Discussion: Paleoneurology of Dromornithids
Ken Ashwell
Limitations of the present study
The fossils used in the present study were excavated from two central Australian formations: the Middle Miocene Bullock Creek local fauna, Camfield station, for Dromornis planei and the Late Miocene Alcoota local fauna for Dromornis stirtoni. The specimens from the Bullock Creek fauna are relatively intact and undistorted, but Alcoota formation sediments had been subjected to repeated hydration and desiccation, so the specimens from that group show some signs of fragmentation of the bone (Worthy et al., 2015). Nevertheless, the distortion of the Dromornis stirtoni specimens is minimal (see Fig. 4A, B) and the endocranial shape is as would be expected for an anseriform (Fig. 4F). Post-mortem distortion is therefore unlikely to have affected the conclusions of the present study.
The endocranial cavities and osseous labyrinths of all the dromornithid fossils were mostly filled with matrix, so the boundaries between cavity and bone had to be determined by differences in density of matrix compared to bone. Nevertheless this was easy to discern for the endocranial cavity (Fig. 4A to D), but only possible for the osseous labyrinth of one dromornithid (Fig. 4E).
In the present study comparisons have been made between the cranial features of the dromornithid fossils and those of modern anseriformes and other birds. ECV could be measured for most of the modern avian specimens used for these comparisons or derived from that stated in the literature (Iwaniuk and Nelson, 2003), but data on body weight of individual museum specimens were usually not available. Where a body weight needed to be assigned to a species, the value for that species as stated in the literature (e.g. Iwaniuk and Nelson, 2003) was used.
Diet and behaviour of dromornithids
The Australian continent underwent significant cooling and drying during the Tertiary and, by the Late Miocene/Early Pliocene, Central Australia was dominated by a mosaic of dry sclerophyll open woodlands and fire-adapted grasslands (reviewed in Murray and Vickers-Rich, 2004). The laterally compressed beaks of the dromornithids with robust and sharp toma appear to be adapted for a diet of tough, woody vegetation and hard seeds (conifers, cycads, palms; Murray and Vickers-Rich, 2004). The present findings on the paleoneurology of the dromornithids suggest that in some respects these huge birds were anseriformes “writ large”, but also that adaptation to the dry environment and hard vegetation of Miocene Central Australia may have led to specializations in the nervous system of the giant birds.
Olfactory bulb size could only be measured in two dromornithid skulls (NTM-P9973-6, NTM-P3249). The findings suggest that olfactory bulb size as a proportion of ECV was quite low, but statistical comparison with other large birds (e.g. ratites) was not possible. Nevertheless, the findings suggest that olfactory specialization was not strong in dromornithids.
The findings concerning orbital size suggest that dromornithids had eyes of a size no greater than would be expected for their brain size, but that these eyes were more recessed within the orbit than in modern anseriformes. This recession may be a reflection of the overall robusticity of the skull, i.e. the large plates forming the roof of the orbit may have achieved that size to bolster the mechanical rigidity of the skull, rather than to protect the eyes. The optic canal area of Dromornis planei was found to be much smaller than would be expected for an anseriform of that body weight and brain size, suggesting that optic nerve cross-sectional area may have actually been quite small. This in turn suggests that retinal area and ocular globe size were certainly unremarkable, and perhaps even small, for an anseriform of such prodigious size. The ocular convergence angles for both species of dromornithid are small and comparable to the modern domestic duck. The Wulst area of the brain is also unremarkable. Taken together, these findings suggest that ocular convergence and stereopsis were minimal in the dromornithids (in fact at the lower end of the avian range) and therefore stereoptic depth perception would have been rudimentary. Selection and targeting of food would most likely have been made by one eye acting independently, with the eye of a given side being used to guide the head so that a seed-pod was brought into the space between the edges of the beak of that side for shearing. These deductions on the visual function of these birds are not consistent with the suggestion that they were carnivores (Wroe, 1999), because they would have lacked the necessary ability to visually target mobile prey by stereopsis.
Oculomotor canal area in one of the dromornithid species was found to be small for its body weight, but appropriate for its brain size. This suggests that eye movement control was not particularly specialized in that bird and is consistent with an ocular size no larger than expected for brain size.
Trigeminal function was assessed from the size of the maxillomandibular canal in the dromornithid fossils. This canal transmits both the maxillary and mandibular divisions of the trigeminal nerve (V2, V3), which serve somatosensory function for the upper and lower beak, respectively, as well as motor axons for control of the muscles of mastication (in the mandibular division). The size of the canal compared to the expected area for an anseriform of that body or brain size was analysed in the present study. Actual areas of this canal in both species fell within the predicted ranges for an anseriform of that body and brain weight, suggesting that the degree of trigeminal sensory specialization in the dromornithids was no greater than would be expected for an anseriform of their size. Nevertheless, the beaks of these birds were prodigious (up to 0.5 m long; Murray and Vickers-Rich, 2004) and the canal large in absolute terms, so the number of sensory axons distributed to the beak could have been very high. A maxillomandibular canal of 25 mm2 could easily accommodate a million myelinated sensory axons of up to 5 µm external diameter.
