AudioMoth with external microphone

In a previous post I mentioned that OpenAcoustic’s AudioMoth was going to be upgraded to allow users to attach an external microphone. This new version was released during winter 2020/21 and in this post I will share first impressions of using this and some comparisons with other recording gear.

Out of the box

The external 3.5mm jack does not come fitted and has to be soldered onto the six contacts on the top of the unit. It’s a bit fiddly, but even with my limited soldering experience I managed to do it OK. It is clearly stated on the OpenAcoustic’s website but worth reiterating that once you have attached the audio jack, the device will no longer fit in the standard AudioMoth case.

To use an external mic you simply plug it in – i.e. you don’t need to enable anything in the Configuration App. I opted for a Primo EM272 capsule mic as these are affordable and good quality. They come with a decent length of cable so can be routed into a parabolic reflector if required.

Comparisons

To field test the AudioMoth with external mic I ran the following equipment side by side over 2 nights in February 2021:

  • Dodotronic Hi-Stereo parabolic mic recording to MixPre3 recorder, set to 28dB gain
  • Wildlife Acoustics SM4mini, set to 24dB gain
  • AudioMoth with EM272 external mic in a parabolic reflector; gain set to medium. The AudioMoth was put inside a small cliplock food storage box with a hole drilled in the end for the cable to exit. I crudely sealed the hole by wrapping a strip of an old bike innertube around the cable.
  • AudioMoth in a standard AudioMoth case; gain set to medium
Three of the devices used in the comparison. In the foreground the EM272 mic in a parabolic reflector; in the background, left to right, the AudioMoth unit with the external mic plugged in, the AudioMoth in standard case, and the SM4mini.

I haven’t done any formal measurements of signal strength. Instead, as for most casual nocmig purposes we visually check spectrograms to detect calls, I have just produced standard spectrograms as they would appear when processing a recording. I think this shows how easily a call recorded on the different gear would be detected by a human observer. In each of the following images the spectrograms are in the order Dodotronic (both channels, or just the strongest), SM4mini, AudioMoth with external mic, AudioMoth in box. Note that the recordings made with a parabolic reflector may not show a major improvement over those without a reflector if the bird was outside the main receptive zone in front of the reflector.

Redwing 27 Feb 2021

A single Redwing call. The note is identifiable on all equipment but weaker in the AudioMoths. There is some marginal improvement with using the external mic.

Little Grebe 27 Feb 2021

A pair of Little Grebe ‘scream’ calls. The upper harmonic was invisible on both AudioMoth recordings. The AudioMoth with external mic is definitely an improvement on the device in the case, but still inferior to the Dodotronic and SM4mini. I suspect I might have missed this call entirely on the recordings from the AudioMoth in the case.

Little Grebe recorded using Dodotronic parabolic mic, SM4mini, AudioMoth with external EM272 mic and AudioMoth in a standard case

Robin 27 Feb 2021

A single Robin flight call, visible on all recordings, but with detail slightly variable across the recordings.

Coot 27 Feb 2021

A sequence of high squeak calls. This bird was off to the left of the garden as evident from the Dodotronic channels where the call is much stronger on the left channel. The AudioMoth with external mic is almost on a par with the right channel of the Dodotronic, and noticeably better than the device in the case.

Water Rail 28 Feb 2021

A sequence of four calls. Again the AudioMoth with external mic shows a small but significant improvement on the device in the case.

Conclusions

Based on these recordings the AudioMoth with external mic showed some small improvements on the AudioMoth in the case, though they weren’t as significant as I was expecting. It is possible this is because the recordings were made using a parabolic reflector, and if none of the birds were in front of the reflector, this might not have provided the best test. The comparison below was made without the parabolic reflector and shows an improvement in detection of the Black-headed Gull call over the AudioMoth in the case. I’ll repeat this test when there are hopefully a few more birds moving and post an update. Personally, I think these small improvements, and the confidence of being able to have the device in a watertight box and microphone external mean this is still a better option than a normal AudioMoth.

Black-headed Gull recorded using Dodotronic parabolic mic, SM4mini, AudioMoth with external EM272 mic and AudioMoth in a standard case
SM4-mini positioned in a parabolic reflector

Nocmig with an SM4-mini

Earlier this autumn I bought a Wildlife Acoustics SM4-mini to use for a long-term nocmig project in North Norfolk. I set this up with a spare fibre glass parabolic reflector that was generously given to me, to improve the range of the microphone. The whole setup was fixed to a garden shed in a coastal village and angled slightly towards the coast and away from nearby trees.

