1967 slides taken by Neal Wright of a putative Ivory-billed Woodpecker in Texas are viewable on Vireo (search Ivory-billed Woodpecker), but high resolution scans have not been widely circulated as far as I know. These images were not made public until after the the Arkansas “rediscovery”, more than three decades after they were obtained. Wright’s story is mentioned in Jackson (2004) “Reynard saw the photo and said that it was fuzzy but definitely of a Campephilus woodpecker.” It’s clear from the context that Jackson had not seen the images at the time of writing.
When I first encountered the Wright slides, I was skeptical, but after seeing some lesser-known Singer Tract photographs as well as other images of Campephilus woodpeckers in cavities, my opinion started to shift. After finding additional ivorybill photographs in the Cornell archives and in Tanner’s dissertation, I thought it would be worth posting some of those images along with one of Wright’s slides for the sake of comparison.
Of course, it’s up to readers to draw their own conclusions, but I think a few things are worthy of note. First, the Wright slides were taken long before the internet, at a time when the only readily available image of an ivorybill in a nest cavity was Tanner’s Plate 1, which is quite similar to Fig. 43b (below). The posture of Wright’s bird is much closer to the ones shown in the then virtually unknown and/or unpublished images, especially those from the 1938 nest. The placement of the cavity is also strikingly similar, just below a major fork. It seems highly unlikely that Wright would have been aware of obscure Singer Tract photographs.
While the image quality is too poor to be certain, there appears to be excavation similar to work found on some Singer Tract nest and roost trees to the right of the nest cavity in Wright’s slide. Again, this is a fine detail that would likely have been unknown to Wright and that would have been difficult to fabricate.
These are very poor quality images; the malar stripe seems a little too extensive, although this could easily be a function of angle and lighting. As with the Fielding Lewis photographs, which were taken several years later, I have to wonder why anyone intent on committing a hoax wouldn’t do a better job. And in the case of the Wright pictures, it would make more sense if the template for such a hoax would have been Plate 1 in Tanner, rather than photos that were unknown to all but a handful of people, most of them at a northeastern university.
Finally, I think the fact that the images were turned over to an ornithologist (George Reynard, scroll down for his obituary) but were kept confidential for so long also tends to support the idea that they’re authentic. Neal Wright may have had an agenda – a desire to protect the area where he took the picture – but the images were not used to serve that purpose.
Edited to add: This fascinating article on a recent, non-ivorybill related hoax suggests that it’s not uncommon for hoaxes to be paradoxically uneven in quality, and that hoaxers’ motives can be murky and bizarre. Nonetheless, I think that other factors point to authenticity for both the Wright and Lewis photos.
Another item I found in Tanner’s dissertation merits comparison with one of Project Coyote’s camera trap photos, since the tree species involved are the same. Plate 7 in Tanner shows ivorybill feeding sign on honey locusts, but the reproduction in the monograph is very dark. The figure from the dissertation is much brighter, making it clearer what Tanner was attempting to show. I think the similarity to the work on our target tree, where I had a sighting a week prior to the capture, is striking.
To enlarge the trail cam photo, click here.
To expand on some of the data included toward the end of the March trip report (which is worth reading in in conjunction with this post), I thought it would be informative to provide a season by season and sector by sector breakdown of the scaling I and others involved with Project Coyote have found since the spring of 2012. To do so, I’ve gone through my notes and photographs and have done my best to reconstruct the data collected. While not complete (I’m quite sure a good deal more scaling was found in Sector 3 during 2013-2014, for example), I think this breakdown is a fairly accurate reflection of what we’ve found over the years.
As discussed in previous posts, I think extensive scaling on hickory boles is the most compelling for Ivory-billed Woodpecker. Bark on this species is thick, dense, and usually remains very tight for a long time. Extensive scaling on sweet gum boles and oaks (upper boles and large branches) is second among work that I’ve found. Work on small boles, and higher and smaller branches is somewhat less compelling and is more significant for its abundance. Some of the high branch scaling and work on smaller boled sweet gums may well have been done by Pileated Woodpeckers (and possibly by Hairy Woodpeckers), but the abundance, the presence of large bark chips in many cases, the way it appears in clusters, and the fact that Pileateds scale infrequently suggest a different source for much of it.
I have excluded all work where squirrels are suspected but have counted one tree, a hickory found this year, on which the work could well have been that of a Hairy Woodpecker. Hairies do forage for Cerambycid beetles just under the bark, but they’re only capable of removing tight bark in small pieces; their work on hickories is perhaps more accurately described as excavation through the bark.
