DEPARTMENT for ENVIRONMENT, FOOD and RURAL AFFAIRS Research and Development CSG 15 Final Project Report (Not to be used for LINK projects) Two hard copies of this form should be returned to: Research Policy and International Division, Final Reports Unit DEFRA, Area 301 Cromwell House, Dean Stanley Street, London, SW1P 3JH. An electronic version should be e-mailed to [email protected] Project title Stocking density and welfare in broilers DEFRA project code AW0219 Contractor organisation and location Department of Zoology University of Oxford South Parks Road Oxford OX1 3PS Total DEFRA project costs Project start date £ 249,435 01/04/00 Project end date 31/03/03 Executive summary (maximum 2 sides A4) Background While there is major public concern over the welfare of broiler chickens, particularly at high stocking densities, up until now there has been no unequivocal and universally accepted evidence that reducing stocking density would significantly increase bird welfare. This study examined the effect of experimentally manipulating stocking density on broiler welfare, including health, hormonal and behavioural measures at five different maximum stocking densities: 30,34,38,42 and 46 kg/m2. The study was carried out entirely on commercial farms, with the full cooperation of the broiler industry. It covers a wide range of broiler housing, from small older houses to large modern ones and involved whole-house experimental manipulation of stocking density. Objectives 1. To document the effect of stocking density on the behaviour, physical health and productivity of broilers within the range 30-46 kg/m2 by experimental manipulation of stocking density on commercial farms 2. To document the effect of season (temperature) on stocking density by conducting the same experimental manipulation of stocking density in summer and again in winter. 3. To identify commercially relevant factors that modify the effect of stocking density by conducting the same experimental manipulation on farms belonging to thirteen different commercial broiler companies at present using a wide range of management and husbandry systems. CSG 15 (Rev. 6/02) 1 Project title Stocking density and welfare in broilers DEFRA project code AW0219 Methods A total of 114 chicken houses on 20 farms involving 2.7 million chickens were studied between August 2000 and November 2002. Two trials were conducted in Northern Ireland, one in Denmark, eight in England and one in Scotland. A wide range of housing systems (different ages and sizes of house) were included in the study. Stocking density was manipulated by altering the number of day old chicks placed in each house, with maximum final stocking density predicted from average company growth rates and mortality against a company-specific required bird weight at a particular age (normally 37/38 days). The ‘target’ stocking density was the maximum stocking density that would be experienced by the birds during their lives. This was either just before clearance or, where thinning occurred, just before thinning. Each company was asked to set aside two whole houses for each of five stocking densities but otherwise not to change its normal practice. A large number of management and husbandry variables as well as environmental correlates were then measured. Bird welfare was assessed through a variety of measures including mortality, downgrades, gait, leg health, faecal corticosterone levels and behaviour. Results and conclusions 1. Experimental manipulation of stocking density of broiler chickens between 30kg/m2 and 46 kg/m2 resulted in a depression of growth rate, an increase in the incidence of birds jostling each other and, at the two highest stocking densities, a decrease in birds with the best gaits. Some other behaviours such as lying stretched out and wing stretching were also affected by stocking density. 2. The effects of stocking density were, however, overshadowed by much larger differences between companies, suggesting that husbandry and management factors are, within limits, more important than stocking density per se. 3. There were differences between birds reared in summer and those reared in winter. In winter, there were higher levels of mortality and birds showed higher levels of faecal corticosterone and a higher incidence of pad dermatitis and dirty pads. 4. Husbandry and management factors that appear to be important include those that affect the litter moisture levels and temperature and humidity during the birds’ growth. These in turn can be related to drinker type, number of drinkers/bird and season of the year. ‘Good management’, however, consists of more than just controlling temperature and humidity since there are many other factors that have significant effects such as the number of stockmen and the number of visits they make each day. 5. The differences between companies in how well they ‘coped’ with higher stocking densities shows the importance of carrying out commercial scale trials with more than one company and not relying on small scale trials that may give misleading results. 6. Legislation to improve broiler welfare should not be directed solely to limiting stocking density as this will detract attention from other more important factors such as improving the birds’ environment and their genetics. 7. Reliance on single measures of welfare such as foot pad dermatitis can be misleading and a variety of different measures are needed to assess welfare, Including physical health and behaviour. CSG 15 (Rev. 6/02) 2 Stocking density and welfare in broilers Project title DEFRA project code AW0219 Scientific report (maximum 20 sides A4) Introduction Worldwide, 20 billion meat chickens (broilers) are killed every year1, over 900 million in the UK alone and 2.2 billion in the US, where almost 90% of animals eaten are chickens. Commercial pressures to produce cheap meat have led to intensification of broiler production, commonly involving large, environment-controlled houses containing 5,000-50,000 birds and selection for rapid growth so that birds reach a slaughter weight of 2-3 kg at 39-42 days of age. There is widespread concern about the welfare of these birds2-4 , particularly their lack of space, but only in Sweden and Switzerland is there any specific legislation about how much space each bird should be given1. In the rest of the world, there are no legal limits on how crowded birds can be and there are now calls for legislation to limit bird density in the UK and elsewhere1,4. Such legislation would have major implications for the competitiveness and viability of broiler production in countries adopting it unilaterally. The key question for policy makers is, therefore, whether giving each chicken more space would genuinely improve welfare. If welfare benefits are demonstrable, then the case for legislation to regulate space allowances would be clear even in the face of economic costs. But if the welfare benefits cannot be demonstrated, then imposing an arbitrary upper limit on stocking density will hinder genuine improvements in bird welfare by deflecting attention from more effective measures, such as improving environment, management, nutrition