Tag Archives: Using GIS In Biology

Paper Of The Week: Lin et al. 2015. Classification of Tree Species in Overstorey Canopy of Subtropical Forest Using QuickBird Images

3 Jun

Remote sensing has become an integral part of using GIS in biological research, and this week’s paper of the week is a nice example of how just how detailed the information you can get from remote sensing can be. This paper is Lin et al. (2015) from PLoS One and is titled ‘Classification of Tree Species in Overstorey Canopy of Subtropical Forest Using QuickBird Images‘. In it, the authors set out to see if they could take high spatial resolution satellite imagery and use them to identify individual trees to the species level in a sub-tropical forest in Taiwan.

This may sound like an unlikely thing to be able to do, but it’s based on the idea that the leaves of different species of trees will reflect different wave lengths of light in slightly different ways, and while it takes some fairly fancy processing, it turn out that it’s possible. In fact, with the right images and the right processing, Lin et al. were able to separate out 40 different species of trees with a pretty high level of accuracy.

While this is impressive stuff in its own right, this isn’t the only reason I selected it as my paper of the week. I also selected it because of the potential that this type of processing offers to biologists. Through the processing and analysis of high spatial resolution images, it should possible to pull out almost any type of information that a biologist might ever need to know.

One example of this is something I’ve been pondering for some time now. This is how to assess the productivity of trees in an oak woodland as part of a study of breeding success in hole-nesting birds. The research question here is whether breeding success is related to the number of caterpillars found in the territory which surrounds each nest box. Now, measuring the number of caterpillars found around 300 nest boxes is just not feasible, but given that the number of caterpillars should be related to the productivity of the trees on which they are feeding, so this could be used as a proxy, and using the type of processing of high spatial resolution satellite imagery done by Lin et al. (2015) potentially provides a way to extract this information automatically. I haven’t yet had the chance to see whether this is, indeed, the case, but I’m looking forward to giving it a go (or more likely finding an eager student and persuading them into taking the project on!)

So, I think the take home point from this week’s paper is that it’s impressive just how much information can be extracted from high spatial resolution satellite images, and in many cases, all that’s needed is a bit of imagination, followed up by some intense work as you work out exactly how to pull out just the information you require to help you answer your research questions.

========================

Dr Colin D. MacLeod,
Founder, GIS In Ecology
========================

Advertisements

Paper Of The Week: Habel et al. 2015. Fragmentation genetics of the grassland butterfly Polyommatus coridon: Stable genetic diversity or extinction debt?

27 May

Back when I was an undergraduate, I was never that interested in genetic analysis: it seemed that it was all too much about the molecules and not enough about the biology. Worse, as someone who was very much a field ecologist even then, it seemed to be carried out by a bunch of pipette pushers who rarely ventured beyond their list of four alternating letters that make up the genetic code of all living things, let alone outside their laboratories. In short, while I could understand its importance for things like taxonomy and cladistics, I found it hard to see how it could be useful for doing the types of research I wanted to do.

This all changed, however, with the introduction of landscape genetics. What is landscape genetics? Well, most simply put, it’s taking the results of genetic analyses and laying them over the spatio-temporal environment from which the samples were taken. By doing this, you can start looking at not just how animals move through their environments, but also how their genes move through it. Suddenly, all that talk of minimum viable populations, genetic bottlenecks and gene flow takes on a spatial perspective, and by including information about the landscapes alongside the genetic information we can start to understand the processes of extinction and speciation in a way that we never have before.

This brings me round to my choice of paper of the week for this week.  It’s Habel et al. 2015. Fragmentation genetics of the grassland butterfly Polyommatus coridon: Stable genetic diversity or extinction debt? Why have I selected this paper? It’s because of the way it combines genetic analyses and spatial information to produce a result that, in my opinion, is greater than the sum of each individual part.

