Hi Gian, You may want to have a look at Clarke (1993) Non-parametric multivariate analyses of changes in community structure, Australian Journal of Ecology, 18: 117-143 You may get a better idea of NMDS approach.
Good luck Federico Date: Thu, 2 Feb 2012 19:30:06 +0100 From: [email protected] To: [email protected] Subject: Re: [R-sig-eco] metaMDS result Dear Jari, Thank you a lot to your precious advices. I am alarmed because the stressplot is indeed a bit strange. Dots are distribute along a line that is almost parallel to X axis (Observed Dissimilarity) and at the end they go up drawing a long pile of dots that is otherwise parallel to the Y axis (Ordination distance). The Non-metric fit R2 is 0.99 and the linear one is 0.995. Is I red well an accurate representation should be that dots must follor the diagonal f unction where x=y It is possible that is because my two datasets of samples share to few taxa one with the other? Thank you very much, G. 2012/2/2 Jari Oksanen <[email protected]> > > On 02/02/2012, at 19:43 PM, Gian Maria Niccolò Benucci wrote: > > > Hi Members, > > > > I tried to use a metaMDS to explore my two fungal communities data. > > That's the result: > > > > Call: > > metaMDS(comm = species.data_log, distance = "bray", k = 2, trymax = > > 200, autotransform = F) > > > > global Multidimensional Scaling using monoMDS > > > > Data: species.data_log > > Distance: bray > > > > Dimensions: 2 > > Stress: 0.0369308 > > Stress type 1, weak ties > > Two convergent solutions found after 13 tries > > Scaling: centring, PC rotation, halfchange scaling > > Species: expanded scores based on species.data_log > > > > species matrix was log10() trasformed before NMDS (otherwise I got really > > similar result with sqrt() trasfromed data) > > > > I wonder if the stress is too low according to dune examples and maybe > > there is something not working in my data or code... > > > > thanks you for helping, > > > > Not many people are worried for a low stress. NB, the stress is taken > directly from the NMDS engine, and the current monoMDS() uses different > stress scaling than the older isoMDS() engine: 0.037 of monoMDS corresponds > to 3.7 of isoMDS. This is a low stress, but not directly alarming. Perhaps > your data are easy and really two-dimensional (e.g., have some few strongly > dominant taxa). > > Looking at the stressplot() graphs may help in seeing if there is > something strange. > > Cheers, Jari Oksanen > -- > Jari Oksanen, Dept Biology, Univ Oulu, 90014 Finland > [email protected], Ph. +358 400 408593, http://cc.oulu.fi/~jarioksa > > [[alternative HTML version deleted]] _______________________________________________ R-sig-ecology mailing list [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
