An Abundance Exchange Model of Fish Assemblage Responses to Changing Physical Habitat on An Embaymen

1
An Abundance Exchange Model of Fish Assemblage
Responses to Changing Physical Habitat on An
Embayment-Stream Gradient of Lake Ontario
Nuanchan Singkran Natural Resources, Cornell
University
2
Conceptual Frameworks
Modeling abundance distribution of fish
assemblages across different habitats and
determining ecotone functions.

• Hypotheses
• Abundance distribution of fish on a physical
  gradient can be determined from their physical
  habitat preference (PHP).
• Diverse abundance exchanges of fish in and
  around an ecotone can reflect the ecotone
  multiple functions.

• Objectives
• Predict abundance exchange of fish assemblages
  across different habitats along a physical
  gradient.
• Explore ecotone functions on fish assemblages
  along the gradient.

3
Modeling Site The Floodwood (FL) Gradient
12
2002
11
10
8
2004
9
7
8
7
6
FL3ltEcotoneltFL6 Gradient TL 28.2 Km
5
5
4
4
3
3
Station
3.2
3.9
21.1
km
2
1
3
6
12
1
4
Occurrence of Selected Fish along the FL Gradient
in the Summer of 2004
Blacknose dace
BND
Fantail darter
FTD
Cutlips minnow
CLM
White sucker
WS
Logperch
LP
Bluntnose minnow
BM
Rock bass
RB
Bluegill
BG
Largemouth bass
LMB
Yellow perch
YP
3
2
1
4
12
5
11
9
10
8
7
6
Station
Downstream
Ecotone
Upstream
5
Abundance Exchange Model (AEM) Assumptions

• Along a physical gradient, a given species has a
  migration rate among heterogeneous habitats
  following non-Ficks law of dispersion (Johnson
  et al. 1992)
• Although a given species acts as a random
  walker, it has a biased decision to migrate to
  its optimal habitat.
• Abundance exchange of each selected fish from
  two life stages, young (0yrs old) and adult
  (1yrs old), within and among habitats are
  related to
• areas-seeking of fish in winter growing
  seasons.
• PHP of young fish in growing season, and PHP of
  adult fish in winter and growing season.
• Abundance exchange of fish is considered on a
  1-D system of the gradient.

6
AEM Development

• Multiplicative PHP model is used to estimate
  PHP for each selected species at each life stage.

Pix preference index on habitat
attribute x x 1, 2 , 3,n Fi observed No.
of fish at an intensity interval i of a habitat
attribute x Ft total observed No. of fish from
all intensity intervals i of a habitat attribute
x Ei No. of observations at an intensity
interval i of a habitat attribute x Et total
No. of observations from all intensity intervals
i of a habitat attribute x

• Habitat attributes putting in the model are
• Water depth (m)
• Current velocity (m/sec)
• Water temperature (0C)
• Substrate () mud, sand, gravel-cobble,
  rock-bedrock
• Cover (level) aquatic plant, algae, woody debris

7
AEM Development

• Finite-difference equations are used for
  simulating the AEM

Where

Pn Fish population, young
(Y) and adult (A), in habitat n
akn (1-aPHPn)/t migration
rate of A out of habitat n
akn-1, n migration rate of A from
habitat n-1 to habitat n aPHPn
PHP of A in habitat n

f (Yn) rb (An) (1- Yn/Cy) net birth
rate of Y in habitat n rb
birth rate fraction of Y

Cy carrying capacity of
Y
z very small No.
(e.g., 10-6) to avoid zero denominator
8
AEM Conceptual Diagram
akwn,n1
akwn,n-1
akgn,n1
akgn,n-1
ykgn,n1
ykgn,n-1
Habitat n-1
Habitat n
Habitat n1
9
Results Discussion
?2 Test Ho No sig. difference between obs.
pred. Ha Sig. difference between obs. pred.

• The AEM showed acceptable prediction.

?Among 12 stations Accept Ho if ?2 lt19.68,
Pgt0.05, df 11 ?Among 3 major habitats Accept
Ho if ?2 lt5.99, Pgt0.05, df 2

• Four groups of fish responses to the ecotone
  reflected the ecotone multiple functions as

A hard barrier BND FTD
A soft barrier BG, YP, CLM
A neutral habitat LMB, RB, BM, WS
An aggregator LP
10
Acknowledgment Committee
Mark Bain Patrick Sullivan
Peter Loucks Cornell Biocomplexity
Project Fieldwork Crew Tom Bell
Erica Fleisig
Mike Brown Kristen
Hahn Friends Kristi Arend
Marci Meixler Takao
Kumasawa Geof Eckerlin Javier Gamarra
Further modeling work
?Stochastic events of the habitat attributes
among years will be created in the AEM to test
the model sensitivity. ?The model accuracy will
be validated against the Hudson River data set.