Unresolved name

Atlantisia elpenor Olson

Comprehensive Description

provided by Smithsonian Contributions to Zoology
Atlantisia elpenor

HOLOTYPE.—Complete left tarsometatarsus, fully ossified (Plate 5e,f,g). Vertebrate paleontological collections of the National Museum of Natural History (USNM 170129). Collected in depression on the east side of the southernmost chamber of the largest fumarole located about 260 meters true north of the summit of Sisters Peak, Ascension Island, South Atlantic Ocean (approximately 7°55′25″S; 14°22′15″W), on 17 June 1970, by Storrs L. Olson and Jesse M. Couch. Directly associated with the type were right and left humeri, ulnae, radii, carpometacarpi, coracoids, scapulae, fibulae, left femur, sternum, furcula, mandible, left quadrate, 13 ribs, 18 vertebrae, and 15 pedal phalanges which have been cataloged under the same number, as I am confident that they are all from the same individual. As this specimen was well exposed and in an excellent state of preservation and had the mandible present but the rest of the skull lacking, it is possible that the skull (Ashmole No. F21) picked up by Ashmole and now in the British Museum (Natural History) may be from the same individual. However, lest there possibly be any confusion, and since there were so many other rail remains in the area, I designate only the tarsometatarsus as the holotype. The associated elements may be considered as especially significant paratypes in the large series of paratypical material from the two fumaroles.

The type tarsus is 38.0 mm in total length; transverse breadth across head 5.5 mm; transverse breadth across center of shaft 2.5 mm; transverse breadth across trochleae 5.7 mm. It is near the maximum length and stoutness for the species and is yellowish white in color. There being so much more material of tarsi and other skeletal elements, I will not further belabor the description of the actual type.

PARATYPES.—All of the various elements of rallid material from the two fumaroles included in the series USNM 169773–170128, 170130–170272, and 175803–175853 may be considered paratypes.

RANGE.—Ascension Island, South Atlantic Ocean.

STATUS.—Extinct; exterminated sometime after 1656.

ETYMOLOGY.—Elpenor was one of Odysseus’ crew members, who, while stranded on Circe’s island, fell from the roof of her palace, was killed, and descended straight to Hades. Odysseus later encountered Elpenor’s shade during his journey through the Underworld. The Ascension rail was also stranded on an island and upon falling off the lip of the fumarole, descended straight to the bowels of the earth and was not known again until its shades were stirred up by inquiring mortals invading its underworld tomb. The specific name stands in apposition to that of the genus and is in the nominative case.

DIAGNOSIS.—A medium-sized flightless rail, roughly the size of a Virginia Rail (Rallus limicola) but with much reduced pectoral girdle and wings, and stouter hind limbs. It is very similar in most respects to Atlantisia rogersi but much larger, the major bones averaging 25 to 35 percent longer. On the other hand, A. elpenor is but half the size of A. podarces (Table 2).

DESCRIPTION.—The skull of A. elpenor (Plate 1b) is represented by several nearly complete crania and cranial fragments, fragments of bill, and one specimen that is nearly complete, with bill, palatal bones, and cranium intact. In this specimen (USNM 170131) the interorbital bridge had been corroded away and when first collected the bill and palate were connected to the cranium by only a thread of the parasphenoid rostrum which with subsequent handling parted. The specimen is superior to the skull collected by Ashmole (1963a, pl. 10), which I have not examined, in that the anterior portions including the palatines, vomer, maxillopalatines, portions of the jugal bar, etc., remain attached. By approximating the two portions of the specimen, I got an overall measurement of 51.6 mm length, which is essentially identical to the 51 mm Ashmole estimated for the British Museum specimen. The interorbital bridge is quite narrow; 7 specimens range from 2.70 mm to 3.25 mm (average 2.97) in width at the narrowest point. The bill is like that of A. rogersi (Plate 1a). The nasal bar is narrow and twisted dorsally in A. elpenor and is broader and flat in A. rogersi. The palate is essentially like that of A. rogersi save that the posterior portions of the palatines are slightly more expanded and squared. The scars for the attachment of M. pseudotemporalis superficialis and M. adductor mandibulae are larger and deeper in A. elpenor, indicating a greater development of the jaw muscles than in A. rogersi.

