Advertisements

Digging Deeper into Tanner: Part 1 of 3 – Tree Size

Although it is thematically quite different, this series of posts is rooted in my recent reexamination of my feeding sign hypothesis that culminates here. It was also inspired by my recent and much closer look at Tanner and the Singer Tract, new insights gleaned from old material, and the input of others that shaped the previous post. My original plan was to make this entry the last in the previous series, but since it has grown to over 5,000 words and addresses different issues, I decided to break it in three and will post the next two installments soon.

I’ve been engaged in an extended dialog with a biologist who is familiar with all the Ivory-billed Woodpecker literature and knows Tanner’s writings specifically. Our back and forth is the primary reason for the long interval between the previous post and this one. This person provided some very important insights that will be included in these posts. At the same time, we have a few points of disagreement. In the interests of transparency and allowing readers to draw their own conclusions, these points of disagreement will be disclosed in the text.

Ivory-billed Woodpecker foraging behavior and diet and what separates this species from Pileated Woodpecker and other North American woodpeckers are issues that have been hotly contested for years. In my view, ivorybills could (and presumably do) forage on any species of tree in any decay condition. However, Campephilus anatomy is specialized, and the only quantitative, observational data that exist on what this species does while feeding young (Tanner 1942) suggest some specialization was in fact occurring at least at the Singer Tract from 1937 to 1939. The problem is that many of the prey items (specifically identified in Tanner and emphasized by others since Tanner’s study), even during breeding, do not seem to match up well with the foraging substrates documented by Tanner as most used by ivory-bills feeding young.

In my view, there are some discrepancies between what Tanner observed and reported and the physical evidence he collected during his study related to ivorybill feeding. I also think there may be discrepancies between Tanner’s observations and those of others from the Singer Tract. At least one thing is clear, Tanner’s observations and the photographic record differ markedly from some of his later recollections. In addition, the monograph itself is sometimes ambiguous, as is evidenced by the disagreements mentioned above. It should become clear that the ambiguity and occasional lack of clarity in Tanner’s monograph have led many, myself and my collaborator included, into misinterpretations. We hope that this series of posts will shed more light and clear up some of the ambiguities.

As most readers already know, Tanner’s observations were restricted to one (and the same) family group each of the three breeding seasons during his study. While a sample size of essentially one family group would normally be a serious constraint for comparing with other information, it is important to point out this information represents the only detailed information we have on prey, foraging behavior, and breeding success for ivory-bills, keeping in mind this family successfully fledged young each of these three years. So the data and information Tanner reported on is directly relevant for understanding what was important for successfully fledging young under the conditions found at the Singer Tract during the late 1930s, but as Tanner himself pointed out in his monograph “…the conclusions drawn from them will not necessarily apply to the species as it once was nor to individuals living in other areas.”

Regarding the observations of others on the Singer Tract, I’ll begin with what may have been the last sighting of the John’s Bayou male. In August 1941, George Bick saw three ivorybills feeding in an ash flat near Sharkey Road, quite likely between the bridges over John’s and Methiglum Bayous, south and west of the John’s Bayou home range as delineated by Tanner. This is the only area along Sharkey Road that Tanner listed as “Ash Flat” on his 1941 map.

According to Bick, “I immediately stopped the car and noticed two Ivory-billed Woodpeckers perched in two small ash trees about eight inches in diameter, having recently killed tops. Only one of the birds was carefully observed. A bright, white bill, flaming red crest, and large white wing patch were all clearly noted as the bird remained at the tree. The second bird in a similar ash tree was observed less carefully . . . [A third bird] flew from a dying water-oak tree ten inches in diameter which had only a few curled brown leaves. A stripped spot about six by eight inches and about seventy feet from the ground was present on the trunk of this tree. This is thought to be a spot where the birds had been feeding and to represent the characteristic Ivory-bill ‘sign.’ In the immediate area were many ash trees with dead tops. Much of the bark was stripped in patches of varying size. This may possibly be old Ivory-bill feeding grounds.”

Logging had taken a significant toll in the Singer Tract by the time of Bick’s sighting. It’s thus possible that the birds were foraging in a suboptimal area due to logging pressure. Nonetheless, it’s still worth pointing out that Bick’s observations were in habitat and on tree species where Tanner observed virtually no foraging activity during his study (which ended two years prior, in 1939; he had no feeding observations on water oaks and only one on an ash). It’s also worth pointing out that Bick made specific reference to sweet gums (what he called “red gums”) as being abundant elsewhere but absent from this location.

