The genus Chamaeza is compose for five species of neotropical birds, where 3 of that have populations in the Atlantic Forest, C. campanisona, C. meruloides and C. ruficauda. In the state of the Rio de Janeiro (Brazil), this species occur in partial simpatry, tending to occur separetadely in specific altitudinals range. My study investigates this pattern, questioning if this is more relate to competitive exclusion or with ecological specialization. Here I expose the scripts used to conduce the experiment.
clean_coords(xy = tibble-like, region = raster::RasterLayer or raster::SpatialPolygonDataFrame, remove_duplicates = Logical , get_centroids = Logical)
Get a coordinate dataset, remove duplicated points and also those outside a desire region, and get centroids (case region is a RasterLayer), returning a more geographically distributed dataset.
xy - The coordinates dataset; a object capable to be coarsed to a tibble object, with longitude and latitude in the first and second columns respectively.
region - The region where the coordinates should match over; a raster::RasterLayer object or a raster:SpatialPolygonDataFrame (polygon shapefile).
remove_duplicates - Should duplicated values be removed from result? TRUE removes, FALSE remains those. This argument is affected by the get_centroids and region arguments, if get_centroids is TRUE and region is a RasterLayer, the return value will be the unique values from centroids columns, case get_centroids is FALSE, the return value will be the unique values from coordinates in xy, what also happen if region is a SpatialPolygonDataFrame.
get_centroids - If region is a RasterLayer, should get the centroids of the cells with occurrence points? TRUE gets, FALSE ignore.
A tibble object with two or five columns, depending of the type of the region and the choose in get_centroids. Returned columns could be, in this order: cells (repective cells in the RasterLayer); long (longitude); lat (latitude); center.x (centroid related to longitude) and center.y (centroid related to latitude).
create_distribution(xy = tibble-like, layers = raster::Raster(Layer, Brick or Stack), kfold_count = Integer , algorithm_name = Character, threshold_name = Character)
Create the distribution model, using k-fold cross validation, returning also -dependent and -independent threshold evaluation.
xy - A data.frame, matrix or tibble with the occurrence points.
layers - The layers of variables to create the model.
kfold_count - The number of subgroups to perform a k-fold cross validation, the minimum are 2.
algorithm_name - What algorithm to use, options are: "domain","bioclim","mahal","maxent". In moment, working well only for maxent :(.
threshold_name - One of the dismo::threshold options: "kappa", "spec_sens", "no_omission", "prevalence", "equal_sens_spec", "sensitivity". See help(dismo::threshold) for details.
A named list with a distribution model ($model), a evaluation model ($evaluation), a predicted distribution ($distribution), a threshold value ($threshold_value) and a binary map of occurrence absence based in the threshold ($threshold_map).