cross species1

Cross-species Analyses of Social Hierarchies

Social hierarchies are found across animal taxa. We are interested in how universal social processes within social hierarchies are, and what types of variation exist in social hierarchies. Recent projects have included examining how different species establish social hierarchies. Working with Dr Ivan Chase, Stony Brook University, we have shown that groups of mice, cichlid fish and chickens show remarkable similarity in the structural evolution of networks of social hierarchies. Specifically, in each species dominant individuals tend to lay down their social dominance relationships and more subordinate individuals fill in their relationships around these. Further, we have demonstrated that social groups at the network level are extremely hierarchically organized and stable but at the triadic and dyadic levels there is much more flexibility in relationship structure.

cross species2

cross species1

It has been over 100 years since Thorleif Schjelderup-Ebbe published his seminal paper: “Beiträge zur Sozialpsychologie des Haushuhns“ (Zeitschrift für Psychologie 88, 1922, 225-252). This was the first study to describe in detail pecking-orders (‘Hackliste’) in chickens. Since then, social hierarchies have been studied across animal taxa. In a recent study, we have curated over 430 datasets from over 130 species including invertebrates, birds, reptiles, fish and mammals. These datasets include edgelists and sociomatrices of competitive dominance data. Our goal with this project is to provide a resource to other researchers who may be interested in cross-species analyses of social dominance. We have created an R package DomArchive that is a living repository of these data. The package also contains a vignette that describes how to use the raw social data and associated metadata.

winner loser effects

Winner-Loser Effects

winner loser effectsMathematical models have demonstrated that winner effects (the increased probability of winning your next fight given you won your previous fight) and loser effects (the increased probability of losing your next fight given you lost your previous fight) can lead to highly linear dominance hierarchies even when individuals do not vary in intrinsic fighting ability. Experimentally, studies in many species have demonstrated winner-loser effects when animals of similar size, who have an experimentally induced differential history of winning or losing, are paired together. Working with Dr Tim Fawcett and Dr Sam Ellis from the University of Exeter, UK, we are developing statistical approaches that can identify winner-loser effects in temporally ordered contest data from free-living and captive species. Further, we are interested in identifying how winner-loser effects vary within a social group according to individual traits.

social dynamics

Statistical Modeling of Social Dominance Interactions

We are interested in applying statistical methods to the study of aggressive social interactions. In one study, we developed methods for determining from all occurrence behavioral data when pairs of mice resolve their dominant-subordinate relationship. In the image you can see when individuals exhibit dominant (red) or subordinate (blue) behavior towards each other over sessions of 20 minutes across five days. We applied Kleinberg’s burst detection algorithm to identify individual bursts of aggression and/or subordination. Within each burst we determined if animals could be described as dominant or subordinate using phi-coefficients. This allowed us to temporally plot the emergence of dominant-subordinate relationships. We also examined Markov transitions and adapted pairwise-correlation methods to assess how dominant and subordinate animals dissociate as relationships resolve in terms of their sequential patterns of social behavior.

social dynamics

In collaboration with the research group of Dr Tian Zheng, Columbia University, we have also examined the temporal sequences of aggressive interactions in groups of mice. By applying network point process models with latent ranks we are able to model features such as winner effects, bursting and pair-flips. We can best model these data using Markov-Modulated Hawkes processes and suggest that these models are suitable for analyzing social interaction event dynamics of various types.

social development

Development of Social Competence

development social competence1In the wild over 90% of Mus musculus females will rear their offspring in communal nests. In the laboratory, the typical method of rearing is one dam with her litter. I have shown that rearing pups in large communal nests (three dams sharing litters) leads to profound changes in the maternal and social behavior of mouse offspring as well as the distribution of oxytocin and vasopressin receptors in several brain regions. Furthermore, these behavioral changes can be transmitted over generations through female offspring.

development social competence2In collaboration with Dr. Igor Branchi, we have shown that male mouse pups reared in communal nests have increased social competency in that they are quicker to recognize their social status when tested in pairs as adults. This effect is only found for pups from older and younger litters who also receive higher levels of maternal care. Both forms of social enrichment (maternal care and peer socialization) appear to be necessary. Increased social competency is associated with higher protein levels of BDNF in the hippocampus, frontal cortex and hypothalamus.

