The following post features the slides and (roughly) what I said when presenting part of my research at the Ecological Society of Australia annual conference held in Alice Springs last week (28 September – 3 October 2014). I was presenting in the ‘invasive species’ session on the Thursday morning which included, amongst others, some great talks by fellow PhD students Melinda Trudgen and Tim Doherty. I present this here for anyone who happened to miss the talk but was otherwise interested in some of the content. Throughout the post the relevant text comes after the slide:
Habitat augmentation drives secondary invasion: an experimental approach to determining the mechanism of invasion success
“Up next we have Luke O’Lofflin who will be telling us about habitat augmentation driving secondary invasion”
Thanks. Its O’Locklin
“Oh I’m sorry”
**laughs ensue (good start)**
Something we all do – and is your classic first step of any research project really – is to quantify a pattern. For example, what is the pattern of tree recruitment following fire? However turning your observations into quantified data leads to another obvious question. Why? Why does recruitment happen like this?
So we move on to describing the mechanism behind these patterns and we do this in a number of ways. We can just compare environmental variables between states and suggest that any difference observed may be driving our pattern. A more robust method is to build a model that looks at your pattern as a function of environmental variables to determine which variable best describes or predicts your observed pattern.
However these methods aren’t deterministic and only infer what the mechanism is.
Consider if you will a situation whereby one environmental variable is the function of another. Your observed pattern may be due to changes in variable 1, or variable 2, or the interaction of both in which one has greater relative importance. Regardless, your inference methods – even robust models – will never answer your question as you cannot collection in situ information on one variable independent of the other.
Inferring the mechanism behind the pattern is out and it’s time to experimentally determine, and it’s this exact situation we have in our study system which brings me to the question of todays talk…
What is the mechanism determining the invasion success of landsnails in rainforest on Christmas Island?
Let me take a couple of minutes to explain the complexities of the unique rainforest ecosystem that exists on Christmas Island – that you may or may not be familiar with.
This is what intact rainforest on Christmas Island looks like. Free of leaf litter for much of the year, minimal understory complexity. Not a particularly resource rich environment for an invasive snail. It looks like this because of the actions of the highly abundant endemic detritivore – The Red Land Crab. Which apart from eating all the litter and seedlings, will also munch down on snails when given the chance.
Unfortunately, over recent decades, areas of the island have become overrun by invasive yellow-crazy-ant supercolonies which actively predate and cause local extinctions of the red crab. This leads to massive structural changes in the rainforest and we go from something looking like this…..
…to something looking like this. A pulse recruitment of seedlings creating a complex understory and the build-up and persistence of leaf litter – and here is that classic picture of Dennis O’Dowd that we all love to keep using. As well as these structural changes, there is also the removal of that biotic resistance which has resulted in the secondary invasion of Giant African Landsnails. Secondary invaders in the sense that they were unable to invade the ecosystem until another invasive species had altered some property of the recipient community.
So we asked; if one species of exotic snail is advantaged by these changes, surely the other 24, albeit much smaller, exotic species would be facilitated in a similar way? After all, these impacted rainforest are now rich with habitat and resources and a key predator has been removed.
So we made our observation and it was time to quantify a pattern – the details of which I presented in Melbourne 2 years ago and so I present to you now in its most simplest form. And from now on, all plus signs (+) mean high and all dashes (-) mean low. Essentially, we found snail abundance was low in intact forest where crab density was high, and therefore leaf litter biomass was low. And in impacted forest where crabs were absent and leaf litter biomass was high, landsnails where up to 10-fold more abundant.
So it appears the mechanism behind this success has something to do with the absence of the red crab and the presence of high amounts of litter. However, as the amount of leaf litter is the product of the amount of red crabs, we cannot make and informed inference about the relative importance of these two variables.
So the question remains outstanding – is the mechanism behind the success of these landsnails the creation of enemy free space (the direct removal of the red crab) or the release of habitat and resources (the indirect build up of leaf litter).
To answer this we set up a manipulative field experiment. One experiment at a site where crab density – and therefore predation – was low, and litter biomass – and therefore resources – was high, and snails were highly abundant. We re-introduced – or enclosed crabs with the aim of inhibiting the snail population.
We set up a full factorial experiment with plots of all combinations of predation and resources, both high and low. We had our un-manipulated control where predation was low and resources were high, and our ‘intact’ forest treatment where crabs were enclosed so predation was high and resources were low. And then we had our other combinations which cannot occur in the field, our ‘unnatural’ combinations. An open plot where litter was removed so both predation and resources were low, and an enclosure plot where litter was added so both predation and resources were high.
Here’s what these treatments looked like in the field at the beginning of the experiment. Plots were 2 x 2 m and these fences enclosed three crabs each, so at a natural density of 0.75 crabs per square meter. Litter was kept low on this (P- R- ) plot via a litter trap that eliminated new litter input and litter was kept high on this plot (P+ R+) via the constant addition of sterile litter so to maintain high biomass while the crabs were consuming it out. Blocks of the four treatments were replicated 10 times across the site and the snail community was sub-sampled monthly over the course of a single wet season, during which time both snails and crabs are active.
So the experiment worked and you’ll have to take it from me that the treatments worked – predation and resources were high where they were required to be high and low where they were required to be low. And so we can confidently look at treatment effects on landsnail abundance.
We used a generalised mixed model approach to look at the interactive effects of treatment and time on snail abundance where the variability of Block, litter properties and crab density where included as random effects in order to produce a stronger model.
If we look at what the model is predicting and here we have landsnail abundance (x) for each month (y) for our four treatments, we found snail abundance was significantly reduced in treatments where resources were low and remained high in treatment where resources were high – irrespective of what the predation pressure was. There was a significant interaction between month and predation low treatments and by months 4 and 5 you can see no overlap of the 95% confidence intervals of the resources low treatments relative to the control.
So it looks like its all about resources.
But if we look at only the larger snails from the samples – and in this case “large” means greater than 2 mm long – we see a different story emerge. Firstly, our model identifies no significant treatment effects and only pulls out the fact that snail abundance changes overtime. However, if you look a bit closer at what’s going on in month 5 at the end of the experiment, the model is predicting landsnail abundance for the two predation low treatments (blue and purple) above the two predation high treatments – albeit with some considerable overlap in our 95% confidence intervals.
Suggestive.. maybe… that there may be some size dependency going on and once you get larger its all about predation pressure?
And for the sake of completion, here is what’s going on for only the small individuals, so those less than 2mm long, and obviously this is more or less exactly the same as what I showed you two slides ago for the total population Basically pointing out here that this community is dominated, at least numerically, by these small individuals.
So to conclude – is the mechanism behind this increased landsnail abundance following the impacts of the yellow-crazy ants the creation of enemy-free space or the increase in habitat and resources. For the small fraction, which is also the majority of the community, we have strong evidence that the presence of the crab doesn’t phase them and they are not invasive in intact rainforest because they are resource limited. However, for your larger species there may be some evidence that they are impacted strongly by crab predation and therefore benefit from the creation of enemy-free space.
Thanks for listening, thanks also to the funding providers and people who have helped me out. And just for reference, this exotic snail here is Subulina octona – the ‘largest’ species from this current study.
The conference itself was fun and productive (as it always is!). Spent heaps of time catching up with friends and colleagues as well as meeting plenty of new people. Some feedback from the talk has resulted in a new theoretical framework for this experiment being considered and I am now looking forward to finalising the results and smashing out the manuscript.
See you all at ESA 2015 in Radelaide. Rock On!