A New Flightless Gallinule (Aves: Rallidae: Gallinula) from the Oligo-Miocene of Riversleigh, Northwestern Queensland, Australia

Flightlessness in birds occurs in a taxonomically diverse array of families, but is best exemplified in the rails (Rallidae). Most flightless species of rails live on islands, where the absence of native mammalian predators may make flight superfluous. Fossil rails from Oligo-Miocene sites at Riversleigh, northwestern Queensland, Australia, are considered to represent a single species of gallinule Gallinula, described here as new. Compared with four Quaternary species of Gallinula from Australasia (two volant, two non-volant), it shows similarities with the flightless species in the development of the foreand hindlimb elements and in other characteristics of limb bone morphology associated with flightlessness. These indicate that the Riversleigh species was non-volant. Its relationships with the Quaternary species, including the flightless Gallinula mortierii, now restricted to Tasmania, but known from Plio-Pleistocene deposits in eastern mainland Australia, are considered. BOLES, WALTER E., 2005. A new flightless gallinule (Aves: Rallidae: Gallinula) from the Oligo-Miocene of Riversleigh, northwestern Queensland, Australia. Records of the Australian Museum 57(2): 179–190. Records of the Australian Museum (2005) Vol. 57: 179–190. ISSN 0067-1975 www.amonline.net.au/pdf/publications/1441_complete.pdf Flightlessness in birds occurs in a taxonomically diverse array of families. Flightlessness per se probably conveys no adaptive value; instead, it is an consequence of morphological changes that convey other selective advantages to the bird (Livezey & Humphrey, 1986). The general consensus is that the loss of volancy results as the flight muscles and pectoral assemblage become reduced as energy-saving processes during ontogeny (Olson, 1973a). Such modifications are advantageous when they do not render the birds more susceptible to predation. The presence of flightless birds on islands is strongly correlated with the absence of predators, particularly mammalian ones. Other factors can moderate this relationship, such as the availability of sufficient cover to avoid or reduce predation, allowing birds to exist in the presence of predators, and the stability of resources, removing the need for far-ranging mobility (dispersal) (Worthy, 1988). The developmental mechanisms involved in the loss of flight have been explored in greatest depth in the rails (Rallidae) (Olson, 1973a), the family that best exemplifies the phenomenon. About a fourth of the world’s 125 or so living or recently extinct species have lost the power of flight. Most, but not all, of these are populations on islands, where the absence of native mammalian predators has reduced the benefit of the ability to fly. Those species that have become extinct usually did so at least in part from the inability to cope with the introduction of exotic predators. Prominent among the few instances of flightlessness in rails 180 Records of the Australian Museum (2005) Vol. 57 on larger landmasses are the three species of native-hens of Australia and New Zealand (Gallinula, subgenus Tribonyx) (Fig. 1). The Black-tailed Native-hen G. ventralis, a volant species, is widespread through mainland Australia except for the east coast and far tropical north, but does not occur in Tasmania, to where the much larger and flightless Tasmanian Native-hen G. mortierii is restricted. The extinct G. hodgenorum of New Zealand, well represented in subfossil deposits, was also flightless (Olson, 1975a). The living volant Dusky Moorhen G. (G.) tenebrosa is common in both Australia and New Zealand. The remains of rails have been recovered from several Oligo-Miocene sites at Riversleigh, northwestern Queensland (Fig. 1). These are considered to represent a single species, a new gallinule of the genus Gallinula. Although only one skeletal element is represented by an intact specimen, there is adequate material of the wings, legs and coracoid to indicate that this form was flightless. Fig. 1. Distribution of Recent and fossil Australian species of Gallinula. Key: black shading, current distribution of G. mortierii; grey shading, current distribution of G. ventralis; Riversleigh (G. disneyi); Quaternary mainland sites producing G. mortierii (taken from De Vis 1888, 1892; Baird 1984, 1985, 1986, 1991b, 1992; McNamara & Baird 1991; Olson 1975b; and this work). Materials and methods Measurements were made with digital callipers and rounded to the nearest 0.1 mm. Length measurements of bones of G. hodgenorum were taken from specimens and from Olson (1975a); all other measurements were taken from specimens. Weight, wing chord and tarsus measurements of living species of Gallinula were taken from Marchant & Higgins (1993) and represent means for adult males. Osteological nomenclature follows Baumel & Witmer (1993), except that as terms of position and direction anterior is used rather than cranial and posterior rather than caudal. Olson (1973b) placed Tribonyx as a subgenus of the gallinules Gallinula; this was accepted by Condon (1975) and Christidis & Boles (1994), and is followed here. Ratio-diagrams of the log differences between measurements of compared taxa were constructed following the method of Simpson (1941), wherein measurements are converted to logarithms, and one taxon is arbitrarily chosen as a standard. The difference between its converted measurements and the corresponding ones for each taxon are calculated (the logarithms of the ratios). The standard taxon thus has all ratios of 0 (zero difference in logarithms), which when plotted along a vertical axis on arithmetic graph scale, form a straight line. The logarithmic ratios for each taxon are plotted such that the points on a single horizontal line represent different values of the same variable across the taxa. Those values larger than the standard fall to the right of the standard line, the smaller ones to the left of it. The points of each taxon are connected with a line. Taxa with proportions identical to those of the standard taxon will have lines parallel to that of the standard. Variations from a parallel line are indicative of variations in the proportions from that of the standard taxon. Because only one of the fossil elements is complete, measurements of other features of the bones were used in lieu of total lengths because these should also reflect the changes in overall sizes of the elements. For the purposes of these comparisons, the values used were the means in Table 1 unless otherwise indicated. Geology and geographical setting The fossils described in this study were collected from the Riversleigh deposits, which are located 5 km west of the Riversleigh homestead (19°02'S 138°45'E), 200 km north of Mt Isa, northwestern Queensland, where they occur as an outcrop of Tertiary limestone overlying the Cambrian Thorntonia Limestone. There are now over 200 named Oligo-Miocene sites at Riversleigh. An informal system of grouping has been used (Systems A–C). These systems are “regionally clustered sites that appear to be superpositionally-related (differing in age but not significantly in position) and/or space-related (spatially isolated but approximately contemporaneous)” (Archer et al., 1989). The principal accumulations are thought to have occurred in several episodes involving large lakes, shallow pools and cave deposits. Rail material has been recovered from six sites, ranging from Late Oligocene to Middle Miocene in age. Current understanding of the age of these sites is taken from Creaser (1997); other birds represented at these sites are taken from Boles (1995, 1997). White Hunter Site, Hal’s Hill Sequence, D-Site Plateau, considered to be part of System A (Late Oligocene), has yielded the greatest diversity of rail specimens. The White Hunter Local Fauna also contains other birds, including the small casuariid Emuarius gidju (Patterson & Rich, 1987) (Boles, 1992), the dromornithid Barawertornis tedfordi Rich, 1979, a stork (Boles, 2005) and several passerines. LSO Site (LSO Local Fauna), from the Verdon Creek Sequence, in the northern section of the D-Site Plateau, is also regarded as System A. Another site from the D-Site Plateau, but considered part of System B (Early Miocene), is Camel Sputum Site, Godthelp Hill Sequence (Camel Sputum Local Fauna). In addition to rails, it has also provided Emuarius, Barawertornis, another new genus and species of dromornithid, a swift Collocalia buday (Boles, 2001) and several passerines. Creaser’s Ramparts Boles: Riversleigh flightless gallinule 181 Site and Dirks Towers Site are in the central and northern sections of the D-Site Plateau, respectively. Both occur at about the same level and are possibly correlated. The ages are still unclear, but