or genetics. Animal welfare includes both physical health and mental well-being5,6 but current evidence is contradictory 1,7-9 and does not answer even the basic question of whether the physical health of broilers is improved by giving them more space. Inconsistencies arise because many studies are conducted on a small scale and so do not accurately reflect the conditions of large commercial broiler houses9-11 , are concerned with only a limited range of conditions in which broilers are kept12, measure only limited aspects of welfare such as foot condition13 or use correlation rather than experimental manipulation 12,13. We here describe a study that for the first time supplies policy makers with the standard of evidence necessary for rational decisions about stocking density and welfare of broiler chickens. The study involved 2.7 million chickens belonging to 10 major broiler producers in the UK and Denmark, and included both large modern houses and smaller older ones. In an unprecedented experimental collaboration, each company stocked whole houses of broilers to five different maximum final stocking densities: 30, 34, 38, 42 and 46 kgm2, with at least two houses at each stocking density, a total of 114 houses were included in the study. By recording a range of welfare measures, plus a large number of environmental and management variables, we provide an evidence base for policy makers seeking to improve bird welfare. Objectives The scientific objectives of this project were: 1) To document the effect of stocking density on the behaviour, physical health and productivity of broilers within the range 30-46 kg/m2 by experimental manipulation of stocking density on commercial farms 2) To document the effect of season (temperature) on stocking density by conducting the same experimental manipulation of stocking density in summer and again in winter. 3) To identify commercially relevant factors that modify the effect of stocking density by conducting the same experimental manipulation on farms belonging to thirteen different commercial broiler companies at present using a wide range of management and husbandry systems. Methods Companies: Initially, there were 13 broiler producing companies participating in this project (Objective 3) but by the time the contract had been signed, one (Marshall’s) had been taken over by Grampian and soon after, Webbs were taken over by Faccendas and Skovsgaard were taken over by Rose Poultry. Subsequently, Premier Poultry became part of the 2 Sisters group. This left 9 independent companies (which now represented an even greater proportion of the industry than before) but both 2 Sisters and Faccendas agreed that we could do trials on farms that had been part of their original companies as well as those that were part of the company just taken over, giving us 11 ‘companies’. These were: Dove Valley, Banhams, Padleys, Moy Park, O’Kane’s, Rose CSG 15 (Rev. 6/02) 3 Project title Stocking density and welfare in broilers DEFRA project code AW0219 Poultry (Denmark), Grampian, 2 Sisters (previously Premier), 2 Sisters (previously not Premier), Faccendas (always Faccendas) and Faccendas (previously Webbs). An outbreak of disease in the Faccenda (previously Webbs) trial prevented us from carrying out this trial but as by this time the previously Webbs farms had been fully integrated into the Faccenda management system, there was no substantive difference between a previously Webbs farm and a farm that had always been part of Faccendas, so that this trial would not have contributed greatly to the ‘between company’ effects reported below. In addition, we carried out two ‘seasonal repeat’ trials – that is, we carried out the experimental manipulation of stocking density with the same stocking densities in the same 10 houses once in winter and once in summer (Objective 2). These seasonal repeat trials were carried out on a 2 Sisters (previously Premier farm) and a Faccenda (always Faccenda) farm. Treatments and replication: Ten major broiler producing companies were each asked to stock whole houses of broilers to five different target maximum stocking densities: 30, 34, 38, 42, and 46 kg/m2, and two houses at each stocking density. Seasonal repeats were conducted with two companies. In total, 114 chicken houses on 20 farms involving 2.7 million chickens were studied between August 2000 and November 2002. Two trials were conducted in Northern Ireland, one in Denmark, eight in England and one in Scotland. The range in densities covered the Freedom Food recommendation15 of 30 kg/m2, the Farm Animal Welfare Council recommendation of 34 kg/m2 and current practice in the UK and EU1, which can be as high as 42.5kg/m2. At the insistence of MAFF (now Defra), the higher stocking density of 46kg/m 2 was included in the study, but as this was beyond normal commercial practice in the UK, companies were given the opportunity for not including it; three companies declined to include this density on welfare grounds. Target stocking density, which was the maximum density experienced by the birds either just before clearance or where thinning occurred, just before thinning, was manipulated by altering the number of day old chicks placed in each house, predicted from average company growth rates and mortality against a company-specific required bird weight at a particular age (usually 37/38 days). Houses within a company were randomly assigned to a given stocking density (one density per house) before the eggs were incubated, to ensure the right number of chicks for a given size house. With five companies, all 10 houses (and therefore stocking densities and replicates) were on one farm, with three companies the trial was split between two sites (with one replicate of each stocking density at each site), and with two companies the houses were split between three and six farms. On the whole, houses used within a company were similar where as houses between companies were diverse. Manipulating stocking density was the only change we asked the companies to make; normal procedures were conducted in all other respects (e.g. sexing chicks or using as hatched, light schedule, feed etc.), and were recorded in detail. Management and husbandry variables: We recorded: company assessment of performance of house, age of house, ownership (company or contract grower), size, orientation, fabric, light pattern, light source, dawn dusk dimming, percent wheat in the finishing ration, feeder type, number of feeders, drinker type, number of drinkers, heater type and number and position in house, type of ventilation system and control, use of misting systems, floor and litter type, number of stockmen and number of visits per house, vaccination programme, feed withdrawal and thinning/clearance programme. Environmental variables: To obtain continuous information about the bird’s environment during their lives, we fitted four randomly (see Appendix 1) positioned Tiny Talk Data loggers in each house at a height of 60cm, recording temperature and humidity every hour. In addition, when target stocking density had been reached, we recorded levels of atmospheric ammonia (Gastec pumpset GV-100s), and