driven_vital_rate {mpmsim}

Calculate driven vital rates

Description

This function calculates new values for a vital rate, such as survival or fecundity that is being influenced by a driver (e.g., weather). It does this by using a driver variable and a baseline value, along with a specified slope for the relationship between the driver variable and the vital rate. The function works on a linearised scale, using logit for survival and log for fecundity, and takes into account the error standard deviation.

Usage

driven_vital_rate(
  driver,
  baseline_value = NULL,
  slope = NULL,
  baseline_driver = NULL,
  error_sd = 0,
  link = "logit"
)

Arguments

driver

A vector of driver values.

baseline_value

A vector or matrix of baseline values for the vital rate (e.g., survival) that is being influenced ("driven") by another variable (e.g. a climatic variable).

slope

A vector or matrix of slopes for the relationship between the driver variable and the vital rate being driven.

baseline_driver

The baseline_driver parameter is a single value representing the baseline driver value. If the driver value is greater than this value and the slope is positive, then the resulting vital rate will be higher. Conversely, if the driver value is less than this variable and the slope is positive, then the resulting vital rate will be less than the baseline value.

error_sd

A vector or matrix of error standard deviations for random normal error to be added to the driven value of the vital rate being modelled. If set to 0 (the default), no error is added.

link

A character string indicating the type of link function to use. Valid values are "logit" (the default) and "log", which are appropriate for survival (U submatrix) and reproduction (F submatrix) respectively.

Details

The relationship between the driver variable and the vital rate is assumed to be linear:

$$V = a * (d - d_b) + x + E$$

Where $$V$$ is the new vital rate (on the scale of the linear predictor), $$a$$ is the slope, $$x$$ is the baseline vital rate, $$d$$ is the driver, $$d_b$$ is the baseline driver and $$E$$ is the error.

The input vital rate(s) (baseline_value) can be a single-element vector representing a single vital rate (e.g., survival probability or fecundity), a longer vector representing a series of vital rates (e.g., several survival probabilities or fecundity values), or a matrix of values (e.g., a U or F submatrix of a matrix population model). The slopes of the relationship between the vital rate (baseline_value) and the driver can be provided as a single value, which is applied to all elements of the input vital rates, or as a matrix of values that map onto the matrix of vital rates. This allows users to simulate cases where different vital rates in a matrix model are affected in different ways by the same weather driver. For example, juvenile survival might be more affected by the driver than adult survival. The baseline_driver value represents the "normal" state of the driver. If the driver is greater than the baseline_driver and the slope is positive, then the outcome vital rate will be higher. If the driver is less than the baseline_driver variable and the slope is positive, then the outcome vital rate will be less than the baseline_value. The error_sd represents the error in the linear relationship between the driver and the vital rate.

Value

Depending on the input types, either a single value, a vector or a list of matrices of driven values for the vital rate(s) being modelled. The list has a length equal to the length of the driver input parameter.

Author(s)

Owen Jones [e-mail]

Examples

set.seed(42) # set seed for repeatability

# A single vital rate and a single driver
driven_vital_rate(
  driver = 14,
  baseline_value = 0.5,
  slope = .4,
  baseline_driver = 10,
  error_sd = 0,
  link = "logit"
)

# A single vital rate and a time series of drivers
driven_vital_rate(
  driver = runif(10, 5, 15),
  baseline_value = 0.5,
  slope = .4,
  baseline_driver = 10,
  error_sd = 0,
  link = "logit"
)

# A matrix of survival values (U submatrix of a Leslie model)
# with a series of drivers, and matrices of slopes and errors

lt1 <- model_survival(params = c(b_0 = 0.4, b_1 = 0.5), model = "Gompertz")
lt1$fert <- model_fertility(
  age = 0:max(lt1$x), params = c(A = 10),
  maturity = 3, model = "step"
)

mats <- make_leslie_mpm(
  survival = lt1$px, fertility = lt1$fert, n_stages =
    nrow(lt1), split = TRUE
)
mats$mat_U
mat_dim <- nrow(mats$mat_U)

driven_vital_rate(
  driver = runif(5, 5, 15),
  baseline_value = mats$mat_U,
  slope = matrix(.4,
    nrow = mat_dim,
    ncol = mat_dim
  ),
  baseline_driver = 10,
  error_sd = matrix(1, nrow = mat_dim, ncol = mat_dim),
  link = "logit"
)