Turkeys used in this study were handled following the guidelines defined by the ‘Guide for Care and Use of Agricultural Animals in Research and Teaching’ (FASS, 2020). This study was approved by the Ethics Committee on Animal Use of the School of Animal Science and Food Engineering at the University of São Paulo (approval number: 2669030122).

The experiment was conducted at an experimental farm in southern Brazil latitude S: 27°09’57.00 °and longitude W: 52°37’11.00 °during the autumn of 2022.

The turkeys were housed in a conventional house with 2.5 mm mesh screens, white curtains, and positive pressure, allowing the turkeys to receive natural lighting. Internally, the facility was divided into 76 boxes of 7.6 m² each. The initial stocking density was 5.4 birds/m².

Each pen was supplied with a hanging feeder (Agromarau, Brazil, developed for poultry) and two bell drinkers (Plasson, Brazil, developed for turkeys), providing standard commercial feed and water ad libitum with no difference between treatments and a maximum final stock density of 21 kg/m² at the time of slaughter. The turkeys received four types of feed throughout their lives according to their age and nutritional needs.

All turkeys had access to wood shavings, which provided them with the opportunity to peck and dust bath, and it was kept dry and fluffy throughout the entire research period.

Shortly after hatching at a commercial hatchery, the turkeys were separated by sex, and half of them underwent a beak treatment. Beak trimming (BT) was performed using Nova Tech® infrared laser equipment, removing one-third of the upper beak.

The total number of animals used in the experiment were transported to the farm on the first day of life and randomly distributed among the boxes, following the experimental design. A total of 3,116 female turkeys from the Nicholas Super Select (SS) lineage were housed and divided into a factorial 2 × 2 design with 19 replications in each combination, totaling 779 turkeys per treatment.

The flock remained housed in the boxes until they reached an average slaughter weight of approximately 5 kg/bird at 62 days of age, after which they were slaughtered.

The experimental design was completely randomized with a factorial 2 × 2, and the distribution pattern of treatments consisted of debeak or non-debeaked turkeys combined with the presence or absence of EE (PET refillable/reusable bottle). The most common management used in turkey farming in Brazil is beak trimming without EE, and the desired one from an animal welfare perspective is non-debeaked animals with EE (without physical alteration and with an enriched environment).

Half of the turkeys had access to an enriched environment during the experimental period. Refillable/reusable plastic bottles (PET) were considered pecking objects and contained approximately 500 ml of water with blue food-grade dye. They were hung inside each pen in the central area, allowing all turkeys access to the enrichment. In cases of leakage, falling, or puncture, enrichment was replenished. The object was maintained at the eye level of the turkeys and adjusted as the animals grew, so that the height remained suitable to keep the turkeys interested in the object.

The choice to use simple EE, such as reusable plastic bottles (PET), even though its effectiveness may be limited, was driven by the need for EE to be accepted by farmers in intensive production. This is because it is a readily available material, posing no danger to animal health or a potential entry point for sanitary issues. It is easily replaceable in the case of breakage and does not increase the labour of the producer because it only requires regular height adjustment. Crowe and Forbes (1999) compared different types of environmental enrichment and concluded that perches and objects were more effective than straw and forage (grain scattered on the litter) in reducing the proportion of birds pecked, corroborating that this type of object could be effective as an EE.

Weekly weighing of all animals was performed to monitor weight gain and to determine the average weight of the turkeys. Weighing began on the seventh day of housing and was repeated weekly until the end of the experiment. The overall data of the flock were considered for the performance calculations.

Feed conversion was measured weekly by using the ratio of feed consumption to weight gain. The weekly feed intake was obtained by dividing the total amount of feed offered and subtracting the number of leftovers for the evaluation days and the number of animals.

Bird mortality was monitored daily, recorded on spreadsheets, and compiled weekly at the same ages when the turkeys were weighed. The need to euthanize injured animals was also analysed daily, and in cases where elimination was necessary, it was carried out by trained personnel, with the compilation performed weekly on the same dates as the previous parameters.

At 39 days of age, flock equalization was performed, meaning that the number of turkeys was standardized in all boxes to maintain a constant number of feeders and drinkers among the boxes. Surplus animals were randomly allocated to border areas.

The average weight of the flocks was determined by calculating the total weight of the turkeys in each treatment and dividing it by the number of animals allocated to each treatment. No health issues were observed in the flock; consequently, there was no need to medicate the turkeys. To reduce experimental errors, all the measurements were performed by the same team.

Behavioural observations were conducted weekly in all boxes, starting when the turkeys were 6 days old and repeated when the birds were 12, 20, 26, 33, 40, 47, 54, and 61 days old, during nine weeks.

