[R-sig-Geo] inquiry on suitable packages

Tomislav Hengl hengl at spatial-analyst.net
Mon Oct 26 13:16:36 CET 2009


Hakim,

There are plenty of (SDM) algorithms that generate species  
distribution maps given occurrence (or density) records and  
environmental maps. For a systematic review, see e.g. Tsoar et al.  
2007 [http://dx.doi.org/10.1111/j.1472-4642.2007.00346.x].

In R, you can try using the adehabitat package that can be used to  
generate Habitat Suitability Index maps using methods such as madifa  
and/or enfa (read more about ecology-analysis packages via the  
"Environmetrics" views). Here are some simple examples:

http://spatial-analyst.net/scripts/bei.R

I myselft have recently been testing MaxEnt  
[http://www.cs.princeton.edu/~schapire/maxent/] and discovered that it  
has several advantages over competition (GARP, ENFA):
- it accepts both categorical and continuous predictors;
- it allows cross-validation;
- it is fast(est);

In short, MaxEnt should be high on your list because it is also the  
most used SDM by biologists at the moment.

There is no package for R to use MaxEnt, but there is probably no need  
for this neither. The whole algorithm comes in a single ".jar" file  
that can be run via batch commands (i.e. from R). Here are some  
examples using the global environmental layers:

http://spatial-analyst.net/scripts/worldmaps.R

Note also that there is also a R mailing list for ecology  
(R-sig-ecology) where you will probably be able to get much more  
feedback.


HTH,

T. Hengl
http://home.medewerker.uva.nl/t.hengl/


Quoting Hakim Abdi <hakim.abdi at uni-muenster.de>:

> Dear All:
>
> I have a set of nine independent variables (extracted Landsat TM band
> DNs, land surface temperature, and NDVI) and a set of eight response
> variables (presence/absence breeding season survey of eight bird
> species). I'm doing a small exercise to try to find a model that best
> fits the relationship between the independent variables and species
> occurrence and then apply that model to another set of independent
> variables from another time period (and even from another area) and see
> what the distribution might be at that time (or place).
>
> I am currently working with the packages BIOMOD Gstat and spatstat. I'm
> wondering if there are any others I'm not aware of that could further
> help me?
>
> Thanks in advance for your time and assistance.
>
> Hakim
>
> --
>
> Mit freundlichen Grüßen/Regards
> __________________________________
> Abdulhakim M. Abdi
> Erasmus Mundus Master's Programme in
> Geospatial Technologies
>
> Institut für Geoinformatik
> Universität Münster
> 32U 5755408.16mN, 404463.35mE
> ifgi.uni-muenster.de
>
> Instituto Superior de Estatística e Gestão de Informação
> Universidade Nova de Lisboa
> 29S 4287095.5mN, 486099.7mE
> www.isegi.unl.pt
>
> www.geospatialtechnologist.com
> www.hakimabdi.com
>
>
> 	[[alternative HTML version deleted]]
>
>



More information about the R-sig-Geo mailing list