The lacrimal in rails is a small delicate bone and is not fused to the skull. Lowe (1928:123) drew attention to the very distinctive nature of the lacrimal of A. rogersi and I therefore felt myself most fortunate to obtain three lacrimals of A. elpenor for comparison. These in no way resemble A. rogersi but are more like the typical rallid form (Plate 2e). In A. rogersi the descending process is long, slender, and twisted, the head of the bone is reduced to a pointed, anteriorly projecting spur, and the horizontal process is a short hook. In contrast, in A. elpenor there is a broad square head with a deep depression on the medial surface and a well-developed area for the attachment to the nasal; the descending process is fairly short and the horizontal process is long, strong, and tapering. Lowe (1928:124) considered the peculiar lacrimal of A. rogersi to represent a remarkable degree of “generalization”; however, I cannot agree with this determination. The condition of the lacrimal of A. rogersi is much more likely some sort of specialization. It is not a retention of a generalized embryonic condition in the Rallidae, as none of the embryos or chicks of rails I have seen possess such a lacrimal. The lacrimal does not provide attachment for muscles, but numerous ligaments do attach to it and in rails it may provide a brace for the jugal bar and protection for the anterior orbital region (Cracraft, 1968). This bone in rails is quite variable in shape as I have previously noted (Olson, 1970), but the adaptative significance of such variations is unknown.

The mandible of A. elpenor (Plate 2c,d) is very like that of A. rogersi, although in the latter the symphysis seems slightly wider and more troughlike.

The furcula (Plate 3h) is slender and in the shape of a narrow U, with the epicleideum represented by only a slight swelling.

The coracoid (Plate 3f,g) like all of the pectoral elements, is reduced in size. The shaft is proportionately wider than in A. rogersi, not thickened and bent as in A. podarces. As in A. rogersi, the brachial tuberosity is in line with the shaft. The sterno-coracoidal impression is considerably shallower than in A. rogersi and the scapular facet is also shallow.

In the scapula of A. elpenor (Plate 3i) the tubercle for the attachment of the dorsal branch of the tendon of M. expansor secondariorum is completely lacking, whereas in A. rogersi a reduced but distinct tubercle is still present.

The sternum of A. elpenor differs greatly from that of A. rogersi (Plate 4). In A. elpenor it is broad and Hat, the carina being reduced to a thickened ridge about 1.5 mm at its greatest depth. Thus, literally speaking, the species barely qualifies as a “carinate” bird and was without question flightless. The carina of A. rogersi, in contrast, is elevated posteriorly. In A. rogersi there is a deep, rounded notch widely separating the coracoidal sulci and completely obliterating the spina externa. This notch is much shallower in A. elpenor although there is still no trace of the spina externa. The sterno-coracoidal processes of A. elpenor are fairly broad and short as in typical rails, not tenuous and elongate as seen in A. rogersi and also in the large, flightless Weka (Gallirallus) of New Zealand. In A. elpenor, the posterior lateral processes are short, scarcely extending beyond the xiphium, and the sternal notches are correspondingly reduced, being confined to the posterior fourth or less of the sternum. The xiphium of A. elpenor is notched, unlike A. rogersi but like Gallirallus. From the anterior end of the carina of A. elpenor, two well-defined ridges extend forward to the internal margin of the ventral lip of the coracoidal sulci, but in A. rogersi the anterior half of the sternum is much more conspicuously flattened. Overall, the sternum of A. elpenor is quite similar to that of Gallirallus, except that the latter has slender, elongate sterno-coracoidal processes like A. rogersi. This similarity is purely convergent as the sterna of flightless rails assume a number of shapes that are not necessarily correlated with phylogenetic affinity (see below).

The humerus (Plate 3a,b) is not as robust as that of A. podarces. The bicipital crest is better developed and shaft slightly more curved than in A. rogersi. The ectepicondylar prominence and external tuberosity are further reduced than in either A. rogersi or A. podarces.

In the ulna (Plate 3c) the external condyle is slightly better developed than in A. rogersi. A distinct depression just proximal to the internal face of the external condyle of A. elpenor is not apparent in the other two species of Atlantisia.

Compared to A. rogersi, the carpometacarpus of A. elpenor (Plate 3d,e) has the proximal metacarpal symphysis somewhat longer and narrower, the distal portion of metacarpal III not as thin, and the external side of metacarpal II slightly narrower and flatter.

In dorsal view of the pelvis (Plate 2a,b) the portion of the posterior iliac crest above the antitrochanter flares rather abruptly outward, giving the area between the crests a squared appearance in A. elpenor. These crests in A. rogersi are less prominent and spread apart gradually and farther behind the antitrochanters, which gives the area between the crests a triangular shape. The anteriormost part of the ilium of A. elpenor forms a sharp, laterally directed point which is less developed and rounder in A. rogersi.

