Insights, Ants, and Old Growth: a Nuanced View of the Ivorybill’s Decline and Possible Survival

I’ve just finished reading Tanner’s dissertation and have gained some new insights into topics that have been discussed in a number of earlier posts.

Conventional wisdom, following Tanner, holds that the Ivory-billed Woodpecker’s decline and possible extinction were caused by habitat loss, specifically the logging of old growth forests during the 19th and early 20th centuries. Birdlife International’s fact sheet on the species suggests “that large contiguous tracts of mature woodland would be required to support a viable population”, referencing Jackson 2002. Snyder et al. have proposed an alternative hypothesis that “human depredation was the primary factor.”  (p.9).

Tanner’s model depends on the idea that food supply was the limiting factor on ivorybill populations, because the species is highly specialized, and that old growth conditions were optimal or essential. While Tanner was aware that ivorybills bred successfully in an area that was predominantly second growth, at Mack’s Bayou, he glossed over this fact in the monograph, and became more dogmatic about old growth as a requirement in later years.

Snyder and some others have contended that the ivorybill is a generalist. According to Snyder, “the data available on diet and foraging methods simply do not provide compelling evidence for strong feeding specialization.” Snyder goes on to suggest that “[i]ts apparent skill in exploiting recently dead timber, coupled with its ability to feed in a variety of other ways, may even have given it some significant foraging advantages over the pileated woodpecker, a species apparently much less capable of bark stripping. Indeed, the pileated woodpecker, like other Dryocopus woodpeckers, may well be more of a food specialist than any of the Campephilus woodpeckers.” (p. 37).

As I see it, there are elements of truth in both models, but neither is complete. In addition, I think that each model relies on at least one flawed premise.

The old growth/virgin forest component of Tanner’s model fails to account for the facts that the Singer Tract population was dwindling even before logging began in earnest and that birds appear to have remained in the Tract until well after it had been extensively logged. Tanner suggested another possibility, “perhaps the greatest factor reducing the rate of ivorybill reproduction is the failure of some birds to nest. One reason for their not breeding is immaturity, for it is probable that ivorybills do not nest until they are two years old. Another possibility is that the quantity of food available to the woodpeckers may determine whether they will nest or not.” (p. 83).

Tanner struggled to account for the fact that the ivorybill population at Singer was dwindling by the mid-1930s, even though overall habitat quality had, if anything, improved relative to what it had been a few decades earlier. He attributed the higher relative abundance in previous years to tree mortality due to fires that took place in 1917 and 1924. Tanner also recognized the probable importance of fire in the pre-contact era, although he seems to have been unaware of the ways pre-contact Native Americans used fire, both for agriculture and habitat management. (The impacts of Native American fire use were almost surely different from what occurred in the 20th century Singer Tract).

Neither Tanner (whose study predates the emergence of the discipline) nor Snyder, take environmental history sufficiently into account. There had been major ‘changes in the land’ long before large scale logging began in the southeast and before the reports of local abundance on which Snyder relies. These changes include: the post-contact collapse of Native American civilizations, the introduction of European plant and animal species, the clearing of log jams on major and secondary North American rivers, habitat fragmentation due to the plantation economy, and the near extirpation of the beaver.

All of these elements likely contributed to a major decline in ivorybill populations. Ivory-billed woodpeckers likely concentrated locally in response to major disturbances, regardless of whether forests were old-growth or advanced second-growth, and this type of specialization caused birds to congregate, making it easier for collectors to kill them in large numbers in short periods of time. Snyder likely misinterpreted this collection of large numbers of Ivory-bills in short periods of time as reflecting a greater regional abundance. In contrast, and more consistent with Tanner, this ecological response to disturbed areas led, in some places, to the collectors extirpating regional populations.

In the latter part of the 19th century, hunting probably sped the collapse of the remaining population, but Snyder’s claim that available data on diet and foraging methods do not provide compelling evidence of specialization fails to account for the anatomical and other evidence that suggests otherwise. It also fails to account for the Pileated Woodpecker’s far more extensive range and ability to thrive in a wider variety of habitats, including badly fragmented and degraded ones. I made some of the case for specialization in a series of recent posts, but there’s more to add, especially with regard to ants.

In one of those posts, I hypothesized that the inability to exploit ants as a food resource was a key component, perhaps the primary component, in explaining the decline of the ivorybill. A commenter asked whether there’s evidence to support the idea that ivorybills and other Campephilus woodpeckers don’t feed on ants and also whether there’s evidence to support the idea that Campephilus woodpeckers don’t regurgitate.

Adult Campephilus woodpeckers rarely feed on ants but do not feed them to their young. They make frequent trips to the nest with food items stored in the bill or at the back of the bill. (M. Lammertink, pers. comm.) Dryocopus woodpeckers and those in closely related genera (the “tribe” Malarpicini) feed their young by regurgitating, while other woodpeckers do not. (Manegold and Topfer, 2012). I think the capacity of Pileated Woodpeckers to consume ants in large quantities and to feed them to their young is a significant distinguishing factor and that Tanner was correct in suggesting that food supply was a major limiting factor on Ivory-billed Woodpecker populations.

Ants comprise up to 33% of the world’s terrestrial animal biomass. In Finland, they comprise as much as 10%. In tropical forests, the percentage is much higher, exceeding vertebrate biomass by 400%. Tanner’s comparative analysis of available ivorybill and pileated food did not include ants, so Tanner’s comparative estimate of available insect prey – suggesting that pileateds in the Singer Tract had access to approximately four times what ivorybills did – was in fact extremely low.

Tanner’s dissertation concludes with a discussion of Audubon’s ivorybill dissection, something that was omitted from the monograph. While I had a passing familiarity with the Audubon material, I had not looked at it carefully. Nor had I compared his ivorybill and pileated dissections.

Tanner wrote: “The proventriculus is both muscular and glandular. Audubon’s drawings and text indicate that the proventriculus of a Pileated is much larger in proportion to the stomach than is the case in the Ivory-bill.” Audubon described the ivorybill proventriculus as being only minimally wider than the esophagus. By contrast, the pileated proventriculus as “an immense sac, resembling a crop, 2 1/4 inches in length and 1 and 5 twelfths in width,” or nearly three times as wide as the esophagus.

Screen Shot 2016-07-11 at 1.47.01 PM

Audubon’s drawing of Ivory-billed Woodpecker digestive tract showing slightly widened proventriculus.

Screen Shot 2016-07-11 at 1.47.28 PM

Audubon’s drawing of Pileated Woodpecker digestive tract showing large, sac-like proventriculus.

The proventriculus and stomach of one of Audubon’s specimens contained “a vast mass of ants and other insects”. According to Bent, Beal found one pileated stomach that contained 2,600 ants. (Others contained fewer, 153 and 469, according to Sutton.) Thus, it’s clear that even if ivorybills sometimes ate ants, they lacked the capacity to consume them in large quantities, let alone feed them to their young.

This supports Tanner’s view that specialization was a limiting factor on ivorybill populations. I’ve previously suggested that this might apply only to breeding season, but it seems reasonable to infer that it’s a factor year-round, based on the differences in proventricular structure.

All of that said, I’d argue that this specialization should not necessarily be read to include dependence on large tracts of mature, contiguous forest. The data from the Singer Tract suggest that even under these ‘optimal’ conditions, breeding was limited. And the fact that the Mack’s Bayou birds bred successfully in an area of second growth suggests that birds could thrive under ‘suboptimal’ conditions. The extent to which survival might be possible in fragmented habitat is less clear, but Snyder (citing Jackson) refers to the Mississippi population of six pairs in a 19.2 square mile forest that Tanner missed; the tract is less than 1/6 the area of the Singer Tract and is smaller than many contemporary wildlife management areas.

The tract, known as Allen Gray Estate, was west of Skene, Mississippi in Bolivar County; some or all of it is now part of Dahomey National Wildlife Refuge; the US Fish and Wildlife Service Habitat Management Plan for the refuge (2013) states that the forested portion of the refuge comprises 8100 acres and provides this historical information, “Dahomey NWR is located on the grounds of the old Dahomey Plantation founded in 1833 by F.G. Ellis and named after the homeland of his slaves. Much of the land west of the refuge was probably cleared for cultivation around this time. The land went through several owners and was purchased by Allen Gray in 1936. The portion that became the refuge was known as the “Allen Gray Woods”. This was the only significant portion of the plantation still forested.”  This 8100 acre figure is 25% lower than the figure reported by Jackson and Snyder.

