Development of a prognostic tool for the occurrence of feather pecking and cannibalism in laying hens

Development of a prognostic tool for the occurrence of feather pecking and cannibalism in laying... Abstract In July 2015, a German voluntary decree stipulated that the keeping of beak-trimmed laying hens after the 1st of January 2017 will no longer be permitted. Simultaneously, the present project was initiated to validate a newly developed prognostic tool for laying hen farmers to forecast, at the beginning of a laying period, the probability of future problems with feather pecking and cannibalism in their flock. For this purpose, we used a computer-based prognostic tool in form of a questionnaire that was easy and quick to complete and facilitated comparisons of different flocks. It contained various possible risk factors that were classified into 3 score categories (1 = “no need for action,” 2 = “intermediate need for action,” 3 = “instant need for action”). For the validation of this tool, 43 flocks of 41 farms were examined twice, at the beginning of the laying period (around the 20th wk of life) and around the 67th wk of life. At both visits, the designated investigators filled out the questionnaire and assessed the plumage condition and the skin lesions (as indicators of occurrence of feather pecking and cannibalism) of 50 laying hens of each flock. The average prognostic score of the first visit was compared with the existence of feather pecking and cannibalism in each flock at the end of the laying period. The results showed that the prognostic score was negatively correlated with the plumage score (r = −0.32; 95% confidence interval [CI]: [−0.56; −0.02]) and positively correlated with the skin lesion score (r = 0.38; 95% CI: [0.09; 0.61]). These relationships demonstrate that a better prognostic score was associated with a better plumage and skin lesion score. After performing a principal component analysis on the single scores, we found that only 6 components are sufficient to obtain highly sensitive and specific prognostic results. Thus, the data of this analysis should be used for creating applicable software for use on laying hen farms. INTRODUCTION The problems of feather pecking and cannibalism in industrially kept poultry, especially in laying hens, are omnipresent. Many factors seem to have an influence on the occurrence of feather pecking and cannibalism during the laying period of farmed laying hens: Some authors say that pecking on conspecifics is induced by stress factors (El-Lethey et al., 2000). Other studies found that the inability to satisfy natural behavior patterns, such as ground pecking or scratching caused by a lack of litter and manipulable material, causes the hens to redirect their needs from the unattractive surrounding towards the other hens (Blokhuis and Arkes, 1984; Blokhuis, 1986; Huber-Eicher and Wechsler, 1997, 1998; Aerni et al., 2000; Ramadan and Von Borell, 2008). This misdirected behavior is often referred to as “frustration behavior,” and Savory (1995) pointed out that frustrated behavior is abnormal behavior, but not of aggressive nature. Furthermore, 2 types of feather pecking can be distinguished (Savory, 1995). Gentle feather pecking seems to be normal behavior for exploration of the environment and can be observed especially in non-enriched surroundings; it is described as soft pecking without pulling and can develop into severe feather pecking (Savory, 1995). Severe feather pecking is defined as intense feather pecking and pulling out feathers, which can lead to bleeding and subsequently to a high chance that this behavior spreads in the flock (Hartcher et al., 2015). Kjaer (2009) replaced the concept of misdirected exploration with the description of a hyperactivity syndrome. He demonstrated that young hens with a high food search activity also show a high probability to develop feather pecking in their adult life. With regard to feather pecking being a result of high stress levels or frustration, several factors of possible influence have been reported: The access to litter material seems to be an important factor for the reduction of stress; in addition, it allows hens to perform natural behavior (Blokhuis and Arkes, 1984; Blokhuis, 1986; Huber-Eicher and Wechsler, 1997; Aerni et al., 2000; El-Lethey et al., 2000; Lugmair, 2009; Brenninkmeyer and Knierim, 2015; Louton et al., 2017). Another factor that could have an influence on pecking at flock mates might be the offering of pecking material made of manipulable substances (Huber-Eicher and Wechsler, 1998; Chow and Hogan, 2005; Steenfeldt et al., 2007). Conversely, Plattner (2015) found that the frequency of feather pecking at conspecifics was much higher during the presence than during the absence of a bale of straw. The enhanced pecking might be explained as a defense of resources. Many studies also found a connection between feather pecking and dust bathing. The possibility for laying hens to perform dust bathing behavior is correlated to the levels of corticosterone, i.e., an absence of a dust bath increases their stress level and is consequently correlated with an increase of stereotypic pecking (Vestergaard et al., 1997). Sanotra et al. (1995) found that the provision of sand as dust bathing substrate could distract the hens from feather pecking, whereas Huber-Eicher and Wechsler (1997) found that only the presence of manipulable pecking material could be correlated with the occurrence of feather pecking behavior. Furthermore, the possibility of using a winter garden and an outside area is negatively correlated to the occurrence of feather pecking (Green et al., 2000; Bestman and Wagenaar, 2003; Bestman and Wagenaar, 2014; Brenninkmeyer and Knierim, 2015; Lenz, 2015). Keeping 2 laying hen strains in one flock was related to an increase of the occurrence of feather pecking (Lenz, 2015). According to Blokhuis and Arkes (1984), cannibalism is simply a redirected behavior of higher intensity than feather pecking. Furthermore, a connection between poor feather condition and the occurrence of cannibalism has been shown (Kjaer and Sørensen, 2002). The period of molting may provoke a transition from feather pecking to cannibalism due to the high chance of bleeding, because the new feathers are very soft and their blood vessels rupture easily upon pecking (Huber-Eicher and Wechsler, 1997). The multiple possible risk factors mentioned above suggest that feather pecking and cannibalism have multifactorial origins (Savory and Mann, 1997; Rodenburg et al., 2013; Schwarzer et al., 2015). These behaviors cause not only pain and suffering for the animals (Allen and Perry, 1975; Spindler et al., 2013) but also economic losses for the farmer. Hen mortality increases (Niebuhr et al., 2006; Stadig et al., 2015), egg production decreases (El-Lethey et al., 2000; Niebuhr et al., 2006), and food intake rises (Tauson and Svensson, 1980). The severity of problems for animals and farmers, especially since the voluntary ban of beak trimming in July 2015 (Zentralverband der deutschen Geflügelwirtschaft, 2015), highlights the need for improved management and early detection of incipient problems. The best way to do so is bringing different flocks and managements onto a comparable level and to evaluate the differences. This is possible by the help of benchmarks, which were already used (Leenstra et al., 2014). With the aid of benchmarks, Holle et al. (2008) investigated risk factors that may cause problems in laying hen farming. Using a benchmarking system, they compared the health status of 20 organic kept laying hen flocks (European Commission, 2007) by assessing various health parameters, namely, feather condition, skin trauma, footpad abscesses, internal parasites, and the existence of Dermanyssus gallinae. Their goal was to establish so called “animal health schemes” that were based on a questionnaire, which was filled out by the investigators with help from the farmers. In order to foster the health of the animals, they generated a plan based on the investigators’ recommendations. This system also facilitated the comparison between different laying hen flocks, highlighting specific skills and drawbacks. They found out that “animal health schemes” work correctly only if the farmer is in a permanent communication with the consultant, and the animal health schemes undergo stable development. Keppler et al. (2014) developed the “MTool” to evaluate the health of non-beak-trimmed laying hens based on selected factors on organic and conventional farms. This tool is as well based on potential risk factors, which got evaluated by comparing the actual hen condition with the condition of what animal health should look like. They scored each factor with 3 different categories: “no need for action”; “medium-term need for action”; and “instant need for action.” If the hen condition got scored with medium-term need for action or worse, the management and husbandry condition needed to be checked and got scored the same way. Both tools were developed to sensitize farmers to closely watch their laying hens and improve the topics for which they did not get good rates. In line with this goal, the present study aimed to establish an easily realizable evaluation system with general management questions and an objective rating of the flocks and barns. It also contains recommendations for farmers to improve their management, especially in topics assessed not good. In the presented study, this was visualized in a traffic light system, similar to the 3 scores from Keppler et al. (2014): Each topic got scored into “1 = no need for action (green),” “2 = intermediate need for action (yellow),” and “3 = instant need for action (red).” All of them would be presented to the farmer, with the respective descriptions. These descriptions can also be used as recommendations for actions to be taken for improvement in a given topic. For example a farmer, who scored 3 in a given topic, can use the description of the better score 2, or score 1, as a guideline for improvement. For this purpose, we developed a prognostic tool based on a questionnaire with 53 benchmarks and simple assessments that allow a comparison between flocks based on combined selected factors. The novel aspect of this study is that we validated the comparison and the correlation with the relative occurrence of feather pecking or cannibalism by re-assessing the occurrence of these problematic behaviors at the end of the laying period. This kind of forecasting system should be a useful tool for laying hen farmers to identify farming conditions that could be improved and to sensitize them to the problems their animals may have. As already mentioned, feather pecking and cannibalism seem to be caused by multifactorial conditions and thus are difficult to eliminate once they occur. Hence, it is sometimes not enough to give farmers general advice or directions on possible actions to mitigate manifested problems to get rid of feather pecking and cannibalism behavior. It would be easier to prevent them before they manifest themselves, if it is possible to forecast their manifestation. It is necessary to analyze and identify the critical aspects of each farm individually to limit and maybe even solve the problems. A prognostic tool that can identify factors before the problems occur is the most preferable course of action for both the laying hens and the farmers. And this procedure was tried to establish in the presented study. MATERIALS AND METHODS Course of the project The data collection ran over 2 yr, from July 2015 until May 2017, and included 43 flocks of 41 commercially run conventional farms in Bavaria, Germany. The project was planned and realized by the Chair of Animal Welfare, Animal Behavior, Animal Hygiene and Animal Husbandry of the Ludwig-Maximilians-University in Munich, Germany. All farmers voluntarily participated in this project. Flocks were assessed twice during the laying period (all visits in each farm were performed by the same 3 investigators), the first time approx. 7 to 10 d after the hens had been transferred to the laying barn facility, i.e., in their 20th wk of life, and the second time on average in the 67th wk of life, before slaughter or molting. Due to an outbreak of avian influenza in November 2016 and the subsequent obligation to keep the animals inside their barns, the time for the second inspection varied (between the 57th and 83rd wk of life). Methods During both the first and the second visits, the same dataset was collected. The investigators filled out a detailed questionnaire, by asking the farmers several questions, including the topics stocking rate, management, rehousing, feeding, pecking material, litter, barn climate, lighting, and animal behavior (for detailed questions and the associated categories, see Table 1). Afterwards, the investigators entered the barn and evaluated all aspects of the flock's behavior, such as nervousness or mutual pecking, the setting of the barn, such as litter quality or the existence of pecking material, and the barn's climate by mutual agreement. The responses from the questionnaire and the results of the flock and farm evaluation were directly recorded in the computer-based prognostic tool. Table 1. The questions for the prognostic tool (53 questions) and the related scores with classifications. The number of flocks, their percentage per value and the total number of flocks per question (n) are given. The 35 scores, which were used for the principal component analysis, are underlined. Question  Score  Classification  Number of flocks  Percentage share  n  Stocking rate  Hens per usable square meter  1  <9 hens per usable square meter  38  88.37  43  2  9 hens per usable square meter  5  11.63    3  >9 hens per usable square meter  0  0    Hens per usable base square meter  1  <18 hens per usable base square meter  28  93.34  30  2  18 hens per usable base square meter  1  3.33    3  >18 hens per usable base square meter  1  3.33    Time with temporarily higher number of hens per usable square meter  1  total area provided by day 1  19  44.19  43  2  total area provided after ≤3 weeks  21  48.83    3  total area provided after >3 weeks  3  6.98    Temporarily higher number of hens per usable square meter  1  total area temporarily not provided with <9 hens per usable square meter  1  4.17  24  2  total area temporarily not provided with 9 hens per usable quare meter  4  16.67    3  total area temporarily not provided with >9 hens per usable square meter  19  79.16    Management  Do you keep laying hen strains mixed?  1  no  32  74.42  43  2  yes  11  25.58    Accessibility of the enclosure  1  good  22  55  40  2  medium  12  30    3  bad  6  15    How long is the cleaning period between 2 laying periods?  1  >2 weeks  28  65.12  43  2  1–2 weeks  10  23.25    3  <1 week  5  11.63    Do you have a cadaver bin?  1  yes  37  86.05  43  2  no  6  13.95    At what time do you open the nests?  1  nest opening ≥1 h before turning on the lights  35  81.39  43  2  nest opening <60 min before turning on the lights  7  16.28    3  nest opening at the same time or after turning on the lights  1  2.33    Do you document the death of animals?  1  yes, always  38  88.37  43  2  irregular  4  9.3    3  never  1  2.33    Did you have problems with feather pecking in previous period?  1  no  13  30.23  43  3  yes  30  69.77    Do you have problems with Dermanyssus gallinae  1  no  9  20.93  43  2  yes  34  79.07    Do you offer permanent access to a winter garden?  1  yes  21  48.84  43  2  no  7  16.28    3  no existing winter garden  15  34.88    Do you offer permanent access to an outside area?  1  yes  9  20.93  43  2  no  15  34.88    3  no existing outside area  19  44.19    Were sick animals found after the morning tour of the person in charge?  1  no  42  97.67  43  3  yes  1  2.33    Were dead animals found after the morning tour of the person in charge?  1  no  42  97.67  43  3  yes  1  2.33    How many times per day does the person in charge investigate the barn?  1  >1x per day  39  90.7  43  2  1x per day  4  9.3    3  <1x per day  0  0    How many times per day does the person in charge investigate the litter area?  1  >1x per day  39  90.7  43  2  1x per day  4  9.3    3  <1x per day  0  0    How many times per day does the person in charge investigate the aviary?  1  >1x per day  31  86.11  36  2  1x per day  4  11.11    3  <1x per day  1  2.78    How many times per day does the person in charge investigate the winter garden?  1  >1x per day  18  69.23  26  2  1x per day  8  30.77    3  <1x per day  0  0    How many minutes per day does the person in charge spend with the laying hens?  1  >3.9 min per 1,000 hens  42  100  42  2  3.9 min per 1,000 hens  0  0    3  <3.9 min per 1,000 hens  0  0    Do you have separate clothing for each livestock?  1  separate clothing for each livestock  16  38.1  42  2  separate clothing but is worn outside of the livestock  23  54.76    3  no separate clothing  3  7.14    Laying hen rehousing  How old were the laying hens at delivery?  1  17th to 18th week of life  34  79.07  43  2  ≤± 3 weeks  9  20.93    3  >± 3 weeks  0  0    Do you weigh the laying hens at delivery?  1  yes, weighing of single hens  10  23.26  43  2  yes, weighing of group of hens  7  16.28    3  no  26  60.46    Average weight of laying hens at delivery  1  ≥normal weight  3  25  12  2  ≤10% under the normal weight  7  58.33    3  >10% under the normal weight  2  16.67    Feeding  How many times does the feeding chain move per day?  1  >6x per day  14  36.84  38  2  4–6 x per day  24  63.16    3  <4x per day  0  0    Do you offer a feeding intermission?  1  always  29  67.44  43  2  sometimes  2  4.65    3  never  12  27.91    How many laying hens come to the moving feeding chain?  