Page 13 - HIMES_2007_

Performance Indicator Importance in MPA Management 613

EIMR. For tourists and researchers, group coherence is higher. These groups could have
Downloaded By: [University of Leeds] At: 13:49 18 January 2008
more coherent opinions because of their origin outside of the EIMR and interests being
centered on either the recreation or research value of the MPA. The most surprising result is
the coherence level of the MPA managers interviewed. Manager coherence was found to be
the lowest of all stakeholder groups. Although this should be expected, as both current and
previous local government officials were interviewed, this is concerning as it underlines
the fact that depending on who is in charge, the management priorities could be extremely
different, thus affecting the consistency of management priorities significantly over time.

Understanding Variability through Cluster Analysis
The next step in understanding this lack of group coherence is to consider the differences
and similarities among individuals both within and between stakeholder groups. This was
accomplished with a hierarchical cluster analysis measuring Euclidean distance between
the preferences elicited in the AHP survey. Cluster analysis is used to classify objects (or
stakeholders in this case) into different groups, or more precisely, to partition a data set
into subsets (clusters), so that the data in each subset (ideally) share some common trait.
Cluster analysis in this study allowed for insight into the apparent lack of coherence among
individuals in stakeholder groups. Figure 5 summarizes the results of the cluster analysis,
where each individual is identified by stakeholder group and an identification number. Six
different clusters are identified, labeled A through F, in Figure 5.
A first glance at the composition of each cluster shows that the clusters are not
homogeneous with regards to the a priori identified stakeholder groups. However, an
analysis of cluster composition in Figure 5 shows a more homogeneous relationship based
on other characteristics. Information from interviews done in other studies conducted by
the author (Himes, 2003; 2005; 2007) was available that identifies the demographics of
each respondent. This information showed an interesting pattern in the clusters. Cluster
A consists of managers that have completely different priorities from the other clusters.
Cluster B represents a local resident that spends half of his time on Favignana and half
in Palermo and thus perhaps views the MPA with different eyes than a full-time resident.
Cluster C is comprised mainly of residents who have been highly involved and vocal in
EIMR management. Cluster D represents individuals whose main interest lies in promoting
tourism and fishers’ well-being, hence the presence of one manager (politically elected),
a fisher cooperative representative, and fishers that have taken an active role in promoting
pescaturismo. Cluster E contains individuals who are more conservation minded: this
includes almost all researchers, more educated residents, and environmentally aware fishers.
Not surprisingly, this cluster also included the recently dismissed ex-director of the EIMR
who is trained as a biologist.
Finally, cluster F represented individuals whose main interests involve the well-being of
the tourism industry. This accounts for all but two tourists and the majority of local residents
that run tourism businesses as well as young residents whose families earn income from
tourism activities. As seen in Table 4, group coherence greatly improves when respondents
are reorganized into these new clusters (i.e., average 0.940).
Although a priori assignment of stakeholder groups has shown to be possible in fisheries
studies (e.g., Mardle et al., 2004), the large variation and low group coherence in responses
that have resulted from the present study begs the question of why individual stakeholders
in the EIMR do not seem to form coherent groups based on self-identified stakeholder
groups. Three points should be made regarding the lack of group homogeneity among a
priori identified groups.