The kit

The SM4-mini is pretty compact (approx. 5″ x 6.5″ x 1.5″), runs off 4 AA batteries and takes standard SD cards. The box is rugged and waterproof and has slots and holes for attaching it to things. The setup includes a single microphone but there is a second port to add an optional second microphone for stereo recordings.

One feature I really like is the mobile app used to configure the device. The device broadcasts a Bluetooth signal which provides basic info like how many recordings have been made to date. Open the box and press the pair button and you connect your mobile app to the device to control all the settings. It is highly customisable. For example, you can configure your own on/off cycles, or allow the devise to use sunset and sunrise times as the basis of the routine.

As you’d expect, you can control the sample rate, the length of files and the gain setting. You can also set a delayed start time. On my last deployment I knew the weather was going to be horrendous for a few days so I delated the start until after the bad weather to save on battery/card life.

Given the particular batteries, card and configuration you’ve selected, the app will estimate when the batteries will be depleted and when the memory card will be full.

The sounds

So what about the recording quality? I think it is pretty good as can be judged from the following recordings:

Common Scoter, Blakeney, 09/10/20 22:26. Unedited

Gadwall, calls of male and female, Blakeney, 29/09/20 01:23. Unedited

Great Bittern, Blakeney, 14/10/20 05:23. Unedited

Migrating thrushes and local Robins. Blakeney, 14/10/20 06:40. Low pass filter to reduce sound of central heating boiler

These are cherry picked interesting examples from my most recent deployment but I think they are representative of the quality the recorder provides. The recorder has run continuously since mid August with no failures or errors. I haven’t done any side-by-side tests of this (with or without the reflector) against other recorders.

Verdict

I’m very pleased with the device for the following key reasons:

  • it’s waterproof
  • it’s easily programmed via a nice mobile interface
  • it runs autonomously for 3 weeks (on my settings)
  • it produces nice clear nocmig recordings

When I retrieve the recorder I will try to make some side-by-side recordings with other equipment.

PS: for anyone considering mounting an SM4-mini in a parabolic reflector, getting the microphone element at the precise focal point of the dish is critical. Without removing the foam windshield of the mic it is hard to know where the mic element is located, but I was told by Wildlife Acoustics that it is 16mm (±1mm) from the outside of the case.

Spotted Flycatcher by Mike Pennington CC-BY-SA

100 milliseconds of confusion

In April 2020 Magnus Robb and The Sound Approach published revised identification criteria for Spotted Flycatcher, Pied Flycatcher and Robin. I find these species, which all produce short, high frequency, sometimes buzzy calls, rather difficult to identify, perhaps partly due to presumptions(*) about what should and should not be flying around in Cambridgeshire in May. I have previously recorded a few of these calls but in May 2020 I recorded 20 individuals, some with multiple calls, providing an opportunity to try out these new criteria on a range of calls. Of course, none of these birds can be identified in any other way, so there is no way of independently verifying the identities.

* Note that in May I assume there should be a reasonable passage of Spotted Flycatchers at this latitude and that Robin migration should have long finished. So my assumption when processing all these calls is that they should be Spotted Flycatchers…

 

6 May

Bird 1: Rather faint. Two bands just visible. Call length = 105 ms, lower band 4.8–5.1 kHz, upper band 5.5–6.6 kHz. No spikes but call length and low frequencies suggest Spotted Flycatcher, although pro-Robin features are that bands approach much closer than 1 kHz. Identity uncertain.

20200506-0251-CB4-Gillings-PS-XX-(flycheck)

7 May

Bird 2: Distant; only one band visible. Longish call (104 ms) suggests Spotted Flycatcher. Frequency of band 5.7–5.9 kHz would be low for even the lower band of a Robin, so suggests this is a Spotted Flycatcher.

20200507-0201-CB4-Gillings-PS-SF-(check)

Bird 3: Another distant call with only one band visible (length = 106 ms, 5.9 kHz) and presumably also a Spotted Flycatcher. Given how variable these calls appear to be, yet how similar this is to the previous bird (which was only 6 minutes earlier), it is possible this is the same individual circling around.

20200507-0207-CB4-Gillings-PS-SF

Bird 4: Two calls 13 s apart, presumably from one individual. Higher frequency (6.3–7.7 kHz) than other birds but call still quite long (99 ms and 110 ms). No obvious spikes but fine shallow modulations throughout both calls. Bands close and tapering, plus frequency range suggest Robin but fine modulations seem not to fit. Identity uncertain.

20200507-0335-CB4-Gillings-PS-SF-(check) A20200507-0335-CB4-Gillings-PS-SF-(check) B

8 May

Bird 5: A long call (127 ms) with suggestion of shallow spikes on lower band. This and lowish frequency 5.3–6.6 kHz suggest probably a Spotted Flycatcher.

