Volume 94 1985 > Volume 94, No. 3 > Nga taro o Aotearoa, by P. Matthews, p 253-272
NGA TARO O AOTEAROA
Colocasia esculenta (L.) Schott (taro) is an ancient crop now grown throughout the tropics and subtropics in subsistence economies (Plucknett et al. 1970, Herklots 1972, Leon 1977). Yen and Wheeler (1968) reported chromosome numbers for taro from eastern Asia and islands of the Pacific as far as Easter Island, and throughout this region, including New Zealand, they found the chromosome number 2n=28. The chromosome number 2n=42, however, while present in eastern Asia and Melanesia, was not found anywhere in Polynesia except New Zealand, and the possibility was raised that taro were introduced to New Zealand from Melanesia. This provoked much interest because early ethnographic records and Maori traditions (Best 1976) clearly establish that taro was introduced to Aotearoa in pre-European times, and it is generally accepted that Maori ancestral and cultural origins lie somewhere in Polynesia (Davidson 1979, Bellwood 1978).
Yen and Wheeler (1968) drew no definite conclusion from their discussion of whether the possible Melanesian introduction occurred before or after the arrival of Europeans in New Zealand. The present research was undertaken in an attempt to answer this question.
Observations are presented here from field work conducted during 1982 and 1983 in New Zealand. Intensive field work was carried out in widely separate areas of the North Island, establishing something of both the general range and local distribution pattern of taro. Three phenotypically distinct variants of taro were found with major distributions. Descriptions and chromosome counts for these and other variants are given in detail elsewhere (Matthews 1984 and in preparation). In order to gain understanding of present-day taro distributions, natural dispersal, survival, and horticultural practices were noted. Particular consideration is given to the suggestions of Rattenbury (1956), and Yen and Wheeler (1968) that taro on the Cavalli Islands, at Spirits' Bay, and on Great Barrier Island may be remnants from pre-European times.
PREVIOUS RECORDS OF DISTRIBUTION
Early written records indicate that taro was once widespread in Maori cultivations of the North Island (Nicholas 1817, Wade 1842, Best 1976). In 1769, Captain Cook and his companions observed taro in plantations - 254 around the East Cape and in the Bay of Islands (Best 1976). Colenso (1880) lists Maori names for taro in two groups according to the geographic origin of the name: (1) the North, including the Bay of Islands, Hokianga, and Kaitaia districts; and (2) Hawkes Bay and the East Coast, south of East Cape.
European crops in Maori cultivations were commonly observed by the early writers after Captain Cook. Hargreaves (1959) writes that by the time the first European settlers arrived in the North Island (between 1800 and 1850), Maori agriculture had developed to such an extent that it was able to provide the settlers with regular supplies of pigs and crops of European origin: potatoes, maize, and wheat. Taro appears to have been displaced by the introduction of the European crop plants. Cheeseman (1900) notes that 50 years before “taro was seen in every Maori cultivation of any size” but had fallen into almost total disuse by 1900. Kirk (1869) briefly describes, among naturalised plants of the Auckland Province, taro as remnant or escaped from field cultivation, found growing in waste places or by roadsides, and in wet soil or water.
Taro was also observed last century in South Island cultivations. Dieffenbach (1843) describes taro growing alongside potatoes by the village of Mokupeka on Arapawa Island, Queen Charlotte Sound, about 1839, while Brunner (1952) observed taro plantations “of former days” near deserted paa at the mouth of the Hokitika River in 1847. It is not clear from these and other reports whether taro was a pre-European crop in the South Island.
Recently, taro chromosome numbers were reported for unspecified sites on the Cavalli Islands (Rattenbury 1956, 1957), Great Barrier Island, Spirits' Bay, and near unspecified settlements of North Auckland and the Bay of Plenty (Yen and Wheeler 1968). Hayward et al. (1979) show taro sites found during an archaeological survey of Motukawanui in the Cavalli Islands.
VARIANTS AND CHROMOSOME COUNTS
The term “variant” has been adopted by the present author for informal reference to phenotypically distinct plants within the polymorphic species Colocasia esculenta (L.) Schott (Matthews 1984 and in preparation). Present knowledge of genetic variation and reproductive behaviour in the genus Colocasia is poor, and good formal descriptions of species or subspecies in terms of phylogeny or reproductive barriers cannot yet be made.
