A total of 167 trout were obtained from the three
streams. From these, a total of 2546 worms belonging to
eight species were recovered: six acanthocephalan
species, one cestode speciesand one nematode species
(table 1). Only four helminth species were common
enough to allow analyses of their co-occurrences and
interspecific associations: Pomphorhynchus laevis, Acanthocephalusanguillae, Echinorhynchus truttae and Cyathocephalus
truncatus. Their intensity of infection varied
significantly among the three streams (Kruskal-Wallis
test: P. laevis, H . 65:52; P . 0:0001; A.anguillae, H .
12:25; P . 0:0022; E. truttae, H . 34:97; P . 0:0001; C.
truncatus, H . 6:63; P . 0:0364.: The mean length of fish
also varied among the three streams (ANOVA: F2;164 .
37:403; P .0:0001.; with fish from the Grimana Nuova
stream being smaller than fish from the other two
streams (table 1).
The observed frequencies of co-occurrence of the four
main helminth species did notdeviate significantly from
those predicted by a null model based on random
occurrences derived from prevalence data (fig. 1). There
was a weak tendency for fewer fish harbouring single
species thanexpected, and more uninfected fish than
expected, but these were not statistically significant (fig.
Numbers of E. truttae per fish were significantly
correlated with fish length in one of thethree streams;
none of the other correlations proved to be significant at
a . 0:05 (table 2). Of the 12 correlation coefficients
computed, six were negative and six were positive.
Fish length alsodid not correlate with either species
richness, i.e. the number of different helminth species per
fish, or total helminth abundance, in any of the three
study streams (table 2).
Of the 17 pairwisecorrelations between the intensity of
infection by two helminth species that could be computed,
16 were negative; of these, only four were
statistically significant at a . 0:05 (table 3). The...