Biodiversity Informatics, 15, 2020, pp. 69-80
69
PRESENCE-ONLY AND PRESENCE-ABSENCE DATA FOR COMPARING
SPECIES DISTRIBUTION MODELING METHODS
1*
2
1
2
3
,
4
5
6
7
8
,
9
10
11
12
13
,
14
15
16
17
,
2
,
2
1
School of BioSciences, University of Melbourne, Australia.
2
Swiss Federal Research Institute
WSL, CH-8903 Birmensdorf, Switzerland.
3
CSIRO Land and Water, Cairns, Queensland, Australia.
4
CSIRO Land and Water, Canberra, Australian Capital Territory (ACT), Australia.
5
CSIRO Land and Water, Tropical Forest Research Centre, Atherton, Queensland, Australia.
6
University of Lausanne, 1015 Lausanne, Switzerland.
7
University of California, Davis, USA.
8
EWHALE Lab, Institute of Arctic Biology, Biology & Wildlife Department, University of Alaska
Fairbanks, Fairbanks Alaska 99775 USA.
9
Universidade de São Paulo, Brazil.
10
College of
Agricultural and Life Sciences, University of Florida, USA.
11
Research School of Biology & Center
for Biodiversity Analysis, Australian National University, Australia.
12
Manaaki Whenua—Landcare
Research, Hamilton, New Zealand (current address: PANTHERA, Floor 18, 8 West 40 St, New
York, USA 10018.
13
Biodiversity Institute, University of Kansas, Lawrence, Kansas 66045, USA.
14
Center for Biodiversity and Conservation, American Museum of Natural History, New York,
USA.
15
Department of Geography, Planning and Environment, Concordia University, Montreal,
Canada.
16
Centre for Tropical Environmental and Sustainability Science, James Cook University,
Townsville, Australia.
17
Manaaki Whenua—Landcare Research, Lincoln, New Zealand
*Corresponding Author: j.elith@unimelb.edu.au
Abstract.
-
-
-
environmental data and can be used to predict distri-
-
-
-
-
-
Jane Elith et al. – Presence-only and Presence-absence Data for Comparing Species Distribution Modeling Methods
70
-
for PA sites were provided, so modelers could predict
-
(detailed in Supplementary Information 1,
1
1
more transparent and repeatable (National Academy
-
-
-
-
-
-
Jane Elith et al. – Presence-only and Presence-absence Data for Comparing Species Distribution Modeling Methods
71
-
model trained and
none are problem-free, and SDM evaluation remains
-
in some areas simply leads to a more or less precise
-
be used to calculate a broader suite of evaluation sta-
-
-
-
-
-
-
2
-
-
spatial resolution (smallest raster cell size) available,
2
-
-
-
-
-
-
2
Jane Elith et al. – Presence-only and Presence-absence Data for Comparing Species Distribution Modeling Methods
72
-
2
-
marised in Table 1, and details of variables in Sup-
3
Records span species of
-
to 19 120 PA evaluation sites (Table 1 and detailed
-
-
-
-
-
-
3
http://hdl.handle.net/1808/30582
below, are available openly
4
OSF data
-
561 MB in total and many users will not want to
-
-
1. Environmental rasters
/data/Environment folder at
/data/Environment
and details of coordinate reference systems, units and
2. Presence-only data—locations and envi-
ronmental samples
/data/Records/train_po fold-
-
-
3. Background data—locations and environ-
mental samples
-
/data/Records/train_bg folder at
4
Jane Elith et al. – Presence-only and Presence-absence Data for Comparing Species Distribution Modeling Methods
73
Code
Region details
Area
(‘000 km
2
)
Area location – red polygons
show locations within countries
/ continents
No. env vars
(no.
categorical)
Approx.
grid cell
resolution
(m)
Biological groups & number
species
Mean no.
records per
species
No.sites:
PA
PO
PA
AWT
Australian Wet
Tropics,
Queensland,
Australia
23.97
13 (0)
80
b: birds: 20
155
97
340
p: vascular plants: 20
35
30
102
CAN
Ontario, Canada
979.34
11 (1)
1 000
birds: 30
253
1 282
14 571
NSW
North-east New
South Wales,
Australia
76.18
13 (1)
100
ba: bats: 7
27
76
570
db: diurnal birds: 8
189
57
702
nb: nocturnal birds: 2
134
142
1 137
ot: open-forest trees: 8
42
164
2 075
ou: open-forest understorey
vascular plants: 8
21
358
1 309
rt: rainforest trees: 7
9
212
1 036
ru: rainforest understorey
vascular plants: 6
18
93
909
sr: small reptiles: 8
84
63
1 008
