Noninvasive, In-pen Approach Test for Laboratory-housed Pigs

Mild injuries to the brain that do not result in overt anatomical and structural changes detectable with state-of-the-art imaging can be difficult to identify and treat²⁸. However, patients with mTBI are especially vulnerable to additional insult that can cause significant damage to the brain, and therefore, it is important for this population to be identified. Behavioral tests developed in a mini-pig model of mTBI are especially relevant to human mTBI patients because pigs have a similar physiology as humans and express similar affective states, such as anhedonia^8,9,10,20. Here, we have developed a noninvasive, in-pen behavioral test (the HAT), and have shown that it is sensitive enough to distinguish mTBI pigs from sham pigs. In addition, a weighted index (the AI) was developed for behaviors observed during the HAT that are ubiquitous across housing and pig types.

Modifications and Troubleshooting:

The methodologies for the HAT were established based on ethology guidelines²⁵ and several trial-and-error strategies for improving the reliability, repeatability, and validity of the test⁶. Reliability measures helped identify the strengths and limitations of the test. Reliability defines the extent to which the measurement is repeatable and consistent and free from random errors^28,29. We have previously reported on the intra- and interobserver reliability of the HAT, and with the additional structural ethograms, reliabilities are similarly high (Pearson's R² > 0.90) for duration⁶. Frequency and latency measures require trained observers, whereas duration measures are less observer-dependent, and therefore, more reliable across laboratories³⁰.

Reliability within a laboratory and repeatability across laboratories is dependent on methods. In our laboratory, the video system recorded continuously, the files were initially stored as 5-minute files, and some HAT sessions occurred over two files. Fewer mistakes were made when the exact time from the data sheet was used to clip and combine the videos. Before developing the ethogram, observers were allowed to pause, stop, and rewind the video footage to timestamp all of the behaviors in the entire ethogram. This method not only caused variation in timestamping each sample, ranging from 3 min to 20 min, but the between- and within-observer reliability was also poor for most behaviors. Therefore, we set the playback speed, and had observers timestamp each category at a time. Therefore, when reliability was low in just one category, observers independently retimestamped just the category rather than the entire ethogram, after they consulted the definitions and footage together. The set playback and category methods allowed for a consistent prediction of how much time was needed to timestamp each sample. For projects that span longer than a month, routine reviewing of the coded videos and within-observer reliability is important to measure.

Another factor that reduces reliability and repeatability is the video set-up. Initially, a handheld camera and a tripod were used, which were moved from pen to pen. When this method was used, the pigs needed to be introduced to the tripod and camera before the HAT; otherwise, the pigs appeared to react to the tripod and movement more than to the test-human. In addition, nonoverhead camera angles limited the observer's view during timestamping and the depth perception of space increased the within- and between-observer variation in the spatial behavior measures; therefore, we developed the protocol with fixed cameras. When this method is used, extra care is needed to make sure the camera is placed correctly before each test, and more time is needed for the set-up in between each pig's session. However, we learned that the continuous video overhead system needed to start its initial recording at midnight at least 24 h before the first HAT. The timestamp display for many video systems are not accurate and in sync down to the frame; therefore, we no longer rely on the display times. The midnight starts allowed for the exact frame-capturing and video editing, and the timestamp display was not used.

In addition, the acclimation of the pigs and setting up a routine was important when troubleshooting this test. In footage of pigs that were not well acclimated to their environments, pacing was observed during the HAT. This is an indicator that the pig may be in an agitated state³¹ rather than in an exploratory state³². Acclimation periods of three or more weeks may reduce the number of pigs that pace in an experiment. However, if pacing persists throughout all the sampling periods, this ethogram may need to be adjusted to include walks and standing still.

