In this study, we conducted two identical replicates each with four rearing (8 groups in total with 24 piglets each) and eight fattening pens (16 groups in total with 12 pigs each). Groups were composed of both sexes and in total 192 undocked crossbred pigs (German Piétrain × German Hybrid) were used (replicate 1: 52 females and 44 castrated males, replicate 2: 57 females and 39 castrated males). The study was carried out from February 2019 to August 2019 in forced ventilated barns at the Bildungs- und Wissenszentrum Boxberg (LSZ), Germany. At weaning at an age of about 4 weeks, piglets weighted on average 7.9 kg (±1.7 kg SD) and were individually equipped with one UHF RFID tag per ear (MS Tag Round UHF, MS Schippers, Netherlands) that we used for data collection during the study. The 96 piglets per replicate were randomly assigned to four rearing pens with 24 piglets each. Rearing pens were continuously numbered across replicates (8 PenIDs in rearing in total) and had 15 m² (5 m × 3 m) floor space. The floors consisted of slatted plastic floor (7.5 m², 38.5% perforation), slatted concrete floor (3.0 m², 17.0% perforation) and partly slatted concrete floor (4.5 m², 7.0% perforation) under a heated covering (Figure 1A). All rearing pens were equipped at any time with a rubber bar, two sisal ropes, a piece of wood hanging on metal chains and a self-constructed dispenser (see below). Mashed feed was offered ad libitum to the piglets in two phases by automatic feeders (animal:feeding place ratio 2.4:1). After 2 weeks of rearing, feed composition changed (detailed information is given in the Supplementary Table 1). Due to upcoming feed change in fattening, piglets were fed pellet concentrate (25 kg per pen) in addition to the ad libitum offered mashed feed during the last week of rearing. Piglets had ad libitum access to water from two nipple drinkers and two drinking bowls (Suevia 92 R, Suevia Haiges GmbH, Kirchheim, Germany) during rearing. Three piglets had to be excluded (one loss due to a disease and two identified tail biters were separated), resulting in 95 pigs in replicate 1 (51 females and 44 castrated males) and 94 pigs in replicate 2 (56 females and 38 castrated males).

After 7 weeks of rearing, piglets weighted on average 30.0 kg (±5.5 kg SD). As in rearing, the floor space of the fattening pens was 15 m² (3 m × 5 m) and consequently the group size had to be reduced. Thus, pigs were divided in eight fattening pens per replicate (Figure 1B) by randomly assigning piglets of each rearing pen to two groups, respectively (12 pigs per fattening pen). By that, pigs of a given fattening pen had the same experience and treatment during rearing and were familiar to each other. Fattening pens were continuously numbered across replicates (16 PenIDs in fattening in total) and were equipped with slatted floor (17% perforation). In each fattening pen pigs had access to two sisal ropes, a piece of wood hanging on metal chains and a self-constructed dispenser (see below) at any time. Fattening pigs had ad libitum access to water from four nipple drinkers in each pen. Pellet concentrate was offered ad libitum from a single feeder to the pigs. Feed composition changed when pigs reached an average weight of about 80 kg (detailed information is given in the Supplementary Table 1). Data collection ended after 11 weeks of fattening, when the first pigs were removed for slaughter.

Throughout the whole observation period, a stock of chopped barley straw with a length of ~10–100 mm was ad libitum provided to all pigs in all pens by a dispenser. At least every second day the straw was manually filled into the storage tube of the dispensers (if necessary, also on weekends).

