|
Numerical
analysis of the relationships between the morphological
and functional characteristics of an atoll
Introduction
The
atolls show a wide range of morphological
diversity, which has resulted in an equally wide
range of functional modes among the lagoon ecosystems.
The Typatoll programme
was designed to determine the
relationships existing between the morphological
characteristics of an atoll, expressed in terms of the
forced variables (FVs), and the
way its lagoon ecosystem functions, expressed in terms
of the internal variables
(IVs). Here we give the
results of a multivariate statistical analysis on the
relationships between the FVs and the IVs.
The
FVs were assessed using teledetection procedures and by
carrying out measurements on the lagoons during
oceanographic campaigns. Each FV, such as the area,
applies to a whole atoll.
The
internal variables (IVs) were collected in the framework
of the Typatoll
project in various parts of
14 Tuamotu atoll lagoons at two times of year
reflecting the annual cycle. The first of these periods
was centred on November so as to integrate the effects
of the dry season, which lasts from May to November, and
the second period was centred on March so as to
integrate the effects of the wet season, which usually
lasts from December to April. As in the case of the FVs,
the data presented here are at the scale of a whole
lagoon. The values adopted were usually the means of all
the measurements carried out at various points on a
lagoon during a given season. The reason why a single
value of each IV was used to characterise a lagoon at a
given season was that the intra-lagoon spatio-temporal
variablility was generally much lower than the
inter-lagoon variability. The latter statement was
confirmed by Dufour & Harmelin-Vivien (1997) as far
as the water column IVs are concerned.
The
values of these variables can be consulted in the database.
The
IVs relating to the
sediments, the
fish and the benthos were collected during a single
campaign on each lagoon centred on the dry season.
Strategy
Data
were collected on a very large number of variables and
there were relatively few values per variable.
This prevented us from performing classical
multivariate statistical analyses, the calculation and
interpretation of which require relatively large numbers
of values (i.e., large numbers of samples and
measurements). We therefore adopted a strategy
consisting of alternating:
-
a
method involving few constraints as to the
nature of the data, with the largest possible
matrices of variables: the multidimensional scaling
(MDS) method, followed by
a comparative analysis with the FVs (using Clarke
and Ainsworth’s method), and
-
a
more classical ranking method (principal components
analysis) on smaller data matrices corresponding to
the minimum conditions of applicability of the
method (number of variables ≤
number of values per variable, in this
case the number of atolls).
Analyses
Classification
of the atolls: the basis of a typology
|
In
terms of the way in which the lagoon biosystems
function, the various methods of analysis used
show that the Tuamotu atolls can be classified
mainly on the basis of their size and
their porosity.
|
The
other FVs analysed, namely the mean depth, the shape,
the area of the coral ring and the latitude were found
to account less satisfactorily for the functional
differences observed between the lagoons. It was
therefore proposed to classify the atolls on the basis
of the above two morphometric variables
|
|
Fig.
10 - The
position of the Tuamotu atolls on the plane
formed by the log of their porosity and the
log of their surface area.
o: Taiaro :
coordinates and name of an atoll included in the
analysis.
▄ : coordinates and name of an atoll not
included in the analysis. |
|
It
can be seen that the atolls included in the
analysis were in fact representative of the
Tuamotu atolls on the whole, although relatively
few of the giant atolls were included.
|
Based
on the above results, it can be concluded that the
nearer to each other the atolls shown in fig. 10
are placed, the more similar their lagoons are.
This property was subsequently used in the kriging
model studies.
|
Atoll_site_