1  all of them  2  14.29  14  2  most of them  11  78.57    3  only a few  1  7.14    Do you document the food consumption?  1  yes, with a food or silo scale  18  41.86  43  2  yes, but only an estimation  11  25.58    3  no documentation  14  32.56    Pecking material  Do you spread grain?  1  regularly  17  40.48  42  2  irregularly  8  19.04    3  never  17  40.48    Do you offer pecking material?  1  regularly  32  74.42  43  2  irregularly  1  2.32    3  never  10  23.26    How much pecking material do you offer?  1  >1 element per 1,000 hens  21  48.84  43  2  ≤1 element per 1,000 hens  9  20.93    3  no pecking material offered  13  30.23    Do you offer a dust bath?  1  yes  20  46.51  43  3  no  23  53.49    Litter  Do you change litter regularly?  1  yes, completly  25  59.52  42  2  yes, partially  16  38.1    3  no change  1  2.38    Do you have problems with litter plaques?  1  no litter plaques  30  75  40  2  partial litter plaques  8  20    3  complete litter plaques  2  5    How deep is the litter at the beginning of the laying period?  1  >2 cm  10  24.39  41  2  ≤2 cm  28  68.29    3  no litter offered  3  7.32    Climate  air quality  1  good  36  83.72  43  2  medium  5  11.63    3  bad  2  4.65    ammonia noticable  1  slightly  38  88.37  43  2  moderately  3  6.98    3  intensely  2  4.65    dust  1  slight  36  83.72  43  2  moderate  7  16.28    3  intense  0  0    noise  1  quiet  35  81.39  43  2  moderate  7  16.28    3  noisy  1  2.33    infiltration  1  slight  30  69.77  43  2  moderate  11  25.58    3  intense  2  4.65    Lighting  Do you have dimmable lighting?  1  yes  31  72.09  43  2  partial  7  16.28    3  no  5  11.63    Do you have sunspots?  1  no  33  76.74  43  2  yes  10  23.26    Are the nests dark?  1  yes  23  57.5  40  2  partially  16  40    3  no  1  2.5    Is the litter area lighted evenly?  1  yes  30  69.77  43  2  no  13  30.23    Behavior of laying hens  How is the visitation of the litter area?  1  intensive  26  61.9  42  2  moderate  11  26.2    3  slight  5  11.9    Is the flock more nervous than usual?  1  no  26  60.47  43  3  yes  17  39.53    Do the laying hens peck on the investigator?  1  no pecking  28  65.12  43  2  slight pecking  11  25.58    3  intense pecking  4  9.3    Do the laying hens peck each other?  1  no pecking  27  62.79  43  2  slight pecking  16  37.21    3  intense pecking  0  0    Do the laying hens chaise each other?  1  no chasing  39  90.7  43  2  some chasing  4  9.3    3  intense chasing  0  0    Ground feathers: Test 3 different areas in the barn and count on each measuring point the ground feathers per 1 square meter  1  ≥10 ground feathers at all 3 measuring points found  17  42.5  40  2  <10 ground feathers found at minimum 1 measuring point  13  32.5    3  minimum 1 measuring point with no ground feathers  10  25    Are the laying hens easy to catch?  1  easy  38  88.37  43  2  medium  5  11.63    3  not easy  0  0    Touch-test: Squat and move slowly; count how many laying hens you are able to touch  1  ≥1 laying hen could be touched on average  29  70.73  41  2  <1 laying hen could be touched on average  12  29.27    Question  Score  Classification  Number of flocks  Percentage share  n  Stocking rate  Hens per usable square meter  1  <9 hens per usable square meter  38  88.37  43  2  9 hens per usable square meter  5  11.63    3  >9 hens per usable square meter  0  0    Hens per usable base square meter  1  <18 hens per usable base square meter  28  93.34  30  2  18 hens per usable base square meter  1  3.33    3  >18 hens per usable base square meter  1  3.33    Time with temporarily higher number of hens per usable square meter  1  total area provided by day 1  19  44.19  43  2  total area provided after ≤3 weeks  21  48.83    3  total area provided after >3 weeks  3  6.98    Temporarily higher number of hens per usable square meter  1  total area temporarily not provided with <9 hens per usable square meter  1  4.17  24  2  total area temporarily not provided with 9 hens per usable quare meter  4  16.67    3  total area temporarily not provided with >9 hens per usable square meter  19  79.16    Management  Do you keep laying hen strains mixed?  1  no  32  74.42  43  2  yes  11  25.58    Accessibility of the enclosure  1  good  22  55  40  2  medium  12  30    3  bad  6  15    How long is the cleaning period between 2 laying periods?  1  >2 weeks  28  65.12  43  2  1–2 weeks  10  23.25    3  <1 week  5  11.63    Do you have a cadaver bin?  1  yes  37  86.05  43  2  no  6  13.95    At what time do you open the nests?  1  nest opening ≥1 h before turning on the lights  35  81.39  43  2  nest opening <60 min before turning on the lights  7  16.28    3  nest opening at the same time or after turning on the lights  1  2.33    Do you document the death of animals?  1  yes, always  38  88.37  43  2  irregular  4  9.3    3  never  1  2.33    Did you have problems with feather pecking in previous period?  1  no  13  30.23  43  3  yes  30  69.77    Do you have problems with Dermanyssus gallinae  1  no  9  20.93  43  2  yes  34  79.07    Do you offer permanent access to a winter garden?  1  yes  21  48.84  43  2  no  7  16.28    3  no existing winter garden  15  34.88    Do you offer permanent access to an outside area?  1  yes  9  20.93  43  2  no  15  34.88    3  no existing outside area  19  44.19    Were sick animals found after the morning tour of the person in charge?  1  no  42  97.67  43  3  yes  1  2.33    Were dead animals found after the morning tour of the person in charge?  1  no  42  97.67  43  3  yes  1  2.33    How many times per day does the person in charge investigate the barn?  1  >1x per day  39  90.7  43  2  1x per day  4  9.3    3  <1x per day  0  0    How many times per day does the person in charge investigate the litter area?  1  >1x per day  39  90.7  43  2  1x per day  4  9.3    3  <1x per day  0  0    How many times per day does the person in charge investigate the aviary?  1  >1x per day  31  86.11  36  2  1x per day  4  11.11    3  <1x per day  1  2.78    How many times per day does the person in charge investigate the winter garden?  1  >1x per day  18  69.23  26  2  1x per day  8  30.77    3  <1x per day  0  0    How many minutes per day does the person in charge spend with the laying hens?  1  >3.9 min per 1,000 hens  42  100  42  2  3.9 min per 1,000 hens  0  0    3  <3.9 min per 1,000 hens  0  0    Do you have separate clothing for each livestock?  1  separate clothing for each livestock  16  38.1  42  2  separate clothing but is worn outside of the livestock  23  54.76    3  no separate clothing  3  7.14    Laying hen rehousing  How old were the laying hens at delivery?  1  17th to 18th week of life  34  79.07  43  2  ≤± 3 weeks  9  20.93    3  >± 3 weeks  0  0    Do you weigh the laying hens at delivery?  1  yes, weighing of single hens  10  23.26  43  2  yes, weighing of group of hens  7  16.28    3  no  26  60.46    Average weight of laying hens at delivery  1  ≥normal weight  3  25  12  2  ≤10% under the normal weight  7  58.33    3  >10% under the normal weight  2  16.67    Feeding  How many times does the feeding chain move per day?  1  >6x per day  14  36.84  38  2  4–6 x per day  24  63.16    3  <4x per day  0  0    Do you offer a feeding intermission?  1  always  29  67.44  43  2  sometimes  2  4.65    3  never  12  27.91    How many laying hens come to the moving feeding chain?  1  all of them  2  14.29  14  2  most of them  11  78.57    3  only a few  1  7.14    Do you document the food consumption?  1  yes, with a food or silo scale  18  41.86  43  2  yes, but only an estimation  11  25.58    3  no documentation  14  32.56    Pecking material  Do you spread grain?  1  regularly  17  40.48  42  2  irregularly  8  19.04    3  never  17  40.48    Do you offer pecking material?  1  regularly  32  74.42  43  2  irregularly  1  2.32    3  never  10  23.26    How much pecking material do you offer?  1  >1 element per 1,000 hens  21  48.84  43  2  ≤1 element per 1,000 hens  9  20.93    3  no pecking material offered  13  30.23    Do you offer a dust bath?  1  yes  20  46.51  43  3  no  23  53.49    Litter  Do you change litter regularly?  1  yes, completly  25  59.52  42  2  yes, partially  16  38.1    3  no change  1  2.38    Do you have problems with litter plaques?  1  no litter plaques  30  75  40  2  partial litter plaques  8  20    3  complete litter plaques  2  5    How deep is the litter at the beginning of the laying period?  1  >2 cm  10  24.39  41  2  ≤2 cm  28  68.29    3  no litter offered  3  7.32    Climate  air quality  1  good  36  83.72  43  2  medium  5  11.63    3  bad  2  4.65    ammonia noticable  1  slightly  38  88.37  43  2  moderately  3  6.98    3  intensely  2  4.65    dust  1  slight  36  83.72  43  2  moderate  7  16.28    3  intense  0  0    noise  1  quiet  35  81.39  43  2  moderate  7  16.28    3  noisy  1  2.33    infiltration  1  slight  30  69.77  43  2  moderate  11  25.58    3  intense  2  4.65    Lighting  Do you have dimmable lighting?  1  yes  31  72.09  43  2  partial  7  16.28    3  no  5  11.63    Do you have sunspots?  1  no  33  76.74  43  2  yes  10  23.26    Are the nests dark?  1  yes  23  57.5  40  2  partially  16  40    3  no  1  2.5    Is the litter area lighted evenly?  1  yes  30  69.77  43  2  no  13  30.23    Behavior of laying hens  How is the visitation of the litter area?  1  intensive  26  61.9  42  2  moderate  11  26.2    3  slight  5  11.9    Is the flock more nervous than usual?  1  no  26  60.47  43  3  yes  17  39.53    Do the laying hens peck on the investigator?  1  no pecking  28  65.12  43  2  slight pecking  11  25.58    3  intense pecking  4  9.3    Do the laying hens peck each other?  1  no pecking  27  62.79  43  2  slight pecking  16  37.21    3  intense pecking  0  0    Do the laying hens chaise each other?  1  no chasing  39  90.7  43  2  some chasing  4  9.3    3  intense chasing  0  0    Ground feathers: Test 3 different areas in the barn and count on each measuring point the ground feathers per 1 square meter  1  ≥10 ground feathers at all 3 measuring points found  17  42.5  40  2  <10 ground feathers found at minimum 1 measuring point  13  32.5    3  minimum 1 measuring point with no ground feathers  10  25    Are the laying hens easy to catch?  1  easy  38  88.37  43  2  medium  5  11.63    3  not easy  0  0    Touch-test: Squat and move slowly; count how many laying hens you are able to touch  1  ≥1 laying hen could be touched on average  29  70.73  41  2  <1 laying hen could be touched on average  12  29.27    View Large Factors that could be a potential risk for feather pecking and cannibalism had been selected before the project started and classified into 3 score categories: “no need for action” (Score 1), “intermediate need for action” (Score 2), and “instant need for action” (Score 3) (Table 1). Hereby, “action” means the recommendation for the farmer to improve his management in this factor's topic. The way to improve in this topic is directly presented in the description of the 3 categories, i.e., he needs to follow the description of the next higher score. For example, if the farmer does not weigh the laying hens at delivery, he receives the score “3 = instant need for action” in this topic, and this question will be presented in the color red. If the farmer wants to improve, he is able to see the description of number “2 = weighing of group of hens” (yellow) and of number “1 = weighing of single hens” (green) and knows what to do. Hereby, the farmer can see his “red topics,” even if the prognostic score is presentable and could show action of improvement also on his own initiative. These score categories were based on law guidelines (German Order on the Protection of Animals and the Keeping of Production Animals, 2006), professional recommendations (LWK-Niedersachsen, 2016; NMELV, 2017), previous studies (Lenz, 2015; Szczepanek, 2016), and scientific literature (Heerkens et al., 2015). For visualization, the score for each factor was marked in green (1), yellow (2), and red (3), and the spreadsheet was presented to the participating farmers on the spot to show them how they were performing in direct but anonymous comparison with other farmers (benchmarking). In this study, the spreadsheet was presented only on the second visit, to avoid that the results and the comparison to the other farmers could falsify the action of the farmers’ management throughout the laying period. An average prognostic score for each flock (best 1.0, worst 3.0) was calculated. It is based on the evaluated average of the given scores (1, 2, or 3) of 53 questions (Table 1). Subsequently, 50 (flock size > 500 hens) or 20 laying hens (flock size ≤ 500 hens) were selected randomly, and their plumage and skin conditions were determined according to a modified “hen score” from Gunnarsson (2000a) and Blokhuis et al. (2007). Hens of mixed laying hen strains were selected according to their proportion in the group. The hens were assessed by all 3 investigators, whose inter-rater reliability was tested in 100 brown egg and 100 white egg laying hen strains. The plumage condition was classified into 5 score categories: 5 = no special abnormality; 4 = >5 damaged feathers or featherless area Ø ≤ 1 cm; 3 = featherless area Ø > 1 − ≤ 5 cm; 2 = featherless area Ø > 5 cm and ≤75% featherless; 1 = >75% featherless. The classification was assessed for 3 body regions (the neck, the back, and the wing coverts) because feather damages in these body regions are likely to be caused by feather pecking (Ramadan and Von Borell, 2008). The sum of the scores for all 3 body regions formed the “plumage score,” i.e., the plumage score 3 equaled the largest plumage damage and therefore the most severe feather pecking problems. The best score 15 equaled an intact plumage and no or only minor feather pecking in the flock. The plumage score was averaged per flock. An average plumage score ≤10 was defined as a flock with severe feather pecking problems. The occurrence of wounds and their size were scaled using a separate score, the “skin lesion score.” In this case, the entire body (except the head and legs) was scanned for wounds, and only skin rupture,, but not discoloration, was considered. A flock that had ≥10% hens with one or more wounds of >0.5 cm was defined as having problems with cannibalism. Statistical Analysis The prognostic tool was designed with Microsoft Excel 2010 (Microsoft Corporation, Redmond, WA). For all analyses, the R language for statistical computing (R Core Team, 2015) was used. The relation between the prognostic score and the plumage or skin lesion score was first analyzed in an exploratory manner using the Pearson correlation coefficient. Prognostic quality of this relation was subsequently investigated in more detail by using linear regression. To this end, the data from 43 flocks were randomly separated into training (70%) and test (30%) datasets. The training dataset was used to estimate the linear model that then was used to predict the plumage and skin lesion scores for the test dataset. Prediction error was measured by the mean squared error and the mean percent error. This procedure was repeated 5,000 times with different training and test datasets to obtain an approximation of the distribution of these errors. In addition, logistic regression models were used to estimate prediction models for feather pecking and cannibalism with the prognostic score as the only predictor and were presented with odds ratio (OR) and confidence interval (CI). Here, prediction quality was measured by 4 criteria: sensitivity, calculated as true positive (TP)/(TP + false negative [FN]); specificity, calculated as true negative (TN)/(TN + false positive [FP]); accuracy, calculated as (TN + TP)/(TN + TP + FN + FP); and the dice coefficient (Dice, 1945), calculated as 2 * TP/(2 * TP + FN + FP). Randomized partition of the complete data into training and test data, as well as repetition, was performed as described above. For analyses of single scores, we first performed a dimensional reduction by means of a principal component analysis. By using this method, the 35 tested scores were transformed into a few components, the principal components. Usually, a small number of components is sufficient to describe the information of the complete data. Up to the first 10 of these components were then used as predictors in logistic regression models for the prediction of feather pecking and cannibalism as described for the prognostic score. Inter-rater reliability was analyzed by means of the PABAK (prevalence-adjusted and bias-adjusted kappa), which is also called kappanor (Byrt et al., 1993). This statistical tool was employed, because as the number of categories used is unpredictable in research with live animals and in this case is rather high; the kappa score should be corrected for prevalence and bias. The equation for more than 2 categories (Gunnarsson et al., 2000b) is   \begin{equation*} {\rm{PABAK}} = \left( {{{\left( {{\rm{k}}{{\rm{p}}_0}} \right) - 1}\, / {\left( {{\rm{k}} - 1} \right)}}}\right. . \end{equation*} Where: k represents the number of assessed categories, and p0 indicates the relation between the observed agreements. For the assessment of skin lesion only, 2 categories were used, in which case the equation shortens to   \begin{equation*} {\rm{PABAK}} = 2{{\rm{p}}_0} - 1. \end{equation*} RESULTS General results The assessed 43 flocks of 41 farms included conventional (79.1%) or organic (20.9%) farms, which could be either floor (32.6%) or aviary (67.4%) barns of varying size (from 200 up to 6,000 non-beak-trimmed animals per barn or barn section); 79.1% of the flocks hosted over 500 laying hens, and 20.9% hosted ≤500 laying hens (here, both flock sizes were hosted by both farming systems). This project included laying hens of the strains Lohmann Brown Classic, Lohmann Selected Leghorn, Bovans Brown, Dekalb White, Lohmann Brown Plus, Lohmann Brown Extra, Lohmann Dual, and Lohmann Sandy; 25.6% of the flocks were flocks with mixed laying hen strains (Lohmann Brown Classic + Lohmann Selected Leghorn, Lohmann Brown Classic + Lohmann Dual, or Bovans Brown + Dekalb White), 67.4% were brown egg poultry flocks, and 9.3% were white egg poultry flocks. For descriptive results for each score (53) of the prognostic tool, see Table 1 in the Appendix. As depicted in Table 2, the results of the PABAK inter-rater reliability test all vary between 0.88 and 0.99, which is within the range suggested by Landis and Koch (1977) for almost perfect reliability scores [0.81-1]. Table 2. Inter-rater reliability scores based on PABAK (prevalence-adjusted and bias-adjusted kappa). The three investigators (A, B, and C) assessed 100 brown and 100 white egg laying hen strains on their plumage condition in 3 body regions and on the existence and size of possible skin lesions in 9 body regions. Agreements and disagreements with the corresponding percentage and results of PABAK are presented for each investigator combination.   