LISTSERV mailing list manager LISTSERV 15.5

Help for ALGAE-L Archives


ALGAE-L Archives

ALGAE-L Archives


View:

Next Message | Previous Message
Next in Topic | Previous in Topic
Next by Same Author | Previous by Same Author
Chronologically | Most Recent First
Proportional Font | Monospaced Font

Options:

Join or Leave ALGAE-L
Reply | Post New Message
Search Archives


Subject:

Re: multivariate analysis advice

From:

mark browne <[log in to unmask]>

Reply-To:

mark browne <[log in to unmask]>

Date:

Sat, 19 Nov 2011 11:49:17 +0000

Content-Type:

text/plain

Parts/Attachments:

Parts/Attachments

text/plain (137 lines)


You may wish to consult this:

Clarke, K. R., P. J. Somerfield and M. G. Chapman (2006). "On
resemblance measures for ecological studies, including taxonomic
dissimilarities and a zero-adjusted Bray-Curtis coefficient for
denuded assemblages." Journal of Experimental Marine Biology and
Ecology 330: 55-80.

Cheers

Mark

2011/11/19 Brezo Martínez <[log in to unmask]>:
> Zeros are always a problem because then the data violated the multinormal
> assumptions of multivariate ordination. More important I think, the
> statistical approach, irrespectively of which one are you using, just have
> not enough information to model the species abundance (guest you have
> abundance data from several sites). I think its worth first to considered
> which species are minimally represented in your data to be included in the
> analysis, if they are very rare, I do not think they can be modeled together
> with more common species. I would considered to remove those that are not
> above a minimum abundance threshold. There is no much reason for those to
> alter the community pattern you are trying to investigate.
>
>
>
> There is a coefficient of dissimilarity, Bray-Curtis, that does not uses
> “shared zeroes”, it does not take as an indication of similarity between two
> samples the absence of a single species in both. It was specifically
> designed to deal with these kind of data with many zeroes and very
> recommended. Last versions of CANOCO software do Bray-Curtis (better option
> that log transformation). Alternatively you can use the software PERMANOVA
> with is non-parametric, based on interactive algorithms, and thus more
> flexible if the data does not fit the parametric assumptions. I think
> log-transform is more a way to reduce biased data showing large variation,
> this is a different situation.
>
>
>
> Hope this help, fell free to contact me if you find these comments useful,
>
> Brezo
>
>
>
> Dra. Brezo Martínez
>
> Área de Biodiversidad y Conservación, Universidad Rey Juan Carlos- URJC
>
> Departamental I, despacho 213, c/ Tulipán s/n, 28933 Móstoles, Madrid, Spain
>
> [log in to unmask]
>
> http://www.escet.urjc.es/biodiversos/engl/staff/brezo/brezo_i.htm
>
> Tf: +34 914888102           Fax: +34 916647490
>
>
>
>
>
>
>
> De: Forum for marine, freshwater and terrestrial algae.
> [mailto:[log in to unmask]] En nombre de Paula Yuri Nishimura
> Enviado el: viernes, 11 de noviembre de 2011 18:28
> Para: [log in to unmask]
> Asunto: multivariate analysis advice
>
>
>
> Dear Algae-L members,
>
>
>
> I am a freshwater phytoplanktologist and I am trying to run some
> multivariate analysis (CCA) in my data.
>
> All phytoplanktologists that I talked with and all articles from my field,
> they always transform the biological data (species density or biomass) by
> log(x+1), or something similar (the explanation is to avoid too many
> “zeros”).
>
> However, all statisticians that are NOT phytoplanktologists that I talked
> with, they told me that transforming the biological data by log(x+1) makes
> no sense.
>
> As I am just a biologist, with no great statistics expertise,  I ask to the
> more experienced ones:
>
> Does anyone have a good explanation of why should I or why should I not
> transform the biological data by log(x+1)?
>
>
>
> Thanks in advance,
>
> Paula
>
> ---
>
> Paula Yuri Nishimura
>
> Laboratório de Limnologia
>
> Departamento de Ecologia
>
> Instituto de Biociências
>
> Universidade de São Paulo
>
> Brasil
>
>
>
> ALGAE-L
> Algae-L is a listerver list for the phycological community; you are
> receiving this message because you subscribed to the list.
> You can get further information and unsubscribe at
> http://www.seaweed.ie/algae-l/
> When posting to the list please add your name, status and an institutional
> or brief postal address. The list will not accept attachments greater than 2
> Mb in size.
>
> ALGAE-L
> Algae-L is a listerver list for the phycological community; you are
> receiving this message because you subscribed to the list.
> You can get further information and unsubscribe at
> http://www.seaweed.ie/algae-l/
> When posting to the list please add your name, status and an institutional
> or brief postal address. The list will not accept attachments greater than 2
> Mb in size.

ALGAE-L
Algae-L is a listerver list for the phycological community; you are receiving this message because you subscribed to the list.
You can get further information and unsubscribe at http://www.seaweed.ie/algae-l/
When posting to the list please add your name, status and an institutional or brief postal address. The list will not accept attachments greater than 2 Mb.


Back to: Top of Message | Previous Page | Main ALGAE-L Page

Permalink



LISTSERV.HEANET.IE

CataList Email List Search Powered by the LISTSERV Email List Manager