My collaborator suggested that Bick’s inference that this ash flat was an “old Ivory-bill feeding ground[s]” is questionable. He suggested that changes in hydrology due to logging may have led to an ash die-off. He also noted that this was Bick’s only observation during his six month stay in the Tract, indicating that he was either not looking hard for ivorybills and/or that ivorybills were not using the ash flat on a regular basis. He added another caveat: it is important to remember that Bick’s observation was in August, well after the breeding season when even Tanner assumed foraging behavior for Ivory-billed Woodpecker likely expanded to different habitats and tree species than used during the time they were feeding young at John’s Bayou.

It’s interesting to note that the last known roost, where Don Eckelberry and young Billy and Bobby Fought famously said goodnight to a lone female ivorybill in April 1944, was apparently located in the ash flat where Bick saw his birds (W. Barrow pers. comm.). Just a few months earlier, in December-January 1943-’44, Richard Pough found a lone female roosting in the heart of the John’s Bayou range, about a mile north and east. According to Pough, who was convinced she was the last ivorybill in the Tract, this bird only crossed the Bayou once “for a brief visit to some trees a few hundred feet west of it . . . confining its activities to an area of hardly more than one quarter of a square mile, within which there were an unusually large number of dying trees.”

In our most recent conversation, my contributor and I touched on the question of whether Bick’s birds (and presumably the one seen by Eckelberry and the Foughts) were from the John’s Bayou family group. Either way, it’s a potentially interesting wrinkle. If the birds did come from John’s Bayou, this points to a heavier use of the ash flat for a period of years than is suggested by the limited information about the family group after 1939. All other observations – Pough, Peterson, Tanner, and Baker – were in the heart of the John’s Bayou home range, and at least one of those birds was reliably present there until shortly before Eckelberry and the Foughts said goodnight. On the other hand, if Bick’s birds were a different family group, it suggests that more ivorybills were in the Singer Tract in 1941 than is commonly assumed. (It’s worth repeating that Peterson wrote that one ivorybill was seen in December 1946, and the last letter to Tanner directly related to the Singer Tract birds says that game warden Gus Willett saw a pair in November 1948 and mentions other reports from around that time.)

To return to the Bick report: all of the trees seem to be in the smallest of Tanner’s size classes, 3-12″ in diameter. This class comprised 75.1% of the forest but was the source of only 12.7% of Tanner’s feeding observations. Tanner believed that ivorybills prefer larger trees because they “have more dead and dying wood” but his own data on this are ambiguous, and what he characterized as large seems problematic. The assumption about older trees having more dead and dying wood may have been true around John’s Bayou during Tanner’s study, but this is by no means always the case – the pine forests of Florida, for example, where Allen and Kellogg found abundant feeding sign on young dead pines, which are more vulnerable to fire than mature trees. And as pointed out in the previous post, even in the Singer Tract, the Mack’s Bayou home range was mostly second growth, so forest composition there must have been quite different.

There are a couple of ways to interpret this data. It’s true that 87% of the feeding was “on trees that are over a foot in diameter”, but this is somewhat misleading. 13-24″ diameter trees are the second smallest size class. They hardly qualify as large and approaching senescence, yet they account for 49% of Tanner’s feeding observations. It’s also true that, relative to abundance, the Singer Tract ivorybills showed a strong preference for trees in the 25-36″ class, but the abundance/observation ratios for 13-24″ trees and over 36″ trees are nearly equal, with a slight preference for the smaller size class not the largest. Thus, I think it’s equally accurate to characterize the data as showing that over 60% of observed ivorybill foraging was on smaller trees, under 24″ diameter at breast height and to reiterate that the most often used feeding trees were in the second largest size category, not the largest. (Tanner pp. 43-45).

On the other hand, there’s a good argument that the data show Ivory-billed Woodpeckers foraged on trees in the 13-24” class at 2.6 times the availability, in the 25-36” class at 6.7 times the availability, and in the 36” plus class 2.57 times the availability; there were very few trees in this size class, most of them sweet gums and a few Nuttall’s oaks (Tanner pp. 43-45). Contrast this with the 3-12” class, when the trees were 5.9 times more available than used.

