Efficacy of 3 Types of Foraging Enrichment for Rhesus Macaques (Macaca mulatta)

Abstract

Primate facilities provide environmental enrichment to improve animal wellbeing, increase opportunities for expression of species-typical behaviors, and decrease the occurrence of stereotypic behaviors. The current study assessed the efficacy of 3 types of foraging enrichment: puzzle balls, supertubes, and shakers. We assigned 48 rhesus macaques to 3 experimental groups, each of which received (after a 3-wk baseline observation period) 1 of the 3 enrichment devices intermittently for 3 wk. Observations were collected during 10-min sessions by using 1–0 sampling with 15-s intervals (480 h total). Observations were collected at the same 10 specified time points each week during the baseline period and after enrichment. Data were analyzed by using generalized linear mixed-effects modeling under the assumption that the underlying response followed a Poisson distribution. Foraging behavior increased significantly in all 3 groups and remained increased in some groups when enrichment was removed after 43 h. The 3 enrichment devices had different effects on individual expression of stereotypy: supertubes decreased it, shakers increased it, and puzzle balls led to a decrease followed by an increase. We present potential reasons for the changes in stereotypy and postulate a likely balance between the beneficial and negative effects of enrichment in any given environment.

The behavior of rhesus macaques (Macaca mulatta) living in captivity can differ vastly from that of their wild counterparts, with a reduced repertoire of species-typical behavior²⁸ and a high frequency of stereotypic behavior,^5,17,29 both of which may indicate decreased welfare.^14,17,20,22 To combat these problems, primate facilities provide environmental enrichment that allows for expression of species-typical activities and promotes psychologic wellbeing, as mandated by the Animal Welfare Act.² Beyond legal requirements, appropriate enrichment is especially important in research facilities that use primates as biomedical models. An impoverished environment and exposure to stressors can lead to development of stereotypies in captive primates.¹⁷ These stereotypies may indicate underlying abnormal physiology, such as neural damage or dysfunction.^11-13 Providing appropriate enrichment might prevent such abnormal physiology, ultimately improving translation of animal research to human clinical applications.^11,26,27

Stereotypies are repetitive, invariant behavior patterns with no obvious goal or function.²⁰ Stereotypies commonly exhibited by macaques include motor stereotypies (for example, pacing, rocking, back-flipping, and swaying), postural stereotypies (for example, floating limb), and self-directed stereotypies (for example, digit suck and eye poke).^17,29 The development and expression of stereotypies have been linked to frustration, lack of stimulation and control, and unavoidable stress.^20,22 Stereotypies therefore cause concern because they are representative of potentially decreased welfare experienced by the animal. Stereotypies are not necessarily a problem in their own right, in that most do not cause harm to the animal. Rather, as an outward manifestation of stress, frustration, and lack of stimulation, stereotypies can be useful as a behavioral indicator of welfare.

Captive primate facilities attempt to decrease stereotypies in 2 ways: prevention and remediation. Numerous risk factor analyses have shown that rearing condition, time spent in individual housing, and several experimental procedures can predict the development of stereotypies in rhesus macaques.^17,24,29,30 In many primate facilities, avoiding such risk factors is problematic due to interference with ongoing studies. Therefore further investigation is needed to identify methods that prevent the development of these behaviors without compromising the integrity of biomedical research.

Enrichment for rhesus macaques typically includes chew toys, foraging devices, audio and visual stimulation, and social enrichment.^4,18 Foraging devices come in a variety of forms, from artificial turf boards with food crumbled on top to more complicated puzzles that require extraction of food from the device.¹⁸ A large body of literature on primate enrichment assessment suggests that enrichment varies in terms of its success in decreasing stereotypies in rhesus macaques,^3,6,7,15,25 with some studies showing no success at all.^8,16 Likewise, enrichment can vary in its ability to promote species-typical behavior, with enrichment usage time varying from minutes to hours to days.^18,28 With limited resources for enrichment at primate facilities, identification and validation of the most efficacious enrichment is crucial for maximizing beneficial effects, particularly given the large number of animals at such facilities and their importance as biomedical subjects.

