Below I give you the outline of my talk, just to give you an idea of what I've been up to.
Predation has been
associated with changes in behaviour and habitat choice in a varied range of species.
Moose shift
their diet dramatically to avoid wolf predation. Birds alter their time
budgets and flock size in the presence of a predator. Here in
Australia, velvet geckos will forego hiding in crevasses previously occupied by
snakes.
Mitigating predation risks represents a trade-off for foraging and reproduction.
Different species come up with different strategies and budgets for survival.
And so predation becomes one of the drivers that helps facilitate coexistence
and high local diversity.
The system is
influenced by abiotic factors like rain
and fire. However, the effects of these large-scale events are not
easily predicted.
In the system
we work in, deserts, rainfall and fire are major drivers of the ecosystem. But
whether they have an impact on predation and if so, how, is not known. These
are questions we’re trying to address and with the Australian desert agamids we find an
unusually good opportunity to do so.
The deserts of central Australia contain richer
communities of lizards than any other arid regions in
the world, with the highest diversity occurring in sand dune habitats dominated
by hummock forming spinifex grasses.
We are here (Alice Springs). This red
blob on the map is the Simpson Desert, we’re on the
western border of it. And the field sites of our work are situated on
the eastern end, right on the border of QLD and NT.
This is an aerial
view of our camp site and surrounds. You can see the road through the middle
which has acted as a fire break. To the left, around our camp there is
old, unburned spinifex habitat. To the right, the fire burned most of the
spinifex, leaving a sparsely vegetated, sandy habitat. Over time, fires create
a mosaic of burned and unburned patches at various stages of regeneration.
Within this setting we find two species of
agamid lizards, the military dragon and the central netted dragon. They both occur abundantly throughout the Simpson Desert
but in contrasting habitats. The military dragon is dominant in areas
where spinifex ground cover exceeds 30%, whereas in areas where spinifex cover
is less than 10%, the central netted dragon is dominant.
In
addition, each lizard species selects different habitat components (micro
habitats). The military dragon selects areas within 30cm of spinifex hummocks,
whereas the central netted dragon selects dead wood within sparsely vegetated
areas.
The difference
in habitat use and behaviour between these species is distinct. Ben Daly has
asked the question ‘why?’ in his 2008 Ecology paper. He identified predation as
a possible driver for this divergence.
If predation is indeed a
driver, the first question
that arises is
whether predation pressure is different in burned and unburned habitats. The second question is whether the use of
different micro-habitats is associated with different predation rates.
These are a selection of predator the dragons encounter: raptors, including the occasional grey falcon if we're lucky, reptiles and mammals.
Just to give you an example of what predation looks like
in the field, here’s a central netted dragon we
encountered only a couple of weeks ago. This one was scooped up by a raptor in
a burned area, thoroughly decapitated and left in a tree
.
Instead of offering up live lizards to
the same fate, we used models made of plasticine clay to test the differences in predation pressures and
predator assemblages.
We used
simplified models of both lizard species.
Using models also has the benefit that we can standardize size, sex,
appearance, and behaviour which makes the method optimal for comparing between
habitats.
300 models,
150 of each species, were set up at 60 sites. Half these sites were burned, the
other half were unburned. Sites were a minimum of 500m apart to make sure
attacks between them were truly independent.
Within a site,
5 models of the same species were set up within 10x10m. Each of these were set
in one of the 5 main micro-habitats the dragons use. One out in the open, one
head-first under spinifex, one parallel and close to spinifex, one set on wood
and the last on leaf litter of a shrub.
After 4 trips between November 2013 and last week, a
total 1200 models and 14,500 checks we put all the results together.
I have to do a
little disclaimer here. This is still hot of the press, the last data was only
collected last week and I’m still working through the full analyses – these are
the preliminary results of a Generalized Linear model (GLM) I ran to analyze
the impact of the variables we were interested in.
For clarity I
decided to forego the R graphs and output and instead go back to basics and use
very simple bar graphs to explain the results.
When we analysed all data
together, this is what we saw:
Burned
vs unburned -> has a significant
impact on attack rate as expected. However, the impact is the complete opposite
of what we expected to see, with unburned, dense habitat incurring a much
greater risk of predation.
Unexpectedly, microhabitat selection did not influence attack rates
Military
dragons and central netted dragons were not attacked at different rates
Birds
and reptiles were by far the most common predators and the only ones
significantly impacting the chance of attack. However, analyses could not
identify which one was the more important predator.
In summary,
only macrohabitat and two sorts
of predator seemed to explain a significant amount of variance. And of those,
the macrohabitat effect was
completely opposite to what Daly found a few years ago in his pilot study; and
on top of that we cannot distinguish which of the two predators is more
important.
All in all, it
felt like we were missing something.
The light bulb moment came when we decided to include rainfall into the model. If almost everything
else in the desert is linked to water, maybe it could explain some of the
variance in our predation model too.
And it did. We
split the data in two, dry times where there had been no substantial rainfall
in the previous 6 months and wet times, where there was either rainfall on the
trips itself or in the two months before.
One of the most marked
changes occurred in attack rates between macro habitats. In the dry
times attacks in open habitats surged, matching the results Daly got in his
pilots (incidentally done at very dry times).
In wet
conditions, the rates completely change. Most attacks occur in denser spinifex
grasslands.
Attack rates on
micro habitats were also greatly affected by rainfall. In dry times,
attack rates in different microhabitats are significantly different, with being
out on wood as the most dangerous spot to be and hiding under spinifex as the
safest.
Rainfall evens out the odds it seems, with no differences
found between the micro habitats.
Dragon species remained un-affected by the split between dry and
wet times, not playing a role in explaining predation events. For the purposes
of this experiment it’s a good thing. It rules out general size and skin
pattern as a cause for predation. If both species inherently incur the same
predation risk, we can assume that their contrasting behaviour and habitat use
are the coping strategies.
Last but not
least, rainfall also explained why we didn’t find a predation difference
between birds and reptiles. One turns out to be the dominant predator
in dry times, the reptiles, and the other in wet times, birds.
It’s also
interesting to look at the difference in raw number of attacks here. Avian
attacks are double the attacks by reptiles in dry times. It seems that wet
periods are much more dangerous than dry times are.
Birds arrive
with the rain and represent a distinct, highly mobile predator group. They seem
to attack several models within a site when they find it – and so they even out
the attack rates within all micro habitats. They also look like a likely
culprit behind the high attack rates in unburned habitat in wet conditions.
In conclusion, desert agamids have to
respond to a changing assemblage of predators. To build a model
explaining predation events on both dragon species we
have to include rainfall as a factor, together with macro habitat, micro
habitat and predator species.
The next
question is how these factors influence the behaviour and habitat use of the
dragons and if both species respond differently to different predators. Who
knows what behaviour they have come up with!