light at bird height (ISO-Tech digital light meter), and litter moisture content ({weight difference in sample after drying/sample weight}*100) CSG 15 (1/00) 4 Stocking density and welfare in broilers Project title DEFRA project code AW0219 Animals: Chicks arrived on farm as day-olds from whatever source the company normally used. We recorded: source, date, time, position of modules on lorry (front, middle, back), number of trays per house, number of chicks per tray, type of vehicle (rigid or artic), ventilation (fans, vents, and position on container) and on board temperature at time of arrival. We counted the numbers of chicks in at least five trays as a check and estimated the number of birds placed in each house. Information on breed, numbers per breed, sex (as hatched, male, female), and age of parent flock per house were provided by the company. Leg Health: Gait score: A single bird was chosen at random from each of 10 grid points (see Appendix 1) in each house. The bird was chosen by viewing the grid through a clear acetate sheet with a pre-chosen point. By approaching the bird slowly and in a circular path, the bird was observed to walk unhindered by other birds, for at least 10 paces and the gait recorded as follows: 0 1 2 birds walks with ease, is well balanced and has regular and even strides bird walks with irregular and uneven strides and appears unbalanced bird is reluctant to move and unable to walk many strides before sitting down Ten birds were recorded per house, so that in total 1140 birds were recorded for gait. Hockburn, pad dermatitis and leg deviation: Lights were dimmed in the house, so that the birds would not move as researchers moved through the house. Groups of 10 birds were caught in a triangular frame at four previously chosen random grid points (see Appendix 1) per house for further visual inspection of leg health (40 birds per house). Individuals were lifted and supported under the breast during inspection of the hocks and pads. They were then inverted (ventral side facing handler) and held by the legs with the handler’s thumbs just below the inter-tarsal joint, pointing upwards with no rotation. Healthy legs are parallel, with foot pads facing directly away from the handler and the middle toes pointing straight up. Deviations were scored as shown in Table 1; the bird was then weighed (Welltech International with sling) and released. In total 4370 birds were weighed and their legs inspected. Table 1. Scoring system for hockburn, pad dermatitis and leg deviations. Foot pads** Angle – in 0 No discoloration or lesions No lesions Legs straight Angle – out Legs straight Hocks* Rotation Toes Legs straight, pads facing away from handler Straight toes 1 <10% hock with lesion <5mm lesion on pad Inward bow at intertarsal joint so that the two legs meet >220 Outward twist at inter-tarsal joint with 300 or more between the legs Rotation of the tibia shaft so that pads face each other >150 One or more toes twisted in the medial or lateral direction * Pink hocks in score 0 were also recorded ** Dirty and swollen pads were also recorded CSG 15 (1/00) 5 2 >10% hock with lesion >5mm lesion on pad Project title Stocking density and welfare in broilers DEFRA project code AW0219 Corticosteroid levels and disease status: The first fresh faecal sample observed in each of five random grids (see Appendix 1) was collected and uniquely labelled. Swabs were taken and sent to the laboratory for the following tests: Salmonella, E.coli, Campylobacter, Clostridia and Coccidia. Faecal samples were dried at 400C and sent to the laboratory for corticosteroid analysis14.15,16. Behaviour: Prior to chick placement we installed four cameras and their operating batteries (supplied by Tracksys Ltd Nottingham, UK) in each house at a height of 155cm according to the randomised grid (see Appendix 1). Each camera battery had a unique switch code and sent a radio signal to a receiver linked to a VCR and switch gear positioned in the ante room of the house. Adjacent houses operated at different radio signals to prevent interference, and the view from cameras at the same location in each house was standardised as much as possible. The cameras were switched on close to maximum stocking density and video records of the birds were made between 10.00am and 12 noon. In total 80 minutes of video was recorded per house in eight 10 minute sequential switches between each camera for two switching cycles. The birds were not disturbed by the stockman or researchers during the recording period. For analysis, one focal bird was chosen at random from each camera and each switch cycle (eight birds per house), via a transparency over the monitor with a predetermined point; the following behaviours were recorded for five minutes: the frequency and duration of stand, lie, feed, drink, preen, rest (eyes closed), lie stretched out; the frequency of walk (including the number of strides), run (including the number of strides), peck litter, peck other bird, scratch litter, scratch head, stretch head, stretch wing, stretch leg, shake body, shake head, dust bathe, wing flap, aggressive interactions, and perch. We also recorded the number of times the bird changed posture (up/down), jostled by or through other birds, was disturbed, was facilitated in its behaviour, and walked-on other birds, or was walked-on by another bird. In total records were obtained from 107 houses and 741 birds. Production data: From audited company records: we analysed: mortality (numbers of birds found dead plus the numbers of birds culled due to illness or leg problems), feed conversion ratio, water intake, date, numbers and weights of birds removed from the house (thinned or cleared), and information on birds subsequently rejected at the processing plant. Bird weight, growth rate and calculated stocking density: Individual bird weight was taken at the time of leg inspection and growth rate subsequently calculated ({individual weight – average chick weight}/number of days). Stocking density at the time of bird weighinga (target) and immediately prior to thinning or clearanceb (maximum) was calculated ({average weight at a or b * number of birds in house at a or b}/area of house). All measures were later correlated to hock burn, pad dermatitis and leg deviation. Post-mortem analysis: On thinning or clearance, we followed the birds to the processing plant and collected 10 birds per house off the processing line immediately pre-evisceration, at randomly predetermined times. The birds were chilled and sent to the laboratory for external, abdominal and thoracic examination, as well as for bacteriological and parasitological analysis (as for the faecal samples previous). Birds were not sent from three companies and in total 807 birds were analysed. Standard Protocol: Our measurements outlined above were taken as close to target stocking density and hence thinning or clearance, as possible. Since the decision to remove birds from each house was determined by each company, often with little notice, a judgement of when to take our measures was taken to ensure no loss of data; this sometimes resulted in our measures being taken a few days