1 INTRODUCTION

The investigation of social hierarchies dates back to Thorlief Schjelderup-Ebbe's work in the 1920s characterizing social hierarchies in populations of chickens. Throughout this work, Schjelderup-Ebbe observed a strict, linear relationship of inter-group aggression, suggesting that each bird maintained a different social rank as a function of their rank within this aggressive order (the finding responsible for the colloquialism “pecking order” when referring to any social hierarchy).1 It has since been well demonstrated that these hierarchies are an integral part of virtually all group-living animal species, and that a subject's rank within a given hierarchy may be characterized by quantifying their access to various resources and/or exhibition of aggressive behaviors, with higher-ranking, dominant animals acquiring larger proportions of said resources2, 3 and/or exhibiting higher levels of aggression toward their lower-ranking counterparts4 and even colony intruders.5

Conceptually, the analyses of hierarchical relationships between rodents can be divided into two main categories: (1) Analysis of agonistic interactions and (2) analysis of differences in access to various types of resources, including territory, mates, standard chow, palatable food, and/or water. Agonistic behaviors can be directly measured in rodents, and highly aggressive rodents do have a tendency for increased access to resources.5-8 However, the unequal distribution of resources in humans is not exclusively regulated by aggression of higher-ranking individuals.9, 10 Similarly, increased access to resources in rodents does not always necessitate overt aggression on the part of the high-ranking subject, suggesting that the assessment of aggression alone is not sufficient to identify a rodent's social rank. It should be noted, though, that the measures obtained during the assessment of differences in access to resources are often indirect. For example, higher scent marking or increased ultrasonic vocalizations in dominant animals (see Sections 4 and 8i) could reflect increased access to territory and mates, but these measures are not necessarily directly proportional to such access. In addition, the behaviors assessed in these resource competition tasks could reflect a state of the animal, rather than a trait: for example, competition for food may not exclusively reflect social rank, as performance in this assay can also be affected by an individual's sensitivity to food deprivation and/or their innate motivation to consume a given resource.

In the sections below, we describe the most frequently used tests of social hierarchy and dominance in rodents, as well as the relationships between them in various rodent species/strains and across both sexes (for a summary of this information, refer to Figure 1). Following the description of these assays, we synthesize this information to provide strategies for future studies and highlight opportunities to develop an improved understanding of social rank among both male and female rodents.

Summary of species and sexes used in each measure of social dominance. (A) Agonistic behavior. (B) Resource competition

2 AGONISTIC BEHAVIORS

One of the most commonly used methods to assess social rank among rodents is the analysis of agonistic behaviors (for schematic of common agonistic behaviors, see Figure 2(A)). Agonistic encounters consist of both offensive and defensive behaviors exchanged between two animals.11, 12 Offensive behaviors typically include lateral attacks, chasing, biting and barbering, whereas defensive behaviors typically include flight, freezing, and exhibition of submissive (lying on back) or defensive (upright with paws raised) postures.12-16 Social rank is often determined following the observation of these behaviors during social interactions. However, experimenters have also used weight loss, the extent of barbering, and the number, severity, and location of scars/wounds following social interactions as indirect measures of dominance. Specifically, an animal is considered subordinate if it exhibits greater weight loss and/or has a higher number of severe wounds, while an animal is considered dominant if it experiences less weight loss and has a lower number of less severe wounds. Additionally, in rats, the wounds of subordinates are often located primarily on the tail, back, and flank, in comparison to dominants, for which wounds are primarily located on the head and snout.17-34 For barbering, also termed fur trimming, whisker trimming, the Dalila effect, or overgrooming, a mouse is considered dominant if all fur and whiskers remain intact, and subordinate if fur loss and/or whisker shortening/loss are observed.35-40 Notably, barbering is one of the only agonistic behaviors readily observed in female C57BL/6J mice.35, 36, 38, 41 On the other hand, barbering has also been shown to be unrelated to social dominance, rather, attributed to factors ranging from an unenriched environment to a manifestation of an obsessive–compulsive-like disorder or stereotypic behavior.42, 43 In fact, barbering is one agonistic behavior that is not thought to occur in a naturalistic setting, but is limited to laboratory mice.43, 44

Schematics of experimental setup for agonistic behavioral analyses. (A) Common examples of agonistic behaviors exhibited by dominant and subordinate subjects. Often conducted in a standard homecage or neutral arena. (B) Example of visible burrow system (VBS), a group-housed, mixed-sex design that promotes strict, despotic hierarchies among male subjects. In addition to the agonistic behaviors outlined in part (A), dominant and subordinate subjects in the VBS also spend differing amounts of time on the surface/in the tunnel system

