Download Soil Temperature Differences & Correlation Coefficient for Soil Depth & Moss Cover and more Slides Biology in PDF only on Docsity!
Statistical Tests for Ecology
Why bother?
Sometimes when we look at data it is easy to spot if there is a definite relationship or not.
Sometimes however it is not so clear cut. Statistical tests help us to decide if the relationship does
exist by removing both the uncertainty and subjectivity from the decision.
How do I start?
All statistical tests start with the formulation of a null hypothesis (H 0 ). This hypothesis states that
there is no significant relationship or pattern within our data. During an investigation we assume
that the null hypothesis is true and we only change our mind, and reject the null hypothesis, if there
is strong evidence to show otherwise. The statistical test helps you decide if the evidence is strong
enough.
Any pattern within data can always be due to chance rather than an actual relationship. The
significance level is used to measure how strong the evidence needs to be before we reject the null
hypothesis. In biology we work with a 5% significance level (P = 0.05). What this means, is that we
only reject our null hypothesis if the probability of the pattern in our data being caused solely by
chance is less than 5 % This allows us to say that there is at least a 95% probability that the data is
statistically significant and that it shows a true relationship.
Student’s t test
We use this test to compare the means of two samples and can be used even if they have different
numbers of replicates. In simple terms, the t - test compares the actual difference between two
means in relation to their variation (the spread of the data about the mean).
You can carry out a t test if:
You have collected measurements from 2 or more sites/populations and are interested in
the differences between them.
The data are normally distributed (symmetrical about the mean).
Data are unmatched (not collected in pairs)
You are using small sets of data (less than 30 in each data set).
Worked example.
If you wanted to find out if there was a significant difference in the soil temperature of the North
and South facing slopes, you could carry out a t test.
Our null hypothesis would be:
There is no difference between the soil temperatures on the two sides of the dale.
In our investigation, six soil temperature measurements were taken from each slope:
Soil temperature:
North facing slope
(degrees C)
1
1
2
Soil temperature:
South facing slope
(degrees C)
2
2
2
1
𝟏
− 𝒙 𝟏
̅̅̅ )
𝟐
= 8.84 Σ𝑥 2
𝟐
− 𝒙 𝟐
̅̅̅ )
𝟐
= 3.
n = 6 n
1
2
- For each data set, add up all of your data (Σ𝑥), then calculate the mean (𝑥̅ ) by dividing
this sum (Σ𝑥) by the number of samples (n).
- For each number, subtract the mean and square the result (𝑥 − 𝑥̅ )
2
- Add up all of the squared differences to find Σ (𝑥 − 𝑥̅ )
2
Find the average of the squared differences by dividing by n-1:
S
2
2
North facing slope South facing slope
1
2
- 84
6 − 1
2
2
=
- 34
6 − 1
- Now find the t value for the data using the formula:
t =
In choosing the 5% significance level we are saying that we would expect to be correct in accepting
or rejecting our null hypothesis 95% of the time. We might get a different result due to chance 5% of
the time.
In this example, our calculated value of 7.075 is greater than the critical value of 2. 228. We can
therefore reject our null hypothesis and be 95% certain that there is indeed a significant difference
between the soil depths on the two slopes.
Spearman’s rank correlation coefficient
We use this to test the strength of the correlation between two variables (i.e. how does one variable
change as the other one changes?). You can carry out this test if:
You have collected quantitative data
The data are not normally distributed (equally spread about the mean).
Have at least 8 pairs of data.
Worked Example :
If you want to find out if soil depth on the slopes is correlated to moss cover you can carry out a
spearman’s rank correlation.
The null hypothesis would be:
There is no relationship between the soil depth and moss cover.
Soil depth
cm 23 25 32.5 13.5 3.5 13 13 11.5 7.5 6.5 6 5
Rank 1
10 11 12 9 1 7.5 7.5 6 5 4 3 2
Moss %
9 5 12 10 1 6 0 1 0 5 2 3
Rank 2
10 7.5 12 11 3.5 9 1.5 3.5 1.5 7.5 5 6
Difference
between
ranks
|R1-R2| 0 3.5 0 - 2 - 2.5 - 1.5 6 2.5 3.5 - 3.5 - 2 - 4
d
2
0 12.25 0 4 6.25 2.25 36 6.25 12.25 12.25 4 16
- Make a table from your data to list your two variables (rows 1 and 3).
- Rank both sets of data from lowest to highest. The lowest number on row 1 gets a rank of 1,
the second lowest gets 2 and so on. Do this for both sets of data individually.
- If two values in a data set are the same (tied ranks) they must be awarded the mean
(average) of their ranks. In the example above two samples have 13 cm of soil depth. These
then share ranks 7 and 8. The mean rank is therefore 7.5 and the next highest value on the
table will get rank 9.
- Calculate the differences in the pairs of ranks i.e. rank 1-rank 2 (row 5).
- Square the differences to remove negative values (row 6).
- Add up the last row of numbers to find the sum of the differences:
d
2
= 111.
- Calculate Spearman’s Rank (r) using the equation:
R = 1 -
n is the number of samples you have taken
R = 1 –
6 × 111. 5
R = 1 –
R = 0.
What does this value mean?
The closer R is to 1, the stronger the positive correlation, the closer this R is to - 1, the stronger the
negative correlation.
Perfect Negative Correlation No correlation Perfect Positive Correlation
How to test the significance
The value of R must now be looked up on a significance table:
Simpsons Index of Diversity
We use this test to find out how diverse a community is. It is better than simply finding the species
richness (number of different species) as it takes into account the abundance (size) of every species
within that community. This, therefore, reduces the impact of species which are not well
represented within the community on the diversity index and hence bases the most value on the
most viable (strong) populations within the community. As species richness and evenness increase,
so does the diversity index.
To work out the diversity index you need to have:
Collected data regarding the numbers of individuals present rather than cover.
Worked example:
If you wanted to compare the diversity of plants growing on the north and south facing slopes we
would calculate the diversity index for each slope.
Note, this test does not require the formulation of a null hypothesis, unless you want afterwards to
test the significance of the difference in values found.
We will work out the index for one slope here as an example.
Species
Number of
plants in
quadrat (n)
n(n-1)
Grass, broad-leaved
Grass, narrow-leaved
Heath Bedstraw
Ladies Bedstraw
Lesser Celandine
Milkwort
Moss
Mouse-ear Hawkweed
Other
Ribwort Plantain
Salad Burnet
Thistle
Thyme
Tormentil
Violet
Total = 100 4940
N = 100 n(n-1)= 4940
- Make a table with your data to show the number of each species found
- Add up the total number of species found in the first column = N
- For each species multiply its number (n) by (n-1).
- Find the sum of the column for n(n-1)
- Use the formula to calculate the index:
D = 1 –
D = 1 -
D = 1 - 0.
D = 0.
What does this value mean?
The value of D ranges from 0 (no diversity) to 1 (infinite diversity). Clearly, on its own the value of
0.501 does not mean that much to us. We use this test o compare the diversity of habitats. So, for
example, you were comparing the diversity of the north and south facing slopes in a limestone dale:
North facing slope = 0. 501
South facing slope = 0.
Conclusion
Then the south facing slope is slightly more diverse than the north facing slope.
