Below are the methodological details used for collecting the data. In your manuscript methods, you will want to describe briefly and in your own words what the data collectors did to collect this data. You will then want to cite the protocol (using a proper in-text citation) in a sentence that refers to the protocol for more detail.
The class was divided into several groups, and each group sampled plots in one of two different forest areas (region 1 or region 2). Ropes were divided into 10-meter units. Data collectors established 100 square meter plots by stretching ropes out on the ground in a square with each of the 10 m ropes stretched as the sides and where any two meet as the corners. The quadrats did not overlap a path or other data collector quadrats. Each group collected data in ~five 100 m2 quadrats from each quadrat as follows:
To estimate the relative “time since a major disturbance”, the data collectors measured the circumference of all pine and hardwood trees found in the quadrats.
In order to determine how these two successional stages differ in species composition, data collectors counted all pines and all hardwood trees in each plot.
By looking at the composition of younger trees, students can also able to make predictions as to what the forest composition will be in the future as succession continues to progress.
There are lots of shrubs and other shorter plants in the plots, so it is important that the data collector used the following definition first define what constitutes a “tree” versus a shrub. When measuring the trees, the trees had to meet the following following criteria:
Must have an obvious single trunk up to at least 3m (~10 ft) off the ground
Must be at least about 25 cm circumference at chest height (~3 inches across in diameter).
All measurements were made at “chest height” (which in the USA means 4.5 ft or 1.37 meters) and all circumference measurements were obtained in centimeters
DATA ANALYSIS
Download the Data Excel File from D2L.
You will be analyzing data collected from a previous semester. The data will be provided as an excel spreadsheet in D2L (download the file). Simple instructions can be found in the Excel spreadsheets and you can watch a video showing the data analysis process here.
To determine whether there is a significant difference in the size of the oldest hardwood trees, you will be conducting a test with which you should be familiar from the Stats & Writing and Fish Labs earlier in the semester- the t-test. T-tests determine whether there is a significant difference between the mean of two different distributions.
In this case, you will be comparing the mean circumference of the top 10 largest trees from region 1 vs. region 2.
1. You will first need to create a pivot table to calculate the means and SDs for the top 10 hardwoods in region 1 and region 2.
You will need to make sure the data is sorted properly (first by Tree type, then by Region, then by Size from largest to smallest). You will then copy the top 10 largest hardwoods from region 1 and paste those values into the sheet. Then repeat this for the top 10 largest hardwoods from region 2. Once you’ve done this, you should select that data and then click insert > pivot table and follow the instructions that you’ve used in previous labs (the statistics lab and the fish lab) to complete the pivot table. If you need a refresher refer to the statistics lab protocol Exercise: Calculate Mean and SD by Group in Excel.
2. Calculate the t-statistic, degrees of freedom and p-value to complete a t-test analysis as you did the statistics and fish labs. You will use the fx function and find T.TEST or type in =T.TEST and feed it the 4 arguments it needs to run the formula which are (group 1 cells, group 2 cells, 2-tail , 3-unequal variances) to compute the p-value.
You will then use the t statistic formula:
=(AVERAGE(H2:H6)-AVERAGE(H7:H12))/ SQRT((VAR.S(H2:H6)/ COUNT(H2:H6))+(VAR.S(H7:H12)/COUNT(H7:H12)))
and the degrees of freedom formula:
=(((VAR.S(H2:H6)/COUNT(H2:H6))+(VAR.S(H7:H12)/COUNT(H7:H12)))^2)/((((VAR.S(H2:H6)/COUNT(H2:H6))^2)/(COUNT(H2:H6)-1))+(((VAR.S(H7:H12)/COUNT(H7:H12))^2)/(COUNT(H7:H12)-1)))
If you need a reminder on how to do any of this refer to the statistics lab protocol Exercise: t-test in Excel.
3. Using the second spreadsheet in the Excel file, make a column chart comparing the top 10 tree circumferences from each forest patch (see results section below for an example of what this looks like) will allow you to see the relative spread of the top 10 trees in both patches. Copy and paste the top 10 largest hardwoods values you used for the pivot table above into the appropriate columns in the second spreadsheet. From there you will highlight all the values and the column names in the second spreadsheet. Then click the insert tab > recommended chart > find the chart that looks like the figure in the protocol below and click that chart> then edit the axis labels and chart title and you have your chart. Don’t forget to include a proper legend when you put it into your manuscript.