prior to clearance/thinning. CSG 15 (1/00) 6 Project title Stocking density and welfare in broilers DEFRA project code AW0219 A strict protocol was followed on farm, to ensure minimum disturbance to the birds and so allow for collection of base line corticosteroid levels and natural behaviour for example. The protocol was as follows: Sample 1 (between 2pm and 5pm, day 1). Enter house and slowly move to the sample grids collecting gait score, atmospheric ammonia and light data, and collecting litter and faecal samples. Video records (10am – 12pm day 2). Enter house and switch on batteries. Allow 15 minutes for birds to settle before commencing recording. Sample 2 (afternoon day 2). Dim lights and enter house moving to the sample grids. Catch birds in frame and inspect legs and weigh. Statistical analysis Initially variables were analysed for the effects of target density and company by analysis of variance (general linear model). Where target effects were shown, the effects of actual stocking density was tested by regression analysis (fitted line model). For the behavioural variables, angular transformation was applied to proportional data, and the numbers of incidents of behaviour were square root transformed. Later, univariate linear correlations were examined between outcome variables and predictors that could be treated as continuous variables. Multivariate linear models were constructed using a stepwise model selection procedure (starting from a model with no predictors) with possible predictors including those continuous predictors with substantial linear correlations (<-0.2 or >0.2) as well as categorical predictors. Results Management and husbandry: A wide range of housing was used in the study. On average houses were 19 years old (5-40 years) and 1144m2 (455-1901m2). Resource was divided almost 50:50 between pan and chain feeders, nipple and nipple with cup drinkers, whilst kerosene powered space heaters were most common (present in 82 percent of all houses). The environment was mostly controlled automatically (82 percent), and litter was in the form of woodshavings, straw or a woodshave/straw mix in 49, 33 and 17 percent of houses respectively. Full details of management data are given in Appendix 2. Animals: Chick numbers placed for each target stocking density is given in Table 2 whilst details of the chicks placed are given in Appendix 3. Most flocks were Ross breed (75 percent) from an average 2.7 parent flocks at 35.3 weeks of age. Audits of numbers into and out of the house (mortality and at processing) were made where possible. Average potential discrepancies of chicks in to the house were 6 birds (n=56 SE31.2 range -639 to +634) and out of the house -24 (n=103 SE 30.5 range -1864 to 672). Table 2. Chicks placed per target stocking density (one density per house) Target SD Mean Minimum Maximum 30 18,939 7,500 35,922 34 21,491 8,550 42,500 38 24,130 9,600 45,200 42 26,235 10,500 48,300 46 29,615 11,550 53,051 Environmental variables: Details of mean temperature and humidity and the percent of time these variables were out of range (according to the Ross broiler manual), as well as the percent variation explained by company, week and other factors, are given in Appendix 4. Company differences were large and there were some target density effects on humidity variables. Average weekly temperature fell from 29.6oC in week 1 to 20.3oC in week 6, and was not affected by target stocking density. Approximately 85 percent of the variation was explained by week, company, drinker type, CSG 15 (1/00) 7 Project title Stocking density and welfare in broilers DEFRA project code AW0219 season and drinker ratio. The percent time temperature was out of range varied from an average 26.9 to 71.6 across the weeks, and 49.6 percent of the variation was explained by week, company, control, number of stockmen and season. Average humidity increased through out the growth cycle from 49.9 to 71.5 percent, and was significantly affected by target stocking density in weeks 1, 3, and 5. Seventy one percent of the variation was explained by week, company and drinker ratio. The percent of time humidity was out of range varied from 20.6 to 62 percent across the growth cycle and 29.9 percent of the variation was explained by week and company. The number of visits made to the house, litter type and age of house helped to explain some of the variation between companies. Average moisture (given in Table 3 and detailed in Appendices 5 and 7) was significantly affected by target density, so that levels were greater in densities of 38 and 42kg/m2 (19.75 percent compared to ~17.7 percent). Fifty six percent of the variation was explained by company and covariate effects (drinker ratio, position of heaters, control, and calculated stocking density). Average ammonia levels were not affected by target density and averaged 10.4 ppm (Table 3). Table 3 presents average figures for the measures taken (except behaviour) along with the uni-variate analysis for the effects of target stocking density, company and calculated stocking density. Effects of stocking density were few (detailed below and further in Appendix 5), whilst effects of company were plentiful and large. Correlations between variables and details of the explanatory statistics are given in Appendices 6 and 7 respectively. Leg health: Gait score: Across all densities and on average, approximately three quarters of the birds examined were awarded a gait score zero. Gait zero was significantly affected by target stocking density, actual stocking density and company, and fell from 81 percent of examined birds in target density 30kg/m2 to 68 and 61 percent in target densities 42 and 46kg/m2, respectively. It also fell linearly with increasing actual density (slope -1.27, R2=11.7% p<0.0001). Graphs of gait deviation (scores 1+2) are given by target density and company in Figures 1 and 2 below. Gaits attaining score 2 which were considered a welfare problem were on average less than one percent of examined birds, and was not affected by density or company. Figure 1. Average gait deviation (with 95 percent confidence levels) by target stocking density Percentage gait deviation 60 Percent incidence 50 40 38.9 30 32.5 26.4 20 23.9 19.2 10 0 30 34 38 Target stocking density CSG 15 (1/00) 8 42 46 Project title Stocking density and welfare in broilers DEFRA project code AW0219 Table 3. The effect of experimental manipulation of stocking density (target), actual achieved density, and differences between companies, on leg health, faecal corticosteroid, important production parameters, litter moisture and ammonia levels. Mean SE Range Company 10-100 Target density p=0.027 p<0.0001 Actual density p=0.002 Gait 0 % 72.6 2.0 Gait 2 % 0.9 0.3 0-20 ns ns ns Hock 0 % 80.6 1.7 20-100 ns p<0.0001 ns Hock 2 % 1.5 0.5 0-47.5 ns p=0.0003 ns Pad 0 % 81.2 2.1 12.5-100 ns p<0.0001 ns Pad 2 % 2.8 0.7 0-47.5 ns ns ns Angle-in % 12.3 0.7 0-35 ns p<0.0028 ns Angle-out % 3.5 0.4 0-17.5 ns p<0.0001 ns Rotation % 6.9 0.7 0-47.5 ns p<0.0001 ns Crooked toe % 0.6 0.1 0-7.5 ns p<0.0001 ns Total leg deviation % Corticosterone ng/ml Calculated growth rate (g/d) Total mortality % Leg cull % Other cull % Dead birds % Downgrades % Litter moisture % Ammonia levels ppm 23.3 1.2 0-62.5 ns p<0.0001 ns 18.9 1.1 3.5 – 50.4 ns p<0.0001 ns 49.4 0.4 39.1–57.9 p=0.011 p<0.0001 ns 4.1 0.2 1.4 – 14.7 ns p<0.0001 ns 0.6 0.1 0 - 2.4 ns p<0.0001 ns 1.5 0.1 0.4 – 4.7 ns p<0.0001 ns 2.0 0.1 0.6 – 4.8 ns p<0.0001 ns 0.9 0.1 0 – 4.24 ns p<0.0001 ns 18.3 0.4 10.7–32.5 p=0.05 p<0.0001 ns 10.4 0.7 1.3 – 29.8 ns p<0.0001 ns CSG 15 (1/00) 9 Project title Stocking density and welfare in broilers DEFRA project code AW0219 Figure 2. Average gait deviation by target density and company. Gait deviation by trial 100 A percent G1+ 90 B 80 C 70 D 60 E F 50 G 40 H 30 I 20 J 10 K L 0 30 34 38 42 46 Target stocking density Gait 0 was negatively correlated to a wide range of variables. Interestingly ‘normal’ gait fell with increasing light levels, more birds in the house, more visits by the stockman, and the more time temperature was out of range in weeks 1 and 4. It also decreased as the incidence of rotation, pad dermatitis 2 and jostling increased. It was positively correlated to hockburn 0 and average temperature in week 1. Fifty two percent of the variation seen in gait score 0 was explained by company and covariate effects. Hockburn, pad dermatitis, and leg deviation: Zero scores for hockburn and pad dermatitis were on average awarded to 81 percent for all birds examined. There was no effect of target or actual density, but a significant effect of company on both measures. Hocks and pads attaining score 2 were low and averaged 1.5 and 2.8 percent respectively. Neither were affected by target density, but company significantly affected hock score 2. On average 25 percent of birds examined had dirty pads (SE 2.57 range 0-95 percent), and 11.5 percent had swollen pads (SE 1.89 range 0-90 percent). Almost a quarter of all birds examined had some form of leg deviation. The most common deviation was angle-in, representing 12 percent of birds examined, rotation was intermediate at 7 percent, whilst angle-out and twisted toes represented 3.5 and 0.6 percent respectively. There was a significant effect of company but no effect of stocking density on leg variables. Corticosteroids and faecal bacteriology and parasitology: Average faecal corticosteroid levels were 18.9 ng/ml and were not affected by target or actual stocking density, but were affected by company. There was no effect of density on disease status of the birds. E.coli and Clostridia were isolated from 13.6 and 10.7 percent of faecal swab samples whilst Salmonella and Campylobacter were isolated in 2.1 and 1.9 percent of samples respectively; no coccidian were isolated. Average faecal corticosteroid was negatively correlated to many temperature and relative humidity readings, and average levels of ammonia and litter moisture. It was strongly positively correlated to mortality levels and percent time temperature was out of range in week 1. Approximately 84 percent of the variation in faecal corticosteroid levels was explained by company and covariate effects. Behaviour: On average birds spent 72 percent of the observation period lying down and just less than half of that time was spent resting (eyes closed head down). The proportion of the observation period spent in various activities is CSG 15 (1/00) 10 Project title Stocking density and welfare in broilers DEFRA project code AW0219 shown in Figure 3. One fifth of the time was spent feeding, drinking and preening, whilst 45 percent of the observation period was spent alert; during this time birds were performing small body movements (stretching head, shaking body etc.), walking, or remaining stationary but watchful. There was no effect of density on the proportion of time spent in these various activities, the length of the activity bout or the number of behaviour incidents per minute. Detailed results of the behaviour analysis are given in Appendix 8. Figure 3. Percentage time observed birds spent feeding, drinking, resting, lying fully stretched out, and alert. Bird activity (percent time) 8.6 7.2 Feed Drink Rest 45.5 Preen LSO Alert 34 0.5 4.2 Apart from jostle and stretch leg, there was no effect of density the incident rate of behaviours. On average birds changed position 0.54 times per minute, pecked at the litter 0.81 times per minute, and stretched head, shook head and dust-bathed 0.12, 0.08 and 0.001 times per minute respectively. The incidence of birds jostling increased from 0.32 per minute at 30 kg/m2 to 0.57 and 0.62 per minute at 42 and 46kg/m2, respectively and is shown in Figure 4. Jostle rate was also significantly affected by actual stocking density and increased with increasing density (Jostle = -0.3613 + 0.0232 density, R2=19.2% F=22.2 p<0.0001). In addition to jostling, birds were also disturbed in their behaviour at an average rate of 0.16 per minute, which was not affected by density. Jostle rate was negatively correlated to gait score 0 and leg cull percentage, and positively correlated to pink hocks (or worse), rotation and angle-out leg deviations, total numbers of birds in the house, various temperature and humidity readings and litter moisture. In total 35.2 percent of the variation in jostle rate was explained by company and co-variate effects. On average birds performed a walking bout 0.43 times per minute which was not affected by target density. The number of strides taken per walking bout was affected by target density, however, and fell from an average 5 strides at 30 and 34kg/m2 to 3.6 strides at 46kg/m2. The rate of leg stretching decreased from 0.056 incidents per minute at 30kg/m2 to 0.028 at 42kg/m2. The following incidents of behaviours were also affected by density in the mixed general linear models: drink (target density p=0.034), lie stretched out (maximum density at thinning p=0.03), shake head (actual density p=0.007), and stretch wing (actual density p=0.038). CSG 15 (1/00) 11 Project title Stocking density and welfare in broilers DEFRA project code AW0219 Figure 4. Average jostle rate of focal birds by target stocking density (with 95 percent confidence bars) Jostle rate 0.9 0.8 Incidenst/minute 0.7 0.6 0.618 0.5 0.5655 0.4 0.4553 0.4314 0.3 0.2 0.3163 0.1 0 1 2 3 4 5 Target stocking density Production data: Growth rate: Growth rate was significantly affected by target stocking density, so that birds grew faster at densities of 30 kg/m2 than at 46kg/m2(50.3g/d over the whole growth cycle compared to 47.7g/d, respectively). Growth rates by target density and company are given in Figures 5 and 6 respectively. On average birds were 1.784kg (SE 0.014 range 1.49-2.27) at 35 days (range 32-40) when weighed prior to thinning or clearance. Growth rate was negatively correlated with humidity in week 1, dirty pads and litter moisture, and positively correlated with flock age, hock and pad scores zero, and the number of visits made per day by the stockman. In total 65.8 percent of the variation see in growth rate was observed by company and co-variate effects. Figure 5. The effect of target stocking density on average growth rate over the crop cycle (with 95 percent confidence bars). Average growth rate 53 52 51 g/d 50 49 50.26 49.9 49.7 48.9 48 47.8 47 46 45 30 34 38 Taregt stocking density CSG 15 (1/00) 12 42 46 Project title Stocking density and welfare in broilers DEFRA project code AW0219 Figure 6. Average growth rate over the crop cycle by company and target stocking density. Average growth rate 60 55 g/d 50 45 40 35 30 30 34 38 42 46 Taregt stocking density Stocking density: Stocking density was calculated at the time of bird weighing and immediately prior to the first birds leaving the house for processing, and is shown in Table 4 below. Both were highly correlated to target density (0.77 and 0.81 p<0.0001, respectively). Table 4. Actual and maximum densities in kg/m2 (calculated at weighing/leg health inspection and immediately prior to birds leaving the house, respectively) by target density. 