1. Introduction
Social housing is an essential tool for promoting the psychological wellbeing of nonhuman primates (NHPs) in research. Studies have shown a variety of physiological and behavioral benefits associated with socialization (e.g., [
1,
2,
3,
4,
5]). Social housing, including continuous full-contact pair housing, provides opportunities for macaques and other NHPs to engage in species-typical behaviors such as grooming and social play [
3] and provides a social buffer against environmental stressors [
6]. Further, full-contact pair-housed macaques have also been shown to exhibit fewer abnormal behaviors when compared to their single-housed counterparts [
7,
8].
The process of pair introduction is often performed in a series of steps [
8], beginning with partner selection. Several studies have identified attributes associated with compatibility between macaque partners, including weight [
8,
9,
10,
11,
12], age [
11,
12], temperament [
9,
13], and previous experience [
8]. Once potential partners are determined, they are typically housed next to one another for a period of time, during which they are given various degrees of visual and/or physical contact. For example, partners may be separated with either a mesh or clear plexiglass partition, allowing visual but no physical contact, or with a “protected contact” partition that allows limited tactile contact. At some point, based on the behavior of the monkeys, the partition is removed, allowing full physical contact between the partners.
The decision about whether to proceed with the social introduction is usually based on the observed interactions between the partners; introductions in which the two partners display many affiliative behaviors toward one another are likely to proceed, while those in which the partners engage in aggressive or overly fearful behavior may be aborted. In order to interact, the monkeys clearly need to be in visual contact with one another. However, in many primate facilities, potential partners may be housed in separate rooms or even different buildings from one another, necessitating the relocation of one or both partners for the introduction. Being moved to a new and unfamiliar environment can be stressful for macaques [
14,
15], and this stress can be compounded if the pair introduction is unsuccessful. In some situations, social introduction attempts may even require the relocation of other monkeys, in addition to those intended for pairing. Therefore, having the opportunity to assess initial behavior
before moving the monkeys may decrease the need for relocations and reduce the stress associated with social introductions.
Video conferencing technology offers the opportunity for animals to communicate virtually. Many organizations have dramatically increased the use of video conferencing in the past couple of years to improve communication, and there is evidence to suggest that macaques may also be able to communicate in this way. Studies have demonstrated the ability of rhesus macaques to recognize and process faces, particularly facial expressions, of conspecifics in still images by measuring activation in the amygdala [
16,
17]. Macaques have also been shown to respond to videos of conspecifics. Rhesus macaques watching unedited videos of unfamiliar conspecifics elicited social engagement with the monkey on video, including looking patterns (gaze aversion and gaze following) and reciprocated facial expressions [
18]. The animals in that study were not provided with treats or other rewards for looking at the screen, suggesting that they were interested in the video and that their reactions were spontaneous [
18]. Further, studies have found that both pre-recorded and live videos of unfamiliar conspecifics were effective rewards for bonnet macaques performing a joystick task [
19,
20]. Taken together, these studies suggest that macaques can recognize and respond to the behavior of conspecifics on video. While promising, the monkeys in those studies were not able to interact with conspecifics in real time. Technologies such as video conferencing provide opportunities for animals to engage in real-time interactions with a conspecific on a video screen.
In this study, we examined the use of the Zoom video conference app (on tablets) as a potential tool to assess pair compatibility in male and female rhesus macaques prior to relocation for pair introduction. Given that prosocial behavior exhibited early in the introduction can predict pair success [
21], we hypothesized that pairs in which the partners demonstrated affiliative behaviors during the Zoom session would be more likely to be successfully pair-housed than those in which there was no prosocial behavior exhibited.
3. Results
3.1. Relationship between Demographic Variables and Behavior on Zoom
3.1.1. Attention
In general, animals spent relatively little time attending to the iPad. All but eight animals (three females, five males) looked at it at least once during the 10 min session. The total amount of attention shown by individuals ranged from 0 to 18 intervals (i.e., 0–60% of intervals), with a median of 4 intervals (i.e., approximately 13% of the 30, 20 s intervals). The best model explaining the amount of attention had an AIC 14.5 points lower than the next best model and was significantly different (chi-square = 20.52,
p < 0.001). Sex and age, as well as a sex-by-age interaction, explained attention shown (
Table 1). However, the coefficient estimate of MtoF is negative (B = −3.26,
p < 0.001), this is offset by the positive coefficient of the age by MtoF interaction (B = 0.22,
p < 0.001), making the overall effect of MtoF positive. Thus, the highest levels of attention were displayed from male to female recipients, and this increased with male age (
Figure 1). The amount of attention males showed to other males (B = −0.04,
p > 0.05), and the amount that females showed to either male or female recipients (B = −0.11,
p > 0.05; B(intercept) = 2.12,
p < 0.001) was not significantly different from one another, and decreased with a main effect of age (B = −0.07,
p < 0.01).