In addition to measuring immediate behavioral response to enrichment, enrichment use and efficacy should be assessed over time. Some previous studies have found foraging devices to be effective in decreasing stereotypies only when subjects were actively using the device and therefore were effective only before all the food was consumed.²⁵ For this reason, evaluating the duration of enrichment use and effectiveness is important for assessing the efficacy of enrichment in mitigating behavioral problems as well as promoting species-typical behavior.

The purpose of the current study was to evaluate the efficacy of 3 foraging enrichment devices—‘puzzle balls,’ ‘shakers,’ and ‘supertubes’—in promoting species-typical foraging and mitigating stereotypical behavior in rhesus macaques. We surmised that providing the macaques an opportunity to engage in a behavior that they are motivated to perform would decrease stress and frustration and thus decrease the expression of stereotypy. Changes in these behaviors were assessed by using a longitudinal approach that allowed us to evaluate both the magnitude and duration of any effects of enrichment on these behaviors. The study design allowed for analysis of the enrichment's immediate effects on macaque behavior and any subsequent changes in behavior over a 2-d exposure period.

Materials and Methods

Subjects.

This study took place from May 2008 through September 2009 at the California National Primate Research Center (Davis, CA). The subjects were 48 (22 male and 26 female) indoor-housed adult rhesus macaques. The mean age of the subjects was 8.9 y, with a range of 3.1 to 20.7 y. Subject animals were selected based on availability and whether they had been observed exhibiting stereotypy. Subjects were housed in cages having a volume of either 4.3 or 6 ft² of floor space, depending on their weight. Forty-two of the subjects were intermittently paired, 4 subjects were continuously paired, and 2 subjects were singly housed with intermittent partial physical contact through a metal mesh divider. Intermittently paired macaques had access to one another from 0800 to 1400 Monday through Friday but were otherwise singly housed. Similarly, singly housed macaques had partial physical access through a metal mesh divider from 0800 to 1400 Monday through Friday but were otherwise singly housed. Subjects were given water ad libitum and fed chow twice daily, once between 0700 and 0800 and once between 1400 and 1500; any food received from the foraging devices was supplemental to their diet. Subjects received additional cage enrichment, including a Nylabone Dental Chew (Nylabone, Neptune, NJ) or Kong Toy chew (Kong, Golden, CO) and both a mirror and forage board affixed to the front of the cage. Forage boards were strips of high-density polyethylene with numerous cavities designed to hold forage material. The forage boards were covered daily with ‘scratch’ made from rice, oats, and granola. These enrichment items are standard to our facility and were in place before the study began. All subject animals were cared for in compliance with protocols approved by the IACUC at the University of California–Davis and adhered to the requirements of the Animal Welfare Act from the US Department of Agriculture.²

Description of enrichment devices.

Each subject was given 1 of 3 enrichment devices: puzzle balls, supertubes, and shakers. Each device was filled with the food item that best suited the device, consistent with how the device would most likely be used in practice. Puzzle balls were hollow, hard plastic balls (diameter, 11.4 cm) that were filled with fruit that could be picked out of multiple small holes in the ball (Figure 1). Puzzle balls were placed directly in cages for the monkeys to hold and manipulate. Supertubes were hollow PVC plastic cylinders (height, 33 cm; diameter, 4 cm) with a long plastic rod on the inside and were hung from a metal hook on the outside of a cage (Figure 2). The inner rod was covered in peanut butter and granola and could be accessed through multiple holes in the outer tube. Shakers were polycarbonate plastic cylinders (height, 7.5 cm tall; diameter, 7 cm) with a single small, tubular opening that were filled with granola (Figure 3). Monkeys had to repeatedly turn and shake the device to retrieve the granola from the opening. Shakers were placed directly in cages for the monkeys to manipulate.

Figure 1.

Puzzle ball enrichment devices.

Figure 2.

Supertube enrichment devices.

Figure 3.

Shaker enrichment device.

Experimental design and data collection.

Subjects were assigned to 1 of 3 treatment groups: puzzle balls, supertubes, and shakers. Each subject animal received only one type of enrichment during the study. Each group comprised 16 subjects, and groups were matched as closely as possible in regard to age, sex, and pairing status of the subjects. Subjects that received shaker enrichment were unfamiliar with the device. Subjects provided with puzzle balls may have had previous experience with the device, and subjects given supertubes may previously have received similar tube enrichment. For all paired animals, both macaques in the pair were study subjects and assigned to the same treatment group.