While I have been unable to find a detailed logging history of Bolivar County, it is in the heart of the Mississippi Delta, which was known for its plantations. Between 1900 and 1940, Bolivar County was more densely populated than Madison Parish: 39.1 people per square mile as opposed to 18.9 in Madison Parish in 1900, 78.92 as opposed to 22.78 in 1930, and 74.57 as opposed to 28.33 in 1940. Based on population density and the number of towns, it seems self-evident that the habitat in Bolivar County was considerably more fragmented than was the Singer Tract.

Thus, there is good reason to question Tanner’s old growth model as well as the idea that large contiguous tracts of mature forest are required. Similarly, there’s good reason to question Snyder’s argument that hunting rather than specialization was the primary cause of the ivorybill’s collapse.

Efforts to reintroduce the beaver in the southeast began in the 1930s, and the population has been growing ever since. Beavers injure trees by partially or fully girdling them and by altering hydrology, which weakens or kills trees at the edges of the ponds they create. Beaver damage renders trees more vulnerable to infestation by ivorybill prey species, something we’ve observed repeatedly in our search area. In Tanner’s day and in the late 19th century, the beaver was barely a part of the southeastern ecosystem, but by the 1950s, beavers again were playing a role in altering southern forests, whether mature or successional.

If the ivorybill was able to survive the logging of the last large tracts of old growth forest, as I think it was, the reintroduction of the beaver may have been central to its persistence. If this hypothesis is valid, there is considerably more potential habitat today than there was in Tanner’s era; much of this potential habitat has been overlooked or dismissed in organized search efforts; and the dismissals of post-Tanner reports based on his habitat model rely, at least in part, on a false premise.




Digging Deeper into Tanner, Part 3 of 3 – Prey Species

Part 1 is here. Part 2 is here.

At an IBWO Recovery Team meeting during 2007, a report was shared by Dr. Nathan Schiff and his colleagues at the USDA Forest Service’s Southern Hardwoods Laboratory that more formally described many of the paradoxes that have been discussed in this series of posts. It provides more information on what is known today about the ecology of the wood boring species documented as having been fed upon by Ivory-billed Woodpeckers. Schiff and his colleagues point out that the larvae Tanner collected from a John’s Bayou nest cavity and those described from stomach contents don’t prefer sweetgums, don’t live in high branches, and spend the bulk of their lives in the heartwood, often in the lower parts of trees.

While it’s not mentioned in the Schiff et al. paper, Mallodon dasytomus or what Tanner called Stenodontes (by far the largest single food source in his sample of remains from the nest cavity described above) is commonly known as the “hardwood stump borer”, and Neandra brunnea, a close relative of another known prey species, Parandra (or Hesperandra polita), is called the “pole borer”. These beetles have a life cycle of 3-4 years.

The authors point to direct evidence that of the six species of insect identified in Tanner’s monograph, none would use wood consistent with the high branch/sweet gum focused foraging model. In addition to Mallodon, and P. polita, these species are: Neoclytus caprea (banded ash borer), Dynastes tityus (Eastern Hercules beetle), Alaus ocualtus (eyed click beetle) or a close relative, and an unidentified Scolytid or bark beetle (not found by Tanner.) We have found both P. polita and A. oculatus adults on suspected feeding trees.

Scolytids are tiny. Neoclytus begins its one year life cycle in early spring; the larvae start feeding under the bark and then burrow into the sapwood, where they pupate and spend the winter before emerging as adults. The species prefers ash but may also occur in hickory, oak, and elm. It is found in downed logs, as well as standing trunks and limbs of stressed to dead trees. Dynastes tityus or Hercules beetle larvae live and feed in the “rotting heartwood of logs and stumps.” Alaus larvae are predatory on Cerambycid larvae and live in decaying stumps and logs; eggs are laid in the ground. In addition, the authors point out that at least some of the larvae Tanner found under bark on higher branches (p. 42) require wood that’s in an advanced state of decay, when bark would be loose.

The insect larvae identified for Tanner (Mallodon, Alaus, Neoclytus, and Dynastes) came from remains he found in nest debris. I think this suggests he may have failed to observe or have unduly downplayed one or more foraging behaviors related to obtaining food for nestlings – excavation of very decayed stumps and logs and extensive scaling on boles in particular – since these are lower dwelling species and two of the four inhabit wood that’s in an advanced state of decay. My anonymous correspondent disagrees with my reading of Tanner but makes a very interesting observation that sheds important new light on the data.

Schiff et al. point to an apparent contradiction; none of the food items found in the nest reflect the preference for high branch foraging that Tanner described. The importance of large Cerambycid larvae (especially Mallodon or Stenodontes dasytomus) in the feeding of young ivorybills at Singer Tract remains unclear. Tanner’s observations indicate that most foraging events involved a substrate (recently dead or dying branches) that doesn’t support these large wood-boring larvae. Tanner reported that Mallodon and other large larvae that were “frequently carried in the bills of adult Ivory-bills”. Some fragments of larvae that were found in the remains of at least one nest cavity had to have come from boles or large, lower branches and were likely to have been obtained from longer dead wood, at least in some cases.

It appears that attention today on the Cerambycid larvae “paradox” may have been founded in part on a misreading of Tanner. The Cerambycid and other large larvae found in the three stomachs reported above were from birds collected in August and November, well after the breeding season. Because Mallodon is so large, was the most abundant prey species found in the nest, was identified in one of the stomachs, and was quite likely the species found by Wilson and others, many have interpreted Tanner as saying that it was the primary prey species. It was undoubtedly an important and calorically rich one, but Tanner’s observations suggest that smaller larvae played a more important role, at least in the case of the John’s Bayou birds.

While he frequently saw adult ivorybills carrying large larvae in their beaks, he observed the birds carrying large numbers of “small” larvae even more frequently. He noted the apparent conflict between his observations and what was found in the nest debris and resolved it by hypothesizing that the smaller insect parts probably remained “imbedded in the feces” and were “removed when the adults cleaned the nest” (pp.40-41). Thus, while there is direct evidence that large wood borer larvae were part of the ivorybill prey base, Tanner’s overall interpretation was that smaller larvae were more important during the breeding season. (pp. 40-41, 51-52).

Tanner admitted that he did not fully understand why ivorybills did not forage more frequently on substrates supporting larger larvae when they were fully capable of doing so. He speculated that the smaller larval woodborers when abundant “are very abundant” for short periods of time, beneath the bark of recently dead or dying wood. In sum, Tanner concluded “The Ivory-bill’s insect food supply is smaller, more variable and erratic, and more unevenly distributed than that of the Pileated.”

To reiterate, Tanner stated specifically that while most of his observations involved scaling of high branches, presumably for smaller larvae, he also observed scaling on boles where larger larvae dwell. Tanner suggested that foraging on trunks took place when trees were “longer dead” and that ivorybills “move downward with the progression of shallow borers” (p. 41) The balance of his observations (27.8%, a not inconsequential number) involved digging for “deeper-living” larvae that spend most of their lives in the heartwood, between the ground and the large lower branches.

In their unpublished manuscript, Schiff et al. concluded that: the “. . . Ivory-billed Woodpecker is an opportunistic feeder with catholic tastes that eats beetle larvae where it can find them and that it probably digs for them with its powerful bill. ” This conclusion was intended to challenge Tanner’s finding that food supply imposed a limitation on ivorybill nesting success, but a close reading of Tanner suggests this conclusion actually is not at variance. It is clear now that Ivory-billed Woodpeckers could and did forage by digging into older boles like Pileateds, though less frequently. It is also clear that they scaled bark at all levels. The new revelation for many of those interested in ivorybills today is that this species showed a preference for stripping bark in pursuit of large numbers of smaller larvae in recently dead and dying trees and carried these smaller larvae en masse to their young.

I think the foregoing observations make a lot of sense. As discussed, perhaps ad nauseaum, I have some doubts about what I take to be Tanner’s conclusions about decay class. I have questions about the way he characterized his data on tree size and wish he had quantified scaling on branches relative to scaling on boles. I also question his suggestion that scaling on boles was done on longer dead trees (and the rationale that trees die from the top down) because it doesn’t account for the fact that the larger bole dwelling Cerambycids can attack injured live trees and hasten or bring about mortality, as was the case with the suspected feeding tree shown on the homepage. I have little doubt about his observations at the nest. For one thing, the number of is considerably greater, 159 as opposed to 101.

There is some reason to think Tanner was at least partly correct with respect to variability and scarcity of this food supply, especially in the higher branches. As noted in the previous post, Tanner found no Cerambycid larvae at all in a random sampling of cutover plots near Horseshoe Lake. The location of these surveys was likely between the Bayou Despair and Greenlea Bend home ranges and not far from where two young birds were seen in 1932. Ivorybills were disappearing from these two home ranges, as well as from the nearby Little Bear Lake range, and the three ranges only produced one successful nest (Greenlea Bend, 1937) between 1934 and 1939. (p. 39), and it seems possible that scarcity of this food supply was a contributing factor.