The starting point for Habel et al. (2015) is the general assumption that habitat fragmentation affects the viability of populations and so has a pivotal role to play in conservation. Given our current understanding, this should be particularly true for habitat specialists that are more likely to have both a naturally patchy distribution, and to be impacted by fragmentation. All this means, given conventional wisdom, that habitat fragmentation should lead a fragmentation of genetic populations, a reduction in local genetic diversity, and a resulting increased risk of extinction.

So far, so straight-forward, but then we get to the results of this particular study: when the authors looked at a specialist grassland butterfly, they found that genetic diversity wasn’t linked with habitat size, habitat
connectivity, or census population size. Thus, they found something very unexpected, given our current understanding of things.

How do they go about trying to explain this? The authors come up with two solutions. The first is that despite the fragmentation between populations and the specialisation in terms of habitat preferences, the species is somehow managing to keep up the levels of gene flow expected of more widespread  generalist species. For a mobile species like a butterfly, this is possible, but if true, it would suggest that dispersion and gene flow are not as limited by the surrounding landscape as we might assume, even for habitat specialists.

The second possibility is perhaps more interesting. This is that there is lag between the fragmentation of a species with a relatively continuous historic distribution into small isolated populations and the loss of genetic diversity. If this is true, then it would suggest that we could easily overlook the negative effects of fragmentation on population viability because we might mistake the effects of historic gene flow for evidence of current gene flow even in the light of habitat fragmentation. Thus, we may make false assumptions about the impacts of specific human activities, such as habitat destruction, on the long term viability of individual species and populations, and that would not be good.

As you will have guessed from the start of this post, I am no geneticist, so I cannot say which of these is more likely to be correct, but simply the possibility that the second option could be true means that measuring the impacts of things like habitat fragmentation on population viability might be much more difficult to do in anything close to real time than we might otherwise assume. And without the further integration of GIS into genetic studies, we will have no way test which, if either, is correct. Thus, because of landscape genetics, GIS has the potential to be as important to genetic studies as it is to many other areas of ecology.

========================

Dr Colin D. MacLeod,
Founder, GIS In Ecology

========================

What’s New In GIS And Biological Research: 26 May 2015

26 May

This week’s summary contains a variety of very different posts which cover topics ranging from how-to tips, to why you don’t get pineapples in the Antarctic. So, without further ado, I’ll begin.

The first post I want to highlight is a tutorial from Saara Pakarinen on creating a PostGIS database for QGIS. While not everyone who does GIS will use the database management systems, for those who do find that they need to use them, then this tutorial will help get you started.

Secondly, and while still on the subject of QGIS, there’s a really nice post from GIS Digest about qquality control, and how you can use the QGIS topology checker to help deal with any quality control problems you come across with your GIS data layers. This is something which is really important to know how to do, not only because it can help you work out why a specific data layer isn’t working properly, but also because, as with all analysis, in biological research, your results will only be as good as the quality of data which goes into your analysis in the first place.

Next, I want to consider a post from Geospatial Wanderings on extracting data layers from OpenStreetMap. For those of you who don’t know, OpenStreetMap can be a great source of spatial information, but sometimes it can be a little tricky to extract exactly the information you need. Geospatial Wanderings’ post provides instructions on how to do this in two different ways. They’re both command-based, rather than graphic user interface-based, but the flexibility that they provided in extracting just the data you want more than pays off the fact that they are a bit difficult to get to grips with the first time you use them.

Fourthly, I want to draw your attention to Volunteered Geographic Information (VGI), and how it can be used to help fill in spatial information where little is currently available. VGI (and the above mentioned OpenStreetMap is an example of this) does exactly what it says on the tin, and uses volunteered information to increase the amount of spatial data available. In ecology and conservation, this can involve things like looking for evidence of changes in forest cover from satellite images, or other similar changes in land use, but the example I’m going to point you towards today is humanitarian in nature and about how VGI is being used to help those caught up in the recent earthquakes in Nepal. It’s a nice case study of how communities of people from around the globe can come together to help others with the skills that they have.