A few additional points should be added to the mix. The numbers discussed above are aggregates, and size preferences were not at all evenly distributed among tree species. Fully 20% of Tanner’s total observations involved sweet gums in the 13-24” class, the most fed upon type. On sweet gums, frequency and abundance ratios are similar for the 13-24” and 25”-36” classes (the latter is the second most fed upon type, comprising around 18% of Tanner’s total observations). For Nuttall’s oak, 13-24” and 25-36” trees were approximately equal in abundance, but Tanner observed considerably more frequent feeding on the larger class.

My collaborator argues that it is more important is to recognize that when combining the data on sweet gums and Nuttall’s oaks, they collectively comprised 31.4% of the total forest and 79.3% of the foraging observations. Trees within the 25-36” class made up 31% and trees within the 13-24” class made up 29% of all foraging observations. Almost all of the trees in the 25”-36” class (5.2%) were in fact sweet gum or Nuttall’s oak, but for trees in the 13-24” class (18.3%) only about 5% (or about a third) were of these two species. This further highlights what Tanner described as heaviest use on sweet gum and Nuttall’s oak for the John’s Bayou family group over all other available trees, and a disproportionately high use of the second largest size class relative to abundance. However, this documented use pattern was not to the total exclusion of other tree species or even the smallest size class available.

This last was a point of contention. I took issue with aggregating sweet gums and Nuttall’s oaks, since they grow and mature at different rates. In addition, I think it’s important to highlight the fact that 13-24″ sweet gums were the single most fed upon type both in terms of frequency of observations and ratio of observations to abundance (albeit it by a small margin). As I see it, this undercuts the misinterpretation of Tanner that ivorybills are ‘large tree specialists’, a misinterpretation I think Tanner invited when he wrote, “The reason for Ivory-bills feeding on the bigger trees is that large, old trees have more dead and dying wood. Young trees grow rapidly and are resistant to the attacks of insects and disease.” As trees ‘mature’ their growth slows and becomes less vigorous, decay begins, insects attack them, and woodpeckers come after the insects.” (p. 43).

In light of this misconception, I also think it’s important to reiterate that in the aggregate, the over 36″ size did not show anything near the disproportionately high use of the 25-36″class. In fact, the rate was very slightly higher on the 13-24″ trees.

Regardless of how one interprets this very limited data set, the idea that Ivory-billed Woodpeckers required ‘large trees’ for foraging has become a truism. The reality is considerably more complex.

The next installment will focus primarily on decay class, and the final one will look at prey species. Stay tuned.

Advertisements

A Brief Update on an Old Image

I’ve been playing around with Topaz photo editing software. I’m a total neophyte and did not track all the steps I took to modify this image from 2009, discussed here, but I think that what I did eliminated some artifacts, specifically the apparent red in the crest of the bird. These pictures were taken in August, so any red in the leaves is clearly an artifact. Note that some red remains despite the processing.

IMG_1096-Edit-Edit-Edit-Edit-Edit-Edit-Edit-Edit-EditTopaz processed neck bird
I’m also attaching the original Reconyx image and the one taken a minute before, so that if people with more skills want to work with the originals, they can. My personal view is that if this were a Pileated Woodpecker, there should be obvious red in one or both of these images, and that red should be in a different place than in the unaltered version. There may also be something of interest in the cavity just below the fork in the background snag.

IMG_1095

IMG_1096

We’ll be back in the field from December 26th-January 3rd.


Habitat Conditions in the Singer Tract

Late last year, I wrote a post entitled “More Minutiae – Habitat Quality and Population Density in the Singer Tract”. I had to follow up with a couple of corrections and elaborations based on insights others shared with me. In the interest of providing more clarity and coherence, I thought I’d do a new piece combining the three posts and expanding on them a bit. I won’t delete the originals, but this one reflects what I think is a more accurate understanding of the material involved.

Mack's Bayou Ivorybill nest tree. Courtesy of the Division of Rare and Manuscript Collections, Cornell University Library

Mack’s Bayou nest tree. Courtesy of the Division of Rare and Manuscript Collections, Cornell University Library

The initial post was inspired by the image above and the caption describing it as the “Third ivorybills’ nest”, one I had looked at but not closely until last year. When I did examine it carefully, I was struck by how open the surrounding area seemed to be. Then I started going through archival photographs and scrutinizing them a little more closely.

In browsing through the Louisiana Digital Library’s collection of Singer Tract photographs, I came across an image that I had missed, one of the bridge over John’s Bayou taken in 1940. Tanner (p.32) includes an ivorybill sighting from this immediate vicinity, just northwest of the bridge. What I find interesting about the photograph is that the forest along the road appears to be fairly even-aged and does not have the characteristics typically associated with old growth. It is similar to what can be found in many parts of Louisiana today. An image from along Sharkey Road taken in 1937 shows similar characteristics, although another shot from 1939 (probably taken east of the bridge) shows more impressive looking habitat.

Richard Pough wrote a follow-up report to the Audubon Society on the Singer Tract in 1944. It’s a very interesting document that raises some questions about Tanner’s work. Pough explicitly accepted “Tanner’s premises as to the feeding habits and habitat preferences of the ivory-bill”, but he also noted “[n]othing in Mr. Tanner’s study indicates that Ivory-billed Woodpeckers show any preference or marked dependence on trees of great size such as one would find only in a virgin forest. He found them doing 14% of their feeding on trees 3”-12” in diameter and 64% on trees under 24” in diameter.”

Pough pointed out that “Maps of the area as of 1846 showed much of the Tensas River in plantations and many cleared fields back from the river on some of the interior ridges. This development continued for another 20 years until the Civil War, by which time Madison Parish was producing 110,000 bales of cotton a year. As the Parish has never produced over 30,000 bales since the Civil War, one gets some idea of how much land is now occupied by second growth forest of approximately 80 years age.”

Pough found only one ivorybill, a female. He assumed, relying on Tanner, that this was the only one left in the Tract. He may well have been mistaken, since Gus Willett, game warden in the Tract, wrote Tanner about seeing a pair in November 1948 (although the exact location is unclear). Correction, Willett did not write the letter, although the report reached Tanner, as discussed here.

The lone bird Pough saw was either the John’s Bayou female or one of its offspring. According to Pough, this bird was probably not feeding in virgin forest, and his report specifically suggests that Tanner might have been mistaken about the maturity of some of the habitat in the John’s Bayou area. In 1941, Tanner had written that the remaining John’s Bayou birds were roosting and feeding in “virgin” timber. Pough’s description of this area (and it seems to be the same patch) suggests it was likely cultivated pre-Civil War. It was devoid of big sweet gums, which Pough deemed to be the best indicator of old growth conditions, but had many dying Nuttall oaks 12-20 inches in diameter. Nonetheless, Pough relied on  Tanner’s premises to conclude that “only a relatively small portion of the total area of the Singer Tract supported a forest suitable as habitat for these birds.”

To return to the material in the earlier blog posts, much of the discussion focused on home ranges and the distribution of nest sites.

This map, drawn by Tanner after the 1935 expedition, lists three nests – designated as nests II, III, and IV –within a mile or two of each other and in the vicinity of John’s Bayou. Nest II is the famous “Elm Rock” nest. The map also shows a tree which is designated “Nest (?) III Squirrel” (also mentioned in Bales, “two miles to the south of the first nest” and fifty feet up p. 45). This was outside the area Tanner designated as prime in 1941 (p. 91); it is approximately a half-mile from the John’s Bayou bridge.

I’ve discussed this issue in depth with someone who’s very familiar with Tanner’s notes. I’m now persuaded Tanner concluded that nests III and IV from 1935 were not nests after all and that he assigned the birds involved to Titepaper (Nest III) and Bayou Despair (Nest IV). Nest IV is apparently one that Kuhn found but was unable to re-locate. Why Tanner changed his mind about it remains a mystery.

It’s very difficult to piece together this fragmentary information, and the monograph muddies the waters a bit by presenting the home ranges of the birds as being quite discreet, perhaps a good deal more than they were in fact. I suspect that Tanner decided the cavities were actually roosts, (although neither one is mentioned in the monograph). If so, they would have been well outside the home ranges Tanner identified and closer to the core of the John’s Bayou range than to the core of Titepaper or Bayou Despair.

