Workshop on the World Ocean Assessment

The important point about the cluster analysis is that the differences being displayed are the sum- marised differences relative to the differences be- tween all the other parameters. This helps to avoid what might be a small relative difference for a small number of parameters being prioritised as impor- tant, when there are other parameters that may be also as (or more) important but not recognised as such because they are measured or reported using different indicators or in a different way. To guide assessment, the cluster analysis is further summarised in a ‘heat map’ diagram. This graphic (Figure 12) depicts the extent to which the groups in the cluster dendrogram are different from each other. The higher differences identify greater rela- tive divergence in the patterns of information, and indicate which groups may be worthy of more de- tailed discussion or investigation. The highest differ- ences in the heat map are linked to Groups 6, 7 and 8 of the cluster. Classification Groups 6, 7 and 8 consist of 22 param- eters: 14 species groups, five physical or chemical processes, and one each of habitat; pests, diseases; and pressure parameters (Table 1). These param- eters have high average levels of condition (Fig- ure 13), and most of the parameters in Most and Best10% places are either stable or increasing (Fig- ure 15), assigned with medium to high confidence (Figures 14, 16).

underlying statistical distributions, and uses only a simple set of well-tested non-parametric statistical tools, available free (or at low cost) in the public domain. The approach used here is cluster analy- sis, which classifies the parameters into coherent groups of parameters with similar information con- tent across all eight of the indicators scored/graded for each parameter. The information pattern for the data provided by the experts for the 104 parameters that were scored at the workshop is shown in the classifica- tion dendrogram (Figure 11). The eight groups of parameters shown in the dendrogram each have unique patterns in condition, trends, confidence and information base, and some examples are dis- cussed below.

12

Figure 11. Classification (average linkage) of scores as- signed at the workshop, resolving the 104 parameters into 8 groups of parameters that share similar character- istics as defined by the scores/grades.

Figure 12. Heat map symmetrical matrix of groups from the classification (Figure 11); the dark blue cells repre- sent lowest difference in information content, red cells represent the highest level of difference in information content. The greatest differences are demonstrated by groups 6, 7 and 8.