slower in their performance during the second trial than the during the fourth trial (Z=-2.086, P=0.037), and tended to be slower during the third trial than during the fourth trial (Z=-1.943, P=0.052) (Figure 15). There was a significant main effect of the trials (F4,216=11.826; P<0.01), but there was no significant main effect of the treatments
(F1,54=0.003; P=0.957) nor significant interaction between the two factors (F4,216=1.163,
P=0.328).
The control group and the hatchery group did not differ in recalling where the open door was one hour after the training session (t26=0.632, P=0.533) and three hours after the training
session (t26=0.733, P=0.470). The control group did not differ in its performance between the
first trial of the training session and its performance one hour after this session (t13=-1.410,
P=0.182) as well as compare to their performance three hours after (t13=0.995, P=0.338).
Similarly, the hatchery group did not differ in its performance between the first trial and its performance one hour after (t13=0.774, P=0.453). However, there was a significant difference
between its performance during the first trial and three hours after (t13=2.572, P=0.023)
(Figure 15).
Finally, the control group did not show difference in remembering where the open door was within the test session between one hour and three hours after training session (t26=0.847,
29
Figure 15: Mean latencies (s) to find the correct entrance during the training session (First, second, third, fourth, fifth trial) and one hour and three hours after that training session (groups mean ± SEM are shown).
4.4 Sociality test
The control group and the hatchery group did not differ in the time taken to exit the start area of the arena (CC Mdn =16, HC Mdn=14.5; U=116, P=0.347) (Figure 12).
Chicks of both groups spent main part of the experiment in the social area and few of them explored the neutral area and the behind-screen area (Figure 13).
The hatchery chicks spent less time than the control group in the behind screen area (F1=6.188, P=0.018) (Figure 13). They also tended to spend more time than the control group
in the social area (F1=3.557, P=0.068) (Figure 13). There was a significant main effect of the
area type (social area, neutral area and behind-screen area) (F2=185.042; P<0.01), no
treatment (Control group, Hatchery group) main effect (F1=2.708; P=0.106) and a significant
30
(F2=3.314; P=0.043). Thus, area type was affecting the time spent in the areas differently
according to the treatments.
Furthermore, the control group and the hatchery group spent more time still (CC M=422.94, SE=17.596; HC M=418.22, SE=12.518) than moving (CC M=140.75, SE=15.074; HC M=159.5, SE=11.734) (Difference in time spent still or moving for CC Z=-3.516, P<0.001; and for HC Z=-3.724, P<0.001). However, there was no significant difference in time spent still between the control and the hatchery group (t32=0.222, P=0.826) and in time spent
moving (t32=-0.993, P=0.328) (Figure 14).
Figure 12: Mean start latencies (s) of the of the control and hatchery group to exit the start zone of the sociality test arena (groups mean ± SEM are shown).
31
Figure 13: Mean time (s) spent in the social, neutral and behind-screen area by the control and the hatchery group during the 10 mn experiment (groups mean ± SEM are shown).
Figure 14: Mean time (s) spent by the control group and the hatchery group moving or not moving during the 10 mn experiment (groups mean ± SEM are shown).
32 4.5 Tonic immobility
The control group was faster to make the first head movement during the tonic immobility preceded by a stress than during the regular tonic immobility (Table 8). There was no other significant difference between the two experiments for the other measurements. The hatchery group tended to be faster to produce their first beep during the tonic immobility preceded by a stress than during the tonic immobility not preceded by stress (Table 8). No other difference was found within the hatchery group.
The control group and the hatchery group did not differ in the time taken to right themselves, to make the first head movement and first beep, or in the number of beeps and in the number of beep per second, both during the tonic immobility and during the tonic immobility preceded by a stressful event (Table 8).
Table 8: Wilcoxon signed-rank test output for the comparison of the time taken by the control group (CC) and the hatchery group (HC) to right themselves (Right), made their first head movement (First head mvt), produce their first beep (First beep), as well as the number of beep (No beep) and the number of beep per second (No beep/s). Mdn TI: Median for the Tonic immobility, Mdn TI-S: Median for the Tonic immobility with prior stress. Z and P values are presented.
33
Table 9: Mann-Whitney output for the comparison of the time taken by the control group (CC) and the hatchery group (HC) to right themselves (Right), made their first head movement (First head mvt), produce their first beep (First beep), as well as the number of beep (No beep) and the number of beep per second (No beep/s) . Mdn CC: Median for the control group, Mdn HC: Median for the hatchery group. Z and P values are presented.
4.6 Weight
The hatchery group weighted less than the control group at day 1 (CC: M=40.378, SD=2.887; HC: M=36.69; SD=2.588; t143=8.558; P<0.01). However, this difference disappeared with
time as no significant differences were found at day 4, 7, 15, 21 and at day 35 (Figure 11).
Figure 16: Mean weight (g) of the chicks form the control group and the hatchery group at day one, four, seven, fifteen, twenty-one and thirty-five (groups mean ± SEM are shown).
34 5. Discussion
Our results showed that the hatchery group was overall slower than the control group in reaching the goal line under the different pictures, suggesting not only a more pessimistic affective state but also a less optimistic affective state. Both characteristics would perhaps be indicative of a depression-like state.
Regarding the delayed matching-to-sample test, the hatchery group made more right choices than the control group after retention of 30 s and made fewer null choices than the control group with that same delay. This could be explained by a higher motivation for social reinstatement in the hatchery group, or by an enhancement of their short-term memory by a mild stress. No differences were found between the two groups in the assessment of their long-term memory, either showing that the commercial hatchery procedures did not impact their ability of retaining information for a longer time or showing that our experiment might not have been suitable to assess this cognitive aspect.
Furthermore, the hatchery group tended to spend more time in the social area and less time exploring the arena during the sociality test compare to the control group, implying a need for social reinstatement in the hatchery group due to past stressful experiences.
Both groups did not differ in their performances during the tonic immobility, perhaps showing that the stress from the commercial hatchery procedures, did not impact their stress state evaluated by tonic immobility.
Finally, the hatchery group weighted less than the control group only at day one after hatching showing a possible effect of the stress from the commercial hatchery.
5.1 Cognitive bias
Our study showed an overall treatments effect, along with pictures effect and an interaction between the two factors, on the latencies to reach the goal line with the hatchery chicks being slower than the control group to reach all the pictures. This suggests that they would perceive all pictures as ambiguous implying that the hatchery group would be more pessimistic than the control group.
Indeed, no differences were found between the two treatments in their start and goal latencies under the mirror. This would suggest than the differences found between the treatments under the other pictures are not due to a more fearful behaviour from the hatchery group toward the arena, but to a difference in the perception of the pictures.