Except for size, the hind limb elements of the three species of Atlantisia are quite similar (Plate 10). The shaft of the femur (Plate 5c,d) of A. elpenor, in lateral view, is more curved than in A. rogersi. Proportionately, the shaft of the tibia (Plate 5a,b) is narrower and the fibular crest is shorter in A. elpenor than in A. podarces. The fibula (Plate 5a,b) of A. elpenor is a long slender splint with the head less stout than A. podarces. None of the 50 appropriate specimens of tarsometatarsus (Plate 5e,f,g) of A. elpenor has the ossified tendinal loop for the tendon of M. extensor digitorum longus on the proximal anterior face. In A. rogersi this loop is mostly cartilaginous and the same may be assumed for A. elpenor; however, this loop is fully ossified in A. podarces. The scar for the hallux is deep in A. elpenor and is longer than that of A. podarces. The internal trochlea of A. elpenor is noticeably less flared medially than in A. podarces.

The toes of A. elpenor, as far as can be determined from assembling the disarticulated phalanges, are of about the same proportions as A. rogersi. The claws are not nearly as long and well developed as those of A. podarces.

In plotting the lengths of the humerus, femur, tibia, and tarsus on a graph, no bimodality is apparent except possibly in the tarsal measurements, so it cannot be conclusively demonstrated that A. elpenor was sexually dimorphic in size. The amount of variation in size (Table 1), may nevertheless be attributable in part to sexual differences, as in most rails the males are larger than the females.

Of the external characteristics of A. elpenor there is only Mundy’s description to go on. From this we may assume the bird had a more variegated plumage than the drab-colored A. rogersi, although A. rogersi does retain some variegation. The red eye noted by Mundy corresponds well with the red eye of adult A. rogersi, although this is a character that appears at random throughout the Rallidae and is without taxonomic importance.

Environment and Extinction of the Ascension Rail

Ascension is a relatively small island (97 km2; 38 mi2), entirely volcanic in origin. Geological reports emphasize its extreme youth (e.g., Daly, 1925) and the oldest rocks have now been dated at only 1.5 ± .2 million years of age (Chace and Manning, 1972:5). Since the discovery of the island in 1501, there has been no volcanic activity reported there, which is considered odd in view of the very fresh appearance of some of the lava formations. The highest point on the island is the peak of Green Mountain (860 m; 2817 ft). The coastline consists mainly of steep rocky cliffs with a few scattered patches of beach composed of shell sand or pebbles. The climate is mild, temperatures ranging from about 26° to 29°C in the lowlands and 6° to 6.5°C lower on Green Mountain. Rainfall on Green Mountain, which is often enshrouded in mist and clouds, averages about 708.6 mm (27.9 in) annually, but in the lowlands, where near-desert conditions prevail, there is only about 127 mm (5 in) per year. Strong trade winds blow with great consistency from the southeast. There is very little seasonal variation in weather but at rare intervals heavy rainstorms occur which may destroy the nests and eggs of Sooty Terns (Huckle, 1924) and may have affected rails. These storms are usually followed by a marked burst in growth of vegetation. Unusually heavy rainfall preceded my second visit to Ascension by several weeks. When I arrived, most of the low areas of the island were covered with rippling waves of grass where before there had been only cinders. Within six weeks this grass had dessicated and turned yellow after having produced a prodigious amount of seeds to await the next rains.

Duffey (1964) summarized the history of the terrestrial fauna and flora of the island (although he failed to cite Mundy’s journal (Temple and Anstey, ed., 1936), which contains several valuable points of information aside from the reference to the rail). Man has drastically altered the environment, probably from his first encounter with the island. In Mundy’s account, which is earlier than any cited by Duffey, the only vegetation described as present in 1656 was “rushes” and “grass” on the higher mountains. Goats were already present and Mundy is probably correct in assuming that they were released years earlier by the Portuguese discoverers of the island. The impressions of the earliest botanists to visit the island largely confirm Mundy’s report. Only two shrubs are known to be native to Ascension, both of which are endemic (Euphorbia origanoides and Hedyotis adscensionis, the latter apparently extinct). A few other flowering plants and grasses, as well as a number of cryptogams, including several endemic ferns, are native to the island, most of these being confined to the Green Mountain area. The disproportion of cryptogams to phanerogams indicates an early stage of floral colonization of Ascension somewhat similar to that reported for the Pliocene of St. Helena (Muir and Baker, 1968). Osbeck (1771, II:98) described “several petrifactions of branches of trees and pieces of wood” and was under the impression that the island had formerly been forested. Murphy (1936:154, 865) assumed that Ascension was forested at one time, mostly on the basis of Osbeck’s testimony. There is no other evidence, however, that this was the case. Palynological studies by Duffey (1964) showed no evidence of pollen from indigenous trees and he discounts the idea that the island ever possessed forest cover. Very likely Osbeck mistook some of the deceptively woodlike lava formations about the island for actual remains of trees. There is very little area of suitable soil for forest growth on the island and much of what is now present resulted from the mulch of introduced plants and from direct importation of soil.

