Archive | September 2012

First Publication Using Data from Whale FM

This post is part of Citizen Science September at the Zooniverse.

I am Laela Sayigh, a researcher at the Woods Hole Oceanographic Institution on Cape Cod in Massachusetts. I have been working with a team from the University St Andrews in Scotland trying to categorize the different calls of pilot whales recorded from the Tongue of the Ocean in the Bahamas. Some of the calls in Whale FM come from the same recordings I use in my research, made on acoustic recording tags attached to pilot whales. This year, our research on the characterization of the pilot whale vocal repertoire used your classifications from Whale FM. In fact, they were really useful!

In our study, which involved over 4,000 pilot whale calls, several observers grouped the calls based on visual assessments of similarity. This is similar to the task Whale FM volunteers perform on the site. We found evidence for repeated call types, which had not been previously reported for pilot whales. However, to make the case, we really needed to see how generalizable our classifications were. We initially tried to use data from additional observers, but the vast and variable nature of the data made this very difficult. We were working on this right at the time that Whale FM launched, and we decided to see if preliminary results from the site might provide the corroboration of our classifications that we needed.

At the time, the website had been available to the public for about one month, and there were 255 instances of users being presented with a categorized call from our study as the “main call” to be matched. Users matched calls according to our categorization scheme in 189, or 74% of these instances. Since we do not have information about how many misclassifications may have resulted from users not being presented with a call from the same category as a possible match, the percentage of classifications that agreed with ours could have been even higher than 74%. So, although these results are preliminary, with any given “main call” only occurring once or twice in the data set of 255 matches, we feel that the level of agreement found is very positive. Thus the Whale FM data provided an independent measure of reliability to our results.

Sayigh L, Quick N, Hastie G, Tyack P. 2012. Repeated call types in short-finned pilot whales, Globicephala macrorhynchus. Marine Mammal Science, DOI: 10.1111/j.1748-7692.2012.00577.x

A First Peek At the Whale FM Data

Hi there! My name is Sander von Benda-Beckmann, one of the members of the Whale FM science team. The Whale FM project is about understanding what whales are saying. On the Whale FM site people are asked to look for similar Killer Whale and Pilot Whale calls and help us establish the repertoire of sounds that these magnificent animals use. In this post I briefly want to take you for a look behind the scene to see what we are doing with the whale calls that have matched.

We have been connecting all whale calls that were matched into a map that we call the Whale FM Mesh. A zoom-in on this map of calls is shown in the picture below. You see a lot of colored circles connected together, some forming larger clusters. We were so happy to see these clusters of calls, because they are exactly what we have been looking for! Let me tell you why…

Each such cluster is a set of whale calls that constitutes a potential call category. If the circles are larger, more people have matched them to their neighbors. The colors indicate different recordings (recorded at different locations and for different groups of animals). If a cluster of calls contains many different colors, it is likely that this type of calls is shared amongst different whales, which is very exciting! Some groups like the top green one consist of primarily one color, indicating that these calls are only used by the single animals, or the group that it was living in. You can still see many little groups of pairs of calls. These are either not really part of a real call category of a haven’t been matched to enough other calls to make sense of yet. We have about 150,000 matches performed by almost 10,000 volunteers. This is already excellent, but still need more matches to link up all the little groups.

So I hope you liked seeing a bit of what we are doing with all the work you are doing. The first look looks very promising! We still need more careful analysis to see what is going on, but we are very excited with the data that the Whale FM volunteers are providing us with. Stay tuned to Whale FM!