20200508-2259-CB4-Gillings-PS-XX-(fly)

11 May

Bird 6: a short call (89 ms), no spikes, slight kinks, tapering, and higher frequency 6.6–7.4 kHz suggest possible Robin.

20200511-2216-CB4-Gillings-PS-XX-(zzz)

Bird 7: long call (134 ms) suggestive of Spotted Flycatcher but lack of spikes and higher frequency (6.6–7.7 kHz) point more towards Robin. Identity uncertain.

20200511-2330-CB4-Gillings-PS-XX-(zzz)

15 May

Bird 8: intermediate call length (96 ms) and frequency 5.9–7.1 kHz. Call a bit distant and broken, especially top band. Possibly a Spotted Flycatcher.

20200515-0016-CB4-Gillings-PS-SF-(check)

Bird 9: Two calls, 13 s apart. Long calls (112 ms) and general shape suggest Spotted Flycatcher but relatively high frequency (6.1–7.7 kHz). A Spotted Flycatcher at upper end of range?

20200515-0128-CB4-Gillings-PS-SF-(check) A20200515-0128-CB4-Gillings-PS-SF-(check) B

19 May

Bird 10: A very long call (141 ms) with possible small spikes on lower band. Above average frequency (5.6–7.3 kHz) but within the range for Spotted Flycatcher.

20200519-0136-CB4-Gillings-PS-SF-(check)

20 May

Bird 11: A very long call (158 ms) with 3 distinct spikes and low frequency range (5.2–6.5 kHz) indicative of Spotted Flycatcher.

20200520-0005-CB4-Gillings-PS-SF

Bird 12: Three calls recorded, first very faint (not shown); calls spaced 17 s and 8 s.  Longish calls (113 ms) with 3 spikes. First closes call has suggestion of third higher band. Structure and mid-range frequencies (5.2–7.6 kHz) matching Spotted Flycatcher.

20200520-0144-CB4-Gillings-PS-SF A20200520-0144-CB4-Gillings-PS-SF B

Bird 13: Two calls 11 s apart. Both calls have multiple bands, which are generally tapering with obvious kinks suggestive of Robin. However, frequencies for strongest bands are low for Robin: lower band 5.0–5.7 kHz and upper band starting 7.0 kHz but dropping to 6.0 kHz. Robin calls at 6 kHz seem rare judging by Sound Approach info. First call also quite long (107 ms; second call apparently only 83 ms). Note that the apparent tapering of the call could be exaggerated by the presence of additional bands; the two strongest bands show tapering similar to other calls above. Identity uncertain.

20200520-0251-CB4-Gillings-PS-SF A20200520-0251-CB4-Gillings-PS-SF B

21 May

Bird 14: Fairly average calls – medium length (95 ms) and  medium frequency 5.8–7.4 kHz suggest a spike-less Spotted Flycatcher.

20200521-0119-CB4-Gillings-PS-XX-(zzz)

25 May

Bird 15: 91 ms length. Frequency range 6.2–7.7 kHz are at the upper end of the Spotted Flycatcher range. Another spike-less Spotted Flycatcher?

20200525-0227-CB4-Gillings-PS-XX-(zzz)

Bird 16: a long call (155 ms) and fairly high frequency (6.9–7.9 kHz) with fine modulations in the upper band. Features seem contradictory, identity uncertain.

20200525-0243-CB4-Gillings-PS-XX-(zzz)

26 May

Bird 17: Medium length call(97 ms) but very high frequency (7.2–8.6 kHz). These features plus broken lower band suggestive of Robin.

20200526-0000-CB4-Gillings-PS-XX-(zzz)

27 May

Bird 18: rather faint, short duration (85 ms) and frequency at upper end of Spotted Flycatcher range (6.2–7.8 kHz). Some hints of spikes, probably a Spotted Flycatcher.

20200527-0305-CB4-Gillings-PS-XX-(zzz)

29 May

Bird 19: Longish call (129 ms), no spikes and moderately high frequency 6.1–7.9 kHz). Presumably a Spotted Flycatcher?

20200529-0302-CB4-Gillings-PS-XX-(zzz)

30 May

Bird 20: Two calls, 23s apart, presumed same bird, though very different in structure. First call lacks spikes but has a big downward kink suggestive of Robin. However, call length (124 ms) and frequencies (4.6–7.8 kHz) more suggestive of Spotted Flycatcher. Second call very long (170 ms) with small spikes and frequencies in Spotted Flycatcher range (5.6–6.7 kHz). Presumed Spotted Flycatcher.

