January 18, 1998

Ethel D. Stanley Beloit College

 Oh Phlox!

A Visual Dataset


User Notes

Table of Contents

 What is a visual dataset?

 What kind of images are found in the dataset?

 What is the source of the Oh Phlox! images?

 Why was this visual dataset constructed?

 Why weren't more attractive phlox plants selected?

 How are individual images identified?

 How can I measure these images?

 How can I use this visual dataset?

What is the Oh Phlox! visual dataset?

The images in Oh Phlox! represent a visual dataset which was constructed to provide the means for extended visual analysis of a population of mature phlox plants. Oh Phlox! consists of phlox and ruler images. The phlox images can be explored through multiple means to gather both qualitative and quantitative data.

Most of the images we interact with daily have the potential for further study, but a visual dataset represents a thoughful selection process and is often annotated to provide additional information such as:

· physical charateristics of subjects in the image
· magnification or a scale bar
· identification of the source material
· description of the process by which the image was produced
· when the image was produced
· the reasons for the production of the image

Oh Phlox! has been described as "the beginning of a potentially rich set of data for students to investigate a myriad of questions." She continues:

"I can easily envision this data set being used with elementary school children as well as graduate students. The difference in what the investigation is and where the investigation will go will be dependent on the interests of the student investigators and teachers/professors and the teaching agendas of the teachers/professors."(Soderberg 1996)

What kind of images are found in the dataset?

Oh Phlox! includes PICT and GIF image files of individual leaves and whole views of 34 garden phlox plants. (Note: Only the aerial portions of the phlox plants were scanned, so the whole views exclude the roots.) The whole plant images are reduced to 60% of their original size; leaf images are 100% of their original size.

This visual dataset can be explored with NIH Image or other "graphics" packages like Adobe Photoshop or Picture Publisher. The PICT file format provides an easy interface with NIH Image Analysis. GIF files were also included to allow both Mac and PC users access to images like those encountered in web searches.

The image resolution of the GIF files is 72x72 dpi and saves on the memory required to manipulate the image. This does limit most examinations to macroscopic features, but this is not automatically problematic. (Ordinary "looking" deserves the same careful processing as things we see under microscopes.) Low resolution does limit the image quality for both examination and printing. For instructors and students who have the resources, it makes sense to assemble their own portfolio of images to address this limitation!

What is the source of the Oh Phlox! images?

These images were obtained by scanning fresh material from a population of garden phlox found in the city of Decatur in Macon County, Illinois during early October, 1995. These 34 plants were all the garden phlox plants found in a 6 foot by 6 foot area chosen at random from a 27 foot by 11 foot area that was undergoing proliferation of this species. The above ground portions of the plants were removed by cutting through the stems at approximately one inch above ground level.

Why was this visual dataset constructed?

I have often used fall specimens of common garden phlox to begin intro biology courses. Students are asked to describe evidence for as many biological interactions as they can using a single phlox plant from the collection. Documented images of mature phlox provide a rich visual dataset for individuals wanting to learn more about biology, images, or teaching/learning with visual tools. Looking at organisms in the field begins as a cursory process, but is likely to become a skillful one through repeated experiences. Working with a visual dataset may help provide much needed visual practice before and after students go to the field.

Why weren't more attractive phlox plants selected?

The plant subjects for these images were not selected on the basis of aesthetics! Every garden phlox plant in a small area was used. Why preselect botanical subjects on the basis of mere physical attraction for a dataset intended for learning about plants? Students often regard plants as passive, non-social organisms, therefore it is important to thwart the tendency to provide them with images of glasshouse specimens!

Images of plants in educational materials are often limited to only attractive, robust angiosperms in full flower. There is almost no trace of the daily interactivity between the plant and its environment. Like images of beauty pagent contestants, pristine plant images convey a false sense of what it means to be that organism. Let's say you are pursuing a degree in medicine. Only images of young, healthy adolescents are available for study, but your first diagnosis requires you to recognize symptoms in an aged adult. The images we provide or withhold direct learning.

It is essential to consider the limitations of images that are being used in the classroom. In the Oh Phlox! dataset, two compelling limitations are the "one-sided" views (upper surface primarily) and the "waist-up" views of the plants.

How are individual images identified?

Specimens were labelled by name and number.

Here are phlox plants #1 and #26


 Phlox plant #1


 Phlox plant #26


Phlox plants that did not fit within the scanning surface were divided and each portion was scanned in separately. Large plants are shown as two images. For example, the large plant #8 appears as PHLOX08A and PHLOX08B instead of just PHLOX08.



 Phlox plant #8 Part A


 Phlox plant #8 Part B


Leaves with obvious leaf miner damage were scanned in individually and were labelled with specific plant and relative position information. Examine the images below.



 Phlox plant #4 Leaf #1


 Phlox plant #4 Leaf #2


CP04LF1 tells you that the leaf image came from plant PHLOX04 and is the first leaf from the top to show leaf miner damage. (Note: It is also the youngest leaf to show leaf miner damage.) CP04LF2 is the second leaf to show damage. This system of labelling is actually quite simple. Try to answer this:

 How would this third leaf from portion A of the large phlox plant #10 be labelled?

How can I measure these images?

Ruler images allow you to measure the images. To measure the leaves, use the ruler image named l-ruler. The " whole" plant images have been reduced to 60% of their original size, so you must use the ruler image named p-ruler .




Use the appropriate ruler image with a phlox image to create a scale bar. An image processing application such as Photoshop can be used to combine the images electronically or you can simply print out the ruler and overlay it on a printout of the image.

How can I use this visual dataset?

Students can sample any of the myriad features of this population, develop hypotheses, and use statistical programs to support their ideas.

1. Standard measures of physical traits that can be done easily include:

- the number of leaves per plant

- the percent of leaves showing leaf miner damage

- the average surface area of the leaves

- total area of damage per individual leaf miner as an estimate feeding required by developing larva

2. Access to image analysis tools such as NIH image permits informative, but non-standard approaches to investigating the biology of the plant. For example:

- measure "green" or pixel density in new vs. "old" leaves

- observe leaf shape as evidence of nutritional deficiency

- analyze the variance of leaf form from "idealized" leaf using a field manual image for comparison

- construct a model of the "average" leaf to compare to standard phlox images from identification keys

3. Behavioral observations can include:

- examination of the directionality of leaf miner trails

- plant responses after "endoparasite" activity

- the interval nature of leaf miner "attacks" in some subset of this population.

4. The leaves can be compared with the plant provided as a sort of timesect or time gradient of growth with the oldest growth approximately at the soil line and the newest growth at the most distal points from the soil line (root tips and shoot tips of plant).

See also:

Stanley, E. D. 1996.

Taking a Second Look:

Investigating Biology with Visual Datasets.

Bioscene 22(3) 13-17.

 Return to Ethel Stanley's web page