|Home | Archives | About | Login | Submissions | Notify | Contact | Search|
Copyright © 2003 by the author(s). Published here under license by The Resilience Alliance.
Go to the pdf version of this article.
The following is the established format for referencing this article:
Drew, C. A. and G. R. Hess. 2003. Online publication enhances integration of current research in the classroom. Conservation Ecology 7(1): r12. [online] URL: http://www.consecol.org/vol7/iss1/resp12/
A version of this article in which text, figures, tables, and appendices are separate files may be found by following this link.
Response to G. S. Cumming (2002). "Habitat shape, species invasions, and reserve design: insights from simple models." Online Publication Enhances Integration of Current Research in the Classroom C. Ashton Drew and George R. Hess
North Carolina State University
Integrating current research materials and issues into graduate courses provides students with exposure to emerging concepts and methods. New online journal formats that allow authors to include raw data and model code provide a unique opportunity to bring current research into the classroom. We developed a graduate-level landscape ecology assignment using data and code provided as appendices to an article in Conservation Ecology. Our assignment required students to engage actively with the published material, was positively reviewed by the students, and prompted valuable discussion.
KEY WORDS: inquiry-guided learning, landscape ecology, modeling, problem-based learning, teaching.
Published: June 13, 2003
During the fall 2002 semester, we developed and taught a graduate-level landscape ecology course that introduced the participants to this field by (1) familiarizing them with the jargon, (2) exposing them to the major conceptual foundations of the discipline, (3) relating these foundations to current issues in conservation and natural resources management, and (4) applying these concepts by analyzing landscapes to provide information that could be used to solve conservation and natural resources problems (Hess and Drew 2003). We designed the course using principles of inquiry-guided learning, which are based on the use of teaching techniques that advance learning through independent examination of questions and problems, rather than the presentation of knowledge in a lecture format. Under the guidance of faculty, inquiry-guided approaches promote a range of skills, including the ability to formulate good questions, collect and analyze appropriate data, present results, and formulate conclusions (Boyer Commission 1998, Faculty Center for Teaching and Learning 2003). In our course, lectures were minimal, designed to introduce concepts, and followed by hands-on activities that challenged students to take responsibility for their own learning and interact directly with the concepts that underpin landscape ecology.
The first two thirds of the semester covered the basic tools of landscape ecology, including landscape metrics, spatial statistics, and neutral models. In addition, the students developed an appreciation of connectivity and patchiness from the perspective of individual organisms. At this point, the students demonstrated a basic understanding of landscape pattern, the processes that generate spatial habitat patterns, and the importance of scale. We wanted to help them integrate their new landscape knowledge with their knowledge of population and community ecology and lead them toward an understanding of how landscape ecology provides important data for conservation management decisions. In particular, we wanted them to think beyond the simple "checkerboard" landscapes of earlier exercises.
While ruminating how best to tackle these teaching goals, we encountered Cumming's (2002) publication in Conservation Ecology. His article provided an excellent example of how simple models are developed to provide insight into landscape patterns and processes. Cumming used cellular automata models to examine interactions between habitat shape, species invasions, and reserve design. The simulated organisms varied in their mortality rate, number of offspring, and dispersal distance, and dispersed through linear, branching, or square landscapes. The model reported population size at the end of each iteration. Cumming examined the effect of varying habitat shape on the population size of an invasive species at equilibrium and the time to reach equilibrium. Appendices included with the article provided the Matlab model code and landscape files.