For each observation day, a scanning session was performed in a random order among the pens. The trained observer remained 1.5 m from the animals to minimize disturbance in the box and ensure that all turkeys could be seen. In each pen, the turkeys were settled for 90 s before starting observations. Initially, the pecking object was assessed by counting the number of turkeys interacting with enrichment. Subsequently, the number of turkeys performing each of the behaviours (Table 1) was counted using scan sampling with a 3-minute interval between each box. The total behaviour in each box must be the number of turkeys. During each observation period of 4 h, each box was evaluated once, considering one treatment replication, totalling 19 replications per treatment per day of observation. Behaviour was evaluated once per age, starting at 8:00 a.m. each week.

Every day, an observer entered the pen and carefully examined all the birds by walking slowly around the pen. The skin lesions recorded daily were totalled by week and pen at the time of weighing. Each bird was assessed by a qualified person and the number and location of the injuries were documented. The injuries considered were wounds resulting from aggressive feather pecking, observed as plumage damage, loss of feathers, and open injuries with or without blood (Nikolov & Kanakov, 2022). The body areas evaluated were snood, neck, head, tail, and wing. Severely injured turkeys were removed from the pen and treated if recovery was possible. If there was no chance of recovery, the bird was culled and a record of the cull was recorded.

At 62 days of age, the animals were caught by a specialized team in loading turkeys and transported to separate crates by treatment in a dedicated vehicle for turkey transportation.

There is a training program regarding the welfare responsibilities of the turkeys for employees in charge of catching, transporting, and pre-slaughter handling. The catching team has a leader to monitor the process to avoid any occurrences related to improper handling of animals.

Turkeys with health problems, fractures, or injuries that compromised their welfare were excluded. Before transportation, all turkeys received water until loading began, and fasting at the farm was for six hours before loading. The turkeys were caught on their backs, two at a time.

To verify peck injuries, the turkeys were slaughtered and observed on-site at a factory located in southern Brazil. The treatments were slaughtered separately with a three-minute interval between them to maintain traceability and prevent mixing between treatments.

The conscious turkeys were hung by their legs on moving metal hooks connected to a rotating shackle, which transported the turkeys to a stunning tank containing electrified water.

After stunning, the turkeys were subjected to manual bleeding, scalding, and plucking. The carcass quality in turkey was analysed through visual inspection at the exit of the plucking tunnel. At this point, the presence of pecking injuries in turkeys was also examined.

The design was a completely randomized factorial model, including the effects of environmental enrichment (presence or absence/without EE), beak trimming (BT or non-BT), their interaction, and time as repeated measures through the growing period, and pen as a random effect. Each pen was considered as a replicate.

Data were tested for normality using the Kolmogorov–Smirnov test. The performance data showed a normal distribution, and no transformation was needed. Mean comparisons were performed using the Tukey-Kramer test at a significance level of 5%.

For behaviour analysis (daily behaviour, injuries), the assumptions of normality, homoscedasticity, and sphericity were checked (p>0.05). Data were arcsine square root transformed to achieve normality (p<0.05). We then returned to the real values to present the results. The daily studied behaviour was expressed as the percentage of animals engaged in the activity at the evaluated moment. Injuries were expressed as the average value per week per replication (n=19). Head lesions were detected only once during the experiment and were not considered in the statistics. Mean comparisons were performed using the Tukey-Kramer test at a significance level of 5%.

For carcass injuries, a descriptive analysis was done, and results are expressed as percentage of the total number of carcasses which presented any lesion. The statistical analysis of this study were conducted using SAS software, Version 9.4 (2022). All data are presented as mean ± Standard Error of the mean (SEM).

No effect of debeaking × EE interaction on performance variables, average final weight, percentage of mortality, and culled animals was observed in the female turkeys studied (p>0.05; Table 2), but an effect was found for feed conversion ratio and feed consumption (p<0.05).

When not debeaked, turkeys presented better FCR when they had access to EE (p<0.05), which was explained by less consumption (p<0.05). However, debeaked turkeys had a better FCR when they had access to EE (p<0.05). When EE was not offered, the FCR was similar between BT and non-BT animals (p>0.05) (Table 2).

Analysing isolated factors, for feed consumption, non-debeaked birds consumed more than the debeaked ones (p<0.05; 380.17±1.964 kg, 374.36±1.964 kg, respectively). No other effects were observed (p>0.05). Although no significant effects were observed on mortality and culling, the main causes of culling were lesions from injurious pecking, heart failure, and other causes.

An interaction of debeaking × EE was not observed in turkeys’ behaviour (p>0.05). Debeaking had an effect on gentle pecking and injurious pecking (p<0.05; Table 3). The presence of EE had an isolated effect on exploratory behaviour, with a higher frequency in turkeys with EE access (14.3%) than in those without EE access (12.6%).

