3.a. Unconformities of Oligocene age

Unconformities of Oligocene age are widespread in sedimentary rocks throughout New Zealand (e.g. Carter, Reference Carter1988; Field & Browne, Reference Field and Browne1989; Turnbull & Uruski, Reference Turnbull and Uruski1990; Turnbull et al. Reference Turnbull, Uruski, Anderson, Lindqvist, Scott, Morgans, Hoskins, Raine, Mildenhall, Pocknall, Beu, Maxwell, Smale, Watters and Field1993; King et al. Reference King, Naish, Browne, Field and Edbrooke1999). One distinctive break, the Marshall Paraconformity, is recognized in sections throughout New Zealand and underlying the surrounding sea-floor (Carter & Landis, Reference Carter and Landis1972; Fulthorpe et al. Reference Fulthorpe, Carter, Miller and Wilson1996). Now recognized as having formed during Oligocene erosion of the sea-floor in association with initiation of the circumpolar current and sediment starvation during maximum transgression (e.g. Carter, Reference Carter1985), this unconformity was ascribed by early workers to emergence and subaerial erosion. For example, the pioneering palaeogeographic maps of Fleming (Reference Fleming1962, Reference Fleming1979) showed the East Otago area as land during the Oligocene (see Silverpeaks to Catlins on Fig. 1). At most localities, the Marshall Paraconformity and related Oligocene paraconformities mark abrupt breaks separating fully marine sequences above and below. Although mainly intra-Oligocene with duration of 2 to 4 Ma, in some areas (e.g. Dunedin and the Canterbury Shelf) the Marshall Paraconformity separates slope–bathyal Late Eocene mudstone from Miocene outer shelf or deeper greensand and represents a gap exceeding 10 million years (Fulthorpe et al. Reference Fulthorpe, Carter, Miller and Wilson1996). In most of North Otago and South Canterbury, Oligocene and earliest Miocene sequences are extremely condensed, the paraconformity itself representing missing Oligocene time of 2 to 3 million years (e.g. Carter, Reference Carter1985; Loutit et al. Reference Loutit, Hardenbol, Vail, Baum, Wilson, Hastings, Kendall, Posamentier, Ross and Van Wagoner1988). Although the possibility of occasional local terrestrial conditions cannot be totally eliminated (e.g. Lewis & Bellis, Reference Lewis and Bellis1984), the presence of offshore marine strata directly above and below an unconformity cannot be construed as supporting the hypothesis of continuous emergence during the intervening period. Recognition of the Marshall Paraconformity within marine sequences beneath the deep sea-floor around New Zealand (Carter & Landis, Reference Carter and Landis1972; Carter, Reference Carter2003) provides additional support for a submarine origin of this same unconformity exposed so widely in on-land New Zealand.

3.b. Inland occurrences

The most-inland occurrences of marine strata in East Otago and South Canterbury coincide approximately with the position of the Oligocene shoreline as portrayed on maps by Wellman (Reference Wellman1953), Fleming (Reference Fleming1962, Reference Fleming1979), Stevens (Reference Stevens1974), Kamp (Reference Kamp1986) and many others. These workers have drawn the shorelines at, or only a short distance inland from, outcropping marine strata. However, evidence inferred from sedimentary facies sequences (e.g. Walther's Law: Middleton, Reference Middleton1973) at these inland-most localities requires that Oligocene–Early Miocene transgression must have advanced further inland, beyond those outcrops. For example, consider the area around Lake Aviemore (Waitaki Valley, South Island; Fig. 7), just inland of which the Oligocene shoreline is shown on numerous palaeogeographic maps. This region, 45–50 km inland from the present coast, is well known for its rich Cenozoic marine molluscan faunas (Marwick, Reference Marwick1935; see also stratigraphic summary in Field & Browne, Reference Field and Browne1989). Correlative marine strata occur from Aviemore to the east coast at Oamaru and beneath the adjoining continental shelf. The Cenozoic sequence exposed north of Aviemore Dam commences with fossiliferous and pebbly glauconitic sands of Eocene age resting unconformably upon planated Permian–Triassic greywacke basement (the Waipounamu Erosion Surface). These basal Cenozoic strata, ravinement deposits, are overlain by a sequence of increasingly offshore transgressive marine units, culminating in 60 m of Oligocene greensand of outer shelf depth and 20 m of richly fossiliferous outer shelf limestone of Late Oligocene to earliest Miocene age (Marwick, Reference Marwick1935; Gage, Reference Gage1957; T. S. Loutit, unpub. report, 1973; Field & Browne, Reference Field and Browne1989; I. McDermid, unpub. report, 1998). Age and thickness of this transgressive sequence imply that submarine accommodation space was available from Middle Eocene through all of Oligocene time, a period of at least 15 million years. Contemporaneous Oligocene terrestrial or clastic shoreline facies are not recognized in this area, and if any actually existed they must have lain far to the west. Comparing the sediment facies characteristic of maximum transgression at Aviemore and inland Canterbury (greensand, marl, foraminiferal limestone) with facies on present-day passive continental margins elsewhere in the world, a depositional setting of at least tens and probably hundreds of kilometres from shore is suggested. It can be concluded that there is no sedimentological evidence for the presence of Late Oligocene–earliest Miocene terrestrial environments being situated anywhere near the Aviemore locality. Similar relations are recognized widely in the foothills of the Southern Alps (e.g. Gage, Reference Gage1970), along the margin of Fiordland (Turnbull et al. Reference Turnbull, Uruski, Anderson, Lindqvist, Scott, Morgans, Hoskins, Raine, Mildenhall, Pocknall, Beu, Maxwell, Smale, Watters and Field1993), in the NW South Island (Grindley, Reference Grindley1961) and elsewhere. These will be discussed later.

