Table 1. Input parameters, sampling distributions and parameter ranges, and brief description of all factors varied in the global sensitivity analysis of the whitepark pine metapopulation population viability analysis.
RAMAS module refers to the specific sub-program where the parameter is specified. Parameters varied in RAMAS Spatial are unique to GRIP 2.0, whereas those varied in RAMAS Metapop were introduced in GRIP 1.0 (Curtis & Naujokaitis-Lewis, 2008). Parameters with RAMAS Spatial/Metapop specified include habitat-specific environment-demography relationships varied in both modules.
| Input factora | Distribution and sampling range | Description | RAMAS module |
|---|---|---|---|
| Habitat suitability (HS) map | GRIP 2 has 4 options for varying habitat suitability values: | Spatially explicit habitat suitability values (i.e. raster files). These correspond to the predictive outputs of species distribution models. HS values are rescaled where the default settings are: the minimum is the HS threshold and the maximum is the highest HS value of the original landscape. Options exist to specify a theoretical maximum. | RAMAS Spatial |
| 1. Random normal: N(mean = mean of HS values in original landscape, SD = standard deviation of HS values in original landscape) | |||
| 2. Spatially-autocorrelated: HS surface derived from a simulated gradient where the degree of spatial autocorrelation between cell values can be modified. Uses functions from the ‘randomFields’ R package | |||
| 3. Ensemble: Uses ensemble predictions and the measure of uncertainty to vary new HS values. Ensembles could be based on multiple types of SDM algorithms used to model species distributions (e.g., GAM, GLM, RF, BRT). The current implementation of GRIP2 resamples HS values using a random normal variate with the mean based on the ensemble prediction for that gird cell with a SD based on the uncertainty estimate from the ensemble model (i.e., the model-based measure of variation) for that grid cell. | |||
| 4. Not varied | |||
| Neighborhood distance | N(Original value, 10% CV) | Used to find distinct habitat patches, represents the spatial scale at which the population can be assumed to be panmictic | RAMAS Spatial |
| Distance measure among habitat patches | D(‘edge to edge’, ‘center to edge’, ‘center to center’) | Measure used to calculate the distance among pairs of patches, edge and center refer to the location on the patch where the measure starts or ends | RAMAS Spatial |
| Habitat suitability threshold | N(original value, 10% CV) | Habitat suitability value used as the threshold to distinguish between non-suitable and suitable habitat on the raster habitat suitability map. Any grid cell value above the threshold will be considered for inclusion as a population (i.e., habitat patch) | RAMAS Spatial |
| Number of patches | N(original number, 50% CV) | RAMAS Spatial | |
| Initial abundance | N(mean value per patch is a function of total habitat suitability, CV = 10%) | RAMAS Spatial/Metapop | |
| Carrying capacity | N(mean value per patch is a function of total habitat suitability, CV=10%) | RAMAS Spatial/Metapop | |
| Rmax | N(original value, 10% CV) | Maximum growth rate | RAMAS Metapop |
| Catastrophe extent | D(local, regional) | Randomly varies spatial extent of catastrophe | RAMAS Metapop |
| Catastrophe probability | N(original value, 10% CV) | Probability of catastrophe occurring | RAMAS Metapop |
| Catastrophe intensity | N(original value, 10% CV) | Magnitude of catastrophe effect | RAMAS Metapop |
| Dispersal survival | U(0, 1) | Proportion of dispersers that live | RAMAS Metapop |
| Dispersal rate | N(0, 0.1)* dispersal rate | Each dispersal rate is varied by a constant value | RAMAS Metapop |
| Number of connections | U(0, number of pairwise population connections possible) | Varies number of population pairs connected through dispersal | RAMAS Metapop |
| Among-population correlation coefficient of vital rates | N(0, 0.1)* correlation coefficient | Varies magnitude of correlations in vital rates among population pairs | RAMAS Metapop |
| Seed survival | L(original value, original value) | Seed stage | RAMAS Metapop |
| Seedling 1 survival | L(original value, original value) | 1 year old seedling | RAMAS Metapop |
| Seedling 2 survival | L(original value, original value) | 2 year old seedling | RAMAS Metapop |
| Seedling 3 survival | L(original value, original value) | 3 year old seedling | RAMAS Metapop |
| Seedling 4 survival | L(original value, original value) | 4 year old seedling | RAMAS Metapop |
| Sapling mortality | L(original value, original value) | Sapling | RAMAS Metapop |
| Infected sapling survival | L(original value, original value) | Infected sapling | RAMAS Metapop |
| Nr adult survivalb | L(original value, original value) | Non-reproductive adult | RAMAS Metapop |
| Infected n survivalb | L(original value, original value) | Infected non-reproductive adult | RAMAS Metapop |
| Class 1fecundity and survivalb | L(original value, original value) | Healthy adult trees | RAMAS Metapop |
| Class 2 fecundity and survivalb | L(original value, original value) | Branch infected adult tree | RAMAS Metapop |
| Class 3 fecundity and survivalb | L(original value, original value) | Bole infected adult tree | RAMAS Metapop |
| Class 4 fecundity and survivalb | L(original value, original value) | 50% crown loss infected adult tree | RAMAS Metapop |
Notes.
a
Selected distributions and their parameters: D = discrete distribution (discrete value1, discrete valuex), where each value has equal probability of selection; N = normal distribution (mean, standard deviation—sometimes expressed in terms of coefficient of variation, % CV); L = lognormal distribution (mean, standard deviation); U = uniform distribution (minimum, maximum).
b
Denotes that the stage was included in the model of density dependence.