Assessment of welfare of Brazilian and Belgian broiler flocks using the Welfare Quality protocol

Assessment of welfare of Brazilian and Belgian broiler flocks using the Welfare Quality protocol F. A. M. Tuyttens,∗,†,1 J. F. Federici,‡ R. F. Vanderhasselt,∗ K. Goethals,† L. Duchateau,† E. C. O. Sans,‡ and C. F. M. Molento‡ ∗

Animal Sciences Unit, Institute for Agricultural and Fisheries Research (ILVO), Scheldeweg 68, B-9090 Melle, Belgium; † Faculty of Veterinary Medicine, Ghent University, Salisburylaan 133, B-9820 Merelbeke, Belgium; and ‡ Animal Welfare Laboratory, Federal University of Paran´ a, Rua dos Funcion´ arios, 1540, CEP 80035-050, Curitiba, Brazil

Key words: animal welfare, country of origin, poultry meat, trade, Welfare Quality 2015 Poultry Science 00:1–9 http://dx.doi.org/10.3382/ps/pev167

INTRODUCTION

all dimensions of animal welfare, and a greater focus on animal-based measures than resource-based measures for assessing each of these dimensions. Animal-based measures are believed to be more directly related to the animal welfare status as they show the outcome of the interaction between the animal and its environment (Blokhuis et al., 2010; EFSA, 2012). The goal of the focus on animal-based measures was to make the assessments independent of the housing and management system. As a consequence, they should be particularly suited for comparing the welfare of animals that are kept in different environments. Most Welfare Quality protocols are also distinguished by the standardized way of integrating the information from all welfare measures, such that farms can be assigned one of four categories of overall welfare status. Welfare Quality defines 12 distinct but complementary

After 5 years research, in 2009 the European Welfare Quality consortium published practical protocols for assessing the welfare of the main livestock species, including broiler chickens, on farms and in slaughterhouses. These protocols have been developed through the collaboration of a large number of research groups, in consultation with stakeholders, and have been proposed as standardized science-based ways of assessing animal welfare (Blokhuis et al., 2010). Innovative features of the Welfare Quality approach include the attention for  C 2015 Poultry Science Association Inc. Received December 12, 2014. Accepted May 7, 2015. 1 Corresponding author: [email protected]

1

Downloaded from http://ps.oxfordjournals.org/ by guest on September 15, 2015

level of the welfare principles, criteria, and measures. Brazilian farms obtained higher scores for 3 of the 4 welfare principles: ‘good feeding’ (P = 0.007), ‘good housing’ (P < 0.001), and ‘good health’ (P = 0.005). Four of the 10 welfare criteria scores were, or tended to be, higher on Brazilian than Belgian farms: ‘absence of prolonged thirst’ (P < 0.001), ‘ease of movement’ (P < 0.001), ‘absence of injuries’ (P = 0.002), and ‘positive emotional state’ (P = 0.055). The only criteria with a higher score for the Belgian farms than their Brazilian counterparts were ‘absence of prolonged hunger’ (P = 0.048) and ‘good human–animal relationship’ (P = 0.002). Application of the Welfare Quality protocol has raised several concerns about the validity, reliability, and discriminatory potential of the protocol. The results also call for more research into the effect of animal welfare legislation as broiler welfare on the south Brazilian farms appeared to be superior to that on the Belgian farms. Animal-based welfare assessments on a larger sample of farms are needed to evaluate to what extent these findings may be generalized.

ABSTRACT The Welfare Quality consortium has proposed a science-based protocol for assessing broiler chicken welfare on farms. Innovative features make the protocols particularly suited for comparative studies, such as the focus on animal-based welfare measures and an integration procedure for calculating an overall welfare status. These protocols reflect the scientific status up to 2009 but are meant to be updated on the basis of inter alia implementation studies. Because only few such studies have been done, we applied the Welfare Quality protocol to compare the welfare of broiler flocks in Belgium (representing a typical European Union (EU) country which implies stringent animal welfare legislation) versus Brazil (the major broiler meat exporter to the EU and with minimal animal welfare legislation). Two trained observers performed broiler Welfare Quality assessments on a total of 22 farms in Belgium and south Brazil. All of the farms produced for the EU market. Although the overall welfare was categorized as ‘acceptable’ on all farms, many country differences were observed at the

2

TUYTTENS ET AL. Table 1. Overview of the measures taken on farm or at slaughter included into the Welfare Quality assessment protocol for on-farm broiler chicken welfare (based on Welfare Quality, 2009). Measures were taken from farm (Fr) or slaughter (Sr) records, or were scored by the trained observers on-farm (Fs). Principles

Criteria

Measures

Good feeding

Absence of prolonged hunger Absence of prolonged thirst

Emaciation (Sr) Drinker space (Fs)

Good housing

Comfort around resting

Plumage cleanliness (Fs), litter quality (Fs), dust sheet test (Fs) Panting (Fs), huddling (Fs) Stocking density (Fr)

Thermal comfort Ease of movement Good health

Absence of injuries Absence of disease Absence of pain induced by management procedures1

Appropriate behavior

/ Free-range (Fs), cover on the range (Fs) Avoidance distance test (Fs) Qualitative behavior assessment (Fs)

Indicates welfare criteria that are not measured for on-farm broiler welfare (because an appropriate measure is lacking).