3.1.2. Anxiety
The total amount of anxiety shown by individuals ranged from 0–15 instances, with a median of 2 instances. Three models explained the amount of anxiety equally well when evaluated by AIC and chi-square tests. The model with the lowest AIC had a main effect on sex but no effect on age. The other two models (one with a main effect of age, the other with a main effect of age and an age by sex interaction) were within 2 AIC points of the sex-only model (1.2 and 1.5 points, respectively) and were not significantly better than the sex-only model (chi-square = 0.8,
p > 0.1; chi-square = 6.5,
p > 0.1). Since neither main nor interaction effects of age were significant in either model, we report only the sex-only model here (
Table 2). Males showed more anxiety toward females (B = 1.43,
p < 0.001) and males (B = 1.29,
p < 0.001) than females showed to females. Tukey’s HSD contrasts additionally showed that males showed more anxiety to females than females showed to males (Tukey’s HSD = 0.72,
p = 0.048). However, there was no difference in anxiety shown by males to either male or female recipients (Tukey’s HSD = −0.14,
p > 0.1) or from females to either female or male recipients (Tukey’s HSD = −0.71,
p > 0.1). Thus, males showed more anxiety than females overall, and anxiety did not depend on the sex of the Zoom partner (
Figure 2).
3.1.3. Prosocial
The monkeys in this study showed little prosocial behavior during the Zoom sessions. The total amount of prosocial behavior shown by individuals ranged from 0 to 12 instances, with a median of 0 instances. Thirty-four subjects showed some prosocial behavior, but the remaining 50 subjects did not. The best model for prosocial behavior was 9.9 AIC points lower than the next best model (chi-square = 15.93,
p < 0.01) and contained the main effects of age and sex, with an interaction between age and sex (
Table 3). The most prosocial behavior was shown between pairs of males (B = 3.47,
p < 0.01), but this decreased with male age (B = −0.45,
p < 0.01) (
Figure 3).
3.2. Relationship between Behavior on Zoom and Pair Outcome
Of the 42 pairs involved in this study, 28 (7FF, 6MF, and 15MM) were successfully socialized (i.e., co-housed for at least 14 days), while 14 pairs (5FF, 3MF, 6MM) showed immediate aggression when introduced and thus were not co-housed.
The initial comprehensive model was significantly different from models with fewer variables (chi-square = 10.5,
p < 0.05) and had the lowest AIC by at least 4.5 points; thus, we report the initial comprehensive model. This model (
Table 4) included significant effects of pair sex and attention similarity, with trends for total anxiety, sex by attention, and sex by anxiety interactions. The model predicted that, on average, MM pairs are more likely to be successful than FF pairs, which matches the pattern of successful pairs within our subjects.
Both in our sample and according to the model, MF pairs have a high likelihood of success. However, the model estimate for MF pairings had an outsized standard error (19.48) and was not statistically significant (p = 0.11). This may be a consequence of the lower number of MF pairs relative to other pair types and the number of successful pairs within that category.
The within-pair similarity of attention was the most impactful behavioral indicator of pair success. Predicted success increased when attention was dissimilar within the pair. FF pairs with similar attentiveness had a 0.4% (95% CI: 0.0–52.8%) likelihood of success, compared to a 30.4% (95% CI: 1.3–93.2%) likelihood of success when within-pair attentiveness was dissimilar. MM pairs with similar attentiveness had a 23.3% (95% CI: 3.4–72.3%) likelihood of success, whereas MM pairs with dissimilar attentiveness had a 97.0% (95% CI: 33.8–99.9%) likelihood of success. Thus, pairs in which attention was primarily shown by one animal had a higher chance of success than those in which both individuals showed similar levels of attention toward the iPad. There was a trend toward success for FF pairs that showed low anxiety (B = −1.95, p = 0.07), but for MM pairs that showed more anxiety (B = 2.09, p = 0.07), although these effects were not statistically significant. Neither prosocial behavior nor lag time were predictors of success in this sample.
4. Discussion
Social housing is widely accepted as an integral part of behavioral management with indoor-housed nonhuman primates [
8]. However, finding appropriate partners when NHPs are housed in separate rooms or buildings adds unique challenges to compatibility assessments. Further, the introduction itself can be more stressful when partners are located apart from one another due to the relocation(s) involved for one or more animals [
15]. Having the ability to assess compatibility prior to relocating animals for introductions may help address these concerns. Macaques show measurable interest in videos of conspecifics [
18] to the point of finding them rewarding [
20]. Therefore, we examined the use of live video conference technology (Zoom) as a tool for the initial introduction and compatibility assessment between potential partners in rhesus macaques.
Overall, the monkeys in our study spent relatively little time looking at the screen during the Zoom sessions. The maximum any animal spent looking at the iPad was 60% of intervals. Males with female partners spent more time looking at the screen than males with male partners or females with male or female partners. There was no difference in the amount of time females spent attending to the iPad screen based on the sex of their partner. Attention toward the screen generally decreased with age, with the exception of males who were in a session with a female; for those males, attention increased with age. However, this result should be interpreted with caution, as the sex of the partner we chose for male monkeys was, in part, based on his age. Males with female partners tended to be older than those with male partners (average age of males involved in MF pairs = 13 years +/− 4.1 years; in MM pairs = 6 years +/− 2.4 years). Thus, the increase in attention seen with age for males in MF pairs likely had to do with the sex of the partner and not the age of the male per se. These results are similar to those of a study examining social vigilance in semi-free-ranging rhesus macaques [
29]. In that study, males showed more social vigilance than females, which the authors attributed to differences in social information gathering in the species. Further, those monkeys showed a decrease in social vigilance with age, and there was an effect of dominance rank, with high-ranking individuals more vigilant than lower-ranking monkeys [
29]. The monkeys in our study were single-housed at the time of the Zoom session but had all been socially housed in the past. We do not know their previous dominance rankings or whether that would affect their current attention to social partners.