4. To determine whether the forests differ in the proportion of mature hardwoods relative to pines, you will be employing a different test, the Fisher’s Exact Test. In this test, a two-by-two contingency table is constructed with two categories: the independent variable category is region: region 1 vs region 2, and the dependent variable category is tree type: pines vs. hardwoods. You will only be using “mature” hardwoods and pines which we define as any hardwood or pine ≥ 80cm. To do this you will need to re-sort the data first by Region, then by tree type, then by Circumference. Once you’ve done this you can highlight all the Hardwoods in Region 1 that are ≥ 80cm and the number of trees highlighted will be displayed in Excel. Place that value in the contingency table in your Excel spreadsheet 1 in the cell corresponding to Region 1 Hardwoods. Do this again for Region 1 Pines by highlighting all the pines ≥ 80cm, seeing how many pines that is and putting it in the cell corresponding to Region 1 Pines. Repeat this again for the Hardwoods and Pines in Region 2. You have completed your contingency table.
5. Now you will need to calculate the Fisher’s exact p-value.
Go to one of the following Websites:
http://www.graphpad.com/quickcalcs/contingency1.cfm
http://vassarstats.net/tab2x2.html
https://www.socscistatistics.com/tests/fisher/default2.aspx
Type in the values from your contingency table into the appropriate boxes. Click on two-tail and then run a Fisher’s exact test. Then run the analysis and it will give you a p-value. Important Note: if it gives you a p-value of 1.0 or 0.0 this is impossible. The reason it gives you that is because it rounded a very large or very small p-value. If you get a p-value of 1.0, you must report it as >0.999. If you get a p-value of 0.0 you must report is as 0.99, so report it as such.
5. A contingency table: Your Fisher’s exact test p-value (number 4 above) should accompany a 2×2 contingency table. This table should include the frequency of hardwoods ≥80cm in region 1, hardwoods ≥80cm in region 2, pines ≥80cm in region 1, and pines ≥80cm in region 2.
Discussion:
Begin by interpreting your results and drawing conclusions from them. You should explicitly discuss the descriptive statistics, the inferential statistic (p-value), whether the p-value is statistically significant, what the actual specific p-value means, how the inferential statistics and descriptive statsitics together relate to the relevant prediction. Do not forget to discuss the directionality of your data here (i.e. which region has the older trees/higher proportion of hardwoods to pines, or if there is no real difference). Put the interpretations of the two analyses together to address overall support for the hypothesis. Avoid using the words “believe” and “belief”, “prove”, “disprove” in science writing, as these words can have unclear and varied meanings to different readers. Instead of writing “I believe the results demonstrate,” write “The results demonstrate.” Nothing can be “proven” beyond all doubt in science. The very nature of science is that new data could potentially come along that would require us to adjust our understanding of something we thought we had “figured out.ʼ As such, you should never set out to ʻproveʼ or demonstrate the ʻtruthʼ about something. Instead, set out to ʻtest,ʼ ʻdocument,ʼ or ʻdescribe.ʼ Instead of saying“our data proves the hypothesis,” say “our data supports the hypothesis.”
This section should have a discussion of how your actual results compare to your expected results. Did the results support your predictions/hypothesis? (discuss in terms of statistical significance and support for your prediction and then how all the results provide support, weaken support, or show mixed support for your hypothesis based on the biology behind your prediction). Because your predictions are non-directional you will also need to tell the reader what direction your predictions went (if there is significant differences). E.g. which region has a mean larger circumference). You can support both predictions, but if they are going in opposite directions, you do not have support for your hypothesis. So be careful!
If your results were unexpected, you may wish to consider and address some of the following: Were the assumptions of the original hypothesis correct? Was the study design valid? While these issues should be considered, do not fall into the common trap of “looking for blame”. All studies have some weaknesses in their design. However, for the purposes of this lab you should assume that your results are reasonable. Negative or inconclusive results can and will occur during this lab course. In such circumstances, you should suggest how further study might clarify the areas of doubt in your data.
Additionally Focus on the following points for your discussion:
Choose the correct biological flaw from the options in the manuscript template that may have influenced the results and/or possible biological reasons as to why the results might differ from what you predict.
Specifically address the following questions: Why did we use only the largest hardwoods and why would you use only trees over 80 cm rather than all the pines and hardwoods for the Fisher’s exact test? Based on the saplings found, assuming the no new disturbance occurs, what do you expect the forest to look like in the future?