Tanner specifically looked for insect larvae “from several situations similar to places where ivorybills fed”. This was presumably not a random sample. While details about these “situations” were not provided, they included: under bark of dead sweet gum and willow oak limbs (presumably downed), under the bark of a Nuttall’s oak (condition and part of tree unspecified but presumably a downed limb or limbs based on the species found), and the trunk of a dead hackberry. As might be expected, he found Mallodon, P. brunnea, and A. oculatus in the hackberry bole.

He found Urographis (now Graphisurus) fasciatus and Leptostylus sp. in both species of oak and in the sweet gum. These are small Cerambycids (adults up to 15 mm). He found small Cerambycids, Aegomorphus decipiens (now modestus) under sweet gum bark and Xylotrechus colonus under the willow oak bark. In addition, he found Pyrochroidae (torch beetle) larvae, possibly Dendroides canadensis, in sweet gum and Nuttall’s oak and unidentified Elaterids and Buprestids in the Nuttall’s oak.

Questions remain. Some of these larvae, the Elaterids and Pyrochroids in particular, are found under loose bark in decayed wood, suggesting that at least some of the infestation took place after the limbs Tanner examined had fallen. Aegomorphus also feeds in “soft, decaying hardwoods.”  Graphisurus fasciatus is a common species that prefers trunks and large branches. Xylotrechus colonus, prefers “recently killed trees” and is described as “one of the commonest eastern Cerambycids”. At the same time, Tanner’s very limited random sample suggested that high branches had considerably less available substrate and food than other tree parts. This may suggest that sporadic, localized outbreaks of larval infestation in high branches are crucial for breeding.

There are a couple of added twists to this story. To restate and expand on the foundation of my hypothesis about diagnostic feeding sign: Campephilus anatomy, and especially that of the northern triad (Imperial, Cuban ivorybill, and U.S. ivorybill), is highly specialized. Members of this genus are built to scale bark with greater speed and efficiency than any other woodpecker species, but they are also certainly capable of digging. When they dig, they may be powerful, but I suspect their morphology makes excavation a less efficient foraging strategy.

In contrast to Pileated Woodpeckers, which have evolved to make perpendicular blows, ivorybills have pamprodactylous feet (an evolutionary adaptation that rivals the opposable thumb in terms of how radically it differs from other picids), longer necks, longer, stiffer tails, and larger, broader bills. All these adaptations enable them to deliver strong lateral blows but probably impact their ability to excavate. This may explain why many of the foraging pits shown in the Pearson photograph and in Plate 11 are skewed and why ivorybill nest cavities are asymmetrical. It might also explain why ivorybills dig relatively infrequently during breeding season and instead undertake long daily circuits to strip bark and gather larvae, both large and small, for their young.

Ivory-billed Woodpeckers don’t eat ants or termites and don’t regurgitate. They must obtain live, and when possible large, beetle larvae or large quantities of smaller ones. It’s beyond dispute that they do this most often by scaling bark and finding these insects at or near the exposed wood. Based on the presence of Neoclytus in the nest, it’s reasonable to infer that some prey species are taken early in the life cycle, before they burrow into the heartwood, while others simply live under bark. In addition, several species (Mallodon and H. polita at least) may be exposed when bark is stripped from the bole and their larvae are digging exit tunnels but have not yet sealed their pupation chambers. This is the time when the larvae are largest and most nutritious. This is the substrate in which Tanner found the highest concentration of food, and ivorybills are uniquely adapted for exploiting this opportunity. I believe we have seen evidence of this behavior on some hickories, sweet gums, and oaks in our search area.

One or both of these foraging strategies may be keystones. Fluctuations in the availability of these particular food sources might have a significant impact on nesting success.

Whether or not I’m exactly right about all this, I think there are several important points that deserve to be reiterated.

  1. The Singer Tract ivorybills “usually” or frequently fed on high, freshly dead sweet gum and Nuttall oak branches; what they were feeding on remains unclear; however, there is no doubt about the importance of the prey collected (whatever it was) at the treetops for raising young. Specifically, on April 23, 1939, Tanner observed both adults feeding “Baby Bunting” from prey captured from the top of a dead pecan (hickory), and also the long flights these three, along with “Sonny Boy” (the previous year’s young, still with adults), made from one foraging tree to another. There is also no reason to doubt that prey from treetops made up a substantial part of what was fed to the young before fledging.
  2. Ivory-billed Woodpeckers in the Singer Tract could and did feed at all levels and on wood in all stages of decay, but during breeding season, at least, they took most advantage of more recently dead and dying trees.
  3. Despite the habitat and tree species preferences documented by Tanner during the 1930s, the last few ivorybills could and did feed in areas and on tree species that Tanner did not document as being heavily used during his study. This was mostly in the 1940s, after massive cutting was under way. Subsequent interpreters of Tanner have inferred that these tree species and areas were unimportant or unsuitable, and some of Tanner’s later statements may have abetted this misunderstanding. The takeaway is that ivorybills will feed on a variety of tree species, provided the trees are stressed and infested with wood boring larvae that can be quickly collected by scaling bark.
  4. Prey species were most heavily concentrated in what Tanner called “hard but partly punky” stumps. Though it’s not explicitly stated, this class is likely to include the boles and large lower branches of standing trees, including Cerambycid infested trees that have not yet succumbed.
  5. Despite popular perceptions, large trees are not a requirement. Notwithstanding our disagreement about how to characterize foraging frequencies and size class explored at length in the first post in this series, my collaborator and I agree that insect abundance, not tree size per se, is the most significant factor. The foraging behavior documented by Allen and Kellogg and the nesting successes in mostly second growth but fire damaged forest (Mack’s Bayou) earlier in the 1930s support this interpretation.

I hope this series of posts will prove useful to other searches and that it provides greater clarity about ivorybill foraging behavior.

Addendum, March 26: A biologist wrote to point out that I may have been regurgitating conventional wisdom on the subject of Campephilus regurgitation. Some of the literature states that they do feed their young in this manner, and there is language in Tanner to suggest this may be so for ivorybills (pp. 74-75). “Often it seemed to be jerking as if working food from the back of its mouth.” As I read Tanner, the number of larvae that may have been regurgitated seems small, a single grub in at least one instance. The passage in Allen and Kellogg (mentioned in the comments) involved termites, and it is highly speculative. And I recall reading that ivorybills were hunted for food specifically because they didn’t taste of formic acid, unlike pileateds.

At present, I don’t think this information calls for a major revision of the hypotheses presented here, but I plan to do some additional research and may have more to say on these subjects in future. I’ll be completing a week in the field today and expect to post a trip report before too long. As a preview, I’ve found an unprecedented quantity of recent high branch and upper bole scaling this week, all of it on sweet gums.

Digging Deeper into Tanner, Part 2 of 3 – Foraging Substrates

Part 1 is here.

The second observation from the Singer Tract I want to discuss took place in May 1932, when Audubon Society President, T. Gilbert Pearson and Audubon Sanctuary Director Ernest G. Holt were the first ornithologists to observe Ivory-billed Woodpeckers (a minimum of three separate individuals) on the Singer Tract. As discussed in an earlier post, a newspaper account describes Pearson’s observations as follows: “The birds were feeding on stumps of rotting trees, the tops of which had been broken off. A favorite place for feeding is also on dead limbs at or near the tops of the very tall sweet gum trees found abundantly in this region.” It seems likely that some of these “stumps” broke due to having been weakened by larval infestation, although other factors undoubtedly were in play.

Pearson described his visit to the Singer Tract in Bird Lore (not available online) and included a photograph of the tree on which an ivorybill was first sighted. The tree has been heavily worked on by woodpeckers, and notwithstanding the poor quality of the image, the excavations look similar to some discussed in this speculative post.

Holt:Pearson Tree

Pearson’s observations present an interesting comparison with Tanner’s from later in the 1930s. Tanner wrote that “Ivory-bills in Louisiana usually feed high in the tops of dying trees, but they are not averse to coming close to the ground.” If there was any question, Tanner’s Plate 11 shows digging similar to that pictured by Pearson above and also shows that foraging occurred close to the ground.

(There are a couple of additional points of disagreement with my contributing biologist here. I think Pearson’s feeding tree is in a more advanced state of decay than the one shown in Plate 11; at the very least it has been far more heavily excavated. I also think that Pearson’s reference to “rotting stumps” implies a more advanced state of decay than what Tanner documented for Ivory-billed Woodpecker foraging at John’s Bayou while feeding young.)

