Functional Diversity - Science topic

Beta diversity can be broken up into the components of nestedness and turnover. Which helps to explain the difference between sites. I believe you can explore this with or without including abundance for the calculations. There are R packages that can be used for the calculations. You can also find papers and packages that explore species and local contributions to beta diversity.

When you have a set of categorical variables, you can calculate Gower's distances between objects and then use common distance-based indices of functional richness, evenness and divergence in the FD package.

Dear Colleagues. I am pleased to share with you the first snake functional diversity study assessing effect scale, landscape composition, and roadkill.

"Functional diversity of snakes is explained by the landscape composition at multiple areas of influence"

James, I agree with you.

Yes, it is a Beta delta-time comparison. It is crucial that the methodology that you use to get your data in each year is entirely the same. Nice question!

Kindly go through the following link, I hope all doubts regarding R^2 shall be cleared.

Oh and I maybe found also what explains your low FRic values, dbFD does not standardize FRic by default, see the parameter stand.FRic

Hi! I think you should prepare two matrix, one is abundance matrix where site or sample in row and species in line, another matrix is trait matrix where species in row and triat indices in line. Note that the order of the species in two matrix is the same. Then you can use dbFD function in FD package in R to calculate the functional diversity indices of each site or sample, including functional richness, functional divergence, function evenness, community weighted mean of trait, etc.

Functional diversity metrics are calculated by associating species-by-site matrices, such as presence-absence or abundance of species, to the species' functional traits; i.e. morphological or behavioural traits that are related to the role the species may perform in the ecosystem. Functional diversity is a type of alpha or Simpson diversity and the link below will show you how to calculate it.

I suggest Klaus Fichter 's innovation community construct - based on Witte's promotor theory - which helps to analyse cross-functionional and cross-organisational (multi-level) innovation processes. We further developed the theory in the context of circular economy (and broader sustainability) innovation here by adding specific collaboration mechanisms: https://www.researchgate.net/publication/344281476.

If you are trying to calculate some sort of index that tells you how far apart species are in functional space, on average, then of course you can't calculate an average when there are only 2 or 1 species, there's one value, or none. Personally I prefer distance-based metrics, as creating a tree and measuring distances through it is very distorting of the actual distances among species. You could calculate a distance between 2 species, but with one it would be undefined.

please see

https://pubs.er.usgs.gov/publication/70193625

Hi, Barbara

First of all, I appreciate your way of previously analyzing data and I endorse all Jean's suggestions. The way of including second-order terms are also the best I know so far.

For clarification, I have some minor perspectives.

First, to understand the statement of Cohen et al. to justify the inclusion of the lower-order terms when dealing with quadratic (or another polynomial) equations, it is easier to drawback to their basic form. Let's take a look at quadratic ones:

f(x) = a + bx + cx²

The lower-order term, in this case, is b, which is, by definition, the coefficient that defines the position of the curve on the X (horizontal) axis. If you remove it from your model fitting, you will declare that it is equal to zero and will force the inflection to occur at (0,y). In practice, your hump-shaped equation will peak when 'habitat complexity' is zero, which may be unrealistic. Thus, it is fundamental for model fitting that your b should be non-zero, and I must say negative, so the curve will trace rightwards.

Second, the likelihood approach actually provides you with some useful guidance. If not sure, confirm it through the AIC.

Best wishes,

Matheus

Thank you Artëm Sozontov and Pedro Luna

Regards,

Sileesh Mullasseri

Please go through the attached file, hope this will serve your purpose.

I agree with Chimi Djomo Cédric 's answer!

Hi all, I've got a similar question with FDis values from 9 mammal assemblages which range from 4.001 to 5.031 (mean ± SD = 4.690 ± 0.290). I'm using a set of 28 traits, mostly binary traits. As far as I know, these values appear to be very high. For the same assemblages, functional evenness ranged from 0.481 to 0.667 (mean ± SD = 0.557 ± 0.048) and and Rao’s square entropy from 17.62 to 26.04 (mean ± SD = 23.150 ± 2.582).

Thank you in advance for any suggestions!

If we assume that phylogenetic diversity generally correlates with functional diversity (of course there are exceptions, especially in extremely morphologically clades) it would indeed be interesting to see if this relationship is still observed with genes under strong selection (as genes coding for functional traits would be). I think the relationship you would observe will depend strongly on the trait.