In addition, at the beginning of rearing, pens were assigned to one of two treatments. Piglets in half of the rearing pens additionally received once per day (1 refilling = R1) chopped barley straw (i.e., the same material as permanently available in the storage tube) into the rooting area of the dispenser. Piglets in the other half of the pens received this additional rooting material four times per day (4 refillings = R4) (Supplementary Figure 1). Provision of this additional chopped straw was done automatically (see Supplementary Video 1) and continued during fattening (four pens per treatment, as mentioned above). In treatment R1 additional chopped straw was daily provided at 0800 (first refilling). In treatment R4 the rooting areas of the dispensers were filled at 0800 (first refilling), 1000 (second refilling), 1400 (third refilling), and 1600 (fourth refilling) each day. Filling times were chosen based on previous evaluations of the activity of pigs in order to avoid disturbing of resting phases around noon (e.g., Kauselmann et al., 2020, 2021b). In a previous study we found that fattening pigs consumed higher amounts of chopped straw from the dispenser compared to weaned piglets (Kauselmann et al., 2021a). Thus, at each provision, additional chopped straw was delivered for 15 s in rearing and 20 s in fattening resulting in ~80 g straw/refilling in rearing and ~110 g straw/refilling in fattening. Thus, in treatment R1 pigs got only a quarter of the additional amount of chopped straw compared to the pigs of treatment R4. However, at any time pigs still had ad libitum access to the chopped straw in the dispenser and the additional provision of straw was intended to trigger/enhance rooting behavior.

The self-constructed dispensers (Kauselmann et al., 2021a) were installed in the middle of each pen (Figure 1) and consist of a 100 cm high storage tube with a diameter of 25 cm (Supplementary Figure 2). Around the base of the storage tube a 10 cm high cement ring with an inner diameter of 63 cm was installed to create a rooting area. The chopped straw could be rooted into the rooting area through a 2.4 cm wide gap between the storage tube and the floor by moving a plastic bolt. According to Averberg et al. (2018) a width of 18 cm per pig at the dispenser was used to calculate the animal:dispenser place ratio for rearing pigs and 33 cm per pig was used for fattening pigs. Thus, we obtained an animal:dispenser place ratio of 1.7:1 in rearing and 1.5:1 in fattening pens.

In the storage tube of the dispenser, we installed a dust- and waterproof UHF RFID antenna (Kathrein MiRa ETSI, KATHREIN Solutions GmbH, Ismaning, Germany) at a height of 50 cm above the floor (Supplementary Figure 2C). The UHF RFID antenna read the UHF RFID tags in the ears of the pigs when they stayed at the rooting area of the dispenser. Data were collected in a database via a converter and an UHF RFID reader. Pigs were assigned to the corresponding UHF RFID antenna and, thus, to their pen and the respective treatment by a software application (Phenobyte GmbH & Co. KG., Ludwigsburg, Germany). As described in Kauselmann et al. (2021a) the antenna output power was set to 22.4 dBm and UHF RFID recordings with a maximum pause of 30 s between two readings were summarized to obtain optimal sensitivity (70.7% in rearing and 79.3% in fattening) and specificity (99.4% in rearing and 99.5% in fattening). With the UHF RFID system, we recorded for each individual pig the exploration times that were summed up to exploration durations per pig and hour, which we used for further statistical analysis.

Four times during the test period (at the beginning of rearing, at the beginning of fattening, after 6 weeks of fattening and at the end of fattening; see Supplementary Figure 3), the pigs were scored according to a scoring scheme developed to assess tail and ear damages of, i.e., the Deutscher Schweine Bonitur-Schlüssel (2017) to evaluate the length of the tails and skin lesions at the tails. Tail length was scored in five categories: (0) full/natural length, (1) length loss up to one-third, i.e., here we also include individuals with only minimal losses at the tassel, (2) distinct damages, i.e., length losses from at least one third up to two-thirds, (3) length loss more than two-thirds, and (4) total loss with a maximum of 1 cm leftover for piglets in the rearing and 2 cm leftover for pigs in the fattening. However, score 4 was never recorded and score 1 often concerned a small part of the tail tip rather than one-third of the tail. Injuries at the skin of the tails were scored in four categories: (0) no injuries, (1) superficial perforation of the skin, punctually or in the form of a line, (2) deeper perforation of the skin, maximum as large as the diameter of the tail at the respective point, and (3) deeper perforation of the skin, larger than the diameter of the tail at the respective point.