Eight variants were collected during the study and chromosome counts were made for five of these. The three variants not counted are known only from a botanical garden. Two variants, with the collection - 255 numbers AKL 34 and AKL 79, are known from only one house-garden each, and have the chromosome number 2n=28. Three variants having major distributions in the North Island were found. These are variant RR (red petiole and rounded blade), variant GR (green petiole and rounded blade), and variant GP (green petiole and pointed blade). Each of these three variants has the chromosome number 2n=42, determined by counts for plants of each from a range of sites.
Surveys were made by observation from the road, questioning farmers and other residents, and exploration by foot. During the initial exploratory trip (Matthews 1982) it appeared that taro is common in rural Maori communities, and subsequent surveys were generally conducted in such communities. The subjective decision to leave a locality was made when either it was felt that too little new information could be gained by further interview or observation, or when the effort of finding people or sites was too time-consuming. Taro specimens and information about sites were also received in response to public notices or other channels of contact, and were provided by a number of archaeological and botanical field workers. Reference was made to site records of the New Zealand Archaeological Association.
Criteria for recording sites as separate
Generally each site record represents taro in a local watershed (stream, gully, rivulet, etc.) separate from other local watersheds. Exceptions to this rule were sometimes made where taro in a larger confluence were found below taro in a tributary or tributaries some distance away. A site on the larger confluence is recorded separately in this case because it is not possible to identify a particular tributary as the only likely upstream source of taro, either because more than one tributary is known with taro or because not all the tributaries have been searched. Exceptions to the rule were also made where taro were found near each other but in different environments. Non-cultivated and cultivated taro near a house may be recorded as separate sites if one group does not appear immediately derived from the other.
To give a summary view of site variation, the sites have been classified according to the scheme shown in Figure 1, and explained below with definitions. In this scheme, it is the environment of the taro plants which determines how the site is described or classified.- 256
FIGURE 1. Scheme for describing taro sites. Four categories (shaded) are identified under the overlapping terms ‘garden’ and ‘non-cultivated’.
Cultivation: The ground about the plants is cultivated at present. The plants may or may not form a clumped growth habit, depending on how they are treated.
Non-cultivation: The ground is not cultivated at present. Plants of each of the variants typically form clumps by natural vegetative propagation. Varying intervention in the natural growth occurs, depending on how people use the plants.
Garden: While cultivation of the soil clearly marks a garden, non-cultivation may make it difficult to distinguish between a garden and a wild site. If information from a local resident is not available, two kinds of circumstantial evidence can indicate a non-cultivated garden: (1) the immediate proximity of structural evidence for human settlement (house or other building of the European architectural tradition); and (2) the presence of other exotic plants characteristic of gardens, garden ornamentals such as Alocasia macrorrhiza and Zantedeschia aethiopica for example.
Derelict: Non-cultivated garden sites are considered derelict if they are in the immediate proximity of only ruined or uninhabitable structures, or if they appear to be garden sites solely because of the presence of other exotic plants.
Non-derelict: Non-cultivated garden sites are considered non-derelict if they are in the immediate proximity of an inhabited or clearly habitable house.
Wild: A non-cultivated site is classed as wild if the circumstantial evidence of association with settlement (of European tradition) is unclear or absent. This requires subjective evaluation of the field evidence and depends on the thoroughness of field exploration, which for practical reasons can never be complete and consistent. Relevant field evidence of - 257 settlement may also have been destroyed. Consequently, the categories of “non-cultivated garden” and “wild” overlap to an unmeasurable extent.
Summary of site records
A total of 300 sites have been recorded, including 243 sites reported by the present author. Three sites which do not have variant RR, GR or GP are excluded from the summary (Table 1). The oldest site recorded was observed in 1930 or 1931, and the oldest site seen again during the present study was first recorded 26 years earlier (an East Cape site with variant GP growing wild). The majority of sites recorded before 1981, and sites recently observed by people other than the author, fall into the “not known” categories shown in Table 1. Because of this, the frequencies of identified variants in known site categories largely represent the present-day distribution of taro.
Occurrence of C. esculenta variants RR, GR, and GP in different kinds of site. Note1: site numbers corrected to account for 12 sites where two variants were recorded within each site. The percentages for known sites are based on the corrected site numbers.
Geographic distribution of taro variants
The distributions of the three variants RR, GR and GP are shown in Figures 2 to 4. The areas shown, North Auckland, the Hauraki Gulf and - 258 East Cape, represent respectively the northern, middle and southern areas of the explored distribution range. Inland East Cape (Figure 4) was not searched since it is largely uninhabited, unroaded, steep hill-country. Some areas which are known through correspondence and archaeological site records to have taro have not be explored: the Kaipara Harbour district, Tauranga, Taranaki, and northern coastal areas of the South Island.