30 28.42 (0.54) 34 32.32 (0.52) 38 35.39 (0.83) 42 38.47 (0.75) 46 40.40 (1.4) 24.6 – 32.4 32.2 (0.97) 26.5 – 35.9 35.69 (0.81) 22.2 – 39.8 37.65 (0.57) 32.0 – 43.4 41.75 (0.48) 24.6 – 46.7 44.83 (0.58) Range 26.4 – 42.3 (standard error) 31.5 – 49.8 32.3 – 43.6 37.5 – 45.5 39.7 – 49.1 Actual density (at weigh) Range Maximum density Mortality: Total mortality across all densities averaged 4.1 percent (range 1.4 to 14.7 percent) and was not affected by target density, shown in Figure 7, or actual density; mortality was affected by company and is shown in Figure 8. For some companies mortality was highest at the higher densities, whereas with others the highest mortality occurred at the lower densities or was not affected by density. Causes of mortality were mostly due to birds found dead (average 2 percent of all birds) or culled for reasons other than leg problems (average 1.5 percent of all birds); overall leg culls were small and represented on average 0.6 percent of all birds. CSG 15 (1/00) 13 Project title Stocking density and welfare in broilers DEFRA project code AW0219 Figure 7. Average bird mortality by target stocking density Average total mortality 7 6 Percent 5 4 3 4.16 3.69 30 34 4.45 4.78 42 46 3.79 2 1 0 38 Target stocking density Figure 8. Bird mortality by company and target density Total mortality Percent 10 9 A 8 B 7 C D 6 E 5 F 4 G 3 H 2 I 1 J K 0 30 34 38 42 46 L Target stocking density Mortality was negatively correlated with ammonia levels, actual stocking density, age of house, and pink hocks or worse percent. It was positively correlated to relative humidity deviation in weeks 3, 4 and 5, and temperature deviation in week 5, as well as number of birds placed, corticosterone levels, and the number of stockmen on the unit. Forty four percent of the variation in mortality was explained by company and co-variate effects. Downgrades at processing plant: Less than 1 percent of birds (range 0-4.2 percent; 17.5K out of 1.89million birds) were downgraded or rejected at the processing plant. Stocking density did not affect the rate of downgrades, where as company did. The reject categories recorded differed between companies making comparisons between categories difficult. However, where categories were recorded 11.2, 14.2, 12.6, and 13.6 percent of rejects were due to ascities, E.coli, Septicaemia, and skin lesions respectively, where as 3, 0.8, and 1.2 percent were attributed to Staphylococci infection, emaciation, and abnormal colouring, respectively. ‘Others’ accounted for 43 percent of rejects. CSG 15 (1/00) 14 Project title Stocking density and welfare in broilers DEFRA project code AW0219 Post-mortem analysis: There was no effect of stocking density on the post mortem findings of the birds sent to the laboratory. Twenty three percent of the birds had hock burn, whilst the incidence of breast blisters, abdominal cavity (what) and thoracic cavity (what) were low at 1.5, 0.7 and 0.1 percent, respectively. There was no effect of density on disease status of the birds. E.coli was most prevalent and was isolated from 29 percent of post mortem birds. Campylobacter, Clostridia, and Salmonella were isolated from 7.6, 8.2 and 0.6 percent of birds. A single egg of coccidian was isolated. Seasonal effects: Detailed results for the seasonal trials with two companies are given in Appendix 9. There was a significant effect of season on pad dermatitis score 0, dirty pads, rotation, total leg deviation, faecal corticosterone levels, leg culls, dead birds and total mortality. Fewer birds attained pad score 0 in winter than summer (71.6 compared to 88.9 percent respectively) and more birds (30.3 percent compared to 8.8 percent) had dirty pads in winter. Fewer birds had rotational leg deviation in winter than summer (3.3 versus 5.8 percent respectively) and overall leg deviation was lower in winter (14.9 compared to 19.6 percent). Leg cull was higher in winter than summer (1.3 versus 0.9 percent) and birds found dead were also higher in winter (2.2 compared to 1.7 percent); total mortality was therefore higher in winter (5.3 versus 4.2 percent). Corticosterone levels were higher in winter (36.2ng/ml) than in summer (8.6ng/ml). Mortality data from 2002 relating to the farms and houses used in the stocking density trial was analysed for a further two companies, to increase the numbers of seasonal comparisons outside the main database, and is given in detail in Appendix 10. First company: Mortality was significantly affected by farm (F=8.01 p=0.006) and crop (F=8.98 p<0.0001). Least mortality occurred in crops placed in March, May, August and October (average range 2.77-3.89 percent); most mortality occurred in crops placed in December, February, and June. Second company: Total mortality was significantly affected by crop (F=12.7 p<0.001) and parent flock age (F=22.6 p<0.001); cull mortality was also affected by crop (F= 19.8 p<0.001) and parent flock age (F=8.3 p<0.001), and birds found dead was significantly affected by crop (F=20.7 p<0.001). There was no effect of house, sex or breed on mortality. Least mortality occurred in the crop placed in July (3.26 percent) and most occurred in the crop placed in May (6.1 percent). Mortality decreased with increasing parent flock age (F=6.6, R2 = 10.1 p<0.05). Discussion Objective 1: To document the effect of stocking density on the behaviour, physical health and productivity of broilers within the range 30-46 kg/m2 by experimental manipulation of stocking density on commercial farms. The most striking overall result of this study was that, while there were detectable effects of experimentally manipulating the stocking density of meat chickens on a number of the variables we measured, these were overshadowed by the much larger and more consistent differences between companies. Growth rate and the number of birds with the best gaits were related to target stocking density (Table 3) as were some behaviours such as jostling and leg stretching (Appendix 8). However, on almost all measures, the way in which the birds responded to the different stocking densities was modified by which company a trial was carried out with. Some companies appeared to ‘cope’ with high stocking densities much better than others, as judged by welfare indicators such as mortality, leg health and downgrades. CSG 15 (1/00) 15 Project title Stocking density and welfare in broilers DEFRA project