We modified the code to prompt the students for variable values, rather than requiring them to directly read and edit the Matlab code (see Appendix 1 for the edited code). This allowed students untrained in model development or Matlab syntax to experiment with different combinations of variables. Working in groups of three or four, students were required to run the model using their own ranges of variables and critically evaluate two of Cumming's conclusions (see Appendix 2 for details of the assignment). Based on their simulations, we asked the students to explain whether they accepted Cumming's conclusions without reservation, to dissect how and why each biological variable responded to changing landscape patterns, and discuss whether they would trust these model predictions in a real-world setting. They were to support their arguments with up to three graphs displaying the data they generated using the model. The stated objectives of the assignment were to:
As a simple reading assignment, this article would have given the students the opportunity to ponder the effects of different habitat shapes and critique the model assumptions. However, such assignments do not fully engage the student. As part of the inquiry-guided learning process, we purposefully left some of the assignment directions open-ended. For example, the students were not told which data to include in the figures or which style of graph to generate. This forced them to sift through the mountains of data that such a model can produce and select the most representative data series. We also did not provide a range for the model parameters, suggesting that the students instead review the paper or run short test simulations to determine the model's limitations. As the students worked on their assignments, different groups selected different variables to test and generated different sets of graphs. As a consequence, the results were not consistent among groups, prompting discussion of how models may be used and abused in landscape ecology.
Initial student responses included several requests for clarification of our expectations regarding the final product. As the students began to implement the model, they wanted to know how they should condense the large volume of data into just three graphics. They also wanted to know the "correct" values to input for each model parameter. To meet our inquiry-guided learning goals, we generally asked the students to discuss these concerns within their groups, come to a consensus, and document the decision in their narrative. All groups ultimately selected appropriate parameter values and summarized their results with three graphics.
The students did very well on this assignment, which contributed 15% to their final class grade. We graded according to specific criteria, with a maximum score for each criterion (Table 1). We eliminated two criteria during the course of the assignment. The "Good Graph" criterion was eliminated because of technical problems that prevented the students from accessing the Good Graph reference Website. We eliminated the "reasonable responses" criterion when we realized that the students were being doubly penalized by this question, because those who failed to understand the model or the landscape ecology principles could not provide reasonable responses. Overall, the students demonstrated a good understanding of the model and provided satisfactory graphical and narrative summaries of their observations (see Appendix 3 for sample grade sheets). The lowest scores (3.8/5 ± 0.7 SD) reflected the students' struggle to move beyond a local understanding of ecological processes to the broader landscape perspective.
Class discussion of the results was enriched by our approach because of differences in findings among the teams. It became very clear that research findings, even when everyone is using the same models to answer the same questions, can differ depending on the particular focus taken. It was also gratifying to see the students work to find the consistent patterns among their results. During the rest of the semester, we followed this assignment with other in-class and laboratory exercises that carried us further from the "theoretical model and abstract results," as one anonymous student put it, toward more complex and applied problems in landscape ecology conservation.
In addition to comments and questions received during the course of the assignment, we distributed a survey after the assignments were completed (Table 2). The class response was generally positive. Most students felt the assignment achieved the stated objectives. As seen below, specific positive comments expressed appreciation for the opportunity to experiment with a model and indicated that generating the graphs forced deeper consideration of the hypotheses and results.
"I feel this assignment ... was quite valuable. I found the assignment appropriately challenging. The questions asked in the assignment really made me think about the model and Cumming's conclusions."
"Overall I liked and learned from this assignment. I'd been wanting to gain some experience in modeling, and this was a nice introduction. A little time-consuming for this time of semester, though."
Criticism of the assignment focused primarily on the intentionally open-ended nature of the questions. Without exact, step-by-step instructions and limits, some students expressed frustration that the assignment was "not always completely clear." Others felt they would have liked to link the assignment to a GIS or other tool to visualize the landscapes and population dispersal patterns.
"It wasn't always clear what we were doing. I think it was confusing for people without [a] modeling background. I think there is still some conceptual connection to do for applications on real-life situations, but I learned a lot from this assignment."
"Would have been good to link actual reserve design (some sort of GIS project) to the results, but that would make the project much longer and more complicated."
Overall, we were satisfied that this assignment provided valuable academic and professional development opportunities that could not have been achieved through a simpler "read and critique" assignment. Based on the students' feedback, we will include this assignment in future offerings of our course and be alert for similar opportunities to adapt online publications to classroom exercises. The adaptation and testing of Cumming's model took about two days of effort. This seemed reasonable given the positive response, and matched the time commitment required to prepare more traditional laboratory and homework assignments. The simplicity of the model and the careful documentation accompanying the appendices facilitated the use of these materials and would be important criteria for selecting future publications.