These questions should be integrated into the discussion where appropriate and not just responded to one after the other.
A final paragraph should address what the all the interpretations mean for your hypothesis, how your conclusions relate to the overall topic, and why we other’s should care (broader impact).
Literature Cited:
For this lab, you should cite the lab protocol, the website you used to perform the Fisher’s Exact Test, and the following website:
http://dukeforest.duke.edu/forest-environment/forest-succession/ . The Duke Forest website will give you a good general overview of old-field succession in the piedmont (Duke is in the piedmont of North Carolina, with the same soil type and very similar plant composition as Atlanta/Kennesaw), and you should even be able to predict how old the trees were in your plots.
Here is how to properly cite the protocol: KSU Biology. 2021. Ecology: Variation within Ecological Communities. Kennesaw State University, GA.
Here is how to properly cite the Duke website: Duke Forest at Duke University. 2021. “Environment: Forest Succession” http://dukeforest.duke.edu/forest-environment/forestsuccession/. 3/29/2021.
Here is how to cite the website for Fisher’s Exact Test: GraphPad Software. 2021. “QuickCalcs: Analyze a 2×2 contingency table” http://www.graphpad.com/quickcalcs/contingency1.cfm 3/29/2021.
Important note: In both cases, 3/29/2021 refers to the date you accessed the website. In general, it’s best to cite research published in peer-reviewed journals and not to cite web pages, but we’re making an exception here since most literature on piedmont forestecological succession is pretty old and harder to access.
If it is cited in the literature cited it must be in the main text and if it is in the main text it must be in the literature cited.
Through secondary growth, tree trunks increase in diameter every year. Grown under similar conditions (soil nutrient, light, and water availability) a 100-year-old tree should have a significantly larger circumference than a 50 year-old tree. However, since hardwood seedlings can survive in the shade, there will always be a range of smaller trees regenerating under a hardwood forest regardless of the time since the last disturbance (see diagram below). While there should never be tree trunks as large as the 100 year-old trunks in the 50 year-old forest, some 50 year-old trees are expected be found within a 100 year-old forest. Therefore, the largesthardwood trees are often the best indicators of the age of a forest since the last major disturbance.
Remnants of an old fence line run through the Kennesaw State Arboretum, separating the forest into roughly two regions. Your task will be to investigate whether there is evidence these forests differ significantly in time since the last disturbance by determining if they are at different stages of secondary succession.
Based on this, why might you expect these two forest regions to differ in age? Think about the fence line, what humans usually use fences for, what the last disturbance was in this region, and how those two things may lead you to think that the regions differ in age.
You will be given tree size measurements for 2 types of trees within the two forest sections to test the following predictions that we expect to be true if the forests are of different ages:
The forests will differ in the size of the oldest hardwood trees.
The forests will differ in the proportion of mature hardwoods relative to pines.
In your manuscript you will want to write a non-directional hypothesis. In other words, you are not stating that one region was disturbed more recently than the other, you are stating that they differ in age since last disturbance. You will notice that your predictions are also non-directional (that the forests differ – not that one will have larger oldest trees). You will use these predictions above in your manuscript, however, you should rewrite them and make them even more specific
Figure 2. images showing the trunk cross sections of hardwoods of different ages for a 50 year old forest (left side) and a 100 year old forest (right side). Notice how the older forest on the right has the largest circumference hardwoods, but it also has many younger hardwoods. There is a lot of overlap in the age and size of hardwoods between the two forests, because the hardwoods grow well in the shade – so even in an older forest many new hardwoods emerge. This helps explain why we only consider the top 10 largest hardwoods in our analysis of the difference in circumference. If we were to use all the hardwoods, what would happen to the mean and standard deviations? Would you expect a significant result?
SAMPLING IN FOREST REGION 1 AND REGION 2
Note: Obviously as a class you did not collect any data. The below methods were the methods used by the data collectors. Before you begin your analysis you are required to use these methods to identify one hardwood or loblolly pine tree and create a video showing how you would measure the circumference of that tree at chest height. In order to do this, you will first need to measure chest height on yourself (see below for what that measurement is), then take a piece of tape and put it at that place on your body. Then go outside with a piece of rope, string, a flexible tape measure, or anything else that can easily be wrapped around the trunk of a tree.