When evaluating the behaviour in relation to age, differences were found for all behaviours (p<0.05). However, in this study, it was important to understand the differences related to the treatments. In this case, it was found an interaction of age × debeaking × EE for resting, use of EE, injurious pecking, and eating (p<0.05; Table 4).

Injurious pecking was higher at 33^rd week-old in birds without access to EE (p<0.05), even though they were debeaked. The presence of EE did not change injurious pecking during the experimental period (p>0.05).

Resting behaviour was lower in the 3^rd, 4^th, and 5^th week-old debeaked turkeys without access to EE (p<0.05). The use of EE presented a frequency of 1.27±0.069%/day, despite the debeaking procedure (p>0.05), and the use of EE varied over the 9 weeks, reducing and increasing between weeks in a random way (p<0.05; Table 4).

Eating behaviour was lower at week 4^th, 6^th, 8^th, and 9^th for debeaked turkeys with access to EE (p<0.05).

There was an interaction between debeaking and EE (p<0.05). Non-debeaked turkeys presented more lesions (p<0.05), despite they had access or not to the EE (p>0.05; Table 5). The number of lesions was higher during the last two weeks before slaughter in non-debeaked turkeys (p<0.05; Table 6).

Debeaking proved to be effective, as the number of injuries was lower (p<0.05) in non-debeaking animals (6.7±0.89/week) compared to debeaking at the hatchery (1.38±0.89/week).

Regarding EE, number of injuries was similar between turkeys that had no access to EE (4.2±0.89) and animals that had access to EE (3.9±0.89; p>0.05).

The areas most affected by aggressive pecking that resulted in a higher number of lesions in turkeys were the wing, snood, tail, and neck (59.2%, 19.8%, 10.9%, and 10.1%, respectively).

At slaughter, the injuries in each treatment group were assessed separately to record the number of damages. The total post-mortem inspection rejection (PMI) showed trends between BT and non-debeaked turkeys and the presence or absence of EE. Of the 2,964 slaughtered turkeys, 66 (2.22%) had carcass injury. Turkeys with intact beaks had more carcass injuries when they had no access to EE (40.9%) compared to the ones with EE access (31.8%). BT turkeys had lower carcass injuries when they had access to EE (10.6%) compared to those without EE (16.7%).

Debeaked turkeys showed better performance when evaluating the FCR. This result is likely linked to the beak trimming procedure, as it prevents feed selection in the feeders, highlighting the need for beak trimming in commercial turkey farming for better commercial viability, as feeding is the costliest item in intensive production systems (Sethy et al., 2018). Although it is considered an invasive method that causes pain, and can reduce feed efficiency (Prescott & Bonser, 2004), the turkeys used in this study were beak trimmed on the first day of life in the hatchery, a practice that apparently has less impact on feed intake and weight gain just after the procedure (Leighton et al., 1985). This suggests that less stereotyped behaviour and greater adaptation occur when turkeys are debeaked at a young age; therefore, feed consumption is not affected.

The results of this study are in line with the findings of Noble & Nestor (1997), who found that BT turkeys waste less feed than non-debeaked ones and consequently achieve better feed conversion and lower feed consumption. However, when non-debeaked birds had access to EE, they presented a better FCR. The use of EE can promote a high-welfare environment and promote higher performance (Sherwin et al., 1999a).

In the evaluated groups, the results regarding mortality and the need for culling were not significantly different, which was not expected, as various studies have reported that non-debeaked animals have much higher mortality rates than debeaked animals (Grigor et al., 1995; Nicol, 2018). The turkeys were managed and inspected several times a day by highly trained staff. The attentive monitoring of animals, in combination with the use of EE, may have affected the non-significant difference in mortality and reduced the need for culling between the treatments.

However, even with constant care and the use of EE in the pen, injurious pecking was observed in both trimmed and non-trimmed culled turkeys, and it was not possible to detect the triggering factor for these aggressions, reinforcing that multiple factors trigger abnormal behaviours.

Apart from these results, it must be considered that productivity is not always an indicator of animal welfare, but certain parameters such as decreased mortality rates and the percentage of animals affected by injuries and pathologies may indirectly indicate improvements in terms of welfare. On the other hand, when considering high-performance flocks, this equation becomes more complex.

Non-aggressive pecking behaviour was more intense in turkeys with intact beaks, as well as injurious pecking behaviour, despite the presence of EE, which aligns with the findings of the study by Busayi et al. (2006), suggesting that harmful feather pecking in turkeys may result from vigorous investigative pecking. However, little is known about the development and relationships among the multiple factors that influence aggressive pecking behaviour in turkeys.

