Trail Cam Captures Redux: Beyond (My) Reasonable Doubt, Part 2

The previous post generated a number of responses, mostly positive, but there were a couple of negative remarks, one of which I thought I should address.

One claim was that this new way of examining the imagery (which was misleadingly described as “enhancement”) confuses the issue and makes the images harder to assess. In reality, the processes used in the previous post and in this one are for the most part not “enhancements.” While I have used various image enhancement tools in the past, most of these treatments use unaltered original images to create composites that make it possible to separate figure from ground and parts of the target from artifacts. It’s not a perfect method, but it’s an illuminating one, even with low resolution images. It requires no special forensic expertise. As should become clear from this post, it can also be useful for estimating size.

The images examined in this post were all captured December 7, 2014 on a Reconyx Trail cam that was showing its age. They were previously discussed here, here, and here.  As in the previous post, I’ve used Luminar to create the composites and apply various effects for this deeper dive. As always, views can be had by clicking on the images themselves; mousing over the images will reveal captions in most instances, and I have added versions with arrows to help orient you. I strongly encourage you to examine the images closely.

One initial comment about these images and the ones discussed in the last post: I’m personally convinced beyond a reasonable doubt that two of those images show Ivory-billed Woodpeckers. The images discussed in this post show fewer details, so they may not be quite as compelling. Either way, the captures under discussion are only part of the mosaic, and the subjective conviction that we have found Ivory-billed Woodpeckers in two locations is not solely dependent on them. The audio from last year, for example, is strong evidence in its own right. Go here for a detailed discussion of evidence gathered since 2009

I’ve reposted the original captures in tiled mosaic form below; I’ll follow that with some composites and comparisons among the images. I think the results are helpful if not quite as revelatory as the ones in the last round.  A very prominent American birder tweeted that one of the images below shows a Blue Jay (9.8-11.8″) and characterized the suggestion that it might be an ivorybill as “crazy”. (This is the capture I think is likeliest for ivorybill in this group.) I think his knee-jerk suggestion is absurd, as this approach to reviewing the images should make more apparent.

 

I’ll begin with the raw image I’ve always liked most for ivorybill, as previously discussed in detail and in comparison with Imperial Woodpecker captures. (Note that the version comparing putative IBWO and IMWO has been enhanced.)

IMG_4195

Suspected Ivory-billed Woodpecker

rotatedcomposite

Rotated, resized image of Imperial Woodpecker in flight. The bird is angled downward and slightly away from the camera obscuring the bill and foreshortening the profile. An automated motion blur reduction feature has been applied to the image of the putative ivorybill.

The next image shows the difference between the foregoing frame and one captured approximately 40 minutes later. The bird to the left of the snag is clearly closer to the camera (which was approximately 85′ from the target snag), between it and the snag. The composite strengthens the impression that the suspected ivorybill is behind the tree and the foliage behind it, since at least some leaves appear to be between bird and camera. Frank estimated the foliage as being 15′ more distant.

 

In the earlier discussion, several Pileated Woodpecker captures, including the one below, were used for comparison. These captures were taken from a slightly different position, but that’s irrelevant to the issue of size relative to the reference object, the jug handle-shaped stub.

IMG_2735-1

Pileated Woodpecker on Snag

It may be even more instructive to contrast the capture with the Red-headed Woodpecker, apparently caught as it was preparing to land on the snag. The comparison suggests that the two birds are approximately the same distance from the camera. The possible ivorybill’s body is considerably longer than the Red-headed Woodpecker’s wingspan (14-17″). Thus, between the Pileated Woodpecker (~16-19″, with southern birds typically on the smaller side) and the Red-headed Woodpecker, it can be inferred that the suspected ivorybill is approximately the right size (19-21″), considerably larger than a Red-headed and somewhat larger than a Pileated. I am unable to think of an alternative ID for a bird of similar size and shape, even leaving aside the fact that a white trailing edge seems to be present even in the original, unmodified version of the capture. (The trailing edge becomes more apparent when the image is processed to reduce motion blur.)

