Andrew Scott- GIS

From BenningtonWiki
Jump to: navigation, search

Return to the GIS Project Ideas main page


Final Writeup

Introduction

Ganoderma applanatum, commonly known as “artist's conk”, is an annual shelving fungus, most commonly saprobic, but when parasitic, causes white rot in hardwood trees, typically leading to death within a few years after initial infection6.This initial infection is caused by the proliferation of G.applanatum in wounds of the bark, most commonly in beech, poplar and, as it occurs most commonly on campus, maple. During the course of this project I mapped the local of several G.applanatum specimens (n=16) and analyzed their locations in the context of seeking ecological correlations using GIS software. My question in regards to designing this project was relatively simple : what ecological factors were most (if at all) predictive of G.applanatum - soil type, forest composition, soil quality, land use, etc. I made almost no concrete predictions in this pursuit - there is very little ecological data about G.applanatum to base such predictions off of and thus predicated only that I would see increased occurrence of G.applanatum fruiting bodies where there was an increase in suitable hosts. This was not, for the most part, the question that I answered in my project. The realities of data resolution and the small scale of my project required a reexamination and a slight shift in the aim of this project. Instead of looking at larger scale factors such as soil or using nationwide forest data I utilized the recent Bennington forestry survey data and examined factors that would likely be related with presence of absence of fungal infection (in the case on the presence of fruiting bodies) - primarily stand age, size and biodiversity2,3,4. Age and size have previously been shown to be strongly correlated to susceptibility to fungal infection5 : an older, larger tree is more likely to be wounded over time, providing a pathway for possible infection. In this project, as exact age of stands is not available, basal are and biaomass were used as indications of age as both a strongly correlated stand age. Biodiversity has also been correlated with infection frequency , which is intuitive, as a stand trends to monoculture a pathogen has more possible hosts, more chances to infect.


Methods

In this study four interactions were measured for correlation to presence of G.applanatum: species richness, sugar maple basal area (nearly all specimens were recorded on sugar maples), total basal area and total biomass of a given plot. To carry out this project I first recorded the locations of G.applanatum specimens using a hand-held GPS unit. Additionally, my analysis required a number of data sets acquired form various sources, primarily the Vermont GIS Data Warehouse and the USGS National Atlas. From the Vermont Data Warehouse I pulled a recent satellite map for reference, as well as several soil related maps - soil quality primarily, which went unused in my final analysis because my entire sample range fell almost entirely within one level of soil quality. From the USGS I employed a nation land cover data set.

The data set which provided the most important piece of my project was from the recent Bennington forestry survey. These data were then manipulated for the purposes of my project. I qualitatively linked each of my noted G.applanatum specimens with the nearest 500sq meter plot recorded from the forestry survey, and assigned each plot a simple species richness score based on the number (2-7) of unique species recorded in the survey. The modified data were then entered into QGIS and statistics were computed with Graphpad Prism 5. It was necessary to create several versions of the forestry survey data in order to display multiple facets simultaneously.

Results and Discussion

Statistical analysis showed that none of the factors I examined were correlated with the presence of Ganoderma applanatum. Species richness appeared to be trending towards significance (p= .18) (figure 1), as did total biomass (p= 0.1) (figure 2) , but the same could not be said for either sugar maple basal area (p=0.39) (figure 3) or total basal area (p= 0.45) (figure 4). I suspect that a larger sample would further refine these results and possibly lead to significance as the minute number of specimens makes any definitive statistical analysis nearly impossible.

Visually, it would appear that species richness is most predictive of clusters of G. applanatum (figure 1) as the largest concentrations of fruiting bodies are found in areas of very low species richness. I suspect that this trend would be more significant with a more appropriate sample size.