20200530-0136-CB4-Gillings-PS-SF-(check) A20200530-0136-CB4-Gillings-PS-SF-(check) B

 

Conclusions

It appears that during May I recorded 2 probable Robins and 13 probable Spotted Flycatchers, with 5 more birds as yet unidentified. For a rapidly declining species it is nice to have recorded several Spotted Flycatchers. For the Robins, if indeed they are Robins, it begs the question, what are they doing migrating in May? As ever, nocmig raises just as many questions as it answers.

Identifying these 20 birds has been challenging, and while some birds fall neatly within the ranges described on the Sound Approach pages, some have conflicting features. In particular, many of the pro-Spotted Flycatcher calls had frequencies that were at the upper end of (or slightly above) the limits quoted by Magnus et al. As he points out, there is still much to learn for these and related species, and hopefully greater sample sizes of recordings from different places will help to clarify the picture. I imagine I may revisit the identities of these 20 birds as more information comes to light.

 

Metrics

The table below summarises the metrics extracted from the calls. Frequencies are measured excluding spikes. Note also for calls 2A and 3A I do not know whether the visible band was the lower or upper band. All measurements were made in Raven Lite. The spectrograms were produced in Python using bespoke code with a window length of 150 samples and overlap of 90%.

Call Number of bands Length (ms) Number of spikes Lower bar Freq. min Lower bar Freq. max Upper bar Freq. min Upper bar Freq. max
1A 2 105 0 4863 5107 5453 6572
2A 1 104 0 5657 5880 NA NA
3A 1 106 0 5901 5921 NA NA
4A 2 99 0 6674 6877 7305 7691
4B 2 110 0 6389 6633 6776 7712
5A 2 127 2 5372 5433 6247 6572
6A 2 89 0 6267 6735 6959 7366
7A 2 134 0 6572 6755 7203 7712
8A 2 96 0 5921 6267 6715 7142
9A 2 112 0 6328 6491 7508 7671
9B 2 112 0 6104 6165 7081 7203
10A 2 141 1 5575 6064 6877 7325
11A 2 158 3 5209 5290 6348 6532
12A 2 113 3 5514 5738 7142 7590
12B 2 113 3 5209 5514 6918 7223
13A 5 107 0 5229 5758 5982 7040
13B 3 83 0 5005 5616 5779 6979
14A 2 95 0 5840 6267 7020 7366
15A 2 91 0 6165 6328 7427 7691
16A 2 155 0 6493 6700 7690 7944
17A 2 97 0 7230 7437 8243 8565
18A 2 85 0 6194 6447 7391 7828
19A 2 129 0 6148 6355 7207 7898
20A 2 124 0 4628 5618 6884 7828
20B 2 170 2 5618 5664 6539 6700

 

 

Spectrogram of Fox Vulpes vulpes

Nocmig Top Tips

It is great to see so many people taking up nocmig recording at the moment. It might seem like getting the equipment sorted is the battle won but there are still some steps that will make your nocmig experience more productive and enjoyable. These ‘top tips’ were compiled in conjunction with David Darrell-Lambert (@birdbrainuk) and James Lidster (@james_lidster), experienced nocmig recorder from London and the Netherlands.

1. Identifying out of context invisible birds is tricky and there’s no shame in not knowing a particular call. Everyone, new and experienced alike gets calls they can’t place but there’s a great community of people keen and willing to help.

2. Take some time to learn the ambient sounds of your local environment. There will be hundreds of clicks and squeaks, not to mention dogs, foxes (see below), cats and other critters (and humans) that will try to fool you. Once you learn the visual signs you can process recordings without listening to every one.

3. Start off focusing on clear, close sounds. Faint sounds are often impossible to identify, even by experienced nocmig recorders.

4. Indistinct notes from the Robin two streets over will sound like almost any rare bird you can image. Get used to what different night singers sound like. Look for regularly spaced song phrases. Also, song from stationary birds often look fuzzy on spectrograms due to slightly delayed echoes off buildings, whereas flight calls are usually crisper, even if they’re quieter.

5. Get Audacity configured in nocmig mode so you can produce clear spectrograms to identify species and share with others.

6. Always include both time and frequency axes when sharing spectrogram images. When sharing sound clips, don’t crop them too tight. Include 2–3 seconds before and after the call so ears can get accustomed to the volume before the main event.