The semicircular canal loop diameters in the dromornithid fossil in which this could be measured (NTM-9973-6 Dromornis planei: horizontal semicircular canal HSCC – 5.9 mm; anterior semicircular canal ASCC – 6.6 mm; posterior semicircular canal PSCC – 7.6 mm) were remarkably small compared to those we have found previously in much smaller bodied accipitrids (e.g. the 18 kg Harpagornis moorei: HSCC – 9.8 to 11.5 mm; ASCC – 8.6 to 10.8 mm; PSCC – 9.0 to 9.6 mm; Scofield and Ashwell, 2009). The log10 square root of ASCC area calculated for NTM-9973-6 is only 0.77, much smaller than that found for non-avian dinosaurs of a similar body size (1.0 and above; Georgi et al., 2013). This suggests that the sensory apparatus for detection of angular acceleration was poorly developed in Dromornis planei, perhaps not surprising for a heavy flightless bird with no predators.
The hypoglossal canal was assessed in the dromornithids to test the hypothesis that the tongue musculature would have been enlarged to enable the manipulation of the hard seed-pods that are the posited diet of these birds. Evidence was found to support this hypothesis, in that the measured areas of the hypoglossal canal/s in Dromornis planei were more than 6 times larger than would have been expected for an anseriform of that brain and body size. This would be consistent with enlarged lingual musculature and/or dense innervation with fine motor control.
Finally, foramen magnum area in the dromornithids was measured and compared with a range of anseriformes. The aim was to determine if foramen magnum area scaled in proportion to body or brain size in a manner predicted from modern anseriformes, or whether it exhibited deviation from predicted size, such as would suggest spinal cord hypertrophy or reduction. It was found that foramen magnum area was actually significantly smaller than expected for anseriformes of the body sizes of both species. On the other hand, the observed foramen magnum areas did agree well with those predicted on the basis of ECV for both species. In other words, spinal cord cross-sectional area for the dromornithids appears to have been appropriate for brain size, but not body size. The spinal cord serves somatosensory, viscerosensory, somatic motor and autonomic functions so it is difficult to interpret this observation precisely. It may be that the density of postcranial somatosensory and somatic motor innervation was lower than for other anseriformes, perhaps reflecting less discriminative somatosensation and/or larger motor units, but this cannot be definitively proven. Nevertheless the small relative size of the foramen magnum stands in contrast to the substantial maxillomandibular canal for trigeminal somatosensation from the beak and the enlarged hypoglossal canal/s for lingual control.
Why are the brains of big birds smaller than the brains of big mammals?
The ECV of Dromornis stirtoni was measured at 126 ml, which makes this the largest endocranial volume ever reported for a bird. Nevertheless this is much smaller than ECV or brain size reported for large mammals (up to 9,000 ml; van Dongen, 1998). Why are the brains of large birds so much smaller than the brains of large mammals, when avian and mammalian encephalization is similar at lower body weights (see Fig. 3B)?
One explanation for this may be that those birds that reach very large body size are drawn from avian groups that are not highly encephalized (i.e. anseriformes and ratites). Therefore one could not expect those large birds to have particularly large brains. In support of this idea, the present study found that the ECV of the dromornithids agreed closely with what would be expected for an anseriform scaled to their body size (Fig. 5A). This naturally begs the question of why highly encephalized avian groups do not develop large body size. Large body size has benefits for herbivores in that it allows the carriage of a large digestive tract, but may not have been of benefit to carnivorous predatory birds, which are usually the most encephalized birds (Iwaniuk and Nelson, 2003).
One could also argue that birds do not reach large brain size in part because they do not reach particularly large body size (Deeming and Birchard, 2008). Of six lineages of large flightless birds that emerged in the Late Cretaceous, estimated maximal body mass has probably only risen above 500 kg in Dromornis stirtoni and Aepyornis maximus and most large birds weigh no more than 250 to 300 kg (Deeming and Birchard, 2008). This stands in striking contrast with the large sizes regularly achieved by non-avian theropods. One of the key reproductive features of birds is the use of contact incubation, whereby eggs are incubated by direct contact with an adult body, usually the male (Birchard and Deeming 2009). This allows good control of incubation temperatures, but requires that the egg be able to withstand the weight of an adult. Even though the incubating parent is usually of lower body weight (hence the reversed sexual dimorphism of large ratites), this requirement for direct contact between adult and egg places significant mechanical constraints on the eggshell. Eggshell must be thick enough to withstand breakage, but thin enough to allow diffusion of gases during incubation and permit successful hatching (Birchard and Deeming, 2009). Limitations on egg size naturally place limitations on size at hatching, and hence fix the starting body weight for the post-hatching life of birds at a much lower level than that for the newborn young of large placental mammals or large non-avian theropods.