The study had 2 phases: a 3-wk baseline phase followed by a 3-wk experimental phase. Observations were conducted at the same frequency and times of day during both phases, to control for behavioral changes related to time of day. During the experimental phase, subjects were given the enrichment items between 1500 and 1600 3 times each week (Figure 4). Focal observations were conducted immediately on providing the enrichment device and at 1, 19, and 25 h thereafter. Enrichment was removed after 43 h, and a final observation was obtained with no enrichment in the cage. Observation times were chosen such that observations were least likely to interfere with husbandry procedures.

Figure 4.

Schedule of enrichment and observations. Enrichment was given on Mondays and Wednesdays and removed 2 d afterward. Five observations (0, 1, 19, 25, and 43 h) were conducted after enrichment distribution. Enrichment also was given on Fridays and removed Monday mornings; however, no observations were conducted during this time.

Subjects were observed by using a 10-min, 1–0 focal sampling design with 15-s intervals.¹ Recorded behaviors included motor stereotypies, postural stereotypies, self-directed stereotypies, self-abusive behaviors, abnormal social behaviors, and foraging behavior (Figure 5). Other species-typical behaviors were not recorded because the devices were intended to increase foraging behavior only, and we did not want to reduce observational accuracy by adding superfluous behavioral observations. With the exception of eating biscuits in the cage, all foraging behaviors (including foraging for crumbs and residual scratch at the bottom of the cage) were recorded. This broad definition of foraging was used for 2 reasons. First, food from the devices often fell into the cage, complicating determination of where the food originally came from. Second, we wanted to determine whether the devices encouraged natural foraging behavior, even if this behavior did not occur while an animal used one of the devices. A total of 480 h of data were collected, including both baseline and enrichment phases. Observations were collected by 5 observers, with a joint probability of agreement interobserver reliability of greater than 85%. Observers were trained to record data with minimal interaction and disturbance to the macaques. All subject animals had previous experience with observations from multiple observers and were accustomed to the observation process.

Figure 5.

Behaviors recorded.

Statistical analysis.

Data were analyzed by using generalized linear mixed-effects modeling in SAS (SAS Institute, Cary, NC) under the assumption that the underlying response followed a Poisson distribution. Abnormal behaviors other than motor stereotypy occurred too infrequently or in too few subjects to yield a significant change and therefore were not analyzed. For each enrichment device, a separate model was implemented. To determine the effect of enrichment device on stereotypy and foraging, data were analyzed by using a binary predictor indicating whether the observation was performed during baseline or with enrichment present. Because repeated measurements from the same animal were likely to be correlated, animal identification was treated as a random effect. Because measurements from multiple animals in the same room were likely to be correlated, room was treated as a random effect. The sample size was too small to include additional parameters, such as pairing status (continuous compared with intermittent), sex, and age, in the model. Instead, efforts were made to balance group demographics as much as possible. Separate models were analyzed at each specified hour of observation, by using stereotypy and forage as outcomes. Counts of behaviors were averaged for each macaque at each hour of observation for both baseline and experimental phases. Subjects that were never observed to exhibit stereotypy during both baseline and experimental phases were removed from analyses with stereotypy as the outcome. For these analyses, the subject size for puzzle balls, supertubes, and shaker groups were reduced to 13 (12 intermittently paired, 1 continuously paired), 8 (6 intermittently paired, 2 continuously paired), and 9 (all intermittently paired) subjects, respectively. For all analyses, treatment effects were considered to be significant at a P value of less than 0.05. Time of observation included 0, 1, 19, 25, and 43 h. The 0-h observation was immediately after the enrichment device was given, with the 1-h observation at 1 h after the macaque received the enrichment device, and so forth. Enrichment was removed just before the 43-h observation, so that all 43-h observations were done without a foraging device in the cage.

Results

Foraging.

Puzzle ball.