 

A different but related method of image processing is helpful for understanding the position of the bird relative to the snag and also suggests that another capture in the series is at least size-appropriate for ivorybill; the tuck and the long, pointed shape of the wings are also intriguing. The first composite shows the unknown bird and the Red-headed Woodpecker. (Note what seems to be a wing, species unknowable, protruding from behind the snag; it appears in the Red-headed frame and no others and had gone unnoticed until now. On looking again, this could also be a squirrel’s tail, albeit unusually positioned.) The second is an overlay of the suspected ivorybill capture and the unknown bird processed using Luminar’s “Color Burn” feature; the two birds appear to be approximately the same size and in the same plane. I’ve added a shot of a Pileated Woodpecker in flight, with wings in a similar position, for comparison.

 

 

 

 

 

imgp3505

Another capture from the morning of December 7 has always interested us. The underwing pattern and shape are suggestive of ivorybill and reminiscent of one of the Singer Tract photographs. But questions about size have lingered. Comparing this image with two others from the same deployment that show Red-headed Woodpeckers may be helpful, and it has pushed me toward favoring Ivory-billed Woodpecker for this capture as well.

IMG_4399

Either Ivory-billed or Red-headed Woodpecker Depending on Size

 

 

 

 

This capture seems more ambiguous to me, but a few factors are more consistent with Ivory-billed Woodpecker than Red-headed Woodpecker. Both Red-headed Woodpeckers, one captured on the same morning under similar lighting conditions, appear to be closer to the camera than the possible ivorybill. My efforts at measuring wingspan were crude, but the suspected ivorybill seems to have a slightly greater wingspan, despite the upturned angle of the wings and apparent greater distance from the camera. This would tend to exclude Red-headed.

The comparison is also useful insofar as it shows that the wing and body shape of the possible ivorybill are not similar to the Red-headed Woodpeckers captured during this deployment. While Red-headed wing and body shapes can resemble ivorybill at certain angles and under certain lighting conditions, as Bill Pulliam documented years ago, the tail in this image looks too long relative to the body, and the wings look very narrow, at an angle that I don’t think would create an illusion of elongation.

As always, I’m open to correction, and my interpretations are provisional. Intelligent and informed comments are welcome.

Advertisements

March Recordings Revisited: A Compilation of the Calls for Easier Listening, Interesting Knocks, and Some Additional Analysis

If you’re interested in possible double knocks, I’ve made what may be some important new finds, so be sure to read the whole post and listen to the clips at the end.

In the meantime, I think this post will be of interest to many readers – from the new ones who’ve found the blog either because of the recent sighting or after reading about Project Coyote on the LABird list (thanks to Jay Huner for the mention) to other ivorybill searchers and aficionados to those who have had trouble hearing the putative ivorybill calls on the March recordings or didn’t want to wade through all the audio.

In the easy listening department, Steve Pagans has made a compilation of the clearest calls on Matt Courtman’s first, 2 hour clip recorded on the morning of March 15 using NCH software.

I tweaked Steve’s version a bit, amplifying it and applying noise reduction using Audacity, an equivalent program. Sonograms were generated using Sonic Visualizer, to my knowledge the best free program of its kind.

Both Steve’s version and mine (immediately below Steve’s) should make it considerably easier to hear many of the calls recorded that morning. Steve’s is somewhat cleaner, and mine is somewhat louder.

Steve’s extracts from Matt’s first clip:

With additional amplification:

Steve has done similar, shorter condensations of the the other two recordings Matt made. (The extract from the second clip adds little, so I’ve opted not to post it.)

Steve’s extract from Matt’s third clip:

With additional amplification:

The calls have a very consistent sonogram pattern (the stacks of three or more parallel horizontal lines), with an emphasis on the second partial (third horizontal line from the bottom). There are apparent tree squeaks in both clips that have a similar quality; the dominant frequency is similar, but the tree squeaks show more energy at that frequency. The second screen cap is a detail of the first, showing both calls and tree squeaks. The latter show a brighter orange, indicating more energy at a similar frequency to the calls’ dominant partial. The sounds are definitely different, but they can be hard to distinguish at the margins.

Screen Shot 2017-12-11 at 8.34.36 AM

Screen Shot 2017-12-11 at 8.38.56 AM

Steve’s effort, which took many hours, inspired me to revisit the recordings and specifically to reexamine possible responses to Matt’s anthropogenic double knocks and Phil’s playbacks over the course of the more than three hours when Matt’s recorder was running. I had concluded that the ADKs did not seem to have had an impact, but on careful re-listening, I’ve amended that view. Matt’s knocks do seem to have stirred up calls in reaction and to have generated several possible single or double knocks.

