EOO is NA for presences in a straight line

[AMBarbosa]

In my use case, it is not too uncommon for some species to have only one or two presence points in the study area. In these cases, the EOO "polygon" is actually a point or a line, so it has an area of zero. IUCN guidelines say that, in such cases, EOO should be set as =AOO. Actually I see this in the code for getAnywhere(.IUCN.comp): if (EOO < AOO) EOO <- AOO. However, this seems to get lost somewhere, as the result of IUCN.eval() still says EOO is NA for these species. Maybe it should be changed to if (is.na(EOO) || EOO < AOO) EOO <- AOO?

I've also tried specifying buff_width to get an actual polygon around the occurrences, but it seems to be ignored, as well as the draw.poly.EOO argument.

Here's a reproducible example:

(example <- structure(list(lon = c(-2.59487105548141, -2.01867048441999), lat = c(39.1635112135211, 38.9798640743379), species = c("myspecies", "myspecies")), row.names = c(1L, 2L), class = "data.frame"))
#         lon      lat   species
# 1 -2.594871 39.16351 myspecies
# 2 -2.018670 38.97986 myspecies

ConR::IUCN.eval(example,
                buff_width = 10,  # apparently ignored
                parallel = TRUE,
                write_results = FALSE,
                draw.poly.EOO = TRUE)
#  |=================================================================| 100%
#           EOO AOO Nbe_unique_occ. Nbe_subPop Nbe_loc Category_CriteriaB
# myspecies  NA   8               2          2       2                 EN
#           Category_code Category_AOO Category_EOO
# myspecies        EN B2a           EN         <NA>

A fix would be greatly appreciated! Cheers

[gdauby]

Hello, thanks for your comment.
In fact the condition if (EOO < AOO) EOO <- AOO is applied only if EOO can be calculated. But if you have one or two occurrences EOO cannot be calculated.
The test for a straight line is only done when there is more than 3 pairs of unique coordinates.
Whether EOO should be set equal to AOO if it cannot be calculated is a good question. I should check what the guidelines says.

[AMBarbosa]

I figure if the condition applies to 3 presences in a straight line, it should also apply to 2 presences (which are necessarily in a straight line), and ultimately to one presence as well. The extent (area) of a line or a point should be zero; it should only be NA if there were no presence points at all. If there is an AOO, to me it only makes sense to have an EOO too.

[gdauby]

You are right, I'll modify this asap

[gdauby]

Dear @AMBarbosa

since I am currently working on a new version of the package, I updated the developmental version which is quite different from the original one. I cannot maintain both packages in parallel. The ConR currently on CRAN will probably be achived at some point and replaced by the new version.

I therefore updated the new version accordingnly and I suggest you try this one

devtools::install_github("gdauby/ConR@devel")

library(ConR)

(example <- structure(list(lon = c(-2.59487105548141, -2.01867048441999), lat = c(39.1635112135211, 38.9798640743379), species = c("myspecies", "myspecies")), row.names = c(1L, 2L), class = "data.frame"))

criterion_B(x = example )

The ConR::IUCN.eval not longer exist and has been replaced by criterion_B.

Let me know if you have any issue or questions.

Concerning this issue, note that EOO is not set equal to AOO if EOO < AOO or if EOO cannot be computed which is the case when there are less than 3 unique pairs of coordinates.
If pairs of coordinates form a straitgh line (tested when at least 3 unique pairs of coordinates), 'noise' is added to coordinates to avoid the straight line (no buffer anymore).