To return to advice about using GIS, MaybeItsAMap has a great post for ArcGIS users which looks at how to select data points or features from a data layer based on their spatial locations, or the spatial locations of features in another data layer. The example they use is selecting features from a line data layer of streams based on things like polygons of management or political areas, and it provides detailed information about how to use this tool effectively to select exactly the subset of data that you want to select.

Still on the advice for using GIS, but going back to QGIS, if you’re interested in making really nice terrain maps complete with shaded elevations and contours, then check out this post by Anita Graser titled How to create illuminated contours, Tanaka-style. This isn’t something I’ve done before, but I could see it being a nice skill to have for creating really smart-looking maps for reports and presentations, especially when overlaid with biological data, such as sampling locations.

The last four posts I’m going to mention this week are brought together under the banner of things that made be stop and think. They are all loosely GIS-based (some, admittedly, more loosely than others), and all connected to various aspects of biology.

The first of these posts comes from Hamilton Ecology Lab, and is about a study of genetic diversity in an invasive species. In this case, it is European stoats in New Zealand. As invasive species often start with a small number of individuals, you’d usually expect their genetics to go through a bottleneck, but this doesn’t seem to be the case in this example. Or rather, it seems, the original source population in Britain has, for various reasons, gone through a greater bottleneck than the introduced one in New Zealand. Where’s the spatial element in all this? Well, it just goes to show that you shouldn’t make assumptions about your samples based on where the came from. Instead, you need to approach them without such preconceived spatial prejudices.

Over at Mashables, in tribute to the recent International Day for Biological Diversity, they put together a post to highlight the five greatest threats that biodiversity faces: Climate change, habitat loss, overexploitation, invasive species and pollution. This list will be nothing new to most biologists, but it strikes me how important spatial analyses are for studying, and managing, all these threats. In fact, without a decent spatial knowledge to underpin our management strategies, it’s unlikely we’d be able to control any of these impacts effectively.

HominidLikeMe has an interesting post on non-infectious epidemics, and while they are rather bizarre examples, they all have one thing in common: spatial clustering of people suffering from unusual symptoms that might, at first, appear to be caused by infections of some kind. However, further exploration results in the identification on non-infectious causes. As with one of the first ever studies that integrated spatial analysis and epidemiology, these examples show how spatial information can help us understand the causes behind human diseases, and how they are spread.

The final post comes from Scientiflix, and is aimed at kids. It’s about why certain types of plants are only found in some places and not others. In other words, it provides a kid-friendly introduction to the idea that there are spatial patterns in where different species occur, and that is a great introduction to one of the reasons why GIS is such an important tool for studying spatial patterns in ecology.  Without GIS, we wouldn’t be able to explore and test the hypothesis we generate to explain such patterns, and if we couldn’t do that, then we’re not doing science. And of course, it’s never too early to get kids interested in GIS!

So these are the GIS-related things that have caught my eye this week, but, as always, I’m sure there’s a lot of other good stuff out there as well.

========================
Dr Colin D. MacLeod,
Founder, GIS In Ecology
========================

Paper Of The Week: Pauli et al. 2015. The simulated effects of timber harvest on suitable habitat for Indiana and northern long-eared bats

20 May

Ecology is sometimes criticised for being an observational science rather than a truly predictive one. That is, much of what we, as ecologists, do is to try to explain existing or past patterns, rather than use our knowledge to predict what is likely to happen in the future or under different conditions. When combined with conservation, this means that ecologists are often left to implement conservation strategies reactively, after something has happened or changed, rather than pro-actively, before the change occurs in the first place. This is unfortunate as, if done properly, a pro-active approach is likely to lead to much better conservation and management outcomes than reactive ones.

However, with the advent of new statistical approaches, the availability of cheap, powerful personal computers and advances in GIS, our ability to make the types of predictions needed to test whether specific changes to the environment, especially those made by humans or those under our control, might have on specific ecosystems or species before they occur is finally within the reach of almost every ecologist. Yes, the techniques and skills might be difficult to get your head round at first, but the results are often more than worth the effort.