Listening through the recordings and especially to the ADKs and their immediate aftermath was a time-consuming and difficult process, especially because the knocks are very loud and tightly spaced. Matt used two wooden blocks (rather than dowels and a tree trunk or a knock box) and did not follow a specific protocol. Overall, his approach was more aggressive than Frank’s or mine, both in terms of volume and number of knocks, and a somewhat more restrained approach is probably advisable in the future. Nevertheless, if you assume (as I do) that ivorybills were present on this morning, these recordings may provide some insights into the efficacy (or lack thereof) of ADKs in generating responses.

This return visit to the recordings also supported my view that the putative ivorybill calls (recorded in the old search area) that Phil played back generated no reaction at all from the suspected ivorybills on the morning of the 15th.  Blue Jays, however, appeared to respond, and I now tend to think those calls (though not all of those recorded in the old area) were Blue Jay, based on the apparent Blue Jay responses and on the harmonics. In retrospect, it’s unfortunate (but totally understandable under the circumstances) that playback of the Singer Tract recordings and other possible attraction methods weren’t tried. Perhaps another opportunity will present itself, although the events of March 15 were singular . . . thus far.

In the first of these two clips, the playback seems to have provoked Blue Jays to call faintly. In the second, in which the Blue Jay calls are easier to hear, they had been calling before the playback began.

Phil’s Playbacks:

Screen Shot 2017-12-09 at 1.13.49 PM

Sonogram of played back calls showing fundamental frequency, strong second partial, and strong higher harmonic at approximately 5000 hz. Presumed Blue Jay call is the faint horizontal line at bottom center with a frequency of around 2000 hz.

By my count, Matt did 7 sets of ADKS and performed approximately 205 knocks in all. I noted six possible double or single knock responses (of varying quality) in four of the six series. These knocks occurred within seconds of ADKs. The temporal proximity between the ADKs and the possible DKs and SK in response make it less likely that Matt and Phil (to a lesser extent) would have noted them in the field.

I’m posting the relevant extracts below in unmodified form followed by clips with the interesting knocks amplified using Audacity. They should be audible through a desktop computer, but headphones will help. To repeat a strong caveat: Matt’s ADKs are very loud (which made this analysis especially difficult). I should also reiterate that I don’t consider myself particularly skilled at analyzing recorded knocks; I don’t have the greatest ear for intervals and have no direct field experience with Campephilus woodpeckers; my ability to interpret sonograms is also limited. Nevertheless, I’m sure these sounds are neither shots nor industrial noise (or duck wingbeats). With one exception (the fifth knock, which sounds like a single to me), I’m also confident that Matt was not the source of the sounds. Some of inadvertent bumping together of his blocks should be audible at various points, and it has a different quality.

The first set of ADKs was the shortest, involving only six, and these were very closely spaced. The first ADK in the series produced a possible DK in response. Matt began the series 15 minutes and 52 seconds into the recording, and there had been a kent call at 14:59. Both a possible DK and a kent call can be heard between the first and second knocks. An additional kent can be heard after the third knock (omitted here).

Unmodified version:

Amplified knock:

I did not find any possible double knocks in the second and third series, and there were only a few kent calls – one during the second and three during the third. There was a possible double knock in the fourth series, at 1:48.22, two seconds after an ADK.

Unmodified:

Amplified:

During the 5th set, on the second clip, I found a possible single knock at 3:16. (Steve Pagans, who has an excellent ear, thinks it’s a closely spaced double.) I can’t rule Matt out completely as the source of this sound, but I think that’s a remote possibility.

Edited to Add: Playing the clip at a slower speed, reveals that there are two distinct knocks, the first louder, and leaves me convinced that Matt was not the source of this sound.

Unmodified:

Amplified:

Also on the second clip in the 6th set, there’s a possible double knock at 30:26, approximately 2 seconds after an ADK.

Unmodified:

Amplified:

On the final recording, during the 7th and final set, there’s possible DK after the 5th ADK in the series. It was preceded by a possible ambient DK about 50 seconds before Matt began the ADKs. The ambient knock is slightly buried behind some rustling, approximately three seconds into the recordings. To my ears, the DK in response sounds as though the first knock is softer than the second, something that’s uncommon but not unheard of for Campephilus woodpeckers. I’ve included both the entire relevant sequence and extracts in which I’ve amplified the knocks.