The impoverished flora of the island has been greatly augmented by man’s introductions, most of which, after the establishment of a permanent British garrison in 1815, were intentional. The invertebrate fauna is likewise depauperate and the presence of many species may well be the result of having been unintentionally introduced along with plants. There are now no native terrestrial vertebrates (the only one known was Atlantisia elpenor). Of the several species of birds that were released on the island, only four have persisted, and these are dependent on the man-made vegetation landscape. They are a francolin (Pternistes afer), a myna (Acridotheres tristis), and two finches (Estrilda astrild, Serinus flaviventris).

Pelagic seabirds that used the island for nesting and roosting were once present in tremendous numbers on the main island of Ascension as all the early accounts testify (Temple and Anstey, ed., 1936; La Caille, 1763). Huge colonies of boobies (Sula spp.), Sooty Terns (Sterna fuscata) and frigatebirds (Fregata aquila) were reported, mostly from the northwestern and southern portions of the island (for an account of the history of the seabird colonies of Ascension, see Stonehouse, 1962). Because of introduced predators, all but the Sooty Terns and a few Gygis abandoned the main island and retreated to Boatswainbird Islet, a 96–meter (315 ft) high trachytic monolith, roughly 4000 m2 in area, lying about 250 meters off the eastern side of the main island. Most of the island’s remaining populations of tropicbirds (Phaethon aethereus, P. lepturus), noddy terns (Anous stolidus, A. tenuirostris), Fairy Terns (Gygis alba), and petrels (Oceanodroma castro) are also found on this crowded refuge and adjacent cliffs. Only large deposits of guano and dung-whitewashed rocks, especially in the area between Sisters Peak and English Bay, remain as mute testimony of the former great abundance of the seafowl of Ascension. In the vicinity of the fumaroles may be found guano and numerous rocks whitened on their leeward side (Figure 3a), and in crevices in the rough lava, skeletons and even feather shafts of boobies (Sula dactylatra) are occasionally encountered.

How did Atlantisia elpenor survive in its seemingly hostile environment? What role did it play in the island’s relatively simple ecosystem? Upon what did it feed? The island’s poor flora could hardly have provided enough, if any, suitable seeds or fruit; nor, presumably, was there ever much of an invertebrate fauna associated with this flora to provide it with sustenance. It is possible that some birds may have been able to glean a little food from the few beaches available for foraging, but these very small areas could hardly sustain a viable population of any bird. Furthermore, the only indications of the bird have come from the interior of the island. Mundy’s (Temple and Anstey, ed., 1936) account implies that their birds were taken up on the mountain. The fumaroles are about 3 km from the nearest beach, over rough country—a long trip for a flightless bird. The locality south of South Gannet Hill, where Ashmole (1963a) collected a tarsus of A. elpenor, is about 1 km from the very small beach at Mars Bay and well over 2 km from the nearest suitable beach at Southwest Bay. The species must have been fairly numerous, as evidenced by the abundant remains in the fumaroles and the fact that Mundy’s men could catch a half-dozen individuals in one afternoon’s foray.

It is obvious that there must have been a good many rails in the neighborhood of the fumarole and that they were finding food in the vicinity. The most probable explanation is that A. elpenor was a scavenger in the large colonies of seabirds. During the process of feeding, the young and adults of boobies and terns often drop food. Thus, the rails would have had an ample supply of pelagic fish, squid, and crustacea available to be picked up, or perhaps even stolen directly, not to mention an abundance of seabird eggs, which were no doubt taken as well. In the carcasses of birds on Boatswainbird Islet, I found numerous dermestid beetles which are probably an original element in the fauna and these could well have provided an additional food source for the rails. With the bird remains in both fumaroles were numbers of dipterous pupae cases which would indicate that flies and maggots were abundant. The Laysan Island Rail (Porzanula palmeri), which lived in a situation somewhat similar to that of A. elpenor, was observed to eat the eggs of seabirds and maggots from carcasses (Rothschild, 1893). The Spotless Crake (Porzana tabuensis) has been seen to eat petrel eggs in the Kermadec Islands (Soper, 1969), so there is ample precedent among rails for using seabird colonies as food sources. In a different order, adults and young of sheathbills (Chionis) during the breeding season subsist almost entirely on krill regurgitated by penguins (Jones, 1963) and seem particularly well adapted to this semi-parasitic existence. As there is seabird nesting activity the year around on Ascension (Stonehouse, 1960:213), the rail would have been able to reap an abundant harvest throughout the year in its commensal role.