The trail cam images toward the end of this post are the best we have (out of many thousands of hours of coverage) showing how these species forage on suspected ivorybill feeding trees.
All trees were live or recently dead (twigs and sometimes leaves attached). All scaling was on live or recently dead wood.
Sweet Gum (Liquidambar styracifula)
Sector 1: 46
Sector 2: 8
Sector 3: 51
~15% had scaling on boles (a few of these were large trees). The majority of work was on crowns, including larger branches. Fallen trees were included when woodpecker involvement was evident and bark was tight.
Bitternut Hickory (Carya cordiformis)
Sector 1: 3
Sector 2: 4
Sector 3: 7
All trees were standing; scaling was on boles and was very extensive (the tree shown on the homepage is one example) with one exception from this year . Insect tunnels were visible in all examples. An additional hickory with a modest amount of high branch scaling was found in Sector 1 this year but was not counted for this analysis.
Oak (Quercus) spp.
Sector 1: 1
Sector 2: 4
Sector 3: 0
All oaks had scaling on large branches; one also had some on the bole. All oaks in Sector 2 were found in a single cluster.
We have some information on forest composition in Sector 3, and it appears that sweet gums make up approximately 19%, oaks upwards of 35%, and hickories somewhere under 10%. Sectors 1 and 2 may differ and be more varied in overall composition.
The overwhelming preference for sweet gums relative to their abundance stands out. The scaled oaks are a mix of species, one Nuttall’s, one willow, the others unidentified.
In Sector 3, I am treating the compact stretch from the location of Frank Wiley’s sighting last spring/downed sweet gum top where we had the camera trap to just south of our current deployment as a cluster. The estimate of 23 trees being found in this area is conservative. I have only found one instance of recent scaling north of the location of the downed limb/Frank’s 2015 sighting. The main cluster has been in the same vicinity this year and last, with additional work scattered around farther south. Two of the hickories are within 30 yards of each other, approximately half a mile from the cluster, and one was on the edge of the concentration.
It also may be significant to note that we found a cluster of old but intriguing cavities in the same vicinity as the Sector 3 concentration in 2013-2014. Most of these seem to have fallen. The difficulty we’re having finding active, suggestive cavities is vexing, and may be the most compelling reason to be skeptical about the presence of ivorybills in the area. At the same time, finding Pileated cavities is difficult, even in defended home ranges.
I’m treating Sector 1 as a single concentration; the vast majority of the work is on a natural levee where sweet gums are abundant. The entire area is considerably larger than the other clusters, but given the abundance and ease with which we’ve found sign there over the last five seasons, I think it constitutes one area of concentration.
In Sector 2, there was a small cluster in the area where I recorded putative kent calls in 2013, with work found in 2012 (spring and fall) and 2013. Because the area is small with open sight lines, I can be confident there has been no recent work there since late in 2013 (I last passed through it with Tom Foti back in March of this year.)
The sweet gum work Tom and I found on that day was perhaps half a mile north of this cluster, within 100 yards of the hickory on the homepage. The other hickories found in the 2013 and 2014 seasons were not far away, no more than 500 yards apart as the crow flies.
There’s obviously some bias here, since there’s a relationship between finding feeding sign in a given area and spending time there. Nevertheless, I have little doubt that the putative ivorybill work tends to be clustered. I also have little doubt about the strong preference for sweet gums, since I’m not looking at tree species when I look for scaling. The degree to which sweet gums are favored has only become clear over the last year or so.
Frank pointed out this data does not reflect most of the scaling that likely exists in relatively close proximity to the Sector 3 cluster but cannot be quantified because it is in an area we have intermittently visited due to inaccessibility. Only two or three examples are from this area, which has been visited a handful of times.
Frank wrote this up and asked me to post it.
Recently, we’ve received several messages, and another blogger has mentioned the use of game cameras – also referred to as trail cams, or camera traps – as aids in searching for Ivory-billed Woodpeckers. We’ve been deploying trail cams since early 2009, and have had some intriguing “hits”. They have been useful in many ways, and we have learned a lot about the feeding habits of other woodpeckers – specifically, Pileated, Hairy, and Red-bellied – that were useful in giving Mark insights into the type of feeding sign that we have come to believe is most likely diagnostic for ivorybills. In dozens of deployments, and hundreds of hours of searching with boots on the ground, and sometimes in the water, we have NEVER observed a woodpecker performing the type of scaling that we have come to suspect is diagnostic.
While trail cams are a valuable tool in our arsenal, they are, by no means infallible. There are many pitfalls involved in their use. Firstly, they are not designed to capture birds – in fact many common species of birds are almost unidentifiable in the images. Birds do not set off the motion triggers that these cameras use, and why should they? The cameras are designed for use along trails and adjacent to food plots used by larger mammals – usually whitetail deer. I have tested the Reconyx “Hyperfire” cameras that we use, and even a large Wild Turkey at fairly close range will not trigger one, even at its most sensitive setting, while a relatively small (terrier size) dog will trigger one from a distance of twenty-five or more yards. The manufacturer has indicated to me that there is “something about the way birds reflect light in the visible and infrared spectrum” that makes the cameras’ triggering units unable to “see” them.
As a result, ivorybill hunters must find a camera that will operate in time lapse mode. For all their disadvantages (3.1 megapixels, low resolution among them) the Reconyx cameras offer the best time lapse mode available at an affordable price. This, in and of itself, though, becomes a handicap, as the card for each camera must be programmed, using a proprietary program provided by Reconyx, on a PC. One programs the card, inserts it into the camera, and hopes for the best – there is no way to check and see if the card/camera combination is functioning properly. This has led to many wasted deployments. Additionally, there is the problem of how often should the camera take a photo vs. storage capacity of the card. The cams are designed for 8 gig cards, sometimes they will function with a 16 gig card, but will universally malfunction with a 32 gig card. 8 gigs is usually enough for a ten-day deployment with the camera time constrained to take a photo every twenty seconds for ten hours a day – you do get to select the hours of operation though. While every twenty seconds would seem to be quite often (and a full 8 gig card will store upwards of thirty thousand images), perform a little test for yourself. Go out to your favorite birdwatching location and see how long a bird – any arboreal non-raptor species – stays in one location for twenty or more seconds. Captures of birds on game cams are a relatively rare event – I have looked at nearly a million Reconyx photographs and have picked up birds of any kind in perhaps a thousand images – identifiable birds in maybe two hundred.
The very first thing that has to be done when using game cams is to select a location where you suspect an ivorybill is likely to show up. This could be a tree with scaled bark and other indications that an ivorybill has visited it, or a cavity with features that seem to match photographs of known ivorybill cavities. Both of these are, at best, iffy propositions. One has to find the tree, geotag it (you do have a good GPS unit, right?) then return to the location – often several miles of hiking through some pretty rough and secluded forest, carrying the camera and its mounting system – which weighs around ten pounds. Then the camera and mount have to be positioned to get the target near the center of the frame, which gets easier with practice, and the camera and mount hidden and intervening vegetation trimmed so as not to interfere with the line-of-sight. Once all this has been done the camera can be turned on, armed, and left to do its thing.
Assuming that everything up to this point has been done perfectly, in ten days or so, it’s time to change cards, or retrieve the cam. Now one is faced with the daunting prospect of going through some thirty-thousand photos looking for anything “interesting”. Often, several days of images will pass without a single “hit” of any kind. It’s often a relief to spot even a small woodpecker or squirrel, to remind one that the target is part of a living ecosystem.
The series shown here is exceptional in terms of quantity of images, quality, and our ability to place the camera. (It may be significant for what it doesn’t show, a Pileated doing the type of scaling we think is diagnostic.) Even so, it was not possible to cover the entire target tree.
As I stated earlier, these cameras, within their limitations, are useful tools, but for my money, nothing really beats the good old MK I MOD I human eyeball. But at this point, that’s just not enough …
In addition to the suggestive photos we’ve already posted, we recently obtained some pictures at the site where last week’s double knock was recorded. Because we find some of these pictures intriguing but inconclusive, we have deployed two cameras in hopes that the double placement will yield an identifiable photograph. The first picture, taken under good lighting conditions, clearly shows a Red-headed Woodpecker (there is a roost at the very top of the snag); the others are ambiguous. We are posting them unedited and leave it to you to speculate about what they may be. We recognize that none of these are of anywhere near good enough quality to be identifiable as ivorybills, but we are doing some further analysis to get a clearer sense of scale. The camera placement is 85’ from the tree; the branches behind the tree are an additional 15” away. My preliminary estimate of the diameter of the tree just above the jug handle on the right is approximately 18”.
I have taken careful measurements using a camera with known lens settings and a rangefinder – when the weather is more congenial, I will make comparison shots at the exact measured ranges. This should give a margin of error of ~1″ or less.