Transcript (RES) model

Mapping world-wide distributions
of marine mammals using a
Relative Environmental Suitability (RES) model
I. Introduction

Delineation of large-scale geographic ranges
mammals is difficult and often subjective (Fig. 1)

We developed a generic approach to map global distributions of
115 marine mammal species based on species-specific habitat
preferences using a GIS-based environmental envelope model
marine
Sea Around Us Project,
Fisheries Centre, UBC, Vancouver
Contact: [email protected]
II. Methods
Expert knowledge &
other source of
habitat preference
information
Sighting
records
Model input 1: Qualitative & quantitative information about
marine mammal habitat preferences

Model input 2: Global 0.5 degree lat/long raster data sets of
bathymetry, annual mean sea surface temperature (SST), annual
mean distance to ice edge (Fig. 1a, 2a , 3a)
distribution
We assigned species to broad habitat predictor
categories (environmental envelopes) in terms of depth,
SST and ice edge association (Figs. 1b, 2b, 3b)

We calculated the combined relative environmental
suitability of each cell based on local environmental
conditions (Illustrated step-by-step in Figs. 2, 3, 4)
4
6

0
2
sightings
STEP I

‘Mainly continental slope’
Pmax
probability
8
2b
10
Fig. 1 –
1
Standard
outline of
max. range
extent of
Sowerby’s
beaked
whale
(Jefferson et
al, 1993) &
known
2a
sightings
of
K. Kaschner, R. Watson,
A.W. Trites & D. Pauly
Fig. 2 a – global bathymetry
0
-200
-1000
-2000
-3000
-4000
-5000
-6000
-7000
-8000
Depth [m]
b – sighting frequency per depth
predictor category & assumed
environmental envelope
III. Results & Validation
2c
High
Predicted RES maps:
c – predicted RES for species
based on depth preferences
alone
Low

match traditional outlines of max. range extent
closely in most cases (Fig. 4c)

provide information about likely heterogeneous
patterns of species’ occurrence (Fig. 4c)
3a
Northern fur seal
Harbour porpoise
Killer whale
Antarctic minke whale
0.54
0.59
0.56
0.71
0
0
0.54
0
10
6
4
0
Fig. 3 a – mean annual SST
2
sightings
STEP II
% of random data
sets with significant
correlations
‘Subpolar – warm temperate’
8
3b
Species
Spearman's
non-parametric
rank correlation
rho
p
-2
0
5
10
15
20
25
Table 1 Validation results
3c
High

c - predicted RES based on depth
& temperature preferences
Low
V. Conclusions
Relative Environmental Suitability modeling


represents a more objective approach than
the standard outlines of maximum range
extents
is useful to address ‘Big Picture’ questions
relating to biodiversity, marine mammalfisheries interactions, speciation, historic
ranges etc.
Acknowledgements
This work was supported- by The Pew Charitable Trusts of
Philadelphia, USA as part of the ‘Sea Around Us’ Project, NSERC
and a Li-Tze-Fong Graduate Fellowship
10
8
6
STEP III
4
allows visualization of hypotheses about
species distribution
‘No association with ice edge‘
2

C
4b
correlate significantly
with observed patterns
of species’ occurrence
based on independent
sighting data sets for
all species tested
(Table 1)
0
utilizes expert knowledge & is independent
of point data for input
4a
sightings

0.0001
0.0001
0.0001
0.0001
30
SST [° C]
b - sighting frequency per
temperature predictor category &
assumed environmental envelope
<
<
<
<
-1
Fig. 4 a – mean annual distance
from the ice edge
0
1
500
1000
2000
8000
Distance from ice edge [km]
4c
b - sighting frequency per iceedge distance predictor category &
assumed environmental envelope
High
Low
c - predicted RES based on depth,
temperature & distance from ice
edge preferences
FINAL MODEL OUTPUT