Groups, competition, aggressive communication

1
Groups, competition, aggressive communication

• MSc ACSB 2005/06
• Session 14 Understanding
• by modelling?

2
Benefits from group-living

• Food
• Food-finding, e.g. colony information exchange
• Lower variance of feeding success
• Co-operative hunting, - lions, hunting dogs
• Safety
• Increased vigilance - pigeon flocks
• Dilution of risk from predators - pelagic fish
• Co-operative defence- buffalo
• Breeding
• Co-operative territorial defence
• Helpers at nest/den - scrub jay, dwarf mongoose
• Eusocial societies - naked mole rat

3
Trade offs in group life

• Benefits must be traded off against costs
• cooperation increases size of prey that can be
  killed BUT low rank big group limits food
  gained
• larger groups reduce predation risk, BUT
  increase competition at food sources
• What currency to use for this trade-off?
• Mean intake per hour, or per day
• Variance in intake rates across time

4
Mean or Variance?

• Group can reduce variance in food intake
• High variance gt a greater risk of going bust,
  even if mean return is quite good
• Thompson, Vertinsky Krebs (1974) JAE 43,
  785-820 Realistically-simulated flocking in
  titmice - flocking had much greater impact on
  variance than on mean return (MODEL 1)
• Safety (low variance) may be more valuable than a
  high rate of intake

5
Badges of status

• Access to resources in a group often depends on
  status
• Larger, darker bib below bill on dominant Harris
  sparrows (studlies) - Rohwer
• Birds take note of the size of the bib in their
  interactions so this badge of status allows a
  dispute about rank to be resolved without
  fighting
• Why is the dark bib an honest signal of status?
• If wearing a large badge gives victory at minimum
  cost, why dont subordinates pretend to be
  dominant by growing larger badges?

6
Rohwers experiments

• Experiments in which badge size was increased or
  reduced experimentally
• Dominants with bleached-out badges had to fight
  hard for access to food etc - badge works
• DO win eventually - so really are better fighters
• Subordinates with enlarged badge dont win
• UNLESS also injected with testosterone at the
  time that the badge is enlarged

7
Møller House sparrows

• Møller (1987) Anim. Behav., 35, 1637-1644
• High-ranked birds had larger badges
• Attack most common between males with equal badge
  sizes (in field, not lab flocks)
• Attacks on birds with a similar-size badge more
  frequent among high-ranking birds so a
  subordinate with a deceptively large badge will
  face more attacks than if honest

8
Control of cheating

• So larger badge imposes a greater cost from other
  birds attacks limits cheating
• Møller showed elsewhere that a large badge also
  gt lower survival in autumn limits cheating
• In many contests, differences between the
  individuals (in size, weight, prowess RHP,
  value of the resource) are likely to be used to
  decide the outcome

9
Skew in animal groups

• Siblys model group is stable (assuming free
  entry/exit) at point where a subordinate fares as
  badly living in the group as it does when living
  independently
• Group size reflects entryexits, or fission
• Dominants get more of the benefits from the
  group, subordinates bear more of the costs. Skew
  within the group

10
Kinship affects skew in groups

• Vehrencamp, 1983, Anim. Behav. 31, 667-682 (MODEL
  2)
• In kin-groups, lowest-ranking members incur
  direct costs to benefit dominants but if
  dominants are related, subordinates get indirect
  benefits
• Subordinates inclusive fitness is increased if
  the breeding success of these dominants is
  enhanced
• The closer the relationship, the greater the skew
  that can be imposed on subordinates
• So a family group may be exploitative, not wholly
  co-operative, especially if very close kin are
  involved

11
Vehrencamps model
12
Helping vs. breeding

• Harassment in W. Fronted Bee-Eaters
• Nest in tunnels in sandy banks
• Shortage of food many young die
• Eggs hatch asynchronously always stronger and
  weaker chicks weakest starve quickly if food is
  scarce while rest remain well-fed
• Harassment halts many breeding attempts

13
Harassment (2)

• Who causes the trouble?
• Incidence Success
• Parents 36-45 75-84
• Brothers 16-19 33
• Other kin 16-17 0
• What is pay-off?
• Young per pair 0.51 (unassisted)
• 0.98 (helper)

14
Harassment (3)

• Harasser gains 0.47 young (at r0.5) 0.235
• loses 0.51 young at r0.25 0.1275
• Benefit to harasser (rounded) 0.11
• Victim Loses 0.51 at r0.5 0.2550
• gains 0.47 sibs at r0.5 0.2350
• Cost of harassment to victim (rounded) 0.02
• Victim (son) has less leverage than harasser
  (dad), and so is constrained to help, not breed

15
Threat can resolve contests

• Combat involves energy costs, and risk of injury
• Threat displays allow animals to resolve disputes
  without physical combat
• But if A gains victory more cheaply by giving an
  intimidating display, why does B not give an
  equally- or more-intimidating display and win
  instead?
• The problem of honest signalling in animal
  contests

16
Phased contests have escalating costs

• DeCarvalho et al, 2004, Anim. Behav., 68,
  473-482, measured energy of 3-phase Sierra Dome
  spider contests
• Phases have expenditure of 3.5x, 7.4x, 11.5x BMR
• Costs are substantial, and later phases are more
  costly, explaining why it pays a
  potentially-losing rival to quit early rather
  than to persist through phase 1 to phases 2 or 3.

