1. The document summarizes a presentation on the contemporary decline of rodents in Australia.
2. It finds that rodent declines in northern Australia have been driven primarily by predation from cats and habitat changes from fire and grazing that simplify vegetation structure.
3. Rodent declines across Australia are caused by a similar set of traits around body mass and habitat preference, suggesting interacting causes of predation, fire, and grazing are driving the declines.
TERN Ecosystem Surveillance Plots Kakadu National Park
Michael Lawes_Contemporary rodent decline and extinction in Australia: feral animals, frogs, felines or fire?
1. Contemporary rodent decline in
Australia
Presentation by Mike Lawes, Diana Fisher, Chris
Johnson, Mike Letnic, Alex Kutt, Anke Frank
2. An output of the ACEAS working group:
Vast lands and variable data: systematic
analyses to understand the patterns and
processes of mammal decline
3. Databases
Collate survey, museum, literature data on mammal presence/abundance
across Australia, especially northern Australia: >600,000 data records
4. Australia: a third (24 species) of all recent
mammal extinctions worldwide
Causes? Invasive
predators, changed fire
regimes, cattle/sheep grazing
pressure…plus other factors
6. But northern savannas have recently
experienced dramatic declines of small
mammals: eg. Kakadu NP (Woinarski et al
2010)
Individuals Species
16 3.5
No. of individuals per plot
14 3
No. of species per plot
12 2.5
10
2
8
1.5
6
1
4
2 0.5
0 0
1996 2001 2008 1996 2001 2008
Year Year
7. Problems in diagnosing cause:
• Lots of contenders:
fire, grazing, cats, climate/weather
change, disease, toads…
• Patchy survey, observational & historical
data, spread unevenly over a vast area
• Almost no in-depth study of declining
populations, and little time
• Declines happening fast
8. Aims:
• Collate, review and analyse all existing
data & evidence
• Evaluate hypotheses/formulate new
hypotheses
• Apply traditional & novel methods for
analysis of disparate data sets
• Define new policy and management
practices for Govts and communities
9. Global Ecology and Biogeography
The historical and current decline of
tropical marsupials in Australia
Diana O. Fisher, Chris N. Johnson, Michael J.
Lawes, Susanne A. Fritz, Hamish
McCallum, Simon P. Blomberg, Jeremy
VanDerWal, Brett Abbott, Anke Frank, Sarah
Legge, Mike Letnic, Colette R. Thomas, Alaric
Fisher, Iain J. Gordon, Alex Kutt.
North & South South North
Body mass medium-sized species small species
Habitat structure
Habitat modification
(reduced cover)
Introduced predators
10. Random forest - Error rates:
Out-of-sample prediction: 21%
conditional inference Non-declining species 16%
Declining species 27%
tree N S
11. Rodents
• Accounted for
phylogenetic non-
independence using
GLS model and
phylogeny derived
from Cardillo et al.
(2004), using Ford
(2006) for resolution
of Pseudomys and
Geffen et al. (2011)
for Rattus.
• Strong phylogenetic
signal.
12. Regression models
Declines: Size in S vs. N
• Strong linear
correlation
between decline
and body mass in
the south but not
the north.
• Declining rodent
species in the
north are 40-270 g.
• This size range in
the south is the
larger species.
13. Declines: Size in S vs. N
Larger rodents
declined more in the
south, and no ‘large’
rodents survived in
the south.
Maximum size of Maximum size
southern rodents is southern spp.
approx. half that for
northern rodents
14. Declines: Aridity S vs. N
• Rodent declines in the south
have been worse in arid
areas.
• No clear trend with rainfall in
the north
• All declined rodents in the
north are from mid-rainfall
savanna and not desert or
forest
• But, many savanna-climate
rodents are also OK
15. Summary: Regression models
• A particular size range (75 – 240 g) of rodents is
at risk.
• These are the larger southern rodents and the
medium sized northern ones.
