1. Introduction
Pigs spend a significant amount of their time using oral manipulation to explore objects around them by sniffing, digging or chewing. This behavior can be redirected to other conspecifics, especially in barren housing systems [
1,
2], and can lead to various behaviors such as tail biting, ear biting or belly nosing [
3]. Tail biting is one of the most important problems in modern pig production. Beside economic losses due to carcass trimming or full loss as consequence of spread infection, it causes stress, painful injuries and thereby reduced animal welfare in affected pigs [
4,
5,
6].
Three different forms of tail biting are distinguished by the behavior of the performing pig: “two-stage tail biting”, “sudden-forceful tail biting” and “obsessive tail biting” [
7]. A tail biting outbreak may occur with usually one pig starting to bite and thereby encouraging further pen mates to join the behavior [
8,
9]. Within an outbreak, switching roles between performers and victims were observed [
10,
11]. Analysis of tail biting behavior is difficult because it occurs relatively rarely and seems unpredictable [
12]. A wide range of influencing factors have been discussed by several authors (e.g., [
4,
7,
9,
13,
14,
15]). For example, refs. [
10,
16,
17] found lighter piglets to be more likely to bite pen mates and heavier piglets were bitten more frequently, whereas [
18] found tail biters to be heavier than non-tail-biters. However, several authors failed to confirm significant correlations between a pigs’ body weight and its frequency of performed tail biting [
11,
19,
20].
Several solutions were evaluated to prevent a tail biting outbreak such as environmental enrichment. Fewer wounded tails, more exploration behavior and improved animal welfare were found when pigs were offered organic objects and substrate which is ingestible, smellable, chewable, deformable, and destructible like roughage, wood shavings, or peas as enrichment material (e.g., [
5,
21,
22,
23]).
Another approach is to establish methods for early detection of occurrence of tail biting such as scoring of tail lesions or changes in behavioral traits. Tail biting outbreaks often seem to begin with first visible tail lesions two to three weeks after weaning, with the development of the percentage of injured pigs and the severity of the injuries depending on the design of the barn, including the enrichment material offered [
9,
24,
25,
26,
27,
28,
29]. In many cases, these lesions are recognized with a delay of a few days after the frequency of tail biting behavior has already increased [
30].
When analyzing the nosing behavior of pigs, Camerlink and Turner [
31] found a low percentage of social nosing behavior that were followed by subsequent potentially injurious oro-nasal behavior. Within these behaviors, tail biting, ear biting and belly nosing were correlated with nosing of the corresponding body parts. In further studies positive correlations between tail biting and other manipulation behavior were found with biters spending more time to manipulate enrichment material than victims [
10,
32] and also tended to bite more parts of the body [
32,
33] than non-tail-biters. In contrast, Larsen et al. [
34] found no alteration of object manipulation prior to an outbreak but overall lower object manipulation rates in pens with tail damage. Furthermore, increased tail-in-mouth behavior [
30,
35] and activity level [
10,
33,
34], as well as an alteration of tail posture prior to a tail biting outbreak [
36,
37,
38,
39,
40,
41], are discussed as signs for early detection of tail biting.
Little is known whether predispositions for later tail biting can be already deduced from the piglets’ behavior during the suckling period. Prunier et al. [
12] reviewed several predisposing factors for biting experienced in early life as for example social stress due to fighting for teats or other resources. The authors pointed out an effect of this stress on aggressive biting during hierarchy formation but could not transfer this effect to non-aggressive forms such as tail biting [
12]. Furthermore, Hoy et al. [
11] did not find an association of certain teat positions during suckling with performed tail biting during rearing.
In the present study we followed individual agonistic behavior from suckling until the end of the rearing period at the age of 70 days. Analyses of suckling behavior is described in detail in Warns [
42]. Here, we report on manipulation behavior of the weaned pigs with a special focus on performers and victims of tail biting across the rearing period of six weeks. Evaluation of general health condition and possible tail lesions of the pigs further provided additional information to the behavioral analysis. In a further step a possible relationship between agonistic behavior shown by a piglet during the suckling bouts and its manipulation behavior during rearing was analyzed. From these results, we wanted to deduce whether agonistic behavior shown during the competitive suckling situation could be used to predict tail biting during the later rearing phase.
2. Materials and Methods
The study was carried out at the Agricultural Test Center VBZL Haus Duesse of the Agricultural Chamber of North Rhine-Westphalia in Bad Sassendorf, Germany, between August 2017 and March 2018. The animals were kept according to the German Animal Protection Law and the Animal Welfare Livestock Farming Regulation.