Figure 7. Aviemore, a mesa-like remnant of the stripped Waipounamu Erosion Surface; to immediate right of the Lake Aviemore Dam (see Fig. 1 for location). Aviemore is the most inland remnant of preserved Cenozoic marine sequence. Gravelly Eocene marine sediments rest directly on greywacke basement. These are overlain by a transgressive marine sequence including Oligocene limestone and greensand. The crest of the Hawkdun Range (Fig. 4; see HR on Fig. 1) is visible in the far distance about 40 kilometres away. Pleistocene alluvial terraces are conspicuous along the southern (left) side of the valley. View looking NW up the Waitaki River valley. Photo: Lloyd Homer, GNS Science.

3.c. Aviemore and Silverpeaks

Also at Aviemore (see Section 3.b above), a distinctive planar erosion surface is well exposed on a table-like hill (mesa) that is underlain by Permian greywacke, one kilometre north of the Aviemore Dam (Fig. 7). We infer that this conspicuous surface is not a peneplain (the traditional interpretation); rather, it is the Waipounamu Erosion Surface that has been cut by Eocene wave planation. The conspicuously flat top of the mesa is an exhumed angular unconformity from which transgressive Palaeogene marine sediments have been largely stripped by late Neogene erosion. However, remnant lenses of cover sediment rest directly on the unconformity and they consist of fossiliferous and glauconitic pebbly marine sands of Paleocene–Eocene age. Similarly planar, but commonly much higher, ridges are well exposed elsewhere in the Waitaki Valley and to the south and west in Central Otago. For example, the Rock and Pillar, Dunstan, Hawkdun, Old Man and Pisa ranges (Figs 1, 4) are all late Cenozoic fold and fault-bounded ranges that are characterized by planar summits. Although stripped of a postulated earlier Cenozoic marine sediment cover, the morphological similarity of these high ranges to the Aviemore mesa is striking, and correlation of these surfaces is confidently proposed.

Planar ridge crests are also conspicuous in East Otago. For example, in the Mountain Road area of the Silver Peaks (Figs 5, 6), a conspicuously flat surface is cut into biotite schist. Previously regarded as forming part of the exhumed Cretaceous peneplain (Benson, Reference Benson1935; Mortimer, Reference Mortimer1993; Bishop, Reference Bishop1994), much of the ridge is thinly veneered by 1–3 m of latest Cretaceous–Paleocene marine sediment (LeMasurier & Landis, Reference Lemasurier and Landis1996; Landis & Youngson, Reference Landis and Youngson1996) and is now interpreted as having been planated by wave processes, not by fluvial erosion. This surface is identified as an early-formed portion of the Waipounamu Erosion Surface.

Wave planation is the only surficial erosional process capable of forming regional surfaces on bedrock as flat as those that characterize the planar landscapes in the Aviemore region, the Silver Peaks tableland (LeMasurier & Landis, Reference Lemasurier and Landis1996, Fig. 6), and similar planated basement in East Otago. Despite being degraded by Pleistocene periglacial processes (Wood, Reference Wood1969; Stirling, Reference Stirling1990), the characteristic flat crests of the Central Otago mountain ranges (1000–1500 m) are strikingly similar to their lower-elevation correlatives at Aviemore, Silver Peaks and elsewhere. They are all regarded as uplifted portions of the Waipounamu Erosion Surface. The absence of marine sediment from these more highly uplifted exhumed planar erosion surfaces is attributed to late Cenozoic uplift and associated terrestrial erosion. Not only that, but the absence of erosion surfaces further to the west can itself be attributed to ongoing Neogene uplift and erosion. Reduced erosion at lower elevations has permitted the outlying marine remnants to be preserved.

