Diapositive 1

1
New proxies to detect changes in copepod
community structure and distribution
Lars Stemmann, Frédéric Ibanez, Maria Grazia
Mazzocchi, Stéphane Gasparini, Caroline
Warembourg, Virginie Raybaud, Gabriel Gorsky
2
Introduction
Objectives to calculate proxies based on image
analysis to assess their relevance for
zooplankton studies.
Why new proxies ? The visual counting and sorting
of plankton samples is time consuming and is a
limiting factor for the study of large amount of
samples. The harmonisation of data gathered by
different experts is difficult.
Which new proxies ?
We use the term new proxies, because they are
derived from a new instrument that does not
detect and classify objects exactly as the human
expertise would do. The proxies in this study
are the total number of copepods and the
diversity of size in two time series .
How to assess the relevance of these new proxies ?
In order to validate the proxies, we need to know
the real changes (infered from visual counting
and taxonomy) in the ecosystem and see if they
are detected by the proxies.
3
Methodology Time Series sites
Villefranche
Naples
Calvi
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
4
Methodology Zooscan and taxonomy
Sizing counting Classification
Abundance
Copepod size spectra
Slope of size spectra
Shannon
C. Warembourg V. Raybaud
Visual counting for copepods
Comparaison Bravais-Pearson correlation (after
Chelton correction)
Abundance Taxonomic diversity
Naples M.G. Mazzocchi Villefranche S.
Gasparini
5
Methodology Automatic sorting with the Zooscan
Dataset (series)
Automatic classification
The training set is specific for a region.
6
Introduction Why do we expect a relation
between zooplankton size spectra and taxonomic
composition ?
frequency
frequency
size
species
ZOOSCAN
EXPERT
7
Methodology Deriving new indices from spectra
Shannon indice calculated from the
spectra (Parson, 1969 Ruiz, 1994)
Slope of the spectra (Platt et Denman, 1977, 1978)
Size class

• Log-Log transformation
• Linear regression

Samples
Shannon (H) (1 size class 1 species)
Log (abondance) en ind.m-3
Equitability, Piélou (J)
Biais due to modes
No biaise due to the position of the modes.
8
Methodology detection of interannual changes
Extraction of interannual component Census II
Raw Time Series
Cumsum
9
Results Abundance at the three sites
Abondance (ZOOSCAN)
4

• Villefranche peaks in march-april, highest
  abundance when compared to other sites.
• Naples maxima in april-september
• Calvi maxima in march april

10
Results Abundance at the two sites
Abondance (Experts)
11
Results Time series of the size spectra slope
in Villefranche
12
Results Correlation between slopes and diversity
13
Results Correlation between slopes and diversity
the value of the slope depends on the choice
of the size classes
14
Results Correlation between size and taxonomic
and diversities
2
r 0,72 Significatif à 5
Villefranche 2002
1
Reduced Diversity indices
0
-1
Taxonomic diversity Size diversity
-2
déc
janv
fév
mars
avr
mai
juin
juil
aout
sep
oct
nov
déc

15
Results Seasonal difference
Naples 1997 à 2001
Sesonal component of Equatibility indice
3
2
1
0
-1
-2
Taxonomic diversity Size diversity
-3
-4
1997
1998
1999
2000
2001

• Spring, summer and autumn similar evolution
• Winter Opposite evolution

Lot of small copepod during winter in Naples
(Clausocalanus, Oithona et Oncea).
16
The size spectra is not linked only to the
taxonomic composition of the zooplankton. It
includes the life cycle of the organisms.
ZOOSCAN
EXPERT
frequency
frequency
X
size
species
So how can we use the new proxies ? What do they
tell us on the system ?
17
Results Comparison between sites
Size diversity (Zooscan)
Equitability (J)
18
Results Interannual changes of the local
environement
Cum sum on interannual fluctuation
Upper salinity
Villefranche Naples
19
Results shift in 1999
Salinity
Copepods Abundance
Villefranche Naples
20

• Conclusions and Perspectives
• 1) Imaging techniques such as the ZOOSCAN provide
  reliable information of the abundance and size of
  copepods, which can be use to derive proxies.
• 2) The ZOOSCAN abundance is a good proxy for the
  total abundance of copepods in a given area. The
  diversity in size is related to the taxonomic
  diversity but also to to the population age
  structure (e.g., adults vs. juveniles) the
  relation is therefore complex and deserves
  further investigations.
• 3) Temporal changes in the copepod abundance in
  Villefranche sur Mer and Naples can be related to
  changes in salinity, which may indicate changes
  in the hydroclimate at larger scale.
• There is a need for
• increasing the taxonomic effort to study the
  correlation between the proxies provided by
  automated analyses and real changes observed in
  the taxonomic composition (notably around winter
  1998/1999), and
• working on other proxies than the ESD (or
  combinations of proxies) provided by the image
  analysis. The use of new proxies will help to
  rapidly detect periods of changes in the plankton
  communities from existing and on going time
  series.

21
END
Thank you C. Warembourg, V. Raybaud, F. Ibanez,
G. Gorsky, MG Mazzochi, S. Gasparini, E.
Antajun, M. Picheral,
Lets work together