After provision of the puzzle ball foraging device, foraging behavior increased compared with baseline levels at time points of 0 h (β = 2.31, P < 0.0001; average baseline percentage foraging, 9%; average enriched percentage foraging, 96%), 1 h (β = 2.47, P < 0.0001; baseline, 7%; enriched, 78%), 19 h (β = 0.75, P < 0.0001; baseline, 12%, enriched, 26%), and 25 h (β = 0.96, P < 0.001; baseline, 5%, enriched, 14%; Figure 6). Average percentage foraging across all observations was 9% during the baseline phase and 45% during the experimental phase.

Figure 6.

Change in species-typical foraging behavior with (A) puzzle ball, (B) supertube, and (C) shaker enrichment devices. Percentage change in foraging behavior when subjects received enrichment compared with baseline observations performed at the same time of day. ⁺, P < 0.01; ^∆, P < 0.001.

Supertube.

After provision of the supertube foraging device, foraging behavior increased compared with baseline levels at time points of 0 h (β = 2.80, P < 0.0001; baseline, 5%; enriched, 77%), 1 h (β = 2.52, P < 0.0001; baseline, 2%; enriched, 27%), 19 h (β = 0.45 P < 0.01; baseline, 13%; enriched, 20%), 25 h (β = 2.06 P < 0.0001; baseline, 1%; enriched, 12%), and 43 h (β = 0.70 P < 0.001; baseline,8%; enriched, 16%; Figure 6). Average percentage foraging across all observations was 6% during the baseline phase and 30% during the experimental phase.

Shaker.

After provision of the shaker foraging device, foraging behavior increased compared with baseline levels at time points of 0 h (β = 2.76, P < 0.0001; baseline, 4%, enriched, 69%), 1 h (β = 1.14, P < 0.0001; baseline, 11%, enriched, 35%), 19 h (β = 1.60, P < 0.0001; baseline, 3%, enriched, 15%), and 43 h (β = 0.75, P < 0.01; baseline, 6%, enriched, 11%). Significantly less foraging occurred at the 25-h time point (β = −0.78, P < 0.001; baseline, 15%; enriched, 7%) as compared with baseline levels (Figure 6). Average percentage foraging across all observations was 8% during the baseline phase and 28% during the experimental phase.

Stereotypy

Puzzle ball.

Subjects given the puzzle ball foraging device were significantly less likely to display motor stereotypy compared with baseline levels at 0 h (β = –2.93, P < 0.05; average baseline percentage stereotypy, 4%; average enriched percentage stereotypy, 0%) and significantly more likely to display motor stereotypy compared with baseline levels at the 1-h (β = 0.67, P < 0.05; baseline, 5%; enriched, 9%), 25-h (β = 0.81, P < 0.05; baseline, 3%; enriched, 7%), and 43-h (β = 0.57 P < 0.05; baseline, 4%; enriched, 8%) time points (Figure 7). No significant change in stereotypy was found at all other observation points.

Figure 7.

Change in stereotypy with (A) puzzle ball, (B) supertube, and (C) shaker enrichment devices. Percentage change in stereotypy when subjects received enrichment compared with baseline observations performed at the same time of day. *, P < 0.05; ⁺, P < 0.01.

Supertube.

Subjects given the supertube foraging device were significantly less likely to display motor stereotypy compared with baseline levels at the 19-h (β = –0.90, P < 0.05; baseline, 8%, enriched, 3%) and 43-h (β = −0.85, P < 0.05; baseline, 10%, enriched, 4%) time points (Figure 7). No significant change in stereotypy was found at all other observation hours.

Shaker.

Subjects given the shaker foraging device were significantly more likely to display motor stereotypy compared with baseline levels at the 1-h (β = 1.39, P < 0.01; baseline, 3%; enriched, 13%) and 25-h (β = 1.97, P < 0.01; baseline, 2%; enriched, 12%) observations (Figure 7). No significant change in stereotypy was found at all other observation hours.

Discussion

The purpose of the current study was to determine the efficacy of 3 foraging devices in increasing species-typical foraging behavior and decreasing stereotypy in rhesus macaques. All 3 devices clearly accomplished the goal of increasing foraging behavior (Figure 6). However, a uniform decrease in stereotypy was not observed, in that subjects responded differently to the 3 different devices (Figure 7).