animal welfare criteria grouped into 4 principles: good feeding, good housing, good health, and appropriate behavior (Table 1). A 3-step integration process is used to derive an overall assessment of animal welfare. The first step transforms the measures collected on farms or at slaughter into criteria scores on a value scale (0 = worst; 100 = best situation possible) to reflect compliance of a given farm with each of the 12 criteria. Next, the criteria scores are aggregated into scores for each of the 4 principles (0 to 100 scale). In the final stage of the integration procedure, the 4 principle scores are aggregated to categorize the overall welfare status as “not classified,” “acceptable,” “enhanced,” or “excellent.” The Welfare Quality protocols reflect the scientific status up to 2009, but they are meant to undergo a continuous process of updating and revision on the basis of new scientific findings, societal developments, and practical experiences gained during implementation (www.welfarequalitynetwork.net). Use of the Welfare Quality protocols by the industry or certification bodies has been limited, however, with especially limited use of the broiler chicken protocol. Furthermore, published scientific reports that present information on the application of the broiler protocol are, however, still scarce (De Jong et al., 2011; Sans et al., 2014; de Oliveira Souza et al., 2015). To gain information about its practical implementation, we applied the Welfare Quality protocol for assessing the welfare of commercial broiler flocks in 2 countries, namely Belgium and Brazil. We chose these 2 countries for several reasons. First, the variation of housing and management conditions could be expected to be greater by focusing on 2 very different and geographically distant countries. Second, a case study that compares broiler welfare in these 2 contrasting countries was expected to yield interesting observations.

Indeed, Belgium represents typical industrial broiler farms in the European Union (EU) that are subject to strict animal welfare legislation. Council Directive 2007/43/EC stipulates basic rules for the protection of chickens kept for meat production in the EU. Standards are stated for stockperson training, drinkers, feeding, litter, ventilation, noise, lighting, inspection, cleaning, record keeping, and surgical interventions. The main provision of this Directive, however, is to limit the stocking density to 33 kg/m2 . Derogation is available for higher stocking densities up to 39 kg/m2 and even 42 kg/m2 , but only if more rigorous requirements related to production, environmental quality, inspections, record keeping, and maximum mortality rate are met. Belgium accounts for 2.4% of the broiler production in the EU (Tuyttens et al., 2014b). Moreover, Belgium is located between the northern and western member states, which often have higher animal welfare standards than the EU Directives require, and the southern and eastern member states which rarely have legislation in excess of the minimums stated in the EU Directives (van Horne and Achterbosch, 2008). Brazil is the largest poultry meat exporter in the world (ITAVI, 2012) and supplies nearly two-thirds of the poultry meat imported into the EU (Van Horne and Bondt, 2013). Although societal calls for animal welfare reforms are growing as citizens in emerging countries become more aware of intensive production systems and potentially controversial husbandry practices, animal welfare does not seem to be yet an equally important issue for the domestic market in Brazil as it is in Europe (Bonamigo et al., 2012; von Keyserlingk and H¨ otzel, 2014). Only basic requirements relating to animal welfare have been legislated. Brazil, like the United States and Canada, mainly relies on establishing voluntary, market-driven good-practice recommendations for animal welfare that target export-oriented

Downloaded from http://ps.oxfordjournals.org/ by guest on September 15, 2015

1

Expression of social behaviors1 Expression of other behaviors Good human–animal relationship Positive emotional state

Lameness (Fs), hock burn (Fs), foot pad dermatitis (Fs), breast lesion (Fs) Mortality and culls on farm (Fr), ascites (Sr), dehydration (Sr), septicaemia (Sr), hepatitis (Sr), pericarditis (Sr), abscess (Sr) /

BROILER WELFARE IN BRAZIL VERSUS BELGIUM

MATERIALS AND METHODS Broiler Flocks In total, 11 Belgian and 11 Brazilian broiler flocks, all from different farms, were included in the study. Farm visits in Belgium and Brazil were carried out in May and October 2011, respectively. Each flock was housed in a different industrial farm that produced broilers for a single slaughterhouse in Belgium and Brazil, respectively. These farms were randomly selected from the slaughterhouse’s producer list, conditional to farmer consent (one Belgian farmer did not grant us access to his farm). Belgian farms were mainly located in Flanders, the region where Belgian broiler production

(predominantly for the national and EU market) is concentrated. All Belgian farms reared mixed-gender Ross flocks in closed poultry houses, as is typical for Belgian broiler production (Tuyttens et al., 2014b). The farms visited in Brazil all produced broilers destined for the European market and were located in the region of Passo Fundo, State of Rio Grande do Sul, south Brazil. The Brazilian farms operated in an integrated system within a major Brazilian food company, as is typical for industrial broiler production in the Brazil. The farms were regularly inspected for quality standards set by the integrator for export to the EU. All Brazilian farms reared male Cobb birds in open-sided poultry houses with curtains on both walls. None of the flocks had access to a free-range area.

Broiler Welfare Assessment The on-farm animal welfare status was assessed using the Welfare Quality protocol (Welfare Quality, 2009). Each flock was assessed during a single farm visit 1 or 2 d before slaughter by 2 trained assessors; one from Brazil and one from Belgium, who scored the animals simultaneously but independently. For all analyses, the mean score of these 2 observers was used. Both assessors had been trained in the theory and practice of Welfare Quality scoring by an experienced trainer accredited by the Welfare Quality Network. Training consisted of 1 d theory followed by 1 d on-farm training. Subsequently, both trainees practiced further using photographs and video-recordings (by scoring independently from one another followed by discussing differences in order to reach consensus) until they no longer disagreed on the scores. The assessors aimed to start the assessment of each flock at around 9:00 AM. A brief farmer’s questionnaire and flock records were used to obtain information on the number of birds in the house at placement and at the date of the visit, their age, breed and average weight, house measurements, and the cumulative mortality and culling percentage. The assessment started by a walking tour of the stable to observe the whole flock, followed by a qualitative behavior assessment, flock-level observations of panting and huddling, avoidance distance testing, recording of housing characteristics (dimensions, litter quality, dust, type, and number of drinkers), and scoring the gait, plumage cleanliness, foot pad dermatitis, and hock burns of a sample of birds as described in the broiler protocol (Welfare Quality, 2009). Breast lesions were also scored on the same sample of live birds, because the observers were not permitted to do measurements at the slaughter line (i.e., the recommended location for scoring breast blisters by the protocol). Moreover, similarly to de Jong et al. (2011) we did not only score real breast blisters (as prescribed by the Welfare Quality protocol) but also any other type of wound, burn, or irritation of the breast area. A small part of the Welfare Quality assessment for broiler on-farm welfare is based on data collected