code AW0219 This result has two important implications. Firstly, it shows that studying the effects of stocking density on broiler chickens can be very misleading if the study covers only a limited set of environmental conditions. Small scale, laboratory manipulations of stocking density provide useful baseline information but it is impossible to reproduce the relevant conditions of a commercial farm under non-commercial conditions so that any conclusions that can be drawn from them are strictly limited. Even commercial trials carried out with just one company could be misleading since they would not necessarily represent conditions in the rest of the industry. For example, our results show (Figure 8) that by picking out different company’s results, it would be possible to conclude that mortality increased with stocking density, that mortality decreased with stocking density or that mortality was unaffected by stocking density. This dependence of the effects of stocking density on the precise conditions in which it is studied probably accounts for the diversity of previous findings and their lack of consensus on the welfare implications of stocking density for meat chickens1,7-9.We hope that this study will set new standards for the use of scientific information as the basis of legislation that affects a whole industry by involving the industry itself and studying the range of conditions found commercially. Secondly, this result shows that bird welfare is greatly affected by factors other than stocking density and so concentration by legislators on stocking density could mean that other more effective ways of improving bird welfare – such as improving the environment or the genetics – are overlooked. Changes in stocking density are relatively easy to legislate: from a legislators point of view, one number rather than another is inserted into a sentence and a producer conforms by falling below rather than above this figure. Our results show that such a simplistic view of stocking density is not in the best interests of bird welfare. Factors other than stocking density have, within limits, more effect on bird health welfare and should form part of any future legislation. Our study has, as discussed below, gone some way to identifying what those factors are, but what makes for ‘good management’ is, as we point out, inconveniently complex. We must not, however, conclude that stocking density is unimportant. At the two highest stocking densities (42 kg/m2 and 46 kg/m2), there were fewer birds with the very best gaits and, as culling rates were not sufficiently greater, this suggests that leg health was compromised above 42 kg/m2. It is noticeable that some behaviours such as jostling and leg stretching were significantly affected by stocking density and we are in process of examining these effects and their relationship to bird health in more detail by further analysis of the videotapes. We are developing the use of these non-invasive ‘behavioural indicators’ of welfare for use in commercial situations, but as this was not an objective of this project, a detailed report is not included here. Furthermore, the multivariate analysis (discussed under Objective 3) showed that there were significant correlations between actual stocking density and a number of other measures such as drinking, head shaking and wing stretching. In this study, we used a variety of different measures of bird welfare. Some of these were obvious and universally accepted measures such as mortality, some were indirect (downgrades) and others behavioural (gait score, ethograms). We also used a new method of assessing corticosteroid levels from freshly collected faeces and, in conjunction with Ross (now Aviagen) developed a measure of leg health that involved several different features of the state of the feet and legs. It is clear from the results that different measures are affected both by stocking density and by company in different ways – in other words, the different measures do not always correlate with each other. This reinforces the now well established view that there is no single measure of welfare5,6. Animal welfare is multi-faceted and assessing bird welfare on single variables such as foot pad dermatitis, as has been proposed13 could be misleading. Objective 2 To document the effect of season (temperature) on stocking density by conducting the same experimental manipulation of stocking density in summer and again in winter. There were some significant differences between trials conducted at different seasons. Total mortality and faecal corticosterone were higher in winter than in summer (Appendix 9). Pad dermatitis and dirty were also higher in winter. Leg deviations were higher in summer but this was possibly due to the fact that there were more leg culls in winter. The temperature and humidity inside the houses is clearly of crucial importance (Appendix 4) and environmental control is a widely recognized by the industry as very important. It is striking, however, that producers monitor temperature but not humidity within the houses. CSG 15 (1/00) 16 Project title Stocking density and welfare in broilers DEFRA project code AW0219 Objective 3. To identify commercially relevant factors that modify the effect of stocking density by conducting the same experimental manipulation on farms belonging to thirteen different broiler producers at present using a wide range of management and husbandry systems The large and highly significant differences between companies we detected on almost all our measures show that the effects of stocking density can be greatly modified by the birds’ environment. Some features of housing, management or husbandry or a combination of all three clearly has a major effect on bird welfare. The very large numbers of measurements we took of how companies operated and the effects of what they did (for example how good their environmental control was) enabled us to ask what it was that correlated best with the different welfare measures. In view of the need to take different measures of ‘welfare’, we were particularly concerned to pick out company features that consistently affected several different welfare measures. One problem we encountered in analysing this data was that different companies tended each to have their own ‘package’ of management and husbandry practices – a particular lighting regime, temperature profile during development, feed composition etc - and so the different environmental factors tended to vary in clusters, rather than independently. This is an inevitable feature of a correlational as opposed to an experimental approach. Whereas the stocking density results (Objective 1) are based on experimental manipulation of just one factor the effect of which can therefore be clearly identified, a correlational approach (Objective 3) can only point to variables that co-vary and to the possibility that they are causally related. A second problem was of our own making – the very large number of variables we measured. For example, the fact that we had data on how humidity varied hourly throughout the birds’ lives as well as litter moisture just before clearance