Ref.3

Hi Peter, just use R! It is free and has several packages to compute functional diversity metrics. One of the most-used packages is FD (function dbFD) that allows computing functional richness, functional eveness and functional divergence, among others.

Cheers

Dear Maria

Thank you for your help

Have a nice week.

Tom Vincent van der Meer

For N functional genes I would use the following primers:

Nitrification:

- amoA bacteria: amoa-1F/amoa2R (Rotthauwe et al 1997)

- amoA archaea: Arch-amoF/Arch-amoAR (Francis et al 2005)

Both pairs of primers are good and you will not really find better pairs even if they are old. The region are really conserved, and the primer show a good coverage.

Denitrificaiton:

It is a bit more difficult because some primer pairs show low coverage, when other are better but so far there is no perfect primers pairs. This is especially the case for nirK that show poor coverage. I would use the following primers pairs:

- nirK gene: nirK876F/nirK1040R (Henry et al 2004)

- nirS gene: cd3aF/R3cdR (Throback et al 2006)

- nosZ clade I gene: nosZ2F/nosZ2R (Henry et al 2006)

- nosZ clade II gene: nosZ-II-F/nosZ-II-R (Jones et al 2013)

N fixation:

nifH gene: PolF/PolF (Poly et al 2001)

There is other target for the N cycle (nxrA/B, NOB, narG...) but with these genes you have a good starting point.

For P cycle it is much more difficult. There are primers which have been fairly recently developed, but based on all our tests and attempts to develop new ones, it is rarely specific with concern on what they actually amplify. So I would first focus on the one for N cycle before moving into the P cycle. Here are few primers pairs you can test that should work:

phoD: phoD-F733/phoD-R1083 (Ragot et al 2015)

phoD: ALPS-F730/ALPS-R1101 (Sakurai et al 2008)

phoC: phoC-A-F1/phoC-A-R1 (Fraser et al 2017)

Regarding the Standard, the way we do, is to perform a PCR with a positive control (from environmental samples rather than culture) for each primers pairs, run an agarose gel to check for specificity, do a gel extraction on the band and a PCR clean up and then quantify the DNA by fluorometer (e.g. Qubit). Then the quantified DNA is used for a 10-fold serial dilution. This is a quick and easy way of doing a standard, it allow you to create a standard from a complex microbial samples, that will behave more closely to your samples than a pure culture. You can calculate the number of gene copies based on DNA concentration following an equation (see Smith et al 2006; Evaluation of quantitative polymerase chain reaction based approaches for determining gene copy and gene transcript numbers in environmental samples).

For the qPCR reagents, you can use a 2 steps reagents, such as QuantiFast that show good results on N and P cycles for us on soil samples. So you can have the same amplifications conditions for all primers, which is practical to run different mix/primers on the same plate.

You can find more info on the standard, amplification, N cycle primers (with full ref) on the following article we published earlier this year.

Dear Charlotte

I think that the topic of gut microbiota functionality is very interesting. I think that with NGS you could get a lot of information about the function of the organisms in your samples. You have at least two options:

1) You could perform a whole metagenome sequencing (WMS) by sequencing the DNA of all the species in your sample. In this case you could identify most of the present species, getting up to the strain depending on the depth of sequencing and the bioinformatics pipeline. In addition you could get an indirect functional analysis because you could identify the genes associated to the different taxa and then infer their function.

2) The most direct way to address the functionality problem would be through a metatranscriptomics analysis. In this case you would sequence all the transcripts expressed by the microbiota. Thanks to this approach you could not only identify the species in your samples (even though their quantification might be tricky) but also quantify the expression levels of the genes they express. Then by associating genes to functions you could "easily" correlate taxonomy with functions.

We have analyzed several metagenomics and metatranscriptomics data and I can tell you that you can obtain a nice picture about the function of the species and the pathways activated in your samples. With a proper experimental design you could also perform a differential analysis to identify functions which are associated to a specific group of samples.

We have recently developed an online and user-friendly tool to analyse amplicon, WMS and metatranscriptome data, if you want you could give a look at it: https://metagenomics.sequentiabiotech.com/

I hope this might be useful.