Downloaded from http://ps.oxfordjournals.org/ by guest on September 15, 2015

companies wishing to comply with customer demands (van Wagenberg et al., 2012). In theory, industry incentives to improve welfare standards could result in a similar welfare level as legal standards. Although there seems to be a general perception in Europe that animal welfare is inferior in third-world countries, some have reported that broilers produced in Brazil for export to the EU may have an equivalent or even better welfare status than their EU counterparts (Bracke, 2009; Van Horne and Bondt, 2013). Evidence-based information is very limited, however. Van Wagenberg et al. (2012) state that basic welfare requirements are fulfilled on many farms in south Brazil, the region where broiler production for export is concentrated. The subtropical climate lends itself to low-energy housing and low stocking densities (Van Horne and Goddijn, 2005; Bracke, 2009; ITAVI, 2012). The indoor climate has been reported to be generally good and the litter on the earth floor to be dry, resulting in a low risk for pododermatitis, contrary to what is observed in European production units (ITAVI, 2012). Moreover, Van Horne and Goddijn (2005) reported that mortality did not appear unusually high even though broilers in Brazil are kept until they reach a higher BW. The integration model is largely adopted in Brazil, which brings efficient and professional production and slaughter methods with modern technologies and strict control of the entire production chain (ITAVI, 2012; van Wagenberg et al., 2012). Thinning and mixed-sex rearing are not common practice in Brazil. Finally, the low labor costs allow for animal-friendly catching and loading of the broilers, and transport distances to the slaughterhouses are relatively short, although the road quality may be bad (Van Horne and Goddijn, 2005). These reports provide anecdotal evidence that casts doubt on the common perception that animal welfare is a food quality attribute that legitimately differentiates broiler meat produced in the EU from that imported from south Brazil. The aims of this study were, therefore, to gain knowledge on the implementation of the Welfare Quality protocol by applying it to compare broiler welfare on industrial farms in Belgium versus south Brazil.

3

4

TUYTTENS ET AL. Table 2. Differences between Belgian and Brazilian broilers flocks and indoor climate. Variable

Belgium Mean (SE)

Brazil Mean (SE)

P

Bird age (d) Average BW (kg) Flock size Surface area housing unit (m2 ) Stocking density (kg/m2 ) Total mortality (%) Temperature (◦ C) Wind velocity (m/s) Relative humidity (%)

38.64 (0.51) 2.34 (0.05) 20,054 (1,957.6) 1,179 (118.7) 39.97 (0.92) 3.95 (0.66) 26.18 (0.44) 0.26 (0.04) 54.71 (2.83)

40.36 (0.77) 2.57 (0.08) 12,859 (848.3) 1202 (82.3) 27.55 (0.90) 5.11 (0.39) 27.88 (0.80) 0.67 (0.09) 49.70 (4.45)

0.075 0.018 0.005 0.874 < 0.001 0.145 0.081 < 0.001 0.356

flocks had access to a free range. For the flock and indoor climate characteristics as well as for the individual welfare measures, the nonparametric Wilcoxon–Mann– Whitney test was used to test for a difference between Belgian and Brazilian flocks.

RESULTS Housing and Management Differences

In addition to the broiler welfare measures, the observers recorded measures that are not included in the Welfare Quality protocol, namely temperature, RH, and wind velocity in the house at bird height using a portable device (Testo 410–2, Testo NV/SA, Ternat, Belgium), as well as the type of heating (wood, gas, or hot air gun). In addition, in each flock, 4 groups of 5 birds were penned for validating the voluntary water consumption test as an alternative measure of thirst (Vanderhasselt et al., 2014). These birds were not used for the welfare assessments of the present study.

Broiler housing and management differed substantially between Belgium and Brazil (Table 2). The birds in Brazil were kept for circa 2 d longer before slaughter and thus reached a higher BW than in Belgium. All Belgian broiler houses had concrete floors, whereas only one Brazilian broiler house had a concrete floor. The litter in Belgium was removed after every production round, whereas in Brazil a thick litter layer was kept for several production rounds and was milled frequently, often daily. In Belgium all houses were closed and heated by hot air guns, whereas Brazilian houses were open-sided (with screens and curtains along the long sides) and heated with wood, gas, or both wood and gas. One Belgian house was equipped with bell drinkers and 2 Belgian houses were equipped with cup drinkers. All other houses were equipped with nipple drinkers. Eight houses in Brazil had standard dimensions (100 × 12 m), one had a different shape but similar surface area (120 × 10 m), one was larger (130 × 14 m), and one was smaller (50 × 19 m). Belgian housing units were more variable in shape and dimensions (Table 2) and were often shorter and wider. On 2 farms in Brazil, the birds were fed manually instead of automatically.