Plumage (n = 300)  Skin lesion (n = 900)    Number of agreements (%)  Number of disagreements (%)  PABAK  Number of agreements (%)  Number of disagreements (%)  PABAK    Brown egg laying hen strain  A/B  280 (93.33%)  20 (6.67%)  0.92  891 (99%)  9 (1%)  0.98  A/C  276 (92%)  24 (8%)  0.9  888 (98.67%)  12 (1.33%)  0.97  B/C  281 (93.67%)  9 (6.33%)  0.92  891 (99%)  9 (1%)  0.98    White egg layng hen strain  A/B  276 (92%)  24 (8%)  0.9  895 (99.44%)  5 (0.56%)  0.99  A/C  270 (90%)  30 (10%)  0.88  892 (99.11%)  8 (0.89%)  0.98  B/C  277 (92.33%)  23 (7.67%)  0.9  893 (99.22%)  7 (0.78%)  0.98    Plumage (n = 300)  Skin lesion (n = 900)    Number of agreements (%)  Number of disagreements (%)  PABAK  Number of agreements (%)  Number of disagreements (%)  PABAK    Brown egg laying hen strain  A/B  280 (93.33%)  20 (6.67%)  0.92  891 (99%)  9 (1%)  0.98  A/C  276 (92%)  24 (8%)  0.9  888 (98.67%)  12 (1.33%)  0.97  B/C  281 (93.67%)  9 (6.33%)  0.92  891 (99%)  9 (1%)  0.98    White egg layng hen strain  A/B  276 (92%)  24 (8%)  0.9  895 (99.44%)  5 (0.56%)  0.99  A/C  270 (90%)  30 (10%)  0.88  892 (99.11%)  8 (0.89%)  0.98  B/C  277 (92.33%)  23 (7.67%)  0.9  893 (99.22%)  7 (0.78%)  0.98  View Large Plumage and skin lesion scores Based on the 2 parameters plumage score and skin lesion score, the assessed flocks were divided into groups with or without problems. Fourteen of the 43 flocks (32.6%) had a plumage score ≤10 at the second visit, indicating a problem with severe feather pecking at the end of the laying period. In comparison, none of the assessed flocks had a plumage score ≤10 at the first visit, i.e., no flock showed problems with severe feather pecking at the beginning of the laying period (approx. 20th wk of life). The average plumage score was 10.65 points with a range from 7.08 up to 13.48 points at the second visit (Figure 1A). Figure 1. View largeDownload slide A (above) + B (below): Average plumage score (1A) and average skin lesion score (1B) (y-axis) of every farm (x-axis) with number of n showing the number of assessed laying hens per flock at the second visit. Each farm has its own number; farms with 2 assessed flocks are identified by different decimal places. The plumage score indicates the average feather condition per assessed flock for the body regions neck, back, and wing coverts. A plumage score ≤10 presents a problem with severe feather pecking in this flock. The skin lesion score indicates the percentage share of animals in the flock with skin lesions >0.5 cm. A flock with ≥10% of animals with big skin lesions shows problems with cannibalism. Figure 1. View largeDownload slide A (above) + B (below): Average plumage score (1A) and average skin lesion score (1B) (y-axis) of every farm (x-axis) with number of n showing the number of assessed laying hens per flock at the second visit. Each farm has its own number; farms with 2 assessed flocks are identified by different decimal places. The plumage score indicates the average feather condition per assessed flock for the body regions neck, back, and wing coverts. A plumage score ≤10 presents a problem with severe feather pecking in this flock. The skin lesion score indicates the percentage share of animals in the flock with skin lesions >0.5 cm. A flock with ≥10% of animals with big skin lesions shows problems with cannibalism. The skin lesion score (as a scale for the percentage of hens in a flock with wounds >0.5 cm) showed that at the second visit, 24 flocks (55.8%) had a minimum of 10% with one or more wounds >0.5 cm, and hence a problem with cannibalism; at the first visit, it was one flock (2.3%). On average, the assessed flocks at the second visit had a skin lesion score of 14%. The highest percentage of hens with big wounds >0.5 cm was 53.3%. In 7 flocks, no laying hen with wounds >0.5 cm was found (Figure 1B). Prognostic tool: Prognostic score The flock with the best (i.e., lowest) prognostic score was scored with 1.29 and the worst with 1.69 points. A correlation analysis showed that the prognostic score was negatively correlated with the plumage score (r = −0.32; 95% CI: [−0.56; −0.02]), meaning that higher (worse) prognostic scores were correlated with lower (worse) plumage scores. Between the prognostic score and the skin lesion score, we found a positive correlation (r = 0.38; 95% CI: [0.09; 0.61]), i.e., higher (worse) prognostic scores were correlated with higher (worse) skin lesion scores (Figure 2). Consequently, we can say that flocks with feather pecking or cannibalism or both have a higher (worse) prognostic score (Figure 3). Figure 2. View largeDownload slide Correlation between the average prognostic score and the plumage score on the left side (negative) and between the average prognostic score and the skin lesion score on the right side (positive). The black line indicates the simple regression line. Figure 2. View largeDownload slide Correlation between the average prognostic score and the plumage score on the left side (negative) and between the average prognostic score and the skin lesion score on the right side (positive). The black line indicates the simple regression line. Figure 3. View largeDownload slide Distribution of the average prognostic scores (y-axis) along categories of feather pecking and cannibalism. The prognostic score has a significant positive relation to the occurrence of feather pecking (P = 0.03) and cannibalism (P = 0.03). These occurrences for each problem are presented with yes/no. Box plots with thick black lines show the median, boxes show the upper and lower quartiles, whiskers show the maximum and minimum values, and circles show the outliers. Figure 3. View largeDownload slide Distribution of the average prognostic scores (y-axis) along categories of feather pecking and cannibalism. The prognostic score has a significant positive relation to the occurrence of feather pecking (P = 0.03) and cannibalism (P = 0.03). These occurrences for each problem are presented with yes/no. Box plots with thick black lines show the median, boxes show the upper and lower quartiles, whiskers show the maximum and minimum values, and circles show the outliers. Further analysis using logistic regression models revealed that the chance of developing feather pecking was 2.5 times higher when the prognostic score rose 0.1 points (OR = 2.5; 95% CI: [1.14; 6.33]). For the chance of developing cannibalism, we got a similar result (OR = 2.4; 95% CI: [1.14; 6.07]). When the prognostic score exceeded 1.57 points, the probability of feather pecking in the laying period was 0.5; for cannibalism, this threshold value was 1.45 points. Although correlations were significant and plausible, further analysis revealed that the prognostic quality of the prognostic score was not good enough. For example, the linear relationship between the prognostic score and the plumage score explained only about 10% of the variation of the plumage score. For the skin lesion score, this value was about 14%. In addition, predictions by the fitted logistic regression models were not satisfactory (average sensitivity and specificity <0.75). Therefore, we tested if other combinations of the possible 53 risk factors might raise the prognostic quality (results are presented in the following section). Prognostic tool: Single scores For the analysis of the single scores, only scores from complete datasets were used. Not every score could be assigned in each flock (e.g., not every poultry farmer offers litter directly), which reduced the 53 scores to 35 (Table 1 underlined scores). We performed a principal component analysis as implemented in R’s “prcomp” function on these scores. This method applies singular value decomposition on the full data matrix, which yields the principal component transformations. Essentially, these transformations are linear combinations of the single scores, but, in contrast to the prognostic score, different weights (loadings) were assigned to each single score, whereas the prognostic score uses the same weights. By default, this analysis yields as many components as single scores, in our case, 35. However, the order of the components is directly related to their importance with respect to the information in the data, i.e., the first component explains the most variation of the data matrix, the second the second most, and so on. Usually, only a relatively small number of components was needed to explain the main characteristics of the original data. This allows us to reduce the dimension of the data even further by choosing only the first few components. Figure 4 shows the prognostic quality that resulted when up to 10 components were used within the logistic regression setup for predicting feather pecking and cannibalism. For feather pecking, sensitivity and specificity yielded excellent results for a minimum of 6 components. For cannibalism, one more component was needed to obtain comparable values. Assessing the overall classification quality by the criteria accuracy and Dice coefficient, we reached the same conclusion, i.e., that 6 and 7 components were sufficient for predicting feather pecking and cannibalism, respectively. Figure 4. View largeDownload slide Prognostic quality of logistic regression models with different numbers of components as covariates (x-axis) for predicting feather pecking (top row) and cannibalism (bottom row). For each component, the quality criteria sensitivity, specificity, accuracy, and Dice coefficient are depicted (y-axis). Box plots with thick black lines show the median, boxes show the upper and lower quartiles, whiskers show the maximum and minimum values, and circles show the outliers. Figure 4. View largeDownload slide Prognostic quality of logistic regression models with different numbers of components as covariates (x-axis) for predicting feather pecking (top row) and cannibalism (bottom row). For each component, the quality criteria sensitivity, specificity, accuracy, and Dice coefficient are depicted (y-axis). Box plots with thick black lines show the median, boxes show the upper and lower quartiles, whiskers show the maximum and minimum values, and circles show the outliers. Figure 5 depicts the corresponding weights for the single scores that were needed to construct the first 6 components. The sign and magnitude of these loadings allowed us to identify the scores that were most relevant for each component. For example, mainly responsible for the construction of the first component were the scores for “air quality,” “ammonia noticeable,” “dust,” “Do the laying hens peck the investigator?” “Do the laying hens peck each other?” and “Do the laying hens chase each other?” Thus, the first component mostly summarized the topics “climate” and “pecking behavior.” However, we also derived smaller positive and negative loadings for the remaining scores so that this interpretation to neglect these can be seen only as a simplification. In addition, the interpretation that the sign of the single score shows which kind of relation, a positive or a negative, it has on the development of feather pecking and cannibalism is unconfirmed. It shows only the relation it has on the corresponding component. Figure 5. View largeDownload slide Loadings of each single score on the first calculated 6 components that are used for prediction. The loadings represent different corresponding weights of the scores for the development of feather pecking and cannibalism and consist of the magnitude (size of the circles) and the sign (color). Each component is based on the 35 single scores with differing estimated loadings. All components have an impact on the following one. Figure 5. View largeDownload slide Loadings of each single score on the first calculated 6 components that are used for prediction. The loadings represent different corresponding weights of the scores for the development of feather pecking and cannibalism and consist of the magnitude (size of the circles) and the sign (color). Each component is based on the 35 single scores with differing estimated loadings. All components have an impact on the following one. DISCUSSION General Results And Plumage And Skin Lesion Scores This study was designed to develop an easy-to-use tool for farmers or their consultants that gives them a prognosis for a possible development of feather pecking or cannibalism in their current laying hen flocks. As the conditions that may be related to these problematic behaviors are multifactorial (Savory and Mann, 1997; Rodenburg et al., 2013; Schwarzer et al., 2015), a choice of possible risk factors related to management types, barn settings, and laying hen behaviors were assessed. In addition, independent of the results that the farmer receives from the prognostic tool or the principal components, he is able to have a closer look at each rate he got in each question and to improve himself by following the description of the next higher score in this topic. By following these descriptions, he would reduce his prognostic score and, as you can see in the results section, a difference in 0.1 points in the prognostic score changes the chance of feather pecking and cannibalism development 2.5 times. Furthermore, the different farming systems, laying hen strains, and flock sizes of the presented study are representative of the conditions in currently practiced laying hen farming, suggesting that the designed tool could be applicable in various laying hen systems (only in loose housing systems). There is no need for the future use in practice of the assessment of the laying hens, since this procedure was necessary only for the establishment of this forecasting system. In addition, the very high inter-rater reliability results of the 3 investigators facilitate the good requirements for the used analysis. The 53 scores we used for benchmarking the different farms by far outnumbered the base data of 43 flocks. Thus, we refrained from investigating further factors such as food ingredients, behavioral studies, and early-life rearing conditions. Therefore, our results cannot reveal whether the propensity for severe feather damage and cannibalism developed during the rearing or the laying period, since it was shown that already the rearing period seems to have an influence on the occurrence of feather pecking in the laying period (Bestman et al., 2009; Schwarzer et al., 2015). We wanted to keep it simple, to allow the realization of the tool by the farmer on its own. Checking the rearing conditions, behavioral observations, and a food ingredients analysis would not provide easy, practical use. Nonetheless, we observed that both feather damage and wounds developed during the laying period. At the first visit, when the hens were approx. 20 wk old and at the beginning of the laying period, only one flock showed problems with cannibalism, and none showed severe feather pecking. During approx. 47 wk, the time between the first and the second visit, 14 flocks developed severe feather pecking, and 24 flocks showed ≥10% of the hens to have skin lesions of >0.5 cm and thus a problem with cannibalism. Prognostic tool: Prognostic score In a first step, we investigated the prognostic value of the newly designed average prognostic score and found a significant correlation between the prognostic score and both the plumage score and the skin lesion score: Assessed flocks with a better prognostic score showed less occurrence of feather pecking and thus a higher (better) plumage score. In addition, they showed less occurrence of cannibalism and thus a lower skin lesion score, which means fewer laying hens with wounds >0.5 cm. Furthermore, when the prognostic score exceeded a threshold value (1.57 for feather pecking; 1.45 for cannibalism), the probability of observing the 2 problems exceeded 0.5. This result could help laying hen farmers as a guide value that indicates a need for action. Spindler et al. (2013) developed a system for laying hen farmers to identify critical parameters they should monitor on their farms and a so-called emergency plan for critical situations, e.g., when the loss rate increases extensively and the plumage and skin condition of the laying hens worsen. In the studies by Holle et al. (2008) and Keppler et al. (2014), farmers could use benchmarks to compare their own management practices and the behavior of their laying hens with those of other farmers and make use of this comparison to improve conditions on their farms. Both studies compared flocks by using various scores and benchmarks with the aim to find possible risk factors of welfare parameters and to sensitize the farmers to their management practices. In the present study, we took a similar approach. First, we selected typical factors with possible risk for the development of feather pecking and cannibalism and scored them into 3 categories. With the help of these benchmarks, the assessed flocks were comparable. A possible inter-rater variability for the assessment of the laying hens got tested and achieved very high results and so can be neglected; in addition, the assessment of the laying hens will not be necessary in practicable use. Based on the comparisons between the assessed flocks, we could identify the factors (scores) that differed and used them as benchmarks to create a prognostic tool. The novel aspect of this study was the validation of the tool by comparing the results with the actual development of feather pecking and cannibalism problems in each flock. The comparisons showed a significant correlation between the estimated prognostic score and the developed problems. A drawback in the course of this study was an outbreak of the avian influenza in November 2016. As a result, the timing of the second visit varied widely (57th to 83rd wk of life). However, this variation can be neglected because the prognostic tool should be used to predict whether feather pecking and cannibalism are likely to occur but not when they may occur. Although the prognostic quality of the significant correlations between the prognostic score and the plumage and skin lesion scores were not satisfying, the identified correlations are an immense progress in the construction of an early alert system that is needed to solve the problems of feather pecking and cannibalism. As pointed out by Spindler et al. (2013), the existing guidelines for critical situations need to be amended with a system that can predict these problems so that farmers can implement preventive measures. Prognostic tool: Single scores As a second step, single scores were investigated in a principal component analysis to achieve a higher prognostic quality by reducing the number of factors (53 scores) relative to the number of assessed flocks (43). Whereas the prognostic score was constructed by using the same weights for all single scores, the principal components were obtained by allowing different weights (loadings). We could reduce the assessed scores of the prognostic tool to the smallest possible number with the highest possible weights (loadings) by means of the principal component analysis (yielding 6 components – Figure 5). The loadings consisted of a magnitude and a sign. These do not allow identifying a possible risk factor, but they have an influence on the components, for which none of the assessed scores, even the ones with low magnitude, can be neglected. This procedure led to a highly prognostic quality of the identified principal components. Based on these results, this novel tool is suitable for the forecast of the development of feather pecking or cannibalism in laying hens. As a next step, software for this tool, accessible for practical use, should be developed and validated. CONCLUSION The prognostic tool presented here is a highly valid and easy-to-use prognostic system, which shows only slight deficits in its prognostic quality. These deficits can be explained by the large number of scores (53) used to assess a small number of base data (43 flocks). A subsequent principal component analysis reduced the number of scores to 6 principal components that yielded a very high probability of a correct prognosis of the development of feather pecking and cannibalism. ACKNOWLEDGEMENTS This work was supported financially by the Bavarian State Ministry of the Environment and Consumer Protection via the Bavarian Health and Food Safety Authority (grant number: Az. StMUV: 47d-G7131-2011/28-30). The project was planned and realized by the Chair of Animal Welfare, Animal Behavior, Animal Hygiene and Animal Husbandry of the Ludwig-Maximilians-University in Munich, Germany, in cooperation with Bavarian poultry consultants from the Departments of Food, Agriculture and Forestry in Pfaffenhofen an der Ilm and Kitzingen, Germany. The funding source played no role in study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the paper for publication. Conflict of interest statement There is no conflict of interest regarding any financial, personal, or other relationships with other people or organizations that could inappropriately influence or be perceived to influence this study. Ethical statement The work described in this article with research on live animals met the guidelines approved by the institutional animal care and use committee (IACUC). Authorship statement The idea for the paper was conceived by Helen Louton and Angela Schwarzer. The experiments were designed by Helen Louton, Angela Schwarzer, and Michael Erhard. The experiments were performed by Miriam Zepp, Franziska Helmer, and Anne-Katrin Kaesberg under the supervision of Helen Louton and Angela Schwarzer. The data were analyzed by Paul Schmidt. The paper was written by Anne-Katrin Kaesberg. REFERENCES Aerni V., El-Lethey H., Wechsler B.. 2000. Effect of foraging material and food form on feather pecking in laying hens. Br. Poult. Sci.  41: 16– 21. Google Scholar CrossRef Search ADS PubMed  Allen J., Perry G.. 1975. Feather pecking and cannibalism in a caged layer flock. Br. Poult. Sci.  16: 441– 451. Google Scholar CrossRef Search ADS PubMed  Bestman M., Koene P., Wagenaar J.-P.. 2009. Influence of farm factors on the occurrence of feather pecking in organic reared hens and their predictability for feather pecking in the laying period. Appl. Anim. Behav. Sci.  121: 120– 125. Google Scholar CrossRef Search ADS   Bestman M., Wagenaar J.-P.. 2003. Farm level factors associated with feather pecking in organic laying hens. Livest. Prod. Sci.  80: 133– 140. Google Scholar CrossRef Search ADS   Bestman M., Wagenaar J.-P.. 2014. Health and welfare in Dutch organic laying hens. Animals . 4: 374– 390. Google Scholar CrossRef Search ADS PubMed  Blokhuis H. 1986. Feather-pecking in poultry: Its relation with ground-pecking. Appl. Anim. Behav. Sci.  16: 63– 67. Google Scholar CrossRef Search ADS   Blokhuis H., Arkes J.. 1984. Some observations on the development of feather-pecking in poultry. Appl. Anim. Behav. Sci.  12: 145– 157. Google Scholar CrossRef Search ADS   Blokhuis H. J., Fiks Van Niekerk T., Bessei W., Elson A., Guémené D., Kjaer J. B., Maria Levrino G. A., Nicol C. J., Tauson R., Weeks C. A., Van De Weerd H. A.. 2007. The LayWel project: Welfare implications of changes in production systems for laying hens. Worlds Poult. Sci. J.  63: 101– 114. doi 10.1017/S0043933907001328. Google Scholar CrossRef Search ADS   Brenninkmeyer C., Knierim U.. 2015. Förderung der Tiergesundheit und des Tierwohls ökologischer Legehennen in Europa – Promoting good health and welfare in European organic laying hens: Abschlussbericht. Accessed Jun. 2017. Available from: http://orgprints.org/29166/7/29166-11OE020-uni-kassel-knierim-2015-HealthyHens-merkblatt.pdf. Byrt T., Bishop J., Carlin J. B.. 1993. Bias, prevalence and kappa. J clin epidemiol . 46: 423– 429. Google Scholar CrossRef Search ADS PubMed  Chow A., Hogan J. A.. 2005. The development of feather pecking in Burmese red junglefowl: The influence of early experience with exploratory-rich environments. Appl. Anim. Behav. Sci.  93: 283– 294. Google Scholar CrossRef Search ADS   Dice L. R. 1945. Measures of the amount of ecologic association between species. Ecology . 26: 297– 302. Google Scholar CrossRef Search ADS   El-Lethey H., Aerni V., Jungi T., Wechsler B.. 2000. Stress and feather pecking in laying hens in relation to housing conditions. Br. Poult. Sci.  41: 22– 28. Google Scholar CrossRef Search ADS PubMed  European Commission. 2007. Council regulation (EC) No 834/2007 of 28 June 2007 on organic production and labelling of organic products and repealing regulation (EEC) No 2092/91. German Order on the Protection of Animals and the Keeping of Production Animals. 2006. Amended and promulgated on August 22nd, 2006, last changed on February 5th, 2014. Verordnung zum Schutz landwirtschaftlicher Nutztiere und anderer zur Erzeugung tierischer Produkte gehaltener Tiere bei ihrer Haltung; Tierschutz–Nutztierhaltungsverordnung (TierSchNutztV) in der Fassung der Bekanntmachung vom 22. August 2006 (BGBI. I S. 2043), die durch Artikel 1 der Verordnung vom 5. Februar 2014 (BGBI. I S. 94) geändert worden ist. Green L. E., Lewis K., Kimpton A., Nicol C. J.. 2000. Cross-sectional study of the prevalence of feather pecking in laying hens in alternative systems and its association with management and disease. Vet. Rec.  147: 233– 238. Google Scholar CrossRef Search ADS PubMed  Gunnarsson S. 2000a. Laying hens in loose housing systems: Clinical, ethological and epidemiological aspects . PhD thesis, Sveriges Lantbruksuniversitet ( Swedish University of Agricultural Sciences), Uppsala. Gunnarsson S., Algers B., Svedberg J.. 2000b. Description and evaluation of a scoring system of clinical health in laying hens. Gunnarsson, S.: Laying hens in loose housing systems . Doctoral thesis, Swedish University of Agricultural Sciences, Uppsala. Hartcher K. M., Tran M. K., Wilkinson S. J., Hemsworth P. H., Thomson P. C., Cronin G. M.. 2015. Plumage damage in free-range laying hens: Behavioural characteristics in the rearing period and the effects of environmental enrichment and beak-trimming. Appl. Anim. Behav. Sci.  164: 64– 72. Google Scholar CrossRef Search ADS   Heerkens J. L., Delezie E., Kempen I., Zoons J., Ampe B., Rodenburg T. B., Tuyttens F. A.. 2015. Specific characteristics of the aviary housing system affect plumage condition, mortality and production in laying hens. Poult. Sci.  94: 2008– 2017. Google Scholar CrossRef Search ADS PubMed  Holle R., Müller-Arnke I., Rahmann G.. 2008. Tiergesundheitspläne in der Ökologischen Legehennenhaltung. Praxis trifft Forschung Neues aus der Ökologischen Tierhaltung . 2008: 143. Huber-Eicher B., Wechsler B.. 1997. Feather pecking in domestic chicks: Its relation to dustbathing and foraging. Anim. Behav.  54: 757– 768. Google Scholar CrossRef Search ADS PubMed  Huber-Eicher B., Wechsler B.. 1998. The effect of quality and availability of foraging materials on feather pecking in laying hen chicks. Anim. Behav.  55: 861– 873. Google Scholar CrossRef Search ADS PubMed  Keppler C., Moeller D., Knierim U.. 2014. Economic analysis of a management tool for the improvement of laying hen health on organic farms. Proceedings of the 6th International Conference on the Assessment of Animal Welfare at Farm and Group Level, L. Mounier, I . Veissier, Clermont-Ferrand, Page 71. Kjaer J. B. 2009. Feather pecking in domestic fowl is genetically related to locomotor activity levels: Implications for a hyperactivity disorder model of feather pecking. Behav. Genet.  39: 564– 570. Google Scholar CrossRef Search ADS PubMed  Kjaer J. B., Sørensen P.. 2002. Feather pecking and cannibalism in free-range laying hens as affected by genotype, dietary level of methionine + cystine, light intensity during rearing and age at first access to the range area. Appl. Anim. Behav. Sci.  76: 21– 39. Google Scholar CrossRef Search ADS   Landis J. R., Koch G. G.. 1977. The measurement of observer agreement for categorical data. biometrics  33: 159– 174. Google Scholar CrossRef Search ADS PubMed  Leenstra F., Maurer V., Galea F., Bestman M., Amsler-Kepalaite Z., Visscher J., Vermeij I., van Krimpen M.. 2014. Laying hen performance in different production systems: Why do they differ and how to close the gap? Results of discussions with groups of farmers in The Netherlands, Switzerland and France, benchmarking and model calculations. Eur. Poult. Sci.  78: 1– 10. Lenz A. 2015. Federpicken und Kannibalismus bei nicht-schnabelgekürzten Legehennen in Praxisbetrieben, Feather pecking and cannibalism amongst non-beak-trimmed laying hens on laying farms: Einfluss von Management und Haltung, influence of management and husbandry . Diss. Dr. med. vet., Lud'wig-Maximilians-University Munich, Germany. Louton H., Bergmann S. M., Rauch E., Liebers C., Reese S., Erhard M. H., Hoeborn C., Schwarzer A.. 2017. Evaluation of welfare parameters in laying hens on the basis of a Bavarian survey. Poult. Sci.  96: 3199– 3213 Google Scholar CrossRef Search ADS PubMed  Lugmair A. 2009. Epidemiologische Untersuchungen zum Auftreten von Federpicken in alternativen Legehennenhaltungen Österreichs . PhD Diss. Univ. Vienna, Austria. LWK-Niedersachsen. 2016. Minimierung von Federpicken und Kannibalismus bei Legehennen mit intaktem Schnabel. Neue Wege für die Praxis: Managementleitfaden. New ways for practice: management guideline. Agricultural Chamber of Niedersachsen, Oldenburg, Germany. Accessed Jun. 2017. https://www.mud-tierschutz.de/fileadmin/SITE_MASTER/content/Dokumente/Downloads/Leitfaden_LH_Minimierung_Federpicken_Kannibalismus_2813MTD003-1.pdf. Niebuhr K., Zaludik K., Gruber B., Thenmaier I., Lugmair A., Baumung R., Troxler J.. 2006. Untersuchungen zum Auftreten von Kannibalismus und Federpicken in alternativen Legehennenhaltungen in Österreich – Empfehlungen für die Praxis. in Online-Fachzeitschrift des Bundesministeriums für Land- und Forstwirtschaft, Umwelt und Wasserwirtschaft, 2006 , Vienna, Austria. NMELV. 2017. Empfehlungen zur Verhinderung von Federpicken und Kannibalismus bei Jung- und Legehennen. Niedersächsisches Ministerium für Ernährung, Landwirtschaft und Verbraucherschutz . Ministry of Food, Agriculture and Consumer Protection of Niedersachsen, Hannover, Germany. Accessed Jun. 2017. https://www.google.de/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&ved=0ahUKEwjl1oKQyPTUAhWIbhQKHUVaAiIQFggmMAA&url=http%3A%2F%2Fwww.ml.niedersachsen.de%2Fdownload%2F118043%2FEmpfehlungen_zur_Vermeidung_von_Federpicken_und_Kannibalismus_bei_Jung-_und_Legehennen_neu_2017.pdf&usg=AFQjCNHG_cmZUW5TOXxxJCur8vjzqZQlww&cad=rja. Plattner C. 2015. Verhalten nicht-schnabelgekürzter Legehennen in Boden- und Freilandhaltung mit Fokus auf das Pickverhalten, Behaviour of non-beak-trimmed laying hens in alternative housing systems with a special focus on the pecking behaviour . Diss. Dr. med. vet., Ludwig-Maximilians-University Munich, Germany. R Core Team. 2015. R: A Language and Environment for Statistical Computing . R Foundation for Statistical Computing, Vienna, Austria. Ramadan S., Von Borell E.. 2008. Role of loose feathers on the development of feather pecking in laying hens. Br. Poult. Sci.  49: 250– 256. Google Scholar CrossRef Search ADS PubMed  Rodenburg T., Van Krimpen M., De Jong I., De Haas E., Kops M., Riedstra B., Nordquist R., Wagenaar J., Bestman M., Nicol C.. 2013. The prevention and control of feather pecking in laying hens: Identifying the underlying principles. Worlds Poult. Sci. J.  69: 361– 374. Google Scholar CrossRef Search ADS   Sanotra G., Vestergaard K., Agger J., Lawson L.. 1995. The relative preferences for feathers, straw, wood-shavings and sand for dustbathing, pecking and scratching in domestic chicks. Appl. Anim. Behav. Sci.  43: 263– 277. Google Scholar CrossRef Search ADS   Savory C. 1995. Feather pecking and cannibalism. Worlds Poult. Sci. J  51: 215– 219. Google Scholar CrossRef Search ADS   Savory C. J., Mann J. S.. 1997. Behavioural development in groups of pen‐housed pullets in relation to genetic strain, age and food form. Br. Poult. Sci.  38: 38– 47. doi 10.1080/00071669708417938. Google Scholar CrossRef Search ADS PubMed  Schwarzer A., Louton H., Bergmann S., Rauch E., Erhard M.. 2015. Endbericht: Maßnahmen zur Verbesserung des Tierschutzes bei Legehennen in Praxisbetrieben , Ludwig-Maximilans-University Munich, Germany. Spindler B., Schulze Hillert M., Hartung J.. 2013. Abschlussbericht Praxisbegleitende Untersuchungen zur Prüfung des Verzichts auf Schnabelkürzen bei Legehennen in Praxisbetrieben , TiHo, Hannover, Germany. Stadig L. M., Ampe B. A., Van Gansbeke S., Van den Bogaert T., D’Haenens E., Heerkens J. L., Tuyttens F. A.. 2015. Opinion of Belgian egg farmers on hen welfare and its relationship with housing type. Animals  6: 1. Google Scholar CrossRef Search ADS   Steenfeldt S., Kjær J. B., Engberg R. M.. 2007. Effect of feeding silages or carrots as supplements to laying hens on production performance, nutrient digestibility, gut structure, gut microflora and feather pecking behaviour. Br. Poult. Sci.  48: 454– 468. Google Scholar CrossRef Search ADS PubMed  Szczepanek A. 2016. Untersuchungen zu Risikofaktoren für das Auftreten von Federpicken und Kannibalismus bei nicht-schnabelgekürzten Legehennen in Praxisbetrieben . Diss. Dr. med. vet., LMU Munich, Germany. Tauson R., Svensson S.. 1980. Influence of plumage conditions on the hen's feed requirement. Swed. J. Agr. Res.  10: 35– 39. Vestergaard K. S., Skadhauge E., Lawson L.. 1997. The stress of not being able to perform dustbathing in laying hens. Physiol. Behav.  62: 413– 419. Google Scholar CrossRef Search ADS PubMed  Zentralverband der deutschen Geflügelwirtschaft, e. V. 2015. Pressemitteilung: Verzicht auf das Schnabelkürzen: Geflügelwirtschaft unterzeichnet freiwillige Vereinbarung mit Bundeslandwirtschaftsminister Schmidt. Accessed: Jun. 2017. http://www.zdg-online.de/uploads/tx_userzdgdocs/Gefluegelwirtschaft_unterzeichnet_Vereinbarung_zum_Verzicht_auf_das_Schnabelkuerzen.pdf. © 2017 Poultry Science Association Inc. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Poultry Science Oxford University Press

Development of a prognostic tool for the occurrence of feather pecking and cannibalism in laying hens

Loading next page...