Ethical approval was not required for the study because the animals were not isolated from conspecifics, not restricted from any resources, remained in their familiar environment and were not manipulated in accordance to the study protocol. Furthermore pigs were housed and managed according to the German legislation for farm animals (TierSchNutztV, 2017) and in accordance to Council Directive 2008/120/EC. The different numbers of refillings with rooting material per day, i.e., once vs. four times, were provided in addition to the rooting material available ad libitum. These additional refillings represented additional environmental enrichment and were the only change in the housing environment with respect to the data collection. The additionally provided rooting materials at the refillings (R1 vs. R4), thus, represented no limited resource which might have triggered competitive behavioral patterns, but rooting material was ad libitum available at any time in the dispersers regardless of the refillings (R1 vs. R4). All data were recorded at a licensed farm that produces, rears, fattens, and markets pigs (VVVO-Number: 08 128 0140 538). Thus, pigs from the study were marketed after the study. Beyond minimum German legal requirements, space allotment per pig in the present study was higher and pigs were offered additional enrichment material on a daily basis (as R1 and R4 treatment), which is usually expected to have a positive effect on pigs' welfare, as additional stimulation of explorative behavior is regarded to reduce potential welfare issues such as tail biting. Data on exploration were automatically recorded via an UHF RFID system which was integrated into the dispenser. Scoring of tail damages was conducted carefully and non-invasively using the Deutscher Schweine Bonitur-Schlüssel (2017). Pigs remained with their peers at any time. Pigs were visually inspected on a daily basis. In case of tail biting, additional enrichment materials (paper bags, mineral feed, squeezed oats, and zeolite) were offered and in case biters were identified (two piglets during rearing) they were separated from the group and housed in a separate pen.

All statistical analyses were calculated separately for rearing and fattening using R version 3.3.1 (R Core Team, 2019) and the packages lme4 (Bates et al., 2015), nlme (Pinheiro et al., 2019), and car (Fox and Weisberg, 2019).

We summed up the recorded exploration duration per pig for each hour. Weeks were consecutively numbered in rearing (7 weeks) and fattening (11 weeks) and days within weeks were additionally considered (7 days). The 7 weeks of rearing were assigned to temporal section 1 and for fattening two temporal sections were considered (section 2: first 6 weeks of fattening and section 3: last 5 weeks of fattening; see Supplementary Figure 3), which corresponded to the intervals between the scorings, as in Kauselmann et al. (2020). Although data were log-transformed [log(x+1)], q-q plots showed minor deviations from normal distribution of the residuals. We used linear mixed effect models (LME) that are robust to such minor deviations (Schielzeth et al., 2020). The LME used for analyses of exploration duration per hour included the exploratory variables (i) refilling treatment (2-level factor: R1 and R4), (ii) week of rearing/fattening (rearing: 7-level factor; fattening: 11-level factor) (iii) day within week, (iv) hour (v) two-way interactions between refilling treatment and hour and between refilling treatment and week. As animals within pen were not independent from each other, AnimalID, PenID, and week were considered as nesting random factors in the model for rearing and fattening, respectively. The hierarchical structure of the statistical model is exemplarily shown in the Supplementary Figure 4. In case of significant differences (P < 0.05), pairwise t-tests were used to calculate post-hoc comparisons.

As we found significant interactions between refilling treatment and hour within day on exploration durations in the above mentioned LME's, during both rearing and fattening we further examined this effect and did a more detailed descriptive analyses of exploration duration in 10-min sequences focusing on the hours after refilling the rooting area (0800–0900, 1000–1100, 1400–1500, and 1600–1700) independent from whether refilling was done four times (R4) or only once (R1) per day. All days during rearing and fattening were considered.

In order to relate the changes in tail length to exploration durations during the three sections, we calculated Δ-tail lengths by subtracting for each pig its tail length score at the end of a section from the score at the beginning of the respective section. For example, if a pig had an intact tail (score 0) at the beginning of rearing (beginning of section 1) and a tail length score of 1 after rearing (after section 1), Δ-tail length was 1 (1–0 = 1) for section 1. If the same pig still had a tail length score of 1 after 6 weeks of fattening (after section 2), Δ-tail length was 0 (1–1 = 0) for section 2. When the score was 2 after a total of 11 weeks of fattening (after section 3), Δ-tail length was 1 (2–1 = 1) for section 3. To analyze Δ-tail lengths and skin injuries at the tails we used generalized linear mixed models (GLMMs) with Poisson distribution separately for rearing and fattening. We used (i) the refilling treatment (2-level factor: R1 and R4) and sex (2-level factor: female or castrated male) as explanatory variable and AnimalID as well as PenID as nesting random factors for the one section in rearing. For fattening we considered the explanatory variables (i) section (2-level factor: section 2 and 3), (ii) refilling treatment (2-level factor: R1 and R4), and (iii) sex (2-level factor: female or castrated male). Two-way interactions between all variables were considered, however, due to convergence problems in the model for skin injuries during fattening, we only considered interactions between refilling and section (sex was not considered as there was no significant effect on tail injuries; see section Effects of Refillings of Additional Rooting Material on Tail Damages During Fattening). For fattening the nesting random factors AnimalID and PenID were considered.

