Interview your friendly Ecologist/Archaeologist

trinker

ggplot2orBust
#1
While helping Bugman and gianmarco with respective problems I became interested in the work they do. I love visualization and I love geographical plotting. But I'm in a field (literacy) that really isn't about geo-spatial representation of data. But then I had a though. I remembered seeing this Deb Roy video about the birth of a word. And I thought really why can't I do this in the classroom? I've started working on a means of doing some of what Deb Roy does in this package called embodied.

Now for what I'm after. I know ecologists and archaeologists do cool spatial analysis but I don't know exactly what you do. I have a feeling you see what deer eat what kinds of foods in what areas but that's my limited understanding. I know there's a spatial component but there's probably a statistical component as well.

Could you all who are in these fields be so kind as to give your take on:

  1. Overview of what you do
  2. The types of research questions you may investigate
  3. Spatial analysis and other visual representations you use
  4. Types of statistical tests you use and how you might use them with your data
  5. Anything else you'd like to share...

Given I see the class room as a ecosystem with particular resources arranged in particular ways, with particular traffic patterns and higher density of things like language use, questions, answers, cross talk, disengagement etc.

  1. How do you see your work connect to what I might observe in an educational setting?
  2. If you suddenly became me and had to observe students in the classroom what tools you have now would you use to better understand how space, engagement, language, and proximity and intersect?

Videos, papers, links, images are welcomed.

Thank you in advance.

EDIT A fake animated video of motion paths
 

bugman

Super Moderator
#2
Trink, I love that you are interested enough to post this!

I'm going to post in more detail very soon to your questions. Right now its late and I need to think about how best to respond!
 

bugman

Super Moderator
#5
1.

I am an aquatic ecologist. Not a researcher, but I guess a practioner. I have a background in mosses and popualtion genetics and some taxonomy experience. I was at a cross roads whether to pursue a PhD in botany or pursue my obsession with rivers and bugs. So left the PhD dream and went for the later.

Most of my work focuses on biossassments and water qulaity monitoring. And most of our projects are dealing with point source imapcts of some description in river systems. We sometimes get projects dealing with impact assessments from. the EPA

2.
As above, not really research per se, but we usually evaluate ecological risks fo certain infrastructure projects, WWTPs and threats to native species etc.

3.
I use arcgis mainly, but thanks to bryangoodrich I have started palying with qgis
qgis, and now also use qmaps package amoungst others in R.
And I have also found it useful to colour code rivers in a given area based on thier flow regimes and basic water quality characteristics. This has been a useful tool for a) a quick fire site selection tool and has also helped non technical people understand how the watershed is organised in terms of these physical and chemical characterisitcs. I also have plotted monthly water quality data on a monthly scale in small resovoirs to track temporal changes. This is also done with fish densities and we are loking at now doing habitat preference maps for turtles.

4.

Luckily or unluckily the type of data I have varies greatly. I have tons of time series data from my water qulaity data and because it ecology, my bug data is usually ,multivariate, unbalanced, skewed, bionomial, continous, counts and pretty ugly.

I use clustering techniques a lot and PERMANOVA for horribly complex data sets. Canonical Correspondence Analysis, Principal Coordinates Analysis and now I have recently started reading u up on Intervention analysis and fourier series analysis (without much joy as of yet). I have also used Boosted Regresstion Trees to predict habitat preferences for some fish species. CART techniques are used (for the fish and turtle work) and I have started analysing some algae data using SEMS.

Jake has helped me massivly with GLMM and mixed models in general and I use the r pacakge lme4 extensively – our most common design is where we have and upstream and downstream location (fixed factor) with sites nested in locations and replicates nested in site. These are either modelled using a gaussian or bionomial distribution.

I have also found this to be very useful for learning about GLMMs

Jack of all trades, master of none.


More to follow...

I need to think a bit more about the last two questions and how it relates to yours trinker...
 
Last edited:

trinker

ggplot2orBust
#6
@Bugman, thanks I can see similarities already, computer station = revers, or ecological risks fo certain infrastructure projects = risk to authentic dialogue or engagement.

Please keep it coming. (PS links are broke for me)
 

gianmarco

TS Contributor
#9
While helping Bugman and gianmarco with respective problems I became interested in the work they do. I love visualization and I love geographical plotting. But I'm in a field (literacy) that really isn't about geo-spatial representation of data.

Now for what I'm after. I know ecologists and archaeologists do cool spatial analysis but I don't know exactly what you do. I have a feeling you see what deer eat what kinds of foods in what areas but that's my limited understanding. I know there's a spatial component but there's probably a statistical component as well.

Could you all who are in these fields be so kind as to give your take on:

  1. Overview of what you do
  2. The types of research questions you may investigate
  3. Spatial analysis and other visual representations you use
  4. Types of statistical tests you use and how you might use them with your data
  5. Anything else you'd like to share...
Thanks @Trinker to create such an interesting thread, devoted to ecologists and to those strange kind of scientists (if they may be defined so) called "archaeologists". I hope I will be able to provide some answers to your questions, provided that: 1) my English allows me to make my points; 2) my English allows me to well understand the main points of your questions.