 
/lp/ou_press/development-of-a-prognostic-tool-for-the-occurrence-of-feather-pecking-freTCbcNfg
Publisher
Oxford University Press
Copyright
© 2017 Poultry Science Association Inc.
ISSN
0032-5791
eISSN
1525-3171
D.O.I.
10.3382/ps/pex369
Publisher site
See Article on Publisher Site

Abstract

Abstract In July 2015, a German voluntary decree stipulated that the keeping of beak-trimmed laying hens after the 1st of January 2017 will no longer be permitted. Simultaneously, the present project was initiated to validate a newly developed prognostic tool for laying hen farmers to forecast, at the beginning of a laying period, the probability of future problems with feather pecking and cannibalism in their flock. For this purpose, we used a computer-based prognostic tool in form of a questionnaire that was easy and quick to complete and facilitated comparisons of different flocks. It contained various possible risk factors that were classified into 3 score categories (1 = “no need for action,” 2 = “intermediate need for action,” 3 = “instant need for action”). For the validation of this tool, 43 flocks of 41 farms were examined twice, at the beginning of the laying period (around the 20th wk of life) and around the 67th wk of life. At both visits, the designated investigators filled out the questionnaire and assessed the plumage condition and the skin lesions (as indicators of occurrence of feather pecking and cannibalism) of 50 laying hens of each flock. The average prognostic score of the first visit was compared with the existence of feather pecking and cannibalism in each flock at the end of the laying period. The results showed that the prognostic score was negatively correlated with the plumage score (r = −0.32; 95% confidence interval [CI]: [−0.56; −0.02]) and positively correlated with the skin lesion score (r = 0.38; 95% CI: [0.09; 0.61]). These relationships demonstrate that a better prognostic score was associated with a better plumage and skin lesion score. After performing a principal component analysis on the single scores, we found that only 6 components are sufficient to obtain highly sensitive and specific prognostic results. Thus, the data of this analysis should be used for creating applicable software for use on laying hen farms. INTRODUCTION The problems of feather pecking and cannibalism in industrially kept poultry, especially in laying hens, are omnipresent. Many factors seem to have an influence on the occurrence of feather pecking and cannibalism during the laying period of farmed laying hens: Some authors say that pecking on conspecifics is induced by stress factors (El-Lethey et al., 2000). Other studies found that the inability to satisfy natural behavior patterns, such as ground pecking or scratching caused by a lack of litter and manipulable material, causes the hens to redirect their needs from the unattractive surrounding towards the other hens (Blokhuis and Arkes, 1984; Blokhuis, 1986; Huber-Eicher and Wechsler, 1997, 1998; Aerni et al., 2000; Ramadan and Von Borell, 2008). This misdirected behavior is often referred to as “frustration behavior,” and Savory (1995) pointed out that frustrated behavior is abnormal behavior, but not of aggressive nature. Furthermore, 2 types of feather pecking can be distinguished (Savory, 1995). Gentle feather pecking seems to be normal behavior for exploration of the environment and can be observed especially in non-enriched surroundings; it is described as soft pecking without pulling and can develop into severe feather pecking (Savory, 1995). Severe feather pecking is defined as intense feather pecking and pulling out feathers, which can lead to bleeding and subsequently to a high chance that this behavior spreads in the flock (Hartcher et al., 2015). Kjaer (2009) replaced the concept of misdirected exploration with the description of a hyperactivity syndrome. He demonstrated that young hens with a high food search activity also show a high probability to develop feather pecking in their adult life. With regard to feather pecking being a result of high stress levels or frustration, several factors of possible influence have been reported: The access to litter material seems to be an important factor for the reduction of stress; in addition, it allows hens to perform natural behavior (Blokhuis and Arkes, 1984; Blokhuis, 1986; Huber-Eicher and Wechsler, 1997; Aerni et al., 2000; El-Lethey et al., 2000; Lugmair, 2009; Brenninkmeyer and Knierim, 2015; Louton et al., 2017). Another factor that could have an influence on pecking at flock mates might be the offering of pecking material made of manipulable substances (Huber-Eicher and Wechsler, 1998; Chow and Hogan, 2005; Steenfeldt et al., 2007). Conversely, Plattner (2015) found that the frequency of feather pecking at conspecifics was much higher during the presence than during the absence of a bale of straw. The enhanced pecking might be explained as a defense of resources. Many studies also found a connection between feather pecking and dust bathing. The possibility for laying hens to perform dust bathing behavior is correlated to the levels of corticosterone, i.e., an absence of a dust bath increases their stress level and is consequently correlated with an increase of stereotypic pecking (Vestergaard et al., 1997). Sanotra et al. (1995) found that the provision of sand as dust bathing substrate could distract the hens from feather pecking, whereas Huber-Eicher and Wechsler (1997) found that only the presence of manipulable pecking material could be correlated with the occurrence of feather pecking behavior. Furthermore, the possibility of using a winter garden and an outside area is negatively correlated to the occurrence of feather pecking (Green et al., 2000; Bestman and Wagenaar, 2003; Bestman and Wagenaar, 2014; Brenninkmeyer and Knierim, 2015; Lenz, 2015). Keeping 2 laying hen strains in one flock was related to an increase of the occurrence of feather pecking (Lenz, 2015). According to Blokhuis and Arkes (1984), cannibalism is simply a redirected behavior of higher intensity than feather pecking. Furthermore, a connection between poor feather condition and the occurrence of cannibalism has been shown (Kjaer and Sørensen, 2002). The period of molting may provoke a transition from feather pecking to cannibalism due to the high chance of bleeding, because the new feathers are very soft and their blood vessels rupture easily upon pecking (Huber-Eicher and Wechsler, 1997). The multiple possible risk factors mentioned above suggest that feather pecking and cannibalism have multifactorial origins (Savory and Mann, 1997; Rodenburg et al., 2013; Schwarzer et al., 2015). These behaviors cause not only pain and suffering for the animals (Allen and Perry, 1975; Spindler et al., 2013) but also economic losses for the farmer. Hen mortality increases (Niebuhr et al., 2006; Stadig et al., 2015), egg production decreases (El-Lethey et al., 2000; Niebuhr et al., 2006), and food intake rises (Tauson and Svensson, 1980). The severity of problems for animals and farmers, especially since the voluntary ban of beak trimming in July 2015 (Zentralverband der deutschen Geflügelwirtschaft, 2015), highlights the need for improved management and early detection of incipient problems. The best way to do so is bringing different flocks and managements onto a comparable level and to evaluate the differences. This is possible by the help of benchmarks, which were already used (Leenstra et al., 2014). With the aid of benchmarks, Holle et al. (2008) investigated risk factors that may cause problems in laying hen farming. Using a benchmarking system, they compared the health status of 20 organic kept laying hen flocks (European Commission, 2007) by assessing various health parameters, namely, feather condition, skin trauma, footpad abscesses, internal parasites, and the existence of Dermanyssus gallinae. Their goal was to establish so called “animal health schemes” that were based on a questionnaire, which was filled out by the investigators with help from the farmers. In order to foster the health of the animals, they generated a plan based on the investigators’ recommendations. This system also facilitated the comparison between different laying hen flocks, highlighting specific skills and drawbacks. They found out that “animal health schemes” work correctly only if the farmer is in a permanent communication with the consultant, and the animal health schemes undergo stable development. Keppler et al. (2014) developed the “MTool” to evaluate the health of non-beak-trimmed laying hens based on selected factors on organic and conventional farms. This tool is as well based on potential risk factors, which got evaluated by comparing the actual hen condition with the condition of what animal health should look like. They scored each factor with 3 different categories: “no need for action”; “medium-term need for action”; and “instant need for action.” If the hen condition got scored with medium-term need for action or worse, the management and husbandry condition needed to be checked and got scored the same way. Both tools were developed to sensitize farmers to closely watch their laying hens and improve the topics for which they did not get good rates. In line with this goal, the present study aimed to establish an easily realizable evaluation system with general management questions and an objective rating of the flocks and barns. It also contains recommendations for farmers to improve their management, especially in topics assessed not good. In the presented study, this was visualized in a traffic light system, similar to the 3 scores from Keppler et al. (2014): Each topic got scored into “1 = no need for action (green),” “2 = intermediate need for action (yellow),” and “3 = instant need for action (red).” All of them would be presented to the farmer, with the respective descriptions. These descriptions can also be used as recommendations for actions to be taken for improvement in a given topic. For example a farmer, who scored 3 in a given topic, can use the description of the better score 2, or score 1, as a guideline for improvement. For this purpose, we developed a prognostic tool based on a questionnaire with 53 benchmarks and simple assessments that allow a comparison between flocks based on combined selected factors. The novel aspect of this study is that we validated the comparison and the correlation with the relative occurrence of feather pecking or cannibalism by re-assessing the occurrence of these problematic behaviors at the end of the laying period. This kind of forecasting system should be a useful tool for laying hen farmers to identify farming conditions that could be improved and to sensitize them to the problems their animals may have. As already mentioned, feather pecking and cannibalism seem to be caused by multifactorial conditions and thus are difficult to eliminate once they occur. Hence, it is sometimes not enough to give farmers general advice or directions on possible actions to mitigate manifested problems to get rid of feather pecking and cannibalism behavior. It would be easier to prevent them before they manifest themselves, if it is possible to forecast their manifestation. It is necessary to analyze and identify the critical aspects of each farm individually to limit and maybe even solve the problems. A prognostic tool that can identify factors before the problems occur is the most preferable course of action for both the laying hens and the farmers. And this procedure was tried to establish in the presented study. MATERIALS AND METHODS Course of the project The data collection ran over 2 yr, from July 2015 until May 2017, and included 43 flocks of 41 commercially run conventional farms in Bavaria, Germany. The project was planned and realized by the Chair of Animal Welfare, Animal Behavior, Animal Hygiene and Animal Husbandry of the Ludwig-Maximilians-University in Munich, Germany. All farmers voluntarily participated in this project. Flocks were assessed twice during the laying period (all visits in each farm were performed by the same 3 investigators), the first time approx. 7 to 10 d after the hens had been transferred to the laying barn facility, i.e., in their 20th wk of life, and the second time on average in the 67th wk of life, before slaughter or molting. Due to an outbreak of avian influenza in November 2016 and the subsequent obligation to keep the animals inside their barns, the time for the second inspection varied (between the 57th and 83rd wk of life). Methods During both the first and the second visits, the same dataset was collected. The investigators filled out a detailed questionnaire, by asking the farmers several questions, including the topics stocking rate, management, rehousing, feeding, pecking material, litter, barn climate, lighting, and animal behavior (for detailed questions and the associated categories, see Table 1). Afterwards, the investigators entered the barn and evaluated all aspects of the flock's behavior, such as nervousness or mutual pecking, the setting of the barn, such as litter quality or the existence of pecking material, and the barn's climate by mutual agreement. The responses from the questionnaire and the results of the flock and farm evaluation were directly recorded in the computer-based prognostic tool. Table 1. The questions for the prognostic tool (53 questions) and the related scores with classifications. The number of flocks, their percentage per value and the total number of flocks per question (n) are given. The 35 scores, which were used for the principal component analysis, are underlined. Question  Score  Classification  Number of flocks  Percentage share  n  Stocking rate  Hens per usable square meter  1  <9 hens per usable square meter  38  88.37  43  2  9 hens per usable square meter  5  11.63    3  >9 hens per usable square meter  0  0    Hens per usable base square meter  1  <18 hens per usable base square meter  28  93.34  30  2  18 hens per usable base square meter  1  3.33    3  >18 hens per usable base square meter  1  3.33    Time with temporarily higher number of hens per usable square meter  1  total area provided by day 1  19  44.19  43  2  total area provided after ≤3 weeks  21  48.83    3  total area provided after >3 weeks  3  6.98    Temporarily higher number of hens per usable square meter  1  total area temporarily not provided with <9 hens per usable square meter  1  4.17  24  2  total area temporarily not provided with 9 hens per usable quare meter  4  16.67    3  total area temporarily not provided with >9 hens per usable square meter  19  79.16    Management  Do you keep laying hen strains mixed?  1  no  32  74.42  43  2  yes  11  25.58    Accessibility of the enclosure  1  good  22  55  40  2  medium  12  30    3  bad  6  15    How long is the cleaning period between 2 laying periods?  1  >2 weeks  28  65.12  43  2  1–2 weeks  10  23.25    3  <1 week  5  11.63    Do you have a cadaver bin?  1  yes  37  86.05  43  2  no  6  13.95    At what time do you open the nests?  1  nest opening ≥1 h before turning on the lights  35  81.39  43  2  nest opening <60 min before turning on the lights  7  16.28    3  nest opening at the same time or after turning on the lights  1  2.33    Do you document the death of animals?  1  yes, always  38  88.37  43  2  irregular  4  9.3    3  never  1  2.33    Did you have problems with feather pecking in previous period?  1  no  13  30.23  43  3  yes  30  69.77    Do you have problems with Dermanyssus gallinae  1  no  9  20.93  43  2  yes  34  79.07    Do you offer permanent access to a winter garden?  1  yes  21  48.84  43  2  no  7  16.28    3  no existing winter garden  15  34.88    Do you offer permanent access to an outside area?  1  yes  9  20.93  43  2  no  15  34.88    3  no existing outside area  19  44.19    Were sick animals found after the morning tour of the person in charge?  1  no  42  97.67  43  3  yes  1  2.33    Were dead animals found after the morning tour of the person in charge?  1  no  42  97.67  43  3  yes  1  2.33    How many times per day does the person in charge investigate the barn?  1  >1x per day  39  90.7  43  2  1x per day  4  9.3    3  <1x per day  0  0    How many times per day does the person in charge investigate the litter area?  1  >1x per day  39  90.7  43  2  1x per day  4  9.3    3  <1x per day  0  0    How many times per day does the person in charge investigate the aviary?  1  >1x per day  31  86.11  36  2  1x per day  4  11.11    3  <1x per day  1  2.78    How many times per day does the person in charge investigate the winter garden?  1  >1x per day  18  69.23  26  2  1x per day  8  30.77    3  <1x per day  0  0    How many minutes per day does the person in charge spend with the laying hens?  1  >3.9 min per 1,000 hens  42  100  42  2  3.9 min per 1,000 hens  0  0    3  <3.9 min per 1,000 hens  0  0    Do you have separate clothing for each livestock?  1  separate clothing for each livestock  16  38.1  42  2  separate clothing but is worn outside of the livestock  23  54.76    3  no separate clothing  3  7.14    Laying hen rehousing  How old were the laying hens at delivery?  1  17th to 18th week of life  34  79.07  43  2  ≤± 3 weeks  9  20.93    3  >± 3 weeks  0  0    Do you weigh the laying hens at delivery?  1  yes, weighing of single hens  10  23.26  43  2  yes, weighing of group of hens  7  16.28    3  no  26  60.46    Average weight of laying hens at delivery  1  ≥normal weight  3  25  12  2  ≤10% under the normal weight  7  58.33    3  >10% under the normal weight  2  16.67    Feeding  How many times does the feeding chain move per day?  1  >6x per day  14  36.84  38  2  4–6 x per day  24  63.16    3  <4x per day  0  0    Do you offer a feeding intermission?  