Unmodified clip with ambient DK and later DK in response to Matt’s 5th in a series of ADKs:

Ambient DK Amplified:

Amplified DK in response to ADK:

Finally, in listening to parts of Matt’s first clip again, I noted that there are a number of very distant kents. These are barely audible at normal volume and only faintly so on the amplified version. This suggests a highly mobile source (or sources) for the calls. It also suggests that there may be more calls on the recordings than the approximately 200 hundred I originally estimated. Steve noticed these calls too but elected not to include them in his compilation because the amount of amplification necessary degrades the sound quality. Thus, it may be difficult for some readers to hear these more distant calls.

The first few seconds of the extract below are unamplified to give a sense of the volume of other ambient sounds at the time. The kent calls come toward the end, and with one partial harmonic showing up faintly on the sonogram (the lighter colored dot near the left margin in the image below).

Amplified Distant Kents:

Screen Shot 2017-12-12 at 9.13.08 AM

 

 

Click the hotlinks for prior posts and pages related to the March recordings.

I owe readers the final installment of the “Bits and Pieces” series (hotlink is to the most recent installment). I anticipate that it will be my final post for this eventful year. Look for it just before Christmas.

 

 


More Recordings from March 15: Approximately Three Hours of Kent-like Calls and Analysis

On March 15, Phil Vanbergen and Matt Courtman recorded numerous kent-like calls at the same location where we heard several calls on March 11 and 12. Phil was able to record two of the March 11 calls. That capture is included in the post, along with Phil’s audio from the morning of the 15th. I heard two of the calls on the 11th; the second one in particular struck me as being consistent with the Singer Tract recordings; the first seemed a little low pitched to my ears, an observation that’s captured on the recording. Steve Pagans and I heard several calls on the 12th, but these were not recorded.

Matt obtained nearly three hours of audio, and to my ears the sounds are coming from 2-4 distinct sources; I had the same impression after listening to some of Phil’s clips. I have now listened all the way through Matt’s recordings several times and will share my analysis below. Matt and Phil are likely to weigh in later with their perspectives. I also have a couple of trip reports pending, so there should be a lot of activity on the blog in coming weeks.

To start with, I counted over 200 kent-like calls in all.

On the long clips posted here, I’ve edited out all of Matt’s ADK (anthropogenic double knock) series, which he did on the half hour. The knocks are very loud, as can be heard from the one trailing sound I’ve left in. I also snipped out several minutes of conversation between Phil and Matt. The ADKs seem to have led to more frequent calling and may have provoked some double knocks, something we may address in a future post.

Edited to add: On further review, there does not appear be a correlation between ADKs and more frequent calling. Clips like the one posted below can be deceptive. One kent-like call that overlaps with a knock has been deleted. There is also one possible knock in response. Caution, ADKs have not been completely spliced out, and they are loud. See bottom of page for brief clip and sonogram.*

Between 6:12 and 6:25 on the last long recording made that morning, there are five calls of differing durations and volumes, followed by what may be a double knock. Similarly, at the end of the full clip, starting at 14:14, 3 calls are bracketed by some potentially interesting knocks, 2 before and 1 or 2 after.

The first four clips below are shorter, amplified extracts on which the calls can be heard easily.

The first two of these are extracted from the final segment described above and include the interesting, tooting sounds and possible knocks.

The third clip includes multiple calls over 2 minutes and nine seconds, along with a wide variety of other sounds.

The fourth is four minutes long (pardon the airplane noise) and should provide additional context while also revealing some of the variations among the calls.

For those, like me, who don’t have professional sonogram software, I recommend using Sonic Visualizer – an easy to use, free program that enables you to watch the sonogram as you listen.

And for those who are unfamiliar with avian bioacoustics, this is a great place to start. I’m on a very steep learning curve myself and am prepared to stand corrected about any misstatements in this post.

Many of the sounds are audible on built-in computer speakers, but playback through headphones, earbuds, or external speakers is highly recommended.

 

I think these calls were likely made by Ivory-billed Woodpeckers. My perspective is based in part on the fact that I have spent all or part of nearly 40 days in close proximity to where the calls were recorded, starting in 2014. A considerable amount of this time was spent sitting quietly, and the total person hours spent in the area is well into the hundreds. We have had occasional kent-like calls, possible double knocks, and possible sightings over the years, but nothing approaching what transpired on the 15th.