Judging from the old reports and the guano that still remains, the seabird population of Ascension was very large, and one might assume a correspondingly large population of rails. Predation would certainly not have been a limiting factor under natural circumstances. The only possible predator of A. elpenor would have been the frigatebird Fregata aquila, a species that does prey to a certain extent on the chicks of Sooty Terns. The agility and speed of rail chicks and adults, plus their relative scarcity in comparison to more easily obtained food would certainly have precluded their capture by frigatebirds on any but the most exceptional of conditions.

Unfortunately, we know nothing of the nesting of A. elpenor. The only typically suitable nesting cover would have been on Green Mountain and adjacent vegetated areas. The birds perhaps could also have nested in small sheltered crannies in the lava flows of the lowlands, using the sparse grass and possibly feathers to construct a nest. Again, Chionis provides a fine example, as it nests in rocky holes and lines its nest with feathers, eggshells, and bones of penguins (Jones, 1963).

One factor that is more difficult of explanation is the seeming ability of the rails to cope with the almost complete lack of fresh water on the island. The only more or less permanent sources of water are on Green Mountain. The nearest of these to the fumaroles is a small seepage area (Dampier’s Drip), the result of an accumulation of water above an impermeable, compact layer of iron oxide (Darwin, 1844), which is 3.2 km away. South Gannet Hill is even farther from a source of fresh water. Rain does not collect in any appreciable amounts in the porous rocks and soil, and rainfall in the area of the fumaroles, for instance, is very slight anyway. Atlantisia elpenor probably ingested regurgitated invertebrates such as squid, which are isotonic to seawater, making some mechanism of salt excretion a necessity. In the salt-marsh dwelling rail Rallus longirostris, the interorbital bridge is narrowed to accommodate hypertrophied supraorbital glands which function in the excretion of salt. The much narrowed interorbital bridge of A. elpenor is a good indication that its supraorbital glands were equally well developed. The birds, therefore, probably derived their water solely from their food.

With the coming of man, Atlantisia elpenor was a doomed species. As mentioned, goats were an early introduction to the island, although they probably had little effect on the avifauna. At some unknown time, rats and mice made their appearance. They were present by 1754 (La Caille, 1763) and Osbeck (1771, II:84) felt that they came to the island with the wreck of William Dampier’s ship Roebuck in 1701, although they may well have reached shore previously with vessels stopping for sea turtles or other supplies. The rats at least must soon have made inroads on the population of ground-nesting rails and probably some of the seabirds as well. Kepler (1968) has documented the grisly effects of direct predation by Rattus exulans on seabirds in the Pacific.

Just how long after 1656 the Ascension rail survived is not known. If present, its numbers must have been considerably reduced by the 1700s as the species was never reported again. If any persisted into the 19th century they could never have survived the coup de grace administered by the British who introduced cats for rodent control soon after 1815. By 1823, feral cats were in abundance and by the 1860s this vicious scourge had eliminated all but the Sooty Terns from breeding in significant numbers on the main island (Stonehouse, 1962:111–113). Thus, the rail and its source of food were both destroyed.

Vagrant Rallids

In recent times, three species of rallids have found their way across hundreds of miles of open ocean to Ascension. The first of these to be reported was a gallinule “killed with a stick near the summit of the island” in July 1836 (Darwin, 1841:134) and which was made the type of a new species “Porphyrio simplex” (Gould, in Darwin, 1841). Sharpe (1894) synonymized this with Porphyrula parva (=P. flavirostris), an uncommon South American species, which has not otherwise been recorded as an ocean-crossing vagrant. Gould’s description is of an immature bird and does not tally with the appearance of P. flavirostris while the measurements given are too small for P. martinica. Therefore, I think that it is highly likely that this specimen represents the African species P. alleni. An immature specimen of Porphyrula alleni was taken on Ascension on 27 May 1920 and reported by Lowe (1924). I have examined this specimen and confirmed its identity.

I have reported and discussed two unexpected instances of the occurrence of the American Purple Gallinule (Porphyrula martinica), both immatures, on Ascension in June in 1970 and 1971 (Olson, 1972).

Stonehouse (1960, 1962) records the capture of a juvenile Common Gallinule (Gallinula chloropus) at Mars Bay during the BOU expedition. Unfortunately, the bird was not made into a specimen and later escaped, eliminating the possibility of a subspecific determination. In view of the fact that both American and African species of Porphyrula have occurred on Ascension, it would seem that Gallinula chloropus, which is native to both continents, could come from either direction. Stonehouse (1960:153) mentions a farmhand on Ascension who had seen other gallinules in previous years, and my own discussions with islanders imply that gallinules (probably of all three species) are of fairly regular occurrence there. The significance of these records is discussed later.