Statistical Analyses

Broiler Welfare Assessment

Data were analyzed using the Statistical Analysis System Version 9.3 for Windows (SAS Institute Inc., Cary, NC). Effect of country on the 0 to 100 principle and criteria scores was tested using Student’s t-test. It should be noted that the Welfare Quality protocol for broiler welfare on-farm does not (yet) allow calculation of a score for criteria ‘absence of pain induced by management procedures’ and ‘expression of social behaviors’ because appropriate measures are lacking, and all scores for the criterion ‘expression of other behaviors’ were the same (score = 13) as none of the

The overall animal welfare category was ‘acceptable’ for all Belgian and Brazilian flocks. In order to refine this overall welfare level, we differentiated 3 levels, i.e., acceptable-plus, acceptable-medium, and acceptableminus. If all principle scores were <50, we categorized the flock as ‘acceptable minus,’ if one principle score was >50 and all others <40 we categorized the flock as ‘acceptable medium,’ and if at least one principle score was >50 and at least one other principle score >40 than the flock was categorized as ‘acceptable plus.’ Using these subcategories, 40% of the Belgian flocks but

Additional Measurements

Downloaded from http://ps.oxfordjournals.org/ by guest on September 15, 2015

at slaughter. Copies of the meat inspection records were available for the flocks in question. From these records the number of birds rejected for the various predefined pathological reasons (i.e., ascites, dehydration, septicemia, hepatitis, pericarditis, and abscesses) was derived. The meat inspection records of one Belgian farm were unavailable, thus the scores for criteria ‘absence of prolonged hunger’ and ‘absence of injuries’ and for principles ‘good feeding’ and ‘good health’ and the overall welfare category could not be calculated for that farm. Data were recorded using personal digital assistants equipped with specialized software developed by Wageningen University & Research centre Livestock Research (The Netherlands). Data were downloaded into an Excel database, checked for errors, and subsequently used as input for calculating the integrated criteria and principle scores and the overall welfare category using the software developed by Welfare Quality.

5

BROILER WELFARE IN BRAZIL VERSUS BELGIUM Table 3. Number of Belgian and Brazilian flocks of which the overall welfare level was categorized as acceptable-minus, acceptable-medium, and acceptable-plus. Country

Acceptable-Minus

Acceptable-Medium

Acceptable-Plus

4 0

6 8

0 3

Belgium (n = 10) Brazil (n = 11)

Table 4. Differences in the mean (SE) welfare principle and criteria scores between the Belgian and Brazilian broiler flocks. Higher scores are better for broiler welfare. Belgium Mean (SE)

Brazil Mean (SE)

P

Good feeding Absence of prolonged hunger Absence of prolonged thirst Good housing Comfort around resting Thermal comfort Ease of movement Good health Absence of injuries Absence of diseases Appropriate behavior Expression of other behaviors Good human–animal relationship Positive emotional state

54.6 83.0 50.8 24.4 26.2 35.4 26.5 28.1 16.8 56.8 29.0 13.0 95.9 32.7

78.6 74.5 90.2 36.0 31.0 42.7 57.2 35.3 23.6 68.5 29.4 13.0 80.6 38.7

0.007 0.048 < 0.001 < 0.001 0.310 0.399 < 0.001 0.005 0.002 0.392 0.741 N/A1 0.002 0.055

1

(6.8) (1.8) (6.7) (2.0) (3.0) (5.3) (3.4) (2.0) (1.5) (11.3) (1.0) (0.0) (1.1) (2.5)

(3.0) (3.5) (1.9) (1.7) (3.5) (6.7) (1.5) (1.2) (1.2) (7.4) (0.7) (0.0) (3.8) (1.5)

N/A = not applicable.

none of the Brazilian flocks were categorized as ‘acceptable minus’ (Table 3). None of the Belgian flocks and 27% of the Brazilian flocks were categorized as ‘acceptable plus.’ The average scores for 3 of the 4 welfare principles (‘good feeding,’ ‘good housing,’ and ‘good health’) were significantly higher for the Brazilian than the Belgian flocks (Table 4, Figure 1). The average scores for the fourth principle (‘appropriate behavior’) were very similar for both countries. Seven out of the 10 criteria scores were numerically higher for the Brazilian flocks than the Belgian flocks, although the difference was significant for only 3 criteria (‘absence of prolonged thirst,’ ‘ease of movement,’ and ‘absence of injuries’) and nearly significant for ‘positive emotional state’ (Table 4). Belgian flocks had significantly higher scores for ‘absence of prolonged hunger’ and ‘good human–animal relationship’. For the individual welfare measures, several differences were found (Table 5). These differences often indicated fewer welfare problems among Brazilian flocks. For example, Brazilian flocks had a better median score for plumage cleanliness, litter quality, foot pad dermatitis, and lameness. The percentage of birds with breast lesions, however, was higher on Brazilian versus n farms. The percentage of birds found dead by the farmer during the production round did not differ significantly between Belgian and Brazilian flocks. However, culling was more frequent on Brazilian than Belgian farms. There was a trend for a higher percentage of birds panting on Belgian than Brazilian farms. Huddling was very rarely observed in either country. Regarding the individual terms that were scored for the qualitative behavior assessment, Brazilian flocks

were perceived as more ‘comfortable’ (87.5 vs. 65.5, P = 0.011), more ‘content’ (88.0 vs. 68.5, P < 0.001), more ‘energetic’ (82.0 vs. 69.5, P = 0.028), more ‘positively occupied’ (76.5 vs. 63.5, P = 0.015), less ‘distressed’ (14.0 vs. 23.5, P = 0.007), and less ‘frustrated’ (25.0 vs. 29.0, P = 0.020) than Belgian flocks.