Matt spent the morning of the 16th in the same location and did not hear any of the sounds, and Patricia and I spent 4 mornings and one afternoon there between the 23rd and 31st and heard no similar calls. I think this militates strongly against the idea that the source of the sounds is a common resident of the area.

Three alternative hypotheses have been suggested.

On the morning of the 15th, Phil proposed that the sounds might be tree squeaks. There appear to be multiple tree squeaks on the recording, some with similar pitches, but they bear little resemblance to the kent-like calls on the sonogram or to the ear. In addition, the calls sometimes come singly, sometimes in groups, and they vary in pitch, volume, and duration and seem to occur independent of wind velocity (on the 11th we noted that there was no wind.)

Screen Shot 2017-04-04 at 9.29.13 AM

Sonogram of first clip, showing apparent tree squeaks before and and immediately after two kent-like calls – the stacked horizontal lines right of center.

The first sound, at just after 4 seconds on the second clip, seems ambiguous. The sonogram is somewhat similar to the kent-like sounds, but the duration is very short, and it has a creaky quality. It’s also associated with the two creaky sounds that follow. These can be heard frequently over the course of the morning, and their appearance on the sonogram is nearly uniform.

Screen Shot 2017-04-07 at 7.51.16 AM

Sonogram of second clip showing tree squeaks.

One reviewer proposed Wild Turkey and Blue Jay as possible sources. I think turkey can be ruled out due to the absence of other turkey-like sounds associated with these very persistent calls.

Blue Jay strikes me as a more plausible alternative. Blue Jays can be heard at numerous times on Matt’s recordings. And Blue Jays are known to make kent-like calls, some of them very similar to known Ivory-billed Woodpecker sounds. This is likely not mimicry, since the most similar recorded calls I know of were obtained in upstate New York. The recordist noted the similarity. On the sonogram, the resemblance between these calls and either known Ivory-billed Woodpecker sounds or Matt’s recording is not as strong as it might seem to the ear. While they share a strong third partial, the Blue Jay fundamental is higher and some of the higher partials are considerably stronger.

Screen Shot 2017-04-02 at 7.03.13 AM

Kent-like, suspected Ivory-billed Woodpecker calls recorded by Matt Courtman on March 15, 2017. Note the relatively strong base frequency and 3rd partial, and weak 2nd, 4th, 5th, and 6ths.

Screen Shot 2017-04-04 at 11.05.39 AM

Kent-like Blue Jay calls. Note the relatively weak fundamental and strong second through fifth partials.

It’s also important to note that on the Cornell Blue Jay recording, the kent-like calls are intermixed with typical Blue Jay vocalizations. Over the three hours of Matt’s recording, kent-like calls occur during periods when Blue Jays are vocalizing and during periods when Blue Jays are silent.

Some of the kent-like calls have harmonics consistent with Cornell’s recorded playbacks of Singer Tract calls at 145 meters. Many are lower in pitch. Most have a considerably longer duration, although to my untrained eye/ear, some seem close to the 80-100 ms duration on the Cornell recording. More on duration below.

On March 28, I did some playback of Singer Tract calls using an iPhone and a bluetooth speaker. Patricia recorded them on a Roland Edirol R09HR digital recorder. I’m including the recording and the sonogram for comparison. Like Matt’s recordings, the second partial is strong throughout, and the 1st, 3rd, and 4th appear to be weaker than for the Blue Jay shown above.

Screen Shot 2017-04-06 at 12.30.00 PM

At various points, Phil also experimented with doing playback, using calls recorded in 2010 at the old Project Coyote site. (He also played back several other species – Red-bellied, Golden-fronted, and, possibly, Gila Woodpecker to gauge Red-bellied response, and Eastern Towhee out of personal interest.) Examples of putative ivorybill playbacks can be heard over the first 3 minutes of the fourth long clip posted above and also during the first part of the fifth. These sounds are longer but similar in tonal quality to the lower pitched calls. Their harmonic structure is different, however, with a fundamental at around 800 hz, a second partial at around 1600, and a fairly strong higher partial at approximately 5000 hz., and should be readily identifiable on the sonogram.

