Wednesday, April 18, 2012

Introducing the Dental Microwear Image Library

Dental microwear, seen in the tiny pits and scratches on a tooth, provides lots of detailed data for inferring diet and chewing behavior in animals. Analyses are often conducted by digitizing highly magnified images of the tooth surface and counting up and classifying the various microscopic features. Animals with a certain percentage of pits and scratches may have browsing habits, whereas those with another profile may be grazers. By measuring extant animals with known diet, we can (hopefully) infer the diet of extinct animals.

Dental microwear; modified from figure 3 in Mihlbachler et al., 2012
In this age of increasingly open science, microwear studies can be problematic. A cornerstone of science is reproducibility - yet, inter-observer variation and error can greatly affect measured data. Furthermore, one study alone may generate dozens or hundreds of images. Even if you wanted to re-analyze teeth, it's pretty tough - how could you get access to the necessary images? Ideally, we want a world where anyone can access the raw image data, make their own observations, double-check published analyses, and add new data for comparison.

Thus, a new project - called the Dental Microwear Image Library, or DMIL - may change things. Assembled by Brian Lee Beatty and Matthew Mihlbachler, the website aims to become a clearinghouse for dental microwear images. This will allow greater standardization of analyses and hopefully better interpretations of paleoecology and diet for extinct organisms and modern organisms. The first data (from a recent paper in Paleontologia Electronica) are now posted, along with many other data sets.

Brian Lee Beatty (who blogs at The Aquatic Amniote and tweets as @Vanderhoofius) was kind enough to answer a few questions about the DMIL. Thanks, Brian!

Was there a particular moment or incident that inspired you to build the DMIL? If so, what was it?
As we set out to test and develop the method that Nikos Solounias and Gina Semprebon started, we found ourselves frustrated by not only the lack of information on methods that were given in most microwear papers, but also the inability for people to check their work. Interobserver error is a major cause of problems for microwear, and the only way for anyone to be aware of those differences is if they compare interpretations of microwear surfaces, not just their numbers on a spreadsheet. The DMIL was the only possible solution to the need to share such images.

How has community response been so far? Is there any particular type of skepticism that you're working to overcome?
The DMIL hasn't yet come up against skepticism, but our first paper on this method that uses it has.

What license, if any, are the data housed under? Or is it on a case-by-case basis?
There is no license for the data. We want it to be completely open-access and simply available.

How would you envision the DMIL 10 years from now? What goals might you have for the long-term?
We hope it will be a place that people can use to learn how to use the methods we are continuing to develop. I most sincerely hope that it will not only be home to our own data, but also be a place for others to deposit their data using similar methods so that more work of this sort is available in a similar, comparable format.
Authors of the recent paper in PE, along with a research assistant. Photos courtesy of Brian Lee Beatty.
For more information, check out the DMIL, or read the recent paper (open access) about the work.

Citation:
Mihlbachler, Matthew C., Beatty, Brian L., Caldera-Siu, Angela, Chan, Doris, and Lee, Richard, 2012. Error rates and observer bias in dental microwear analysis using light microscopy. Palaeontologia Electronica Vol. 15, Issue 1;12A,22p.

No comments: