Title: BSC 417517 Environmental Modeling
1BSC 417/517 Environmental Modeling
- Predator-Prey Oscillations on the Kaibab Plateau
2The Predator-Prey Relationship
- Predator-prey relationships have always occupied
a special place in ecology - Ideal topic for systems dynamics
- Examine interaction between deer and predators on
Kaibab Plateau - Learn about possible behavior of predator and
prey populations if predators had not been
removed in the early 1900s
3Deer and Predators on Kaibab Plateau
- Information on deer population irruption is not
reliable - Data on predators is even more sketchy
- Gain insight into predator prey relationship on
the Plateau from a more well-documented system
the snowshoe hare-lynx system in Canada - Time series available on number of lynx pelts
purchased by the Hudson Bay Co.
4Snowshoe Hare-Lynx System
7
6
5
4
3
2
Hares
Lynx
1
5Snowshoe Hare-Lynx System
- Records show peak in number of lynx pelts every
9-10 years - Data suggest that populations have oscillated in
a cyclical manner for over 100 years - Data are viewed as a classical example of
predator-prey interaction - Oscillations are not related to seasonal or other
obvious annual changes - Best examples of predator-prey oscillations in
mammal populations show periodicity of 3-4 or
9-10 years
6Reference Mode for Kaibab Deer-Predator System
- Use hare-lynx example to draw a reference mode
for deer-predator relationship - Should the oscillations be sustained, damped, or
growing? - Intuition says sustained, but many other types of
behavior have been observed - For sake of simplicity, go with sustained
oscillation with 9-10 year periodicity as
reference mode - Peaks in predator (cougar) populations should lag
behind peaks in deer population by a few years
7Initial Model Equilibrium Conditions
2000
50
4000
2000
0.0
0.5
800
0.0
40
5
1.0
8Model Structure
- Ignore biomass impact of deer growth
- Assume ample forage is present by setting
fraction forage needs met equal to 1.0 - Predator stock is dependent on deer density
vis-à-vis deer density-dependent kill rate and
kill-rate dependent net birth rate
9Predator Kill Rate Functional Response
- Number of deer killed per predator per year is 60
if there are more than 10 deer/1000 acres 1
kill/week satiation limit - Shape of graphical function reflects a
combination of Type I and Type II functional
response
Kill rate
Kill rate
Type II
Type I
Prey density
Prey density
10Predator Kill Rate Graphical Function
11Predator Birth Rate Response
- Net birth rate is dependent on kill rate higher
kill rate gt higher net birth rate - Maximum net birth rate 0.45/yr
- Cougars start to breed young (2-3 years age)
- Breed every 2 years with an average of 3 kittens
- Maximum net birth rate for predators and prey are
comparable and relatively highimplications for
potential oscillation?
12Predator Birth Rate Graphical Function
13Initial Model Results Verify Equilibrium
Conditions
Initial predator density 50
14Initial Model Results - Nonequilibrium Initial
Prey Density
- Set initial predator density at 45
- System displays unstable behavior (as illustrated
by 30 vs. 50 year simulation) - Predators virtually annihilate prey after ca. 25
year, which lead to ensuing unstable behavior - Question why doesnt such unstable behavior
typically occur in nature?
15Initial Model Results - Nonequilibrium Initial
Prey Density
16Natural Predator-Prey Systems
- Predators dont normally hunt prey to zero
- Rather, select individuals from prey population
that are easiest to catch (young, old, weak) - Minimum threshold concept prey density limit
below which predators would no longer find it
profitable to hunt the prey and would switch to
different prey - Threshold is determined by availability of prey
hiding places (refuge) and prey social behavior
17Revising The Model
- Should we revise the model to take into account
the threshold concept, effect of prey refuge, and
prey social behavior? - Perhaps expand deer population to multiple stocks
to simulate deer age structure, and then allow
predators to concentrate on young and old deer - Sounds good, butcomplexity would increase
dramatically in face of limited data - Better to consider if combined effect of these
factors could be taken into account within
existing, simple model structure
18Revised Model
- Try using a different functional response for
density-dependent kill rate which incorporates
the concept of threshold prey density - No kills if deer density falls below 2 deer per
1000 acres, e.g. because of the ability of deer
to find safe refuge when overall density is low - S-shaped function response corresponds to Type
III functional response
19Type III Functional Response
20Revised Model Results
21Revised Model Results
- Initial predator population is set at 100
- Large predator population causes an initial
decline in deer population, but predator
population declines quickly - Damped oscillatory behavior ensues with
periodicity of ca. 10 years - Result essentially corresponds to the original
reference mode
22Further Interpretation
- The initial dynamic hypothesis was that the
cougar and deer populations could interact to
produce stable cycles with a period similar to
the classic 9-10 year cycle observed in other
mammalian predator-prey systems - Requirement for a Type III functional response to
produce stable behavior can be interpreted as an
indication of the importance of prey refuge or
threshold levels
23State Space (Phase Plane) Diagram
Point attractor
24Patterns of Oscillation
- Previous simulations show possibility for both
damped and growing oscillations, depending on the
nature of the predator functional response - What about potential for sustained oscillation,
as state in the reference mode? - Could random disturbances lead to persistent
cycles?
25Influence of Random Variation
- Introduce randomness into the deer net birth rate
via the following equations - net birth rate 0.5 random factor
- random factor random(-0.2,0.2,123)
- The random factor allows net birth rate to vary
randomly from a low of 0.3 to a high of 0.7 - The value 123 is a seed for the random number
generator
26Influence of Random Variation
- System shows sustained oscillation over long time
scales, with periodicity of ca. 10 years - Reference mode has been generated
27Policy Test Selective Removal of Predators
- Results of revised model with random variation in
deer birth rate suggests that stable
predator-prey interactions would have been
possible if the predators had not been removed
from the Kaibab Plateau - Although predator population averages 50,
substantially higher numbers occur in some years,
which could pose problem for ranchers livestock - Test influence of allowing hunters to kill some
predators to protect live stock
28Model With Selective Removal of Predators
29New Equations
predator_kills IF(TIMEgtstart_year) THEN
(predator_population-maximum_acceptable_predators
) ELSE 0 start_year 1920 maximum_acceptable_pr
edators 55
30Simulation Results With Selective Removal of
Predators
31Interpretation
- Results suggest that it might have been possible
to reduce peak values of predator population
without destroying the stability of the
predator-prey system - However, managers in early 1900s had essentially
no knowledge of predator-prey dynamics - Even today, other factors besides predator-prey
population dynamics are know to be important in
governing response of the system
32Current Interpretation of the Hare-Lynx Predator
Prey System
- Krebs et al. (Bioscience 2001) (see PDF on
web-site) conclude that Lotka and Volterra were
only partly correct when the concluded that the
snowshoe hare cycle was the product of a
predator-prey oscillation - Missed critical point that the cycle can only be
understood by considering three trophic levels
rather than just two - Hare cycle is produced by interaction between
predation and food supplies - Dependence on food supply ripples across many
species of predators and prey in boreal forest