The local distributions of taro (variant RR) on two offshore islands of North Auckland are shown in Figures 5 and 6. These correspond to two of the three clusters of sites indicated in Figure 2.
The three major variants can be differentiated by the combination of their over all frequency, general geographic distribution, and occurrence in different kinds of site. Variant RR is the most common over all (75 percent of identified taro), is common in all areas, and is found most commonly in non-derelict, non-cultivated gardens and in the wild; variant GR is less common over all (13 percent of identified taro), is most common in North Auckland, and is found most commonly in non-derelict, non-cultivated gardens; and variant GP is also less common over all (12 percent of identified taro), is also most common in North Auckland, but is most commonly found in the wild (Table 1, and Figures 2 to 4).
Natural dispersal and survival
The distribution of taro in New Zealand is the outcome of both natural dispersal and transfer by people. Natural dispersal by vegetative means occurs by the movement of corms and side-corms by water down watersheds (variants RR and GR), and extension by attached sideshoots (all variants) or spreading stolons (variant GP). Seed production has not been observed, despite numerous observations of flowering (Cooper 1969, Matthews in preparation), and contrary to previous statements by the author, based on a mis-identification (Matthews 1984).
Taro were observed in frost-prone and in exposed sites and apparently survive such harsh conditions because of the immediate shelter of stream banks, or the protection of apical and lateral shoots in clumps by closely growing corms and leaves. A major factor in the survival of all taro is the presence of axillary buds, each having the potential to develop into a new lateral, vegetative shoot. These represent a reservoir of growth points from which regrowth can occur if exposed parts of the plants are damaged. A systematic comparison of the hardiness of the different variants in New Zealand has not been made, but during field work an impression of considerable hardiness was gained with respect to the three major variants. Apart from noting that variants AKL 34 and 79 have the common trait of axillary buds, nothing certain can be said at present about the natural survival ability of these 28-chromosome taro in New Zealand.- 259
FIGURE 2. C. esculenta in North Auckland: distributions of variants RR, GR, and GP. Unless otherwise indicated each symbol represents a single site, in this and subsequent figures.- 260
FIGURE 3. C. esculenta in the Hauraki Gulf, Auckland: distributions of variants RR and GR. Variant GP has not been observed here.- 261
FIGURE 4. C. esculenta around East Cape district: distributions of variants RR, GR, and GP.- 262
FIGURE 5. Distribution of C. esculenta (variant RR) and archaeological site on Motukawanui, Cavalli Islands, North Auckland. Eight taro sites have been recorded, although only five separate tributaries are populated. See locality diagram, Figure 6. Map adapted from Hayward et al., 1979.- 263
FIGURE 6. Distribution of C. esculenta (variant RR) and archaeological sites on Mahinepua, North Auckland. Four taro sites have been recorded, one for each populated watercourse. Survey by A. E. Wright (personal communication, 1982).- 264
Use and horticultural practice
Unless specifically stated otherwise, the following observations are best regarded as applying to the variant RR, which predominates in both gardens and the wild, and to which informants directly referred in many instances.
The use of blades from leaves was rarely mentioned. Younger leaves are preferred for eating, but one informant said that older leaves can be de-ribbed (central and lateral veins of blade removed) to get rid of poisons. Contrary to the general opinion that the leaves of variant GP are inedible, even after cooking, one informant said they are good, but need longer cooking. Two Maori informants gave Pacific Island methods of cooking using canned coconut milk, and reports were made of immigrant Islanders harvesting leaves from local taro.
Use of petioles was never mentioned. Corms are the part generally eaten, after either boiling or roasting to remove poisons. Feeding taro to pigs was frequently mentioned, particularly in North Auckland, where variant GP was described as inedible and good only for pigs. In one instance, poor ground conditions (infertile clay) had resulted in variant RR corms which were unpleasant to eat, but were cooked for pigs.
Although tangi were most frequently mentioned as occasion for taro consumption, taro also appears useful for any large, temporary gathering demanding large amounts of food. For some families who have many rows of taro in their cultivations, domestic use of the plant as a starchy staple may be important. A number of informants remarked upon a general upsurge of interest in cultivation and use of the plant in recent years, or expressed increased interest themselves. Nevertheless, taro appears generally regarded with disfavour or disinterest, particularly by younger people.
