Skip to contents

Fit previously-developed species distribution models for riparian landbird species and waterbird groups during the fall or winter to a new set of predictors, such as those derived from a new scenario of landscape change.


  constants = NULL,
  factors = NULL,
  landscape = NULL,
  unsuitable = NULL,
  overwrite = FALSE


pathin, SDM, landscape_name

Character strings defining the filepath (pathin/SDM/landscape_name) containing new predictor rasters to include in the model, such as those created from running python_focal_finalize()


List of model objects of class 'gbm' representing the distribution models to which new predictors should be fit.


optional dataframe containing predictors with a constant value that should be applied to all pixels. See Details.


optinal list of named lists defining categorical predictors included in distribution models. See Details.


optional SpatRaster corresponding to the landscape represented by the predictors contained in pathin/landscape_name, used to identify the locations of unsuitable land covers. Must be provided if unsuitable is not NULL.


optional vector of numerical values representing the land cover classifications that should be considered unsuitable a priori. If not NULL, landcape must also be provided.


Character string defining the filepath (pathout/SDM/landscape_name) where output rasters should be written


Logical; passed to terra::writeRaster()


Nothing returned to R environment. Writes rasters to pathout for each model in modlist


This function is designed to fit multiple distribution models to the same set of predictors describing a given landscape. New predictors must first be created and named to match the predictors included in the original models, e.g. using python_focal_prep(), python_focal_run(), and python_focal_finalize().

constants are passed to terra::predict() and provide a way to include constant values for one or more predictors that should be applied to all pixels. For both riparian and waterbird models, this will include a predictor representing effort ('area.ha' for riparian landbirds and 'offset' for waterbirds). For riparian landbird models applied only to the Delta, constants should also include a region predictor used as a categorical predictor representing the Sacramento Valley (0) or the Delta or San Joaquin Valley (1). (See vignette)

factors are also passed to terra::predict() and provide a way to define categorical predictors. For waterbird models, this is necessary to define the 'covertype' predictor. (See vignette)

unsuitable land covers will be presumed to have a predicted value of zero. The locations of unsuitable landcovers will be extracted from landscape, assigned a value of zero, and overlaid on the model predictions.


# See vignette