Beneath the upper and central green circle, the wild valleys are quite desolate and untenanted.

Darwin, 1846

Darwin’s impression of St. Helena is as true today as it was in 1836, but at one time the desolate valleys were well tenanted with multitudes of birds. The island of St. Helena has undergone a drastic alteration of its fauna and flora since its discovery by man in 1502. At present there are only two species of landbirds (an endemic plover, Charadrius sanctaehelenae, and a gallinule) not introduced by man, and only a handful of seabird species remain. Such was not always the case.

The BOU Ascension expedition also visited St. Helena and while there collected a small series of bird fossils from several sites on the island. These and other collections, which were reported on by Ashmole (1963b), indicated that there were many other species of birds on the island in the past. The deposits from which these fossils came range in age from Pleistocene to recent. In 1971 I visited these same localities, discovered a few other sites, and collected thousands of fossil specimens, mostly of seabirds. All of these specimens are now in the National Museum of Natural History, Smithsonian Institution. These sites and most of the specimens will be described in detail in a later paper covering the entire former avifauna of the island. Here I will concern myself only with the rails. An assortment of fossils from Prosperous Bay and Dry Gut collected by Arthur Loveridge and forwarded to the Museum of Comparative Zoology also contains a few bones of rails and these are included in the following discussions.

Redescription and Relationships of “Aphanocrex” podarces

From the extensive sediment deposits at Prosperous Bay, the BOU expedition collected several elements of a very large rail. These were subsequently examined by Wetmore (1963) and described as a new genus and species, Aphanocrex podarces. The material available to Wetmore consisted of a left tarsometatarsus (the type), one right and one left femur, a left tibiotarsus, a proximal right ulna, a distal left humerus, and a partial pelvis.

My own collections contain a number of specimens of this species from four localities in all parts of the island. From the same deposits on the east bank of the gorge at Prosperous Bay came a right and left tarsus, proximal and distal left tibiae, a right femur, a cervical vertebra, and a pedal phalanx. In addition, on the west bank I unearthed much of a skeleton of a single individual associated with the bones of a frigatebird (Fregata sp.). This specimen (USNM 175878) is represented by a right femur, right and left tibiae, the head of a fibula, right and left tarsi, seven pedal phalanges (including two claws), right and left humeri (both lacking distal ends), right ulna (lacking proximal end), right and left carpometacarpi, digit II phalanx 1 of the wing, right and left coracoids (both lacking the sternal ends), 3 cervical vertebrae, 2 fragments of pelvis, and a fragment that appears to be part of the posterolateral process of the sternum. In the MCZ collections are a right and a left femur, right ulna, proximal right tibia, and a fragmentary right tarsus. The Prosperous Bay specimens do not appear to be mineralized and are probably of no great age. Including the material examined by Wetmore, there are at least four, and very likely more, individuals represented from this site.

From the eroded washes of Dry Gut, near Bencoolen, I collected the following material of the large rail: two left femora and a proximal right lemur, a proximal right tibia, three proximal right tarsi, a fragment of a distal left tarsus, a distal left humerus and a proximal left coracoid. There is a distal right tibia in the MCZ collections. Some of these specimens are quite well mineralized and are of obviously greater age than those from Prosperous Bay. Not less than three individuals are represented from this site.

Near Sugarloaf Hill are aeolian deposits of calcarious sand that are much dissected by erosion. These are thought to be Pleistocene in age (Baker, 1970) and contain fairly abundant bird remains. Here I collected a single, well-mineralized and nearly complete right tarsus of A. podarces.

Ashmole (1963b) has described all of the above localities but listed remains of A. podarces only from Prosperous Bay. However, a worn proximal end of a right tibia (BMNH S/1963.25.24) from the Sugarloaf site which Ashmole (1963b:405) thought “may represent another species of columbid,” on examination proves to be from A. podarces, and in a lot of unidentified specimens collected from this site by the BOU, I found a distal left ulna of this species.

I found more bones of this species at one additional site that was discovered quite by accident. While walking along the well-used trail next to a small embankment immediately south of the old battery ruins in Sandy Bay Valley, I saw an avian tarsus which because of its size, location near houses, and new appearance, I assumed to be of a barnyard fowl or a pheasant. Picking it up out of force of habit, I was more than pleasantly surprised to find that it was a complete left tarsus of A. podarces. Further searching on this low hillside disclosed the bones of other bird species and a left carpometacarpus, the shaft of a tibia, a right fibula, a cervical vertebra, a pedal phalanx, and a right quadrate of the large rail. By far the best find, however, was a complete cranium of this species. It unfortunately is fractured off at the naso-frontal hinge so that none of the bill or palate is present, but the remainder is nearly perfect, the parasphenoid rostrum, interorbital septum and ectethmoids being retained (Plate 1c). All of the specimens from this site appear very recent and are possibly from one individual.

