Canoco 5 Gallery

Creating new analyses

Many analyses that were very complex to setup and interpret in the old versions of Canoco (4.x) are now easily accessible in Canoco 5. The Canoco Adviser uses an expert system technology to match your data properties with appropriate analyses. This is a two-stage process illustrated below.

First snapshot shows the New Analysis Wizard at the point you have already chosen the data tables of your project that you want to use and after the expert system generated matching analysis templates. Offered analyses are divided into separate sections (only part of them is visible in the snapshot) and as you select particular analysis, more detailed description is provided in the lower part of the page (introduced by the words In the selected analysis, you would).

After you select particular analysis template, new analysis is created and you can specify its details in an Analysis Setup Wizard sequence. Good news for the Canoco 4.x users is that this sequence is more compact than in the older versions and also that the Canoco Adviser provides valuable advice, as illustrated in the following snapshot, where the unimodal method is recommended. You can also note how easy is to calculate more than four ordination axes for the basic analyses. The analyses in Canoco 5 may contain multiple steps so that e.g. variation partitioning with three groups of explanatory variables is still a single analysis.

Finally, Canoco5 also offers you – after the analysis is executed – valuable suggestions for visualizing your results using ordination diagrams and attribute plots. After you select the type of graphs you want to be created, contents of each graph can be then further specified within a page similar to the following one (for a graph visualizing the variation of species diversity with the values of explanatory variables).

Analysis execution proceeds mostly silently, but the interactive stepwise selection already present in Canoco for Windows 4.x was visually enhanced as illustrated in the following snapshot.

Worth of mentioning is the visual presentation of the proportion of variation explained by selected predictors (the cyan-and-red bar in the dialog centre), the colour coding of candidate predictors based on the significance value, and the three available methods of p value adjustment: beside the selected false discovery rate, the Holm and Bonferroni corrections are also offered.