Since Tanner’s monograph was published, a misunderstanding has arisen – that the Singer Tract birds were canopy specialists that rarely or never foraged on boles or near the ground. Tanner contributed to this misunderstanding after the publication of the monograph by dismissing reports of birds being flushed from near the ground and using this as one of his arguments for designating the species as extinct. If there was any doubt that Ivory-billed Woodpeckers foraged on boles (all the way to base), at least two of the four plates in the monograph (7 and 9) show scaled trunks; another (11) shows a hackberry bole (or stump) that has been scaled and excavated. Allen and Kellogg (1935) reported watching a female ivorybill foraging on the ground “like a Flicker”, and their photographs of foraging show feeding to near the ground on multiple small, rotting pines in Florida and on a “gum” in the Singer Tract.

Tanner did not quantify the frequency with which the Singer Tract birds foraged at different levels beyond saying that they “usually” fed high, but these remnant populations were clearly not averse to feeding low when circumstances required it, and as will become clear in the next post, there’s good reason to suspect that foraging lower on trees might have happened more frequently outside of breeding season.

In the back and forth leading up to this post, my interlocutor had an important insight. He suggested that some misunderstanding has arisen regarding the species of wood boring beetles that would be most important as prey for Ivory-billed Woodpeckers, at least during breeding season. Anecdoctal reports from many familiar with ivorybills, including Alexander Wilson and John James Audubon, have led to an emphasis on large Cerambycid larvae as a major prey item. Of the three Ivory-billed Woodpecker stomachs examined (p. 50), Cerambycid and other large beetle larvae made up high percentages of the animal matter identified in each stomach. It’s clear that ivorybills do feed extensively on large Cerambycid larvae and may prefer them at times, but as with all things ivorybill, nothing is straightforward. There will be more on this in the next post.

Tanner was the first and only observer to attempt to quantify at the availability of wood boring larvae and foraging substrates used by both Ivory-billed and Pileated woodpeckers. As I read Tanner, ivorybills, but not Pileateds, fed under the tight bark of dead on high limbs; both species feed on “hard but partly punky” stumps, and Pileated Woodpeckers (but not ivorybills) feed on “punky, and punky and rotten” limbs, stumps and logs.

I think Tanner was categorical about this in both the table and the text on p. 52. My contributor strongly disagrees and reads Tanner as not dismissing the possibility that ivorybills foraged on more decayed wood, including “punky and rotten” limbs, stumps, and logs, only that he did not observe this for the successfully reproducing Ivory-bills at John’s Bayou during his three year study.

Either way, crucial to the hypothesis that food was limiting to Ivorybills during the breeding season was Tanner’s sampling of wood boring insects among these three different substrates in the Singer Tract.

In May of 1939, Tanner did a survey in a freshly cutover area near Horseshoe Lake. He sampled eight .25 acre plots. He expressed some doubts about the data because he had difficulty finding dead limbs among the smashed tops but only to the extent that the “amount of Ivory-bill food discovered was undoubtedly less than was actually there” and the “calculated proportion of Pileated to Ivory-bill food was quite a bit greater than normal.” (pp. 50-51.) While two young birds were seen in the vicinity of Horseshoe Lake in 1932, this area was not in any of the home ranges delineated by Tanner, and he did find Cerambycid larvae under the bark of dead limbs that he apparently collected, presumably in non-random fashion, from downed wood in the John’s Bayou area. I’ll discuss these findings in the next post.

These caveats aside, the findings are dramatic because they seem to conflict with the observations mentioned above and even some of Tanner’s observations and data, particularly with respect to Cerambycid larvae.

The more decayed type of wood was over 12 times more abundant than the hard but punky stumps, the scarcest substrate in the sampled area but also the one with the highest relative abundance of Cerambycid larvae. Thus, this class seems to have offered by far the highest return on foraging investment, 237 cubic centimeters of insect volume in a mere 30 cubic feet of wood.

The higher, freshly dead limbs contained no Cerambycid larvae at all. Tanner estimated the total volume of insects for this class was 27 cubic centimeters in 80 square feet of foraging surface, a truly minuscule quantity. (He measured the surface area rather than the volume in this class.) The volume of insects in the most decayed substrates was 1154 cubic centimeters in 386 cubic feet of wood, abundant but not nearly as concentrated as in the “hard but punky” stumps.

This led Tanner to estimate that Pileated Woodpeckers in the Singer Tract had access to 40 times more food than ivorybills (p. 51) and to infer that this explained the much higher density of Pileateds (about 36 for each ivory-bill) there. I suspect that there’s another major factor that accounts in part for this difference.

I’m puzzled by the fact that Tanner never mentioned ants and termites when comparing ivorybills and pileateds. Long before he conducted his study, ants and termites had been documented as a primary Pileated Woodpecker prey species.

“F. E. L. Beal (1911) gave the results of examination of the contents of 80 stomachs collected far and wide throughout the range of the species . . . Beetles made up 22.01 percent of the total, and ants 39.91 percent. As many as 2,600 ants were counted in a single stomach. The ants were “mostly of the larger species that live in decaying timber.”

In another study of 23 PIWO specimens, also cited by Bent, ants comprised 60% of the stomach contents.


Pileated Woodpecker Feeding on Ants, Tensas NWR, 2010

There’s no evidence to suggest that ivorybills ever preyed on ants. Pileateds can gather and regurgitate ants, termites, and their larvae in great numbers, whereas Ivory-billed Woodpeckers must bring live beetle larvae to their young. It does not seem farfetched to suggest that the Pileated Woodpecker’s ability to exploit this abundant resource is a major reason for its relative success and that the ivorybill’s inability to exploit it could have been a major limiting factor on population and fecundity.

Tanner may have proceeded from the assumption that if there was competition between these two species, it would be for available “borers” (meaning beetle larvae), but the omission seems problematic and hence puzzling to me because it leaves a distorted picture of the degree of competition between the species and makes them seem more similar than they are in fact. It also skews the data Tanner presented and discussed on p. 51, since the volume of food available to Pileated Woodpeckers, but not ivorybills, in more decayed substrates would be substantially greater if ants were included.

There’s no reason to doubt Tanner’s observations with respect to high branch scaling, but it still seems paradoxical that the John’s Bayou birds “usually” foraged on parts of trees that apparently contained a lot less food. We’ll explore this paradox in more detail in the next installment.

Trip Report: February 5-11, 2016 and More on the Singer Tract

I flew into Houston on February 4 and arrived at the search area on the morning of the 5th. Frank’s work schedule had precluded him from returning to the search area during my absence, and he was unable to get time off to join me this trip.

Tommy Michot visited on the February 5th; we went to the northern sector, and passed the downed sweet gum top (actually a limb) found in April of last year. Project Coyote had a camera trained on it for some time but took it down due to equipment failure. The main stem, which reaches from the ground to about 20 feet up, had been scaled extensively, down to the base, over the course of the last month. Some of the work had been done no more than a few days prior to my arrival based on the condition of bark chips found at the base.


We have a camera back on this top but have low expectations, since so much bark has been removed that it makes a much easier target for other species of woodpecker. While I don’t believe in the “curse of the ivorybill”, individuals and small groups of self-funded searchers face enormous obstacles and are dependent on equipment that’s often unreliable.

Tommy and I measured a number of the largest trees in the area, and the biggest oaks and sweet gums are around 4’ DBH, with many more in the 3’ range. Here are some of the highlights: two Nuttall oaks: 137 cm/53.94”, 119 cm/46.85″; swamp chestnut oak 110 cm/43.31” four sweet gums: 124 cm/48.82”, 123 cm/48.43”, 110 cm/43.31″, 109 cm/42.91”.

While ours was not a random sample, this table from a 1986 paper by Tanner (on data collected in the Singer Tract in 1938), is interesting for the sake of comparison.Screen Shot 2016-02-19 at 7.37.40 AM

In his 1944 report on the Singer Tract, Richard Pough described sweet gums in the 5’-6’ DBH range as being characteristic of old growth conditions, and such trees were not uncommon in the 19th century. Impressionistically, at least, most of the ~4’ DBH sweet gums in our area are moribund and are likely to have lost their tops. I know of only one gum that appears to be in the 5′ DBH category. As of 2009, the national champion sweet gum had a DBH of 5’4.6″. The tree below could be close to that.


Frank Wiley with massive sweet gum

Many, perhaps most, gums have at least some beaver damage. This may be contributing to the earlier mortality, both by stressing the trees directly and by creating the opportunity for beetles to infest them. I have long suspected that the decline of the beaver could have contributed to the IBWO’s disappearance, since beavers directly damage trees by gnawing and also stress or kill them by altering hydrology.