1  always  29  67.44  43  2  sometimes  2  4.65    3  never  12  27.91    How many laying hens come to the moving feeding chain?  1  all of them  2  14.29  14  2  most of them  11  78.57    3  only a few  1  7.14    Do you document the food consumption?  1  yes, with a food or silo scale  18  41.86  43  2  yes, but only an estimation  11  25.58    3  no documentation  14  32.56    Pecking material  Do you spread grain?  1  regularly  17  40.48  42  2  irregularly  8  19.04    3  never  17  40.48    Do you offer pecking material?  1  regularly  32  74.42  43  2  irregularly  1  2.32    3  never  10  23.26    How much pecking material do you offer?  1  >1 element per 1,000 hens  21  48.84  43  2  ≤1 element per 1,000 hens  9  20.93    3  no pecking material offered  13  30.23    Do you offer a dust bath?  1  yes  20  46.51  43  3  no  23  53.49    Litter  Do you change litter regularly?  1  yes, completly  25  59.52  42  2  yes, partially  16  38.1    3  no change  1  2.38    Do you have problems with litter plaques?  1  no litter plaques  30  75  40  2  partial litter plaques  8  20    3  complete litter plaques  2  5    How deep is the litter at the beginning of the laying period?  1  >2 cm  10  24.39  41  2  ≤2 cm  28  68.29    3  no litter offered  3  7.32    Climate  air quality  1  good  36  83.72  43  2  medium  5  11.63    3  bad  2  4.65    ammonia noticable  1  slightly  38  88.37  43  2  moderately  3  6.98    3  intensely  2  4.65    dust  1  slight  36  83.72  43  2  moderate  7  16.28    3  intense  0  0    noise  1  quiet  35  81.39  43  2  moderate  7  16.28    3  noisy  1  2.33    infiltration  1  slight  30  69.77  43  2  moderate  11  25.58    3  intense  2  4.65    Lighting  Do you have dimmable lighting?  1  yes  31  72.09  43  2  partial  7  16.28    3  no  5  11.63    Do you have sunspots?  1  no  33  76.74  43  2  yes  10  23.26    Are the nests dark?  1  yes  23  57.5  40  2  partially  16  40    3  no  1  2.5    Is the litter area lighted evenly?  1  yes  30  69.77  43  2  no  13  30.23    Behavior of laying hens  How is the visitation of the litter area?  1  intensive  26  61.9  42  2  moderate  11  26.2    3  slight  5  11.9    Is the flock more nervous than usual?  1  no  26  60.47  43  3  yes  17  39.53    Do the laying hens peck on the investigator?  1  no pecking  28  65.12  43  2  slight pecking  11  25.58    3  intense pecking  4  9.3    Do the laying hens peck each other?  1  no pecking  27  62.79  43  2  slight pecking  16  37.21    3  intense pecking  0  0    Do the laying hens chaise each other?  1  no chasing  39  90.7  43  2  some chasing  4  9.3    3  intense chasing  0  0    Ground feathers: Test 3 different areas in the barn and count on each measuring point the ground feathers per 1 square meter  1  ≥10 ground feathers at all 3 measuring points found  17  42.5  40  2  <10 ground feathers found at minimum 1 measuring point  13  32.5    3  minimum 1 measuring point with no ground feathers  10  25    Are the laying hens easy to catch?  1  easy  38  88.37  43  2  medium  5  11.63    3  not easy  0  0    Touch-test: Squat and move slowly; count how many laying hens you are able to touch  1  ≥1 laying hen could be touched on average  29  70.73  41  2  <1 laying hen could be touched on average  12  29.27    Question  Score  Classification  Number of flocks  Percentage share  n  Stocking rate  Hens per usable square meter  1  <9 hens per usable square meter  38  88.37  43  2  9 hens per usable square meter  5  11.63    3  >9 hens per usable square meter  0  0    Hens per usable base square meter  1  <18 hens per usable base square meter  28  93.34  30  2  18 hens per usable base square meter  1  3.33    3  >18 hens per usable base square meter  1  3.33    Time with temporarily higher number of hens per usable square meter  1  total area provided by day 1  19  44.19  43  2  total area provided after ≤3 weeks  21  48.83    3  total area provided after >3 weeks  3  6.98    Temporarily higher number of hens per usable square meter  1  total area temporarily not provided with <9 hens per usable square meter  1  4.17  24  2  total area temporarily not provided with 9 hens per usable quare meter  4  16.67    3  total area temporarily not provided with >9 hens per usable square meter  19  79.16    Management  Do you keep laying hen strains mixed?  1  no  32  74.42  43  2  yes  11  25.58    Accessibility of the enclosure  1  good  22  55  40  2  medium  12  30    3  bad  6  15    How long is the cleaning period between 2 laying periods?  1  >2 weeks  28  65.12  43  2  1–2 weeks  10  23.25    3  <1 week  5  11.63    Do you have a cadaver bin?  1  yes  37  86.05  43  2  no  6  13.95    At what time do you open the nests?  1  nest opening ≥1 h before turning on the lights  35  81.39  43  2  nest opening <60 min before turning on the lights  7  16.28    3  nest opening at the same time or after turning on the lights  1  2.33    Do you document the death of animals?  1  yes, always  38  88.37  43  2  irregular  4  9.3    3  never  1  2.33    Did you have problems with feather pecking in previous period?  1  no  13  30.23  43  3  yes  30  69.77    Do you have problems with Dermanyssus gallinae  1  no  9  20.93  43  2  yes  34  79.07    Do you offer permanent access to a winter garden?  1  yes  21  48.84  43  2  no  7  16.28    3  no existing winter garden  15  34.88    Do you offer permanent access to an outside area?  1  yes  9  20.93  43  2  no  15  34.88    3  no existing outside area  19  44.19    Were sick animals found after the morning tour of the person in charge?  1  no  42  97.67  43  3  yes  1  2.33    Were dead animals found after the morning tour of the person in charge?  1  no  42  97.67  43  3  yes  1  2.33    How many times per day does the person in charge investigate the barn?  1  >1x per day  39  90.7  43  2  1x per day  4  9.3    3  <1x per day  0  0    How many times per day does the person in charge investigate the litter area?  1  >1x per day  39  90.7  43  2  1x per day  4  9.3    3  <1x per day  0  0    How many times per day does the person in charge investigate the aviary?  1  >1x per day  31  86.11  36  2  1x per day  4  11.11    3  <1x per day  1  2.78    How many times per day does the person in charge investigate the winter garden?  1  >1x per day  18  69.23  26  2  1x per day  8  30.77    3  <1x per day  0  0    How many minutes per day does the person in charge spend with the laying hens?  1  >3.9 min per 1,000 hens  42  100  42  2  3.9 min per 1,000 hens  0  0    3  <3.9 min per 1,000 hens  0  0    Do you have separate clothing for each livestock?  1  separate clothing for each livestock  16  38.1  42  2  separate clothing but is worn outside of the livestock  23  54.76    3  no separate clothing  3  7.14    Laying hen rehousing  How old were the laying hens at delivery?  1  17th to 18th week of life  34  79.07  43  2  ≤± 3 weeks  9  20.93    3  >± 3 weeks  0  0    Do you weigh the laying hens at delivery?  1  yes, weighing of single hens  10  23.26  43  2  yes, weighing of group of hens  7  16.28    3  no  26  60.46    Average weight of laying hens at delivery  1  ≥normal weight  3  25  12  2  ≤10% under the normal weight  7  58.33    3  >10% under the normal weight  2  16.67    Feeding  How many times does the feeding chain move per day?  1  >6x per day  14  36.84  38  2  4–6 x per day  24  63.16    3  <4x per day  0  0    Do you offer a feeding intermission?  1  always  29  67.44  43  2  sometimes  2  4.65    3  never  12  27.91    How many laying hens come to the moving feeding chain?  1  all of them  2  14.29  14  2  most of them  11  78.57    3  only a few  1  7.14    Do you document the food consumption?  1  yes, with a food or silo scale  18  41.86  43  2  yes, but only an estimation  11  25.58    3  no documentation  14  32.56    Pecking material  Do you spread grain?  1  regularly  17  40.48  42  2  irregularly  8  19.04    3  never  17  40.48    Do you offer pecking material?  1  regularly  32  74.42  43  2  irregularly  1  2.32    3  never  10  23.26    How much pecking material do you offer?  1  >1 element per 1,000 hens  21  48.84  43  2  ≤1 element per 1,000 hens  9  20.93    3  no pecking material offered  13  30.23    Do you offer a dust bath?  1  yes  20  46.51  43  3  no  23  53.49    Litter  Do you change litter regularly?  1  yes, completly  25  59.52  42  2  yes, partially  16  38.1    3  no change  1  2.38    Do you have problems with litter plaques?  1  no litter plaques  30  75  40  2  partial litter plaques  8  20    3  complete litter plaques  2  5    How deep is the litter at the beginning of the laying period?  1  >2 cm  10  24.39  41  2  ≤2 cm  28  68.29    3  no litter offered  3  7.32    Climate  air quality  1  good  36  83.72  43  2  medium  5  11.63    3  bad  2  4.65    ammonia noticable  1  slightly  38  88.37  43  2  moderately  3  6.98    3  intensely  2  4.65    dust  1  slight  36  83.72  43  2  moderate  7  16.28    3  intense  0  0    noise  1  quiet  35  81.39  43  2  moderate  7  16.28    3  noisy  1  2.33    infiltration  1  slight  30  69.77  43  2  moderate  11  25.58    3  intense  2  4.65    Lighting  Do you have dimmable lighting?  1  yes  31  72.09  43  2  partial  7  16.28    3  no  5  11.63    Do you have sunspots?  1  no  33  76.74  43  2  yes  10  23.26    Are the nests dark?  1  yes  23  57.5  40  2  partially  16  40    3  no  1  2.5    Is the litter area lighted evenly?  1  yes  30  69.77  43  2  no  13  30.23    Behavior of laying hens  How is the visitation of the litter area?  1  intensive  26  61.9  42  2  moderate  11  26.2    3  slight  5  11.9    Is the flock more nervous than usual?  1  no  26  60.47  43  3  yes  17  39.53    Do the laying hens peck on the investigator?  1  no pecking  28  65.12  43  2  slight pecking  11  25.58    3  intense pecking  4  9.3    Do the laying hens peck each other?  1  no pecking  27  62.79  43  2  slight pecking  16  37.21    3  intense pecking  0  0    Do the laying hens chaise each other?  1  no chasing  39  90.7  43  2  some chasing  4  9.3    3  intense chasing  0  0    Ground feathers: Test 3 different areas in the barn and count on each measuring point the ground feathers per 1 square meter  1  ≥10 ground feathers at all 3 measuring points found  17  42.5  40  2  <10 ground feathers found at minimum 1 measuring point  13  32.5    3  minimum 1 measuring point with no ground feathers  10  25    Are the laying hens easy to catch?  1  easy  38  88.37  43  2  medium  5  11.63    3  not easy  0  0    Touch-test: Squat and move slowly; count how many laying hens you are able to touch  1  ≥1 laying hen could be touched on average  29  70.73  41  2  <1 laying hen could be touched on average  12  29.27    View Large Factors that could be a potential risk for feather pecking and cannibalism had been selected before the project started and classified into 3 score categories: “no need for action” (Score 1), “intermediate need for action” (Score 2), and “instant need for action” (Score 3) (Table 1). Hereby, “action” means the recommendation for the farmer to improve his management in this factor's topic. The way to improve in this topic is directly presented in the description of the 3 categories, i.e., he needs to follow the description of the next higher score. For example, if the farmer does not weigh the laying hens at delivery, he receives the score “3 = instant need for action” in this topic, and this question will be presented in the color red. If the farmer wants to improve, he is able to see the description of number “2 = weighing of group of hens” (yellow) and of number “1 = weighing of single hens” (green) and knows what to do. Hereby, the farmer can see his “red topics,” even if the prognostic score is presentable and could show action of improvement also on his own initiative. These score categories were based on law guidelines (German Order on the Protection of Animals and the Keeping of Production Animals, 2006), professional recommendations (LWK-Niedersachsen, 2016; NMELV, 2017), previous studies (Lenz, 2015; Szczepanek, 2016), and scientific literature (Heerkens et al., 2015). For visualization, the score for each factor was marked in green (1), yellow (2), and red (3), and the spreadsheet was presented to the participating farmers on the spot to show them how they were performing in direct but anonymous comparison with other farmers (benchmarking). In this study, the spreadsheet was presented only on the second visit, to avoid that the results and the comparison to the other farmers could falsify the action of the farmers’ management throughout the laying period. An average prognostic score for each flock (best 1.0, worst 3.0) was calculated. It is based on the evaluated average of the given scores (1, 2, or 3) of 53 questions (Table 1). Subsequently, 50 (flock size > 500 hens) or 20 laying hens (flock size ≤ 500 hens) were selected randomly, and their plumage and skin conditions were determined according to a modified “hen score” from Gunnarsson (2000a) and Blokhuis et al. (2007). Hens of mixed laying hen strains were selected according to their proportion in the group. The hens were assessed by all 3 investigators, whose inter-rater reliability was tested in 100 brown egg and 100 white egg laying hen strains. The plumage condition was classified into 5 score categories: 5 = no special abnormality; 4 = >5 damaged feathers or featherless area Ø ≤ 1 cm; 3 = featherless area Ø > 1 − ≤ 5 cm; 2 = featherless area Ø > 5 cm and ≤75% featherless; 1 = >75% featherless. The classification was assessed for 3 body regions (the neck, the back, and the wing coverts) because feather damages in these body regions are likely to be caused by feather pecking (Ramadan and Von Borell, 2008). The sum of the scores for all 3 body regions formed the “plumage score,” i.e., the plumage score 3 equaled the largest plumage damage and therefore the most severe feather pecking problems. The best score 15 equaled an intact plumage and no or only minor feather pecking in the flock. The plumage score was averaged per flock. An average plumage score ≤10 was defined as a flock with severe feather pecking problems. The occurrence of wounds and their size were scaled using a separate score, the “skin lesion score.” In this case, the entire body (except the head and legs) was scanned for wounds, and only skin rupture,, but not discoloration, was considered. A flock that had ≥10% hens with one or more wounds of >0.5 cm was defined as having problems with cannibalism. Statistical Analysis The prognostic tool was designed with Microsoft Excel 2010 (Microsoft Corporation, Redmond, WA). For all analyses, the R language for statistical computing (R Core Team, 2015) was used. The relation between the prognostic score and the plumage or skin lesion score was first analyzed in an exploratory manner using the Pearson correlation coefficient. Prognostic quality of this relation was subsequently investigated in more detail by using linear regression. To this end, the data from 43 flocks were randomly separated into training (70%) and test (30%) datasets. The training dataset was used to estimate the linear model that then was used to predict the plumage and skin lesion scores for the test dataset. Prediction error was measured by the mean squared error and the mean percent error. This procedure was repeated 5,000 times with different training and test datasets to obtain an approximation of the distribution of these errors. In addition, logistic regression models were used to estimate prediction models for feather pecking and cannibalism with the prognostic score as the only predictor and were presented with odds ratio (OR) and confidence interval (CI). Here, prediction quality was measured by 4 criteria: sensitivity, calculated as true positive (TP)/(TP + false negative [FN]); specificity, calculated as true negative (TN)/(TN + false positive [FP]); accuracy, calculated as (TN + TP)/(TN + TP + FN + FP); and the dice coefficient (Dice, 1945), calculated as 2 * TP/(2 * TP + FN + FP). Randomized partition of the complete data into training and test data, as well as repetition, was performed as described above. For analyses of single scores, we first performed a dimensional reduction by means of a principal component analysis. By using this method, the 35 tested scores were transformed into a few components, the principal components. Usually, a small number of components is sufficient to describe the information of the complete data. Up to the first 10 of these components were then used as predictors in logistic regression models for the prediction of feather pecking and cannibalism as described for the prognostic score. Inter-rater reliability was analyzed by means of the PABAK (prevalence-adjusted and bias-adjusted kappa), which is also called kappanor (Byrt et al., 1993). This statistical tool was employed, because as the number of categories used is unpredictable in research with live animals and in this case is rather high; the kappa score should be corrected for prevalence and bias. The equation for more than 2 categories (Gunnarsson et al., 2000b) is   \begin{equation*} {\rm{PABAK}} = \left( {{{\left( {{\rm{k}}{{\rm{p}}_0}} \right) - 1}\, / {\left( {{\rm{k}} - 1} \right)}}}\right. . \end{equation*} Where: k represents the number of assessed categories, and p0 indicates the relation between the observed agreements. For the assessment of skin lesion only, 2 categories were used, in which case the equation shortens to   \begin{equation*} {\rm{PABAK}} = 2{{\rm{p}}_0} - 1. \end{equation*} RESULTS General results The assessed 43 flocks of 41 farms included conventional (79.1%) or organic (20.9%) farms, which could be either floor (32.6%) or aviary (67.4%) barns of varying size (from 200 up to 6,000 non-beak-trimmed animals per barn or barn section); 79.1% of the flocks hosted over 500 laying hens, and 20.9% hosted ≤500 laying hens (here, both flock sizes were hosted by both farming systems). This project included laying hens of the strains Lohmann Brown Classic, Lohmann Selected Leghorn, Bovans Brown, Dekalb White, Lohmann Brown Plus, Lohmann Brown Extra, Lohmann Dual, and Lohmann Sandy; 25.6% of the flocks were flocks with mixed laying hen strains (Lohmann Brown Classic + Lohmann Selected Leghorn, Lohmann Brown Classic + Lohmann Dual, or Bovans Brown + Dekalb White), 67.4% were brown egg poultry flocks, and 9.3% were white egg poultry flocks. For descriptive results for each score (53) of the prognostic tool, see Table 1 in the Appendix. As depicted in Table 2, the results of the PABAK inter-rater reliability test all vary between 0.88 and 0.99, which is within the range suggested by Landis and Koch (1977) for almost perfect reliability scores [0.81-1]. Table 2. Inter-rater reliability scores based on PABAK (prevalence-adjusted and bias-adjusted kappa). The three investigators (A, B, and C) assessed 100 brown and 100 white egg laying hen strains on their plumage condition in 3 body regions and on the existence and size of possible skin lesions in 9 body regions. Agreements and disagreements with the corresponding percentage and results of PABAK are presented for each investigator combination.   