Foraging behavior is a large part of the activity budget of wild rhesus macaques, accounting for as much as 48% of their daily activity.^19,23,31 In contrast, foraging behavior was recorded in less than 10% of all observations during the baseline phase for each device; however, this figure does not include time spent eating the daily ration of chow. Foraging increased to 28%, 30%, and 45% during the experimental phase for the shaker, supertube, and puzzle ball devices, respectively. This increase in foraging and change in time budgeting is promising because it closely represents the activity budgets recorded for wild rhesus macaques.

An unexpected result was the significant increase in foraging behavior at the 43-h time point compared with baseline in subjects that received the supertubes and shakers. This change is surprising because the 43-h observation occurred after the devices were removed from the cage. This result indicates that supertubes and shakers were successful in encouraging foraging behavior and modifying behavioral time budgets even when the devices were no longer present. This finding may suggest that subjects became primed to more fully engage with their environment even after the stimulating object had been removed. However, subjects using shakers exhibited a significant decrease in foraging at hour 25 compared with baseline. We speculate that this result might reflect a rebound effect caused by the previous increases in foraging behavior.

Although the devices were able to increase foraging behavior to a near species-typical level, this increase alone does not necessarily imply that they improved animal welfare. Foraging behavior of rhesus macaque troops in the wild can vary from as low as 11% to as high as 48%,^19,23,31 yet this range does not indicate that the welfare of troops that spend more time foraging is higher than that of others. To confidently show that the provided devices improved welfare would require demonstration that having the opportunity to increase foraging to a species-appropriate level was intrinsically rewarding, decreased frustration, or was in some way beneficial to the animal.

To evaluate whether the increase in foraging opportunity was beneficial to the macaques, we looked for an accompanying decrease in stereotypy. Once an animal has developed stereotypy, multiple environmental triggers can lead to the expression of the behavior, including frustration, unavoidable stress, and low levels of stimulation.²⁰ We expected all 3 foraging devices to decrease frustration by providing opportunities to express a desired behavior, to decrease stress by distracting subjects from potentially aversive environmental stimuli such as other animals or room activity, and to provide stimulation through mental engagement. Therefore, we expected all 3 foraging devices to lead to decreases in stereotypy. However, we also expected that the decreases in stereotypy would be restricted to a brief period after distribution of the devices, because foraging devices are often depleted of food rapidly. Unexpectedly, only puzzle balls led to a significant decrease in stereotypy immediately on their distribution. In contrast, supertubes led to decreases in stereotypy at 19 and 43 h after distribution. Even more surprising, both puzzle balls and shakers actually increased stereotypy during certain hours of observation.

The variability in the devices’ effects on stereotypy indicates that a simple relationship of enrichment decreasing stress leading to decreases in stereotypy is not sufficient to explain our results. As previously mentioned, there are known environmental triggers for stereotypic behavior in animals that have developed stereotypies. ²⁰ In observations where the enrichment led to decreases in stereotypy, we might conclude that the use of enrichment decreased the negative effects of these undesirable environmental triggers. By the same token, when stereotypy increased, we might conclude that the enrichment led to an increase in the effects of these undesirable triggers. Because foraging enrichment is unlikely to decrease stimulation, we must turn to other explanations related to these environmental triggers and the circumstances that might influence them.

First, foraging devices might have led to frustration if the devices were too challenging for animals. Although problem solving can be desirable in an enrichment program, complications arise when animals are unable to solve the problem, leading to distress.²¹ Therefore challenging enrichment must be provided at the optimal level of difficulty that requires problem solving without causing frustration and distress. Increases in stereotypy might have occurred because some of these devices were too difficult for all or some subjects. We considered that shaker enrichment was the most difficult, in that it required manipulation of the device beyond simple picking of food out of holes. This high level of difficulty may explain why there was never a decrease in stereotypy associated with shaker devices.