124 cm dbh sweet gum with beaver damage at the base, measured this week. The photo was taken in 2013, and the top of the tree has fallen in the interim.


Beavers were extirpated from much of Louisiana by the early 19th century. As of 1931, populations were restricted to the Amite and Comite rivers in the southeastern corner of the state; they were reintroduced in other areas in 1938 and had established themselves in 21 parishes as of 1951. (Wylie Barrow, pers. comm.) Range expansion continued into the 1990s and after. They’re now considered a pest animal and appear to be found in all parishes.  A recent paper suggests that the introduction of beavers into Magellanic Woodpecker habitat may have benefitted that species.

I was on my own on February 6th, and I went and staked out the downed top for the better part of the morning. Nothing landed on it except for a Red-bellied Woodpecker that pecked and gleaned but did not scale bark or do any excavating. At approximately 9:30, I did a very aggressive series of ADKs. I heard a couple of loud single knocks that seemed to come from no more than a couple of hundred yards away and also a possible double knock. These came during a period when I was standing, moving around, and doing the ADKs, so I did not hear them very well. In addition, there were a few distant gunshots within about 15 minutes after the series, so I’m not very confident about what I heard. (These were the only shots heard all day.) I found some scaling the next day a couple of hundred yards away (discussed below). This gives me some reason to think the SKs were a reaction, not shots. Still I’d place these in the weak possible category.

One highlight of the day was watching a pair of Red-shouldered Hawks in the act of mating.

I returned to the same area on the 7th, with two cameras to deploy. One is aimed at a large sweet gum stub, about 20’ tall and well over 3’ DBH that I found last trip. The top had broken off shortly before my arrival. While it’s not discussed in Tanner, T. Gilbert Pearson, who was the first modern ornithologist to observe the Singer Tract IBWOs, described this type of “stump” as one of the species’ preferred feeding sites. This is a tree on which I found high branch scaling last year, before the top broke off. I expect this to be a long-term deployment.


I also redeployed a camera on the downed top, although we’re not very hopeful about that location, since the scaling is so extensive and the bark has been loosened in many of the remaining unscaled areas.

I walked south for a couple of hundred yards and found very fresh, large bark chips at the foot of a live sweet gum (there were two large gums ~3’ dbh about 10’ apart). There was extensive scaling on live or recently dead high branches of one or both of these trees. Because there had been a major rainstorm and accompanying minor flooding a week before and the chips were mud free, I can be sure the scaling took place after the rain, and since Tommy and I had spent considerable time in the area examining some other nearby scaling two days before, I strongly suspect this work was done on the 6th. I can’t help but wonder whether the possible single knocks came from whatever was doing the scaling; that would be consistent with my immediate impression when I heard them, both in terms of distance and direction. Nonetheless, my confidence level about the SKs is low given the gunfire.



I don’t think the scaling and bark chips are consistent with squirrel; the chips are large and thick and do not show signs of having been chewed off; the ones collected weighed over five pounds.


There was a little excavation and exit tunnel expansion (visible in the first image above) associated with the scaling; and it has the generally clean edges and lack of layered, flaked off appearance around the edges or on the chips. The leaves and gumballs are attached on most of the limbs, indicating that they’re alive or very, very recently dead, so the bark is almost certainly tight. This is about as good as it gets when up-close examination is not an option

I met Tom Foti, who came in from Arkansas, on the morning of the 8th. Winds were high, with gusts approaching 50 mph. We decided it would be unsafe to venture into the woods, so we drove around the edges of the search area looking at the surrounding upland forest, much of which is impressive and mature. Tom is very enthusiastic about the area, ivorybills or not, and we’re hopeful that steps will be taken to protect and manage it appropriately. The car ride was a running lesson on southern forest ecosystems, and as I told Tom, I’ll count myself lucky if I retain 10% of what I learned.

The next morning, the winds had dropped enough to make it safe to head for the swamps, and Tom and I visited the southern sector, an area where we haven’t spent much time lately. As mentioned in some previous posts, there has been a significant uptick in four-wheeler activity in the area, and it’s heartbreaking to see the destruction these callous individuals are causing. Fortunately, the damage is almost entirely limited to the periphery, and the deeper parts of the bottom are unscathed. The habitat types here are somewhat different, and the logging date is more recent, but it remains very impressive. We walked a long way and went to places I had never been, including a lower-lying flat with tree species I haven’t noticed elsewhere – shagbark hickory, bitter pecan, and overcup oak.


We saw no recent feeding sign in any of these areas, except for some older work on a small sweet gum that I described as being about a grade B-.

We then looped back along a different track, passing the spot where I recorded calls in March of 2013 and where we’d had a concentration of feeding sign in 2012 and 2013. We found nothing until we reached a location farther north that is within 100-200 yards of the tree shown on the homepage. Tom spotted a group of trees with bark scaling, some on boles and some on branches. Once again, this was not “grade A” work, but the concentration makes it more interesting than if it were one isolated example. We did not find any chips at the base of the snag that had been scaled on the bole, and the high branch work is not as extensive some.




It’s worth pointing out that on many days, I’ll walk for hours and see nothing and then find either a dramatic example of scaling or a small cluster of it. Tom and I had probably walked 3.5 miles or more before finding this little cluster.

I was on my own again on the 9th, and I opted to go on a death march to retrieve a trail cam from a tree deep in the swamp and proceed north from there. The tree is a large blown down sweetgum discussed and shown here. There was some fresh scaling on it that I suspect was done by a Pileated Woodpecker. There are nearly six weeks of images to go through, so it will take some time before we find out if there were any captures.

As on the previous day, I walked for a couple of hours without seeing or hearing anything suggestive until I got to a part of the area we haven’t visited since last year, perhaps a quarter mile south of the southernmost point Tommy and I had reached earlier in the week. I found old sign, some of which was fresh last winter and some of which was older. I then found some fresh work on two trees in close proximity to one another. Some of the scaling was on a downed tree but was clearly done by a woodpecker, with chips and other characteristics that I consider to be suggestive. Since the chips were caked with mud, the  scaling was a little over a week old. The other work was on one high branch, but conditions made it impossible to look for chips.



Bark chips found beneath downed sweet gum. iPhone shown for scale is 5″x2.5″


On the return hike, I found what I’m quite sure is Pileated Woodpecker work on a recently dead or dying hickory. Since we’ve found a number of hickories that we suspect have been scaled by Ivory-billed Woodpeckers, this was an unusual opportunity to do a direct comparison. In my view the work on hickories is the most compelling for ivorybill due to the density and tightness of the bark and the hardness of the wood. There are pronounced differences in the presumed Pileated and suspected ivorybill work on this species.

The work on the homepage is suspected ivorybill. It is extensive, with huge contiguous areas, perhaps 20% of the entire surface, completely stripped, with evidence of bill strikes targeted at exit tunnels. The Pileated work is spotty by comparison. The bark chunks scaled from the tree on the homepage were large, dense, and thick, and there were no pieces of sapwood among them. By contrast, the suspected Pileated work involves very small pieces of bark that appear to have been removed by digging rather than scaling; there were also a few pieces of punky wood among the chips.


Presumed Pileated Woodpecker work on recently dead or dying hickory


Detail of presumed Pileated Woodpecker work on hickory


Small chips at base of hickory



Different view of scaled dying hickory that appears on the homepage. Work is suspected Ivory-billed Woodpecker


Bark chips from base of hickory shown immediately above

The next morning, I drove to the Wetlands and Aquatic Research Center (formerly the National Wetlands Research Center) in Lafayette and met with Wylie Barrow, Heather Baldwin, Tommy Michot, and Philip and Eric Vanbergen. (Two young enthusiasts who will be helping us out.) Frank joined us briefly, and then Wylie, Tommy, the Vanbergens, and I went out to lunch. It was an exciting and thought-provoking day, and the Research Center is an incredible facility. Wylie and Heather shared their comprehensive and in-depth analysis of conditions in the Singer Tract in Tanner’s day. They’ve amassed an array of materials encompassing land records, Civil War era maps, and stereographic aerial photographs. Their research far surpasses my own speculative effort. It covers the finest details – roads, improved and unimproved, snag densities, tree mortality, conditions around roost and nest sites, as well as conditions in other locations where ivorybills were seen. Tom Foti has done complementary research on hydrology, soils, and vegetation.

Their presentation convinced me that I’ve been too hard on Tanner in some respects. There was a little more old growth in the Singer Tract than I had inferred from the Pough report and some of the historical documents. Nonetheless, the characterization of the Tract as a whole as “virgin” forest is somewhat misleading, since over a quarter of it was second growth, and some of it fairly young. Heather and Wylie have graciously given me permission to summarize some of their findings.

