CHAPTER 35
Comparison between Modern Clustering Techniques and Original
Macrobenthic Community Classification on A. Vatova's Adriatic
Data Set (1934-36)
P. Di Datol, E. FresiI, and M. ScardP
ABSTRACT
We carried out a comparison between the results of several clustering procedures and the
zoocoenoses defined by A. Vatova after his studies on the Adriatic macrozoobenthos (1934-36).
The original data set was analysed by means of hierarchical and non-hierarchical clustering
algorithms, with and without contiguity constraint, as well as with different combinations of distance/dissimilarity indexes and data normalization. None of the partitions obtained from these
procedures matched satisfactorily the original classification, which had probably been based on
information that is not available in the numerical abundance data reported by the Author.
Therefore, we recommend that only binary information from this data set should be taken into
account for further analyses involving recent data sets.
Introduction
The most comprehensive study on the benthic
fauna of the Adriatic Sea is certainly the one carried out from 1934 to 1936 by Aristocle Vatova,
an Italian marine biologist.
During four cruises he collected more than
four hundred macrozoobenthos samples in 390
stations, both in the central and in the northern
part of the Adriatic basin, using Petersen grabs
(0.2 or 0.1 m 2 ). Almost three hundred species
were identified and their abundance was reported both as number of individuals and as biomass. Qualitative information about the characteristics of the sea floor in each station was also
reported (Vatova 1949).
On the basis of his data, A. Vatova defmed
fourteen different zoocoenoses. He classified the
stations mainly according to a dominance criterion and named the zoocoenoses after their
dominant species. This kind of classification,
inherently subjective. was performed in order to
outline large areas where the composition of the
benthic fauna showed similar features.
The Author's concept of zoocoenosis was
close to the one expressed by Clements and
Shelford (1939), since he considered zoocoenoses
as real and very sensitive entities, which were
supposed to adapt to environmental changes by
modifying their species composition.
The present paper provides a methodological framework for the exploitation of Vatova's
data in view of future comparisons, with the
objective to know:
1. If and how these data can be used as an historical reference for more recent data set.
2. Which technique provides results that more
closely approximate Vatova's point of view in
the description of the community.
In order to verify to what extent Vatova's
classification was supported by the macrozoobenthos data, we compared it with the
results of modern clustering techniques. which
should provide, from a theoretical point of view,
a more objective way to describe the spatial heterogeneity of the macrozoobenthic communities than as recorded more than 60 years ago. Of
course, the purpose of our work is not to criticize Vatova's subjective approach to the classification of zoocoenoses. On the contrary, we
think that Vatova's classification summarized
valuable information and we want to check the
possibility of comparing it to the classifications
that can be obtained by means of the objective
1 Dipartimento ill Biologia, UniversitA ill Roma "Tor Vergata", Via della Ricerca ScientifIca, 00133 Roma, Italy
2 Dipartimento ill Zoologia, Universita di Bari, Via Orabona 4J A, 70125 Bari. Italy
EM. Faranda" L. Guglielmo, G. Spezie (eds)
Mediterranean Ecosystems: Structures and Pmce88eS
© Springer-Verlag Italia 200 1
Comparison between Modern Clustering Techniques and Original
Macrobenthic Community Classification on A. Vatova's Adriatic
Data Set (1934-36)
P. Di Datol, E. FresiI, and M. ScardP
ABSTRACT
We carried out a comparison between the results of several clustering procedures and the
zoocoenoses defined by A. Vatova after his studies on the Adriatic macrozoobenthos (1934-36).
The original data set was analysed by means of hierarchical and non-hierarchical clustering
algorithms, with and without contiguity constraint, as well as with different combinations of distance/dissimilarity indexes and data normalization. None of the partitions obtained from these
procedures matched satisfactorily the original classification, which had probably been based on
information that is not available in the numerical abundance data reported by the Author.
Therefore, we recommend that only binary information from this data set should be taken into
account for further analyses involving recent data sets.
Introduction
The most comprehensive study on the benthic
fauna of the Adriatic Sea is certainly the one carried out from 1934 to 1936 by Aristocle Vatova,
an Italian marine biologist.
During four cruises he collected more than
four hundred macrozoobenthos samples in 390
stations, both in the central and in the northern
part of the Adriatic basin, using Petersen grabs
(0.2 or 0.1 m 2 ). Almost three hundred species
were identified and their abundance was reported both as number of individuals and as biomass. Qualitative information about the characteristics of the sea floor in each station was also
reported (Vatova 1949).
On the basis of his data, A. Vatova defmed
fourteen different zoocoenoses. He classified the
stations mainly according to a dominance criterion and named the zoocoenoses after their
dominant species. This kind of classification,
inherently subjective. was performed in order to
outline large areas where the composition of the
benthic fauna showed similar features.
The Author's concept of zoocoenosis was
close to the one expressed by Clements and
Shelford (1939), since he considered zoocoenoses
as real and very sensitive entities, which were
supposed to adapt to environmental changes by
modifying their species composition.
The present paper provides a methodological framework for the exploitation of Vatova's
data in view of future comparisons, with the
objective to know:
1. If and how these data can be used as an historical reference for more recent data set.
2. Which technique provides results that more
closely approximate Vatova's point of view in
the description of the community.
In order to verify to what extent Vatova's
classification was supported by the macrozoobenthos data, we compared it with the
results of modern clustering techniques. which
should provide, from a theoretical point of view,
a more objective way to describe the spatial heterogeneity of the macrozoobenthic communities than as recorded more than 60 years ago. Of
course, the purpose of our work is not to criticize Vatova's subjective approach to the classification of zoocoenoses. On the contrary, we
think that Vatova's classification summarized
valuable information and we want to check the
possibility of comparing it to the classifications
that can be obtained by means of the objective
1 Dipartimento ill Biologia, UniversitA ill Roma "Tor Vergata", Via della Ricerca ScientifIca, 00133 Roma, Italy
2 Dipartimento ill Zoologia, Universita di Bari, Via Orabona 4J A, 70125 Bari. Italy
EM. Faranda" L. Guglielmo, G. Spezie (eds)
Mediterranean Ecosystems: Structures and Pmce88eS
© Springer-Verlag Italia 200 1