Second, methods of distribution of the enrichment devices may have led to stress and fear in some of the macaques, thereby leading to increases in stereotypy. Human–animal interaction increased during the enrichment phase, because humans distributed and removed enrichment devices. The cage door was opened to place or remove puzzle ball and shaker enrichment from cages. This activity might have been especially stressful for some macaques and may have negatively affected the human–animal relationship, thereby increasing the likelihood that an animal would express stereotypy in the presence of a human observer. There were no significant increases in stereotypy immediately after the devices were placed in the cage, but this result is likely because the observed high level of foraging out of the newly filled devices was incompatible with stereotypy. In contrast to puzzle balls and shakers, supertubes were latched onto the outside of the cage, without opening the cage door. This arrangement may explain why supertubes were the only device in which no significant increase in stereotypy was observed.

Third, exposure to novel enrichment may have been particularly stressful for neophobic subjects. It is interesting to note that shaker enrichment led only to increases and not decreases in stereotypy, and this enrichment device was the least familiar to the subjects. Perhaps if the study had continued longer than 3 wk, the macaques would have had time to become familiar with the enrichment and stereotypies may have decreased.

Fourth, selective distribution of enrichment may have conflicted with the preexisting social relationships in the room. Rhesus social groups are characterized by a strict linear dominance hierarchy³² and frequent unidirectional aggression.¹⁰ Although the animals used in the current study were not part of a traditional social group, they had visual access to multiple monkeys. In addition, all subjects included in the stereotypy analysis were socially housed and thus shared a cage with a dominant or subordinate partner for a portion of the day. A recent study on group-living rhesus found that high-ranking macaques were likely to punish low-ranking counterparts simply for showing interest in high-quality resources.⁹ Subjects that selectively received foraging enrichment may therefore have received punishment in the form of aggression from a dominant partner or increased threats from nonparticipating animals with visual access. Although room activity and aggression were not evaluated empirically, observers noted that general room activity seemed heightened during the experimental phase of the study. If it is the case that distribution of enrichment to only part of a room leads to aggression, it may be important to distribute enrichment to all, and not just selected, animals in a room. This factor might have implications for both daily management as well as future studies evaluating enrichment.

Fifth, time of day might have influenced response to enrichment. No single enrichment had the same effect at all times of day. Puzzle balls resulted in a decrease in stereotypy at 0 h followed by increases in stereotypy. With most of the fruit removed during the first hour, the beneficial effects of the device may only outweigh its negative effects, such as increased human–animal interaction or aggression from other animals, for a short period. Supertubes decreased stereotypy only at 19 and 43 h, both 1000 observations, indicating that this enrichment may be especially beneficial during morning hours. Shakers increased stereotypy at the 1- and 25-h observations, both of which are 1600 observations and the time at which intermittent pairs are separated from their partners. Therefore we may be seeing an interaction between time of day and enrichment device, such that some devices are particularly stressful or beneficial at specific times of the day. For example, both devices that were placed inside the cage resulted in an increase in stereotypy at the time when intermittent pairs were separated from their partners. During separation, subordinate partners may have been unable to receive visual ‘approval’ from their dominant partner and therefore were fearful of using the high-valued monopolizable resource. Further research is needed as to why specific time interactions were seen with each enrichment device.

In conclusion, multiple factors are likely necessary to explain individual macaques’ responses to enrichment. The variations observed in stereotypy can be thought of as resulting from an ongoing balance between beneficial effects of the enrichment and stressful interactions with the environment. Depending on the time of the day and current environmental activities, the net effect of the enrichment may lead to different levels of stereotypy. The scope of the current study was to determine whether the enrichment devices were successful in increasing species-typical behaviors and decreasing stereotypies. As such, the design did not test which of the aforementioned factors were implicated in the subjects’ responses to the enrichment. If evidence demonstrating that enrichment can have beneficial as well as undesirable outcomes continues to accumulate, future research should investigate specifically what aspects of enrichment devices are positive and which are negative. This differentiation can include studies on the optimal level of challenge for enrichment, the effect of human distribution of enrichment, how dominance relationships affect enrichment use and stress, and how all of these factors change depending on the time of day, pairing status, and so forth. In addition, interindividual differences should be accounted for in future studies, in that what is stressful or frustrating to one animal may be challenging and enriching to another. Determination of the positive and negative effects of enrichment can then be used to design devices and implementation methods that maximize the beneficial effects yet minimize the negative effects of enrichment on an animal-by-animal basis.