When Tanner began his study, 72% of the Singer Tract was old growth. (Tanner estimated it at over 80%.) Logging in 1938 reduced that percentage to 67%. The ridges, which Tanner deemed to be the best ivorybill habitat, were actually the least likely areas to be old growth. (Tom Foti’s analysis also points to a preference for higher, drier locations.) The regrowth percentages for each landform in Tanner’s day are as follows:

Ridge (43%)

Low ridge (23%)

Total on ridges (32%)

Flat (9%)

Low flat (4%)

Cypress brake (4.5%)

For the most part, the second growth forests were not particularly old, as has been suggested in previous posts. According to Heather, most of these areas only started to regrow in the 1880s and 1890s, “due to consecutive depressions and low cotton prices”. Thus, parts of the Singer Tract were relatively young second growth, and this included one of the ivorybill home ranges and one that Tanner deemed to be “best” – Mack’s Bayou.

The nature of the habitat in the Mack’s Bayou area is immediately apparent from the 1938 aerial photos, which suggest forest conditions that are present in many parts of Louisiana today. Nevertheless, Ivory-billed Woodpeckers nested there in 1934 and 1935, at minimum, and did so successfully at least once. This fact alone refutes the idea that Ivory-billed Woodpeckers are old growth dependent. Heather informs me that there was an abundance of dead and dying trees on the eastern side of the Mack’s Bayou range, due to a fire caused by logging activities. In any event, the home range Tanner delineated in this primarily second growth area is no larger than the home range he delineated around John’s Bayou, which had more mature forest. In fact, the area he designated as “best” for ivorybills around Mack’s Bayou was slightly smaller than its older equivalent near John’s Bayou.

Tanner knew that a significant portion of the Mack’s Bayou home range was not old growth, since his 1941 map shows “old fields” in the heart of it. He seems to have been unaware of the resurgence of cotton growing during the 1870s and 1880s, so he may have overestimated the age of the forest on that basis. I can’t help but wonder if he glossed over the conditions in the Mack’s Bayou range in part for the sake of protecting the Singer Tract and (as Heather suggested) in part based on what he deemed to be best for the birds from a conservation standpoint, an approach that later ossified into a categorical set of beliefs about old-growth dependence.

As I and others have been arguing for years, extensive forest cover, sufficient dead and dying wood, and enough large trees for roosting and nesting are probably the main requirements, even if old growth or near-old growth conditions are optimal.

I plan to return to the search area in late March and have another post or two in mind in the interim.




Not So ‘Virgin’ Forest: The Singer Tract Myth Debunked

Update: This post includes hard data about the extent of old growth in the Singer Tract (scroll down past all photographs) and in ivorybill home ranges. The general points made below remain valid, although some of the wording is perhaps too strong; Tanner overestimated the amount of old growth in the Tract ( at “over 80%” v. 72% in fact), and the Mack’s Bayou home range was predominantly second growth.

This post is a companion to the previous one and to others discussing habitat conditions in the Singer Tract. Those posts reference Richard Pough’s 1944 report to the Audubon Society. Pough, whose study was never published, noted that much of the Singer Tract had actually been under cultivation prior to the Civil War. But it’s worth taking a closer look at just how much.

Tanner characterized the Singer Tract as “the largest tract of virgin timber in the Mississippi Delta,” contending that it contained “120 square miles of virgin forest in 1934”. He also wrote that the largest plantation “had about 3000 acres under cultivation,” while suggesting that “some of the early settlers along the Tensas River cleared land along the river banks for cotton fields.” Thus, the myth of the Singer Tract as virgin forest was born.

It’s not clear where Tanner got his information, but some of his characterizations are not supported by the historical record; the language about “early settlers” almost seems disingenuous when one looks at the history of Madison Parish. (Rootsweb has many pages devoted to this subject, and I’ve drawn heavily on them for this analysis.) As should become evident, there was a great deal of human activity in and around the Singer Tract, especially prior to the Civil War. I will suggest that most, perhaps virtually all, of the arable land in the parish, had been cleared for agricultural purposes and that the Singer Tract was a mix of second growth and remnant old growth, most of which was in the lower-lying, wetter areas that Tanner deemed to be less suitable for ivorybills.

A 1937 Masters thesis in economics by Robert L. Moncrief, “The Economic Development of the Tallulah Territory”, provides a great deal of information about the parish and its history. In the post-Columbian era, the area was very sparsely settled until the 1830s. Madison Parish was established in 1839, and in 1840, steamboats began plying the Tensas River. A major population influx began in 1836, and the population kept growing until the Civil War, going from 5,142 in 1840 to 14,133 in 1860. The war led to a dramatic decline to a mere 8,600 in 1870. Over the next couple of decades, the numbers grew again to 14,135 in 1890. Changing economic conditions and the boll weevil outbreak caused another decline that was only reversed between 1920 and 1940, when the number of residents reached 14,826. By 2010, it had fallen to 12,093.

Cotton and the quality of the soil drove this influx. By 1850, there were 27 landowners in the parish who owned more than $20,000 (over $590,000 in 2015 dollars) worth of real estate. The largest holding was valued at $140,000 (well over $4,000,000 in today’s dollars).

According to Moncrief, “the newcomers cleared away the heavy forests and planted the new ground in the favored crop then, as now –– cotton. They cleared all the lands fronting water courses (which are the highest and most desirable lands for cultivation in this region) to form a continuous line of plantations along the streams.” Streams in this context refers not just to the Tensas but also to the smaller non-navigable bayous. Cotton raised along the smaller streams was brought down to the Tensas in flat-bottomed boats.

Moncrief’s thesis also includes figures for cotton and corn production in Madison Parish. Cotton production peaked at over 46,000 bales in 1858. (Pough was apparently incorrect in stating it was over 100,000 bales; he may have combined the total with that of an adjoining parish.) It had fallen to 1,830 by the end of the war. Production recovered between 1870 and 1875 and reached a postwar/pre-boll weevil peak of 25,981 bales in 1890, about the same level of production as in 1936. Corn production peaked at 618,620 bushels in 1859, falling dramatically after the war, peaking at 836,000 bushels in 1909, and then falling to 320,000 by 1936. My crude, back of the envelope estimate based on  yields of 5 bales per acre for cotton and 15 bushels per acre for corn, suggests that between a quarter and a third of the total acreage in the Parish was under commercial cultivation prior to the war.

While Moncrief’s paper evokes Tanner by describing the Singer Tract as 81,102 acres (126 square miles) of virgin timber, it also notes, “The tract includes several abandoned and grown up plantations, which after the Civil War, reverted to the state and were later sold to the present owners.” The ruins of one plantation house are still standing, deep within the Tensas National Wildlife Refuge.

While it was adjacent to and not strictly part of the Singer Tract, the story of the Frisby Plantation is illustrative. The plantation was established in the early 1850s with land acquisitions taking place over the next decade. Norman Frisby, the founder, was murdered by his nephew by marriage in an 1863 in a dispute over property. When Frisby’s widow was forced to sell the plantation in 1870, it totaled 19,479 acres, and its crops generated over $77,000 in revenue (the equivalent of $1.36 million in 2015). Tanner visited the site of the plantation and photographed one of its old fields. I haven’t been able to pinpoint the location of the old house; one 19th century survey survey seems to place it in Tensas Parish, near Fool’s River. Another account (from the history of the Sharkey Plantation discussed below) says it borders Disharoon (or Dishroom) Bend, much closer to the core of the Singer Tract. As shown on this overlay of 1875 land ownership on a modern map, the Frisby holdings included parcels adjacent to Mack’s Bayou and on Dishroom Bend.

The maps help flesh out the story. An earlier and incomplete parish-wide map of patentees shows that many parcels in the Singer Tract were purchased well before Frisby started acquiring land in the 1850s. Lands purchased in the 1840s include parcels along John’s and Mack’s Bayous, which makes sense since frontlands along streams were most desirable. This history of the Sharkey Plantation reveals that land sales began in the heart of Tanner’s search area during the 1840s. The author explains that the Sharkey plantation and others like it were more like communities, with a cluster of families (and presumably their slaves) living in close proximity, near a watercourse.  The 1875 map shows that much of the Singer Tract remained in private hands even after many parcels were abandoned during the Civil War.

