Summary

The zoogeography of moas is compared with that of other bird groups within New Zealand. Moa species tend to be restricted to one or other of the two main islands, whereas the other birds, flightless as well as flying, tend to occur on both islands. The ecology of moas provides no explanation for the contrast. Rather, the anomaly probably reflects faulty taxonomy, several subspecies of moas having been ranked as true species.

In 1873 Haast, drawing largely on the descriptions of Owen, made the first serious attempt to classify the moas. Subsequently, Lydekker (1891) and Parker (1895) refined the generic definitions, and in 1930 Oliver produced a fully integrated taxonomy. It was further improved by Archey (1941) and Scarlett (1972). There would be general agreement with this skeletal history of the evolution of moa taxonomy. Other writers might add a few more names and dates — they are unlikely to subtract any — but they would recognise the progression I have listed. It passed through a descriptive phase powered by Owen's genius, to a classificatory phase dominated by Parker and Oliver, to a consolidatory phase in which Archey and Scarlett rationalised and economised the previous classifications.

Meanwhile, Fleming (1962) placed the extinction of moas in ecological and historic perspective. Far from being a puzzling and protracted episode reflecting an obscure conjunction of unknowable circumstances in the early Holocene, the extinction stood revealed as an event modern in its timing and abrupt in its duration. Most species of moas were still extant in 1600: while Cromwell was making Ireland safe for potatoes the Polynesians were making New Zealand safe for kumaras.

Classification of incomplete specimens usually goes through a stage when species proliferate. The moas provide no exception. The trend reached its apogee with Rothschild's (1907) 38 described species, many of them based on single bones. Oliver (1930) reduced the number to 22 which he divided between two families, the Dinornithidae and the Anomalopterygidae. Subsequently Archey (1941), who had access to a wider range of material, continued this rationalisation by reducing the number of species to 19. His sensible scheme was made possible by his study of variation within species, thereby enabling him to reject those taxa that fell within the range of variation of others. He accepted Oliver's (1930) families, and subdivided the Anomalopterygidae into the sub-families Anomalopteryginae and Emeinae.

McDowall (1973) complained that there were still too many species for his liking. He looked at the problem as a zoogeographer rather than as a taxonomist, pointing out that the late tertiary history of New Zealand did not provide the necessary conditions for a radiation of this magnitude, and that none of the other groups of birds radiated in this way.

Taken at face value Table 2 suggests that the zoogeography and hence ecology of moas was quite different from that of the other birds. Circumstantial evidence as to the ecology of moas is thin, but what data are available on distribution, diet and the species of birds associated with them in sub-fossil deposits, suggest that most forms, if not all of them, lived in forest and along the forest edge (Simmons 1968, Gregg, 1972). Moas ranged from sea level to the upper timber line and penetrated the alpine grassland beyond. Habitat requirements for most species were probably generalised in much the same way as are those of kiwis today. Ecological counterparts of the moas are more likely to be found with the forest-dwelling cassowaries of northern Australia and New Guinea than with grassland forms such as emus and ostriches. There is no evidence to suggest that the ecology of moas was in any way strange or remarkable; ecology provides no explanation for the strange zoogeography implied by their current taxonomy.

A more likely explanation is that the zoogeography of moas is much the same as that of the rest of the New Zealand avifauna but that this is not reflected in their current classification. The dearth of species occurring on both islands need reflect no more than an unwarranted assigning of species status to North Island and South Island subspecies. Interestingly enough, Archey (1941, p. 62) came within a whisker of this interpretation in his treatment of the genus Dinornis. He remarked that the series in the South Island — D. torosus, D. robustus, D. maximus, ranked in ascending size — was paralleled by the analogous series D. struthoides,* D. novaezealandiae^* and D. giganteus in the North Island. Scarlett (1972) explicitly invoked subspeciation to explain these similarities, but then awarded species status to all six forms in his informal list of species. Since, in each case, the South Island form is a little more robust than its North Island equivalent, there seems little reason to suspect anything more than straight-forward subspeciation conforming to Bergman's rule. Similar comparisons can be made with species-pairs of the other genera.

I am actively dissatisfied with the checklist classification, but my dissatisfaction is founded on distributional anomalies rather than on morphology. Hopefully taxonomists will take a closer look at the taxonomic relationships deduced from the size and shape of bones. Luckily they are not entirely satisfied with the prevailing scheme witness, for example, Scarlett's (1972) drubbing of Oliver's (1949) genus Zelornis and Cracraft's (1976) conclusion that only 13 species should be recognised. As Dwyer (1976) and others have argued, if it is to have any evolutionary meaning a classification must make sense according to ecological and zoogeographic criteria as well as according to morphological criteria. My considerable fund of osteological ignorance does not, of itself, disallow a critical interest in the classification of moas. Nor does it disqualify me or any other ecologist from questioning the biological reality of this or that ‘species’ which, although perhaps morphologically distinct, is ecologically anomalous.

There are four areas in which ecology can say something useful about the classification of moas:

Sexual dimorphism: We know from extant species that ratites can be sexually dimorphic, particularly with regard to size. So far no bimodality of size has been detected within any species of moa. Strange? An ecologist would suspect strongly that he was dealing with sexual differences if two forms shared a common distribution and were found together consistently in ratios not significantly more disparate than 1:4.