7. Familiarise yourself with the range of calls of the common nocturnally active species. Tawny Owls produce a wide array of calls (see here). In the UK the following account for over 75% of all nocmig calls logged on Trektellen during 2016–19:

  • Redwing (61,000 calls logged on Trektellen)
  • Song Thrush (14,000 calls)
  • Blackbird (5000 calls)
  • Coot (2000 calls – the Great Tit of the night, making lots of different calls)
  • Moorhen (5000 calls)
  • Water Rail (380 calls – scarce but variable and confusing)
  • Little Grebe (380 calls – beware of this Whimbrel impersonator)
  • Grey Heron (1800 calls)
  • Wigeon (2000 calls)
  • Oystercatcher (1500 calls)

8. If the spectrogram looks like a Redwing but doesn’t sound like one, it’s probably a Redwing. If it sounds like a Redwing but the spectrogram doesn’t look like one, it’s probably also a Redwing! They vary massively, see here for examples.

9. Do you have an odd sound that is clearly a flying bird? Then check out the range in Coot calls as well.

10. You may pick up wing beats. It is usually not possible to identify these without calls.

11. Register on xeno-canto and upload mystery recordings to get feedback and suggestions.

12. Log interesting records on BirdTrack and when you’ve really gotten hooked, consider submitting structured counts via Trektellen.

EM172 lapel mic, Zoom H4n Pro audio recorder, Audiomoth and mini USB mic (photo: Simon Gillings)

Equipment Comparison February 2020

On one of the few calm nights in February 2020 I tried to make simultaneous recordings of the same flight calls using a variety of equipment. I tested:

  1. A cheap USB microphone connected to a desktop PC. Mic level set to 90/100.
  2. An AudioMoth in a plastic bag. Gain set to Medium
  3. An AudioMoth in a homemade waterproof case. Gain set to Medium.
  4. A lapel microphone (EM172) with a digital audio recorder (Zoom H4n Pro; record level set to 80/100).
  5. A Dodotronic parabolic microphone with a digital audio recorder (Sound Devices MixPre 3; gain set to 25dB)

There was very little bird movement but I got useful comparisons for a Redwing and a Moorhen pass. Below is a recording of each call and a side-by-side comparison of the spectrograms.

Redwing

A single Redwing call at 2010. The call is a bit atypical but still conveys the quality of the recording using the different equipment. Without adjusting the gain on the USB recording this call is inaudible. It is visible and audible on all the other equipment setups but is stronger and more clearly defined on the final two recordings. The bird was not directly overhead, which explains why the lapel mic and parabolic mic have similar performance. Had the bird been directly above the parabolic dish that recording would have been much louder.

Cheap USB mic and computer:

AudioMoth in bag:

AudioMoth in case:

Digital audio recorder and EM172 lapel mic:

Digital audio recorder and parabolic reflector:

Spectrogram comparison
Redwing comparison 2

 

Moorhen

The Moorhen call was picked up and visible on all recorders, though it was still faint on the USB. On the two AudioMoth recordings only the first kek-kek-kek series is clearly visible whereas on the Lapel mic and Parabolic mic recordings the subsequent kek-kek notes are also clearly visible.

Cheap USB mic and computer:

AudioMoth in bag:

AudioMoth in case:

Digital audio recorder and EM172 lapel mic:

Digital audio recorder and parabolic reflector:

Spectrogram comparison
Moorhen comparison 2

 

Conclusion

Based on this very limited comparison the USB mic would miss some birds, though I should note that the mic used here is quite old and newer models might be more sensitive. More distant calls, calls of quiet birds, or calls where fine detail is needed for identification (e.g. flycatchers) might be easily missed or hard to identify with the AudioMoth. I did not run the AudioMoths on the high gain setting which might partly explain the slightly poorer detections. Previously I have found high gain generated too much white noise (hiss) in the recordings. Next time there is a break in the wind I will try again…

Postscript

The wind dropped again briefly and I was able to make a comparison of two AudioMoths, one with gain set to Medium and another with gain set to High. Again not many birds moving so not many comparisons possible. Of three Redwings detected by parabola: i) one was not detectable via either AudioMoth; ii) one was missed on Medium gain and barely visible on High gain; iii) one was picked up by both with negligible difference in sound quality. Of two Song Thrush calls detected by the parabola: i) one was missed by the Medium gain AudioMoth and just visible on the High gain one; ii) the second call was visible on all as can be seen in the following comparison:

ST call cf

Note this bird was not directly overhead, as evident from differences between the left and right channels of the parabolic recording. Without knowing these calls were present from the parabolic recording I may well have missed some of them in the AudioMoth recordings.

The lack of birds prevented a more detailed comparison but this does suggest the high gain setting might be better for the AudioMoths for nocmig.

Common Scoters by Jon Heath

Common Scoter Nocturnal Flight Calls: long calls?

Since Magnus Robb posted a Sound Approach article describing the short ‘peep‘ calls of migrating Common Scoters, this species has been a target for many UK nocturnal migration recorders. Late March and early April 2018 were memorable for the number of scoters crossing southern and central England, with many people recording the species for the first time. Here’s one such recording, from a typically noisy night in Cambridge (03:13, 14th April 2018).