Nests I and V were located near Mack’s Bayou. Nest I is dated May 14, ’34 on the map, and Nest V is dated May 10, ’35. The ’35 nest failed. It is the one referenced above. Allen and Kellogg described it as being 45’ feet up, in a pin oak snag, in a natural clearing, although it has been suggested that the snag may have been a remnant large tree in area that had been cleared prior to the Civil War. The nest designated Nest I and dated May 14, 1934 appears to be the one Tanner described on p. 81 of the monograph, “located within 100 yards of the second nest found in 1935”; however, in the monograph, he gave the date as May 13, 1933.

It’s worth pointing out that the Mack’s Bayou nests were in an area that Tanner designated as “best” for ivory bills (even if it’s not clear whether it was truly old growth). Nonetheless, nests failed in 1933 and 1935, and the adult birds had disappeared by 1938, apparently after producing one fledgling in 1936 or 1937. This was before the logging began.

My intention in writing those initial posts was to get a clearer handle on population densities and habitat requirements in the Singer Tract. In retrospect, I’m not sure that’s possible, since Tanner’s observations were almost entirely limited to one family of birds in a population that was dwindling for unknown reasons. At the very least, Tanner’s statement that 7 pairs of birds required 120 square miles of virgin forest in 1934 is based on an inflated estimate of the amount of old growth in the Tract and his minimum estimate of 6.25 square miles per pair also rests on that flawed premise.

Pough observed, “ . . . the ivorybill problem puzzles me exceedingly, and I do not feel that Tanner’s report begins to explain the reasons for the drastic decline in this species.” As the 2014-2015 search season approaches, I can only hope that the question of how the species persisted will puzzle people exceedingly in future years.


Cavities and Context

As many readers of this blog are aware, I’ve been actively searching for ivorybills since 2007 and have been obsessed with foraging sign and cavities since my first days in the field. Over the years, I’ve looked at bark scaling and cavities both in and out of suspected ivorybill territory and have developed and refined a hypothesis about what constitutes diagnostic ivorybill foraging sign. I have come to believe that no such diagnostic category exists for nest or roost cavities, although size, shape, and other contextual elements can support a suspicion that a given cavity is an ivorybill nest or roost.

The cavity cluster we found in May 2014 has characteristics that are encouraging – size, shape, tree species, bark scaling in the vicinity – and one of the cavities resembles a known Ivory-billed Woodpecker nest. This is more than ample reason to survey the surrounding area very carefully (and I’m eager to get back out there), but whereas I feel personally confident that the dramatically scaled hardwoods in our search area were fed on by IBWOs, the mere presence of intriguing cavities is nowhere near as compelling for me.

Similarly, while I suspect that Frank Wiley found a recent nest and at least a couple of roost holes in our old search area, this view is informed more by the associated conditions than by the qualities of the cavities themselves. The suspected nest was in a tree that was heavily scaled and excavated in a way that was strikingly similar to a nest tree found in the Singer Tract, and this is more important than the size and shape of the cavity itself. The two suspected roost holes were in close proximity to one another, and were certainly large enough, but were dramatically different in appearance. There are several facts that are more significant than size or shape. We had a camera trap that covered both trees for a period of several months, and no Pileated Woodpeckers were photographed using the trees in question; the only image we obtained was suggestive of ivorybill but inconclusive (no images of either species were obtained from the suspected former nest tree, even though there were active PIWO roost cavities within 20 or 30 yards); in addition, we found an abundance of feeding sign in the area and had multiple auditory encounters and possible sightings within a few hundred yards of the suspected roosts between August 2009 and April 2010.

oldareacavity2

cavity 2 reportWillow Cavities, East-Central Louisiana 2009. Photos by Frank Wiley.

In examining the images of the suspected roosts and comparing them with other images of woodpecker cavities, I came across a photograph of a Crimson-crested Woodpecker (Campephilus melaneoleucos) near a cavity and was struck by the fact that it was almost identical in appearance to one of our suspected roost holes. I did some further research and found an image of a Pileated Woodpecker nest that was quite similar, if not quite as close a match.

All of this leads me to think that, while it’s important to look for suggestive cavities, this should not be the top priority. I’d encourage other searchers to focus on feeding sign, habitat characteristics, and local reports above all. The limited available information suggests that feeding sign is what led Allen and Kellogg to the Singer Tract birds and that it’s what Kuhn and Tanner looked for. Pileateds will use big cavities (and I’ve even seen White-breasted Nuthatches nesting in holes that look perfect for ivory bills). While finding a nesting or roosting ground will be central to documenting the ivorybill if it persists, cavities alone will not point the way.


