Seed predation behaviors

Seed predation occurs after seed dispersal has occurred and serves to reduce the number of seeds by simulating seed consumption by predators.

In this document:
Seed predation parameters
Functional response seed predation behavior
Neighborhood seed predation behavior
Functional response seed predation behavior (linked)
Neighborhood seed predation behavior (linked)

Seed predation parameters

Functional response seed predation

This behavior simulates functional response seed predation, where the number of predators is a function of the amount of food that has been consumed. Since seed predator life cycles are often very short, this behavior runs as a mini-model within the context of the larger simulation. It simulates weekly timesteps of seed fall, seed consumption, and predator response for as much of the year as the user desires. The mini-model simulates a single year, which starts at the beginning of seed fall. It is possible to define two seasons during the year with different parameters for seed consumption. All species are assumed to drop seeds at the same time, at a constant rate. You set the number of weeks that seedfall should occur. Once seedfall is over, there are no new additions to the predator food pool. There is then an optional period of germination in which the food pool of seeds is further reduced by some proportion of those remaining seeds germinating into seedlings (seedlings cannot be eaten). Once the model has finished running, the leftover seeds that were not consumed by predators, or those seeds that escaped through germination, are available as input to the Establishment behaviors.

How it works

The behavior's mini-model begins with the number of seeds of each species to which it is applied that are available in each cell of the Dispersed Seeds grid. Each cell gets its own run of the mini-model. The pool of seeds in one cell for all species to which this behavior applies is treated as a single food pool for one year's time (even if the model timestep length is greater than one year). For all the species to which this behavior is applied, the seeds are assumed to be equally available; however, each species can have its own parameters for actual consumption rate. The initial number of predators is calculated from the Predator initial density - num/sq m parameter or, if the mini-model has run before and the Preserve predator densities between SORTIE timesteps parameter is set to true, from the final density of the last mini-model run.

The behavior mini-model run begins at the part of the year in which seedfall occurs. The number of seeds in the seed rain is the total seed pool, which is evenly divided over the user-defined seed rain length. The predator population has as a food source the number of seeds added during the current week's rain (if the rain is going on) plus any leftover seeds from previous weeks which have not been consumed. Beginning at a certain week in the spring, the number of seeds available to the mice is further reduced by a certain percentage each week to simulate germination. Once germination begins, it continues until the predator model finishes running. In order to correctly calculate mouse consumption and ensure that the seeds which germinate are actually available later, the behavior keeps track of the seeds actually consumed; it is this number which is subtracted from total seeds at the end.

Seed offtake for each week is calculated as
O = Σ ( IRs * N )

where Per capita seed offtake for a single seed species is
IRs = cs(1 - e-(S*D)) * ps

where

Predator response to food availability can be different for two seasons in the year. The dividing line between the seasons is given with the Func Resp - Week Season 2 Begins parameter. If you want a uniform response with no seasonal differences, you can set this value to 1 or 52, or set the predator response parameters to be the same for both seasons.

The number of predators in each cell's population is calculated as


Nt = Nt-1 * ert-1

where The instantaneous rate of change per week, r, is calculated as
r = (a + d * Σ ( IRs ) + g*N) / 12

where

How to apply it

This behavior may be applied to seeds of any species. Any species to which it is applied must also have a Disperse behavior applied as well.

Functional response seed predation (linked)

How it works

This behavior is used together with the Neighborhood seed predation behavior (linked) behavior. The two behaviors work together to model seed predation.

This behavior performs the functional response model in almost exactly the same way as the Functional response seed predation behavior. However, instead of removing the eaten seeds, it calculates a whole-plot offtake rate for the group of species to which it has been applied. This rate is always for the plot as a whole, not for individual cells of the seed grid. This offtake rate is then available for use by the Neighborhood seed predation behavior (linked) behavior, and no further action is taken.

This behavior can be used in the same run with the non-linked version of this behavior. The two sets of species are kept completely separate, and there are two separate sets of parameters.

How to apply it

This behavior may be applied to seeds of any species. Any species to which it is applied must also have a Disperse behavior applied as well. Presumably, the Neighborhood seed predation behavior (linked) behavior will be applied to the same set of species, and should be placed after this behavior in the ordered list of model behaviors.