To assess social dominance relations between singly housed animals, a pair of subjects, typically same-sex, is placed in a neutral cage or arena to allow for social interaction, during which agonistic behaviors are scored. This design is especially useful for more aggressive species or strains, in that the length of agonistic interactions can be controlled by the experimenter. This model has been used to characterize social rank in male mice,45, 46 male gerbils,47 male and female rats,13, 14, 48, 49 male and female mandarin voles,50 male root voles,51 and male and female hamsters.52-58 These studies have revealed that dominant-subordinate relationships are readily formed for all male mice, rats, hamsters, gerbils, and mandarin voles under these conditions, and that these relationships are typically stable over time. In contrast, female rats are rarely reported to form these relationships, due to less frequent and severe agonistic interactions.13, 14 On the other hand, female hamsters and mandarin voles are naturally more aggressive, which results in the observation of strict dominant-subordinate agonistic relationships during same-sex interactions.50, 54, 55 It must be noted that while these studies identify potential dominant-subordinate relationships between pairs of subjects, it cannot be conclusively stated that these relationships are reflective of the social hierarchies that emerge in socially-housed animals.

Thus, agonistic behaviors have also been analyzed in rodents that are housed in the same cage, but separated by barriers or partitions to prevent constant, direct physical contact. When these partitions are removed by the experimenter, subjects are able to interact directly, allowing for the observation of agonistic behaviors.32, 59-68 This assay has been used to identify dominance hierarchies in male Long Evans rats,32, 66 male BALB/cJ mice,59 male NMRI mice,63 male CD1 mice,64 male large vesper mice (Calomys callosus),65 male and female hamsters,60-62, 68 and male gerbils.67 As in single-housed designs, these experiments are particularly useful when investigating social rank in highly aggressive species and strains, that if housed in standard group-housing conditions, would present the risk of severe injury or death of subordinate subjects.

Several studies have also assessed agonistic behaviors and social rank in pair- or group-housed rodents allowing for continuous, direct social interaction. These designs allow for agonistic interactions to be analyzed under two potential conditions: (1) Among the group and colony members only, or (2) following the introduction of an intruder rodent. In the former example, animals exhibiting the highest number of offensive behaviors toward their own cagemates are considered dominant,69 whereas in the resident-intruder model, animals exhibiting the highest number of offensive behaviors toward the intruder are considered dominant.21, 70

Analysis of spontaneous agonistic interactions and evaluation of wounds or barbering among cagemates has been conducted in group-housed male mice,16, 27, 28, 30, 33, 38, 39, 46, 69, 71-90 female mice,16, 38, 39, 86, 91, 92 male rats,4, 14, 22-24, 31, 32, 34, 93-98 female rats,14 male bank voles,99 and female hamsters.100, 101 The majority of these studies observed strict linear dominance hierarchies in male mice, male rats, male voles, and female hamsters; whereas markedly less strict hierarchies—or no hierarchies at all—were observed among female mice and rats. In this context, linear refers to the structure in which there is an alpha (dominant over all cagemates), beta (dominant over all cagemates except alpha), and so forth, within the hierarchy.86 In contrast, studies using the resident-intruder model have demonstrated that despotic, or exclusive, dominance hierarchies are formed in male rats and male and female mice,21, 91, 102-104 where despotic refers to a single animal maintaining dominance over all other cagemates, with no differences in rank between these subordinates.86

Additional studies of social rank have been conducted using a mixed-sex design. These models are typically employed to potentiate agonistic interactions among male subjects and/or to provide more naturalistic housing conditions (refer to Figure 2(B) for a schematic of the visual burrow system, or VBS, a system often used in these types of experiments). As in same-sex studies, social rank can be determined by analyzing the spontaneous agonistic behaviors among the colony members6, 29, 105-108 or the agonistic behaviors of colony members toward a stranger, “intruder” rodent in a resident-intruder test.109, 110 The vast majority of mixed-sex colony studies analyzing spontaneous agonistic encounters within a colony have revealed that strict, despotic social hierarchies are readily established among male rats and mice, while no hierarchies are observed among female rats and mice.5-8, 15, 17, 25, 26, 86, 87, 111-129 In contrast, female hamsters housed in mixed-sex pairs establish strong dominant-subordinate relationships, with females typically maintaining dominance over the male.130 However, a resident-intruder study using mixed-sex colonies of Long Evans rats has demonstrated that one male and one female in each colony exhibit social dominance evidenced by increased aggression toward an intruder rat introduced to the colony.109 Importantly, though, a female was only identifiable as dominant when all male colony members were removed from the cage during the resident-intruder test.109 This finding demonstrates that the lack of observed social ranks among females in previous VBS studies is likely attributable to the testing conditions used: It seems that for this assay, all males must be removed from the environment during testing for a dominant female to emerge. These data suggest that social hierarchies can, in fact, form among female rats, but that the tests traditionally used to assess dominance in males may prove insufficient in detecting these dynamics in females.