 

At the time I didn’t pay much attention to the long, flat beeps, passing these off as an anthropogenic sound — maybe a reversing siren of a truck! But on 30th August 2018 I recorded a similar flat whistle, accompanied by wing beats that sound distinctly duck like. In the original Sound Approach article, Magnus Robb mentioned that Black Scoter give longer calls, but that would be an outrageous claim for Cambridgeshire…

 

Some Common Scoter recordings on Xeno-canto include longer calls, but these are at the same frequency as the peep calls, around 1.8 kHz, and sound like drawn out peep notes, as in this example I recorded in Skagen, Denmark in May 2018. Notice also the typical “steam train” wingbeat sounds.

 

In contrast, the calls from August were around 2.2 kHz, with slight modulations and slightly dropping in frequency.

Common Scoter long flat whistles 0121 30/08/2018 Cambridge (Simon Gillings)
Common Scoter long flat whistles 0121 30/08/2018 Cambridge (Simon Gillings)

Looking back through my scoter recordings I realised that “reversing vehicle noise” from 14th April was actually the same long flat calls. And then on 25th September 2018 I recorded another group of Common Scoter over Cambridge, again giving typical peep calls and 2.2 kHz long calls. I’ve since recorded these calls in combination again, in October 2018.

 

 

 

Conclusion

Unless each of these flocks had a second species lurking amongst the scoters, these recordings suggest that nocturnally migrating Common Scoters sometimes give long calls instead of, or as well as the more familiar peep notes. I don’t know how common these calls are or which ages or sexes produce them, but I would interested to hear from anyone else who has recordings of similar calls.

 

Credits: recordings © Simon Gillings; Common Scoter photo © Jon Heath.

Juvenile Goldfinch, by Simon Gillings

Nocturnal Goldfinches: never say never

The ‘ink’ had barely dried on my last post about the rarity of nocturnally migrating Goldfinches when I heard that Tim Jones had recorded nocturnal Goldfinches at Spurn twice in October 2017. Then Magnus Robb mentioned he’d recorded them in southern Portugal 13 times over nine nights in October and November 2017, plus once in April 2018. Aat Schaftenaar also provided details of his two nocturnal birds mentioned in the previous post. Tim, Magnus and Aat kindly shared their recordings, providing an opportunity to confirm what nocturnal Goldfinches sound like and how similar their spectrograms are to Ortolan calls.

First, here’s a recording of 5 seconds of Goldfinch nocturnal flight calls, probably a single bird, recorded in Portugal in November 2017 by Magnus Robb. The overall sequence is reassuringly similar to diurnal calls in consisting of single or paired upward-inflected notes, with consecutive notes often varying in frequency, giving a ‘bouncy’ impression.

 

Here’s one of Tim’s from Spurn, recorded at 00:51 on 17th October 2017:

 

And one of Aat’s from the Netherlands, recorded in March 2016:

 

Of the 18 files I examined, all consisted of sequences of these upward-inflected call notes. In 12 cases the passing bird(s) gave both paired and single notes (as in the example above); two passes consisted of only paired notes, and the remaining four consisted of only single notes. All recordings were of single birds except a small flock recorded by Tim Jones at Spurn (00:20, 17/10/2017).

I described the call notes as upward-inflected, and this becomes more apparent when the spectrogram is zoomed in. Each call note starts low and rises in frequency. Chris Batty described these as looking like the forward slash ( / ) character on a computer keyboard. They vary somewhat in length and slope, but they all rise in frequency. None of the notes begin with a descending part, and while some have a terminal drop in frequency, this is very small in comparison to the range of the rising part (e.g. penultimate note below). Overall they lack the characteristic reversed ‘N’ shape of Ortolan Bunting plik calls.

Goldfinch nocturnal flight calls
Goldfinch nocturnal flight calls (Portugal, 16/11/2017, Magnus Robb)

 

Conclusion

These recordings confirm that Goldfinches do, on occasion, migrate at night and call whilst doing so. The calls they give are identical to diurnal calls and both the pattern of calling and the structural of individual call notes are distinct from Ortolan Bunting plik calls. These results suggest it should be straightforward to eliminate Goldfinch as a potential confusion when identifying a nocturnal Ortolan Bunting.

Thanks to Tim Jones, Magnus Robb and Aat Schaftenaar for sharing their Goldfinch NFC recordings.