means that the analysis of how humidity interacts with stocking density could be carried out in a large number of ways, such as humidity on Day 34, mean humidity in each week of the bird’s life, amount of time humidity was outside the recommended (Ross) levels, interaction with temperature and so on. This analysis and its interpretation are still in progress and we believe it will be a very valuable guide to the important environmental variables in a chick’s life. What we are reporting here are the results that meet Objective 3 – namely, the factors that modify the effects of stocking density. We therefore concentrate on the variables that we found to be most affected by stocking density (gait score 0, growth rate and jostle rate) or season (faecal corticosterone). Because of its importance as both a welfare and production indicator, we also included mortality. It was clear from this analysis (Appendix 6) that factors relating to the temperature, air humidity and litter moisture are among most important predictors for these variables. The extent to which companies are able to control this aspect of their birds’ environment is thus one of the explanations for the major between-company differences apparent throughout this study. Components of variation in humidity and temperature include drinker type and the numbers of drinkers/bird (Appendix 4), as well as season of the year. However, it would be a mistake to conclude that control of temperature and humidity are the only management factors that can modify the effects of stocking density. For example, the number of stockmen in charge of a site and the number of visits per day made to each house also recur as predictors of the variables most affected by manipulation of stocking density. Summary and Conclusions 1. Experimental manipulation of stocking density of broiler chickens between 30kg/m2 and 46 kg/m2 resulted in a depression of growth rate, an increase in the incidence of birds jostling each other and, at the two highest stocking densities, a decrease in birds with the best gaits. Some other behaviours such as lying stretched out and wing stretching were also affected by stocking density. 2. The effects of stocking density were, however, overshadowed by much larger differences between companies, suggesting that husbandry and management factors are, within limits, more important than stocking density per se. 3. There were differences between birds reared in summer and those reared in winter. In winter, there were higher levels of mortality and birds showed higher levels of faecal corticosterone and a higher incidence of pad dermatitis and dirty pads. 4. Husbandry and management factors that appear to be important include those that affect the litter moisture levels and temperature and humidity during the birds’ growth. These in turn can be related to drinker type, number of drinkers/bird and season of the year. ‘Good management’, however, consists of more than just controlling temperature and humidity since there are many other factors that have significant effects such as the number of stockmen and the number of visits they make each day. CSG 15 (1/00) 17 Project title Stocking density and welfare in broilers DEFRA project code AW0219 5. The differences between companies in how well they ‘coped’ with higher stocking densities shows the importance of carrying out commercial scale trials with more than one company and not relying on small scale trials that may give misleading results. 6. Legislation to improve broiler welfare should not be directed solely to limiting stocking density as this will detract attention from other more important factors such as improving the birds’ environment and their genetics. 7. Reliance on single measures of welfare such as foot pad dermatitis can be misleading and a variety of different measures are needed to assess welfare, including physical health and behaviour. Knowledge transfer We intend to publish the results in high quality scientific journals, in the popular farming press (e.g. Poultry World) and for a more general audience (e.g. New Scientist). The following papers are planned: 1. The main findings will be submitted shortly to Nature or another rapid publication journal. 2. At least two papers are planned on the behaviour of the birds, one to Animal Behaviour, the other to Applied Animal Behaviour Science. 3. A paper on the role of environmental factors, particularly temperature and humidity, will follow further analysis of the data set. References 1. European Commission (2000) The welfare of chickens kept for meat production (broilers). Report of the Scientific Committee on Animal Health and Animal Welfare. 2. Webster J. (1994) Animal Welfare: A cool eye towards Eden. Blackwell Scientific, Oxford. 3. Farm Animal Welfare Council (1992) Report on the Welfare of Broiler Chickens. FAWC, Tolworth, Surrey. 4. Stevenson, P. (1995) The Welfare of Broiler Chickens. Compassion in World Farming. Petersfield, Hants. 5. Broom, D.M. (1986) Indicators of poor welfare. British Veterinary Journal 142: 5246. Dawkins, M.S. (2001) How can we recognize and assess good welfare? In: Coping with Challenge: Welfare in Animals including Humans ed. D.M. Broom. Dahlem University Press, Berlin. Pp 63-78. 7. Hall, A. (2001) The effect of stocking density on the welfare and behaviour of broiler chickens reared commercially. Animal Welfare 10: 23-40. 8. Grashorn, & Kutritz (1991) Effect of stocking density on performance of modern broiler breeds. Arch Geflugelk 55: 84-90. 9. McLean, J.A., Savory, C.J. & Sparks, N.H.C. (2002) Welfare of male and female broiler chickens in relation to stocking density, as indicated by performance, health and behaviour. Animal Welfare 11: 55-73. 10. Sørensen, P., Su, G. & Kestin, S.C. (2000) Effects of age and stocking density on leg weakness in broiler chickens. Poultry Science 79: 864-870. 11. Feddes, J.J.R., Emmanuel, E.J. & Ziudhoff, M.J. (2002) Broiler performance, body weight variance, feed and water intake and carcass quality at different stocking densities. Poultry Science 81: 774-9 (small pens). 12. Heier, B.T., Hogasen, H.R. and Jarp, J. (2002) Factors associated with cumulative mortality in Norwegian broiler flocks. Preventive Veterinary Medicine 53: 147-165. 13. Martrenchar, A., Boilletot, E., Huoinnic, D. & Pol, F. (2002) Risk factors for foot-pad dermatitis in chicken and turkey broilers in France. Preventative Veterinary Medicine 52: 213-226 14. Cockrem, J.F. & Rounce, J.R. (1994) Faecal measurements of oestradial and testosterone allow noninvasive estimation of plasma steroid concentrations in the domestic fowl. Bristish Poultry Science 35: 433-443. 15. Lord, A (2001) D. Phil. Thesis, Oxford. CSG 15 (1/00) 18 Project title Stocking density and welfare in broilers DEFRA project code AW0219 16. Denhard, M., Schreer, A., Krone, O., Jewgenow, K., Krause, M. & R. Grossmann (2003) Measurement of plasma corticosterone and fecal glucocorticoid metabolites in the chicken (Gallus domesticus), the great cormorant (Phalacrocorax carbo), and the goshawk (Accipiter gentilis). General and Comparative Endocrinology 131: 345-352. CSG 15 (1/00) 19
© Copyright 2026 Paperzz