Neighborhood seed predation

This simulates seed predation as a function of tree neighborhood and masting events. The same equations are used to calculate the amount of seed eaten, but there are different parameters for masting and non-masting timesteps. Masting events can be determined in one of two ways: by seed levels rising above a threshold that you set, or by masting having occurred as defined by one of the masting disperse behaviors.

How it works

Masting timesteps are those with a heavy density of edible seeds. The first step in a given time step is deciding whether or not masting is occurring. There are two methods available for making the masting decision. You set your chosen method using the Neighborhood Predation - Mast Event Decision Method parameter.

If the event decision method is set to "Seed threshold", masting events occur whenever seed density rises above a certain threshold. You set the threshold for this density in the Neighborhood Predation - Masting Seed Density, #/m2/yr parameter. You then set which species are included in the mast count with the Neighborhood Predation - Counts For Masting? parameter. (Species which do not count towards masting may still be predated.) The seed density is set as an annual average so the density will be calculated the same way for different-length timesteps. If the total plotwide average annual seed density is greater than this value, the timestep is treated as a masting timestep. Only those seeds of species to which this behavior applies are considered in the density.

If the event decision method is set to "Ask disperse", masting events occur whenever any of the species to which this behavior applies masts according to either the Masting non-spatial disperse behavior or the Masting spatial disperse behavior. You can use either of these behaviors, both of them, or neither (in which case masting never occurs). If any species to which this behavior is applied has masted with either behavior, then this is a masting time step.

If there are no adult trees in the plot, then masting does not occur according to either method.

The seeds in each grid cell of the Dispersed Seeds grid are treated separately for predation, according to their local neighborhood composition. The relative basal area of each species is calculated, from the total basal area of individuals within Neighborhood Predation - Neighbor Search Radius (m) meters of the grid cell center that have a DBH greater than Neighborhood Predation - Minimum Neighbor DBH (cm). The amount of seed eaten for each species is calculated as:

where Y is the proportion of that species' seed that is eaten, and p0 is either the Neighborhood Predation - Masting "p0" or the Neighborhood Predation - Non-Masting "p0" parameter for that species. X is calculated as follows:

where pn is either the Neighborhood Predation - Species i Masting "pn" or the Neighborhood Predation - Species i Non-Masting "pn" of species n, and RBAn is the relative basal are of species n.

The eaten seeds are removed from the Dispersed Seeds grid.

In order to make results more verifiable, Neighborhood Seed Predation produces a grid called Neighborhood Seed Predation. This grid stores the pre predation seed rain and amount of seeds eaten for each cell in the Dispersed Seeds grid. This grid has no effect on calculations but can be saved in the output file for review.

How to apply it

This behavior may be applied to seeds of any species. Any species to which it is applied must also have a Disperse behavior applied as well.

Neighborhood seed predation (linked)

How it works

This behavior is used together with the Functional response seed predation behavior (linked) behavior. The two behaviors work together to model seed predation.

The actual amount of seed eaten is calculated by the Functional response seed predation behavior (linked) behavior. This behavior then distributes the offtake according to neighborhood composition.

The behavior begins by calculating the Y values for each grid cell as in the non-linked version (masting is ignored - there is only one set of parameters). Then the Y values are adjusted so that their mean is equal to the amount of whole-plot offtake Z as calculated by the Functional response seed predation behavior (linked). The adjustments are made as follows:

  1. Calculate logit(Z)
  2. Calculate the logit(Ys) and subtract the minimum value from each (as well as the logit(Z)) so they will all be positive
  3. Average the logit(Y)s
  4. Divide logit(Z) by average logit(Y) to get a correction factor
  5. Multiply each logit(Y) by the correction factor
  6. Add back the same minimum value formerly subtracted
  7. Back transform logit(Y) to Y and use when removing seeds

Then the seeds are removed according to the adjusted Y values.

This behavior can be used in the same run with the non-linked version of this behavior. The two sets of species are kept completely separate, and there are two separate sets of parameters.

This behavior may be applied to seeds of any species. Any species to which it is applied must also have a Disperse behavior applied, as well as the Functional response seed predation behavior (linked). This behavior should be placed after that behavior in the ordered list of model behaviors.

In order to make results more verifiable, Neighborhood Seed Predation produces a grid called Neighborhood Seed Predation. This grid stores the pre predation seed rain and amount of seeds eaten for each cell in the Dispersed Seeds grid. This grid has no effect on calculations but can be saved in the output file for review.


17-Oct-2008 12:50 PM