Have a nice day

Riccardo

Bioinformatics data analyst at www.sequentiabiotech.com

In addition to what Tyler Bourret pointed out, there is an increasing body of literature that the ITS region(s) vary in their utility for identification of a large number of fungi (see some of the previous discussions elsewhere on ResearchGate such as https://www.researchgate.net/post/What_primer_sets_are_good_for_fungal_community_profiling_by_pyrosequencing) and that multilocus methods are much better for many important and diverse groups of fungi (e.g. http://www.westerdijkinstitute.nl/fusarium/). In addition, there are numerous errors in the identifications and classifications associated with ITS in some of the major databases (e.g. DOI 10.1007/s13225-014-0291-8, doi.org/10.1101/288654, etc.) as well as database biases (e.g. http://geoffreyzahn.com/limitations-of-the-unite-fungal-database/). These types of complications and errors make assigning a functional prediction based on ITS difficult and such assignments questionable.

If you want to avoid using functional indices then you could gather functional data in association with your human population intensity gradient and use an Non-metric multidimensional scaling (NMS) ordination of sites in trait space. This is similar to what Damaris is suggesting with her option #1. You can then run a PERMANOVA to test for significant dissimilarity of group centroids -- however this is not the best test to run if you have convergence/divergence of functional traits as a response of population intensity.

I will mention that Facundo is right in that this technique is usually used as a method of indirect gradient analysis, relating environmental variables with functional traits. But it depends on the matrices you are using.

Lastly, I would suggest you read this paper that outlines functional trait diversity metrics using the FD package in R, and reconsider avoidance of functional diversity indices. Visual inspection of your data using NMS should be followed up by different statistical tests of variable FD metrics depending on the what you want to measure.

Is most cases, I use only the dominant feeding group in the computation of percentage of each FFG. This is not just a question of stage/age but it is also a question of avalaibility of ressources according to the season considered.

The use of species diversity indexes is a very complicated issue. in being able to select one or the best. At present there are many works that make references to the use of several indices. for example I recommend the following work:

Morris, E. K., Caruso, T., Buscot, F., Fischer, M., Hancock, C., Maier, T. S., ... & Socher, S. A. (2014). Choosing and using diversity indices: insights for ecological applications from the German Biodiversity Exploratories. Ecology and evolution, 4(18), 3514-3524.

Personally, I prefer to use abundance-dominance curves. You can see his employment in the following work:

No, your functional groups are the functional richness, for the index FD and others you need traits. Acording to your research question, you can measurament your traits or find traits in data bases for example https://www.try-db.org/TryWeb/Home.php

This publication was not available on Sci-Hub but I fortunately got it from another universitary library in South Africa !

If someone also needs this paper, don't hesitate to contact me, I would be glad to send it to you ;)

I concur with the prior answers. A methodological answer to why you're unable to calculate the metrics is because the FRic (at least in R package "FD") is calculated on a principal coordinates analysis that requires more species than traits in each sample. There are corrections (see ?dbFD or ?calc_metrics in R package "ecospace") you can use to try to get around the limitations. But if some samples really have less than 3 species, then you will not be able to calculate this metric. (Technically, all Villeger's metrics are based on PCoA space, but only FRic requires the "more species than traits" requirement for the convex hull calculation to work correctly. If using dbFD to calculate, you can "turn off" FRic calculation with dbFD(... calc.FRic=FALSE), and you'll still get the others.

See this file:)

Thanks Xavi for sharing your paper.

There is no problem to use diversity index on OTU table, and with this you can avoid problems of bad taxonomical affiliation. In fact, the most common is do diversity analysis at OTU level.

I think the editor is asking you to think on why you went to sample a bacterial community at that site in first place. Why you did not assume that the bacterial diversity there was going to be the same than in other saline sites? if you (or whoever designed the experiment) would have thought that it was going to be same, then you would have not invested the money and the time doing that. So you had an hypothesis, that was probably that the diversity was going to be different from other similar sites. Ergo you have to prove it, and in order to do that you will need to compare your results with the results that have been obtained in other similar ecosystems. 

