Angelique,

Many thanks for explanations and for the link, I will try the software for
sure.
I have on year more untill my PhD end, maybe I will disturb you for more
helps... I am also planning to come to Germany for the final foreign period
of my PhD! How small is the world? :)
Cheers,

Gian





2010/6/9 Angelique Gobet <[email protected]>

> Hi Gian,
>
> I'm happy if I could help you a bit :) Multivariate analyses are not easy
> (it's been almost 4 years I work with it and I am still learning...), so you
> are completely right to ask for help and the R-sig-ecology mailing list is a
> great source of help. About having a workshop on community ecology
> statistics, I think it is a really good idea but unfortunately I do not have
> that much time right now to be involved in such a thing...I am currently
> finishing my PhD and I think I will be quite busy until this Autumn.
> However, I would be glad to try to help if you have any other questions on
> community ecology statistics :)
> By the way, in case you don't know it, there is a cool website where you
> can download an easy software (with the according documentation) to do
> quickly some multivariate analyses (this is what I use to propose to people
> just starting with statistics and I still use it when I need a quick
> overview on my analyses): http://folk.uio.no/ohammer/past/
>
>
> Cheers
> Angélique
>
> Gian Maria Niccolò Benucci wrote:
>
>> Hi Angélique,
>>
>> Many many thanks about your help, I will read the paper carefully. It's
>> always a pleasure to find matherial where to study above all for me that I
>> am at least the only one in my faculty that study fungal community ecology.
>> I am doing the best I can with my own skills and with a big passione for
>> that job.
>>
>> The overlapping is sure indeed, the two communities are form the same
>> site, but they come from 2 different host species. It's interesting for me
>> to elucidate the effect of host in structuring the community, and that's
>> seems to be true in reason of that some species I found, live only
>> associated with one host tree only. As you argued some other species are
>> shared, and indeed the communities have some species in common.
>> I would try to do again the ANOSIM test, maybe with differend
>> standardizaion of the dataset and also with the rough data matrix.
>>
>> Thanks :)
>>
>> PS. For Rookies like me, would be interesting to have a workshop on
>> community ecology statistics, or a work group for  talking and exchanging
>> experiences, opinions and skills. don't you think?
>>
>>
>> G.
>>
>>
>>
>>
>>
>> 2010/6/9 Angelique Gobet <[email protected] <mailto:
>> [email protected]>>
>>
>>
>>    Hi Gian,
>>
>>    I just read the thread about the ANOSIM test and I would like to
>>    give you a bit more insights about the interpretation of the R
>>    value. You probably have read it already but the ANOSIM tests for
>>    the significance of difference between two or more groups of
>>    samples, based on a dissimilarity matrix. The ANOSIM compares
>>    distances among groups with distances between groups. An /R/ value
>>    > 0.75 indicates clearly different groups, an /R/ value > 0.5
>>    indicates separated groups but with some overlapping between the
>>    groups while an /R/ value < 0.25 indicates almost no differences
>>    between the groups. The significance of the ANOSIM can then be
>>    computed with permutation tests (as you did). In your case, I
>>    would say that the two fungal mycorrhizal communities are
>>    different but with some overlapping.
>>
>>    Also, I would also use the same approach as you did: first and
>>    NMDS followed by an ANOSIM. As you said, in an NMDS, you can only
>>    compare the groups (here, the two fungal mycorrhizal communities)
>>    on a 2 dimension plot and an ANOSIM test would give you more
>>    information about the relationships between the two fungal
>>    mycorrhizal communities.
>>
>>    I attached some papers that might also help. The Ramette paper is
>>    a clear review of multivariate analyses. For more details, you can
>>    still have a look at the Legendre and Legendre (1998) book.
>>
>>    Hope this helps :)
>>    Cheers
>>    Angélique
>>
>>
>>
>>
>>
>>    Gian Maria Niccolò Benucci wrote:
>>
>>        Hi Dulce,
>>
>>        Unfortunately isn't possible to post graphs here, anyway I
>>        have a  nMDS
>>        graphs where you can see the spread of community data. For one
>>        community
>>        samples are more close one to each other, while for the second
>>        communty
>>        samples look more scattered but no one fills in the area of
>>        the first one.
>>        The stress is ok for 2 dimensions, it is 24 if I do not
>>        remember bad. I just
>>        talked about it in another post, and I was suggested to look
>>        to graph more
>>        than to the stress value.
>>        Anyway, I looked for ANOSIM test just to include it as a
>>        support for my
>>        data. I think that could work for that aim, don't you think?
>>        Thanks a lot!
>>
>>        G.
>>
>>
>>
>>
>>        2010/6/9 Maria Dulce Subida <[email protected]
>>        <mailto:[email protected]>>
>>
>>
>>
>>             Hi Gian,
>>
>>            In an ANOSIM the value of the R statistic is at least as
>>            important as its
>>            statistical significance. So you must always look at both.
>>            Besides, if you
>>            have a lot of replicates you increase the chance of a
>>            small value of R being
>>            significantly different from zero.
>>            I would say that 0.15 is a rather small value for R but
>>            one must always
>>            look at the data. In your particular case, with an R =
>>            0.27, I would take a
>>            look to the nMDS of the data (using the same similarity
>>            measure as for