An additional confounding factor to consider is the nature of most of the G. applanatum specimens recorded : they were found primarily, though not exclusively, on already dead trees. This could muddle the use of basal area and biomass as predictors of G. applanatum infection as they have a somewhat paradoxical relationship with the fungus: an older, bigger stand should be more likely to susceptible to infection, but a stand which has been infected would be predicted to have a lower number of possible hosts, and therefore a lower basal area and biomass as compared to stands without infection. Thus we might expect that in stands which have already been infected biomass and basal area would decrease as the presence of G. applanatum increases, but in stands uninfected, or lightly infected stands we would expect biomass and basal area to increase as presence of the pathogen decreases. It would be more helpful, though out of my scope, to examine the change in pathogen distribution over time - ie: in five years I would expect that the areas with the highest biomass and sugar maple basal area today would be more likely host the largest G. applanatum populations, and the areas with higher rates of infection today would have markedly lower rates as the pathogen has fewer viable hosts. With this in mind a time based study is an obvious do-next. Additionally, it may be helpful, or even more informative to create a risk map1 from the given forestry data, combining biodiversity and host-species basal area to predict the areas which would be most prone to infection. This kind of map/data, when paired with a time based element of study could prove to be rather informative.

A third consideration would be for something that would be relatively simple to do though, though likely fruitless in the context of my sample size, would be to examine the spatial relationship between G. applanatum specimens - that is, to see if there is any relationship based on proximity to another specimen. A glance at figure 1 indicates that there may indeed be some clustering effect, but I did not examine this, though it might prove to be a fruitful area of inquiry. With that said a very through, and more methodical sampling nature would likely be required. In gathering my initial data I missed several samples in initial passes through areas because they resided over 15 feet from ground level, and were therefore easy to overlook. To thoughtfully examine a purely spatial relationship between fruiting bodies a great amount of attention would be required to provide a clear picture and accurate representation of the population.

Conclusion

While I did not find any particular relationship between any factors examined in this project and G. applanatum location, this project does display the possibility, and useful nature of GIS software for approaching and examining pathogen populations. I believe a similar approach to mine, on a larger scale, and when combined with a temporal element may well prove to be highly informative, and may provide a method for accurately predicting the spread of G. applanatum across Bennington College’s campus.


Figures

Figure 1
Figure 2
Figure 3
Figure 4


Works cited

1. Nelson, M.R., Orum, T.V., Jaime-Garcia, R. & Nadeem, A. Applications of Geographic Information Systems and Geostatistics in Plant Disease Epidemiology and Management. Plant Disease 83, 308-319 (1999).

2. Mitchell, C.E., Tilman, D. & Groth, J.V. EFFECTS OF GRASSLAND PLANT SPECIES DIVERSITY, ABUNDANCE, AND COMPOSITION ON FOLIAR FUNGAL DISEASE. Ecology 83, 1713-1726 (2002).

3. Conner, R.N., Rudolph, D.C., Saenz, D. & Schaefer, R.R. Heartwood, Sapwood, and Fungal Decay Associated with Red-Cockaded Woodpecker Cavity Trees. The Journal of Wildlife Management 58, 728-734 (1994).

4. Carlsson, U., Elmqvist, T., Wennstrom, A. & Ericson, L. Infection by Pathogens and Population Age of Host Plants. Journal of Ecology 78, 1094-1105 (1990).

5. Holmgren, J. Prediction of tree height, basal area and stem volume in forest stands using airborne laser scanning. Scandinavian Journal of Forest Research 19, 543-553 (2004).

6. Campbell, W.G. The chemistry of the white rots of wood. Biochem J 26, 1829-1838 (1932).



Original Ideas

  • This is in many ways similar to both Kayta and Mara's questions, but specifically concerning the map of tree populations from the early 70s (that Kerry mentioned in class) what changes have occurred in distributions over the last forty years, and what, if any, are the geographic and environmental correlates that correspond with changes...land use could be one, but I'm more interested in soil types, previous populations, shade nutrient availability, densities...etc. I could see updating these maps as a wonderful project.
    • Okay -- interesting things to be interested in -- but still a bit amorphous? Maybe choose one or two of these things that can be most effectively looked at from geospatial perspective AND would have readily available or create-able data-sets and focus there?Kwoods 00:01, 3 October 2011 (UTC)
  • Again, very similar to Joe's idea, but I would be interested in mapping infected (by some to be determined pathogen) plant populations to the end of attempting to determine what environmental factors are most predictive of a pathogen's range/success/virulence.
    • I like it, but am not sure of the viability in terms of data availability; hard to get thorough mapping of pathogen effects at local or regional scales and, while you might find broader-scale maps of occurrences of pest/pathogen, it would be hard to find corresponding environmental data with appropriate resolution. Would be interesting to think about exactly what kinds of data you'd need, thoug.Kwoods 00:01, 3 October 2011 (UTC)
    • Mapping entire infected plant populations might be prohibitively difficult, but you might be able to address the question in a similar manner to the first paper we read-- sample a handful of areas on campus, look for a specific pathogen (maybe G. applanatum, which we've already got some data on from the biodiversity project), and try to find factors that contribute to differences in distribution. The difficulty with this would be dealing with all the statistical noise- really obvious factors, such as the presence of viable hosts, would be far more significant than factors such as soil quality, land use, etc.

