Inferring patterns of herbivory from leaf injury

BACKGROUND
Much plant tissue in the world is consumed by a myriad of herbivorous animals. Some herbivores specialize
on particular species and on particular parts of the plant, while other species are more general. Not surprisingly
plants have evolved defenses to counter herbivory, involving both mechanical and biochemical responses. For
example, plants whose tissues have been injured by herbivores can respond by translocating toxic secondary
metabolic products to these injured areas. Herbivores can in turn evolve behaviors to respond to these defenses.
The Plain Tiger Danaus chrysippus caterpillar carves out circular trenches in the leaves of milkweed, severing leaf
veins to prevent defensive chemicals from entering this tissue in the middle.
Often times in the study of ecological systems, important events are too rare to be observed, or are hidden
from view of the ecologists. For example, how likely would it be that one would one would observe owls preying
on rodents during a dark night of observation? As a result, ecologists must often rely on indirect measures to
estimate the nature and intensity of ecological events. In this lab, we will survey patterns of injury to leaves as a
way of inferring processes associated with herbivory. Alternatively, we could bring an herbivore into the lab and
observe it feeding on a given plant, but removed from its natural setting we could not be confident on whether
what we observed in the lab actually occurs in nature.
Often in science, exploratory quantitative data sets are generated in order to establish trends from which
hypotheses can then be formulated. By sampling quantitatively, we can be more confident that the trends we
thought we observed really do exist (i.e. are statistically significant) before we spend a lot of time testing ideas on
why those trends exist. In other words, we may establish that there is a difference before testing why there is a
difference.

INSTRUCTIONS
Work in groups of two. From the forest near the science building, collect at least 30 leaves of one plant
species, placing them in a plastic bag. Try to collect these as haphazardly as possible (in other words do not base
your selection on whether they have damage or not). Collect only leaves on the surface of the forest floor (leaves
deeper down in the litter are more likely to have tissue missing due to decomposition rather than from herbivory).

Return to lab and measure the following on each of 25 of the leaves you collected, then switch your leaves for
another group’s leaves (MUST BE DIFFERENT SPECIES) and repeat the measurements:

 Taxon (type) of leaf

 Length of leaf (in centimeters); measure to nearest mm (ex: 12.4 cm)
 Number of injuries along margin of the leaf
 Number of injuries (holes) away from the leaf margin (count only those that penetrate all the way through
the leaf)
 Total percent of tissue removed from each leaf (estimated as falling in one of the three injury categories
listed on the following data sheet).
Example Data Sheet – Record your data in Excel for each leaf species

Estimated amount of tissue missing

(indicate with check mark √)
Taxon Length Number Number No tissue Minimal Substantial
(species (cm) of injuries of injuries missing damage damage
or type of along away 0-10% >10%
leaf) margin from
margin
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
ASSIGNMENT

1) During lab examine the data set. “Explore” the data by plotting two variables against one another, or by
calculating and comparing some means and standard deviations. From this, identify two trends
(relationships) or lack of trends among variables. Do NOT compare average leaf length between two species
(you will get a 0 on analyses comparing mean leaf lengths).

From your rough analysis of the data, identify two possible hypotheses that might explain the trends you
found.
Submit a document on D2L describing what trends you explored and which hypotheses you think apply. This
is due by the end of the lab period. If working in pairs make sure all members names are in the document.
Grading Rubric:
Each trend described 2 points
Each hypothesis identified 2 points

2) Outside of lab, using an appropriate statistical technique, determine whether the relationships are significant.
For each trend or lack of trend, you should:
a. Prepare a graph
b. Describe the trend in a sentence or two.
c. Report the type of the statistical test used and the results of the test. Be sure to report the P-value if
needed and whether it is considered significant or not. SEE NEXT PAGE FOR REQUIREMENTS FOR FULL
CREDIT FOR EACH STATS TEST TYPE. Do NOT use same stats tests for the two trends! Change trends if
this is the case (-10% if you use same stats test)

3) For each of the two hypotheses you identified in lab, explain why you chose the hypothesis and how you could
further test that hypothesis using the Arboretum where the leaves were collected.

4) Turn in a final report document AND the Excel document (or R script) showing your raw data and statistical
tests (including the cells with the formulas/ commands to run the stats test). Be sure all contributors’ names
are on the documents, for example “Smith and [Link]”

Flowchart of assignment requirements:

 Identify two
Describe trend
stats test hypotheses and
(or lack of)
explanation
Explore trend 1

graph

Dataset
Identify two
Describe trend
stats test hypotheses and
(or lack of)
explanations
Explore trend 2

graph
Required elements for each stats test type (you will choose which test you should run on your data for the
questions you are asking)
NOTE: remember to round off statistics as described in Lab 1 Part 2.
(half-credit given if you do not round off and provide a bazillion decimals)

T-test
 Report values for:
o Means
o Standard error (half credit if standard deviation is given instead)
o P-value
 Graph
o Appropriate type graph (probably a bar or column chart)
o Axis labels
o Proper error bars
 Do not choose Excel’s default error bar option, calculate standard error for each mean
and then add the error bars as “custom values”
o Title or caption
Chi-squared test (only for “counts”)
 Report values for:
o Contingency tables
 Two separate tables
 Observed
 Expected
o P-value
 Graph
o Appropriate graph type (bar or columns charts)
o Axis labels
o Title or caption
o Legend

Correlation
 Report values for:
o Correlation coefficient (r)
 make sure to discuss whether this is positive or negative in your discussion and what
that means
o Correlation coefficient from table needed for significance
 Graph
o Appropriate graph type (x vs y AKA scatter plot)
o Axis labels
o Title or caption

Linear Regression
 Report values for
o R2
o Y=mx+b equation (AKA trendline, line of best fit, line of regression)
o Any predicted values you may have made (ex: if X=…, what will Y equal?)
 Graph
o Appropriate graph type (x vs y AKA scatter plot)
o R2 and equation on graph
o X-axis is independent variable
o Title or caption