Of the areas surveyed, North Auckland was where cultivations of more than just one or two plants were most frequent. Taro is cultivated as a summer root crop, although in some cultivations clumps more than one year old were seen. In the Hokianga, informants regarded October (late winter) as the time to plant and March as the time to harvest, after the leaves die off. (Generally, the older leaves die, and growth of new leaves slows down or stops, with the onset of winter.) These dates are just a loose rule; late frosts in October are best avoided, and later plantings in November and December were reported. One informant gave March or April as harvesting times, and a harvesting in May was observed in the Bay of Islands. Further south, in the East Cape area, planting is said to be in November, after the late frosts.
Side-corms and the cut tops of corms (top of the corm together with leaf bases, preserving the central shoot) were both mentioned as planting - 265 material or “seed”. Planting material is initially obtainable from a wide range of sources: from locally known wild taro, from non-cultivated derelict or non-derelict gardens, and from the gardens of friends or relatives. The statements of many informants about the original source of their plants indicated that transfer of planting material about the North Island occurs over long distances between friends and relatives. Planting material is also collected when corms and larger side-corms are harvested for eating. Very little was learned about winter storage methods. In two reports plants or pieces of plants were said to be kept outside on the ground, under loose cover, over winter.
The occurrence of non-cultivated taro is often the result of neglect, either because a house and its associated garden have been abandoned, or because plants have been discarded or have naturally escaped from a cultivated area and now grow in an adjacent area of swampy ground, a stream or a ditch. It can be easily imagined that many wild taro sites have arisen by such a gradual transition of taro out of cultivations into natural watercourses, although this cannot be observed in a short-term study such as the present. The statements of informants show, however, that plants may be moved by many different routes between different types of site. A non-cultivated clump of taro may be kept in a garden as an ornamental, as a reminder of the past and for an occasional taste, or as a potential source of planting material for cultivation. One informant suggested that present wild taro had originally been planted in streams to preserve planting stocks, noting that they produce too much leaf and not a good “root” (corm) in streams.
The harvesting of corms from non-cultivated sites, garden or wild, can be at any time of year, although there are said to be times of the year when the corms are less poisonous. A number of informants emphasised the need to replant at the time of harvesting to maintain plants in non-cultivated sites.
Living archaeological specimens?
Taro has often been recorded in the New Zealand Archaeological site recording scheme as comprising all or part of an archaeological site. Archaeological sites are legally defined as more than 100 years old (Historic Places Act 1980), but it is generally impossible to prove such antiquity for taro sites, since recent or present settlement is also commonly associated with the archaeological landscape of paa, pits, terraces and middens.
Rattenbury (1956) suggests that taro from the Cavalli Islands may have survived there “since earliest known settlement”, while Yen and Wheeler (1968) note that specimens from Great Barrier Island and Spirits' Bay - 266 “cannot be ascribed confidently to transfer in European times”. Following these suggestions and the report of taro sites on Motukawanui by Hayward et al. (1979), opportunities were taken to visit the islands in the hope that detailed surveys would establish whether taro there can be regarded as “archaeological”. From Spirits' Bay taro specimens have been obtained, but a detailed survey was not made there. The settlement at Kapowairua, at the eastern end of Spirits' Bay, was occupied earlier this century, according to former residents now living at Te Hapua nearby (O. Sutherland, personal communication 1982), so it is quite possible that recent introductions have been made to the area.
Motukawanui (Figure 5) is the largest member of the Cavalli Islands. The Cavallis were home to numerous Maori from pre-European times up to about 1900, and the first European encounter was with Captain Cook in 1769 (Hayward 1979). Cruise (1824) records Cavalli Islanders offering vegetables for sale in 1820, and photographs at the Auckland Institute and Museum Library (Hayward et al. 1979) show, at about 1900, a small village of Maori dwellings behind the southern part of Papatara Bay. The island was farmed from the latter part of the 19th century up until 1974, and a homestead at the northern end of Papatara Bay was last permanently occupied about 1953 (Hayward 1979). Since 1978, the island has been part of the Bay of Islands Maritime and Historic Park.
On Motukawanui, taro (variant RR) is found in five tributaries and may therefore represent five separate plantings (Figure 5). Archaeological pit and terrace sites are distributed along the ridges, and the distribution of taro high in tributaries among remnant patches of native bush suggests an association with past Maori occupation of the archaeological landscape. This occupation may have continued for some time after the encounter with Cruise in 1820. The history of occupation of Mahinepua (Figure 6), an island not far from Motukawanui, is not known. Until recently it has been farmed without occupation, and it is largely covered by ungrazed grass (A. Wright, personal communication 1982). The distribution of taro (variant RR) in four separate streams here is also suggestive of an association with past occupation of the archaeological landscape.
