Redwing Blackbird Guidelines

Conspicuous color patches are thought to evolve as signals for use during social interactions. However, as with all signals, there are costs and benefits of continuously exposing the signal. Clearly predation costs will be higher for an individual with a conspicuous color patch compared to one without. There may also be social costs that constant exposure of a signal carry. With the exception of chromatophores and vascularized skin, most color patches can not be rapidly altered, so these social and predation costs may be significant. Some species, however, have gotten around this problem by being able to behaviorally change their conspicuous signals by covering and uncovering them. Red-winged blackbirds are an excellent example of this behavior, as they frequently cover and uncover their bright red epaulets, depending on the social context. This behavior leads to the obvious question of what are the social advantages of covering one’s epaulets. We used model (stuffed bird) presentations to investigate this question.

This is a short report (1-2 pages + graphs if necessary). You should follow the approach used in the Ensatina Model report: write brief introduction and methods sections and combine your presentation of results along with your interpretation and discussion of those results.

Here is a first reference to get you started on your citations:

Hansen and Rohwer (1986) Coverable badges and resource defense in birds. Animal Behaviour 34: 69-76

DATA:

The summarized data have been put on the lab computers for you in the ‘Redwing’ folder. You should complete your analysis using the Statview file (redwing.svd). The other two excel files (redwings_final.xls and redwings_extracted.xls) are simply there for you to look at or manipulate if you want more information.

The summarized data reflects a ‘weighted’ averaging (we threw out outliers) of the data that you all entered. While we all observed the exact same playbacks there was considerable variation in the data that different people gathered! Some of this variation may be due to individuals with and without binoculars. However, one challenge in behavioral ecology is to gather accurate, repeatable data and to minimize biases introduced by the observer. Sometimes it takes practice to notice specific behaviors or to manage gathering many types of data simultaneously, but you should make a special effort to perfect these skills. Observer bias is a major pitfall if a real effort is not made to minimize it. While we do not want you to focus on this in your discussion of the experiment, please look at the raw data file to see this for yourself and think about these issues as you conduct your independent projects!!

ANALYSIS:

The data should be analyzed using a "matched pair" comparison. We randomized the order of presentations to control for the effect of previous exposure to a model. However because each male may differ in their reaction to a model (say if one male is much more aggressive or dominant that another), we gain "statistical power" (i.e. easier to actually show true differences with smaller sample size) by comparing each male’s response to the two model types and asking if the direction and magnitude of the difference in each males response the to two model presentations is consistent across males, even if some males respond more aggressively to both types than other males. This type of comparison is called a "matched pair comparison" and is very powerful. There are both parametric (t-test) and non-parametric (Wilcoxon) versions of the test.

Example:

In this example each male receives the two treatments (un/covered). Males vary greatly in their response, so by comparing each male’s change in response, rather than the average response to covered vs uncovered, we gain power to detect a pattern – uncovered treatments elicit a larger response than covered do.

In Statview, you should do a “Paired T-test” or a Wilcoxon Sign Rank test (while the data are not normally distributed, results are the same for both tests in this particular case). To complete this analysis, you must compare how the same variable differed between the two treatments (U=uncovered, C=covered treatments). Do this by selecting the two data columns (C and U) that reflect the same behavioral measure. Additional variables calculated for you are: Latency: the time it took the male to first approach the model; Closest approach: the closest distance a male approached (15 m is assumed to be the max); and Latency*approach: a composite of how long it took a male to approach the model and how close he got.

While many, if not all, of these results are not significant you should still report p-values for the analysis you did. In your discussion, you may want to consider why these results are not significant (HINT: the birds don’t care if the signal is covered or uncovered is only one answer – you should come up with more than this). You may also want to think about any natural history observations you made while we were observing the blackbirds that may have some bearing on the issue of why the epaulets are occasionally covered (some of you may even have written notes in your field notebook!!).