An alternative explanation may be that avian brains do not have the capacity to reach the large sizes that mammalian brains do. There may be limitations arising from nutritional, vascular or intrinsic proliferative factors during development that restrict the ability of the avian brain to expand beyond an adult size of 120 ml. The developing avian brain has a limited nutrient supply in the egg, whereas the developing brains of placental mammals have access to ongoing nutritive support for as long as gestation can be maintained. The extensive pallial sheet organization of the developing mammalian forebrain may permit greater expansion during development than does the ganglionic eminence-based proliferation that dominates the developing avian forebrain (Butler and Hodos, 2005). Studies of avian brain development are currently confined to quite small-bodied domestic species, so nothing is known about brain development in large-bodied species like the ratites. There may also be differences between developing mammals and birds in the arterial supply to the brain that limit brain growth, although this has never been quantitatively analysed. Our analysis suggests that carotid canal size in Dromornis planei was at the upper limit of what would be expected for body size and larger than expected for brain size, so at least carotid supply appears to be more than adequate in the adult of that species. Nevertheless, no information is currently available on arterial supply to the developing brain of large birds and how it might match or limit brain development.
Finally, it may be that large avian brains have much more processing power per unit volume than large mammalian brains. Olkowicz and colleagues have recently shown that the forebrains of some birds (corvids and parrots) have a more than two-fold higher neuronal density in the forebrain compared to similarly sized brains of rodents or primates (Olkowicz et al., 2016). Nevertheless, current data suggest that this high neuronal packing density does not apply to galloanseriformes and ratites (Olkowicz et al., 2016), groups to which the largest birds belong.
Concluding remarks
The findings indicate that in many respects the paleoneurology of these giant Miocene dromornithids is much as one would expect for an anseriform (duck/goose) of that size. Endocranial volume was close to what would be expected for that body weight and many osteological features of dromornithids that have neurological significance (e.g. orbital diameter, Wulst area, oculomotor canal area, maxillomandibular canal area) were of a size that would be expected for an anseriform of such a body weight and ECV. Visual and auditory function seem to have been comparable to modern anseriformes, although vestibular detection of angular acceleration may have been poor. On the other hand, hypoglossal canal area was larger than expected, consistent with an ability to precisely manipulate large seed-pods and their crushed remains in the oral cavity.
Acknowledgements
I would like to thank Peter Murray, formerly of the Alice Springs branch of the Museum and Art Gallery of the Northern Territory for access to the fossils, and Louise Lock of Alice Springs Hospital Radiology for her help scanning the dromornithid specimens.
References
Birchard GF, Deeming DC (2009) Avian eggshell thickness: scaling and maximum body size in birds. Journal of Zoology279, 95–101.
Butler AB, Hodos W (2005) Comparative Vertebrate Neuroanatomy. Evolution and Adaptation. second ed. Wiley, Hoboken, New Jersey.
Deeming DC, Birchard GF (2008) Why were extinct gigantic birds so small? Avian Biological Research 1, 187-194.
Georgi JA, Sipla JS, Forster CA (2013) Turning semicircular canal function on its head: dinosaurs and a novel vestibular analysis. PLoS ONE 8, e58517.
Iwaniuk AN, Nelson JE (2003) Developmental differences are correlated with relative brain size in birds: a comparative analysis. Canadian Journal of Zoology 81, 1913-1928.
Murray PF, Vickers-Rich P (2004) Magnificent Mihirungs: The Colossal Flightless Birds of the Australian Dreamtime. Indiana University Press, Bloomington & Indianapolis.
Olkowicz S, Kocourek M, Lucan RK, Portes M, Fitch T, Herculano-Houzel S, Nemec P (2016) Birds have primate-like numbers in the forebrain. Proceedings of the National Academy of Science USA. 113, 7255-7260.
Scofield RP, Ashwell KWS (2009) Rapid somatic expansion causes the brain to lag behind: The case of the brain and behaviour of New Zealand’s Haast’s eagle (Harpagornis moorei). Journal of Vertebrate Paleontology 29, 637-649.
Van Dongen PAM (1998) Brain size in vertebrates. In The Central Nervous System of Vertebrates (ed. R. Nieuwenhuys, H. J. Ten Donkelaar, C. Nicholson). Berlin: Springer. pp. 2099–2134.
Worthy TH, Handley WD, Archer M, Hand SJ (2015) The extinct flightless mihirungs (Aves, Dromornithidae): cranial anatomy, a new species, and assessment of Oligo-Miocene lineage diversity. Journal of Vertebrate PaleontologyDOI:10.1080/02724634.2015.103134.
Wroe S (1999) The bird from hell? Nature Australia 26, 56-63.