you can try with rda (Redundancy analysis), you can assess

relationships between the phytoplankton community structure and environmental variables. i also use partial triadic analysis for find spatial-temporal structures (in phytoplankton community and environmental variables) in three dimensional database (Sampling stations-Date-Community or variables), and coinertia analysis

I'd try a simple principal component analysis to see where things 'shake out.'

I think that the best is the test of ANOVA, MANOVA and ANCOVA.

But if your data are not folow a normal distribution you must folow the non parametric statistics.

There are many possible analyses, and as stated above, I think RDA is a good one to summarize relationships. Yet, I would advise you to start plotting yout time series and then performing PCA for exploration of what you have in hand. Depending on what you observe and what your question is, you may need to detrend or deseasonalize your datasets before performing a multidimensional analysis.

CCA, wtih Cannon software!!

your questions sound like factor analysis in each block conditions (that is the graphical resolution of a PCA)

...A way to made more clear the analysis you can use a standard anomalies by each variable...if you have several several sampling trip with a different observations numbers you can calculated a Frankenstein Variable (With a empirical orthogonal function)

Please take with some considerations the anovas manovas ancovas and mancovas tests because the phytoplankton abundance are in nominal scale and this analysis need minimum a variables interval scale. also Gaussian distribution and homocelasticity can be a proof to use this test

None until you just put your data in a graph. If you do not see nothing...is because there is nothing (I am old fashioned scientist)

I would recommend you to use a PCA analysis too, so you can identify which are variables are related etc. But if you want to assess the temporal variability such as seasonal or interannual variability probably you will need to decompose your variables in different componnets...seasonal, interannual trend, white noise etc. You can read more about this in the following paper:

Nogueira, E., Pe´rez, P. P. and Rı´os, A. F. (1997) Seasonal patterns and

long-term trends in an estuarine upwelling ecosystem (Rı´a de Vigo,

NW Spain). Estuarine Coastal Shelf Sci., 44, 285–300

It sounds like a transfer function problem used in palaeoecology. You can produce your own function (weighted in diffrent ways) using your assemblage and physico chemical data. In this way you can predict the reverse i.e the physico-chemical data from the measure (or preserved) assemblage. I suggest Down Load PAST (free) with its easy to understand Pdf explanation with examples togehter with all of the above suggestions.

It could be appropiate to use the STATICO analysis in assessing the temporal stability of species–environment relationships. Some articles you can review:

SIMULTANEOUS ANALYSIS OF A SEQUENCE OF PAIRED

ECOLOGICAL TABLES. JEAN THIOULOUSE,1,3 MONIQUE SIMIER,2 AND DANIEL CHESSEL1 Ecology, 85(1), 2004, pp. 272–283

as the paper presenting the tecnique,

and here you have some applications to similar data:

http://dx.doi.org/10.1016/j.jembe.2011.05.015

http://dx.doi.org/10.1214/10-AOAS372

http://dx.doi.org/10.1016/j.ecss.2011.11.018

DOI: 10.1080/02664763.2011.634393

It is not a simple question. I will recommend you to have a deep read of the book: Numerical Ecology by Legendre and Legendre. You will find many directions to go though depending on yout data and goals.

If the study is only on a temporal gradient then definitely as many suggested PCA is good but if there is some spatial aspect also then I would also perform the ANOVA, and MANOVA.

Whether spatial or temporal, PCA is still suggested. It depends on software being used and the required arrangement of data according to spatial categories and then according to temporal categories.

MANOVA will not highlight specific differences across the intervals being compared.

the evaluation of a long-term data collection has some difficulties. This is my own experience. When I started such an evaluation some years ago I did not know what kind of objections I would meet. The first presentation of the results from a small lake in a meeting was frustrating: no question from the audience. The exchange started after I submitted a paper with these data and the reviewers recommended me to calculate the biomass. Since then the paper is not yet accepted for printing, but I see that each submit is an opportunity to receive more suggestions. I am in need for this exchange because of my solitary existence as a freelancer. So, I went through a kind of PCA, computing regressions with Excel in order to look for correlations between physical and chemical parameters as determinants and the biological features like size and biomass as variables. This means the part on bottom-up forces. But I had also some data about fish stocking and I found them suitable to cover the part of top-down forces. I think that someone should be concerned about predators and disease agents like viruses, bacteria or fungi may act in the system and that positive or negative feedbacks are possible.

Most important was the exchange with other people. Once, I received a question and did more work with excel resulting in a new equation. I could see the result as a good solution of the dreadful, distracting and confusing data. Frequently, young people give me valuable suggestions for the work with Excel.

I hoped for more exchange and wrote a posting in this forum and forgot about. Twenty two months later, John B G from Tasmania answered and suggested to use the PAST as free software. It comprises a collection of several statistical methods. I had the feeling that I was not alone in the world, from the other end of earth valuable suggestions were sent to me. PAST has also the cluster analysis with different filters and I could corroborate the finding that there is at least one more parameter given by the season. But I am aware that a major difficulty is the arrangement and transformation of the biological data. There are successions of species, size classes, and biomass volumes depending on the acting forces. Reviewers got angry about my results and claimed: too many eliminations. At the end, I approached my contractee, asked for the permission to publish the original data on a website and I got it: http://www.phosph-frctns.de/html/buchtzig.html

These are my personal experiences. I think that you and other people have to find their own way of work in such evaluations. A major incentive to me was the word of a reviewer: “You have an analytical paper.” Data from long-term investigations are an occasion for system analysis. I think so.

Next time, I will go to a meeting again and present my data.

I wish you good luck and success in your evaluations.

Regards,

Albert