Biology 141   Lewis & Clark College

Lecture 3 Designing Ecological Studies

K.E. Clifton   Dept. of Biology

From last Friday's lecture:

What are the variables of interest... note the term: Variable!

Theoretical models

Comparative observational studies

Experimental manipulations

Now... Where should scientific studies be done?

Laboratory versus "field" studies

 

What labs offer: Far easier to control some conditions.

Lower between individual variation (most of the time)

Observations also often easier (but also sometimes less fun)

 

But lab studies also have limitations:

Controlled environment vs. nature: can you make nature in a bottle?

Conclusions apply only for the conditions tested.

Generalizations may be better seen from field studies.

 

Revisit the scale of biological inquiry, causation, and the way to think about the design of a study

Now on to thinking about sampling and experimental design

First, an acknowledgement about variation in nature... it is everywhere

What causes this variation?

sorting out the "pattern" (variation caused by something) from the noise (random variation)

General principles of designing ecological studies... use replication and randomization to overcome the influence of random variation

Replication

Think about a study of human height for men and women

What is the general perception?

Are there exceptions?

Are there limits?

What do the data often look like?

Consistent patterns are unlikely to be explained by chance

Detailed example – beetle larvae and competition

Replication and statistics

 

Pseudoreplication?

Taking samples from larger populations reduces problems of spatial scale

Replicates must be independent

Common causes of pseudoreplication

Shared enclosure

Relatedness

Issues of measurements through time... temporal scale is as important as spatial scale

 

Dealing with pseudoreplication

Randomization       

Definition

Why do you want a random sample?

Haphazard versus random

Pitfalls associated with randomization procedures

Random versus "representative"

How many replicates?

Is "the more, the better" a good rule?

Educated guesswork

Factors affecting the "power" of an experiment

How much the "effect" influences the results

Amount of random variation

Number of replicates

Controls

What are they and why are they so important

Take me to the course moodle page