These local distributions, however, provide only circumstantial and inconclusive evidence. Even without recent occupation it is possible that plantings have been made on both Motukawanui and Mahinepua in recent times by Maori living nearby on the mainland and visiting the islands for various purposes (D. Simmons, personal communication 1983).
On Great Barrier Island, taro was not found in the now unoccupied forested area of the island's northern end, despite an intensive search and - 267 a high density of archaeological sites. Although taro (variant RR) was found in settled areas nearby (Figure 3), this may be of recent (post-First World War) introduction from the eastern Bay of Islands, according to a Maori resident of Motairehe, northern Great Barrier Island. Taro was present on the island in earlier times, since Kirk (1868) records it (as Colocasia antiquorum) in a species list for the island. The present absence of wild taro in an unfarmed part of the island is possibly because of the very dense populations of pigs and goats inhabiting the native forest. Neither of these herbivores of taro have been mentioned as wild inhabitants of Motukawanui, where taro survives, although other islands in the Cavalli group have had pigs or goats (Hayward, 1979). The survival of taro in the wild may require protection from dense populations of these herbivores, while pasture-fed stock probably pose little threat.
It is apparent that cultivation is not the only way in which people aid survival of taro in New Zealand. Deliberate dispersal of taro into non-cultivated sites is also important, though not the only means by which it may arrive in such sites. The replanting associated with the harvest of taro from non-cultivated sites may well be necessary to the plant's perpetuation in more frequently harvested sites.
It is in the practice of replanting that the distinction between cultivated and non-cultivated sites becomes unclear, since planting to some extent alters the immediate environment of the plants. Zeven and de Wet (1982) observe that people maintain cultivated plants in an artificial habitat if the plants are of sufficient value, and that cultivation includes all kinds of agricultural practices, from merely protecting individual plants to actual planting or sowing and tending of planted populations. The definition of a cultivated site used in the present study has been deliberately restricted to sites in which the soil appears cultivated, since consistent observation of horticultural practices attendant to the plants, over time, was not possible for every site.
Site categories and analogy from the present to the past
The different categories of taro sites have been defined here partly on the basis of their present relationship to settlement of European design, to help distinguish between garden and wild sites and between derelict and non-derelict gardens. It seems quite likely that an analogous range of sites also existed in pre-European times in relation to Maori settlement. The relative importance of the different kinds of site would have been quite different, though, since the cultivation of taro before the arrival of European crops was undoubtedly of much greater importance than after- - 268 wards. Nevertheless, non-cultivated taro in a variety of settings, garden or wild, could have had similar uses in the past as a less labour-demanding, less controlled source of food and planting material.
The general geographic distribution of each variant provides, in itself, little evidence regarding history. The variant GR, for example, with a distribution apparently centred in North Auckland (Figure 2), could be a relic from a wider distribution in the past, or it could have been of such recent origin that only limited dispersal has taken place. It is apparent that almost any explanation can be proposed for any geographic distribution, in the absence of other kinds of evidence.
Horticultural practices and site types
Variant RR is clearly the most important variant as a food crop, and its dominance in all types of site (Table 1) suggests that it has been more important in the past than the other variants, whatever the routes by which plants reached these sites.
Speculation about variant GR is difficult, since almost no information was obtained specifically on its use. This lack of information might not simply be due to a small chance of encountering both informants and the plant at the same time, since variant GP, with a similar low overall frequency, was quite often spoken about. It is quite possible that variant GR is not commonly recognised as a distinct type, since it appears indistinguishable from variant RR in leaf shape and corm morphology (Matthews 1984, and in preparation). The distinctive petiole colours of these variants may be generally ignored if, perhaps, there is little or no difference in the cooking and eating qualities of the two variants. Variant GR was mostly found without cultivation in non-derelict gardens (Table 1), but the historical significance of this is unknown to the present author (who is unable to speak for the gardeners themselves).
The general absence of variant GP from garden sites undoubtedly reflects its poor eating qualities. Another reason for exclusion from gardens is suggested by the observation that stolon-producing taro (like variant GP) are unpopular as cultivars, in areas outside New Zealand, because they require the repeated removal of stolons to promote growth of the central corm (Wilson 1982).