 

 

Redwing in flight by Tim Jones

Autumn thrushes

The first tseee calls of Redwings arriving into Britain and Ireland have been heard in recent nights and Song Thrushes have been on the move too, with more Continental immigrants soon to arrive to supplement our local breeders. As the BirdTrack reporting rate graph below shows, Fieldfares arrive a little more gradually and Blackbirds continue to arrive through November. If you’re lucky there’s always the chance of a Ring Ouzel. This blog summarises the main nocturnal flight calls of these species.

thrush_phenology_BT
Graph showing the percentage of BirdTrack complete lists in Britain & Ireland on which different thrush species were reported each week.

 

All the nocturnal flight calls described below are also given in daytime, during flight and sometimes when perched. They were all recorded as nocturnal migrants over Cambridge, UK.

 

Redwing

The characteristic tseee of an autumn evening. Calls vary a little in length and frequency but they all descend slightly in frequency. In the following recording you can also hear some soft puk calls too.

 

Blackbird

Blackbird srrri flight calls are a little like Redwing calls but are shorter and show less of a frequency drop. They also have a noticeable fluctuating or modulated quality, making it sound like the call is slower than a Redwing tseee call. In the following clip, a Blackbird call is followed by a Redwing call.

 

Blackbird sometimes pair these calls into a double call. The second note is often slightly lower pitch, as in the following example

 

Rarer still, Blackbirds will string together three of these calls into a triple call. The following example (apologies for quality) contains single, double and triple calls.

 

Song Thrush

A single high pitch zit call. A migrant might only give one of these calls as it passes overhead, but sometimes they can give two or three calls a few seconds apart. Very occasionally the zit call is repeated with almost no gap.

 

Fieldfare

At night, much scarcer than the previous three species. Main call a dry chattering, slightly chuckled schack-schack-schack, sometimes preceded by a wheezy rising eep call.

 

Ring Ouzel

Significantly rarer than the other autumn thrushes. Two main call types. The first, which is often heard from migrants in the day, is a very distinctive hard tac-tac-tac call.

 

The Ring Ouzel tac notes are shorter and sharper than corresponding tup notes of Blackbird, which can be heard in the following clip.

 

The second Ring Ouzel call is a rising squeaky bubbling call. It is superficially similar to Fieldfare calls, though it is faster and softer, with a rising quality.

(drake Teal calling in background)

Ring Ouzels also give a rolling prrrt call (e.g. here) though I have not yet recorded it myself.

© all recordings copyright Simon Gillings. Redwing photo by Tim Jones.

Ortolan Bunting by Jon Heath

Ortolan Buntings and Goldfinches

It’s peak season for migrating Ortolan Buntings in northwest Europe and in addition to several sightings and ringed birds, several have been recorded by the UK’s fledgling ‘nocmig network’. As well as the now regular locations of Portland/Weymouth and Poole Harbour area, birds have also been recorded (so far) over London, Surrey and Cambridge. They’re being recorded in several locations that don’t have any history of visual records of Ortolans, prompting lots of head-scratching about the likelihood of detecting such species and questions about identification criteria. The Sound Approach have produced two very detailed blog posts on the identification criteria (here and here), describing the common call types and their potential confusion species. With reference to the commonly used plik call they say:

European Goldfinch Carduelis carduelis has a call similar to Ortolan Bunting’s plik, although it normally uses this call in combination with others. Goldfinches usually move around in tight flocks, and we have never recorded one migrating at night, despite it being such a common species…Note the variation in pitch and rhythm, giving a bouncing effect. On odd occasions when a single goldfinch repeats only its most plik-like calls, these are likely to be given in twos as well as singly, and with much shorter gaps than an Ortolan Bunting

The suggestion that Goldfinch calls are a potential confusion for Ortolans has been raised as an issue, particularly in those counties where Ortolan Buntings have never been recorded in the daytime.

Setting aside for a moment how similar the calls are, it is helpful to quantify the potential for confusion from night flight calls of Goldfinches. I checked on the timing and reporting rate of diurnal and nocturnal Goldfinches and Ortolan Buntings on Trektellen. The many thousands of hours of day and night recording submitted there should give a good indication of the behaviour of these species.

Diurnal movements

First, here’s the timing of Goldfinch diurnal autumn migration (all countries, all years), showing a very low level of movement through late summer picking up in late September to a broad peak through October to November.

Trektellen_GO_diurnal
Figure 1. Timing of Goldfinch autumn diurnal migration, from Trektellen

For comparison, here’s the diurnal autumn phenology of Ortolan. Ortolans in daytime are clearly much earlier than Goldfinches and at the time when most Ortolans are moving, Goldfinches movements have barely got started. These results are largely driven by data submitted in the Netherlands, Germany and France where Ortolan passage is well understood.