Now, you can represent a diversity from one site with a profile. For example:

1500 sequences from species-A, 765 sequences from species-B, etc.. 

and if you arrange your number of species (or any other taxonomic rank of interest) in a particular order then you can have a vector like (1500, 765, ... ) 

You can get the same from other sites, so you will have a number of vectors that you can compare with the vector representing the profile of your site. And here you a lot of statistical methods to compare such data, for example constructing a similarity matrix with those vectors and do some cluster analysis, etc.

I hope that helps.

Best regards.

I would like to use beta diversity across all the sites. Can I use beta diversity for group level taxonomy of meiofauna such as nematoda, harpacticoida, polychaeta, oligochaeta, tanaidacea, etc.

Hi Daniel

If you have the data on which arthropod species are from which location, you could use rarefaction to rarefy to the smallest number of locations you have for a particular plant species. That is, if I sampled a smaller number of locations for plant species x, how many arthropod species would I expect on average?

This approach is commonly used for alpha-diversity patterns but not beta-diversity. However, in principle, it could be done by repeatedly subsampling the data and recalculating the dissimilarity matrix (I assume you are using dissimilarity because you mentioned dendrograms).

A much simpler alternative is to use Simpson's Dissimilarity. If increased sampling effort leads to more arthropod species per plant species (a very likely assumption), then Simpson's Dissimilarity is designed to correct for this difference in species richness between samples. In practice, I find it an over-correction but it might work fine for your data.

Indeed there are other aspects that need to be considered, such as species abundance and response diversity. Looking at correlations between response and effect traits can help assessing the importance of functional redundancy: in case of a strong correlation, there is limited response diversity within functional groups and so functional redundancy does not insure resilience of the function. At the contrary, higher functional redundancy should provide with higher response diversity in case there is no correlation between response and effect traits. This I believe can guide local active restoration.

I think you are right Koenraad, I don't need to determine the part of the correlation that is due to the phylogeny (also it is very interesting).

Megan, I work on coral reef species, so yes, they are closely related.

Cheers,

First you need to annotate your sequences to find out from which genes/functions each sequence belong. Then you can rarefy the gene count table directly, it is much more efficient, fast and as roubust to do it like this. Then you need to run proper statistical analysis to mine your data properly. Good luck :-)

OK, there are two possible concepts of functionality: one is for genes encoding enzymes acting in important metabolic steps, or another is about the genes that are actively transcribed vs the ones that are present and had been selected for a long time in the community. Is your analyses on total metagenomic/metatrasncriptomic datasets or is it on PCR amplicons targeting some genes in DNA and RNA (cDNA templates) environmental extracts or is one single organism? Depending of the complexity of the community (in soil is extremely diverse and a huge number of sequences will not cover the diversity of the metagenomes), then it is possible that you may run comparison of the reads that are found in both datasets. But this results will have always the limitation about how representative they are of the total RNA or DNA content. Is not an easy question to solve. As an starting point I would say, try to get the same amount of sequences on each dataset (metagenomic and metatranscriptomic to avoid normalizations and loosing information). And try to get as much as possible. As a guide, if you want to have a representative assembly of a genome, you need a minimum of 20X coverage with a 2x150 pair end library (read average 300 b). If an average bacterial genome is 4 Mbp, so you may need 80 Mpb of information for that single genome. Now, assuming that in soil community you have around 5000 different species, and 4 or 5 are above 1% in that community, to have a, let's say 20X representation of one single genome of one of the abundant species in the community, you may need a metagenomic dataset of 8000 Mbp (8Gbp, a full MiSeq flow cell). In the case of metatranscriptome is more difficult to do calculations as there the expression would be predominantly masked by rRNA over mRNA. And from the mRNA derived sequences (after RT PCR sequencing), can be classified with predicted function or mapped to the metagenomic dataset. 

You can use the R package randomForest by Liaw and Wiener.

Hi There,

Thanks for your responses (Rana & Sarah). I am planning to communicate both bacteria and fungi (NGS works) at the same time two different journals. The physioc-chemical data is common for both. Is there any other option ? If not, I have to publish one paper and cite the physico-chemical information in the second paper. 

Dinesh

Sorry that I answer this query so late as I just saw it. 

The R code for Chiu and Chao (2014) has been uploaded in my website:

Thanks for your interest in my work. 

Please go through these papers.