Validity is the extent to which a measurement represents the intended scope of the question being asked²⁵. When first developing the HAT, we used only a spatial ethogram. Definitions of spatial ethogram behavior describe accurately and specifically the proximity to the human subjects and they tell the observer directly how much space the pig leaves between itself and a human. However, once these methods are needed to be applied to a new laboratory set-up, we recognized that spatial ethograms are laboratory specific. Pen dimensions and the placement of other objects influence the outcome of the spatial ethogram; therefore, a diagram with measurements and specifics of the pen will need to be published if the pen set-up has not been previously reported. In addition to reporting on the pen environment, the structural behaviors were added to the ethogram. Unlike spatial behaviors, structural behaviors can be evaluated more easily across laboratories; these behaviors have validity because they specifically describe the pig's level of active state. When a pig is resting, it is likely not motivated to approach and is unable to change positions to approach as quickly as a standing pig. Similarly, a pig displaying NNOB is in an exploratory state, but a pig with its head still while standing is more likely in a catatonic state. The nose orientation helps with validity because the nose, ears, and then eyes are what the pig uses to gather information about the human.

Limitations of the Technique:

A potential concern with this technique is the variability in the pigs' responses to the test-human. In addition, pigs will look at the test-human's hands, which can cause unintentional cueing by that person. Therefore, these limitations were expressed through the experimentally testing of 1) the pigs' responses to a familiar human and unfamiliar humans, and 2) standardizing that, after the pellet is dropped, the test-human stands still and places their hands out of the pig's sight. Data showed that there were no treatment or treatment x time differences during the HAT (Table 2), suggesting that the HAT could be administered by either familiar or unfamiliar humans. Other researchers suggest that pigs tend to generalize about humans based on previous interactions^11,12,13; therefore, a pig's previous experiences with humans need to be positive. This challenge can also be remedied with a vigilant experimental design; for each block, a sufficient number of experimental units represented for each treatment of interest is needed.

In this study, although there were only two experienced observers timestamping all the videos for all three housing types, there were differences among the housing types for specific behavior outcomes (Table 1). For example, pigs in housing type B entered the closest area more often than those in housing types A and C. This is likely due to a difference in the pen material; in housing type B, the front of the pen was a chain-linked gate with horizontal bars that allowed the pig to climb the gate during the HAT. Housing types A and C, on the other hand, had vertical bars and fewer horizontal surfaces for the pigs to climb on. This variability can be remedied by adding the duration spend in the close and closest areas before comparing them across housing types (Table 1; P > 0.10). However, pigs in housing type C spent more time in the far area than those in housing types A and B (Table 1; P < 0.05), which was likely due to the placement of waterers in the back of the pen rather than at the front of the pen. This is a limitation that can be remedied if laboratories choose to standardize the placement of waterers, bowls, and toys and make sure that they are fixed so that the pig does not move the object into another area.

This test has great accessibility for laboratories of all types, but, as mentioned previously, the manually stamped spatial ethogram and measurements will vary more across laboratories. Nonetheless, the body- and head-structural ethograms are ubiquitous. Laboratories that have access to validated, automated tracking for pigs may benefit by having the spatial ethogram tracked automatically rather than manually because the distance moved and the rate of movement may be additional outcomes of behavior measures from the HAT. The limitations from the in-pen set-up and traditional technologies rather than test areas and automatic tracking technologies may be remedied by adapting the AI formula. The AI provides standardized measurements and terminology for how individual pigs use their pen space and express interest in a human. This calculation, derived from common behavior measures, is sensitive to porcine models of subconcussive mTBI and, possibly, other states of subclinical injury or sickness. In addition, the AI reduces random variations during experimentation and may be more easily compared across experiments and laboratories than methods that rely on more experiment-specific measurements. The structural behaviors provided the foundation for this formula, because these behaviors are standard measurements across treatments, whereas the spatial behaviors are dependent on the pen set-up, number of pigs in a pen, and tracking system. For example, we observed that when two healthy pigs are tested in a pen, they will perform similar spatial behaviors by approaching together, but the pig that follows the first may orient its nose more towards its pen mate than towards the human and express more NNOB, because the pig who leads serves as the sentinel. Nonetheless, the AI aids in reducing this variation even from paired behaviors.