Cavity Comparison

In late 2011, I visited the Museum of Natural History at Harvard and took some iPhone photos of the Ivory-billed Woodpecker nest that’s on display there. In reviewing the pictures of cavities I took last week in Louisiana, I was struck by the similarity in appearance, especially between the Harvard cavity and the cavity in the stack, the most significant difference is in orientation; one angles left and the other right. I’ve cropped and zoomed the Louisiana cavity, while boosting the contrast and details, and I’ve flipped the Harvard cavity so the orientation is the same.  ImageImage

 

The Louisiana cavity could, of course, have been gnawed by squirrels, but the similarity is intriguing.

Edited to add: while I believe there is a type of feeding sign that is diagnostic for IBWO, cavities are another matter. In my view, ivorybill-ish cavities only become interesting when associated with other indicators, especially feeding sign.
 


More Minutiae – Habitat Quality and Population Density in the Singer Tract

Last week I linked to this image and pointed out that the caption described it as the “Third ivorybills’ nest”. That discovery sent me back to the online archives. What I found raises some interesting questions about the Singer Tract, the population density of ivorybills, at least around John’s Bayou in 1935, and about habitat quality in general.

Most ivorybill aficionados are familiar with images from the Singer Tract. I think most of us know the ones that show dramatic, ‘primeval’ forest, but I suspect we are all prone to overlooking those showing less impressive habitat (just as I am prone to taking photographs of the biggest trees and the most dramatically mature forest), even if we are familiar with the debate about how much actual old growth existed in the Tract in the 1930s. In browsing through the Louisiana Digital Library, I came across an image that I had missed, one of the bridge over John’s Bayou taken in 1940. Tanner (p.32) includes an ivorybill sighting from this immediate vicinity, just northwest of the bridge. What I find interesting about this photograph is that the forest along the road appears to be fairly even-aged and does not have the characteristics typically associated with old growth. It is similar to what can be found in many parts of Louisiana today. An image from along Sharkey Road taken in 1937 shows similar characteristics, although a 1939 shot shows more mature looking habitat and was probably taken to the east of the others.

What makes this even more interesting is that the “Third ivorybills’ nest” appears to have been found to the south of Sharkey Road in an area that was either outside or on the very edge of what Tanner considered to be one of the “Best areas for Ivory-bills” (p. 91). This map shows the location of the nest tree, which is designated “Tree III Squirrel”. (also mentioned in Bales p. 45, “two miles to the south of the first nest” and fifty feet up p. 45.) The image of the snag itself suggests  it might have been outside of prime habitat, since the cavity was 45′ (caption), 47′ 8″ (Tanner) or 50′ (Bales) up, and the snag is considerably taller than the surrounding trees.

On Edit: The oak snag could have been a super-dominant tree, which would account for the height difference and be consistent with the location being in mature forest. It’s also possible that the tree was an older tree that had been left behind. A close look at the two maps suggests it was outside the area Tanner designated as prime in 1941; it is approximately a half-mile from the John’s Bayou bridge.

On further Edit: Some of my interpretations discussed in this post have proven to be incorrect. Go here and here for clarification.

Things grow even more puzzling since what appears to be the “Elm Rock” nest, the famous one, is listed as “Nest II” on the 1935 map, and there is no “Nest I” (Nest I was near Mack’s Bayou. See following post for correction.) There is also a “Nest IV” shown. This is somewhat closer to the “Elm Rock” nest and about 2/3 of the way toward Methiglum Bayou.

In his monograph, Tanner apparently recognized nests II and III but not nest IV, and why there is no reference to a nest I remains a mystery. It seems that 1935 was a bad year for ivorybills, and there’s nothing to suggest that any of these three or four nests succeeded. It’s possible that there never was a “Nest I” and that Tanner later decided that “Nest IV” was not a nest after all. Nevertheless, it appears that at least two pairs of ivorybills attempted to nest in the John’s Bayou area in 1935 and that one of those nests was in or near the border of what Tanner would have deemed to be inferior habitat.

This raises many questions and may have some implications in terms of how the IBWO might have been able to persist. At the very least, it calls into question Tanner’s estimate that “six and a quarter square miles” per pair is “probably close to the maximum density.”