Sitting on a Cliff Waiting for some Dolphins

Notes From the Field

Hello all. I've just returned from a month in the Scottish highlands studying dolphins. Go ahead, you can be jealous and I'll forgive you. But it wasn't all fun and games, even before the tick bites and sunburns. I was there on a collaborative project with the University of Aberdeen Lighthouse Field Station (check them out and donate money) who are also studying dolphins using passive acoustic recorders.

The goal was to estimate the detection functions for  CPOD and SM acoustic detectors that comprise the majority of the data collection for my PhD.

Sitting on a cliff waiting for some dolphins...

I discussed detection functions and probabilities in my last post but, in case you forgot (or had better things to do than read my posts) these values are useful critical for estimating the total area monitored by acoustic recorders; it is impossible to produce accurate density and abundance estimates without them. Therefore, this is an important piece my PhD research.

About that PhD...

It occurs to me that I haven't really discussed what the hell I'm doing in Scotland in the first place. Well, you will be stunned to find out that I'm working on a Phd (gasp!) studying bioacoustics (double gasp!). The project is looking at the spatial and temporal habitat use of a marine top predator using two types of acoustics detecting devices.

Ugh.. fine. I study dolphins.
For an example of what studying dolphins does NOT look like see below:

I often just say I study bioacoustics.

Right, anyway what I do do is look at how wild animals use the Eastern Scottish Coast. My study is specificlly looking to characterize the habitat use of these top predators before the construction of offshore windfarms. Then, if any long term changes are observed they can be characterized and used to understand other potential anthropogenic effects.

The data I'm working with primarily comes from 30 CPOD and 10 SM units deployed up and down the coast. The CPODs report when dolphins are present (but don't tell you what type) and the SM units record sounds for approximately 20 minutes every hour and require special detectors and classifiers to tell you what type of dolphins were present and when. Morover, the CPODs can't detect dolphin whistles while the SM units miss a large percent of the dolphin clicks.

Below is a schematic of the deployment locations of all acoustic recorders.

The 30 deployment locations of the acoustic recorders used for my PhD
 (See Marine Scotland Science for full details)

As you can see, with such a broad area covered by the detectors, knowing how far the detectors can "hear" the animals is critical to understanding the outputs of the research. This again might depend on the ambient noise level, the local bathymetry (seascape) and how loud the dolphins call.

Getting the Detection Function

In the last post I showed an oversimplified version of a detection function. As great as it would be, it's not anything like what we would expect from real data. In general, detection functions tend to follow half-normal or hazard rate curves. We fit those curves to observed data that tells us how far away we can see or hear an animal. Below is one such example taken from terrestrial data. The tricky part is fitting those curves to the data as best as possible without over or under adjusting the curve, but I'll leave that to a statistician to explain.

Detection function (blue line) estimated from detection probabilities at
various distances (x-axis) from the observer

For me, what I need to do is determine how far away the detectors can hear the animals and that's where the fieldwork sets in.

The Cromarty Firth is a relatively deep channel in North East Scotland. The area is also a hotspot for pseudo-resident bottlenose dolphins. Dr. Paul Thompson and the rest of the crew from the Lighthouse Field Station have been studying these animals for many years now. In recent  years they've deploying CPOD dolphin/porpoise detectors in order to monitor changes resulting from the construction and operation of large industrial plants along the firth. These are the same type of detectors used in my study. Thus, if I can estimate the detection function in a known area I'll be better prepared to apply it to the unknown areas (all of the eastern scottish coast).

Sitting on a cliff... 

Cromarty Firth is special because, in addition to the deep water channel, it is guarded by two large sea-cliffs. These cliffs provide an excellent platform from which to watch dolphins as they enter and exit the Firth. So, to figure out what the detection functions might look like, I sat on one of those cliffs for several hours a day and waited for the dolphins. When the animals came by, I took their picture and using a special device recorded the bearing and pitch of the camera. Once you the height of the cliff, pitch of the camera and bearing it is a relatively simple matter to figure out where the animal is. All this to say that the 7th grade geometry does have a place in life.

How to estimate the detection function

In a few weeks the crew there will retrieve the acoustic detectors and has graciously offered to share the data with me. My plan is to combine my visual observations with the acoustic detection to ask what proportion of dolphins were detected at what distances? This will give me an idea of what the detection function for the CPODs (and SM units) will look like. Better yet, because the SM unit is capable of recording the ambient noise level, this will help us understand how the detectability of the dolphins will change with increasing noise level. Hopefully, the results will look something like this:

Graph showing different detection functions. I hope to find similar results where
green dots would represent low noise conditions and red ones high noise conditions

However, even if all the data I collected are lovely and clean (HA HA HA!!!) the results from this small study will unlikely be the be-all end all to the detection function story. Like all mammals, dolphins change the amplitude of their calls based on the noise they hear. In humans, this is called the cocktail party effect; where as a room full of people talking becomes louder, we instinctively raise our voices such that our friends can hear us. If dolphins do this in the presence of ambient noise, then the total area monitored by the detectors will remain relatively unchanged. 

Will The Detection Function Change of Not?

That's the question I'm after here. In my last post I specifically said it could, in this post I'm proposing that it might. Unfortunately, we just don't know (hence the study). I suspect that, like humans, it will be a bit of both. At lower noise levels the detection function won't change because the animals can compensate by increasing their call amplitude. But at a certain point the animals will no longer be able to do so. Not unlike a noisy pub where you finally give up trying to talk to your friends and just listen to the music.  

So, anyway, you've made it through a fairly dense post so here, have a dolphin picture taken from the top of the Cromarty South Suitor.

Dolphins in the Cromarty Firth

Of course a special HUGE thank you to the Lighthouse Field Station Crew (in no particular order: Paul, Barbra, Georgia, Becky, James, Karen, Annabelle, Kelly, Ewan, Tim, Max and Jim). Thanks for reading! 

---

* Just because making the area monitored as a function of ambient noise animation was so much fun, here it is again