may be System A or B. Other birds recovered from Dirks Towers Site are Emuarius and passerines. Ringtail Site forms part of the Ray’s Amphitheatre Sequence on Gag Plateau. This site is included in System C, considered to be of Middle Miocene age. The Ringtail Local Fauna includes waterfowl and a number of passerines. Systematic palaeontology Order Gruiformes Bonaparte, 1854 Family Rallidae Rafinesque, 1815 The fossils are referred to the Rallidae and subordinate taxa on the following suites of characters (adapted in part from Baird, 1992; Gilbert et al., 1981; McCoy, 1963; Olsen, 1979; and Worthy, 1997). Coracoid. The tuberculum brachiale is not undercut. The processus procoracoideus is pronounced, extending further medially than the processus acrocoracoideus. The cotyla scapularis is large and deep. The facies articularis humeralis is round, about as wide as long, and flares strongly laterally. The impressio m. sternocoracoidei is deep, extending far anteriorly. Humerus. The incisura capitalis is deep and forms a shallow angle with the main axis of the shaft. The fossa pneumotricipitalis is shallow. The long axis of the caput humeri is roughly parallel with that of the incisura capitis. The sulcus ligamentosus transversus is shallow. The crista deltopectoralis is high, triangular and directed anteriorly. The processus supracondylaris dorsalis is small and blunt. The distal end of the element is narrow, not markedly produced laterally or medially. The processus flexorius extends further than the condylus ventralis. The fossa m. brachialis is shallow. Carpometacarpus. The processus extensorius tends slightly proximoventrally (in anterior view). The dorsal rim of the trochlea carpalis extends far proximally and is acute at its proximalmost point. The fossa infratrochlearis is distinct, deep and circular. The os metacarpale minus is slightly curved. The facies articularis digitalis major and minor extend distally to the same extent. Femur. The crista trochanteris is low but strongly developed proximolaterally and curves medially; there are strong ridges extending distally from its distal edge onto the anterolateral face of the shaft, where it joins the linea intermuscularis cranialis, and from near the anteriormost projection to the anterior border of the facies articularis antitrochanteris. The impressio m. iliotrochantericus caudalis is restricted to the extreme proximal end of the trochanter femoris. The collum trochanteris is distinctly narrowed both anteriorly and posteriorly. The shaft curves to meet the caput femoris in a broad, gentle curve (in anterior view). The proximal half of the shaft has a distinctive posterior inflection (in lateral view). The linea intermuscularis cranialis extends far distally from the ventral side of the crista trochanteris. The sulcus intercondylaris is moderately shallow. The fossa poplitea is shallow. Tibiotarsus. The cristae cnemialis are strongly developed. The crista cnemialis lateralis is flattened proximodistally. The crista fibularis is well developed. The pons supratendineus is well developed. The condylus lateralis is much broader than the condylus medialis. The incisura intercondylaris is narrow and displaced medially by a broad condylus lateralis. The facies lateralis of condylus lateralis is rounded. Tarsometatarsus. The hypotarsus is an elongated triangle (in proximal view). The crista lateralis hypotarsi is prominent, extending distally. There are two sulci hypotarsi (usually one or both enclosed), and a proximal pons tendineus on the medial side of the dorsal face. The dorsal surface of the shaft is flat, not concave. The trochlea metatarsi IV is shorter than the trochlea metatarsi III. Another useful character, which cannot be assessed on the fossil, is that the trochlea metatarsi II is much shorter than the trochlea metatarsi IV and is recessed plantarly. Genus Gallinula Brisson, 1760 Gallinula Brisson, 1760, Ornithologia sive Synopsis Methodica vol. 1, Paris: Ad Ripam Augustinorum [50], vol. 6[2]—type species: Gallinula Brisson = Fulica chloropus Linnaeus, 1758. Tribonyx has been distinguished from Gallinula by the shorter, heavier toes, longer tails, absence of white on the undertail coverts (Olson, 1973a) and short, wide bills with very short premaxilla (Olson, 1975a), all characters that cannot be assessed from the available fossil material. Osteological comparisons of a “typical” gallinule, G. tenebrosa, and