With so much better material now available, it is possible to make more extensive comparisons. Wetmore (1963) in his analysis of the species compared it with Aramides and certain other large rails, for the purposes of description only. He did not suggest any relationships of the St. Helena bird other than to subfamily (Rallinae). A relative exists, however, but in all fairness one would hardly expect to look for the closest living relative of one of the largest species of the family in one of the smallest of the family that is also flightless and on an even more remote island over 1500 miles to the south. But with the perspective offered by the nearly intermediate-sized A. elpenor, the similarity of A. podarces to A. rogersi becomes apparent. These three species are so alike in most proportions (Table 3; Plate 10) and in most details of the skeletal structure and so recognizably different from related mainland genera, that in my opinion they must be regarded as congeners. Aphanocrex Wetmore then falls into synonomy with Atlantisia Lowe and its single species will hereafter be referred to as Atlantisia podarces.

The generic characters of Atlantisia I have set forth in Appendix 2. Here I will attempt to redescribe the species A. podarces, especially in relation to the other two members of the genus.

Atlantisia podarces is a large rail, among the largest of the family, but considerably more slender and less robust than is usual in the large flightless forms such as Gallirallus or Nesotrochis. In linear measurements it is double the size of A. elpenor and nearly thrice that of A. rogersi (Table 2) but is similar in proportions except that it appears to be slightly though consistently stouter, with somewhat longer wing elements (Table 3). The head, however, does not increase by the same proportion as body size, and, as in other birds, the larger species has a proportionately smaller head. Inasmuch as volume increases as the cube of linear dimensions, the bulk of A. podarces must have been in the neighborhood of 8 times that of A. elpenor and 15–20 times that of A. rogersi.

The cranium (Plate 1c) of A. podarces is massive, about the size of Aramides cajanea and heavier than, though not as elongate as, Gallirallus australis. The interorbital bridge is wide (10.7 mm) and exceedingly thickened and inflated, being approached only by the ponderous Tribonyx mortierii in this respect. In comparison with A. elpenor, this wide bridge certainly indicates a much lesser development of the salt glands which reflects the abundant supply of fresh water on St. Helena. Apart from the interorbital bridge and the great disparity in size, the cranium of A. podarces is otherwise nearly duplicated on a smaller scale in that of A. elpenor. The great similarity in the shape of the scar for M. pseudotemporalis superficialis is especially striking. In A. podarces the cranium is more domed postorbitally than A. elpenor. The fossa for M. adductor mandibulae is deeper and the zygomatic area is enlarged, both, along with the reinforced interorbital bridge, probably correlated with a heavier bill and more massive jaw musculature.

The quadrate of A. podarces is similar to that of A. elpenor save that the shaft of the otic process is wider.

The coracoid (Plate 6d,e) is quite large and heavy. Its head is much more massive than in A. elpenor or A. rogersi and the brachial tuberosity is turned internally and is not in a line with the shaft. The scapular facet is deep and cuplike in contrast to A. elpenor and A. rogersi. The stout shaft is bent and swollen at the midpoint, appearing almost pathological; but since both the right and the left coracoids of the same individual appear identical, this is probably the normal condition. From what remains, the sternocoracoidal impression seems to have been quite deep, more so than A. rogersi and unlike A. elpenor in which this impression is shallow.

The sternum of A. podarces is unknown except for most of a right posterior lateral process with an attached piece of sternal plate. This is wide and thickened and would seem to indicate a large, heavy sternum.

The head of the humerus (Plate 6f,g) is more bulbous, with the capital groove less transverse and closer to being parallel with the shaft than in A. rogersi and A. elpenor. The ligamental furrow is much deeper than in the latter two species. (See Wetmore [1963, pl. 9] for an illustration of the distal humerus). The ulna (Plates 6h,i; 10) of A. podarces is heavier than in A. elpenor or A. rogersi, with a larger internal condyle and deeper tendinal pit on the external condyle; the distal radial depression is not as deep as in A. elpenor. The carpometacarpus (Plates 6j,k; 10) is like that of A. elpenor but proportionately longer, with a longer, more tapered distal symphysis.

