Lecture 3: Animal Social Networks Flashcards
What is an animal social network
The pattern of social relationships in an animal population
Consists of nodes and links
- We usually study larger group, because we are interested in general patterns rather than specific individuals
Nodes (also called vertices)
Each node represents an individual
Links (also called edges)
Each link represents the social relationship between two individuals
Why are animal social networks studied so much?
Animals have complex (non-random) social structure
- Not only ‘cognitively advanced’ species, but all taxa (insects, birds, reptiles, fish, mammals..)
The structure have important consequences for ecological and evolutionary processes
- For example, they affect the spread of disease and information
The network approach gives us a way to look at social structure in detail
- We can measure different structural aspects, and we can test hypotheses about the structures statistically
Two common types of research questions about animal social network structures
1) How do things influence social structure?
- Traits of individuals (age, sex, body size, relatedness, etc)
- Ecological factors (food, season, climate, etc)
- Human activities (hunting, habitat loss, etc)
2) How social structures influence things?
- Transmission processes (the spread of information and diseases)
- Evolutionary processes
Two main approaches for studying animal social networks
1) Analysis of real network data
- We collect data on a social network and analyze it statistically
2) Modelling of simulated networks
- We create artificial networks with computer algorithms and study their properties
Common steps in studies of (real) animal social networks
- First we collect data on social relationship
- Then we construct the network
- Then we measure structural aspects of the network
- Then we test hypotheses about the network with statistical tests
- We can create a network matrix with network data and then plot the data and create a network graph
- Collecting data on social relationships (network links)
We need to be able to recognise each individual
- natural markings
- tags
- electronic devices
We collect data over a period of time (the ‘observation period’)
We note the number of times that each pair of individuals interacted or associated (were near each other)
The data collection method depends on what works with our species
Two common data collection methods for animal social networks
- Focal follows and direct interactions:
- The observer follows each individual in turn, and notes down all its interactions with others
- Useful for small populations with frequent interactions - Daily surveys and ‘gambit of the group’:
- The observer surveys the whole population each day, and notes down who is grouping together
- Useful for populations where individuals often change groups
- Constructing the network
To construct the network (the matrix) we need to calculate the strength of all the links:
* Pairs that interacted or associated more should have a stronger link (social relationship)
* We usually use the rate of interaction of association
* This is calculated in different ways depending on the method of data collection
We can then plot network graph with special software (for example in R)
- Measure structural aspects of the network
We can measure network structure with measures called ‘network metrics’
Two main types:
* Local (node-based) network metrics
* Global network metrics
Local (node-based) network metrics
- Measures of the social connectedness of an individual
- One measure for each individual (each network node)
Examples of local network metrics:
- Degree= number of links an individual has
- Betweenness = The extent to which the individual connects different network communities
- Eigenvector centrality = The extent to which the individual and its neighbors are strongly connected
Global network metrics
- Measures of the overall structure of the network
- One measure for the whole network
Examples of global network metrics:
- Network density = number of links that are present out of all possible links
- Assortativity = The extent to which individuals are primarily connected to others of the same class (ex: sex or age)
- Testing hypotheses about the network structure
Two common types of statistical analysis of animal social networks:
- Node-based analysis
- We test for a correlation between a node-based network metric and individual traits (age, sex, etc.).
- This tells us whether the social position of individuals depends on their traits. - Link-based analysis
- We test for a correlation between link strength and traits of pairs (relatedness, age difference, etc.)
- This tells us whether the strength of social connections depends on such things
We often use special types of statistical tests because network data are different
Modelling social networks - Common steps
- Create or select a model
- Create networks
- use the model to make a set of networks
- also make a set of networks with random structure (for comparison) - plot results
- plot results for the network sets and compare
Case study: Social connectedness and information access in tits
Question: Does information access depend on network position?
Methods:
- Quantified multi-species social networks in two cases
- Measured individual connectedness (eigenvector and betweenness)
- Measured whether each individual discovered new food patches
Result:
- Socially central birds were more likely to discover food patches
–> Information flows along the links of the social network
Case study: social environment and sexual selection in finches
Question: Do males select social environments that make them look
more attractive and increases fitness?
Methods:
- Quantified the social networks and measured betweenness of males
- Quantified the attractiveness of males (plumage colour)
Result:
- For the less colourful males, higher betweenness lead to higher mating success
→ Less attractive males increase their fitness by seeking environments where they
look more attractive
→ The social environment affects the selection pressure on male colourisation
Case study: networks and conservation in Tasmanian devils
Question: Can the Tasmanian devil be helped by removing ‘super-spreaders’?
Method:
- Quantified the social network and measured social connectedness of individuals (degree, betweenness and information centrality)
- Tested for differences in social connectedness between age and sex classes
Result:
- ” The scope for directing management at highly connected individuals is [] limited as no particular sex or age class was obviously more highly connected than others”
Why are animal social networks important?
They affect processes such as the spread of disease and information
They affect the fitness of individuals and thereby affect evolution
They can help in conserving species and improving wellbeing
They are natural complex systems
