Chapter 2 Balcony birdwatching


Bird observation, from a distance.

Learning outcomes

  1. Identify common species of birds locally.
  2. Collect a dataset.
  3. Connect principles of experimental design to implementation.
  4. Write clear and reusable meta-data.
  5. Contribute to open science by publishing data and meta-data.

Steps

  1. Scout out a location with more than a single species of birds and a frequency of a few different individuals of birds over a 5-10 minute duration.
  2. Select a good spot to the observe birds at your designated location. It can be a balcony or quiet spot. Vegetation such as trees or shrubs can facilitate observation of birds by providing habitat.
  3. Choose a distance that permits enough resolution to see plumage and what an individual bird is doing (depending on whether you are using binoculars, a spotting scope, or unassisted with your vision). There are considerable merits to observing birds more simply (Wilkinson, Waitt, and Gibbs 2014). You are also welcome to address any visibility or spotting challenges using bird calls to record frequency of birds in a sampling region.
  4. Specify a duration to sample, for instance, 60 minutes when you have observed the most birds in your scouting exercise. Remember, this is a pilot experiment. Take qualitative notes, sketch, and complete this datasheet.
  5. Use your notes to complete a meta-data file, i.e. a description of how the data were collected, whether, when, and what each attribute in your dataset means.
  6. Sign out a bird guide for your region from the library or university or try out a free app for now to support identification.

Data

Here is a sample datasheet for the pilot experiment. This is set up as species-level observations, i.e. each row or replicate is a species of bird you observe. This datasheet is for the pilot experiment, and it is a stepping stone for the deeper dive experiment if you choose to complete this experiment for your first report. A more detailed datasheet can consider duration or start and stop times of each individual bird, more details on the environment, or record interesting ecological or environmental variables that are present in the environment too - noise, disturbance, squirrels, other birds, etc.

Data can be organized in many different formats depending on the approach to collecting the data in the field or the lab, the instrument or method used, preference, or accepted standards within the domain of study. In many modern data science endeavours, data are also formatted according to the principles of ‘tidy data’ (Wickham 2014). The following three rules define data as tidy (Grolemund and Wickham 2016).

  1. Each variable must have its own column.
  2. Each observation must have its own row.
  3. Each value must have its own cell.

Most scripting languages such as R or Python can resolve and tidy up data to adhere to these principles, but with a little planning, your data can be set up to facilitate this process and enable easier data visualization and models.

Sample data set
In this example, the field observations were coded as one species per behaviour per row. A compromise between tidiness and ease of collecting the tallies per species in the field.

rep date researcher location species frequency behaviour
1 15/9/2020 cl High Park, Toronto House sparrow 12 flying
2 15/9/2020 cl High Park, Toronto Blue jay 2 foraging
3 15/9/2020 cl High Park, Toronto Cedar waxwing 1 perching
4 15/9/2020 cl High Park, Toronto Cedar waxwing 3 foraging
5 15/9/2020 cl High Park, Toronto Dark-eyed junco 2 flying
6 15/9/2020 cl High Park, Toronto Black-capped Chickadee 3 foraging
7 15/9/2020 cl High Park, Toronto Black-capped Chickadee 2 posturing
8 15/9/2020 cl High Park, Toronto Black-capped Chickadee 3 interacting with another bird
9 15/9/2020 cl High Park, Toronto Black-capped Chickadee 2 sitting
10 15/9/2020 cl High Park, Toronto Wood thrush 2 flying
11 15/9/2020 cl High Park, Toronto Wood thrush 5 on ground
12 15/9/2020 cl High Park, Toronto Northern flicker 1 perching in a tree

Meta-data

In many disciplines of science, meta-data are the descriptive elements of the dataset. They provide a clear means for discovery and reuse of data collected - by you in future and for others (Heidron 2008; Reichman, Jones, and Schildhauer 2011). For the purposes of our practical learning in experimental design here, describe what each column in our dataset means, describe the structure of your dataset (i.e. each row is a species-level observation, or plot, or transect), describe the duration of sampling, location, and provide a bit of guidance for someone to use in inspecting the dataset. It is very similar to the methods in conventional publications or standard reports, but it ensures each attribute in the dataset has a brief description. It is also superb preparation for the methods if you choose to write a report.

Deeper dive

If you choose this adventure, your goal is to experiment with the method of animal observation to test a hypothesis and predictions. The text ‘Experimental Design for the Life Sciences’ does an excellent job of explaining how to set up hypotheses and predictions (Ruxton and Colgrave. 2018). Pilot experiment first, think, explore your data and notes, then write your ideas down that you want to test. A hypothesis is a clear explanation of how a system works (LaPlaca, Lindgreen, and Vanhamme 2018; Bains 2005). The predictions are logical and reasonable outcomes if the hypothesis is a good approximation of how the system works, i.e. the key variables that make it work. Predictions should be testable and read like simple sentences that describe results. The goal of the deeper-dive experiment is to take your pilot experiment, examine what worked and did not work so well in your experiment, and do a deeper and more thorough job of testing a key idea that you are interested in associated with bird communities in your backyard or neighbourhood. The goal should be to explore one key factor that describes how the species locally interact within one another, the environment or other species, or resources.