Subspeciation: If two forms, one on each of the main islands, are morphologically closer than is either to other species of that genus with which it is sympatric, then subspecies status is the most economic and conservative interpretation.

Misclassification: If a form is well represented on one island but occurs rarely on the other we can reasonably question its occurrence on the latter island. It may be misidentified there or the locality label may be in error. Neither circumstance is unprecedented.

The one-off problem: If a species is erected on a single specimen (e.g. Dinornis gazella, Pachyornis murihiku, and Megalapteryx benhami), or a particular bone (Dinornis hercules), we should wonder why, with the wealth of material available, further specimens have not been discovered. Does the specimen represent a morphological extreme of some well-known species? Ecologists tend to distrust this kind of data for two reasons: they are conditioned to question the significance of an unreplicated event, and they would view as anomalous the corollate implication that several species of moa were extraordinarily rare. It might be as well to keep these ‘species’ on a suspense list, transferring them to the list of species we are sure of only after further material is unearthed in support of the original diagnosis.

Probably the most efficient instrument for tightening up the current classification is a multivariate analysis of a large number of bones collected, without judgement as to species, from both islands. The first question we might ask of it is how many distinct morphological groups (clusters) can be recognised within the moas. These groups would then be investigated further to determine which are sexes, which races and which species, by use of distributional criteria similar to those listed above. The problem of deciding which bones belong to which specimen can be avoided by restricting the analysis to one bone (femur, tibia) or a single fused bone sequence (cranium, pelvis, sternum, metatarsus).

As a means of determining how large a sample might be required for this exercise I ran a preliminary analysis on the measurements of 60 pelves and 85 crania recorded by Oliver (1949). Three multivariate techniques (Principal Component, MULTBET, MULCLAS) generated clusters mainly on the basis of the absolute size of specimens. The measurements were therefore converted to ratios and re-analysed. This time the clusters produced interesting relationships at and above the level of genus, but these samples were too small to allow resolution at the level of species. A sample of around 500 is needed, and that requirement rules out crania, and probably also pelves. The answer seems to lie with leg bones, but if these were used the sample size should be increased further or the number of measurements taken from each should be expanded well beyond the standard set.

But does it all really matter? Is the questioning of the number of moa species a first cousin to the debate over the number of angels that can dance on the head of a pin? It does matter. The New Zealand avifauna provides unparalleled opportunities for cracking some of the larger general problems of zoogeography and evolution. Because the tertiary and quaternary geological history is known with fair precision, and that knowledge improves each year, before long we should be able to calculate rates of speciation and subspeciation for the birds. Few other countries provide this opportunity. If there is a pattern of distributions common to several species in New Zealand it can be summarised as a zoogeographic model of how this came about in terms of what is known of the geological history. The hallmark of a useful zoogeographic model is its generality, the extent to which it accounts for the distributions of many species, not just one or two. But any attempt to produce such a model runs headlong into 25 species of moas doing something else. Until we know if the moas are, in fact, zoogeographically distinct from the rest of the avifauna, avian zoogeography in New Zealand will hang fire.

Acknowledgements

I am very grateful to the following people who criticised a previous draft of this note: J. A. Caughley, J. Cracraft, P. D. Dwyer, C. A. Fleming, F. C. Kinsky, R. M. McDowall, R. J. Scarlett, N. G. Stephenson, R. B. Sibson, G. R. Williams and J. C. Yaldwyn.

Literature Cited

Archey, G., 1941: The moa. A study of the Dinornithiformes. Auck. Inst. Mus. Bull. 1: 1-145.

Checklist Committee, 1970: Annotated checklist of the birds of New Zealand. Reed, Wellington.

Cracraft, J., 1976: The species of moas (Aves: Dinornithidae). Smithsonian Contributions to Palaeobiology 27: 189-205.

Dwyer, P. D., 1976: Systematics, ecology and biological resources. Search 7: 294-298.

Fleming, C. A., 1962: Extinction of moas and other animals during the Holocene period. Notornis 10: 113-117.

Gregg, D. R., 1972: Holocene stratigraphy and moas at Pyramid Valley, North Canterbury, New Zealand. Rec. Canterbury Mus. 9: 151-158.

Lydekker, R., 1891: Catalogue of Fossil Birds of British Museum. Brit. Mus., London.

McDowall, R. M., 1973: Zoogeography and taxonomy. Tuatara 20: 87-96.

Oliver, W. R. B., 1930: New Zealand birds. Fine Arts, Wellington.

——, 1949: The moas of New Zealand and Australia. Dominion Mus. Bull. 15: 1-206.

Parker, T. J., 1893: Classification of the Dinornithidae. Trans. N.Z. Inst. 25: 1-3.

Rothschild, W., 1907: Extinct birds. Hutchinson, London.

Scarlett, R. J., 1969: On the alleged Queensland moa Dinornis queenslandiae De Vis. Mem. Qld. Mus. 15: 207-212.

——, 1972: Bones for the New Zealand archaeologist. Canterbury Mus. Bull. No. 4: 1-64.

Simmons, D. R., 1968: Man, moa and the forest. Trans. Roy. Soc. N.Z., Gen. 2: 115-127.