DISCUSSION This study provides information about the application of the Welfare Quality broiler welfare assessment protocol and indicates that on-farm broiler welfare appeared to be superior among the limited sample of south Brazilian flocks as compared to the equally limited sample of Belgian flocks.

Welfare Quality Protocol for Assessing Broiler Welfare Our study relies heavily on the assumption that the Welfare Quality protocol correctly quantifies broiler welfare. The completeness and validity of the Welfare Quality measures deserve critical examination. Our first observation is that the broiler protocol does not include a measure for 2 welfare criteria (‘absence of pain induced by management procedures’ and ‘expression of social behaviors’). Consequently, the protocol could not have detected if broilers from the Belgian and Brazilian farms differed with regards to these welfare criteria. Second, it is sometimes claimed that the Welfare Quality measures have been carefully selected on the basis of their reliability, validity, and feasibility (Blokhuis et al., 2010). However, for some measures, such as drinker space as an indicator of prolonged

Downloaded from http://ps.oxfordjournals.org/ by guest on September 15, 2015

Principles & Criteria

6

TUYTTENS ET AL.

Table 5. Differences in the median (range) scores for separate welfare measures between Belgian and Brazilian broiler flocks. Variable

Belgium Median (min to max)

Brazil Median (min to max)

P

Plumage cleanliness (0 to 3 scale) Litter quality (0 to 4 scale) Dust (0 to 4 scale) Panting (% birds) Huddling (% birds) Lameness (0 to 2 scale) Hock burns (0 to 2 scale) Foot pad dermatitis (0 to 2 scale) Breast lesions (% birds) On-farm mortality (% birds) Culls on farm (% birds) Ascites (% birds) Dehydration (% birds) Septicemia (% birds) Hepatitis (% birds) Pericarditis (% birds) Subcutaneous abscesses (% birds)

1.16 (1.06 to 2.28) 2.08 (1.33 to 3.00) 3.00 (2.00 to 4.00) 45.90 (7.07 to 96.40) 0.00 (0.00 to 1.67) 1.23 (1.12 to 1.40) 0.85 (0.54 to 1.44) 1.71 (0.95 to 1.99) 0.00 (0.00 to 3.96) 3.00 (1.30 to 8.70) 0.00 (0.00 to 1.60) 0.14 (0.00 to 1.10) 0.36 (0.10 to 1.01) 0.00 (0.00 to 0.58) 0.00 (0.00 to 0.00) 0.00 (0.00 to 0.00) 0.46 (0.14 to 4.26)

0.75 (0.36 to 1.51) 1.25 (1.00 to 2.80) 3.00 (2.00 to 4.00) 33.33 (3.00 to 59.10) 0.00 (0.00 to 0.00) 1.12 (1.02 to 1.22) 0.85 (0.47 to 1.20) 1.07 (0.97 to 1.80) 4.81 (2.65 to 6.38) 4.70 (2.20 to 1.70) 0.60 (0.20 to 1.70) 0.17 (0.00 to 0.54) 0.00 (0.00 to 0.04) 0.02 (0.00 to 0.18) 0.00 (0.00 to 0.00) 0.00 (0.00 to 0.00) 0.00 (0.00 to 0.01)

0.002 0.010 0.274 0.076 0.363 0.002 0.844 0.026 < 0.001 0.324 < 0.001 0.622 < 0.001 0.732 1 1 < 0.001

thirst, these claims appear poorly supported and highly contested (Sprenger et al., 2009; Vanderhasselt et al., 2013a, 2014). Third, for some measures the comparability when applied in very different production systems can be questionable. For example, based on the avoidance distance test we found a better score for the criterion ‘human–animal relationship’ for the Belgian

flocks than their Brazilian counterparts. However, the observers reported that the lower stocking density (e.g., making it easier for the birds to flee) and greater light intensity (e.g., making the birds more responsive to their surroundings) on the Brazilian farms may have confounded this test, which calls for further validation studies. Fourth, many welfare indicators of the broiler

Downloaded from http://ps.oxfordjournals.org/ by guest on September 15, 2015

Figure 1. Principle and criteria scores for the Belgian (black circles) and Brazilian flocks (grey circles) in the left panel. The right panel shows the mean differences (with 95% CI) for these scores between Belgian and Brazilian flocks. If the CI values do not include zero, the difference is significant (P < 0.05). Differences < 0 indicate a better score for Brazilian flocks and differences > 0 indicate a better score for Belgian flocks.

BROILER WELFARE IN BRAZIL VERSUS BELGIUM

Differences between Belgian and Brazilian Flocks Despite these critical remarks, the Welfare Quality protocol is, to the best of our knowledge, the best tool available at present for assessing overall broiler welfare and we are quite confident that the comparison of broiler welfare generally indicated better conditions for the Brazilian flocks versus their Belgian counterparts. The differences found were more often in favor of the Brazilian than the Belgian flocks. This was true at the level of the individual welfare measures as well as at the various stages of aggregation such as the criteria and principle scores. Only at the very last step of the integration procedure, in which flocks are allocated to an overall welfare category, we could not differentiate between flocks as they were all categorized as ‘acceptable.’ This is not surprising, though, as this category is meant to include the majority of conventional farms where the level of animal welfare is above minimal levels of acceptability (i.e., compliant with the minimum standards in EU legislation) but insufficient to enter a certification scheme (Botreau et al., 2009; Veissier et al., 2011).