A great example of this is a recent paper by Pauli et al. (2015) from the journal Ecosphere titled The simulated effects of timber harvest on suitable habitat for Indiana and northern long-eared bats. Bats, like many other species, can be impacted heavily by human actions in forestry management, but what timber harvesting strategies are likely to be best for their conservation? And which are likely to be the worst?

While I’m a marine biologist at heart, this is a subject I’ve dabbled in before. This is one of the reasons I found Pauli et al.’s paper so interesting. While, in our own study, we looked at historic effects of forest management on bats, they looked at future ones, and specifically, they combined species distribution modelling and forest succession models to compare the likely impacts of nine different timber harvesting strategies on two bat species over a prolonged period of time. They also considered the differences that impacts might have on nocturnal foraging habitat and diurnal roosting habitats, an interesting extension to the more traditional approach of just looking at one, or a combination of, habitat requirements rather than looking separately at each individual component.

So what did they find? Well, you’ll have to read the paper to get the full details, but in summary, they found that the overall suitability of habitat was primarily driven by the requirements for diurnal rather than nocturnal habitats, that what might be the best strategy for one species may not necessarily be the best strategy for another, and that if you wish to have the best outcome for multiple species, you might want to select a timber harvesting strategy that was somewhere between theses two.

While these results are relatively complex and at times contradictory in terms of their impacts on the two species, they do provide concrete information that can be used to ensure that any timber-harvesting strategies are implemented in such a way as to have the best outcome possible for all species being considered, and that is something that is always better to know before you implement them, rather than afterwards.

Of course, as with any predictions which you’re going to use for setting conservation or management strategies, you have to ensure that your predictions are accurately, but as long as you have the appropriate spatial and temporal validation as part of the investigation process, this issue can often be easily avoided. Then all that’s left is to say: welcome to the world of predictive ecology.

========================

Dr Colin D. MacLeod,
Founder, GIS In Ecology
========================

Paper Of The Week: Buschke et al. 2015. Simple mechanistic models can partially explain local but not range-wide co-occurrence of African mammals

13 May

When doing GIS, sometimes you can get so caught up in all the fancy things that you can do with it, that you forget that, just like statistics, while GIS can do many amazing things, it’s still just a tool which you can use to answer research questions which interest you. This is why some of my favourite examples of the practical application of GIS to biological research don’t necessarily contain huge amounts of complex GIS-based data manipulations, but instead just use some basic GIS processing to bring data together so they can be analysed to help investigate interesting questions.

In the case of this week’s paper of the week, the interesting question being investigated is: why are variations in species diversity often linked with variations in climate? It seems a straight-forward enough question, but it’s one that we really don’t yet have a satisfactory explanation for. This is where a recent study by Buschke et al. 2015 comes in. In this study, titled ‘Simple mechanistic models can partially explain local but not range-wide co-occurrence of African mammals‘ (Global Ecology and Biogeography doi: 10.1111/geb.12316), the authors sought to explore this very issue (a summary of this paper, containing some nice animations, written by the lead author, can be found here on the Solitary Ecologist blog).

As is often the case, the GIS work involved in this is well-hidden within the methodology, but it involves calculating species richness within individual grid cells so that the local diversity of mammals can be compared to local climatic conditions. As such, while the actual spatial processing is relatively basic and straight-forward, without GIS, this study would not have been possible. However, the term GIS isn’t even mentioned once. I point this out not as a criticism of the authors, but rather to highlight how the importance of GIS to individual studies, and to biology as a whole, can often be overlooked in preference to detailed descriptions and discussions of statistical analyses.

So, with that personal bugbear out of the way, what did the study actually find? Well, they found that the evidence suggests that climate doesn’t influence the ranges of individual species, which in turn determines species richness. Rather, they suggest that climate may influence how many species can persist in a local area, and it is this that then determines the variations in local biodiversity linked to variations in climate.

This is an interesting idea, and one which is worthy of further investigation, especially with other taxa, in parts of the world and in environments (this study only looked at terrestrial mammals in Africa). In particular, it would be interesting to see how this hypothesis transfers into the marine environment, and whether patterns of biodiversity in marine mammals follow the same rules as their terrestrial relatives. In marine mammals, there are clear relationships between biodiversity and water temperature, but no one has really come up with a decent explanation as to why and maybe this hypothesis could provide one.