>>            ANOSIM, of course!). If it has a low stress, it gives a
>>            reliable picture of
>>            the positioning of your samples and if your samples are
>>            definitely grouped,
>>            you should also be able to see it in the nMDS. More
>>            important than any
>>            significance level is what you think your data is telling you!
>>
>>            Kind regards,
>>
>>            Dulce
>>
>>             Maria Dulce Subida
>>
>>
>>
>>            ~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*
>>
>>
>>
>>            Instituto de Ciencias Marinas de Andalucía (ICMAN)
>>
>>            Consejo Superior de Investigaciones Científicas (CSIC)
>>
>>            Campus Universitário Río San Pedro
>>
>>            11510 Puerto Real - Cádiz. España.
>>
>>
>>
>>            www.icman.csic.es <http://www.icman.csic.es>
>>        0034 956832612 ext. 316
>>
>>
>>
>>
>>            ~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*
>>
>>
>>
>>
>>            Gian Maria Niccolò Benucci escribió:
>>
>>            Barbara,
>>
>>            Thank you very much, indeed I red the two values on a
>>            paper, that were of
>>            R=0.15 and P=0.006, and they states that the difference
>>            between the two
>>            groups is significant with them. So, my values, that are
>>            aloso higher, are
>>            better for sure to support the difference between my two
>>            community. Don't
>>            you agree?
>>            Thank you really much for all,
>>
>>            G.
>>
>>
>>
>>
>>            2010/6/8 <[email protected]
>>            <mailto:[email protected]>>
>>            <[email protected] <mailto:[email protected]>>
>>
>>
>>
>>              Hi Gian,
>>            I'm not sure if anyone else has answered your question yet.
>>            But basically: the R statistic is a measure of the
>>            strength of the
>>            difference between your 2 groups. It considers the
>>            variance both within,
>>            and between, your groups. It can be from 0-1, with 1 the
>>            strongest
>>            difference.
>>            The significance value tests the significance of the
>>            difference, like
>>            any other p-value. The statistics you derived from your
>>            ANOSIM analysis
>>            are somewhat analogous to an F-statistic, and a p-value, for a
>>            univariate ANOVA test.
>>            When including the results in a paper or dissertation, it
>>            would be
>>            normal to quote both values (R and significance).
>>            I hope that's helpful,
>>
>>            Barbara
>>
>>
>>
>>            ----- Original Message -----
>>            From: Gian Maria Niccolò Benucci <[email protected]
>>            <mailto:[email protected]>> <[email protected]
>>            <mailto:[email protected]>>
>>            Date: Tuesday, June 8, 2010 4:57 am
>>            Subject: [R-sig-eco] ANOSIM test
>>
>>
>>
>>             Hi everybody,
>>
>>            I am tring to get some statistical evidence of differences
>>            between
>>            my two
>>            fungal mycorrhizal communities (samples from two different
>>            Host
>>            species)with ANOSIM test, that's my code:
>>
>>
>>
>>             dist.ABCD <- vegdist(sqrt.ABCD.1)
>>            attach(env.table1)
>>            ABCD.ano <- anosim(dist.ABCD, Host)
>>            summary(ABCD.ano)
>>
>>
>>             Call:
>>            anosim(dat = dist.ABCD, grouping = Host)
>>            Dissimilarity: bray
>>
>>            ANOSIM statistic R: 0.272
>>                Significance: 0.001
>>
>>            Based on  999  permutations
>>
>>            Empirical upper confidence limits of R:
>>             90%    95%  97.5%    99%
>>            0.0407 0.0554 0.0724 0.0952
>>
>>            As I read in a paper, also an R value of 0.27 could be
>>            considered
>>            significant. So similar samples are inside the communities
>>            and not
>>            outsidethem, is it roght?
>>            Thank you for you time and patience,
>>
>>
>>            Gian
>>
>>
>>
>>            --
>>            Gian Maria Niccolò Benucci
>>            Ph.D. Candidate
>>            Department of Applied Biology - University of Perugia
>>            Borgo XX Giugno, 74
>>            I-06121 - Perugia, ITALY
>>            Tel: +39.0755856433
>>            Fax: +39.0755856069
>>            Email: [email protected] <mailto:[email protected]>
>>
>>
>>
>>            ----- Do not print this e-mail If you do not need it. Save
>>            paper
>>            and protect
>>            the environment! -----
>>
>>                 [[alternative HTML version deleted]]
>>
>>
>>
>>
>>             _______________________________________________
>>            R-sig-ecology mailing
>>            [email protected]://
>> stat.ethz.ch/mailman/listinfo/r-sig-ecology
>>            <http://stat.ethz.ch/mailman/listinfo/r-sig-ecology>
>>
>>
>>
>>                   [[alternative HTML version deleted]]
>>
>>
>>
>>            ------------------------------
>>
>>            _______________________________________________
>>            R-sig-ecology mailing
>>            [email protected]://
>> stat.ethz.ch/mailman/listinfo/r-sig-ecology
>>            <http://stat.ethz.ch/mailman/listinfo/r-sig-ecology>
>>
>>
>>
>>
>>
>>
>>
>> ------------------------------------------------------------------------
>>
>>
>>
>>        _______________________________________________
>>        R-sig-ecology mailing list
>>        [email protected] <mailto:[email protected]>
>>
>>        https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
>>
>> ----
>>
>

        [[alternative HTML version deleted]]

_______________________________________________
R-sig-ecology mailing list
[email protected]
https://stat.ethz.ch/mailman/listinfo/r-sig-ecology

Reply via email to