The rules for classifying farms are supposed to be a compromise between theoretical aspirations and what can be achieved in practice. Contrary to the cattle protocol, however, the Welfare Quality rules for categorizing broiler farms were not checked for their agreement with the general impressions of experts who had visited the farms nor for how realistic they were in practice for farms to achieve higher welfare categories (Veissier et al., 2011). These rules appear to correspond poorly with researchers’ general impressions and do not allow sufficient discrimination between farms (De Jong et al., 2011). For this reason, we differentiated three subdivisions within the ‘acceptable’ category: acceptableminus, acceptable-medium, and acceptable-plus. The use of these subdivisions supports the conclusion that overall broiler welfare on average appeared to be superior in the Brazilian than Belgian farms. These subdivisions have been based on rather arbitrary decision rules; however, expert and stakeholder support for such rules would be desirable. The questionable quality of the only 3 welfare indicators (percent emaciated birds, percent birds with breast lesions, and avoidance distance test) for which the Belgian flocks scored better than the Brazilian flocks also led us to believe that Brazilian flocks fared better in general. The percentage emaciated birds is based on nonverified slaughterhouse rejection records. To our knowledge, the reliability of such records has not yet been tested. The recommended place to record the percent birds with breast lesions is the slaughter line after the birds are defeathered (Forkman and Keeling, 2009). However, the observers were not granted access to the slaughterhouses, thus this measure was scored on-farm on a sample of living birds. The correspondence between on-farm and at-slaughter scores has not been tested, however, and it cannot be ruled out that the condition of the plumage (which was better among the Brazilian birds) confounds live scoring (Forkman and Keeling, 2009). As mentioned previously, the avoidance distance test may have been confounded by the agility of the birds (Brazilian birds had a better gait score), by the stocking density (which was lower in Brazil) and the light intensity (which the observers reported to be higher in Brazil). For example, De Jong et al. (2011) remarked that, although slower-growing broiler strains did not appear to be more fearful according to the researchers, they obtained a worse score on the avoidance test. The researchers argued that the test measured the birds’ ability to walk more than their degree of fear. Similar remarks can also be made for some of the welfare indicators for which the Brazilian flocks scored better: for instance, slaughterhouse records of pathological rejections have also not been tested for reliability. However, the welfare indicators for which the Brazilian flocks obtained a better score were far more numerous and did include several important indicators that have been validated and tested for reliability, such as lameness, foot pad dermatitis, and plumage cleanliness

Downloaded from http://ps.oxfordjournals.org/ by guest on September 15, 2015

protocol are subjective to some degree. Tuyttens et al. (2014a) have documented that such welfare indicators may be prone to observer bias due to conscious or unconscious expectations of observers about the study outcome. Blinding the observers is very effective in preventing such expectation bias but this was not possible in the present study. Lesser debiasing techniques (Reese, 2012; Croskerry et al., 2013) were used instead of blinding; all farms were assessed simultaneously but independently by 2 observers, one from Brazil and one from Belgium. The different background of both observers increased the likelihood that they would have different expectations about the study outcome. Moreover, the knowledge that their own scoring could always be compared with the scoring of the other observer is likely to have increased their own bias awareness, perspective-taking, and accountability. The Welfare Quality step-wise procedure to integrate the information from all welfare measures to assign flocks to one of the 4 categories of overall welfare status is claimed to have societal support (Blokhuis et al., 2010: Veissier et al., 2011). Due to lack of a goldstandard reference value for overall welfare status, however, it is difficult or even impossible to test the validity of these integrated scores. Indeed, some have expressed concerns that the overall welfare categories sometimes do not reflect expert opinion (De Jong et al., 2014; Van Eerdenburg et al., 2014), and that the relative weights attributed to the various welfare criteria or principles are not always in line with the quality of the underlying measures (Tuyttens et al., 2009; Heath et al., 2013) or with the opinion of key stakeholders (Tuyttens et al., 2010, 2014b).

7

8

TUYTTENS ET AL.

sumed to be equal (but see de Oliveira Souza et al., 2015). Before these findings can be possibly generalized, a much larger and more diverse sample of flocks would need to be assessed during different seasons. A much larger study using validated and reliable animal-based measures to compare the welfare status of broiler flocks in the EU versus countries exporting meat to the EU would be very interesting. Not only would this allow a proper risk-factor analyses (and hence a better understanding and identification of preventive and remedial actions) for important broiler welfare problems, it seems warranted to clarify whether or not concerns about animal welfare are a valid argument to promote the consumption of EU-produced instead of imported poultry meat. Such evidence-based information would be highly relevant in the context of negotiations to further liberalize international trade by abolishing quotas and import levies and the future competitiveness of the EU meat industry. Furthermore, evidence-based evaluations of the impact of animal welfare legislation on the actual welfare status of the animals on commercial farms, as well as the consequences of these norms if trade barriers were to be further liberalized, must be enabled. In conclusion, this comparative study on a limited number of farms in Belgium versus south Brazil has revealed some concerns about the validity and reliability of the Welfare Quality protocol for assessing broiler welfare. Nevertheless, the findings challenge the common perception held by European consumers that the welfare of broilers reared in the EU is superior to that of broilers reared in south Brazil for the EU market. On the contrary, our results suggest that although overall welfare status was categorized as ‘acceptable’ on all farms studied, specific welfare aspects indicate that conditions were often better for the Brazilian flocks than the Belgian flocks. Animal-based welfare assessments on a larger sample of farms are needed to evaluate to what extent these findings may be generalized.

