Ecological Experiments

1
Ecological Experiments
2
Lecture Goals

• Review Basic Principles of Experimental Design
• control, treatment, etc.
• Distinguish among types of ecological experiments
  
• mensurative
• manipulative
• Discuss the role of replication
• relationship to environmental variation
• identify pseudoreplication
• Describe the relationship between
• precision and sample size

3
Is It Science?

• Not meant to say "science good, other methods
  bad"
• What makes it "scientific"?
• "I think that there will be more people in
  the world by this time next year."
• How can I say this?
• How can I convince others?
• How do I know that I am correct?
• What distinguishes scientific inquiry from other
  forms?

4
Elements of an Experimental Study

• Hypothesis
• "The number of people...
• Experimental Design
• plan repeated census
• Experimental Execution
• train staff, prepare forms, walk
• Statistical Analysis
• If we count everyone, skip(how likely is this?)
• Interpretation
• more or less?

?
5
Variability and Bias

• a.k.a. precision and accuracy
• Why measure more than once?
• Which darts are more precise (less variable)?
• Which are more accurate (less biased)?
• How does this influence the design and execution
  of experiments?

Experiments are designed and analyzed to
distinguish between results caused by the factor
being examined and natural _______________________
___________.
variability and biases
6
How Many Replicates?

• Experimental Unit
• focus item of study
• plant, pot, plot, etc.
• Example gtgtgt Fertilizer Clover Density
• Treatment 100 g fertilizer / m2
• equal seed density everywhere
• (Control Treatment) X Replication
• Case 1 Pots in a greenhouse
• Case 2 Plots in an open field
• Which case would most likely require more
  replication? Why?

7
Types of Ecological Experiments

• Mensurative Experiments (sensu Hurlbert 1984)
• a.k.a. sampling or census
• based upon observation alone
• 'treatments' are location and/or timing of
  measurements
• Manipulative Experiments
• experimental units assigned (usually randomly) to
  different treatments
• often employ explicit control
• to remove common temporal change in experimental
  units from the analysis (HOW DOES THIS WORK?)

8
Mensurative Experiment(comparative)

• Hypothesis
• Bird diversity declines in urban areas
• Design
• Estimate species their abundances in urban and
  rural areas
• experimental unit 100m transect (visible from
  walking)
• replicates?
• location of transects?
• all in the same city block?
• all in same rural section?

9
Manipulative Experiment

• Hypothesis
• Plant species diversity increases after a burn
• Design
• experimental unit plot
• Randomly assign plots in a field to either
• A. burn or B. no-burn
• replicates?
• estimate species diversity in all plots
• apply burn to appropriate plots
• wait 1 week, month, or year
• re-estimate species diversity

10
External Intrusions

• Non-Demonic Intrusion
• chance events that affect experimental units
  while an experiment is in progress (increase
  variability)
• death of an animal (disease unrelated to
  experiment)
• hole in a single enclosure due to weathering
• Demonic Intrusion
• unforeseen event that confounds the
  interpretation of results (unexpected bias)
• may or may not be detected
• treatment (all non-control units) that attracts
  predators when mortality is being estimated (e.g.
  mark or tag)

11
Sources of Confusion
Protocol
Problem
Temporal Change Procedure Effects Experimenter
Bias Variability among Units Non-demonic
Intrusion Demonic Intrusion
In some cases interspersion may be preferred to
simple randomization
12
Beware (and be aware of) Pseudoreplication

• Experiment
• fertilizer influence on plant density
• 4 pots (2 control, 2 fertilized)
• replicates 4 pots ?
• gtgtgt OR ltltlt
• count plant densities in each pot twice
• replicates 8 pots ?
• Pseudoreplication
• sample size artificially inflated by repeated
  measurement of the same experimental units
• often properly dealt with through a "nested"
  analysis or repeated measures analysis

One pot or eight plants?