Overview of what you do
I believe that the majority of people does not suspect that statistics can be something that archaeologists actually use. On the contrary, quantitative approaches have steadily gained popularity since 1960s, since American archaeologists started a revolution in the field of archaeology with that which would have been named Processual Archaeology or New Archaeology. This new wave of archaeological thinking introduced into archaeology, among other thing, the use of statistics and of a "scientific" methodology. Many things have changed since than, and New Archaeology has been criticized form many reasons, but I think that the focus on the utility of quantitative methods has remained alive into the discipline.

As for me, I concur with many scholars (i.e., archaeologist) that archaeology, like other social sciences, inescapably has numerical components, and that one cannot escape statistics, provided that the use of the latter and the application of statistical approach to the archaeological documentation are faced with some inherent difficulties.

The types of research questions you may investigate
Overall, I think that there are as many research questions I may investigate as are the questions that may stem in my mind. These, in turn, are stimulated by the theory(ies) and theorethical frameworks one can read from literature. For this reason, I think it is of utmost importance to read a lot, not only what is closer to the own field(s) of inquiry, but also what regards other geographical, cultural, chronological contexts. For example, even though I am specialist in Italian Bronze Age, I benefitted from reading about Neolithic China, Bronze Age Greece, Mongolian Steppe, pre/post contact North and South America, and the like.

Types of statistical tests you use and how you might use them with your data
In recent years, the use of statistics in my studies has steadily increased and has been centered mainly on the analysis of categorical data (i.e., cross-tabulation) and, to a lesser extent, on descriptive statistics, simple hypothesis testing (t-test, Mann-Whitney, 1-way ANOVA, Kruskal-Wallis), correlation/regression and, sometime, on the use of bootstrap approaches.

The predominance of the interest in the analysis of cross-classified data should come with no surprise since in archaeology things are often put into categories. So, it is natural for archaeologists to deal with cross-tabulations. And it has become natural for me to extensively use Correspondence Analysis as an exploratory tools for seeking pattern(s) of association in large contingency tables. I have also, as @Trinker now, created a website on the use of that technique in archaeology. The site has gained a good popularity, with accesses from all over the world (also thanks to Trinker's post on his blog). I coupled Correspondence Analysis also with Cluster Analysis. I am familiar also with other scaling techniques, like Principal Component Analysis and Multidimensional Scaling, even though I did not happen to actually use them.

Since 2011, I got involved into Bayesian approaches to radiocarbon chronology. This does not mean that I have deep knowledge of Bayesian computation. Quite fortunately, there are out there two on-line facilities (namely, OxCal, BCal) which actually perform all the needed calculations once the data have been fed into them. In a nutshell, I happen to use Bayesian approaches with the aim to derive posterior distributions of radiocarbon dates once some prior constraining information are available (e.g., stratigraphy). I have published three articles dealing with such an approach, in 2011 (Radiocarbon), 2012 (Journal of Archaeological Science), 2013 (Journal of Quaternary Science).

I have also recently approached Logistic Regression (especially binary LR), and I have found it very interesting. As always, I approach statistics from a practical standpoint. Since I do not have an extensive math background, I "limit" myself to grasp the principles on which statistical techniques are based, understand the contexts and situation in which they are used, understand the outcomes and results, and how the latter can be used to gain knowledge from data. I have also recently dared to write a journal article that uses Binary Logistic Regression, whose revised version is currently under evaluation.

Spatial analysis and other visual representations you use
Like Bugman, I have used GIS for my studies, and I have recently turned to open platforms like qgis. As one can imagine, tying quantitative information to spatial data is essential to archaeology. Recently, I have played with the possibility to use qqplot2 to build some spatial representation of a dataset I am studying, in an attempt to graphically visualize the distribution of functional classes of objects within a Bronze Age settlement here in Sicily.


That's all for now….
I have to think some more about your other questions.

Cheers,
Gm
 

bugman

Super Moderator
#10
@ Here's an example of what I'm working now that is related but those heat maps are the next goal and then apply some of the techniques you all describe here:
This must be the sentence drawing I read about on your blog? How do you interpret this?
 

trinker

ggplot2orBust
#11
@bugman not the sentence drawing. I'm not sure if sentence drawing is a great way to show data. It was more of an exercise of can I do this. What I have here is made up data of individuals' movements within the room. I'm interested in that movement as related to dialogue us (amount and quality) . And in turn how that effects self reported engagement levels. So right now this is me developing my skills to show the motion paths of participants in a similar way to the "worms" Deb Roy discusses. In your contact this may be migration patterns.
 

bugman

Super Moderator
#12
very cool. this would also be like temporal trajectories of bug communities in ordination space. this would be perfect -,the animation part - for a forthcoming presentation that I have to give.