Student comments were helpful for modifying the exercise for future classes. In this version of the assignment, we did not integrate an effective landscape or dispersal visualization tool. Such visualization would have been a valuable aid to understanding the population dynamics for several students. To address concerns of clarity, future versions of this assignment will expressly state the professional development goals to pre-empt confusion about the selection of parameter values and graphics. In particular, we would emphasize the importance of effective data consolidation and presentation, and the need to clarify broad, "unclear" questions by examining and focusing on the objectives.
Responses to this article are invited. If accepted for publication, your response will be hyperlinked to the article. To submit a comment, follow this link. To read comments already accepted, follow this link.
Thanks to the students of Landscape Ecology in fall 2002 for their participation and feedback, to Graeme Cumming for responding to our questions about the model and encouraging us to submit this note to highlight the value of the Conservation Ecology format to teaching, and to Virginia Lee for comments on an earlier version of this manuscript. Our efforts to design this course using an inquiry-guided philosophy were greatly enhanced through participation in a semester-long professional development seminar on inquiry-guided learning led by Virginia Lee at North Carolina State University.
Modified Cumming (2002) Code for Student Assignment
% Cellular automaton model - stochastic dispersal, reproduction and mortality
% Original program written in Matlab 5.3 by Graeme Cumming, November 2000;
% Published in Cumming, G. 2002. Habitat shape, species invasions, and reserve
% design: insights from simple models. Conservation Ecology 6(1): 3
% Modified by Ashton Drew in Matlab Student 6.0, September 2002 to add data entry
% prompts to serve as a teaching tool.
% To run this program in Matlab, simply copy following code then paste and save it as an
% m-file in Matlab.
% These commands read in x,y coords from landscape file
filename = input('What do you want to name the output file? >', 's');
landscape = input ('Select a landscape > ', 's');
fid = fopen (landscape,'r');
[xcoord,ycoord] = textread (landscape,'%f %f');
status = fclose (fid);
count = size (xcoord);
% Initialise the variables for the program
cell = zeros (count);
cell2 = zeros (count);
cell (500) = 1; % Starting value
dispdist = input('How far do organisms disperse in each step? >');
% dispersal distance in m
inc = input('How many offspring do organisms have in each step? >');
% number of 'offspring' per iteration
mortality = input('What is the mortality rate (x/1000)? >');
% death rate / 1000
area = ceil (3.141592654*(dispdist^2)); %max possible number of neighbours
neighbour = zeros (count(1),area);
reps = input('How many iterations should the model run? >');
% this is the number of iterations, or time steps, that the model runs
% first we write a matrix containing identifiers of all cells in dispersal range
for x = 1:count
match = 1;
for i = 1:count
distance = sqrt(((xcoord(x)-xcoord(i))^2)+((ycoord(x)-ycoord(i))^2));
if distance <= dispdist
match = match + 1;
neighbour(x,1) = neighbour(x,1)+1;
neighbour(x,match) = i;
% now begins the main loop of the program
trials = input ('How many times do you want to run the model? >');
for z = 1:trials %this is the trials, or repetitions, of the model
cell = zeros (count);
cell2 = zeros (count);
cell (500) = 1;
results = zeros (reps,1); % this is the matrix that holds the results
for r = 1:reps % iterations
for j = 1:count(1)
if cell(j) == 1
cell2(j) = 1;
for m = 1:inc % remember "inc" is offspring, short for "increase"
rnum = rand * neighbour(j,1);
p = ceil(rnum);
p = 2;
plusone_id = neighbour(j,p);
cell2(plusone_id) = 1;
cell = cell2;
cell2 = zeros (count);
for k = 1:count
if round(rand*1000) < mortality
cell(k) = 0;
tally = sum(cell);
results(r,1) = tally;
% ends the main loop
if z == 1
final = results;
if z > 1
final = cat(2,final,results);
Assignment Based on Cumming (2002) Article
Habitat Shape, Species Invasions, and Reserve Design Exercise
Read: Cumming, G. 2002. Habitat shape, species invasions, and reserve design: insights from simple models. Conservation Ecology 6(1): 3. [online] URL: http://www.consecol.org/vol6/iss1/art3/
Until now, most of our example landscapes have been squares. What happens when our landscapes are arranged in more realistic shapes? In this article, Cumming uses models to explore how habitat shape can influence the dispersal and abundance of invasive populations. We will use the Matlab code provided in this article to modify and reproduce Cumming's cellular automata experiments.