Jkendrick 14:16, 27 September 2011 (UTC)

      • I'm kind of attached to this project idea, so bringing it down to a level that's do-able for this class is a priority...I like the idea about looking at something like G.applanatum, one because it's easily detectable year round, but there are, as always, problems, not just what Kerry mentioned, but there also isn't a whole lot of literature on it as a pathogen. Ascott2 13:50, 4 October 2011 (UTC)
  • This is a bit obvious in the wake of Irene, but inspired by my harrowing experience being evacuated from the Hampton : What areas of Bennington and North Bennington are most prone to flooding, and what preventitive measures could be taken to lessen environmental and property damage in the event of further flash flooding?

Ascott2 02:11, 20 September 2011 (UTC)

    • Well, second part might not be directly a GIS thing, but first part certainly could be -- but now put in terms of WHAT YOU'D NEED TO PUT TOGETHER to get at it (also, see Ella's ideas).Kwoods 00:01, 3 October 2011 (UTC)

Project

  • More concrete idea

After reviewing the available data sets it does look like there's enough data, at a high enough resolution to go trough with the idea of mapping pathogen populations. As it stands I'm inclined to look at G.applanatum as per Joe's suggestion, because it's relatively widespread on campus and easily identifiable, and because it is known to be parasitic on both conifers and hardwoods. At the moment I am looking to map at least two distinct populations (this should be possible based on my observations over the course of the Biodiversity project this term), and examine the pathogen populations in relation to environmental variables : forest type, soil type primarily, possibly over all biodiversity of the selected areas (might require more data collection than can be warranted in the context of this class). What I expect, though is that when calculated the data will show that the pathogen, despite being a generalist, thrives in certain hosts, which are more common in area X because of Y more than pathogen is successful in area X because of Y - separating the two could be tricky, but do-able, or so I think now.

  • Cool project. It will be interesting to see how this pathogen looks on a map. So you are going to make your own shapefiles right??
    • That's the plan.
    • Annotated Bibliography
      • Nelson, Merritt R., Thomas V. Orum, Ramon Jaime-Garcia, and Athar Nadeem. "Applications of Geographic Information Systems and Geostatistics in Plant Disease Epidemiology and Management." Plant Disease 83.4 (1999): 308-19. Print.
        • This is just a basic, older introduction to the uses of GIS and geostatistics in mapping pathogen populations for agricultural purposes. They're working on a much larger scale, but some of the methods discussed are relevant to my interests.
      • Plantegenest, M., C. Le May, and F. Fabre. "Landscape Epidemiology of Plant Diseases." Journal of The Royal Society Interface 4.16 (2007): 963-72. Print
        • This is another agriculturally focused review, but it really gets at the heart of the questions I'm interested in for the purposes of this class: what landscape features have the greatest effects of pathogen success/dispersal/etc, on a large regional scale and a smaller scale : local diversity or lack thereof, local landscape characteristics and so on.
      • Ostfeld, R., G. Glass, and F. Keesing. "Spatial Epidemiology: an Emerging (or Re-emerging) Discipline." Trends in Ecology & Evolution 20.6 (2005): 328-36. Print.
        • Another review. This one is rather similar to the first, looking mostly at human disease and the applications for spatial mapping in prevention and risk assessment. One of their figures, Fig.1 is an almost perfect representation of what I want to do, roughly, aside from the fact that on this small a scale it is unlikely that I can create a risk map (especially when there is so little data describing the tendencies of this pathogen - these are the data that I am looking to build), but the it's a good model to follow otherwise.