Video showing three mouse dams communally rearing their ten day old pups.

meaning of weaning1

The Meaning of Weaning

meaning of weaning1We have shown that the age at which animals are weaned has significant effects on their social development. Usually in the laboratory, mice are removed from their mothers at day 21 postnatally. We show that mothers will nurse and lick/groom their offspring beyond this period up to day 28 postnatally. During this fourth week postpartum dams will actively wean offspring by pinning and mounting them in response to pups’ nipple solicitations. The image shows a 4 week old pup sucking from its mother. We have also shown that female dams that provide less licking/grooming to their pups in the first week postnatally are more likely to actively wean their pups sooner. Pups that are weaned earlier exhibit increased levels of juvenile social play behavior likely as a compensatory mechanism for receiving less tactile stimulation early in life. Later actively weaned offspring are faster to exhibit other social behaviors. Further, early weaned female offspring are more likely to wean their own offspring earlier.

meaning of weaning2

Figure from Franks et al 2015 showing how low licking dams wean offspring earlier and those offspring engage earlier in social play.

Video: Mouse dam mounting and pinning 24-day-old pup in response to nipple solicitation.  At this time, dams start to actively wean pups by decreasing the proportion of times that they allow suckling. Pups transition from engaging in suckling to side-by-side contact with the dam.


Behavioral Plasticity

behavioral plasticity

Studying complex social behavior in the laboratory is challenging and requires analyses of dyadic interactions occurring over time in a physically and socially complex environment. In our laboratory we conduct long-term behavioral observations of animals housed in groups in large vivaria that mimic the burrow systems of their ancestral species Mus musculus (pictured).

We have studied groups of males and females. We have found that both male and female mice form highly linear dominance hierarchies where animals occupy unique social ranks.  This is illustrated in the pictured sociomatrices. These aggregate observed aggressive behaviors between individuals. Numbers represent the total number of agonistic behaviors that individuals in rows directed towards individuals in columns. This matrix can be converted into a dichotomized 1/0 matrix (on the right). In this example, a 1 represents that the individual in the row is a clear winner against the respective individual in the column. A 0 indicates a clear loser or tied relationship. From these sociomatrices many statistical analyses of social dominance can be applied. We have made many of these methods available in our compete R package

behavioral plasticity 2
behavioral plasticity 3

We have shown that mouse hierarchies emerge rapidly and are stable over several weeks. Within each group of 10-12 male mice, typically one clear alpha individual becomes dominant over all overs. There are an additional 2-4 sub-dominant individuals and the remaining individuals are subordinate. However, even subordinate animals determine their relative rank order.

Notably, we have shown that male mice are even able to form linear hierarchies when housed in groups of 30 individuals. In larger groups, we have also observed using social network analysis that animals form distinct sub-communities.

We have also demonstrated that mice are able to attend to their social context and adjust their behavior accordingly. In particular, using pairwise-correlation statistical methods, we show that males monitor the behavior of the alpha male in a hierarchy and avoid engaging in aggressive behavior whilst the alpha male is being actively aggressive. Only the alpha male mouse is monitored and avoided in this fashion. Indeed, the more despotic that each individual alpha male mouse is, the more strongly anti-correlated are the aggressive behaviors of other animals with his behavior.

Below are representative tick plots from six cohorts that show each aggressive behavior by each animal by rank. For readability the x-axis is the number of each aggressive act rather than time. As can be seen, there is inter-hierarchy variability in the degree of activity of animals – but across all groups, those who can be classified as ‘alpha’ or ‘sub-dominant’ engage in the highest levels of activity.

behavioral plasticity5

Social & Physiological Plasticity

Physiological Plasticity

Individuals must adapt to their current social environment. Dominant and subordinate animals each face unique challenges that require them to adjust their physiology as well as behavior. Dominant male mice increase not only their levels of aggression, but also their patrolling behavior and scent-marking. We have shown that dominant alpha males increase their urination frequency and importantly deposit large numbers of major urinary proteins (MUPs) into their urine. These proteins are non-volatile proteins that confer dominant social status and are used to mark territories by male mice. We also show that dominant males increase the expression of all Mup genes in the liver. These genes are metabolically expensive to produce and require dominant animals to increase their food and water intake to maintain their production and excretion. Further, dominant animals show constantly increased activity and reduced rest. Consistent with these changes, we have also found using Tag-Seq that the livers of dominant males show an up-regulation of genes involved in the production of energy such as fatty acid catabolism. Recent projects have also investigate the immunophenotypes of dominant versus subordinate animals.  We have found that dominant animals invest more in relatively cheap adaptive immunity (e.g. lymphocytes – B and T cells) and less in innate immunity. Conversely, subordinate animals invest more in immune cells and processes that promote inflammation and wound-healing. Future work in the lab is exploring the metabolic pathways of dominant and subordinate animals and how dominant animals plastically shift these to maintain their status.

physiological plasticity

Figure 1. Coordination of Physiological and Behavioral Plasticity in Response to Social Status in Mice.