Overall, these studies demonstrate that agonistic behaviors serve as a useful measure of social dominance in males of many rodent species, as well as in female hamsters and mandarin voles, and even female rats under certain testing conditions. As such, this method has proven generally less reliable in female mice, rats (when tested in the presence of males), and gerbils, suggesting that different measures should be utilized when assessing social rank in these populations. Lastly, it must be noted that severe injury or death of subordinate subjects is a considerable risk in studies using more aggressive strains and species. Therefore, in these subjects, single-housing, or modified group-housing settings where subjects are separated by partitions need to be used, in that they allow for constant supervision of agonistic encounters.

3 TUBE TEST

The tube test was first developed by Lindzey and colleagues in 1961 to characterize social dominance in male and female mice,131 and has since been employed to assess social rank in different mouse strains and other rodent species (see Figure 3(A) for tube test schematic).

Schematics of experiment setup for resource competition assays. (A) Tube test. (B) Scent-marking. (C) Standard chow/water competition. (D) Palatable food competition. (E) 70-mHz ultrasonic vocalizations. (F) Warm spot test. (G) Shock avoidance

For this assay, experimenters use a clear, plastic tube, typically 30 cm in length for smaller rodents. The appropriate diameter is determined so that there is only room for one subject to pass through the tube at a given time.131, 132 Before testing, rodents are first habituated to the apparatus and trained to cross through the tube individually. Subjects are often presented with a food reward during this training process to promote crossing through the full length of the tube.131, 133-135 Animals are then paired to undergo testing, during which each subject is placed on opposite ends of the tube and allowed to approach its conspecific toward the center of the tube. The subject that subsequently forces its competitor out of the tube is considered the winner (dominant), and the subject forced out of the tube the loser (subordinate). If examining social rank among a group of 3 or more subjects, multiple tube tests are conducted using a round-robin design to ensure that tests take place between all members of said group.132

The tube test has been used to evaluate social rank in male and female mice,36, 37, 39-41, 45, 70, 131, 132, 136-149 male and female prairie voles,150 male and female rats,133, 134, 151, 152 male hamsters,135 and male gerbils.135 In these models, a point system is often used to express the social rank of a subject relative to other group members. For example, the winner of each test can be assigned 1 point, and the loser 0 points. Following the completion of all tests between group members, each subjects' points are added together to determine its social rank within the group. Therefore, in a group of four subjects, the most dominant animal would obtain 3 total points, and the most subordinate 0 points following tube tests with all other group members.37, 131, 132 Alternatively, David's Score (DS) can be used to calculate social rank following tube tests. DS is a slightly more complex measure that calculates dominance score based on a subject's proportion of wins to losses following repeated social interactions (for detailed information on DS calculation, refer to Reference 153).

Importantly, social ranks determined using this measure have been shown to positively correlate with ranks determined from other dominance measures, such as the warm spot test, urine-marking, and ultrasonic vocalization measures.37, 132, 138, 142, 143 The relationship between tube test ranks and food/water competition ranks remains less clear, however, and tends to vary between species and the type of reward used. For example, a negative correlation between tube test and food competition ranks has been reported in male DBA and albino mice when the food reward is standard rodent chow.45 In contrast, in studies using male Lister rats151 or male ICR mice,145 ranks obtained following palatable food/liquid competition were positively correlated with tube test ranks.145, 151 These inconsistencies are also observed in water competition tasks, in which no relationship between water competition rank and tube test rank is seen in male gerbils, though this relationship is positively correlated in male hamsters.135

Contradictory results have also been obtained for the relationship between tube test rank and ranks determined from displays of agonistic behaviors, with certain studies revealing a positive correlation,37, 40, 142 and others finding a negative correlation or no relationship between these measures.41, 45 Of note, the studies that observed positive correlations between agonistic behavior rank and tube test rank were conducted using male C57BL/6 mice. In contrast, the studies that observed a negative correlation45 or no relationship between these measures41 were conducted using male DBA and albino mice45 or male and female C57BL/6 mice,41 respectively. Taken together, these findings suggest a potential role of sex and/or strain in the correlation between social ranks determined from the tube test and those determined from agonistic behavior measures.