Does this matter? Well, quite frankly, yes. If we’re to understand how climate change will affect local diversity of marine mammals (or indeed any other taxa), we need to understand why species diversity is linked to climate in the first place.  So, if there are any students out there looking for an interesting GIS-related project, this would be one that, following the methods laid out in Buschke et al., would be both relatively easy to conduct, and that could potentially play an important role in understanding how climate change is likely to impact marine mammals.

========================
Dr Colin D. MacLeod,
Founder, GIS In Ecology
========================

Paper Of The Week: Benito-Calvo et al. 2015. First GIS Analysis of Modern Stone Tools Used by Wild Chimpanzees (Pan troglodytes verus) in Bossou, Guinea, West Africa

15 Apr

GIS is an extremely powerful tool that can be put to a very wide range of uses in biological research, and this week’s paper of the week is a really nice example of just how wide this range can be. While most studies which use GIS seek to make maps of some kind or other, Benito-Calvo et al. (2015) use GIS to examine and model the surface of stone tools used by chimpanzees to crack open nuts.

This might, at first, seem like a strange thing to do, but when you think about it, examining the topography of the surface of a stone tool is no different from examining the topography of a mountain, it’s just done at a much, much smaller scale. This means that the same GIS tools that can be used to quantify the physical characteristics of mountains (such as the creation of Digital Elevation Models, and the calculation of slope and aspect) can also be used to quantify the shape of stone tools. This, in turn, can help identify patterns in wear, damage and other characteristics that can provide insights into how a tool was used.

Why is this important? Well, it not only helps us understand how our closest relatives use tools (something that not so long ago was thought to be a purely human occupation), it also provides a potential window into our own archaeological past. By using GIS to analyse stone tools from archaeological sites, and comparing them to what has been found in the analysis of chimpanzee tools, we can potentially get a much greater understanding of how tool use in humans developed over time.

So, if you’re interested in seeing how GIS can be used in biological research for purposes other than creating maps, then take a look at Benito-Calvo et al. (2015) as it’s a great example of this type of ‘outside-the-box’ usage of GIS.

========================
Dr Colin D. MacLeod,
Founder, GIS In Ecology
========================

Paper Of The Week: Loehle et al. 2015. Range-wide analysis of northern spotted owl nesting habitat relations

8 Apr

For my choice of paper of the week this week, I’ve selected one on species distribution modelling (SDM). SDM is becoming  increasingly  widely used in management and conservation, and it is often seen as a cheaper and faster alternative to conducting dedicated surveys to look for a specific species in hard-to-reach parts of the world. However, while species distribution modelling is really simple in theory, applying it in a biologically meaningful way can be very difficult. As a result, great care needs to be taken when running an SDM study, and it is important that the predictive ability of the model is properly and thoroughly validated in some way, preferably using an independent data set, before the predicted distributions are used for any conservation or management purposes.

It is for this reason, that I’ve selected Loehle et al. 2005. Range-wide analysis of northern spotted owl nesting habitat (Forest Ecology and Management, 342: 8–20) as the paper of the week. This paper takes an existing SDM for northern spotted owls (a threatened species from western North America) which was created to help conservation and management efforts and investigates it to see if it is actually suitable for this purpose. What they found as part of their detailed analysis was that the existing SDM model did not perform well when tested against an independent data set, and that caution should be used when using the SDM’s predicted distribution for conservation and management purposes.

Loehle et al. (2015) is a really good example of how model validation should be done as part of any SDM study, and it demonstrates nicely how a SDM study does not necessarily end when the initial prediction is made. If you are considering using SDMs for any conservation or management purposes, I would highly recommend reading this study to help you understand why independent validation is so important, and how it can be conducted.

========================
Dr Colin D. MacLeod,
Founder, GIS In Ecology
========================