The mis calculation of diet indices

1
The mis calculation of diet indices
Robert J. Latour, Christopher F. Bonzek, and
James Gartland Virginia Institute of Marine
Science College of William and Mary Gloucester
Point, VA 23062
2
Why examine diets of fishes?

• Define structure of food webs

• Evaluate competitive interactions

• Estimate food web energy flow

• Infer prey-selectivity

3
Why predation?
4
Synthesizing raw diet data
5
Fish diet studies

• General approaches to collecting stomachs

6
Calculation of W
(1.) Haphazard or opportunistic sampling
- Every fish in the population has equal
probability of being captured at the time of
collection pi n/N
- Each fish stomach is assumed to be an
independent sampling unit
7
Calculation of W
(2.) Directed sampling (active gears)
- Fish distribute in temporal, spatial
aggregations
- Biological, ecological characteristics reflect
this underlying spatio-temporal pattern
- Diets of fish collected in same tow are more
similar than those collected at different
time/place
- Tow is assumed to be the independent
sampling unit
8
Diet index

• Use of cluster sampling estimator in fish diet
  studies

Pennigton 1985, Pennington and Volstad 1994,
Bogstad et al. 1995, Buckel et al. 1999, Link
and Almeida 2000, Pennington et al. 2002..and
others
9
Example 1 Hypothetical data

• Sample a predator population on 3
• occasions using an active gear (e.g.,
• trawl, seine, etc.)

• Follow a 2-stage design - subsample
• 3 individuals from each catch for diet
• analysis

• Consider 2 prey types
• Prey p1 approx. 10g per individual
• Prey p2 approx. 2g per individual

10
Example 1 Hypothetical data
11
Example 2 Striped bass in Chesapeake Bay
Common prey

• Striped bass (Morone saxatilis)

• Life history well documented
• Principle piscivore in Chesapeake Bay
• feeds on a variety of taxa,
• ranging in mass by 105 grams

12
Striped bass sampling

• Chesapeake Bay Multispecies Monitoring and
  Assessment Program (ChesMMAP)

Maryland
Data on late juvenile adult fishes in support of
fisheries assessments and ecosystem-based
approaches to fisheries mgmt

• 80 stations, 5 cruises/yr region and
  depth-stratification 45ft otter trawl
• Traditional species-level data collected,
  including diet information from stomachs

Virginia
13
Striped bass distribution/abundance

• July 2005 cruise

• Which station-level diet data should the overall
  index follow?

14
Striped bass diet data

• Prey types that should be emphasized by diet
  index
• Bay anchovy
• Weakfish
• - Atlantic menhaden
• Spot

15
Striped bass diet, July 2005
16
Striped bass diet, 2002 - 2005
Percent W
Percent Weight
17
Central discussion point

• Our understanding of the trophic ecology of
  fishes can be sensitive to the underlying
  assumptions associated with study design

Haphazard collection simple random
estimator
cluster sampling estimator
Directed collection (active gears)
18
Acknowledgements

• Funding
• NOAA Chesapeake Bay Office
• Virginia Environmental Endowment
• Virginia Marine Resources Commission

• ChesMMAP Staff
• RaeMarie Johnson
• Eric Brasseur
• Debra Parthree
• Melanie Chattin
• Students
• Patrick Lynch
• Andre Buchheister
• Andrij Horodysky
• Justine Woodward