Table 1.
Farm characteristics.
Fig 1.
Stacked bar graphs of frequency of tail injury scores.
Tails were scored on severity (0, 34.9%; 1, 5.6%, 2, 19.6%; 3, 23.6%; 4, 16.4%), freshness (0, 59.8%; 1, 27.1%; 2, 9.8%; 3, 3%; 4, 0.3%; 5, 0.1%), length loss (0, 85.4%; 1, 13.5%; 2, 0.8%; 3, 0.3%), and presence of swelling (0, 93.2%; 1, 6.8%) (See S1 File for score descriptions).
Table 2.
Summary of how scores were combined to generate variables used in modelling.
Fig 2.
Stacked bar graphs of frequency of flank injury scores.
Severity (0, 96.2%; 1, 0.3%; 2, 1.1%; 3, 2.4%), freshness (0, 96.2%; 1, 1.9%; 2, 1.2%; 3, 0.5%; 4, 0.1%) and size (0, 96.4%; 1, 1.7%; 2, 1.9%) of flank biting lesions are shown. (See S1 File for score descriptions).
Fig 3.
Stacked bar graph of frequency of ear injury scores.
Severity (0, 30.0%; 1, 56.4%; 2, 13.4%; 3, 0.2%), freshness (0, 86.9%; 1, 9.2%; 2, 3.0%; 3, 0.9%; 4, 0.0%), swelling (0, 99.8%; 1, 0.1%; 2, 0.1%) and size (0, 32.8%; 1, 66.8%; 2, 1.2%; 3, 0.0%) of ear injury are shown. (See S1 File for score descriptions).
Fig 4.
Stacked bar graph of frequency of other health scores.
Lameness (0, 91.9%; 1, 7.0%; 2, 1.1%), Nasal discharge (0, 99.9%; 1, 0.1%), Ocular discharge (0, 12.3%; 1, 87.7%) and skin lesions resulting from social aggression (0, 49.1%; 1, 49.6%; 2, 0.9%; 3, 0.4%) (See S1 File for score descriptions).
Fig 5.
Example of categorisation of scoring periods as either increasing or decreasing tail injury.
The percentages 20, 40 and 30 are included for illustration purposes to show the weighted total tail injury severity, and how changes in this metric would lead to a period being classified as increasing or decreasing. E.g. from 20% to 40% is an increase, while 40% to 30% is a decrease.
Fig 6.
Timeline diagram of how focal periods were selected and how camera data was validated against injury / health scoring.
Coloured squares indicate separate scoring events. The period of time between the first and last scoring event was identified as the focal period, and the space of time between consecutive scoring events were referred to as the scoring periods. The end of the batch (Unknown 2) was removed as no further injury scoring was gathered in order to validate the tail posture changes during this time. The start of the batch (Unknown 1) was removed due to variable data quality at the start of some batches.
Table 3.
Correlations between health and injury scores used in statistical modelling.
Fig 7.
Example of tail posture data for a typical batch- stacked bar graph of proportion of low (red), mid (yellow) and high (green) tails by day.
Fig 8.
Example of tail posture data for a batch in which tail injury disrupted tail posture-stacked bar graph of proportion of low (red), mid (yellow) and high (green) tails by day.
Table 4.
Results of linear mixed models of tail posture by injury and health scoring metrics.