We focused on anxiety and prosocial behaviors for this study because those behaviors have been shown to predict pair success when seen early in the pair attempt. MacAllister and colleagues found that prosocial behaviors exhibited by the pair on the first day of the introduction were predictive of pair success, while pairs that exhibited anxiety behaviors were often unsuccessful [
21]. Monkeys in our study showed relatively little anxiety or prosocial behavior, which is not surprising given the minimal amount of time they attended to the tablet. Males showed more anxiety than females overall, and anxiety did not depend on the sex of the Zoom partner. Interestingly, males with male partners showed more prosocial behavior, but less attention, than males with female partners. It is possible that this difference was related to the age of the monkeys; as stated above, males in male–male pairs were younger than those in male–female pairs. It could be that younger animals are more likely to show prosocial behavior.
Our focus for undertaking this study was to examine whether allowing animals to interact via video conference technology could help us predict pair compatibility. For example, if one or both animals threatened the other, that could indicate that they might not be compatible. We predicted that pairs that spent time paying attention to each other, particularly if they demonstrated prosocial behavior, would be more successful than those that did not. While the total amount of time animals spent looking at the iPad did not predict pair success, the within-pair similarity of attentiveness (e.g., amount displayed by the partners relative to one another) did. Pairs in which attention was primarily shown by one animal had a higher chance of success than those in which both individuals showed similar levels of attention toward the iPad for both MM and FF pairs. It is possible that this discrepancy is related to a clear dominance relationship for the potential partners, which has been found to help predict pair success in some studies (e.g., [
3]).
Interestingly, prosocial behavior exhibited by the pair did not predict pair success. This finding may be due to the low amount of prosocial behavior displayed in general. Our Zoom sessions were relatively short and thus may not have been long enough to allow for the expression of prosocial behavior. Alternatively, it is also possible that the inability to follow prosocial behavior with physical contact diminished its effect on pair success.
We had also predicted that pairs that demonstrated aggressive or anxious behaviors would be less likely to be successful. We saw very little aggressive behavior; only three animals demonstrated any aggression at all during the Zoom sessions. There was a trend toward pair success in FF pairs that showed low anxiety (B = −1.95, p = 0.07) and MM pairs that showed increased anxiety (B = 2.09, p = 0.07), although these effects were not statistically significant. Anxiety behaviors could indicate either stress due to the inability to avoid the Zoom session entirely or frustration at an inability to have physical contact with the Zoom partner. Thus, anxiety could have actually indicated compatibility for some monkeys. Future studies should examine anxiety in conjunction with social vigilance to parse out these differences.
There were a number of potential confounds and limitations to this study that may have prevented us from finding more striking results. We only provided animals with one ten minute Zoom session due to logistical challenges and intermittent Wi-Fi availability. Longer and/or more frequent sessions most likely would have resulted in more robust findings. The iPad was novel for most individuals, and it is possible the animals could have been wary of the tripod set-up or unaware that they could interact with the tablet. A short training or introduction session to the equipment could potentially eliminate this particular confound. Similarly, the Zoom session was conducted on relatively small tablets. Larger iPads may have allowed the animals greater visual access. In addition, attention to the screen was scored from recordings. While the observer was highly trained, she was not in the room with the monkey during the Zoom session and only scored attention when it was clearly directed toward the screen. Thus, we may have underestimated the amount of time the animals spent looking at the iPad. Using eye-tracking technology may provide more precise data regarding social attention (e.g., [
30,
31]). Other confounds include the time of day when sessions were held. We did not keep to a strict window of time in which to perform Zoom sessions due to husbandry or project-related scheduling conflicts. Some sessions were performed close to normal feeding and enrichment times, which could introduce more opportunities for distraction.
Despite these confounds, our initial data suggest that video conferencing technology may be useful as a tool for introducing unfamiliar partners ahead of their pair attempt. There are also other uses for this technology. In many facilities, partners need to be temporarily separated from one another for research or clinical-related reasons. In these situations, Zoom sessions may be set up between the familiar animals so that they may communicate, thus minimizing the impact of physical separation. At the ONPRC, group-housed rhesus macaques who are injured or sick may be removed from their pen and brought to the clinical hospital. We have used Zoom to allow these monkeys to communicate with their groupmates while they are under clinical care. While we have only used this for a handful of animals so far, we have found that both the sick/injured monkey as well as the groupmates pay a great deal of attention to the screen. More work is needed to examine whether the ability to interact with groupmates affects how well an individual responds to medical treatment and the individual’s integration back into the group after medical treatment is completed.