17
Why do weaker rivals quit?

• Huntingford et al. (2000) data on shore crabs
  and cichlid fish
• Show glycogen reserves depleted, and increase in
  lactate in muscle, as contests progress so
  these contests impose costs
• Partial energetic Wars of Attrition - where
  opponents fight until one has reached the
  critical value of (lactate accumulationglycogen
  depletion) for giving up. Fight is equivalent
  to vigorous exercise (5 min of being chased)

18
Bowl doily spider contests

• Austad, 1983, Anim. Behav., 31, 59-73
• Weight/size affects probability of injury
• When males introduced together, weight -gt result
  (larger won 82 contests longer where weights
  nearly equal) (K)
• When weight equal, fights settled by female
  value. Owner fought for longer if has
  information that makes female more valuable (V)
• Most serious escalations when V/K equal for both
  contestants, e.g. smaller owner vs. larger
  intruder

19
Assessment-based models

• Many think that differences in size/weight/etc.
  are assessed during contest
• Larger/heavier rival wins in fish, spiders, etc.
• Later phases of contest likely to provide more
  information about weight-difference than earlier
  phases
• BUT see Taylor Elwood 2003, covered later in
  session
• Leimar, Austad Enquist, 1991, Evolution 45,
  862-874

20
Enquists honest-signals model MODEL 3

• Several alternative display actions can be ranked
  in terms of effectiveness and cost the most
  effective acts also the most costly
• When high-V contest, most costly act has the
  greater payoff (P3VH is high so gt C3)
• When low-V contest, least costly act gives
  greater payoff (P1VL gt C1 but P3VL lt C3)

21
Fulmar Contests

• Fulmars feeding on fish scraps behaviour
  includes some (but not many) ritualised displays
• Cost P(pecks, breast-to-breast, physical
  combat)
• Effectiveness P(opponent gives up)
• E.g. full wing display more costly and more
  effective than half/no wing display
• More effective/costly acts used for higher V
  resource
• Typically high positive rank correlations between
  cost and effectiveness of options used at each of
  successive decision points

22
But will information transfer evolve?

• Contests longest when opponents are equal gt pick
  up info about size difference from interaction
• Taylor Elwood (2003) show that same pattern can
  occur if each rival persists for a time related
  to its size, with no assessment need to tease
  apart effects of largers size, smallers size,
  and difference in size, to test assessment models
• Noble no information transfer evolved in two
  A-life models designed to allow info pickup
• http//www.comp.leeds.ac.uk/jasonn/Research/Aggres
  sion/index.html

23
Nobles simulations MODELs 4-5

• (1998) Tough guys dont dance
• Animats in contest change their position on the
  attack-flee continuum (? ) over time
• Being aggressive carries cost being attacked
  carries cost. Animats differ in fighting ability
  (f)
• Neural network implementation
• Animats know own fighting ability, their place on
  the attack-flee dimension ?, and what they are
  doing and can see what opponent is doing

24
Tough guys (2)

• Blind get zero info about opponents ?
• Exptal can evolve to make use of input that
  gives info about opponents ?
• Unfakable full info - given opponents fighting
  ability, f, so dont really need ?
• Did not evolve to use this info about ?, although
  over time fewer contests were settled by fights,
  and 41 of fights were between opponents with
  well-matched f-values

25
Nobles next attempt

• (2000) Talk is cheap
• a production system implementation of Enquists
  strong/weak signal honest-signalling model (1985)
  
• Strategies that evolved from a random starting
  point rarely employ information about opponents
  signal
• Enquists honest strategy was vulnerable
  displaced by mutants that didnt use information
  about signal
• If Noble increased the cost of just being
  attacked to unrealistic levels, information
  about opponents signal WAS now employed in some
  simulations

26
Understand behaviour through modelling?

• Models of various types
• Grounded simulations using time-steps
• Algebraic analysis of costs and payoffs
• Algebraic ESS models
• Evolutionary A-life simulations

27
References

• Pulliam Caraco (1984) Chapter 5 in Krebs
  Davies (Eds) Behavioural Ecology (2nd Ed only
  this edition). Blackwell, Oxford.
• Krebs Davies (1993) An introduction to
  behavioural ecology (3rd Edn) Chapter 6, 7
• Bradbury, JW and Vehrencamp, SL (1998) Principles
  of animal communication. Chapter 21 (and parts
  at least of Chapter 20, esp. pp. 649-658,
  662-665, 668-676)
• Enquist et al. (1985) Anim. Behav., 33, 1007-1020
• Nobles papers and additional info at
  http//www.comp.leeds.ac.uk/jasonn/Research/Aggres
  sion/index.html