two species of native-hens Gallinula (Tribonyx), G. ventralis and G. mortierii, found few useful characters that might permit separation of skeletal elements to subgeneric level. In the humerus of Gallinula (Gallinula), the proximal end of the condylus dorsalis extends over the condylus ventralis. Brodkorb (1967) used this character to distinguish Gallinula (s.s.) from Fulica, in which the ventral extent of the condylus dorsalis is less, just reaching the dorsal border of the condylus ventralis. The species of Gallinula (Tribonyx) are somewhat intermediate between typical Gallinula and Fulica in this character, with a shorter, but still overlapping ventral extent of the condylus ventralis. The condylus dorsalis on the one distal humeral fossil fragment is abraded and the state of this character is equivocal. In the tarsometatarsus of Gallinula (s.s.), the lateral border of the shaft is about even with that of the trochlea metatarsi IV and they thus join smoothly with little lateral flaring. In contrast, species of Gallinula (Tribonyx) have the trochlea metatarsi IV curving outwards laterally, away from the border of the shaft. This character cannot be evaluated for the fossil material. Until skull material is recovered, it cannot be ascertained with certainty whether the Riversleigh flightless rail was indeed a native-hen. This taxon is here diagnosed only as Gallinula. In the following discussion, however, extensive reference and comparisons are made to native-hens Gallinula (Tribonyx) because these are the only gallinules in Australia and New Zealand in which flightlessness occurs and the only ones well represented in the fossil record. Gallinula (s.l.) can be diagnosed on the material available by the following suite of characters. 182 Records of the Australian Museum (2005) Vol. 57 Coracoid. The processus procoracoideus extends about half way along the shaft posteriorly, joining it gradually. The impressio m. sternocoracoidei is deeply excavated mediodistally, and has a rounded border to the facies articularis sternalis. Humerus. The crista bicipitalis extends distally only 1⁄2–2⁄3 as far as the crista deltopectoralis. The border of the epicondylus ventralis is concave (in anterior view). The proximal end of the condylus dorsalis extends over the proximal end of the condylus ventralis. Carpometacarpus. The processus alularis is broad (in proximal view). The os metacarpale minus is somewhat curved proximally, less so distally, but overall more than in Fulica. The distoposterior corner of the symphysis metacarpalis distalis is obliquely angular, rather than square. Femur. The junction of the impressiones obturatoriae and trochanter femoris forms an acute angle of c. 50°. The trochanter femoris is deep, flaring from the shaft both anteriorly and posteriorly (in lateral view). There is general agreement in the size and location of the impressiones iliotrochanteria. The condylus lateralis is well produced (in lateral view). The sulcus intercondylaris is situated near the midline of the element. The condylus medialis is moderately robust (in posterior view) and well produced posteriorly (in medial view). Tibiotarsus. The impressio lig. collateralis medialis is deep. The crista fibularis is short. There is a deep U-shaped notch on the posterolateral margin of the area interarticularis. The crista cnemialis lateralis extends to a point well proximal to the proximal end of the crista fibularis. The crista cnemialis cranialis continues as a crest along the anterior medial edge of the shaft. The fossa retropatellaris is moderately deep. The anterior surface of the proximal end is slightly convex. The condylus lateralis extends far proximally, overlapping the distal 1⁄3–1⁄2 of the pons supratendineus. Tarsometatarsus. The hypotarsus extends distally, rather than being truncate (in posterior and lateral views). The shaft has roughly parallel sides throughout its length. The foramen vasculare distale is situated far distally. Other useful characters, which cannot be assessed on the fossil, are that the dorsal margin of the trochlea metatarsi II is roughly even with the plantar margin of the trochlea metatarsi III; the area proximal to the trochlea metatarsi III, medial to the foramen vasculare distale and lateral to the trochlea metatarsi II, is broad with parallel sides; and the incisura intertrochlearis lateralis is wide. Gallinula disneyi n.sp.