There are two datasets (plus one optional Bonus Dataset) with which students can work. The one(s) you choose will depend upon your goals for this exercise, available time, and your students’ skills with spreadsheets plus their level (e.g., sophomores versus seniors). Each dataset is explained below along with suggestions on how to use the data. In addition, you may choose to have students complete a writing exercise involving critiques of the Cedar Creek and other biodiversity experiments (the exercises for Datasets 1 and 2 should be completed before attempting this writing exercise). The ideas behind these critiques are also discussed below.
Dataset 1: contains average plant biomass and standard errors for each species richness level. These data show an increasingly positive relationship between species richness and productivity over time.
Dataset 2: shows the percentage of plots that exceeded the maximum monoculture biomass. This second dataset allows a test of the sampling effect hypothesis (as stated below), and finds that the initial years are consistent with the sampling effect hypothesis while three of the last four years are not.
Bonus Data: presents the raw data (as opposed to the averages presented in Dataset 1) and presents information on the presence/absence of functional groups. Graphs from these data are rather cluttered and difficult to interpret, illustrating why we use the averages. There are a large number of analyses that are possible with the functional group data. We have generally found significant effects of both C4 grasses and legumes and of their interactions. We have included this richer data set as an opportunity for teachers and students to be creative and have therefore left the exercise quite open-ended. Teachers electing to use these data should determine what exactly they want students to do with the data and provide more explicit instructions than we do here.
Species List: shows the functional groups, genera and species, plus common names of the plants used in the experiment.
As explained in the Overview, the key issue addressed here is the relationship between species diversity and productivity as a measure of “ecosystem function.” The history and importance of this research is explained on the Cedar Creek website (http://cedarcreek.umn.edu/research/biodiversity.html) which also leads you to numerous papers and figures (see also Tilman et al., 1996, 1997, 2001, listed below).
You can use these data in a basic ecology class so that students better understand why the diversity/productivity relationship is of so much interest. Having students work with this dataset is a good preparation for a lecture on species diversity; you could also use it as a follow-up on a lecture about this topic.
From nature shows and other sources students may believe that high species diversity is “good,” but they likely could not explain why this might be so. “Ecosystem function” can be a pretty abstract idea, so working with and discussing these data can help to make that idea more concrete. The Resources (see links below) section of this TIEE dataset provides links to numerous Web sites on species diversity. You can use these to get ideas to present in class. (There are also websites about grasslands in the Resources section if you know little about grassland ecology.) Alternatively, you can ask students to look at one or two Web sites for homework; this would be a useful set-up for discussion following the lab.
These data also show students the importance of long-term data and you can discuss this with them using the context of data interpretation and experimental design (e.g., how can scientists come to similar or different conclusions using similar experiments if the time course is different?)
The Excel directions are fairly explicit, but anticipate the time your students will need to make their figures. (To save time in class, you can have them work on Excel for homework or give one of the Excel tutorials in the Resources section ahead of time.) As a note, briefly explain what a standard error is and why it is useful.
The student questions (repeated below) emphasize data description and interpretation. You can use these as discussion points, homework essay questions, and/or test questions. Depending on your students’ experience with data analysis, you may need to help them see all the details in the figure they developed (see question 2). The third question can stimulate discussion about the trade-offs of controlled ecological experiments and foreshadows issues discussed further in the “critiques” section.
These data are the next “step up” from the first dataset. They move students from describing the observed pattern to testing a hypothesis about the mechanism behind a pattern. This dataset is appropriate for a more advanced ecology class or with basic ecology students if you have time to discuss the issue with them.
As the Student Instructions section explains, the sampling effect hypothesis suggests that diverse plots have more biomass because they are more likely to contain a dominant species with high biomass. Under this hypothesis, one or a few high biomass species are good competitors and dominate the plots in which they are present. Consequently, the total biomass of communities containing these species is predicted to be the same as the biomass of these species when grown alone. In this scenario, there should be an increase in the average biomass across diversity gradients, but no increase in the maximum amount of biomass across diversity gradients.
The questions and answers for Dataset 2 are discussed here.
The answer depends upon the year. Specifically the years 1996, 1997, 1998 and 2001 show no change in the percentage of monoculture-biomass-exceeding plots across the diversity gradient. The years 1999, 2000, and 2002 show an increase in the percentage of monoculture-biomass-exceeding plots (P < 0.05 for all three years).