The refilling treatment (i.e., R1 vs. R4) at which the chopped straw was additionally distributed into the rooting area of the dispensers did not affect the exploration durations of the piglets during rearing (LME, factor refilling, F_{(1, 6)} = 0.039, P = 0.85). Mean exploration durations per pig and hour were 0.56 min (i.e., summing up to 13.4 min per pig and day) when the rooting areas were additionally filled once daily (R1) and 0.54 min (12.8 min per pig and day) when they were additionally filled four times a day (R4). However, the weeks within rearing had a significantly but not systematic effect on exploration durations of the piglets (LME, factor week, F_{(6, 1131)} = 17.19, P < 0.0001; Supplementary Table 2). Piglets spent most time at the dispenser in week 7 (0.62 min/pig and hour) followed by week 1 (0.60 min/pig and hour). Lowest mean exploration durations were recorded in week 2 (0.44 min/pig and hour). The days within week significantly affected exploration duration of the piglets (LME, factor day, F_{(6, 222677)} = 99.53, P < 0.0001; Supplementary Table 3) but also did not show any consistent pattern. The exploration duration of piglets was also affected by the hours within day (LME, factor hour, F_{(23, 222677)} = 1005.93, P < 0.0001) and showed one peak in the morning and one higher peak in the afternoon. There was a significant interaction between refilling and hour (LME, factor refilling^*hour, F_{(23, 222677)} = 8.39, P < 0.0001; Figures 2A, 3A). Exploration duration of piglets in treatment R1 and R4 increased after refilling the rooting area at 0800, while exploration duration of piglets in treatment R4 also increased after refilling at 1000, 1400 and 1600. Furthermore, between 0800 and 0900, piglets from treatment R1 showed more exploration (0.84 min/pig) compared to pigs from treatment R4 (0.64 min/pig). However, compared to R1 treatment, piglets in treatment R4 showed more exploration within the hours after refilling at 1000 (R1: 0.70 min/pig; R4: 0.80 min/pig) and 1600 (R1: 0.92 min/pig; R4: 1.07 min/pig). The interaction between refilling and week was not significant (LME, factor refilling^*week, F_{(6, 1131)} = 0.66, P = 0.69).

When having a closer descriptive look on the 10-min sequences within the hours after refilling (i.e., 0800–0900, 1000–1100, 1400–1500, and 1600–1700), no clear differences in the course of exploration duration could be found between the refilling treatments during rearing (Figures 2B–E, 3B–E).

There was neither an effect of the refilling treatments (GLMM, factor refilling, X(1)2 = 0.94, P = 0.33; Figure 4A), of sex, nor of the interaction (GLMM, factor sex, X(1)2 = 0.23, P = 0.63; factor refilling^*sex X(1)2 = 0.20, P = 0.65) on Δ-tail lengths. Likewise, skin injuries of tails were neither affected by refilling treatment (GLMM, factor refilling, X(1)2 = 0.13, P = 0.72), by sex, nor by the interaction (GLMM, factor sex, X(1)2 = 1.65, P = 0.20; factor refilling^*sex X(1)2 = 0.49, P = 0.49) during rearing. After rearing, 57.8% of the piglets had intact tails without length losses (category 0).