Hi Sabrhina,

It is possible the papers you refer to may have been talking about ecologically relevant traits in the sense of ones that can clearly be related to aspects of ecology and life history (also called 'functional traits'). Examples would include (for plants) traits such as leaf mass per unit area or seed size (cf. the paper by Diaz et al. for how only six such traits were used to explain a 'global spectrum of plant form and function'). 'Ecologically irrelevant traits', as you put it, could be thought of as traits for which either an ecological function cannot easily or clearly be ascribed, or where the relevance of the trait to ecology and life history is indirect (i.e. manifest via one or more other traits) or is simply unknown - there are an enormous number of these (cf. for example the trait table on the TRY Plant Trait Database  https://www.try-db.org/de/TabDetails.php; cf. also http://traitnet.ecoinformatics.org/traits-and-protocols). An alternative way of thinking about 'ecologically irrelevant traits' (or 'non-functional traits') is that they are neutral in an evolutionary sense, i.e. have no adaptive value. Hope this helps. Cheers, Matt

Essentially all of the lineages with C4 photosynthesis have been identified. So, if you have the species names and know their families, etc., you can look up whether they are likely to be C4 species or not. See the paper by R. F. Sage, 2016, Journal of Experimental Botany 67: 4039-4056 which provides a wonderful outline of these relationships and identifies C4 lineages. If you do not have this journal in your library, the author's contact email is r.sage@utoronto.ca and perhaps you can ask for a pdf of the paper.

See YouTube: SPSS for newbies: Changing a scale/continuous variable to a categorical variable

Yes, you can divide your dataset into partitions and apply different sequence evolution models to those partitions. A common framework might first involve using PartitionFinder (http://www.robertlanfear.com/partitionfinder/) to delineate partitions and their respective "best" models.

Sounds like you are using Mr. Bayes? You would then define your partitions in the Mr. Bayes command block of your nexus file.

For some help in configuring your nexus file for a partitioned dataset, see: http://mrbayes.sourceforge.net/wiki/index.php/Analyzing_a_Partitioned_Data_Set

Phylogeny of R-project is introduced in http://www.phytools.org/eqg/Exercise_3.2/

If you are asking whether percent cover (by species) is an appropriate abundance measure, then Yes.

Most bacteria experience at least some horizontal gene transfer, usually between relatively related species (Salmonella-Shigell-Escherichia for example) but also between quite distantly related organisms.  If you choose "housekeeping" genes sich as robosomal RNA genes, ribosomal protein gene, DNA Polymerase and so on, the gene trees usually agree with the majority of the genome.  But as Alex noted, many other genes are very often transferred between species.

Dear Mekdes Ourge,

The outcome of an nMDS analysis depends on the similarity/distance index you choose. The choice, in turn, depends a lot on the data you have. Generally, abundance data is more informative because it contains more information than just presence or absence. Apart from that, indices are designed to specific needs; e.g., you might want to use an index applying data normalization when having different data types or large ranges. A frequently used index for abundance data in ecological studies is the Bray-Curtis similarity index.

Alternatively, you could also run a Correspondence Analysis (or a Detrended CA), which is specifically designed for investigating species composition vs. environmental data.

Best regards,

Thomas

It may be better to perform stratified sampling. A stratified sample is a mini-reproduction of the population. Before sampling, the population is divided into characteristics of importance for the research. For example, by main species type. Then the population is randomly sampled within each category or stratum. Random sampling has a very precise meaning in that each community has an equal probability of selection, which it may in fact not have.

Since communities may be of different sizes, perhaps you should compare composition by percentage, for example using Shannon's or Simpson's diversity indices.

See: On sampling procedures in population and community ecology, Vegetatio 83: 195-207, 1989.

I hope this helps :-) 

 

If you install library vegan (including its dependencies), run the command library(vegan), and then load the data set using data=read.csv(directory,header=FALSE), you can analyze it.

I just experimented with it and using the function adipart(data,"shannon") produced the exact same result as the Excel sheet I referenced before on the same data.

I also spoofed it with a 50 x 312 data set and it functioned (I cannot verify the accuracy of its measurements mind you, but it produces alpha, beta, and gamma diversities with statistical significance).  Given that it worked for the smaller set, it should work for the larger one.