Screen Shot 2017-04-04 at 2.15.09 PM

Higher pitched playback calls, at middle left and far right.

Phil’s playbacks do not seem to have provoked any kent-like replies. Blue Jays can be heard during the same time frame, but it’s not immediately apparent whether the recorded Blue Jay calls are responses or merely contemporaneous vocalizations. This segment includes some Blue Jay vocalizations.

With regard to Blue Jays and the numerous kent like calls heard from March 11, 12, and 15: to reiterate, many hours have been spent in this vicinity, with close attention being paid to kent-like sounds. These are heard infrequently and never before in such quantity or over such an extended period. If Blue Jays in the area were making these sounds, we almost certainly would have heard and recorded many of them over the years. In addition, both Steve Pagans and Matt Courtman are experienced and skilled ear birders and neither thinks these are Blue Jay calls.

It’s also worth pointing out that another potential confusion call can be heard on the recordings – White-breasted Nuthatch. The sound is similar but much weaker, as should be clear from the amplified March 15 excerpt. On the sonogram, just below the amplified recording, the calls show up very faintly, with dominant frequencies of around 2200-2400 hz and a relatively strong partial at 1700-1800.

 

 

I encourage people to listen through and draw their own conclusions. Input from those with expertise is welcome.

While I can’t say if any of these calls are a perfect match for the Allen and Kellogg recordings (some may be), many of them are close on the sonogram and similar to the ear. It’s important to bear in mind that the Singer Tract birds were likely agitated when those recordings were made, even though Tanner described some them as being good examples of kents. It’s also important to read Tanner’s descriptions carefully, though as is so often the case, his writing can be opaque. Perhaps his most important observation was that “all of the notes have the same nasal, trumpet-like quality.”

According to Tanner, “The notes of the nuthatches are the only bird calls I know that sound like the voice of an ivorybill; the Ivory-bill’s calls are much longer and pitched higher than the calls of a White-breasted Nuthatch, are more in the range of a Red-breasted Nuthatch.” (Emphasis added.) By contrast, Hasbrouck, writing in the 1890’s, described it as being “exactly like the note of the White-breasted Nuthatch” only much louder and stronger.

Tanner’s reference to Red-breasted Nuthatches has always confused me. I’m very much in Hasbrouck’s camp; I think the Singer Tract kents sound far more akin to White-breasted than Red-breasted Nuthatch. Either way, most of the Allen and Kellogg kents are lower pitched than typical White-breasted Nuthatch calls, as are the ones on Matt’s recording. In addition, the Allen and Kellogg kents seem to be of similar in duration to typical nuthatch calls, rather than longer or “much longer”. This too suggests that they are not typical but are more rapid and perhaps higher pitched due to agitation. Tanner wrote further, “[t]he kent note, given in monotone or infrequently, is the ordinary call note. When the bird is disturbed, the pitch of the kent rises, and it is repeated more rapidly, frequently doubled, kent-kent, with the second note lower. The prolonged and slurring, kient-kient-kient call I always heard when two or more birds were together.” This call was never recorded.

According to Allen and Kellogg, “kenting varied a great deal” and a male bird called “loudly and deliberately”, again suggesting that many calls were of longer duration than those on the recording. Tanner’s notes also point to this variability. At one point, he wrote of “1 and 2 syllable yaps”; he has the Mack’s Bayou bird (whose voice he claimed he could recognize) making a “kient-kient” and also transliterated calls with “keent keent” and “yeenh yeenh” (Bales). These renderings all suggest a more drawn out call than those on the Allen and Kellogg recordings. George Miksch Sutton described the Singer Tract birds’ calls as, “strange, bleat like, not quite sharp enough for a woodpecker’s cry. It was slightly nasal in quality and it sounded to me like ‘Gip!’, with a hard g“. Sutton’s description also suggests that many kents had a fairly long duration.

Edited to add: similarly, several observers (Audubon, Beyer, Hoyt)  described ivorybill calls as “plaintive”; this too seems to imply calls of longer duration than what’s heard on the Singer Tract recordings.

Given the resemblance to the Singer Tract recordings and the lack of plausible alternatives, I posit that these calls are at worst highly suggestive of Ivory-billed Woodpecker.

*Possible knock and kent-like calls temporally associated with ADK series. Caution remnant ADKs are loud.

Screen Shot 2017-04-09 at 7.37.53 AM