Perhaps even more telling is this hand drawn map of Madison and Carroll Parishes from 1862. It shows the locations of towns, roads, ferry crossings, and plantations in the Singer Tract area. While it is incomplete and John’s Bayou is not shown, Sharkey Road is there, cutting in a southwesterly direction from the Richmond-Carthage road, crossing Alligator Bayou, and the Swearingen parcel. Another road crosses the heart of the Mack’s Bayou home range and the Tensas itself. The map delineates abandoned plantations and appears to show that, except for those abandoned areas, some cotton was being grown in every division of the Singer Tract. However limited the agricultural activity may have been in these sectors, the area was hardly a primeval wilderness; habitat had been fragmented; and old growth conditions were likely restricted for the most part to areas unsuitable for farming.

This passage from an 1885 article from the New Orleans Times-Picayune sheds some additional light on conditions in the area both before and after the Civil War. “But little has been said about Tensas River and Joe’s Bayou as, but little interest is there as compared with former years. Before the war there was a continuous planting interest all along those two streams but overflows and the war left them to grow up into weeds and bushes. In 1870 Mason, and later Loyd bought cattle from other parts of the country and carried them to those bayou places for pasturage, wherein a few years they made large sums of money. This was in the neighborhood of Quebec, which before the war was a flourishing little city, shipping 7090 bales of cotton. It was at the junction of the Tensas River and the railroad. It is now a waste place and to pass there on the railroad you would never know that a town had been there.” (In The Race to Save the Lord God Bird, p.76 Hoose plays into the virgin forest myth by claiming that railroads “finally reached the Tensas River sometime around 1900.”) Quebec was just a few miles outside the tract, near Bayou Despair, where Tanner listed a pair from 1934-1936.

The Rootsweb pages provide a couple of additional and important pieces of information.

Theodore Roosevelt visited Madison Parish in 1907.  Roosevelt’s descriptions provide added detail about conditions in and around the Singer Tract several decades after the Civil War. According to Roosevelt:

“Beyond the end of cultivation towers the great forest. Wherever the water stands in pools, and by the edges of the lakes and bayous, the giant cypress loom aloft, rivalled in size by some of the red gums and white oaks. In stature, in towering majesty, they are unsurpassed by any trees of our eastern forests; lordlier kings of the green-leaved world are not to be found until we reach the sequoias and redwoods of the Sierras. Among them grow many other trees–hackberry, thorn, honeylocust, tupelo, pecan, and ash. In the cypress sloughs the singular knees of the trees stand two or three feet above the black ooze. Palmettos grow thickly in places. The canebrakes stretch along the slight rises of ground, often extending for miles, forming one of the most striking and interesting features of the country. They choke out other growths, the feathery, graceful canes standing in ranks, tall, slender, serried, each but a few inches from his brother, and springing to a height of fifteen or twenty feet. They look like bamboos; they are well-nigh impenetrable to a man on horseback; even on foot they make difficult walking unless free use is made of the heavy bush-knife. It is impossible to see through them for more than fifteen or twenty paces, and often for not half that distance. Bears make their lairs in them, and they are the refuge for hunted things. Outside of them, in the swamp, bushes of many kinds grow thick among the tall trees, and vines and creepers climb the trunks and hang in trailing festoons from the branches. Here, likewise, the bush-knife is in constant play, as the skilled horsemen thread their way, often at a gallop, in and out among the great tree trunks, and through the dense, tangled, thorny undergrowth.”

The most salient point here is that Roosevelt’s “great forest” applied to low-lying areas in which there was standing water (something that Hoose glosses over). Roosevelt also saw three Ivory-billed Woodpeckers:

“The most notable birds and those which most interested me were the great ivory-billed woodpeckers. Of these I saw three, all of them in groves of giant cypress; their brilliant white bills contrasted finely with the black of their general plumage. They were noisy but wary, and they seemed to me to set off the wildness of the swamp as much as any of the beasts of the chase.”


A photograph from the hunt is here. Tanner seems to have been unaware of the Roosevelt encounter. Roosevelt’s visit came just 17 years into cotton farming’s second decline and 52 years after the end of the Civil War. Habitat conditions are likely to have been poorer in general than when Tanner was there 3 decades later. The relative ease with which Roosevelt saw three ivorybills (despite their wariness) suggests they were not uncommon in 1907 and calls Tanner’s assumptions (pp. 48-50) about fire, tree death, and population influxes between 1911 and 1930 into question.

There’s another gem in the Rootsweb pages. It’s not directly on topic, but it relates to Tanner’s later dogmatism. In arguing for extinction and dismissing post-Singer Tract reports, many of which involved birds being flushed from tree stumps or other locations near the ground, Tanner characterized this behavior as being characteristic of pileateds not ivorybills.

Rootsweb has a newspaper account of  T. Gilbert Pearson‘s  visit to the Singer Tract in 1932. Pearson (who was President of the Audubon Society at the time) was the first ornithologist to confirm the presence of ivorybills in the Tract. He saw, “The birds . . . feeding on stumps of rotting trees, the tops of which had been broken off. A favorite place for feeding is also on dead limbs at or near the tops of the very tall sweet gum trees found abundantly in this region.”

The evidence that relates directly to Tanner’s study area and its immediate environs suggests that claims about “virgin forest” and IBWO dependency on old-growth are based on flawed premises. The Singer Tract was no doubt a remarkable place, a huge area of contiguous and relatively undisturbed forest, but it’s clear that much of it was not old growth or “virgin”.

It’s more useful to think about what the Singer Tract is likely to have offered Ivory-billed Woodpeckers – some measure of seclusion, enough big trees for roosting and nesting, and an abundance of standing and fallen deadwood. The myth that the Ivory-billed Woodpecker required vast tracts of “virgin” forest may be emotionally compelling, but it’s not based on evidence; it’s time to put it to rest.

I’m looking forward to spending a week in the field starting just after Christmas.

Tanner and Population Density

I recently gave a talk to the Rockland County Audubon Society, and a member raised what I think is the strongest question about our evidence and about the persistence of the Ivory-billed Woodpecker in general. “How could the species have survived in such low numbers and at such low densities?”

In other posts, we’ve pointed to evidence that Tanner missed a population in Mississippi and was somewhat cavalier in his approach to evaluating potential habitat, disregarding advice Herbert Stoddard gave him in 1936, “The area where they (Ivory-billed Woodpeckers) may occur at present is simply tremendous, not restricted as many believe.”

A recent study on Magellanic Woodpeckers points to another factor that raises even more doubt about Tanner’s estimated population of 22 in 1939.  The study was conducted in an old growth Lenga forest in Patagonia. According to this study: “Our results show that Magellanic Woodpecker family groups require a minimum of 100 ha in old-growth forest habitat; thus, forest patches in less favourable forest conditions (e.g., younger, managed, fragmented, mixed forests) should probably be much larger to support a resident pair or family.”

The specific criteria that Tanner used for estimating the 1939 population at approximately 22 are unclear, but he assumed a maximum carrying capacity of six square miles per pair. The Wikipedia entry on the IBWO is even worse and is generally rife with bad information; the editors there expand Tanner’s six square miles to “9.7“.  Elsewhere, Tanner suggested a minimum home range of two and a half to three square miles. These numbers are somewhere between six and 16 times the minimum for a large southern congener that, like the ivorybill, lives in more temperate habitat than others in the genus.

Thus, there is a strong possibility that Tanner severely underestimated ivorybill populations in Florida. If he was so badly wrong about home range, he’s more likely to have missed populations in areas that he rejected for being suboptimal and not expansive enough.

Double the minimum acreage required by the magellanic in optimal habitat and apply that number to the ivorybill, and Sherburne, a large Louisiana WMA, could theoretically be home to just under 90 family groups. (We don’t think this is the case.) Even at 450 hectares per family group, the carrying capacity in Sherburne would be nearly 40. Such numbers are improbable in the extreme, but 9 or 10 family groups in an area that size would be very hard to detect.

With significantly smaller home range requirements, a substantially larger population in 1939, and a recognition that Singer Tract-like conditions are not a requirement (as Tanner himself made clear), various survival scenarios become considerably more plausible, assumptions about low densities become more questionable, and the quantity of potential habitat is far greater than anyone has imagined.

Edited to add: Although the study referenced above is more recent, Noel Snyder made the same basic argument about magellanics and other congeners in The Travails of Two Woodpeckers (2009). Snyder posits that hunting pressure, not habitat loss, was the primary cause of the ivorybill’s decline.

Snyder (who to the best of my knowledge has little hope for the ivorybill) does not fully address how Tanner’s assumptions might have affected his population estimates and habitat evaluations. He also doesn’t consider how taking Tanner at face value has influenced both search protocols and the “credibility” of post-Singer Tract reports. Nevertheless, he does hint at what I suspect is the key to the species’ survival: “With food supplies degraded, not eliminated, a reasonable possibility appears to exist that many ivory-bill populations in logged regions might still have found enough food to persist and might have endured at modest densities, had they been free of shooting pressure. The long persistence of the ivory-bill in one quite thoroughly logged region in Cuba supports this possibility . . .”