It is more difficult to suggest how or why variant GP has become dispersed in numerous wild sites (Table 1). Two possible contributing reasons can be noted: (1) that variant GP has been planted outside garden areas as a cheap source of pig food not requiring special care; and (2) that since its introduction the curiosity of people about a potential - 269 new food source has been sufficient reason for them to plant it, in or out of gardens, though subsequently discarding the undesirable plants, and letting them survive and disperse by their own means in natural watercourses.
As a food source for pigs, variant GP may have become dispersed during the early 19th century, since pigs were raised by the Maori as an item of trade with the early European settlers (Hargreaves 1959). There is also some suggestion from present-day naming and written historical records that variant GP is the taro reputed (Best 1976) to have been introduced since the time of Captain Cook's visits, but there is no certainty in this (Matthews 1984).
Because the distributions of the three major taro variants together span a wide geographical range in which other taro appear almost completely absent, it seems likely that at least one of them is of pre-European introduction. Variant RR is predominant numerically, in the range of settings it occupies (possibly including archaeological sites) and in its present-day use as a food crop. It therefore seems the most likely taro variant to have been introduced in pre-European times. Inherent in this argument is the suggestion that taro with the chromosome number 2n=42 are of pre-European introduction to New Zealand, since each of the three major variants has this number.
This argument rests on two assumptions: (1) that taro was introduced in pre-European times (apparent in the first eyewitness European records and in Maori traditions), and (2) that pre-European taro did not completely disappear after the recent introduction of the variants found today.
Just as surprising as the wide distribution of taro with 42 chromosomes is the apparent general absence of plants with 28 chromosomes. It seems most unlikely that Polynesian ancestors did not introduce taro with this chromosome number, since such taro are widespread in Polynesia (Yen and Wheeler 1968). Further surveys within New Zealand might reveal a wider distribution of plants with 28 chromosomes. Alternatively, it is possible that 28-chromosome taro are obligate cultivars in temperate climates such as New Zealand. The second assumption noted above cannot be satisfactorily justified or discredited without field trials of Polynesian 28-chromosome taro under New Zealand conditions.
Inquiries made to date with New Zealand and European herbaria have failed to locate any taro specimens that early explorers may have collected in New Zealand. Such evidence from the time of first contact might definitely establish the pre-European status of variants present today.
Detailed genetic descriptions and comparisons of New Zealand and - 270 foreign taro are needed. Previously recorded taro with the chromosome number 2n=42 in eastern Asia and in Melanesia may be genetically quite dissimilar from any of the New Zealand variants. More thorough sampling in the Polynesian region could reveal genetically identical populations of 2n=42. Publication of chromosome counts already made in various agricultural research institutions would significantly improve the overall regional views of taro populations. Future surveys outside New Zealand must also take into account possible recent movements of planting stocks.
Genetic studies are also needed to reveal local phenotypic variants which have arisen by vegetative mutation or sexual reproduction. Little is known about the situations in which natural flowering, seed production and seedling survival take place, but there is no longer any doubt that flowering occurs in many countries including New Zealand (Sunell and Arditti 1983). Taro pollen is certainly produced by more than one of the New Zealand variants (Matthews 1984, and in preparation), but apparently in Highland Papua New Guinea it has proved impossible, or extremely difficult, to distinguish claimed fossil pollen from the pollen of other genera (Spriggs 1982). Similar difficulty can be expected elsewhere.
Finally, the suggestion has been made that New Zealand taro with the chromosome number 2n=42 could have originated in the Melanesian region (Yen and Wheeler 1968). The present observations, confined to New Zealand, cannot be related to the question of geographic origin. Nevertheless, it is interesting to speculate that voyagers from Melanesia, or even further afield, brought their plants to Aotearoa in ancient times without leaving descendants or traditions.
Grants for field work were provided by the Mappin Trust, through the Royal Society of New Zealand, and by the New Zealand Historic Places Trust. Special thanks are extended to all my correspondents and to the many people who responded so kindly to my inquiries in the field. Much is owed to the volunteer field assistants Jane Connor, Phillip Anderson, Grace Jones, Ruth Marsh, and Clayton Fredrickson, and to Michael Ford for the analysis of field data. Thanks also to Anthony Wright, Botanist, and David Simmons, Ethnologist, both of the Auckland Institute and Museum, and to Oliver Sutherland, of the Department of Scientific and Industrial Research, Mount Albert. Essential support was provided by the Departments of Botany and Anthropology, University of Auckland, and by the Department of Prehistory, Research School of Pacific Studies, Australian National University.