GEOG/ES&P 330

California Ecosystems

Testing Natural and Sexual Selection among Feral Pigeons

==========

Background:

In this lab, you'll test a hypothesis about natural and sexual selection among feral pigeons, based on data collected by your predecessors following the Cornell Ornithology Lab's Project PigeonWatch protocol.

PigeonWatch was a citizen science project, designed to collect data on the prevalence of different pigeon morphs (blue-bars, red-bars, spreads, reds, checkers, pied, and white) and observations of pigeon males courting other pigeons. The question is why are feral pigeon flocks so diverse in color? Natural selection in natural habitats tends to encourage the propagation of the morph best suited for a given habitat (think of all those identical mourning doves, crows, and seagulls!). This is called stabilizing selection.

One possible exception is sexual selection, where sexual preferences can actually counter natural selection, as long as the animal is so popular with the opposite sex that it leaves more offspring than a more boring but physically fitter animal (e.g., peacocks, cichlid fish, Irish elk). The Cornell lab would like to learn if sexual selection is somehow maintaining the diversity of feral pigeon flocks.

Here are materials from PigeonWatch:

Data:

So, your colleagues and predecessors who couldn't make the regular training field trip went out on their own and did six PigeonWatches for each trip they had to miss. The result is a ton of pigeon data. Here are the raw data from the Fall 2011 section of GEOG 442:

Hypotheses:

Using these data, each group will analyze one of the following three hypotheses:

  • Hypothesis 1: Natural selection favors significantly different mixes of pigeon morphs in different habitats (data are all pigeons counted in six different habitats, grouped into urban versus beach/park/"natural" habitats)
  • Hypothesis 2: Sexual selection favors a significant difference between the morphs male pigeons were observed courting versus the supply of pigeon morphs available at the sites where courting was observed (are the males hitting up more of the exotic colors, such as red, white, or pied?)
  • Hypothesis 3: Pigeon males of particular morphs may sexually select targets similar (or different) from them in coloring (we're matching courting male morphs with target morphs)
Figure out the problem your hypothesis addresses (revised for COVID conditions...):

Figure out how your assigned hypothesis relates to the problem described in the Backgrounder reading on the course home page (LaBranche 1999): Why study pigeons? Your hypothesis may have to do with birds blending into their habitats to avoid being spotted and eaten by predators (garden-variety natural selection). It could have to do with sexual selection. Do male pigeons in general find certain female morphs "hot" and, so, court disproportionately more of the "hot" morphs than you would expect from the mix of birds in the regional population? Or maybe the male morphs themselves differ in their sexual tastes? Maybe certain male morphs prefer hens that resemble themselves or are different from themselves ("birds of a feather" vs. "opposites attract"?) while other male morphs have different inclinations. As the article points out, maybe there's even some power politics going on here. Maybe males are looking for hens of a dominant morph so their offspring can better duke it out when food gets scarce?

Take notes on how the data were collected. You can look at the data collection forms at the bottom of the course home page to have an idea what went on. Also, please note the names of those who actually collected these particular data (on the bottom of the data forms for each hypothesis).

You picked an alpha level last week in order to run the Chi-square on your data. Review why you picked 0.05 or 0.10 in terms of the consequences of making a Type I error (false positive, getting excited over, basically, nothing) or a Type II error (false negative, failing to see something that might be there). Scientists try to minimize Type I errors (smaller alpha but they don't want to preclude seeing something that might actually be there in their data (larger alpha). So, picking alpha is about finding what you think is the right balance between these two kinds of errors and their consequences. Jot down a few reflections on that right now.

You ran the Chi-square and got results. Note down each of the following:

  • Was there a significant difference or association?
  • How do you know? (there are two ways to test your null hypothesis: Calculated test statistic vs. critical test statistics or, more directly, p-value vs. alpha)
  • No matter whether your null hypothesis was rejected or not, what was the size of the effect found? Was it small (~0.10 to ~0.30)? Medium (~0.30 to ~0.50)? Or pretty strong (at least 0.50)?
  • How much power did your study have? That is, how well does it mitigate a Type II error? If you had a moderate effect but too high a prob- value to reject the null hypothesis, it might be because you have an underpowered study (too few observations). On the other hand, if you have non-significant results, low power, AND a really tiny effect size, maybe there really is nothing going on and it would be a waste of work to get more observations in.
Now that you know what your results mean, you are ready to write this up! Please refer to the How to Organize a Scientific Paper or Talk link on the home page.

Write up a mini-paper, an abstract or précis of your lab investigation. This paper should be very short: Aim for about a page or page and a half. It should have five sections with headers:

  • Introduction: what PigeonWatch was trying to investigate and what your specific hypothesis is (state it as the testable null hypothesis)
  • Data and Methods: a couple sentences stating that these are archival data (high quality secondary data), how the data were collected, and credit the student team who collected them in Fall 2011. State that you are going to use Chi-square to compare your counts because Chi-square is designed for counts in categories. State your alpha and why you picked that level.
  • Results: significance, effect size, and power, basically what happened when you ran the Chi-square. You could illustrate it with a small table.
  • Discussion: State what your results mean for your hypothesis. Were you able to dismiss the null hypothesis or not? How strong was the effect?
  • Conclusions: Review whether you successfully rejected the null hypothesis. If you could not reject it, what was the achieved power? In light of the effect size and power, would you recommend that this might be worth collecting more data?
Include a formal citation to the LaBranch article. See link on home page to show you how to do this.

Do a last pass through and edit for spelling, grammar, punctuation/capitalization, and sentence structure. Here's a quick guide for that: https://home.csulb.edu/~rodrigue/writmech.html.

So, this is your first practice with using the standard scientific paper format!

==========

first placed on the web: 12/08/11
last revised: 10/05/20
© Dr. Christine M. Rodrigue

==========