Overall, the tube test serves as a useful model for assessing social dominance in males and females and across various rodent species, proving especially useful if the subject population is less prone to agonistic behaviors. Even among rodents that readily exhibit aggression, the tube test presents certain advantages, in that animals are not susceptible to injury as they would be if using assays such as agonistic behavioral assessment. It must be considered, though, that species, strain, and sex can determine the generalizability of tube test results to other dominance measures.

4 SCENT-MARKING

Patterns in scent-marking, or urine- and flank-marking, in rodents have also been used as measures of social dominance (for schematic of scent-marking assay, refer to Figure 3(B)). Urine-marking is most often used to characterize the social rank of male mice,28, 37, 84, 85, 138, 142, 154-156 but it has also been used to assess dominance in in female mice,157 male bank voles,99, 158, 159 and male root voles.51

In this assay, two subjects are placed in a neutral, clean cage separated into two compartments. The rodents remain in their respective compartments for a test period allowing for urine collection, ranging from 2 to 22 h.28, 37, 84, 85, 142, 158 However, urine-marking can also be assessed in a single subject, placed in a test cage alone for 3 min to 2 h.51, 154-157 This design also allows for the assessment of counter-marking, or scent-marking over an experimenter-presented urine sample from a conspecific, which can denote social rank as well.155, 157

Regardless of the setup used for testing, cages are lined with filter paper so that urine markings can be subsequently analyzed by visualization with UV light. The number of urine marks reflects social rank, with a higher number of marks indicating dominance, and a lower number of marks indicating subordinance.28, 85, 99, 156 Counter-marking studies have also revealed that subjects identified as dominant based on agonistic behaviors will urine-mark over virtually any urine sample presented, regardless of the rank of the animal from which the sample was collected, while subordinate subjects will not.155, 157 Additional studies have shown that dominants, again the rank of which is determined based on displays of agonistic behavior, typically scent-mark throughout the entirety of the test arena, with concentrations close to the partition, while the urine marks of subordinates are often confined to the corners of the arena.28, 158, 159 Interestingly, in the one study using females, urine-marking patterns of dominants were similar to those observed in dominant males, in that female dominants, identified based on agonistic behavior analysis, exhibited a higher number of urine-marks than subordinates, while also exhibiting counter-marking behaviors.155, 157 Each of the studies mentioned above has demonstrated that social rank determined by urine-marking positively correlates with that determined by agonistic behavior analysis or the tube test, with the exception of the study using male root voles. In this experiment, no relationship was observed between social rank determined by agonistic behavior analysis and social rank determined by urine-marking,51 a finding suggesting that urine-marking assays may not be a reliable measure of social dominance in all species. Notably, of the one study conducted using female mice, urine-marking was compared between breeding and nonbreeding females,157 demonstrating that additional urine-marking studies in females of the same breeding status should be conducted to further investigate the effect of social rank on this behavior in females.

In contrast to urine-marking, flank-marking has been used to characterize social rank in male hamsters,53, 60, 62, 160-162 female hamsters,52, 61 and male gerbils.121 Flank-marking involves the rubbing of an animal's dorsolateral flank glands on objects and/or areas within their environment.161 As such, this behavior is typically scored as the total number of flank-marks exhibited by each subject during social interaction,61, 62 when placed in an open field,121 or when placed in the empty, dirty homecage of a conspecific.52, 161 It has been consistently demonstrated that male and female hamsters and male gerbils that exhibit the most flank-marks are also dominant in agonistic encounters. Notably, while female gerbils also possess flank glands, they appear to use scent-marking behavior not to communicate dominance, but rather as a means to defend the nest during gestation and lactation,163 a finding suggesting that while flank-marking is a useful measure for social dominance in male and female hamsters and male gerbils, this assay would not be useful in assessing dominance for female gerbils.

Collectively, these data suggest that while scent-marking assays can prove useful in characterizing social rank, their validity is dependent upon the sex and species—and potentially breeder status—of subjects used.

5 STANDARD FOOD AND WATER ACCESS PRIORITY/COMPETITION

Resource competition assays, or food and water access priority/competition, are additional means of characterizing social dominance in rodents. These tasks almost always involve food or water deprivation, most often for 22–23 h periods,135, 164-168 or food restriction, during which subjects are maintained at approximately 80%–85% free-feeding body weight.91, 169, 170 Animals are then given access to standard rodent chow or water for a short period of time, typically 5 min,45, 164-166, 171, 172 to induce resource competition in a water- or food-deprived state. In fact, only one study of food/water competition has been conducted in non-deprived conditions.100 Overall, these assays have been conducted in the homecage of group-housed subjects,96, 135, 173, 174 or in a neutral test cage, allowing for testing between novel conspecifics45, 165-