• As in marsupials, all rainforest species are safe.
• The most declined rodents are in savanna and
arid environments, but also many species in all
rainfall bands are fine – climate is NOT the
primary driver of declines.
17. Northern rodent species
Of the three most important variables
(mass, habitat openness, rainfall) the declines
among northern rodents partition by mass
only and even then this is a weak inference.
Rodents larger than 76g are likely (P=0.3) to
decline, while smaller rodents are unlikely to
decline.
18. Southern rodent species
Species in semi-arid and arid regions are
significantly more likely to decline than those
in mesic regions, and in particular herbivorous
rodents. Insectivores or generalists are slightly
less likely to decline than herbivores.
19. Conclusions
• Drivers of decline consistent with findings for
marsupials
• Decline predicted by a distinct suite of traits: body
mass and habitat structure (climate and grazing
dependent)
• Like the earlier southern declines, but size-range
shifted down in the south and the north – predation?
• Suggests cat predation in
the north vs. foxes in the
south?
• Indicates interacting
causes: cats/foxes-fire
-grazing
20. Causes? (1) predation (cats)
• Declined species are prey
of cats
• Persistence on cat-free
islands …
• … but declines follow
arrival of cats on islands
• Anecdotal evidence for
local increase of cats (?)
And…
22. Causes? (3) grazing
Recovery of mammals at Mornington
Sanctuary AWC, Kimberley
(data from Sarah Legge)
Small mammal abundance
Years since removal of cattle
23. Single overall process causing rodent
declines throughout the continent
• Declines caused primarily by predation
(cats/foxes)
• Habitat changes by fire and grazing
simplifies vegetation structure
amplifying the effects of predation
24. Bigger messages
1. TERN/ACEAS process worked well to develop
understanding of a rapidly unfolding ecological
change
2. We might be looking at a classic ‘tipping point’ story
3. CWR for rodents
4. Control feral predators
5. Comparative analysis
useful for identifying
potential causes of
multi-species decline
25. Further work enabled by the ACEAS
working group
• ARC Linkage project on mechanisms of
mammal decline
• NERP field projects on landscape
management for northern small mammals
• AWC adaptive landscape management
• New sub-projects by WG members (fire
scale, trends in dingoes)
• Data inputs for Red List reassessments of
mammals [Australia Global]
Editor's Notes
Add icons from ACEAS presentation 2010
The most important variables in random forest models were body mass, habitat type, rainfall, diet, litter size and whether the species was from northern or southern Australia (Fig. 2). A conditional inference tree with these six traits indicates that the effects of habitat, body mass, diet and rainfall on decline probability differ between northern and southern marsupials, and that the crucial (first) dichotomous split was whether a species was northern versus southern (Fig. 2). The regression tree based on the Australia-wide data set indicates that body mass is the critical trait associated with declines in southern Australia, and reveals a threshold at 40.7 g, below which species were unlikely to decline. Species heavier than 6310 g had a lower risk than intermediate body sizes (40.7 – 6310 g).In contrast, the regression tree model suggests that medium body mass is not a critical factor in distinguishing species most at risk in the tropics, but among grassland and savanna dwelling marsupials, herbivorous species (which are relatively large: 6763 1186 g) were significantly less likely to decline than were carnivorous or omnivorous species (which are small: 625 156 g). The regression tree indicates that in southern Australia, rainfall is the next most important predictor of decline after body mass. All marsupials in the intermediate body mass range (40.7 – 6310 g) in the south declined if they were in a climate zone with less than 789 mm of rainfall a year. Most species in this size range also declined in higher rainfall regions in the south.In contrast, habitat structure is the critical trait that distinguishes declining marsupials in the tropics. Grassland and woodland (tropical savanna) species are at greatest risk, and almost no marsupials in forest or rainforest are declining. Among non-herbivores in the relatively open habitats of grassland and savanna, most marsupials in regions with more than 911 mm of annual rainfall are declining.