The refilling treatment of additional rooting material provision did not affect exploration durations of fattening pigs at the dispenser (LME, factor refilling, F_{(1, 14)} = 0.20, P = 0.66). Mean exploration duration at the dispensers per pig and hour was 1.28 min (i.e., summing up to 29.7 min per pig and day) when they received additional chopped straw in the rooting area once a day (R1) and 1.22 min (27.9 min per pig and day) when they were additionally filled four times a day (R4). There was a significant interaction between refilling treatment and week (LME, factor refilling^*week, F_{(10, 1804)} = 3.54, P = 0.0001). The week affected exploration duration of the pigs (LME, factor week, F_{(10, 1804)} = 84.06, P < 0.0001; Supplementary Table 2) but showed no consistent pattern over the 11 weeks of fattening. Highest exploration durations per pig and hour were recorded in week 9 (1.76 min) and lowest in week 1 (0.76 min). Exploration duration was also affected by day of week (LME, factor day, F_{(6, 334465)} = 16.96, P < 0.0001; Supplementary Table 3) but again showed no consistent pattern. Highest exploration durations at the dispenser were recorded on day 2 (1.33 min/pig and hour) and lowest on day 4 (1.19 min/pig and hour) and day 6 (1.17 min/pig and hour). The hours within day also affected exploration duration of the pigs (LME, factor hour, F_{(23, 334465)} = 4594.53, P < 0.0001). Exploration within day showed two peaks, i.e., one peak in the morning and one higher peak in the afternoon. During night time, exploration duration remained at a low level. There was also a significant interaction between refilling and hour within day (LME, factor refilling^*hour, F_{(23, 334465)} = 53.73, P < 0.0001; Figures 5A,B). Exploration duration of pigs that received additional chopped straw once per day (R1) showed a peak within the hour after refilling the rooting area at 0800 and showed more exploration within the hour after refilling at 0800 (1.55 min/pig) compared to pigs in treatment R4 (1.36 min/pig). Pigs that received additional chopped straw four times daily (R4) showed peaks within the hour after all four additional fillings, i.e., after 0800, 1000, 1400, and 1600. Compared to R1 treatment, pigs in treatment R4 showed more exploration within the hours after refilling at 1000 (R1: 1.09 min/pig; R4: 1.63 min/pig), 1400 (R1: 2.03 min/pig; R4: 2.57 min/pig), and 1600 (R1: 3.16 min/pig; R4: 3.45 min/pig).

A descriptive exploration of the 10-min sequences showed that after refilling the rooting area at 0800 (R1 and R4 received additional chopped straw), exploration duration decreased within the first three 10-min sequences and remained at a constant level thereafter in both, R1 and R4 (Figures 5B, 6B). However, pigs of R1 showed higher exploration durations during the first two 10-min sequences compared to R4 pigs. During the hours after refilling the rooting areas only in R4 (i.e., 1000–1100, 1400–1500, and 1600–1700), R4 pigs had higher exploration durations in the first two 10-min sequences (Figures 6C–E), while R1 pigs did not show clear changes in exploration durations (Figures 5C–E) and remained at a low level.

During fattening neither refilling treatment (GLMM, factor refilling, X(1)2 = 1.65, P = 0.20; Figure 4B) nor sex (GLMM, factor sex, X(1)2 = 0.07, P = 0.79) affected Δ-tail lengths. However, there was an effect of section on Δ-tail lengths (GLMM, factor section, X(1)2 = 4.75, P = 0.029). After section 2 Δ-tail lengths of 0 occurred less often (75.3% of pigs) compared to after section 3 (93.6% of pigs), i.e., most losses of tail length occurred in the first weeks of fattening. In total, 31.2% of the pigs had intact tails without losses of tail length after fattening. None of the interaction between the factors significantly affected Δ-tail lengths (GLMM, factor refilling^*section, X(1)2 = 3.02, P = 0.08; factor refilling^*sex, X(1)2 = 0.53, P = 0.47, factor section^*sex, X(1)2 = 1.26, P = 0.26).

Injuries at the tails of the pigs were neither affected by refilling treatment (GLMM, factor refilling, X(1)2 = 0.48, P = 0.49), by section (GLMM, factor section, X(1)2 = 0.40, P = 0.53), nor by sex (GLMM, factor sex, X(1)2 = 0.14, P = 0.71). The interaction of refilling treatment and section also did not significantly affect tail injuries (GLMM, factor refilling^*section, X(1)2 = 0.22, P = 0.64).