Hope that helps.

Can we achieve higher levels of emotional intelligence through metacognitive practices?

Yes. Not only that; emotional intelligence as identified by H. Gardner can ONLY be achieved through a meta-cognitive process (reading both "my emotion" and that of "others".

I suggest Non Metric Multidimensional scaling provides a better interpretation

Hi Amalia,

You can separate rRNA and mRNA reads in silico using sortmeRNA or another tool to align the reads against a rRNA database (e.g., Silva). Depending on your community and its activity the ratio rRNA/mRNA can be really unbalanced and it is not rare the 90% of the reads are rRNA reads.

Hopefully that’s helpful.

Cheers,

Flo

Dear KUMAR R.

Your explanation results is small. Also it is difficult for others who have not been associated in planning, execution and data collection processes. I t will be justified if you could discuss with your supervisor and senior colleagues for help.

in R there is also ade4 in which RLQ is implemented. This method  is useful in functional analysis as well

Hello Reza,

I strongly recommend you reading this article:

Mouchet, M. A., S. Villéger, N. W. H. Mason, and D. Mouillot. 2010. Functional diversity measures: an overview of their redundancy and their ability to discriminate community assembly rules. Functional Ecology 24:867-876.

One way to calculate the functional redundancy is to apply the MFAD, but I  recommend seeing some "problems" with this metric in the article quoted above. MFAD = Modified Functional Attribute Diversity (Schmera, Eros & Podani 2009).

Schmera, D., Eros, T. & Podani, J. (2009) A measure for assessing functional diversityinecologicalcommunities.AquaticEcology,43,157–167.

In the R-program exist several packages that calculate the functional diversity in its different facets, such as, FD and SYNCSA packages.

Best regards

Also see this study on global tropical reef fishes: http://www.pnas.org/content/111/38/13757.short

Dear Jorge,

If your main interest is in the trait differences between habitats, then you have two additional options:

(1) Treat all traits as a matrix of response variables (multivariate normally distributed) and enter them in a multivariate linear model e.g. using lm() followed by Anova() in John Fox´s "car" library:

model1=cbind(trait1,trait2,trait3...)~habitat

Anova(model1)

You can set up contrast matrices for the response matrix to test individual differences and hypotheses on sets of traits. You may also introduce random effects if desired.

See an example provided in the link below.

(2) Alternatively, you could just look at individual traits or groups of traits in structural equation models, e.g. using the lavaan or piecewiseSEM packages.

I hope these suggestions are helpful in some way for you - don´t know if they work but maybe worth a try.

Best wishes,

Christoph

Dan

Check out this paper. I'm not if it is of help to you.

Anesh

Hi Santiago: You could try with the R package 'spatstat'. See the code below for an example. Hope this helps :)

Hola Santiago: Echa un vistazo al paquete en R 'spatstat'.  Mira el codigo debajo. Espero te sea de ayuda.

library("spatstat")

library("VGAM")

n <- 100 # number of row and columns in the grid

# For instance assuming an overdispersed model. Using Poisson-lognormal distribution.

X <- rpolono(n , meanlog =-1, sdlog = 1) # generating random number from Poisson-lognormal distribution

# If your model is not overdispersed then replace the above line with X <- rpois(n, lambda=1) for example or use another distribution.

par(mfrow=c(sqrt(n),sqrt(n)),mar = c(0,0,0,0),oma = c(0,0,0,0), mai = c(0,0,0,0))

for(i in 1:n){

pp <- runifpoint(X[i])

plot(pp, main="")

}

This is a realistic problem. How many values do you have per measurement?

Question: Do you have many data with many repetitions? how many?   n

If yes, then you make first the 3 averages.

This gives you one point in a 3D Coordinate system. number your samples from 1 ...to n

Then you plot ALL points in the three coordinate systems: X Y Z as:

Y versus X, then Z versus X, then Y versus Z. Plot all measurements.

May be you get distinct groups in the 2D planes or in the 3D space. It's just a question of habit to plot your points in a three dimensional space, like in a cube with the coordinates X Y Z.. It may be helpful to normalize your experimental data as deviation of the average values. The diversity you just are numbering from 1 to 2 to 3 ... Invent a smart presentation for your experimental data.