Plumage (n = 300)  Skin lesion (n = 900)    Number of agreements (%)  Number of disagreements (%)  PABAK  Number of agreements (%)  Number of disagreements (%)  PABAK    Brown egg laying hen strain  A/B  280 (93.33%)  20 (6.67%)  0.92  891 (99%)  9 (1%)  0.98  A/C  276 (92%)  24 (8%)  0.9  888 (98.67%)  12 (1.33%)  0.97  B/C  281 (93.67%)  9 (6.33%)  0.92  891 (99%)  9 (1%)  0.98    White egg layng hen strain  A/B  276 (92%)  24 (8%)  0.9  895 (99.44%)  5 (0.56%)  0.99  A/C  270 (90%)  30 (10%)  0.88  892 (99.11%)  8 (0.89%)  0.98  B/C  277 (92.33%)  23 (7.67%)  0.9  893 (99.22%)  7 (0.78%)  0.98    Plumage (n = 300)  Skin lesion (n = 900)    Number of agreements (%)  Number of disagreements (%)  PABAK  Number of agreements (%)  Number of disagreements (%)  PABAK    Brown egg laying hen strain  A/B  280 (93.33%)  20 (6.67%)  0.92  891 (99%)  9 (1%)  0.98  A/C  276 (92%)  24 (8%)  0.9  888 (98.67%)  12 (1.33%)  0.97  B/C  281 (93.67%)  9 (6.33%)  0.92  891 (99%)  9 (1%)  0.98    White egg layng hen strain  A/B  276 (92%)  24 (8%)  0.9  895 (99.44%)  5 (0.56%)  0.99  A/C  270 (90%)  30 (10%)  0.88  892 (99.11%)  8 (0.89%)  0.98  B/C  277 (92.33%)  23 (7.67%)  0.9  893 (99.22%)  7 (0.78%)  0.98  View Large Plumage and skin lesion scores Based on the 2 parameters plumage score and skin lesion score, the assessed flocks were divided into groups with or without problems. Fourteen of the 43 flocks (32.6%) had a plumage score ≤10 at the second visit, indicating a problem with severe feather pecking at the end of the laying period. In comparison, none of the assessed flocks had a plumage score ≤10 at the first visit, i.e., no flock showed problems with severe feather pecking at the beginning of the laying period (approx. 20th wk of life). The average plumage score was 10.65 points with a range from 7.08 up to 13.48 points at the second visit (Figure 1A). Figure 1. View largeDownload slide A (above) + B (below): Average plumage score (1A) and average skin lesion score (1B) (y-axis) of every farm (x-axis) with number of n showing the number of assessed laying hens per flock at the second visit. Each farm has its own number; farms with 2 assessed flocks are identified by different decimal places. The plumage score indicates the average feather condition per assessed flock for the body regions neck, back, and wing coverts. A plumage score ≤10 presents a problem with severe feather pecking in this flock. The skin lesion score indicates the percentage share of animals in the flock with skin lesions >0.5 cm. A flock with ≥10% of animals with big skin lesions shows problems with cannibalism. Figure 1. View largeDownload slide A (above) + B (below): Average plumage score (1A) and average skin lesion score (1B) (y-axis) of every farm (x-axis) with number of n showing the number of assessed laying hens per flock at the second visit. Each farm has its own number; farms with 2 assessed flocks are identified by different decimal places. The plumage score indicates the average feather condition per assessed flock for the body regions neck, back, and wing coverts. A plumage score ≤10 presents a problem with severe feather pecking in this flock. The skin lesion score indicates the percentage share of animals in the flock with skin lesions >0.5 cm. A flock with ≥10% of animals with big skin lesions shows problems with cannibalism. The skin lesion score (as a scale for the percentage of hens in a flock with wounds >0.5 cm) showed that at the second visit, 24 flocks (55.8%) had a minimum of 10% with one or more wounds >0.5 cm, and hence a problem with cannibalism; at the first visit, it was one flock (2.3%). On average, the assessed flocks at the second visit had a skin lesion score of 14%. The highest percentage of hens with big wounds >0.5 cm was 53.3%. In 7 flocks, no laying hen with wounds >0.5 cm was found (Figure 1B). Prognostic tool: Prognostic score The flock with the best (i.e., lowest) prognostic score was scored with 1.29 and the worst with 1.69 points. A correlation analysis showed that the prognostic score was negatively correlated with the plumage score (r = −0.32; 95% CI: [−0.56; −0.02]), meaning that higher (worse) prognostic scores were correlated with lower (worse) plumage scores. Between the prognostic score and the skin lesion score, we found a positive correlation (r = 0.38; 95% CI: [0.09; 0.61]), i.e., higher (worse) prognostic scores were correlated with higher (worse) skin lesion scores (Figure 2). Consequently, we can say that flocks with feather pecking or cannibalism or both have a higher (worse) prognostic score (Figure 3). Figure 2. View largeDownload slide Correlation between the average prognostic score and the plumage score on the left side (negative) and between the average prognostic score and the skin lesion score on the right side (positive). The black line indicates the simple regression line. Figure 2. View largeDownload slide Correlation between the average prognostic score and the plumage score on the left side (negative) and between the average prognostic score and the skin lesion score on the right side (positive). The black line indicates the simple regression line. Figure 3. View largeDownload slide Distribution of the average prognostic scores (y-axis) along categories of feather pecking and cannibalism. The prognostic score has a significant positive relation to the occurrence of feather pecking (P = 0.03) and cannibalism (P = 0.03). These occurrences for each problem are presented with yes/no. Box plots with thick black lines show the median, boxes show the upper and lower quartiles, whiskers show the maximum and minimum values, and circles show the outliers. Figure 3. View largeDownload slide Distribution of the average prognostic scores (y-axis) along categories of feather pecking and cannibalism. The prognostic score has a significant positive relation to the occurrence of feather pecking (P = 0.03) and cannibalism (P = 0.03). These occurrences for each problem are presented with yes/no. Box plots with thick black lines show the median, boxes show the upper and lower quartiles, whiskers show the maximum and minimum values, and circles show the outliers. Further analysis using logistic regression models revealed that the chance of developing feather pecking was 2.5 times higher when the prognostic score rose 0.1 points (OR = 2.5; 95% CI: [1.14; 6.33]). For the chance of developing cannibalism, we got a similar result (OR = 2.4; 95% CI: [1.14; 6.07]). When the prognostic score exceeded 1.57 points, the probability of feather pecking in the laying period was 0.5; for cannibalism, this threshold value was 1.45 points. Although correlations were significant and plausible, further analysis revealed that the prognostic quality of the prognostic score was not good enough. For example, the linear relationship between the prognostic score and the plumage score explained only about 10% of the variation of the plumage score. For the skin lesion score, this value was about 14%. In addition, predictions by the fitted logistic regression models were not satisfactory (average sensitivity and specificity <0.75). Therefore, we tested if other combinations of the possible 53 risk factors might raise the prognostic quality (results are presented in the following section). Prognostic tool: Single scores For the analysis of the single scores, only scores from complete datasets were used. Not every score could be assigned in each flock (e.g., not every poultry farmer offers litter directly), which reduced the 53 scores to 35 (Table 1 underlined scores). We performed a principal component analysis as implemented in R’s “prcomp” function on these scores. This method applies singular value decomposition on the full data matrix, which yields the principal component transformations. Essentially, these transformations are linear combinations of the single scores, but, in contrast to the prognostic score, different weights (loadings) were assigned to each single score, whereas the prognostic score uses the same weights. By default, this analysis yields as many components as single scores, in our case, 35. However, the order of the components is directly related to their importance with respect to the information in the data, i.e., the first component explains the most variation of the data matrix, the second the second most, and so on. Usually, only a relatively small number of components was needed to explain the main characteristics of the original data. This allows us to reduce the dimension of the data even further by choosing only the first few components. Figure 4 shows the prognostic quality that resulted when up to 10 components were used within the logistic regression setup for predicting feather pecking and cannibalism. For feather pecking, sensitivity and specificity yielded excellent results for a minimum of 6 components. For cannibalism, one more component was needed to obtain comparable values. Assessing the overall classification quality by the criteria accuracy and Dice coefficient, we reached the same conclusion, i.e., that 6 and 7 components were sufficient for predicting feather pecking and cannibalism, respectively. Figure 4. View largeDownload slide Prognostic quality of logistic regression models with different numbers of components as covariates (x-axis) for predicting feather pecking (top row) and cannibalism (bottom row). For each component, the quality criteria sensitivity, specificity, accuracy, and Dice coefficient are depicted (y-axis). Box plots with thick black lines show the median, boxes show the upper and lower quartiles, whiskers show the maximum and minimum values, and circles show the outliers. Figure 4. View largeDownload slide Prognostic quality of logistic regression models with different numbers of components as covariates (x-axis) for predicting feather pecking (top row) and cannibalism (bottom row). For each component, the quality criteria sensitivity, specificity, accuracy, and Dice coefficient are depicted (y-axis). Box plots with thick black lines show the median, boxes show the upper and lower quartiles, whiskers show the maximum and minimum values, and circles show the outliers. Figure 5 depicts the corresponding weights for the single scores that were needed to construct the first 6 components. The sign and magnitude of these loadings allowed us to identify the scores that were most relevant for each component. For example, mainly responsible for the construction of the first component were the scores for “air quality,” “ammonia noticeable,” “dust,” “Do the laying hens peck the investigator?” “Do the laying hens peck each other?” and “Do the laying hens chase each other?” Thus, the first component mostly summarized the topics “climate” and “pecking behavior.” However, we also derived smaller positive and negative loadings for the remaining scores so that this interpretation to neglect these can be seen only as a simplification. In addition, the interpretation that the sign of the single score shows which kind of relation, a positive or a negative, it has on the development of feather pecking and cannibalism is unconfirmed. It shows only the relation it has on the corresponding component. Figure 5. View largeDownload slide Loadings of each single score on the first calculated 6 components that are used for prediction. The loadings represent different corresponding weights of the scores for the development of feather pecking and cannibalism and consist of the magnitude (size of the circles) and the sign (color). Each component is based on the 35 single scores with differing estimated loadings. All components have an impact on the following one. Figure 5. View largeDownload slide Loadings of each single score on the first calculated 6 components that are used for prediction. The loadings represent different corresponding weights of the scores for the development of feather pecking and cannibalism and consist of the magnitude (size of the circles) and the sign (color). Each component is based on the 35 single scores with differing estimated loadings. All components have an impact on the following one. DISCUSSION General Results And Plumage And Skin Lesion Scores This study was designed to develop an easy-to-use tool for farmers or their consultants that gives them a prognosis for a possible development of feather pecking or cannibalism in their current laying hen flocks. As the conditions that may be related to these problematic behaviors are multifactorial (Savory and Mann, 1997; Rodenburg et al., 2013; Schwarzer et al., 2015), a choice of possible risk factors related to management types, barn settings, and laying hen behaviors were assessed. In addition, independent of the results that the farmer receives from the prognostic tool or the principal components, he is able to have a closer look at each rate he got in each question and to improve himself by following the description of the next higher score in this topic. By following these descriptions, he would reduce his prognostic score and, as you can see in the results section, a difference in 0.1 points in the prognostic score changes the chance of feather pecking and cannibalism development 2.5 times. Furthermore, the different farming systems, laying hen strains, and flock sizes of the presented study are representative of the conditions in currently practiced laying hen farming, suggesting that the designed tool could be applicable in various laying hen systems (only in loose housing systems). There is no need for the future use in practice of the assessment of the laying hens, since this procedure was necessary only for the establishment of this forecasting system. In addition, the very high inter-rater reliability results of the 3 investigators facilitate the good requirements for the used analysis. The 53 scores we used for benchmarking the different farms by far outnumbered the base data of 43 flocks. Thus, we refrained from investigating further factors such as food ingredients, behavioral studies, and early-life rearing conditions. Therefore, our results cannot reveal whether the propensity for severe feather damage and cannibalism developed during the rearing or the laying period, since it was shown that already the rearing period seems to have an influence on the occurrence of feather pecking in the laying period (Bestman et al., 2009; Schwarzer et al., 2015). We wanted to keep it simple, to allow the realization of the tool by the farmer on its own. Checking the rearing conditions, behavioral observations, and a food ingredients analysis would not provide easy, practical use. Nonetheless, we observed that both feather damage and wounds developed during the laying period. At the first visit, when the hens were approx. 20 wk old and at the beginning of the laying period, only one flock showed problems with cannibalism, and none showed severe feather pecking. During approx. 