Note that the percentage of plots that have a higher biomass than any of the monoculture plots is sensitive to sampling variation in monoculture biomass (i.e., some years may have unusually low or high values for monoculture biomass). Therefore, it is important to test for biodiversity effects by asking whether this percentage increases with biodiversity, rather than whether this percentage differs from zero. For example, the data from the year 1996 show that the percentage of plots was significantly greater than zero (P = 0.003), but this percentage did not increase with biodiversity (P = 0.45). This may be explained by sampling error causing a low monoculture biomass for that year rather than by a biodiversity effect.
These data show, as did the data in Dataset 1, that productivity increases with species richness.
In the early years, the percentage of plots that have higher biomass than any of the monoculture plots was constant across species richness, but this percentage increases with species richness in 3 of 4 later years.
There are a number of possible answers to this question. The most common explanations for this phenomenon are:
The Sampling Effect Controversy
The topic of Sampling Effect has been quite controversial (Huston 1997, Wardle 1999, Kaiser 2000, Wardle et al. 2000, Loreau 2001). You may want to have a classroom discussion based on students’ answers to Question 3: “What does this suggest about the effect of the loss of biodiversity in natural systems? What are the problems that need to be considered when extrapolating the results of this experiment to natural systems?”
Emphasize to students that they are working with real data from an experiment at the heart of this controversy. The following material may be useful in presenting some of the issues raised by this and similar studies. You may choose to handout the following critiques and responses and have the students write a one page argument in favor of one of the critiques or responses. Students can also debate the issue by choosing one side or the other (Wardle et al. 2001 and Kaiser 2000 are quite readable for less advanced students).
A number of critiques have been leveled towards biodiversity experiments. Three critiques and responses are summarized below.
Critique: The positive relationship between plant species richness and productivity can be attributed to the higher probability of one or a few high biomass species occurring in diverse plots. This assumes that these high biomass species are good competitors and dominate the plots such that the total biomass of communities with these species is essentially the same as the biomass of these species when grown alone. In this scenario, there should be an increase in the average biomass across diversity gradients but no increase in the maximum amount of biomass across diversity gradients.
Response: Such a pattern was observed initially, but later years show that the maximum biomass increases across the diversity gradient. Thus, this long-term experiment suggests that the mechanisms behind the biodiversity-productivity relationship may have changed over time.
Critique: If a positive relationship between biodiversity and productivity is due to a “sampling effect” and can therefore be attributed simply to a higher probability of being present at high diversity, this relationship should be considered an experimental artifact with no relevance beyond the experiment.
Response: Because by definition diverse communities contain more species, the higher probability of any given species occurring in diverse communities is a phenomenon that could have effects in many habitats outside of experimentally assembled communities.
Critique: Experimentally assembled communities with a design of random species loss cannot predict the effects of real species loss, which will not be random.
Response: While real species loss will be biased, it is difficult to predict how it will be biased. Rare species are vulnerable to extinction, while disease may cause abundant species to become functionally extinct (e.g., Chestnuts) and nutrient pollution may eliminate slow growing or high root-to-shoot ratio species. Given this uncertainty, it is reasonable to first address the effects of random loss. An interesting question for future research is whether biased species loss will show similar effects as the random species loss simulated in this experiment.
This dataset contains the “raw” data from the first three years of the experiment. These data were the source of the averages and the standard errors in Dataset 1 and can be used if you want student to calculate averages and errors. This would be a good idea in a more advanced ecology class.
As explained above, this richer dataset will allow teachers and students to be creative and the data work-up to be open-ended. These data would be appropriate for students in a plant ecology course or other advanced ecology class. For undergraduates or beginning graduate students, figure out what you want the students to do with the data and give them clear and specific instructions.
The Student Instructions provide guidance on using this dataset to assess whether an increase in biomass with increasing diversity may be due to presence of particular functional groups. Functional groupings are explained in the student section. As noted above, we have found significant effects of C4 grasses and legumes.
Possible assessments include the accuracy and clarity of figures students make using the Excel spreadsheet, written description and analysis of these figures, discussion or analysis of one of the papers listed below, and a short essay on one of the discussion questions above or in the "Student Instructions" section.
Resources include links to sites about the Cedar Creek research, Biodiversity, Grasslands and Prairie, and Excel tutorials.