Live from the Bird Cams Lab: Q&A Session about Hawk Talk Data Visualizations

February 5, 2020

Red-tailed Hawk cam viewers and Hawk Talk participants joined the Bird Cams Lab team on February 3rd to discuss the project’s preliminary findings. 

Charles Eldermire, the Project Leader, kicked things off by welcoming everyone and orienting them to the Zoom tool we were using for the session. Then he went off screen to monitor the chat so that everyone’s comments and questions could be noted. Ben Walters, the Communications Specialist, and Rachael Mady, a graduate student, then took the reins. 

We launched a poll and were excited to find out that there were people listening in who had been involved with Bird Cams Lab as well as newcomers. To get everyone up to speed, we explained how Hawk Talk fits into the larger project of Bird Cams Lab. 

Bird Cams Lab is a Cornell Lab of Ornithology project funded by the National Science Foundation and aims to create an online space for scientists and cam viewers to co-create scientific investigations. We want to see what happens when participants do more than just gather data and have the opportunity to participate in each part of the scientific process, from observing and coming up with a question to publishing the results. 

Hawk Talk is the first investigation out of Bird Cams Lab, beginning with the observation and brainstorming phase in the summer of 2018. We then took a vote as a community, designed a data collection scheme on Zooniverse, and then spent 8 months collecting data from 10-second clips that spanned the first week after the nestlings had hatched in 2018. We set out to answer the question, “Do hawks use different kinds of calls in different situations at the nest?” Once we finished collecting the data in August 2019, the Bird Cams Lab team set to work to turn the data into visualizations for everyone to explore together. 

Before Rachael walked everyone through the visualizations themselves, we opened the floor for those listening in to let us know if they noticed anything in the visualizations they wanted to share. A participant, Cindy, chimed in to let us know that she liked the visualizations and their interactivity, great feedback to hear!

We then explored a visualization that connected nestling and adult vocalizations to different situations at the nest. We fielded questions and dug into what the graph was showing us. One participant, Kaliopi, noticed the big difference in how many clips there are of nestlings vocalizing versus adults: nestlings were found to vocalize in over 1,000 clips while the adults only did so in 45 clips. A very important note when considering what the data can tell us.

In the second visualization we started to get at how the data could answer the research question, “Do hawks use different kinds of calls in different situations at the nest?” by looking at the association between the types of nestling vocalizations and situations at the nest. In the comment section under the visualization, Kaliopi and miksohca had both noticed that peeps are really the dominant category, even if it fluctuates when we consider arrivals and departures. Potentially even cooler was that there is a large portion of the clips that include this “unsure” and potentially undocumented vocalizations. 

The third visualization was just like the second, but looked at the association of situations at the nest with adult vocalizations. Unlike the nestling graph, we only were able to document “chwirk” calls and “other” vocalizations. One participant was surprised by how many clips we were unsure about. We talked through why that might be, and how categorizing something like vocalizations can be tricky, even with practice, because of things like background noise. Also, Ben pointed out it was really exciting to see as high agreement as we did for “chwirk” vocalizations. Classifying vocalizations is tricky, even for those that watch the cam a lot! 

Based on these visualizations we concluded that while we may not be certain that certain types of vocalizations are associated with certain situations, we do think that there is a potential association between adult and nestling vocalizations in general happening more often with certain situations. And an important thing to keep in mind is that certain activities, like brooding, simply happen more often, which could drive some of the patterns we see in the raw data. 

Before we jumped into more visualizations, we opened the floor again to allow any questions from those watching. Understandably, one of the first questions we got was about the future of the project. For Hawk Talk, the data exploration was the second-to-last phase. We’ll be writing up a formal summary of the findings to share with the Red-tailed Hawk community soon, which will be the last phase of the investigation. There is potential to build future investigations based on these results or to investigate another question with the Red-tailed Hawk cam. 

For example, many participants have pointed out that it would be great to look at vocalizations at the nest later in the season because maybe the vocalization types change. One of the visualizations, “Do Nestling Vocalizations Change?” does suggest that there might be a change! However, with only a week of data, we can’t be sure if this pattern continues or if this is driven by the number of nestlings hatched in the nest. Ben pointed out this was one of his favorite preliminary findings and really allowed us to see things by toggling on and off the data with “Choose What to Display.” 

We then continued the discussion as more questions came in. 

  • How many of the classifications on the clips were by research team members? Very little! The Bird Cams Lab  team would periodically classify a few clips to make sure we were on the same page as the community with understanding how things were going, but really the community of over 1,400 people was the driving force in collecting the data. 
  • Why are the sample sizes (N=) changing from visualization to visualization? The sample size changes because the categories are changing. It’s important to consider and remember because 60% of 45 clips is different than 30% of 1,100 clips. 
  • Will you ever be doing research on vocalizations off the nest? Within the Bird Cams Lab framework we are concentrating on investigations that are centered on what the nest cam can capture. However, if the community is interested in documenting vocalizations then that should happen! We already know that there are birders on the ground following the hawks. Off-nest vocalizations could be a great supplement to what we’ve already found. 
  • Were the clips taken from 24/7 time periods and how were they collected? We had recorded footage from the daylight hours and for the week after the first nestling hatched. Charles was able to clip the footage into 10-second clips and then the clips were served up randomly to participants in Zooniverse. Importantly, we took a sample of the clips from the first week; we didn’t collect data on every second, but about a quarter. By doing this we could get a good picture of what was going on without having to watch clips infinitely. Just like researchers go out and sample, we realized that sampling is something we can do with cam footage.
  • Were there any cases where the moderators disagreed with the consensus for a clip? The Bird Cams Lab team hasn’t gone back to compare what the agreement between us and participants would be, but a previous Zooniverse project, Arizona BatWatch, did do that and found that there was no difference between participants and experts. Six participants were just as good as six experts, which points to how neat and powerful this type of data collection. Having multiple people watch each clip increases the accuracy of the data collected. 

We wrapped things up by sharing one more visualization that shows how vocalizations vary with time and Kaliopi chimed in one more time to point out that vocalizations by the nestlings may be related to aggression. We didn’t collect data on this, but now know which clips have nestling vocalizations thanks to this investigation and there is potential to go back to document this in a deeper dive in the future.

Before we signed off, we shared some exciting news about Big Red and Arthur getting ready for the next breeding season. We’ve seen them visit and bring sticks to the nest on the Red-tailed Hawk cam. The birders on the ground (BOGs) have noticed Arthur has been defending his territory and even expanding it. Both Big Red and Arthur are pair bonding by hunting, flying, and sitting together. Plus, Big Red sometimes has been getting into the mating posture. We’re excited to see the hawks return and follow their breeding season on the newly installed cams! 

Thank you to everyone that has been a part of the Red-tailed Hawk community, Bird Cams Lab, and Hawk Talk. A special shout out to Karel and Bogette who took the picture of Arthur that is featured on the Bird Cams Lab website. 

Please share any remaining questions you may have or ideas for more visualizations that we could create below. This call goes out to all those reading this and watching the recorded session as well. Every Hawk Talker can be a co-pilot in this investigation!