BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Using Bayesian Network Inference to Examine Zooplankton Ecology an
 d Herring   Interactions in the Irish Sea - Emily King (British Antarctic 
 Survey)
DTSTART:20081127T103000Z
DTEND:20081127T113000Z
UID:TALK15156@talks.cam.ac.uk
CONTACT:Christian Franzke
DESCRIPTION:Bayesian network inference algorithms can provide a robust\nfr
 amework to help understand biological systems.  In this study we\napply th
 ese algorithms to food web modeling\, for the zooplankton\ncommunity in th
 e Irish Sea.  We test the viability of extending this\napproach to include
  modeling herring stocks in the Irish Sea.  For\nthis reason particular at
 tention is paid to the dietary preferences of\nherring and historical vari
 ations in zooplankton abundance (using CPR\ndata).\n\nWe consider networks
  containing individual species\, and networks\ngrouped by herring food con
 sumption\, and find that the latter provides\na robust method for inferrin
 g causal relations between the zooplankton\ngroups and environmental data.
   Using naïve Bayes classifiers we show\na change in the prey preference 
 of herring during the summer months of\nApril to September.   We consider 
 the differences between networks\nfrom 1970-85 and 1985-2000 for data grou
 ped by herring food\nconsumption and the changes of herring prey preferenc
 e during this\ntime. We find that the influence of copepods and decapod la
 rvae\ndecreases\, while the influence of alternative food sources increase
 .\nWe show that herring is more abundant at higher temperatures\, which\nc
 orresponds to yearly peaks in the summer.  The period 1985-2000 has\ngreat
 er probabilities of higher abundances than 1970-1985\, with the\nexception
  of medium low temperatures for which abundance decreases.
LOCATION:British Antarctic Survey\, Room 330B
END:VEVENT
END:VCALENDAR