The femur (Plate 10; Wetmore, 1963, pl. 9) of A. podarces differs from A. elpenor and A. rogersi mainly in having the ligamental attachment above the fibular condyle rounder and deeper, with a prominent ridge around it. The fibula (Plate 6c) is essentially like that of A. elpenor but heavier and thicker, In A. podarces, the tibia (Plates 6b,c; 10) is very similar to A. rogersi and A. elpenor except that in dorsal view, the outer cnemial crest is thicker and curved more posteriorly, and the shaft is somewhat stouter. The tarsus (Plate 10; Wetmore, 1963, pl. 9) is of nearly identical proportions to A. elpenor. It differs from that species and from A. rogersi in that the loop for the tendon of M. extensor digitorum longus is completely ossified, the posteriormost ridge of the hypotarsus is broad, flat, and more prominent, the scar for the attachment of the hallux is deeper but shorter and more rounded, and the internal trochlea is flared out farther medially.

The toes (Plate 6a) are long and strong (not as heavy as Gallirallus) with exceedingly long and sharp claws (longer than Tribonyx mortierii, Gallirallus australis, or any other rail I have seen). One claw core, which I assume by its size to be from the middle toe, is fairly straight and measures 18.5 mm in length. With its sheath it must have been well over 22 mm long. It composes 25.6 percent of the total length of the middle toe (with claw) whereas the claw of A. rogersi constitutes only 15.6 percent of the total length of the middle toe.

Of only the tarsus is there enough material to give some indication of individual size variation. These measurements are as follows: length 70.1, 72.4, 74.0; width of head 10.7, 10.7, 11.0, 11.4, 11.5, 11.6; width distal end 10.6, 11.4, 11.8, 12.2. Here, and in Tables 2 and 3, I have incorporated the measurements given by Wetmore (1963). Though there is little in these measurements to indicate the extent of variability in size, visual comparison of some of the elements (for instance the proximal tarsi from Dry Gut) shows a discernible difference in individual dimensions and this was likely correlated with sex.

A Second Species of Extinct Rail

From three of the four major collecting localities on St. Helena I uncovered numerous remains of a second species of flightless rail, much smaller than A. podarces. Considering the relative abundance of this species in these deposits, it is somewhat surprising that it had not been discovered before, but Ashmole (1963b) mentions no such bird.

Prosperous Bay (deposits on the east side of gorge): In addition to the nearly complete skeleton which I have designated as the type, the following elements were found: 3 right (one is a shaft only) and 1 left femora, 2 complete and 1 distal right tibiae, 1 distal left tibia, 3 right and 1 left tarsi, 1 left humerus, 1 distal mandible, 1 rostrum and the tip of another. Another individual, from the west bank of the gorge, is represented by a right humerus, distal mandible, and the left posterior lateral process of the sternum. In the MCZ collections are the proximal and distal ends of a left tibia and a distal right tibia. At least five individuals are included in the Prosperous Bay collections.

Dry Gut: One complete and 1 proximal right femur, 1 complete and 2 distal left femora, 2 distal and 2 proximal left tibiae, 2 distal and 2 proximal right tibiae, 1 complete and 2 distal left tarsi, 1 proximal right tarsus, 1 radius, 1 right carpometacarpus, and 1 left lower jaw articulation. In the MCZ collections is a complete right femur. A minimum of four individuals from this site.

Sandy Bay Valley: An articulated right tibia and tarsus, the latter with ossified tendons associated (later lost); 1 right and 1 left femur, 2 complete and 1 distal left tibiae, 1 proximal right tibia, 2 right and 1 left humeri (the latter from a juvenile). These remains represent not less than three and probably at least four individuals.

This small rail is a short-billed crake-like form. As the other landbirds of St. Helena of known derivation are African, and the prevailing winds on St. Helena are from the southeast, it is logical to expect the small rail also to be of African origin. However, the American species Porphyrula martinica has now been taken at least twice on St. Helena (see below). The small St. Helena rail does not agree with any of the species examined of the American genus Laterallus, as these all have markedly longer and more slender hindlimbs and pelves. It in no way agrees with Sarothrura, being heavier throughout, longer billed, and having a different palatal structure among other things. It is much smaller than Aenigmatolimnas, Crecopsis, Crex, and Limnocorax and does not have the peculiar bill shape or long legs of Aenigmatolimnas, nor the narrow, twisted nasal bar of Crecopsis and Crex, nor any of the hindlimb modifications of Limnocorax. Of African “crakes,” this leaves only the genus Porzana, with which it quite agrees in bill shape, palate, pelvis, and hindlimb structure. It represents an undescribed species.
bibliographic citation
Olson, Storrs L. 1973. "Evolution of the rails of the South Atlantic islands (Aves: Rallidae)." Smithsonian Contributions to Zoology. 1-53. https://doi.org/10.5479/si.00810282.152