Do traits of freshwater species
predict vulnerability to climate
change?
Bruce Chessman
Climate Change Science, NSW DECCW
1
Climate change is likely to be the
greatest driver of ecological change in
rivers and wetlands of the MurrayDarling Basin
 Climate models generally project:
-
Increased temperatures
Increased evapotranspiration
Reduced runoff, especially in the south-east
Shift from winter/spring to summer/autumn
runoff
2
Likely traits of vulnerable freshwater
species
 Intolerant of higher temperatures
 Intolerant of stagnant conditions
 Lacking mechanisms to survive drying
 Inflexible timing of life-cycle events
 Small geographic range
 Specialised habitat requirements
 Low mobility
 Low fecundity
3
Management for vulnerable species
 Protect normal habitats
 Protect refuge habitats
 Maintain dispersal corridors
 Assist dispersal
 Control non-climatic threats
 Ex-situ conservation
4
Trends in NSW river invertebrates from
1994-2007 as a test of traits as
predictors
19
 Environmental
changes in NSW from
1994 to 2007 broadly
mirror those projected
by climate models
 Large quantity of
invertebrate data
available from river
bioassessment
projects
18
Average air
temperature (C)
17
1994 1997 2000 2003 2006
700
600
500
Average
discharge at
254 gauges
(GL)
400
300
200
100
0
1994
1997 2000 2003 2006
5
Invertebrate data
 6582 kick or sweep samples from various
habitats at 1818 river sites across NSW
 Identification only to family level, not species
 Data are essentially presence/absence per
sample (non-quantitative sampling)
-28
Latitude (°S)
-30
-32
Sydney
-34
-36
-38
140
142
144
146
148
Longitude (°E)
150
152
154
6
Data analysis
 Logistic regression used to identify
invertebrate families with significantly
increasing or decreasing probability of
capture in samples taken over the whole
State in the period 1994-2007
 Trends of individual families tested for
correlation with their traits
7
Trait 1: thermophily
(preference for higher temperatures)
 Thermophily of an invertebrate family =
average water temperature for samples in which
the family was recorded
divided by
average water temperature for all samples
8
5
Trait 2: rheophily
waterfall or
cascade
(preference for flowing
water)
4
 Habitats sampled scored
from fast-flowing to still
rapid
3
riffle
 Rheophily of a family =
average score of samples
in which the family was
recorded
divided by
average score of all
samples
2
run
1
glide
0
still water
9
Constraints on analysis
 Family level only
 In order to detect long-term Statewide trends
it was necessary to adjust for extraneous
variation associated with:
-
Shifting site locations (many sites but few
sampled often)
-
Different habitats sampled
Sampling at different times of the year
Changes in sampling methods
10
Overall results
 33 families had significantly increasing
probability of capture
 37 families had significantly declining
probability of capture
 54 families had no significant trend (mostly
rare)
11
Traits and trends
2
2
Increasers
Odds ratio
Odds ratio
Increasers
1
1
Decreasers
Decreasers
0
0
0
0.5
1
Thermophily
Heat-loving families
were significantly more
likely to have increased
1.5
0
1
2
3
4
Rheophily
Current-loving families
were significantly more
likely to have declined
12
Conclusions
 Trends did reflect traits as expected
 BUT single traits were weak predictors of
trend
 Traits and trends are likely to vary among
species within a family and among life history
stages
 Trait combinations may predict trends better
than single traits
13
Future directions
 Consider a wide range of traits
 Derive traits for each life-history stage of
each species (not families)
 Vulnerability estimates based on
combinations of traits rather than single traits
14
Thank you
And thanks to all those who collected the
monitoring data for this analysis
Further reading
Chessman BC 2009. Climatic changes and 13-year trends in
stream macroinvertebrate assemblages in New South
Wales, Australia. Global Change Biology 15, 2791-2802.
15