Injurious pecking is a form of aggression that usually follows a social disturbance and is often used to maintain dominance, involving repeated pecks or plucks at the feathers, usually targeting the feathers on the back, tail, base of the tail, and wings. This sometimes includes the removal and consumption of feathers (Dalton et al., 2013). Feather pecking and cannibalism, similar to all forms of harmful pecking in turkeys, are controlled by multiple influences of the environment, diet, and underlying genetic composition (Martrenchar et al., 2001). Stock density, ambience, lighting, diet, and genetic composition variables were the same between the treatments and were not considered in this study. The stocking density used in this study was lower than that used in commercial establishments, yet it was still not sufficient to prevent fights, which is most likely related to social competition.

Environmental enrichment has been suggested to reduce stress and feather pecking, thus leading to an increase in overall flock health (Lindenwald et al., 2021). In this study, EE only altered exploratory behaviour, increasing it when present. Although they are considered strategically important, EE objects should not be seen as the final solution to turkey welfare problems, and if they are not combined with sanitary and nutritional parameters, their benefits may not be observed. Therefore, it is necessary to consider the type of material used, as well as its method of availability, to avoid possible frustration and lead to the belief that EE is not a useful tool for promoting turkey welfare.

Debeaking of commercially raised animals resulted in fewer injuries in the turkeys observed in this study. These results are in line with those of Grigor et al. (1995), who found that all debeaking techniques resulted in significant reduction in injuries due to cannibalism and feather pecking. The adverse effects of beak trimming were minor and short-lived, and were clearly outweighed by the benefits of reducing injuries, as the treatments where the animals underwent BT showed fewer injuries and better FCR results.

The insertion of EE in turkey pens was effective when compared to the injuries and lesions found in the carcasses of turkeys at slaughter with those that did not have access to the tool, confirming the findings of Dalton et al. (2018), which suggests that both feather pecking and cannibalism occur due to the lack of environmental stimuli in sterile housing and an inherent tendency to peck. Furthermore, although the severity of the lesions was not evaluated, as there was no significant difference in mortality and the need for culling between treatments, it can be assumed that the lesions caused by aggressive pecking were milder and did not provide entry points for other opportunistic agents, although the turkeys still presented with pecking injuries. High rates of carcass condemnation in groups of animals that did not undergo the BT process is common (Glatz & Rodda, 2013), the same result found in the current study.

The area most affected by pecking injuries in this study was the wings, which can be explained by the appearance of ultraviolet (UV) markings on the feathers that are temporally and spatially linked to the initial targets of injuries. Most commercial plants operate under fluorescent or incandescent lighting with minimal UV; therefore, the distorted appearance of emerging feathers in young turkeys may have made them targets for investigative pecking (Dalton et al., 2013; Duggan et al., 2014). Sherwin & Devereux (1999) suggested that the 'unnatural' appearance of these markings under conventional lighting, which emits minimal UV radiation, may attract or prolong the harmful pecking.

Several species of poultry, including domestic turkeys, have the ability for UV vision. This can function as a signal among turkeys, for example, in individual recognition or even in hierarchy organization, which would suggest the presence of visible feather markings under UV radiation. When feathers were viewed under conventional white fluorescent lighting (which emits minimal UV light), the turkeys presented a uniform yellow or white colour according to the stage of feather emergence. However, when viewed under a lamp that emits radiation with peaks in the UV spectrum, distinct fluorescent and non-fluorescent spots were observed in various parts of the body, including wings, tail, thighs, neck, chest, and dorsal surface (Sherwin & Devereux, 1999).

Bartels et al. (2017) observed that in turkey housing systems, providing light covering the long-wave UV spectrum together with EE can help reduce the prevalence of harmful pecking. This aligns with the findings of this study, since the turkeys were under incandescent lighting, which may generate image distortion in turkeys and consequently curiosity, leading an investigative peck to become more aggressive, causing injury to the animals.

The infrared irradiation debeaking procedure proved to be more advantageous than not debeaking the turkeys, resulting in a better feed conversion ratio, less feed consumption, and fewer injuries from injurious pecking, as EE was not effective on non-debeaked turkeys.

Adding only reusable plastic bottles with color liquid as an environmental enrichment tool did not provide the expected result as they caused more injuries to the carcasses, perhaps because it was not the appropriate equipment to be used as the only form of EE and avoiding the beak trimming process. To developed practical solutions, it is suggested that more research must be conducted to clearly understand the development of injurious pecking behaviour. The evaluation of different objects of EE, different breeds of turkeys, and stocking density should be carried out to clarify doubts about the benefits of EE and its impact on both performance results and welfare, as practices involving physical changes should be avoided and even banned in some countries. Providing enriched facilities is recommended, even when birds are beak-trimmed.