3.d. Siliciclastic sands

The presence of siliciclastic sands in Late Oligocene–Early Miocene sediments may also be used to argue for proximity of contemporaneously eroding land areas consisting of basement rock (King et al. Reference King, Naish, Browne, Field and Edbrooke1999). For example, terrigenous sands within the Late Oligocene–Early Miocene Milburn Limestone of East Otago might be taken to imply derivation from outcrops of schist basement proximal to the area of outcrop. Thus, maps of Fleming (Reference Fleming1962), Kamp (Reference Kamp1986) and others place the Oligocene shoreline on Haast Schist basement only a few kilometres inland from Milburn (Fig. 1). Petrographic study of the non-carbonate fraction of the Milburn Limestone is instructive in this respect. Samples collected from a 22 m section exposed in the old Milburn Lime Quarry were dissolved in dilute hydrochloric acid and analysed in 12 grain-mount thin-sections (CAL and S. Wilson). Insoluble residues ranged from 2–3% near the base to 20–40% high in the section. Quartz is the dominant mineral with glaucony pellets also common, especially in fine sand fractions where they may form up to 50%. Both quartz and glaucony tend to be well rounded and polished. The glaucony is interpreted to be a transported intrabasinal constituent and is excluded from the terrigenous siliciclastic fraction.

On a calcite- and glaucony-free basis, feldspar is the only other abundant constituent, accounting for 13–21%. Heavy minerals (hornblende, garnet, opaques) combined with altered rock fragments range from trace amounts to 5%. Potassic feldspars (orthoclase and microcline) comprise 13–19% with Na–Ca plagioclase up to 3.5%. Significantly, potassic feldspar and hornblende are absent from the Haast Schist basement throughout Otago, where end-member albite is the sole feldspar mineral. The terrigenous sand component of the Milburn Limestone is thus arkosic in composition and granitic provenance is implied. The nearest compatible crystalline basement exposed today is in Fiordland and Stewart Island (Fig. 1). However, the high degree of rounding strongly suggests a re-worked sediment source. Eocene shelf sediments in East Otago (e.g. Green Island Sandstone) and correlative units in the offshore Great South Basin (Beggs, Reference Beggs1978) are rich in potassic feldspar and may suggest contemporaneous erosion of a distant siliciclastic source area. Regardless, there is no evidence that adjacent schist basement contributed detritus to the Milburn Limestone. Potassic feldspar and hornblende are found widely in basal marine Cenozoic sediments around the margins of the schist. We conclude that schist basement in East and Central Otago was already fully inundated by the transgressive Oligocene sea.

3.e. Local occurrences of shallow marine and estuarine sediments

Local occurrences of shallow marine and estuarine sediments have also been taken to provide evidence for the location of palaeo-shorelines in previous palaeogeographic maps. In itself, this may be a sound interpretation. However, it is not valid to assume that exposed sediments actually represent sustained or precisely dated periods of sediment accumulation or that these sediments did not extend further inland.

A good example is found in the region of Pomahaka in West Otago (Fig. 1) where estuarine beds, comprising mainly shelly mudstone, rest unconformably on late Palaeozoic–Mesozoic basement (Wood, Reference Wood1956; Isaac & Lindqvist, Reference Isaac and Lindqvist1990). A general Oligocene age is widely accepted. It is likely that they represent a depositional interval occurring at the beginning of a prolonged regional transgressive marine episode. Locally, Oligocene marine strata (Chatton Marine Beds; Wood, Reference Wood1956) are present above the Pomahaka Beds but these late Palaeogene strata are rarely exposed, the area being dominated by low-relief deeply eroded Mesozoic basement mantled by Neogene fluvial gravel. Fluvial gravel rests unconformably upon localized erosion remnants of the earlier marine and estuarine strata. Thus the absence of more widespread middle Cenozoic marine deposits overlying these basal transgressive strata cannot be taken as evidence that marine conditions never extended into the more uplifted and deeply eroded areas adjoining Pomahaka.

To the contrary, the nature and thickness of (about 90 m) of the Pomahaka and Chatton formations imply gradual regional subsidence with middle Cenozoic estuarine and marine strata originally extending inland from Pomahaka. No mechanism that would permit the Oligocene–Miocene shoreline to have remained in the Pomahaka area during deposition of the documented estuarine marine sequences (90 m thick) can be envisaged. For example, any basin formed by flexure of the Otago continental crust at this locality must have been on a scale requiring subsidence (and complimentary marginal uplift) to have extended further inland by at least tens of kilometres. Thus sediments at Pomahaka, while recording transgression along the margin of an Oligocene marine basin, also imply that the basin margin must have migrated further inland to permit accommodation of the exposed sequence. Late Oligocene and Early Miocene limestone and greensand, while not exposed at Pomahaka or directly to the north, do occur at all but one of the named South Island locations surrounding Pomahaka as shown on Figure 1. It can be confidently inferred that marine conditions covered the entire area.