Figure 2. From Lee et al. 2021 in review. Results of Weighted Gene Correlation Network Analysis (WGCNA) and Coexpression Differential Network Analysis (CoDiNA). (A,B) Heatmaps representing linear regression estimates of eigengene expression of WGCNA modules against other measured traits. Only significant associations (p-value <0.05) are represented.

Social Neuroscience

Processing of Social Status Signals

Many species use social cues or signals to guide the expression of contextually appropriate behavior, yet little is known about how the brain processes such information. We are currently investigating this question by exposing mice to social cues and analyzing neural excitation and the expression of other receptors and neurotransmitters in the brain using various histological techniques.

Chemical signals, such as pheromones and urinary proteins, are perhaps the most common mode of communication used by social species to convey identifying information including sex, species, status, and individual identity. The urine of mice contains various chemicals including Major Urinary Proteins (MUPs), some of which are crucial for territory marking and are vastly more prevalent in the urine of dominant mice.

In one study (Lee et al. 2021), we exposed mice of dominant or subordinate status to the urine of familiar (from their home-cage) or unfamiliar (from a different home-cage) mice of dominant or subordinate social status and subsequently examined the immunoreactivity of the Immediate Early Gene (IEG) cFos in their brain tissue. IEGs are expressed upon depolarization of a neuron’s cell membrane, and thus they serve as a proxy for neuronal excitation. Briefly, we found that several brain regions respond with more or less excitation depending on the subject animal’s own social status as well as their familiarity with the animal who provided the urine and that animal’s social status.


Figure 1. From Lee et al 2021. Examples of cfos immunoreactivity (cfos-IR) results. In the MEApv (posteroventral medial amygdala), mice of both dominant and subordinate social status exhibited greater excitation to dominant urine compared to subordinate urine and familiarity with the urine source had no effect. Alternatively, in the PMv (ventral premammillary nucleus), both dominant and subordinate mice responded to familiar alpha urine more than familiar and unfamiliar subordinate urine, but dominant mice exhibited significantly greater excitation to unfamiliar dominant urine.

We are further investigating how the brain processes social information using a technique called cellular Compartment Analysis of Temporal activity by Fluorescence In Situ Hybridization (catFISH) to trace back when neurons have been activated at multiple time points. By targeting the expression of different IEGs with non-overlapping expression timescales we can determine which neurons were activated by two temporally distinct social stimulus events.


Figure 2. An example image of in-situ hybridization tagging of IEGs Homer1a (green) and Arc (red) in a brain tissue slice. Neurons are labeled with DAPI (blue) Cells with punctate intra-nuclear green dots indicate it was excited ~30-25min ago. Cells with punctate intra-nuclear red dots indicate it was excited ~5-10min ago.


Social Status Transitions

Subdominant male mice are able to rapidly respond to the emergence of power vacuums. When an alpha male is removed from a hierarchy, subdominant males rapidly (within 3 minutes) recognize that there exists a social opportunity and they aggressively exert their own dominance over all other animals in the group (see Williamson et al 2017 ). These males socially ascend to become the new alpha males and are able to stay at the top of the hierarchy. This demonstrates great social competence on behalf of these males to be able to so quickly respond to a change in the social context of the group. We have also found that this change in social context leads to changes in neural gene expression and activity. When the alpha male is removed, we observe that subdominant and subordinate males show increased GnRH mRNA expression compared to when the alpha male has not been removed.

Notably, we see increased cFos immunoreactivity throughout several regions of the forebrain and midbrain of socially ascending sub-dominant individuals. In particular, ascenders have profoundly increased cFos in the frontal cortices (infralimbic, prelimbic, retrosplenial) as well the CA1 and dentate gyrus subregions of the hippocampus. Future work will identify the functional role of this increased immunoreactivity in enabling mice to make social decisions to ascend a hierarchy.


We are also actively exploring the global transcriptomic changes that occur as a function of social ascension and descension to support transitions of social status.