The experiment was carried out in two repetition groups with a time interval of five weeks between trials. During suckling, twelve litters were housed in conventional farrowing crates (4.58 m
2 per crate) and included 174 piglets across both repetitions. The environmental temperature during suckling was automatically regulated by forced ventilation and set at 23 °C. The piglets had free access to a piglet nest with a heated floor plate, which was completed by a heat lamp for the first 48 h after birth. Usually, the animals had artificial light between 07:00 h and 16:00 h in addition to daylight via windows. During the entire suckling period the piglets had free access to water by a drinking trough. No enrichment material was offered to the piglets during suckling. Agonistic suckling behaviors of the piglets were analyzed individually (further details see [
42]). Out of these piglets, we chose 59 weaner pigs per repetition for analysis of rearing behavior with a total of 118 pigs (59 castrated males and 59 females;
Table 1) across both repetitions. The non-tail docked pigs were crossbreds of a Piétrain sire and a Topigs20 or Topigs70 dam and had an average birth weight of 1.44 ± 0.34 kg. Pigs were weaned with an age of 28 days and an average weight of 7.80 ± 1.44 kg. From day 21 after birth the pigs had ad libitum access to a feeding trough with a commercial rearing feed (feed A; 13.6 MJ ME, 15.0% CP, 1.40% Lys, 0.20% Na).
From all pigs who reached a weaning weight of 6 kg or more, 59 pigs per repetition were chosen and distributed to rearing pens with nine (one pen) to ten pigs (five pens) each. By composing the groups, we considered similar weaning weight, a maximum of three pigs from the same litter and a balanced sex ratio in each group (
Table 1).
The conventional rearing pens for ten pigs were sized 2.1 × 1.7 m (0.36 m2 per pig) with a fully slatted floor and no bedding material. In two pens, the available space was smaller because of a ventilation shaft, so only nine pigs were housed in these pens (0.37 m2 per pig). The environmental temperature was automatically regulated by forced ventilation and set on 29 °C at the beginning of rearing (day 29 of life). During the subsequent weeks, it was decreased stepwise until 24.5 °C at the end of rearing (day 70 of life). Usually, the animals had full artificial light (intensity: 90 lux in average) between 07:00 h and 16:00 h additionally to daylight. At the end of the rearing period, the pigs had an average weight of 22.80 ± 3.10 kg.
During the rearing period, the animals had ad libitum access to feed in a feeding trough and fresh water by drinking nipples. Feeding took place automatically by a dry feeding system. Until day 49 of life, the pigs received rearing feed A which they were used to from the last week of the suckling period. From day 50 of life, the diet was gradually changed during the next four days to rearing feed B (13.4 MJ ME, 16.0% CP, 1.31% Lys, 0.20% Na) which was fed until the end of the rearing period. The animal to feeding place ratio was 3:1 in pens with nine pigs (3.31 m2) and 3.3:1 in pens with ten pigs (3.56 m2). In each pen there was a metal chain with a plastic piece and a cotton rope provided as enrichment material. Twice a day, about 200 g of pellet mix of alfalfa, hay, corncob, and straw pellets were added to a pig bowl in each pen for oral manipulation. If tail biting occurred, pigs were offered a jute sack attached to the wall of the affected pen as additional enrichment material.
The individual body weight was measured weekly outside of the pens on a commercial stationary digital pig weight scale to the nearest 100 g (piglet scale, Meier-Brakenberg).
Possible tail lesions of each individual pig were scored weekly by the same observer during weighing. The individual general health condition was scored weekly in the pen using a scoring scheme modified from the Welfare Quality Assessment Protocol
® for Pigs (Ref. [
43],
Table 2) and included lameness, injury of the ears, the carpal joints, and the body sides.
The scoring scheme for possible tail lesions (modified from FLI [
44];
Table 3) included tail length, hair coat, cleanliness, skin perforation, blood, and necrosis.
During the entire suckling period each pen was equipped with a stationary HD camera (Dallmeier DF4820HD-DN/IR, 720 p, 6 fps) and the pigs were filmed continuously over the entire suckling phase of four weeks. The behavior of the pigs during 30 suckling bouts was analyzed for each pig individually as described in Warns [
42]. In brief, based on the occurrence of agonistic traits during suckling, a dominance index [
45] and social tension index [
46] was calculated for each piglet individually. The dominance index was calculated by dividing the difference between the number of winds and defeats of social interactions by the sum of wins and defeats of social interactions [
45]. For the social tension index, the difference between the sum of all the aggressive actions an animal has performed and the sum of all the aggressive actions an animal has received was calculated [
46]. A positive dominance index of a piglet was the result of a high rate of success in teat disputes, whereas a high positive social tension index expressed a high willingness to initiate a teat dispute to gain access to a teat.