Trektellen_OB_diurnal
Figure 2. Timing of Ortolan Bunting autumn diurnal migration, from Trektellen

Nocturnal movements

So now looking at the phenology of nocturnal movements of Goldfinch, here’s the corresponding plot.

Trektellen_GO_nocturnal
Figure 3. Timing of Goldfinch autumn nocturnal migration, from Trektellen

 

There’s no error in the graph. In 17,564 hours of nocturnal monitoring during autumn, no Goldfinches have been recorded. You might notice in the date range of the graph above that I have omitted 2011. That’s because there is a handful of sound-recorded Goldfinches submitted from Besh Barmag, Azerbaijan by Michael Heiss (note that totals here are currently numbers of calls, not numbers of birds). Michael records from sunset to sunrise, so includes civil twilight which most other nocturnal surveyors don’t consider. The handful of birds he has were all just pre-dawn (e.g. here) and represent the initiation of diurnal movements. More generally within Trektellen there are just two other “nocturnal” Goldfinches, recorded in March 2016 in the Netherlands, though I can’t tell what time of night they were.

For comparison here is the phenology of nocturnal Ortolans, which matches very nicely the timing of Ortolan diurnal migration as expected. Again, much of the data come from the Low Countries where Ortolan calls are well understood.

Trektellen_OB_nocturnal
Figure 4. Timing of Ortolan Bunting autumn nocturnal migration, from Trektellen

 

Possible interpretations

There are two ways of looking at these results – that nocturnal Ortolans are all/mostly misidentified Goldfinches, or that the identification of nocturnal Ortolans is all/mostly correct.

To take the view that all (or even some proportion) of the claimed nocturnal Ortolans are actually Goldfinches raises some tricky questions. Why are these nocturnal “Goldfinches” moving about a month or so before diurnal movements. Perhaps they are recently fledged juveniles? But if that is the case, why are they not apparent over a much wider window as juvenile Goldfinches can be around from late June onwards. Why are nocturnal Goldfinches not detected at any other time of year despite being present year-round at these locations. Further, these nocturnal “Goldfinches” would have to be giving only an unusual call and none of their more familiar calls at night. Otherwise in 17,500 hours of recording, nocturnal recorders would be picking them up more regularly and we would see Goldfinches on Figure 3. This all seems very implausible.

Alternatively, if we accept the identifications as being correct, the clear conclusion is that Goldfinches migrate at night extremely rarely (if ever), or they do not call while doing so. Hence they represent an extremely unlikely confusion species for nocturnal Ortolans. That’s not to say that Goldfinches should be ignored as a potential confusion species. It just means they should be weighed appropriately when considering putative Ortolans.  Nocturnal Goldfinches may in fact be rarer than Ortolans!

 

Thanks

Thanks to Jon Heath for the Ortolan photo and to the contributors to Trektellen on whose data this post is based.

Standardised Nocturnal Flight Call Monitoring

Recording nocturnal flight calls is seeing a big surge in interest in Europe and all this effort scrutinising hours of audio has the potential to generate invaluable data on a large scale for understanding movement patterns of birds. Witness the amazing nocturnal movements of Common Scoters through inland England recorded in April-May 2018, the tantalising records of nocturnal Ortolan Buntings in southern England and continental Europe, and the amazing breadth and magnitude of nocturnal movements through Besh Barmag, Azerbaijan. There’s much to learn if the data can be collected and collated in a consistent manner.

protocol_coverBTO, Sound Approach and Sovon have teamed up to produce a Protocol for Standardised Nocturnal Flight Call Monitoring (available here), and in parallel, significant improvements have been made to Trektellen for the submission and sharing of nocmig data. The protocol aims to highlight simple ways that recordings and the data extracted from them can be standardised whilst still allowing flexibility for local circumstances. It details aspects such as where and when to record, times of night to cover and which species to log. For example, we recommend that call logging starts no earlier than civil dusk and continues no later than civil dawn to exclude diurnal migrants. Ideally counts should be submitted in hourly blocks to allow future analyses of timing and weather, and all birds flying over should be logged whether they are considered to be migrants or local birds. These suggestions will be familiar to many seasoned recorders and should act as a guide for those just starting out.

Trektellen has been collecting nocturnal flight call data for several years and with the latest modifications it is now possible to submit numbers of individuals and numbers of calls, which will be invaluable for species where estimating bird numbers is problematic. It is now easier to indicate which records relate to flying versus stationary birds and sound files uploaded to xeno-canto can now be embedded within Trektellen lists, which will be especially useful for sharing recordings of unusual or unverified records.

trektellen_dayview
Alongside the advances being made by the Sound Approach and others in species identification, we hope this protocol promotes further interest in recording and submission of data from nocturnal flight call monitoring.