If your are interested in exploring changes in composition of the plant communities - then probably an ordination technique such as non-metric multidimensional scaling, or even principal components analysis would be useful for discerning trends.

As already suggested graphs are useful to look at trends in cover and height of the total community, functional/structural groups, and common species.  Once you have identified tendencies, you can perform further analyses to evaluate trends using regression or mixed models.

Hi

please check this paper.

http://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&uact=8&ved=0CBwQFjAA&url=http%3A%2F%2Fwww.unulrt.is%2Fstatic%2Ffellows%2Fdocument%2Frajabov-t.pdf&ei=4vTsU6OEMcyR0QWzp4GwDg&usg=AFQjCNGoQTd8zpf8T9f3a_aAYdHk2bXQTw&sig2=9qmTWC2rGrrfFHQ3Fx8KbA&bvm=bv.72938740,d.bGE

Hi, Candida, first, you have to find parameters you would like to compare. This is dependenht from the aim your investigation: is it practical (like productivity), or rather academic (like photosynthesis) etc.

Thereafter you have to construct algorithm of your calcuoation, with ranging of each parameters. Thereafter you build some basical graphs. Thereafter you will now basical tendencies over 3 years.

But first choose parametrs you plan to analysis. 

If your are interested in exploring changes in composition of the plant communities - then probably an ordination technique such as non-metric multidimensional scaling, or even principal components analysis would be useful for discerning trends.

As already suggested graphs are useful to look at trends in cover and height of the total community, functional/structural groups, and common species.  Once you have identified tendencies, you can perform further analyses to evaluate trends using regression or mixed models.

The graphs are definitely important for well defined parameters. For choosing the parameters the IVI as well as the dominant and co-dominant species and the regeneration status of the species along with their mean survival rate need to be worked out. The number of viable seedlings and saplings and the regeneration pattern of the species plays a significant role in predicting any possible change in the forest community.

As Philip says - PCA, DCA, CCA (for the latter you will need quantified environmental data such as climate, soil characeristics but time also can be used)

Hi Cándida, you could try a couple of other techniques that might give you a simpler answer to your question along with options to explore the detail of individual species’ responses. Can I suggest you investigate two analytical methods: multivariate control charts and/or Principal Response Curves. Both techniques allow you to compare whole (treatment) site behaviours (say your abandoned pasture and  recovered formation) to a control site (or set of sites) (your natural peatland).The methods use analyses that others here have described but take them one step further and allow you to compare whole site (multi species) trajectories to a reference (control) site(s) using all of the information contained in you multi-species multi-plot samples. Multivariate control charts (see Anderson and Thompson, 2004) uses bootstrapping of similarity measures to compare amongst sites. You could have three site types (control-natural peatland; and treatment - abandoned pasture and recovered formation) over three times (your 3 annual sampling occasions) and compare. From such a comparison you could graph ‘state’ (difference between control and treatment sites at any (of the 3) times, and transition (trend) as the treatment sites converge on or diverge from the control site(s). I use the optional plural site(s) here as it may be possible to partition your 250 plots into several replicates to increase your plot sample size (be careful here though to not include spatial autocorrelation into you measures). Which brings me to Principal Response Curves (see Van den Brink and ter Braak, 1999). This uses an ordination method to do a similar task – compare control sites to treatment sites through time. For this you will need replicates for each of your site sets (say 3 replicates of the natural peatland sites, 3 replicates of the abandoned pasture and 3 of the  recovered formation). The method does in addition allow you to tease out (and plot) the influence of individual species on the trajectories of the sites compared to the control. To get the replicates you could randomly or spatially partition your plots into 3 groups and call them replicates. Here you will need great care to not create autocorrelation – your sampling method may not allow you to do it but give it some thought and it may be possible. I won’t describe the detail of the methods – you can read the papers and see if they are right for your problem.

Regards

Along with the other answers given above, a simple measure of the change in the abundance of key species or functional groups will be useful. The multivariate methods can tell you by how much, and in what direction, you communities have changed, but you also want to show how they have changed - in other words, what are the key characteristics that have changed.

You need to look at seasonal climate (usually rainfall) over the three years that you studied, especially in relation to the long-term mean. You can use a regression approach using environmental gradients, where you include all the measured environmental variables and narrow them down to the most parsimonious using selection techniques like Aikaike's Information Criterion.

And remember that three years is just a snapshot in a much longer-term pattern. You probably want to focus more on treatment effects than on trends over time. 

Yes, definitely NMS works, but you might also think of using mixed models to analyze changes in abundance of individual species. See my attached paper for an example.

Article Fish Assemblage Response to a Small Dam Removal in the Eight...

As has been said above, tehre are many ways, depending on available data and objectives. Regular remeasurents in permanent observational studies are essential for understanding different kinds of changes.  I am attaching a file that provides some background

Hello Candida, yes with NMSD you can assume that sample separation in space is connected to ecological dissimilarity.

Also all the other methods are good. In general, however, I believe is better to keep it simple otherwise the conservation aim of your paper could be misunderstood.

Try also frequency distribution of life forms, corology, families, or Ellenberg indicators, maybe using two way kolmogorov smirnov. Simple packages such a past can solve the problems.

Cheers

Roberto

ps a similar study on:

https://www.researchgate.net/profile/Roberto_Crosti/publications?pubType=article&ev=prf_pubs_art

I'd suggest these readings.

Article Principal response curves technique for the analysis of mult...

Article A simple test for alternative states in ecological restorati...

It might also be useful to select some specific species know as indicators of site health in that area and apply the techniques suggested by Philip and Graham above. The kind of data collected might influence the selection analytical techniques (for example: composition by estimated weight, composition  by actual weight, composition by estimated cover, composition by measured cover (basal, areal, etc.)), presence/absence).

Actually I thought I already send a suggestion on this question...however.

You can transform in density data and do a univariate statistics (paired or unpaired according to your initial design).

NMDS is good because it assumes that sample separation in the space is directly related to ecological dissimilarity (then you have to assess which species effected more the results).

And to keep it simple just do difference in frequency distribution on the following parameters: families, life forms, home range-corology, Ellenberg indicators (if available) using two sample Kolmogorov-Smirnow.

See this study on vegetation changes

https://www.researchgate.net/publication/264081859_Changes_in_the_structure_and_species_dominance_in_vegetation_over_60_years_in_an_urban_bushland_remnant?ev=prf_pub

Cheers

Roberto

Article Changes in the structure and species dominance in vegetation...