In my view, Snyder goes a little too far in downplaying specialization as a factor, even if Tanner overplayed it. It’s pretty clear – from range, habitat, and morphology – that ivorybills are more specialized than pileateds. But if the IBWO did persist after World War II and Snyder is right that hunting was a major factor in the species’  decline (even in the Singer Tract), there may be even more room for optimism, since hunting practices changed considerably in the post-war era.



Trip Report November 23-28, 2015, Part One

This post will be something of a departure from previous ones in that we’re writing it jointly.

Though we met in 2008 and started calling our effort “Project Coyote” in early 2010, in many ways this week marks the 6th anniversary of our collaboration. It was the first time we visited the old search area together and everything grew from there. We’re an odd couple, with very different cultural backgrounds, personalities, and worldviews. There have been many strange ups and downs over the years but remarkably few major disagreements. One thing we’ve shared from the start is a similar approach to putting the pieces of this puzzle together – trying to glean what knowledge we can from those who were able to find ivorybills in the past, especially J.J. Kuhn. We think we’re on the cusp of obtaining something definitive for reasons that should become clearer in the post, if they aren’t already.

It’s remarkable that we’ve come this far. The obstacles involved in documenting the ivorybill are enormous. We’re just two individuals with limited time and resources searching in a fairly large, remote, and challenging patch. We have a small circle of supporters and trusted people who visit our area when time allows.

There are huge swathes of potential habitat in the southeast that get little human traffic, especially outside of hunting season, and many of these have not even been considered, let alone visited. It’s not uncommon for foreign (and some domestic) ornithologists to assume that conclusive imagery should have been obtained by now, just because it’s the US and birding is popular here and that extinction is likely because several organized searches have failed to come up with something definitive. Many American birders with little knowledge of or experience in the rural South jump to similar conclusions.

The mere fact that there have been formal, funded searches matters very little. The difficulties in obtaining documentation of an extremely rare, wary bird species that requires a large home range in secluded, difficult habitat are monumental. We think that camera traps are the most promising avenue for obtaining something conclusive (as is frequently the case with cryptic animals). The problem has been to place the cameras in a location that birds are likely to visit. This has been our approach from the start, but it’s only now that we think we’ve solved the problem. While we will continue to use other methods and to host visiting biologists and trusted supporters, camera traps will be our primary focus in the coming months.

On November 23, we were joined by Travis Lux, a freelance radio reporter who contacted us a few months ago and who has promised to keep our location confidential. We visited the downed sweet gum top discussed in several previous posts, most recently this one, and were elated to find that there was some new scaling at the top of the snag and decided that we’d return to Frank’s house to review the footage.


Fresh Scaling on Downed Sweet Gum Top

As soon as we looked at the data stored on the card, our elation turned to alarm and then almost to despair. Travis had been recording the whole day, and we’re sure this will make for some dramatic radio. Frank will pick up the narrative to explain.

Trying to understand what “goes wrong” with the various types of game cameras is a guessing game. Of the three cameras that we had deployed, two of them – upon reaching the 32 gb storage limit on the cards – began to overwrite the files rather than shut off automatically (as the instructions imply but do not directly state they should). The instruction manual was also misleading about the duration of a deployment based on the delay time set by the user. The instructions implied that a 32 gig card would not be completely full at the end of a 60 day deployment.

In reality, the card filled up after fifteen days. The camera continued to operate, but it overwrote the earliest files with the newest files, rendering the earlier files unrecoverable. With this hard learned bit of knowledge, I increased the time lapse to ten seconds, from five, and cut out about 45 minutes of “on” time at each end of the day.

According to the data gathered thus far and calculating data storage capacity vs. time deployed and time lapse setting, this SHOULD give us about 45 days with 50% battery and a fresh 32 gig card. The reprogrammed camera that we pulled the card on Mark’s last day in the field appeared to bear this finding out.

With very few exceptions, these cameras are manufactured and assembled in The People’s Republic of China. The instructions (and this has been true of several different brands we’ve tried) are generally translated from Cantonese or Mandarin…Poorly. Fractured syntax, and confusing usage of common words often leaves the guy programming the camera guessing what the instructions REALLY mean.

One of the cameras shut itself down for unknown reasons after taking just a few images. When checked, the batteries were still above 90%. I put a new card in it and conducted a 30 minute field test; it seemed to be functioning properly. Of the four cameras and three locations where we now have cams deployed, this one is in what we feel is the least likely to be visited by woodpeckers in the near future. Hopefully, the glitch will not reappear.

To add to Frank’s comments, unless one has well over $1000 to spend per unit, there are major tradeoffs involved in selecting trail cams. The brand we’ve selected stores individual frames as the equivalent of deinterlaced video stills. This allows for greater storage capacity and longer battery life but lower image quality, especially at a distance. Fortunately, all of our current deployments are at close enough range to produce a definitive image or series of images, and we now know that the cameras themselves do not scare off whatever is doing the scaling, something we thought possible in the past.

To return to the main topic, after the initial shock and disappointment wore off, we realized that there had in fact been relatively little scaling, except at the very top of the downed crown. The main trunk is almost untouched, and return visits remain a strong possibility. We have redeployed the camera and will leave it in place indefinitely.

Tuesday the 24th was a more encouraging day. We were joined by Tom Foti, formerly of the Arkansas Natural Heritage Committee and a member of the Ivory-billed Woodpecker Recovery Team’s Steering Committee. Tom is perhaps the foremost expert on bottomland hardwood ecology. He was very impressed with the habitat. He jokingly commented that if he were still with the Arkansas Natural Heritage Committee, he’d try encourage his state to annex the area.

While we were unable to show Tom any feeding sign, we did hear a couple of possible double knocks. In addition, we found a recently dead small sweet gum that had apparently been killed by ambrosia beetles, similar to others we’ve found, but as yet untouched.

While there’s no suggestive feeding sign in the immediate vicinity, the location is approximately a mile from the camera deployment discussed above and a few hundred yards from where we recorded an apparent double knock and obtained several intriguing game cam images. Given the absence of recent scaling in the immediate vicinity, we think this is the least promising of the three current locations.


We called it a day early because Tom had to return to Arkansas. What he saw and heard left him enthusiastic and eager to return. It’s a privilege to be around someone who’s so knowledgeable, and we look forward future visits and to learning from him.

If the 24th was a good day, the 25th was even better, though considerably more challenging. Brian, Frank’s son, came along and helped carry some of Frank’s gear. We went into an area that we’ve only visited once before. The area is approximately 1.5 miles from the nearest road, and as it turned out, we did that three mile round trip twice.

The habitat in this part of our patch is magnificent. There’s a good deal of old scaling high on live sweet gums. While this isn’t the type of work that we consider highly suggestive, it is consistent with what Tanner described and photographed (more on Tanner in Part 2).


Old High Branch Scaling at Lower Left and Center

We found a huge and recently downed (leaves attached) sweet gum, part of which fell between two recently dead saplings that both showed signs of ambrosia beetle infestation. Some  of the scaling on the huge downed gum seems consistent with Pileated work (having a layered appearance), and some of it comes close to what we think is diagnostic for ivorybill. This is the first freshly dead tree we’ve found that has feeding sign suggestive of both species.



This Huge Sweet Gum Blowdown Shows Sign Suggestive of Both Ivory-billed and Pileated Feeding.

After finding the sign, we decided the location merited our deploying two cameras. (It would take four or five to cover the whole blowdown.) Brian and I hiked out to retrieve the two that Frank and Bob Ford deployed in October, while Frank went back to his house to get additional cards. We hiked back in and reached the location at around 3 pm.

Over the next hour, we heard several double and single knocks that seemed to be coming from no more than three hundred yards away. A couple of the double knocks were what we consider the best (most Campephilus-like) we’ve ever heard.

It was getting late, so we hiked out as quickly as we could, stopping to rest a little before sunset. As the sun was going down, there was a small burst of shooting from the direction of the road. A PIWO started scolding in apparent response, and Brian and I heard a sharp single knock from the direction of the bottom (away from the shooting) and more distant than the squalling pileated. (As an aside, Frank is a very experienced hunter and can easily distinguish between shots and knocks. It’s not difficult for me either, except when the sounds are obscured by crunching leaves, etc.)

We now have all four cameras deployed on recently dead trees or parts thereof that have a good chance of attracting woodpeckers in the near future. This along an approximately three mile line, with about a mile between each camera.

To be continued . . .