47 wk, the time between the first and the second visit, 14 flocks developed severe feather pecking, and 24 flocks showed ≥10% of the hens to have skin lesions of >0.5 cm and thus a problem with cannibalism. Prognostic tool: Prognostic score In a first step, we investigated the prognostic value of the newly designed average prognostic score and found a significant correlation between the prognostic score and both the plumage score and the skin lesion score: Assessed flocks with a better prognostic score showed less occurrence of feather pecking and thus a higher (better) plumage score. In addition, they showed less occurrence of cannibalism and thus a lower skin lesion score, which means fewer laying hens with wounds >0.5 cm. Furthermore, when the prognostic score exceeded a threshold value (1.57 for feather pecking; 1.45 for cannibalism), the probability of observing the 2 problems exceeded 0.5. This result could help laying hen farmers as a guide value that indicates a need for action. Spindler et al. (2013) developed a system for laying hen farmers to identify critical parameters they should monitor on their farms and a so-called emergency plan for critical situations, e.g., when the loss rate increases extensively and the plumage and skin condition of the laying hens worsen. In the studies by Holle et al. (2008) and Keppler et al. (2014), farmers could use benchmarks to compare their own management practices and the behavior of their laying hens with those of other farmers and make use of this comparison to improve conditions on their farms. Both studies compared flocks by using various scores and benchmarks with the aim to find possible risk factors of welfare parameters and to sensitize the farmers to their management practices. In the present study, we took a similar approach. First, we selected typical factors with possible risk for the development of feather pecking and cannibalism and scored them into 3 categories. With the help of these benchmarks, the assessed flocks were comparable. A possible inter-rater variability for the assessment of the laying hens got tested and achieved very high results and so can be neglected; in addition, the assessment of the laying hens will not be necessary in practicable use. Based on the comparisons between the assessed flocks, we could identify the factors (scores) that differed and used them as benchmarks to create a prognostic tool. The novel aspect of this study was the validation of the tool by comparing the results with the actual development of feather pecking and cannibalism problems in each flock. The comparisons showed a significant correlation between the estimated prognostic score and the developed problems. A drawback in the course of this study was an outbreak of the avian influenza in November 2016. As a result, the timing of the second visit varied widely (57th to 83rd wk of life). However, this variation can be neglected because the prognostic tool should be used to predict whether feather pecking and cannibalism are likely to occur but not when they may occur. Although the prognostic quality of the significant correlations between the prognostic score and the plumage and skin lesion scores were not satisfying, the identified correlations are an immense progress in the construction of an early alert system that is needed to solve the problems of feather pecking and cannibalism. As pointed out by Spindler et al. (2013), the existing guidelines for critical situations need to be amended with a system that can predict these problems so that farmers can implement preventive measures. Prognostic tool: Single scores As a second step, single scores were investigated in a principal component analysis to achieve a higher prognostic quality by reducing the number of factors (53 scores) relative to the number of assessed flocks (43). Whereas the prognostic score was constructed by using the same weights for all single scores, the principal components were obtained by allowing different weights (loadings). We could reduce the assessed scores of the prognostic tool to the smallest possible number with the highest possible weights (loadings) by means of the principal component analysis (yielding 6 components – Figure 5). The loadings consisted of a magnitude and a sign. These do not allow identifying a possible risk factor, but they have an influence on the components, for which none of the assessed scores, even the ones with low magnitude, can be neglected. This procedure led to a highly prognostic quality of the identified principal components. Based on these results, this novel tool is suitable for the forecast of the development of feather pecking or cannibalism in laying hens. As a next step, software for this tool, accessible for practical use, should be developed and validated. CONCLUSION The prognostic tool presented here is a highly valid and easy-to-use prognostic system, which shows only slight deficits in its prognostic quality. These deficits can be explained by the large number of scores (53) used to assess a small number of base data (43 flocks). A subsequent principal component analysis reduced the number of scores to 6 principal components that yielded a very high probability of a correct prognosis of the development of feather pecking and cannibalism. ACKNOWLEDGEMENTS This work was supported financially by the Bavarian State Ministry of the Environment and Consumer Protection via the Bavarian Health and Food Safety Authority (grant number: Az. StMUV: 47d-G7131-2011/28-30). The project was planned and realized by the Chair of Animal Welfare, Animal Behavior, Animal Hygiene and Animal Husbandry of the Ludwig-Maximilians-University in Munich, Germany, in cooperation with Bavarian poultry consultants from the Departments of Food, Agriculture and Forestry in Pfaffenhofen an der Ilm and Kitzingen, Germany. The funding source played no role in study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the paper for publication. Conflict of interest statement There is no conflict of interest regarding any financial, personal, or other relationships with other people or organizations that could inappropriately influence or be perceived to influence this study. Ethical statement The work described in this article with research on live animals met the guidelines approved by the institutional animal care and use committee (IACUC). Authorship statement The idea for the paper was conceived by Helen Louton and Angela Schwarzer. The experiments were designed by Helen Louton, Angela Schwarzer, and Michael Erhard. The experiments were performed by Miriam Zepp, Franziska Helmer, and Anne-Katrin Kaesberg under the supervision of Helen Louton and Angela Schwarzer. The data were analyzed by Paul Schmidt. The paper was written by Anne-Katrin Kaesberg. REFERENCES Aerni V., El-Lethey H., Wechsler B.. 2000. Effect of foraging material and food form on feather pecking in laying hens. Br. Poult. Sci.  41: 16– 21. Google Scholar CrossRef Search ADS PubMed  Allen J., Perry G.. 1975. Feather pecking and cannibalism in a caged layer flock. Br. Poult. Sci.  16: 441– 451. Google Scholar CrossRef Search ADS PubMed  Bestman M., Koene P., Wagenaar J.-P.. 2009. Influence of farm factors on the occurrence of feather pecking in organic reared hens and their predictability for feather pecking in the laying period. Appl. Anim. Behav. Sci.  121: 120– 125. Google Scholar CrossRef Search ADS   Bestman M., Wagenaar J.-P.. 2003. Farm level factors associated with feather pecking in organic laying hens. Livest. Prod. Sci.  80: 133– 140. Google Scholar CrossRef Search ADS   Bestman M., Wagenaar J.-P.. 2014. Health and welfare in Dutch organic laying hens. Animals . 4: 374– 390. Google Scholar CrossRef Search ADS PubMed  Blokhuis H. 1986. Feather-pecking in poultry: Its relation with ground-pecking. Appl. Anim. Behav. Sci.  16: 63– 67. Google Scholar CrossRef Search ADS   Blokhuis H., Arkes J.. 1984. Some observations on the development of feather-pecking in poultry. Appl. Anim. Behav. Sci.  12: 145– 157. Google Scholar CrossRef Search ADS   Blokhuis H. J., Fiks Van Niekerk T., Bessei W., Elson A., Guémené D., Kjaer J. B., Maria Levrino G. A., Nicol C. J., Tauson R., Weeks C. A., Van De Weerd H. A.. 2007. The LayWel project: Welfare implications of changes in production systems for laying hens. Worlds Poult. Sci. J.  63: 101– 114. doi 10.1017/S0043933907001328. Google Scholar CrossRef Search ADS   Brenninkmeyer C., Knierim U.. 2015. Förderung der Tiergesundheit und des Tierwohls ökologischer Legehennen in Europa – Promoting good health and welfare in European organic laying hens: Abschlussbericht. Accessed Jun. 2017. Available from: http://orgprints.org/29166/7/29166-11OE020-uni-kassel-knierim-2015-HealthyHens-merkblatt.pdf. Byrt T., Bishop J., Carlin J. B.. 1993. Bias, prevalence and kappa. J clin epidemiol . 46: 423– 429. Google Scholar CrossRef Search ADS PubMed  Chow A., Hogan J. A.. 2005. The development of feather pecking in Burmese red junglefowl: The influence of early experience with exploratory-rich environments. Appl. Anim. Behav. Sci.  93: 283– 294. Google Scholar CrossRef Search ADS   Dice L. R. 1945. Measures of the amount of ecologic association between species. Ecology . 26: 297– 302. Google Scholar CrossRef Search ADS   El-Lethey H., Aerni V., Jungi T., Wechsler B.. 2000. Stress and feather pecking in laying hens in relation to housing conditions. Br. Poult. Sci.  41: 22– 28. Google Scholar CrossRef Search ADS PubMed  European Commission. 2007. Council regulation (EC) No 834/2007 of 28 June 2007 on organic production and labelling of organic products and repealing regulation (EEC) No 2092/91. German Order on the Protection of Animals and the Keeping of Production Animals. 2006. Amended and promulgated on August 22nd, 2006, last changed on February 5th, 2014. Verordnung zum Schutz landwirtschaftlicher Nutztiere und anderer zur Erzeugung tierischer Produkte gehaltener Tiere bei ihrer Haltung; Tierschutz–Nutztierhaltungsverordnung (TierSchNutztV) in der Fassung der Bekanntmachung vom 22. August 2006 (BGBI. I S. 2043), die durch Artikel 1 der Verordnung vom 5. Februar 2014 (BGBI. I S. 94) geändert worden ist. Green L. E., Lewis K., Kimpton A., Nicol C. J.. 2000. Cross-sectional study of the prevalence of feather pecking in laying hens in alternative systems and its association with management and disease. Vet. Rec.  147: 233– 238. Google Scholar CrossRef Search ADS PubMed  Gunnarsson S. 2000a. Laying hens in loose housing systems: Clinical, ethological and epidemiological aspects . PhD thesis, Sveriges Lantbruksuniversitet ( Swedish University of Agricultural Sciences), Uppsala. Gunnarsson S., Algers B., Svedberg J.. 2000b. Description and evaluation of a scoring system of clinical health in laying hens. Gunnarsson, S.: Laying hens in loose housing systems . Doctoral thesis, Swedish University of Agricultural Sciences, Uppsala. Hartcher K. M., Tran M. K., Wilkinson S. J., Hemsworth P. H., Thomson P. C., Cronin G. M.. 2015. Plumage damage in free-range laying hens: Behavioural characteristics in the rearing period and the effects of environmental enrichment and beak-trimming. Appl. Anim. Behav. Sci.  164: 64– 72. Google Scholar CrossRef Search ADS   Heerkens J. L., Delezie E., Kempen I., Zoons J., Ampe B., Rodenburg T. B., Tuyttens F. A.. 2015. Specific characteristics of the aviary housing system affect plumage condition, mortality and production in laying hens. Poult. Sci.  94: 2008– 2017. Google Scholar CrossRef Search ADS PubMed  Holle R., Müller-Arnke I., Rahmann G.. 2008. Tiergesundheitspläne in der Ökologischen Legehennenhaltung. Praxis trifft Forschung Neues aus der Ökologischen Tierhaltung . 2008: 143. Huber-Eicher B., Wechsler B.. 1997. Feather pecking in domestic chicks: Its relation to dustbathing and foraging. Anim. Behav.  54: 757– 768. Google Scholar CrossRef Search ADS PubMed  Huber-Eicher B., Wechsler B.. 1998. The effect of quality and availability of foraging materials on feather pecking in laying hen chicks. Anim. Behav.  55: 861– 873. Google Scholar CrossRef Search ADS PubMed  Keppler C., Moeller D., Knierim U.. 2014. Economic analysis of a management tool for the improvement of laying hen health on organic farms. Proceedings of the 6th International Conference on the Assessment of Animal Welfare at Farm and Group Level, L. Mounier, I . Veissier, Clermont-Ferrand, Page 71. Kjaer J. B. 2009. Feather pecking in domestic fowl is genetically related to locomotor activity levels: Implications for a hyperactivity disorder model of feather pecking. Behav. Genet.  39: 564– 570. Google Scholar CrossRef Search ADS PubMed  Kjaer J. B., Sørensen P.. 2002. Feather pecking and cannibalism in free-range laying hens as affected by genotype, dietary level of methionine + cystine, light intensity during rearing and age at first access to the range area. Appl. Anim. Behav. Sci.  76: 21– 39. Google Scholar CrossRef Search ADS   Landis J. R., Koch G. G.. 1977. The measurement of observer agreement for categorical data. biometrics  33: 159– 174. Google Scholar CrossRef Search ADS PubMed  Leenstra F., Maurer V., Galea F., Bestman M., Amsler-Kepalaite Z., Visscher J., Vermeij I., van Krimpen M.. 2014. Laying hen performance in different production systems: Why do they differ and how to close the gap? Results of discussions with groups of farmers in The Netherlands, Switzerland and France, benchmarking and model calculations. Eur. Poult. Sci.  78: 1– 10. Lenz A. 2015. Federpicken und Kannibalismus bei nicht-schnabelgekürzten Legehennen in Praxisbetrieben, Feather pecking and cannibalism amongst non-beak-trimmed laying hens on laying farms: Einfluss von Management und Haltung, influence of management and husbandry . Diss. Dr. med. vet., Lud'wig-Maximilians-University Munich, Germany. Louton H., Bergmann S. M., Rauch E., Liebers C., Reese S., Erhard M. H., Hoeborn C., Schwarzer A.. 2017. Evaluation of welfare parameters in laying hens on the basis of a Bavarian survey. Poult. Sci.  96: 3199– 3213 Google Scholar CrossRef Search ADS PubMed  Lugmair A. 2009. Epidemiologische Untersuchungen zum Auftreten von Federpicken in alternativen Legehennenhaltungen Österreichs . PhD Diss. Univ. Vienna, Austria. LWK-Niedersachsen. 2016. Minimierung von Federpicken und Kannibalismus bei Legehennen mit intaktem Schnabel. Neue Wege für die Praxis: Managementleitfaden. New ways for practice: management guideline. Agricultural Chamber of Niedersachsen, Oldenburg, Germany. Accessed Jun. 2017. https://www.mud-tierschutz.de/fileadmin/SITE_MASTER/content/Dokumente/Downloads/Leitfaden_LH_Minimierung_Federpicken_Kannibalismus_2813MTD003-1.pdf. Niebuhr K., Zaludik K., Gruber B., Thenmaier I., Lugmair A., Baumung R., Troxler J.. 2006. Untersuchungen zum Auftreten von Kannibalismus und Federpicken in alternativen Legehennenhaltungen in Österreich – Empfehlungen für die Praxis. in Online-Fachzeitschrift des Bundesministeriums für Land- und Forstwirtschaft, Umwelt und Wasserwirtschaft, 2006 , Vienna, Austria. NMELV. 2017. Empfehlungen zur Verhinderung von Federpicken und Kannibalismus bei Jung- und Legehennen. Niedersächsisches Ministerium für Ernährung, Landwirtschaft und Verbraucherschutz . Ministry of Food, Agriculture and Consumer Protection of Niedersachsen, Hannover, Germany. Accessed Jun. 2017. https://www.google.de/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&ved=0ahUKEwjl1oKQyPTUAhWIbhQKHUVaAiIQFggmMAA&url=http%3A%2F%2Fwww.ml.niedersachsen.de%2Fdownload%2F118043%2FEmpfehlungen_zur_Vermeidung_von_Federpicken_und_Kannibalismus_bei_Jung-_und_Legehennen_neu_2017.pdf&usg=AFQjCNHG_cmZUW5TOXxxJCur8vjzqZQlww&cad=rja. Plattner C. 2015. Verhalten nicht-schnabelgekürzter Legehennen in Boden- und Freilandhaltung mit Fokus auf das Pickverhalten, Behaviour of non-beak-trimmed laying hens in alternative housing systems with a special focus on the pecking behaviour . Diss. Dr. med. vet., Ludwig-Maximilians-University Munich, Germany. R Core Team. 2015. R: A Language and Environment for Statistical Computing . R Foundation for Statistical Computing, Vienna, Austria. Ramadan S., Von Borell E.. 2008. Role of loose feathers on the development of feather pecking in laying hens. Br. Poult. Sci.  49: 250– 256. Google Scholar CrossRef Search ADS PubMed  Rodenburg T., Van Krimpen M., De Jong I., De Haas E., Kops M., Riedstra B., Nordquist R., Wagenaar J., Bestman M., Nicol C.. 2013. The prevention and control of feather pecking in laying hens: Identifying the underlying principles. Worlds Poult. Sci. J.  69: 361– 374. Google Scholar CrossRef Search ADS   Sanotra G., Vestergaard K., Agger J., Lawson L.. 1995. The relative preferences for feathers, straw, wood-shavings and sand for dustbathing, pecking and scratching in domestic chicks. Appl. Anim. Behav. Sci.  43: 263– 277. Google Scholar CrossRef Search ADS   Savory C. 1995. Feather pecking and cannibalism. Worlds Poult. Sci. J  51: 215– 219. Google Scholar CrossRef Search ADS   Savory C. J., Mann J. S.. 1997. Behavioural development in groups of pen‐housed pullets in relation to genetic strain, age and food form. Br. Poult. Sci.  38: 38– 47. doi 10.1080/00071669708417938. Google Scholar CrossRef Search ADS PubMed  Schwarzer A., Louton H., Bergmann S., Rauch E., Erhard M.. 2015. Endbericht: Maßnahmen zur Verbesserung des Tierschutzes bei Legehennen in Praxisbetrieben , Ludwig-Maximilans-University Munich, Germany. Spindler B., Schulze Hillert M., Hartung J.. 2013. Abschlussbericht Praxisbegleitende Untersuchungen zur Prüfung des Verzichts auf Schnabelkürzen bei Legehennen in Praxisbetrieben , TiHo, Hannover, Germany. Stadig L. M., Ampe B. A., Van Gansbeke S., Van den Bogaert T., D’Haenens E., Heerkens J. L., Tuyttens F. A.. 2015. Opinion of Belgian egg farmers on hen welfare and its relationship with housing type. Animals  6: 1. Google Scholar CrossRef Search ADS   Steenfeldt S., Kjær J. B., Engberg R. M.. 2007. Effect of feeding silages or carrots as supplements to laying hens on production performance, nutrient digestibility, gut structure, gut microflora and feather pecking behaviour. Br. Poult. Sci.  48: 454– 468. Google Scholar CrossRef Search ADS PubMed  Szczepanek A. 2016. Untersuchungen zu Risikofaktoren für das Auftreten von Federpicken und Kannibalismus bei nicht-schnabelgekürzten Legehennen in Praxisbetrieben . Diss. Dr. med. vet., LMU Munich, Germany. Tauson R., Svensson S.. 1980. Influence of plumage conditions on the hen's feed requirement. Swed. J. Agr. Res.  10: 35– 39. Vestergaard K. S., Skadhauge E., Lawson L.. 1997. The stress of not being able to perform dustbathing in laying hens. Physiol. Behav.  62: 413– 419. Google Scholar CrossRef Search ADS PubMed  Zentralverband der deutschen Geflügelwirtschaft, e. V. 2015. Pressemitteilung: Verzicht auf das Schnabelkürzen: Geflügelwirtschaft unterzeichnet freiwillige Vereinbarung mit Bundeslandwirtschaftsminister Schmidt. Accessed: Jun. 2017. http://www.zdg-online.de/uploads/tx_userzdgdocs/Gefluegelwirtschaft_unterzeichnet_Vereinbarung_zum_Verzicht_auf_das_Schnabelkuerzen.pdf. © 2017 Poultry Science Association Inc.

Journal

Poultry ScienceOxford University Press

Published: Mar 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off