Although the AI is an excellent toolset for standardizing the test across laboratories, researchers may still want to examine test-specific behavior outcomes within a laboratory or experiment, especially if they have enough power (i.e., experimental units and repeated tests) in a single experiment. Therefore, Table 1, containing all behavior outcomes, the variance, the distribution, and a calculated test for the number of animals for each specific behavior, was included here. For example, if researchers have pen environments that allow pigs to consistently climb during the HAT and they know their treatment causes over a 75% difference in climbing behaviors, then they can justify animal numbers based on the variance measured. If new behaviors are added to the ethogram, scientists will need to justify which behaviors are indicative of approach or withdrawal before incorporating them into the index. For example, if most of the animals within an experiment pace along the walls of the pen (i.e., thigmotaxis)³², the duration of this behavior could be incorporated into the body-structural ethogram category. The behavior could be represented in the detailed stacked bar graph (i.e., Figure 5) or tabular form, and then, it can be summed up with stand-still before applying the index calculation. The AI, therefore, can represent behaviors that are ubiquitous across laboratories, but additional unique behaviors can still be represented separately.

Significance with Respect to Existing Methods:

The existing methods for the HAT were established for pigs on commercial farms to assess animal welfare. Here, a protocol for laboratory pigs has been established, which can help researchers to assess animal welfare and distinguish mTBI pigs from sham-treated pigs. An alternative traditional test could be to use an open-field test. This test was previously used to assess pig emotionality and welfare³³. Open-field tests were originally designed for testing the affective states of rodents by measuring their natural aversion to open space and light. In contrast, healthy pigs may view the same stimuli as appetitive¹⁰, and after a sickness, injury, or stress treatment, they likely express fear. This test requires more laboratory space and will require pigs to acclimate to being handled and placed in an open-field arena. If laboratories have space and protocols for handling the pigs are in place, repeated HAT sessions, in addition to one open-field test, may help further distinguish treated animals from sham-treated animals.

Critical Steps Within the Protocol:

The first three steps in the protocol are the most critical for successful HAT measures. The sessions per pig only take 3 min; however, adequate preparation will help make this test reliable. As stated above, the camera location and recording set-up is crucial for clarity and replication. Improper camera angles can limit the observer's vision, which will add errors to the measurements. Another often overlooked step is fixing the objects in the pen. The pig will move unfixed objects, and this may affect its motivation to approach the human. The set-up and managing system are important because pigs need to be acclimated to their environments before they can perform the test consistently. Pigs that are not well-acclimated to their home pens or routine or are experiencing stress will defecate in other areas rather than at the back of the pen³⁴. The defecation area may affect their motivation to approach. From the camera view, the observer should be able to identify individual pigs; however, it is important that the marking scheme does not provided information about the animal's treatment, as this will bias the observer²⁵.

Identifying pigs is very important for obtaining the correct behavioral data for the right pig, even when they are single-housed. Pigs are often moved for their treatments, and a marking reinsures the observer that they are watching the same pig after it has been removed and placed back in its pen. Pigs can be housed in pairs, as in housing type C, and therefore, it becomes very important to identify the pigs. Livestock marking paints and markers require daily application; therefore, this protocol requires the use of a medical grade tape and a smear of tag cement. The tape sticks best to pigs with longer hair. Pigs with short hair and dry skin will slough off the tape more often than pigs with longer hair.

Future Applications:

In summary, the noninvasive in-pen HAT test described here is sensitive enough to detect mild and temporal dependent changes in pigs after mTBI. Moreover, we have developed a weighted index called the AI to evaluate changes in pigs housed in different pen types, as well as in different types of pig. Although the HAT has been used to detect changes in pigs exposed to mTBI, this behavioral test can be useful for detecting measurable behavioral changes in animals experiencing stress or prepathological conditions.