I’m currently taking a GIS course and I need to submit a project proposal next week. I’m considering doing a GIS project on an invasive species, the Argentine ant, which is invading New Zealand. The project will (ideally) predict the future distribution of the invasive ant, possibly taking climate change into account.
The problem is, would this project be too hard? I only have a basic knowledge of GIS. A quick google search gives me the impression that using GIS for invasive biology seem to be quite “advanced”, constructing models and such. Do you think such project is “doable”? If this project is too hard then I will have to switch a topic.
1200 words is just for the **proposal **. The report itself has no word limit.
As for your question, whichever species I’m going to investigate is not important. The same technique can be used for any species. It is the difficulty of understanding how to use these techniques that is difficult.
The problem is, that report was done by real professionals and I’m not sure if I can do a similar one on ants. Regresion analysis for instance, could itself be very hard or even impossible for an undergraduate student to devote time to learn efficiently.
Since the project is supposed to “showcase your expanded GIS skills that you learned over the course of the project”, then you really need to tell us what GIS skills you are supposed to have learned. Without that its; kind of hard to say whether this is possible.
But look, this is an undergrad project, not a graduate thesis. You shouldn’t even be considering original research. The project is intended to demonstrate your GIS skills, not your research skills or your self-directed research ability. So don’t get over-ambitious, and don’t even *consider *attempting any work that requires you to over-extend yourself in areas aside from GIS.
A project of this type should be reverse engineered. In real research you would ask the question then set out to collect and analyse the data to answer it. In the case of these sorts of showcase projects, you should be considering what data you already have available, then considering how that can be related to what you have learned in the course, and then constructing the research project around those two considerations.
And take note of the advice on the assignment sheet. You do not want to be wasting time collecting data, compiling databases or entering data. You won’t be getting marked on that. Ideally you want a project where there are large datasets already available and you can just play around with the overlays and correlations
Having no idea what datasets you have available or what teh course is trying to teach kind of limits any suggestions I can make. But I would start with what is freely available. Google Earth is a free source of aerial photos that can be used for overlays. In Australia the vegetation community maps are all freely available in GIS format, and I assume the same is true of NZ. If you want to study ants then find out what existing datasets are available. Those are the sorts of things you should be building your project around.
I don’t pretend to be a GIS guru, but if you can tell me what the course covers and what sort of data, specifically ant data, you have available I can probably suggest something that is achievable by an undergrad. But without that info it’s impossible to say whether you can achieve what you have suggested.
BTW, regression modelling isn’t that difficult if you have the software. It’s largely a case of entering the data correctly and pressing the button.
I agree with Blake. Prediction of future invasive species spread sounds more like a graduate thesis than an undergrad project. This type of research is quite advanced and would take a year or more of data gathering and modelling, minimum.
Keep it in mind if you ever go on to graduate studies though - sounds interesting and would be valuable research!
The lecturer posted some project “ideas”. The project I want to do is based on the one below, I think I want to do the Argentine ant though, big-headed ant’s distribution is very limited in NZ which means less data available.
I pasted the disclaimer by the lectuerer too, in the beginning.
But it seems if I want to use regression or whatever tools, I need the “absence” data too, not just “presence” data? I’m not sure about that.
As for the temperature, land cover, land use, terrain etc. I think these should be relatively easy to get, as I’m sure the NZ government has national data of them and they should have already haven them available as GIS layers for people to download. (I can be wrong though, hopefully not…)
I think what the lecturer means is that we need to show that we are expected to learn new skill during the project and we must show that from the project we do. The GIS skill I “learned” last year was quite basic, it was basically an introductory GIS course, the closest thing to a project that we did last year was some computer lab work to find out the optimal locations for landfill in a NZ region. The course I’m doing now is an “advanced” GIS course.
If you are talking about specific operations then I think maybe buffer, union, intersect, some attribute table selection and stuff, basic formulas etc. But we are expected to learn new things by self-study in this current course, you can still ask the lectuerer questions but you are expected to work on your own.
I think the key is staying focused on what you want to learn (interesting GIS techniques), and be willing to cut corners or even completely make things up on the science (just as long as you clearly admit where the science is weak). I’d make sure your instructor is OK with this, but I think she or he will be.
The general idea – look at climate and a couple of land use and land cover datasets to predict invasive species movements – sounds like a great framework for doing a a few different interesting things. Just decide what interesting things you want to learn to do, then figure out how to fit that into the framework.
I think that with a littel luck and some hard work you should be able to do what you want to do. I also think you should be able to do it with either species.
You don’t have a very big data set for either species. That’s not a fatal flaw for an undergraduate project, but you will need to stress this in both your proposal and your conclusions in the final write up.
What you can do is combine this with other data sets. You can use data from outside NZ, that’s perfectly valid.That same site also provides worldwide collection points, though they are obviously far from complete. Your other option is to include environmental constraint data in addition to or instead of collection data. A few quick [Google searches](“Linepithema humile” survival temperature) show quite a bit of info on the environmental constraints of these species.
The point being that you aren’t just limited to the obvious NZ collection data sets. You can validly extrapolate from other data sets of you can get your hand on them.
I’m not sure what you mean by this. You can run a basic regression analysis using just the NZ collection data set, but it’s not too small to give good results. Once gain though, that’s not a problem for an undergrad project provide that you stress that in the real world you would have obtained a much more robust data set. These projects are more about demonstrating that you know how to do it, rather than that you can do it.
Well that’s great. that should make your original idea quite possible.
Rather than trying to do this within the GIS software, you are better off following the methodology of that first paper you cited. Construct your model in SPSS or Mathematica or whatever other statistical software you have. Once the model has been tested then apply it using the the spatial analyst in ArcGIS.
Basically what you want to do is construct a model that says that the potential habitat is defined by sites with a minimum temperature >x, a mean daily temperature >y, precipitation >c <d, slope <50%, distance to undisturbed land >e metres and so forth. It will be a little more complex than that because you want probability of invasion as well as just whether it can be invaded. But that’s the basic idea.
You then use the spatial analyst to compare the rasters for those variables, adding them, mutliplying them and so forth so that you create a map of you model.
This should be perfectly possible with the data that is available, and if you also include OS data and environmental constraints you should produce quite a nice map.
I assume that there is a dataset somewhere that shows the predicted effects of climate change on rainfall and temperature in NZ, so to show the effect on ant distribution, just compare the rasters of the altered climate with your model.
This ought to be ideal for that. It’s not going to be publishable obviously, you just don’t have the data you need. But in terms of demonstrating your mad leet GIS skills it’s should be ideal. You ought to be able to produce maps showing current potential and climate change distributions with relative ease (learning how to generate models in your stats package may be the most difficult part). And if those basic outputs are finished early, you can play around with all the other functions in ArcView. Basically, you should be able to find some way to apply almost any of the whiz-bang functions to this project.