Description Usage Arguments Details Value Author(s) References See Also Examples
boyce.classification
function aims to investigate accuracy of the map obtained from function enirg.predict
, by using the continuous Boyce index (CBI; Boyce et al., 2002).
Curves from P/E ratio give the possibility to reclassify the predicted niche map (enirg.predict
) and thus distinguish unsuitable, marginal, suitable and optimal habitats. In accordance with the method proposed by Hirzel et al.(2006), it interprets predicted-to-expected ratio (P/E) by partitioning habitat suitability predictions into classes and by calculating frequencies. If model properly delineates suitable areas for the studied species, Spearman rank correlation coefficient of the ratio F_i, will be 1.
Evaluation of habitat suitability model accuracy is made by means of n-fold cross-validation (Fielding and Bell, 1997), partitioning data evenly but randomly into cv.sets
partitions. Once ratio is calculated, Spearman correlation coefficient allows to estimate fitting for the predicted-to-expected relationship.
Categories allows to use function classify.map
, in order to perform a classification on the HSM (enirg.predict
).
1 2 |
prediction |
vector. Predicted suitability values from observations or from a second validation data set. |
prediction.map |
vector. Predicted suitability values for the entire area of study. |
categories |
vector with desired categories. NULL if |
cv.sets |
integer, indicating the number of subsets to use for the cross validation. |
type |
string. If "manual", a GUI assists the classification process. If "none", |
outcat |
string. Name for object which will contain the results. |
A GUI allows a manual adjustment of suitability classes. An ideal model would give a straight P/E curve. Curve shape and its confidence interval can be used to define boundaries of habitat suitability classes (as suggested by vertical dashed lines).
This function displays predicted/expected ratio curve shapes. Also enirg.predict
returns a list object, containing the following components:
coefficients. A vector of two: spearman rank coefficient and adjusted r squared.
intervals. Suitability intervals for later using with classify.map
Fernando Canovas fcgarcia@ualg.pt
Boyce, M.S.,Vernier, P.R.,Nielsen,S.E.,Schmiegelow, F.K.A. (2002). Evaluating resource selection functions. Ecological Modelling 157, 281-300.
Fielding, A., Bell, J. (1997). A review of methods for the assessment of prediction errors in conservation presence/absence models. Environmental Conservation, 24, 38-49.
Hirzel, A.H., Le Lay, G., Helfer, V., Randin, C., Guisan, A. (2006). Evaluating the ability of the habitat suitability models to predict species presences. Ecological Modelling 199, 142-152.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 | data(apis.hsm)
# vector of predictions for observations:
apis.predictions <- apis.hsm$predictions[, 2]
# vector of predictions for the entire predicted map:
apis.predictions.map <- as.vector(na.exclude(apis.hsm$African_predicted_hsm@data@values))
# Try with intervals:
# unsuitable = 0.65
# marginal = 0.84
# suitable = 0.96
# Note that this species has an optimal distribution
# in the study area, resulting in a wide unsuitable
# interval and narrow suitable ones.
boyce(prediction = apis.predictions,
prediction.map = apis.predictions.map,
categories = c(0, 0.65, 0.84, 0.96, 1),
cv.sets = 10, type = "none")
|
Loading required package: ade4
Loading required package: miniGUI
Loading required package: tcltk
Loading required package: raster
Loading required package: sp
Loading required package: rgrass7
Loading required package: XML
GRASS GIS interface loaded with GRASS version: (GRASS not running)
Warning message:
no DISPLAY variable so Tk is not available
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.