In Science you are allowed to do what ever you like....but: You have to explain what you did....

Good luck, Reto

You always can contact me : StrasserReto @Bluewin.ch     

I would measure the tissue flammability. See more at Cornelissen J, Lavorel, Garnier, Díaz, Buchmann, Gurvich, Reich, ter Steege, Morgan, van der Heijden M, et al. 2003. A handbook of protocols for standardised and easy measurement of plant functional traits worldwide. Australian Journal of Botany 51: 335.

See attached.  Hurlbert 1971,  This is a good starting point for ecologists.  Many papers since!

English would be appreciated, but French and Spanish are both Ok...

Dear Rama

There is considerable literature regarding the approaches of use in the preparation of future health professionals to work more effectively with Indigenous peoples. You may find some details of interest that are translatable to education of future teachers. You could look at the Leaders in Indigenous Medical Education web site:

Particularly the resources / publication section.

The PhDs of two colleagues could be of interest here too:

Shaun Ewen 2011 Cultural competence in medical education: a university case study available at https://minerva-access.unimelb.edu.au/handle/11343/36708

Suzanne Pitama (2013). “As natural as learning pathology”: The design, implementation and impact of indigenous health curricula within medical schools (PhD). University of Otago, Christchurch, New Zealand. 

Hope this helps, Dave

Dear Klara, it is important to clarify your question:

You could compute functional diversity of each community at time point A and compare to functional diversity of the same communities at time point B. For this I suggest using Rao entropy, for which you will need to compute a dissimilarity matrix between species based on their traits (if the traits are of mixed type, the Gower index may be useful). In this case you can only tell about the temporal change in the overall functional diversity of each community, and nothing about the functional identity of the community components. This is a limitation analogous to when communities are compared by their species diversity (e.g., by using Shannon diversity). 

Another option is to compare the communities based on their functional composition, for which you may consider the definition of fuzzy-weighted community composition. This is described in Pillar et al. (2009, http://dx.doi.org/10.1111/j.1654-1103.2009.05666.x). See also Pillar & Duarte (2010, http://dx.doi.org/10.1111/j.1461-0248.2010.01456.x) in the context of phylogenetic analysis. The fuzzy-weighting requires as input a similarity matrix between species based on their traits (which could be the above mentioned Gower index), which after proper standardisation to unit column will define a fuzzy set matrix U of species by species. The fuzzy-weighted community composition is given by matrix X obtained by matrix multiplication (X = UW), where W is the matrix of species composition in the communities.

The patterns of diversity depends on what your focus is in the study of diversity. There are various levels of diversity study: habitat diversity, species diversity and genetic diversity. The determinants depend on what level you want to focus. If in a forest as an ecosystem, I guess you wanted to see the patterns of a community which generally use the index of diversity, index of dominance, index of evenness and the species distribution as measures of how diverse the ecosystem is. If so, you still need to look at habitat fragmentation, habitat quality and habitat preference of the key species in the forest. 

Thank you all for your suggestions!

Best regards,

Thanasis

Indices are quite similar to those for other kinds of diversity, but the selection of the traits that you will consider in the functional characterization of your communities is indeed the complex issue here...

Thank you very much for your answer, Thiago,

I will read these paperes as well.

Best wishes,

Mika

Using the term 'validation' here does not appear appropriate. Bootstrap, that you mention, will provide you a measure of confidence in your clustering results. See if Chapters 5 and 6 of these notes may possibly help to get some idea

The only way to handle this is to know how the spatial correlation is distributed BEFORE setting up the sampling protocol to guarantee that observations will be more or less spatially independent. Spatial autocorrelation statistics measure and analyze the degree of dependency among observations in a geographic space. Classic spatial autocorrelation statistics include Moran's I, Geary's C, Getis's G and the standard deviational ellipse. These statistics require measuring a spatial weights matrix that reflects the intensity of the geographic relationship between observations in a neighborhood, e.g., the distances between neighbors, the lengths of shared border, or whether they fall into a specified directional class.The Mantel and partial Mantel tests can be flawed in the presence of spatial auto-correlation and return erroneously low p-values See e.g. Guillot and Rousset, 2013.

For the new nosZ group, two related publications:

In Nature Climate change

in ISME J