Similarly, during rearing, each rearing pen was equipped with a stationary HD camera (Dallmeier DF4820HD-DN/IR, 720 p, 6 fps) and the pigs were filmed continuously over the entire rearing phase of six weeks. For evaluation of the individual behavior of the pigs, each animal was marked with individual geometric symbols on its back using a commercial animal marking spray (Raidex®, Raidex GmbH, Dettingen, Germany; red, green and blue). The first marking took place directly after transfer to the pens. The pigs were remarked every two to three days to avoid fading of the marks. To ensure individual identification of pigs, only videos taken during the time with artificial light (07:15 to 16:00 h) were chosen for further analysis. In each repetition, we analyzed two days per week resulting in a total of twelve observation days per pig and repetition (total of 24 observation days across both repetitions and pens).
Videos were analyzed by the same observer for each pig individually by using the software Mangold Interact
® (Version 14.3.9.0; Mangold International GmbH; Arnstorf, Germany). Six different behavioral patterns were coded, separated in animal directed and object directed behavior with three behavioral patterns each. The animal directed behavior included tail biting, ear biting and belly nosing. Tail and ear biting were recorded continuously within the first ten minutes of an hour from 08:00 h to 15:10 h, including information of the performer and the victim. Additionally, the reaction of the victim of tail biting was noted. For belly nosing and the three object directed behavioral patterns (manipulation of the rope, the pellets in the pig bowl, and the jute sack) the frequency of occurrence was determined by instantaneous scan sampling with a 30 sec interval within the above-mentioned time frame. For these four behaviors, the performer was noted (
Table 4).
The statistical procedures were performed with R (Version 1.2.1335, RStudio, PBC, Boston, MA, USA). For analyzing general health condition and tail lesions of the rearing period we converted the evaluated scores to binary data and compared the different scoring times by chi-square tests (function “chisq.test”; [
47]). For illustration of the development of behavioral traits during the entire rearing period, manipulation behaviors were analyzed by calculating average frequencies per pig, hour, and observation day. For the continuously recorded tail and ear biting, we determined the individual frequency of the behavior within the first ten minutes of an hour, extrapolated this result to the total hour by multiplying the result by six (according to Zonderland et al. [
30]) and then calculated the average frequency of behavior per pig, hour and observation day. The remaining manipulation behaviors (recorded by scan sampling) were calculated as percentage of sampling points per hour and day in which behaviors were shown. For this purpose, we divided the determined frequency of behaviors shown at sampling points by the amount of sampling points per observation day and calculated an average per pig, hour and observation day.
Further statistics were based on the frequencies of behavior per individual, which were summarized per behavior over the entire observation period. Analyses of variance were performed for performer and victim of tail and ear biting, belly nosing, rope, pellets and sack manipulation (function “aov”; [
48]) with repetition, sow, sex and pen of the pigs as fixed effects. Parameter “sow” referred to the sow pigs were raised at, so it described not necessarily the genetic mother of the pig. Tukey post-hoc tests were performed by function “TukeyHSD” [
49]. Additionally, Spearman rank correlation coefficients were calculated (function “cor.test”; [
50]) with a confidence interval of 95% between weaning weight, weight at the end of rearing period, frequency of analyzed manipulation behaviors during rearing and dominance index and social tension index of the suckling piglets. Scatterplots were used to visualize the relationships between dominance index, social tension index and performer and victim of tail biting by the function “ggplot” (package “ggplot2”; [
51]). Principal component analysis was performed to evaluate relationships between the behavior traits of the suckling and rearing period (function “PCA” of package “FactoMineR”; [
52]) and its result was visualized in a scree plot for percentages of explained variances (function “fviz_eig” of package “factoextra”; [
53]) and bar plots (function “fviz_contrib” of package “factoextra”; [
53]) for contribution of variables and individuals to Dimension 1–2. Cluster analysis was performed with function “eclust” (package “factoextra”; [
53]) to determine two clusters. Results of cluster analysis were visualized with function “fviz_silhouette” (package “factoextra”; [
53]). A high positive silhouette value (S
i) represented the similarity of an individual with its own cluster and poor matching with other clusters. Finally, principal component analysis was performed within the two calculated clusters as described above. Level of significance was set at
p < 0.05. Values are means ± SD if not otherwise stated.