Products to Turn In for Grading:
Graphs: You should develop three graphs (dispersal distance, number of offspring, mortality rate) that demonstrate the effect of shape on population growth.
Narrative: A review of the model, based upon your observations. The review should incorporate your answers to the above questions. Use the figures to support your arguments where appropriate. Three page absolute maximum, Times font, 1.5 line spacing, 1" margins all around.
As we grade this assignment, we are primarily looking for evidence that you evaluated the landscape – organism biology interactions and thoughtfully evaluated the observed trends within the context of landscape ecology principles. We will also be grading your ability to present a clear analysis of the observed trends both in graphical and written format.
Did you produce the required graphs?
Did you follow the Good Graph criteria?
Did your graphs clearly illustrate the relationships between habitat shape and each variable?
Did you fully respond to each question?
Are your responses reasonable?
Do the responses give evidence of consideration of landscape ecology principles?
Do the responses give evidence of clear understanding of the model?
Is the document well organized with good grammar and sentence structure?
Has the length limit been respected?
How to run the model:
Copy the cellaut.zip file from the class calendar to the desktop. Unzip this file to obtain a folder named Cellaut which contains five files. (Matlab works from one working directory – if you move files around, Matlab will no longer be able to find them.)
Open Matlab. Reset the Current Directory to be the Cellaut folder by clicking the "..." key near the top right. You can now run the model simply by typing cellaut at the Matlab command prompt.
You will be prompted to enter the following information:
What do you want to name the output file?
Create a unique name for the output file. The output file will be an ascii text file that you can open in Excel. (e.g. stream0_off2 or stream0_mort3)
Select a landscape.
Input the name of a landscape shape file. (e.g. stream0.txt). Your landscape files are: stream0 (linear), strm10c (complex branching with 10 nodes), strm20c (complex branching with 20 nodes), and grd40x40 (square)
How far do organisms disperse in each step?
Always enter 3, unless you are evaluating the effect of this parameter.
How many offspring do organisms have in each step?
Always enter 2, unless you are evaluating the effect of this parameter.
What is the mortality rate (x/1000)?
Always use 300, unless you are evaluating the effect of this parameter. (This is equivalent to a mortality rate of 0.3).
How many iterations should the model run?
You can use the figures in Cumming's article to gauge how many iterations you will need to allow for the model to reach equilibrium. **** After you enter this value, you will need to wait patiently for a few minutes while Matlab runs through some functions. *****
How many times do you want to run the model?
Enter 1. Since this is a homework activity, you will only run each model once. If you were interested in a thorough evaluation of these questions, you would run many trials of each model and use the average. After you enter this value you will watch numbers scroll by on the screen. These show the iteration numbers so you can gage how long it will take the model to run and ensure that it is in fact running.
When the model finishes running (the numbers stop scrolling by and command prompt returns), open the output file in Excel. In Excel you can graph the results to observe how the population behaved under the conditions you specified and verify that your model did reach equilibrium.
If at anytime you want to interrupt the model, just hit Ctrl-C. This will give you an error message that tells you where you interrupted the program, which you can ignore.
It may help you to create a chart similar to this for each variable (offspring, movement, and mortality) to keep track of your data:
Samples of Instructors' Grade Sheet with Comments