Sr. Schmolke et al., Multivariate statistical approach to the temporal and spatial patterns of selected bioindicators observed in the North Sea during the years 1995-1997, HELG MAR R, 53(3-4), 1999, pp. 257-266
A comprehensive database, containing biological and chemical information, c
ollected in the frame work of the bilateral interdisciplinary MARS project
("biological indicators of natural and man-made changes in marine and coast
al waters") during the years 1995-1997 in the coastal environment of the No
rth Sea, was subjected to a multivariate statistical evaluation. The MARS p
roject was designated to combine a variety of approaches and to develop a s
et of methods for the employment of biological indicators in pollution moni
toring and environmental quality assessment. In total, nine ship cruises to
four coastal sampling sites were conducted; 765 fish and 384 mussel sample
s were analysed for biological and chemical parameters. Additional informat
ion on the chemical background at the sampling sites was derived from sedim
ent samples, collected at each of the four sampling sites. Based on the ava
ilable chemical data in sediments and black mussel (Mytilus edulis) a pollu
tion gradient between the selected sites, was established. The chemical bod
y burden of flounder (Platichthys flesus) from these sites, though, did not
reflect this gradient equally clear. In contrast, the biological informati
on derived from measurements in fish samples displayed significant a region
al as well as a temporal pattern. A multivariate bioindicator data matrix w
as evaluated employing a factor analysis model to identify relations betwee
n selected biological indicators, and to improve the understanding of a reg
ional and temporal component in the parameter response. In a second approac
h, applying the k-means algorithm on the data matrix, two significantly dif
ferent clusters of samples, characterised by the current health status of t
he fish, were extracted. Using this classification a temporal, and in the s
econd order, a less pronounced spatial effect was evident. Tn particular, d
uring July 1996, a clear sign of deteriorating environmental conditions was
extracted from the biological data matrix.