ACKNOWLEDGMENTS Study Limitations and Implications The main limitations of our study are that the welfare status of broilers was investigated during a particular time of the year, during the on-farm rearing period only and on a limited number of farms. We cannot assume our restricted sample of flocks to be representative for the conventional broiler industry in Belgium, let alone the EU, although we have no reasons to assume otherwise and although general farm, flock, and production characteristics of our sample of Belgian farms are comparable to those reported from a recent survey (Tuyttens et al., 2014b). The same consideration applies to the Brazilian sample, which was deliberately restricted to farms producing broilers for the EU market. Because these farmers ought to comply with the norms agreed upon with EU retails, the welfare status of broilers produced for the domestic Brazilian market cannot be as-

The authors thank all the broiler producers and slaughterhouses for their cooperation within this study. We also thank T. Martens [Institute for Agricultural and Fisheries Research (ILVO), Melle, Belgium] and T. Decroos (ILVO, Melle, Belgium) for technical assistance, M. Levenson (ILVO, Merelbeke, Belgium) for English-language editing, and S. Buijs (ILVO, Melle, Belgium) and A. P. de Oliveira Souza and B. M¨ uller (Laborat´ orio de Bem-estar Animal/Universidade Federal do Paran´ a, Curitiba, Brazil) for useful input. We are grateful to the Brazilian National Research Council (Conselho Nacional de Desenvolvimento Cient´ıfico e Tecnol´ogico) for funding.

REFERENCES Berg, C. C. 2004. Pododermatitis and hock burns in broiler chickens. Pages 37–49 in Measuring and Auditing Broiler Welfare. C. A.

Downloaded from http://ps.oxfordjournals.org/ by guest on September 15, 2015

(Forkman and Keeling, 2009; EFSA 2012; Vanderhasselt et al., 2013b). The number of flocks assessed in Belgium and Brazil were so limited and the number of dissimilarities between production systems in Belgium and Brazil were so numerous, it is impossible, and beyond the scope of the present study, to disentangle underlying causes for the documented differences in broiler welfare. Nonetheless, it seems likely that better scores for many broiler welfare indicators on Brazilian farms were related to a combination of management and housing differences. For example, the lower stocking density in conjunction with the open-sided housing units—which allowed natural lighting and ventilation—likely contributed to a lower panting score (as a sign of heat stress; McLean et al. 2002) and to better litter quality. The better litter quality on Brazilian farms was probably also related to the frequent (often daily) milling of the litter which, in turn, likely contributed to cleaner plumage and a reduced level of foot pad dermatitis. Hock and breast burns, however, are less-sensitive indicators of litter problems (Green et al., 1985; Berg, 2004; Meluzzi and Sirri, 2009) which may explain why these types of contact dermatitis were not less frequent on the Brazilian versus Belgian farms. In fact the percentage of birds with breast lesions was higher in Brazilian farms. The observers did report that the 2-point scale (absence vs. presence of lesions) did not allow them to adequately differentiate the differences in breast lesions they observed. On Belgian farms, lesions were rare but were more severe, whereas on Brazilian farms lesions were more common but were mostly superficial or limited to some small crusts (which appeared related to the different type of litter—chopped wood sometimes mixed with rice hulls, coffee bean hulls, or straw—which may have a higher number of sharp edges). The higher culling rate of presumably moribund birds in Brazil may also have contributed to lower percentages of birds with clinical problems at the time of the on-farm assessment and at slaughter.

BROILER WELFARE IN BRAZIL VERSUS BELGIUM

Tuyttens, F. A. M, S. de Graaf, J. L. T. Heerkens, L. Jacobs, E. Nalon, S. Ott, L. Stadig, E. Van Laer, and B. Ampe. 2014a. Observer bias in animal behaviour research: Can we believe what we score, if we score what we believe? Anim. Behav. 90: 273–280. Tuyttens, F. A. M., D. Maes, N. Geverink, P. Koene, and T. B. Rodenburg. 2009. Assessing animal welfare at farm and group level: Introduction and overview. Anim. Welf. 18:323–324. Tuyttens, F. A. M., F. Vanhonacker, E. Van Poucke, and W. Verbeke. 2010. Quantitative verification of the correspondence beR operational definition of farm antween the Welfare Quality imal welfare and the opinion of Flemish farmers, citizens and vegetarians. Livest. Sci. 131:108–114. Tuyttens, F. A. M., F. Vanhonacker, and W. Verbeke. 2014b. Broiler production in Flanders, Belgium: Current situation and producers’ opinions about animal welfare. World’s Poult. Sci. J. 70:343– 354. Vanderhasselt, R. F., S. Buijs, M. Sprenger, K. Goethals, H. Willemsen, L. Duchateau, and F. A. M. Tuyttens. 2013a. Dehydration indicators for broiler chickens at slaughter. Poult. Sci. 92:612–619. Vanderhasselt, R. F., K. Goethals, S. Buijs, J. F. Federici, E. C. O. Sans, C. F. M. Molento, L. Duchateau, and F. A. M. Tuyttens. 2014. Performance of an animal-based test of thirst in commercial broiler chicken farms. Poult. Sci. 93:1327–1336. Vanderhasselt, R. F., M. Sprenger, L. Duchateau, and F. A. M. Tuyttens. 2013b. Automated assessment of footpad dermatitis in broiler chickens at the slaughter-line: Evaluation and correspondence with human expert scores. Poult. Sci. 92:12–18. Van Eerdenburg, F., J. Hulsen, S. Snel, and A. Stegeman. 2014. Monitoring welfare in practice on Dutch dairy farms. Proc. Benelux Int. Soc. Appl. Ethol. Conf., Eersel, The Netherlands, 23. (Abstr.) Van Horne, P. L. M., and T. J. Achterbosch. 2008. Animal welfare in poultry production systems: impact of EU standards on world trade. World’s Poult. Sci. J. 64:40–52. Van Horne, P. L. M., and N. Bondt. 2013. Competitiveness of the EU Poultry Meat Sector. LEI Report 2013–068, LEI Wageningen UR, The Hague, The Netherlands. Van Horne, P. L. M., and S. T. Goddijn. 2005. De Braziliaanse Pluimveevleessector (The Brazilian Poultry Meat Sector). Project 30221. LEI Wageningen UR, The Hague, The Netherlands. van Wagenberg, C. P. A., F. M. Brouwer, R. Hoste, and M. L. Rau. 2012. Comparative Analysis of EU Standards in Food Safety, Environment, Animal Welfare and Other Non-Trade Concerns with Some Selected Countries. European Parliament, Directorate General for Internal Policies, Agriculture and Rural Development. Accessed May 2014. http://www.europarl.europa.eu/RegData/etudes/join/2012/ 474542/IPOL-AGRI˙ET(2012)474542˙EN.pdf. Veissier, I., K. K. Jensen, R. Botreau, and P. Sandøe. 2011. Highlighting ethical decisions underlying the scoring of animal welfare R scheme. Anim. Welf. 20:89–101. in the Welfare Quality von Keyserlingk, M. A. G., and M. J. H¨ otzel. 2014. The ticking clock: addressing farm animal welfare in emerging countries. J. Agric. Environ. Ethics. doi:10.1007/s10806-014-9518-7. R 2009. Welfare Quality R Assessment Protocols Welfare Quality. for Poultry (Broilers, Laying Hens). Welfare Quality Consortium. Lelystad, The Netherlands.

