Page 4 - Russo_Terlizzi_1998

Giovanni Fulvio Russo & Antonio Terlizzi

of other sites ; station
100% points have saturations
which follow one anoth-
80% ~ v ~ er consistent with their
~ Marettimo were ordered
geographic al position:
60% l S. Domino and Gal-
lipoli, Lampedusa and
respectively, proceeding
40% ~ v from the positive to the
negative portion of the
20% factor, which co uld
therefore be imerpreted
as an East-W est longiru-
0%
S. Dommo Gal h poli Lampedusa Marettimo Medas dinal gradient. Along
F2, apart from th e
D air-lift exclusive D Common species D hand-net exclusive strong polarisation of the
samples collected by air
lift a t Galli poli, a clear
Figure 4. Percent of species collected by both che methods and exclusively by each single method. separation of the samples
according to the collect-
ing method was obtained:
the samples, collected by both a1r lift and hand-towed net. ali the samples by air lift had positive scores, while all those by
Alvania discors, Vitreolina philippi and Gibbemla philippii were hand-towed net were in the negative part of the factor.
present in all air lift samples, while Jujubinus striatus and Bit-
tiltm retimlatmn were presem in ali hand-towed net samples. The DISCUSSION
bivalve species, almost exclusively collected by air lift, were Remarkable differences in the strucrure of the mollusc associa-
well represented, in quamitative terms, in ali the stations and tions living in the fìve seagrass beds of the Mediterranean were
especially in Medas. observed. This high structural variability may be imerpreted as
Each prairie was well characterised by the high quamitative the result of a number of imeracting factors.
dominance of particular species: S. Domino by Rissoella inflata; First of all, differently-strucrured malacological assemblages
Gallipoli by Bittium retimlatum, Bittium latreillii, Nassarius were obtained by the two sampling methods, as clearly shown
incrassatus, Calliostoma conulum, }ujubinus striat11s and Venericardia by the ordination model, in which along F2 the samples by net
antiquata; Lampedusa by Rissoa auriscalpimn, Tricolia pullm and have an opposi te polarisation with respect to those by air-lift.
Aft,ania lineata; Marettimo by Rissoa variabilis, Alvania discors The material obtained by the two sampling tools was differ-
and Vitreolina philippi end Medas by Bittium retimlatum, Nassar- em in both qualitative and quantitative terms. Despite the fact
im incrassatm, Gibbemla philippii and Mmmlm costulat11s. that the hand-net covers a larger sampling area (about 20 m2
compared to l m 2) and collects many more individuals, the suc-
Structural analysis tion-sampler gives more complete information on the overall
A fìrst strucrural analysis was carried out starting from a 82 x species composition living in a particular prairie stand. Samples
10 matrix (species x (stations x methods), mathematically treat- collected by air-lift are richer in species and more variable, the
ed to obtain a CA ordination model. The model showed, along appararus being more effective at the base of the leaves, where
the first factor, a strong opposition between the samples from the malacological stock is richer in species and reflects the vari-
the Medas (both hand-net and air lift) and ali other samples ety of the substrate. Conversely, samples collected by hand-net
and, along the second factor, the opposition of the samples of S. are less rich in species and less variable, this tool being more
Domino to ali others (both factors were signifìcam). effective on the leaves which are inhabited by a more homoge-
A second 54 x 9 matrix (species x samples) was then set up neous malacolog ical stock . Therefore, the suction-sampler
by erasing the samples of the Medas, in order to eliminate dis- would be preferable when the qualitative aspects are to be
tortion in the first factor of the model. In the new CA ordina- emphasized with respect to the quantitative ones, as is the case
tion obtained from this matrix (Fig. 7) the fìrst two factors were for collections for strucrural analyses of communities; the hand-
significant an d showed, along F l, the sa me pattern prese m net may be more useful when the quamitative aspects prevail
along F2 in the previous model, that is, the opposition of the over the qualitative ones, as in the case of population analyses.
samples from S. Domino to ali others. In addition, some other Secondly, in the large-scale strucrural comparison of the
imeresting ordination patterns can be observed in this latter associations from the five Posidonia beds, both sampling meth-
model. Along Fl, the two sample poims (air-lift and hand-net) ods were effective and gave similar information. This is clearly
of the same si te have a saturation which is more similar to that shown by the AC (Fl /F2) ordination model, where the intra-site

>92<