Predatorprey interaction between harp seals and Atlantic cod:

1
Predator-prey interaction between harp seals and
Atlantic cod An exploration of sources of
variation
by Alejandro Buren1, Mariano Koen-Alonso1,2 and
Garry Stenson1,2
1. Memorial University of Newfoundland 2.
Fisheries and Oceans Canada
2
Diet variability
It is a fair assumption that harp seals diet can
vary over time, age, and geographical region.
Time variability can be considered as a surrogate
for changes in prey availability over time. These
changes can be seasonal or long-term (prey
population dynamics).
Age variability is associated with ontogenetic
changes in the diet (e.g. behaviour, changes in
energetic requirements, learning).
Geographical variability is associated with
foraging habitat and the spatial structure of
prey assemblages. This variability can involve
both latitudinal changes (e.g. north-south
clines) and depth-related changes (e.g. inshore,
and offshore areas).
3
Objectives
From a predators perspective we will
a) Examine the effect of sources of variation on
the overall diet
b) Look at the effect of these sources of
variation on Atlantic cod predation
4
Data
Harp seal stomach contents collected between 1986
and 2001 in the NAFO divisions 2J3KLMNO
5
Methods (Part I) The effect of sources of
variation on the overall diet
Multicategory logit model
Response variable (yi) Number of seals that had
prey i as principal prey item (by weight) in the
stomach
6
Methods (Part I) Sources of variation
7
Results (Part I) Main prey
8
Results (Part I) Analysis of effects
9
Results (Part I) Contrasts for age
10
Results (Part I) Visualizing model fit
Lets call a cell to the set defined by Age X
Zone X Area X Season (e.g. one cell is
Juveniles,2J3K,Inshore,Summer). There are 24
cells.
The model estimates the probability pi of prey i
to be a main prey (by weight) within a given year
and cell.
and the expected number of seals that had a given
prey i as main prey (by weight) in any given year
can be calculated as
where Nyear,cell is the total number of seals
observed in a given year and cell.
Comparing over time the expected and observed
number of seals that had a given prey as main
prey can give us a visual impression of the
goodness of fit of the model.
11
Results (Part I) Model fit (1/3)
Artic cod (Boreogadus saida)
Capelin (Mallotus villosus)
Number of seals
Zooplankton
Atlantic Herring (Cuplea harengus)
Year
12
Results (Part I) Model fit (2/3)
Other fish
Gadiformes
Number of seals
Pandalus shrimp (Pandalus spp)
Sandlance (Amodytes dubius)
Year
13
Results (Part I) Model fit (3/3)
Other shrimp (not Pandalus spp)
Atlantic cod (Gadus morhua)
Number of seals
Pleuronectidae
Other invertebrates
Year
14
Results (Part I) Close-up on Atlantic cod
Is this trend associated with sampling issues?
NOTE Although this simple approach removes
sampling effects, its magnitude does not
necessarily represent the actual population trend
because all cells have equal weights in the
calculation. This trend is consistent within
cells.
15
Results (Part I) Further Close-ups
Annual average (SE) of the predicted probability
of being main prey among the cells in the model
16
Results (Part I) Summary
All the sources of variation considered (age,
zone, area, season and year) had significant
effects on the probabilities of being a main prey
(by weight) in a harp seal stomach.
From the 12 prey groups considered, Atlantic cod
ranked 10th it is a minor prey for harp seals.
The probability of Atlantic cod to be a main prey
increased in the period 1986-2001, while Arctic
cod has declined and Capelin has shown a slight
increase.
17
Methods (Part II) Harp seals as predators of
Atlantic cod
Binomial logit model
Response variable (ycod) Number of seals that
had Atlantic cod in the stomach, regardless if it
was the main prey or not.
18
Methods (Part II) Sources of variation
19
Results (Part II) Did eat Atlantic cod?
20
Results (Part II) Analysis of effects
21
Results (Part II) Odds ratios for significant
effects
22
Results (Part II) Non-significant year effect?
23
Results (Part II) Summary
There is no significant effect of year and season
on the probability of a harp seal eating cod
(i.e. the proportion of harp seal stomachs with
Atlantic cod did not show any significant trend
over time).
However, there are significant effects of age,
zone and area on the probability of eating
Atlantic cod.
The probability of eating cod increases with age,
it is lower in 2J3K than in 3LMNO, and
it is much higher in the inshore than in the
offshore.
Simply put, the probability of eating cod depends
on both ontogenetic and geographical factors.
24
Part III Where is the Atlantic cod coming from?
25
Methods (Part III) Sources of variation
26
Results (Part III) Analysis of effects
27
Results (Part III) Contrasts and Odds ratios
28
Results (Part III) Summary
Most of the Atlantic cod in the diet comes from
the Inshore area.
When the analysis is constrained to the inshore,
as expected, the ontogenetic pattern still holds,
and the latitudinal pattern becomes more clear.
The probability of eating Atlantic cod increases
from North to South.
29
Part VI Tying up loose ends
We have seen that
a) the probability of Atlantic cod being a main
prey increases over time, but
b) the probability of eating Atlantic cod did not
change over time.
This means..
30
that for those seals which actually eat Atlantic
cod, this prey has become more important over
time.
31
Part VI A couple of hypotheses
Although not mutually exclusive, we can postulate
two simple hypotheses to explain this
observation.
For those seals that eat Atlantic cod,
a) the average number of Atlantic cod in a
stomach has increased over time, or
b) the average weight of the Atlantic cod
consumed has increased over time
32
Methods (Part VI) Is it numbers or is it weight?
Data
Stomach content data from the Inshore of 2J3KL
between 1986 and 2001.
Analysis
Descriptive statistics and simple non-parametric
correlation analysis (Spearman Rank correlation)
33
Results (Part VI) Is it numbers?
34
Results (Part VI) Is it weight?
35
Results (Part VI) Sampling issues
Age composition looks good..
Proportion
but the geographical distribution of the sampling
may cause some confounding effect
36
Concluding Remarks
Overall diet
All factors considered (Year, Age, zone, Area and
season) have significant effects on the diet of
harp seals
Atlantic cod is a minor prey for harp seals
The probability of Atlantic cod to be a main prey
increased over time, while Arctic cod has
declined and Capelin has shown a slight increase.
Predation on Atlantic cod
Year and season have no significant effects on
the probability of eating Atlantic cod,
but the probability of consuming cod increases
with a) age, b) from North to South and c) from
offshore to inshore.
The increase in the probability of Atlantic cod
being a main prey appears to be associated with
an increase in the average size of the cod
consumed in the inshore.
37
The End
for now