Downloaded from http://ps.oxfordjournals.org/ by guest on September 15, 2015

Weeks, and A. Butterworth, ed. CABI Pub. Int., Wallingford, UK. Blokhuis, H. J., I. Veissier, M. Miele, and B. Jones. 2010. The Welfare R project and beyond: Safeguarding farm animal wellQuality being. Acta Agriculturae Scand. Section A 60:129–140. Bonamigo, A., C. B. S. S. Bonamigo, and C. F. M. Molento. 2012. Broiler meat characteristic relevant to the consumer: Focus on animal welfare. R. Bras. Zootec. 41:1044–1050. Botreau, R., I. Veissier, and P. Perny. 2009. Overall assessment of R Anim. Welf. cow welfare: Strategy adopted in Welfare Quality. 18:363–370. Bracke, M. 2009. Animal Welfare in a Global Perspective. Report 240. Wageningen UR Livestock Research, The Netherlands. Croskerry, P., G. Singhal, and S. Mamede. 2013. Cognitive debiasing 2: Impediments to and strategies for change. BMJ Qual. Saf. 22:65–72. De Jong, I. C., H. Gunnink, and V. Hindle. 2014. Implementation R broiler monitor on Dutch broiler farms. of the Welfare Quality Proc. 6th Int. Conf. Assessment of Anim. Welf. at Farm and Group Level, Clermont-Ferrand, France, 70. (Abstr.) De Jong, I. C., T. Perez Moya, H. Gunnink, H. van den Heuvel, V. A. Hindle, M. Mul, and K. van Reenen. 2011. Simplifying the Welfare Quality Assessment Protocol for Broilers. Report 533. Wageningen UR Livestock Research, The Netherlands. de Oliveira Souza, A. P., E. C. de Oliveira Sans, B. R. M¨ uller, and C. F. M. Molento. 2015. Broiler chicken welfare assessment R certified and non-certified farms in Brazil. in GLOBALGAP Anim. Welf. 24:45–54. EFSA Panel on Animal Health and Welfare (AHAW). 2012. Scientific opinion on the use of animal-based measures to assess welfare of broilers. EFSA J. 10:2774. doi:10.2903/j.efsa.2012.2774. Forkman, B., and L. J. Keeling. 2009. Assessment of Animal Welfare R Reports No. 9, Cardiff, Measures for Poultry. Welfare Quality UK. Green, J. A., R. McCracken, and R. T. Evans. 1985. A contact dermatitis of broilers: Clinical and pathological findings. Avian Pathol. 14:23–28. Heath, C. A. E., W. J. Browne, S. Mullan, and D. C. J. Main. 2013. Iceberg indicators: Fact or fiction? Abstr. Welfare Quality Network Worksh., Lille, France, 8. (Abstr.) ITAVI. 2012. La comp´etitivit´e agricole du Mercosur: Le cas des fili`eres d’´elevage br´esiliennes. Rapport DGPAAT 11–02. Institute de l’´elevage, IFIP and ITAVI, France. Accessed April 2014. http://agriculture.gouv.fr/competitivite-agricole-Mercosur. McLean, J. A., C. J. Savory, and N. H. C. Sparks. 2002. Welfare of male and female broiler chickens in relation to stocking density, as indicated by performance, health and behaviour. Anim. Welf. 11:55–73. Meluzzi, A., and F. Sirri. 2009. Welfare of broiler chickens. Ital. J. Anim. Sci. 8:161–173. Reese, E. J. 2012. Techniques for mitigating cognitive biases in fingerprint identification. UCLA Law Rev. 59:1252–1290. Sans, E. C. O., J. F. Federici, F. Dahlke, and C. F. M. Molento. 2014. R proEvaluation of free-range broilers using the Welfare Quality tocol. Braz. J. Poult. Sci. 16:297–305. Sprenger, M., C. Vangestel, and F. A. M. Tuyttens. 2009. Measuring thirst in broiler chickens. Anim. Welf. 18:553–560.

9