3.f. Evidence that is permissive but not compelling

Evidence that is permissive but not compelling also appears to have played a major but tacit role in postulating substantial Oligocene–Early Miocene land areas in New Zealand. Examples can be cited from the Southern Alps, Fiordland, Catlins and central and western North Island.

For example, a middle Cenozoic land area in the region of the present Southern Alps is shown on all published palaeogeographic maps. While extent and location of this land varies from map to map it is invariably shown to be present but not accompanied by evidence or explanation. For instance, the area around Haast Pass (Fig. 1) is shown as lying above sea-level continuously from the Cretaceous to the present day (e.g. Fleming, Reference Fleming1962; Stevens, Reference Stevens1985; Kamp, Reference Kamp1986; King et al. Reference King, Naish, Browne, Field and Edbrooke1999). Since no pre-Pleistocene sediments are present in this area, the inferred landmass can neither be proved nor disproved. However, this area of postulated Oligocene land is also a region of maximum Neogene uplift and deep erosion along the modern plate boundary (Koons, Reference Koons1990). Metamorphic basement is exposed throughout the region and any Cenozoic cover has been eroded away. Marine sequences are documented from areas of lesser uplift (and lesser Neogene erosion) on all sides of the Haast Pass area. Typically these are transgressive sequences showing increasingly offshore character from Early Cenozoic to Oligocene–Early Miocene, culminating with limestone sequences. We can therefore confidently assume that transgression proceeded well beyond all of these limiting outcrops. It probably covered the Haast Pass region as well. Thus, whereas there is no geological evidence to suggest a persistent early–middle Cenozoic landmass anywhere in the Haast Pass area, all available geological data are compatible with middle Cenozoic submergence.

3.g. The Moa's Ark heritage

The history of New Zealand's biota has been presented from the viewpoint of evolution of plants and animals stranded on an isolated emergent ‘raft’ for millions of years. This raft, a fragment of the Gondwanaland supercontinent, has been surrounded by oceans since Late Cretaceous time. It is seen as an isolated natural laboratory carrying a precious cargo of Mesozoic plants and animals. In this view, evolution in response to changing environmental pressures operating on the descendants of the original (Mesozoic) plants and animals has resulted in creation of the modern New Zealand biota.

This concept, popularized recently as ‘Moa's Ark’ (Bellamy, Spingett & Hayden, Reference Bellamy, Spingett and Hayden1990) has great public appeal and has become deeply rooted in New Zealand's national identity. It has been backed by many proponents for more than forty years (e.g. Fleming, Reference Fleming1962; Wardle, Reference Wardle1963) and is found widely in modern popular science literature (e.g. Wilson, Reference Wilson2004; Stevens, McGlone & McCulloch, Reference Stevens, McGlone and McCulloch1988; Flannery, Reference Flannery1994). Although a small degree of long-distance dispersal following Gondwanaland break-up is widely accepted (Fleming, Reference Fleming1979; Mildenhall, Reference Mildenhall1980; Wardle, Reference Wardle1984; Stevens, McGlone & McCulloch, Reference Stevens, McGlone and McCulloch1988), the distinctiveness and lack of mobility of the modern terrestrial biota has resulted in a prevailing understanding that assumes that substantial terrestrial habitats must have existed continuously since Cretaceous break-up, and that the unique biota of New Zealand is a product of ancient isolation.

However, accumulating biological evidence contradicts this view of the biogeographic history in New Zealand. Pole (Reference Pole1993) argued, using fossil data, that a significant proportion of New Zealand plants have evolved from Cenozoic colonists. In a later paper, he suggested (Pole, Reference Pole1994; see also Pole, Reference Pole2001) that the entire flora may have developed since the Oligocene. Most molecular studies have suggested that groups found in New Zealand, rather than being remnants from Gondwanan times, are actually much more recent in origin and have colonized across the significant water gaps around New Zealand. Examples include southern beeches (Swenson et al. Reference Swenson, Backlund, McLoughlin and Hill2001; Knapp et al. Reference Knapp, Stöckler, Havell, Delsuc, Sebastiani and Lockhart2005), spiders (Griffiths, Paterson & Vink, Reference Griffiths, Paterson and Vink2005), moths (Brown, Emberson & Paterson, Reference Brown, Emberson and Paterson1999), kiwis (Cooper et al. Reference Cooper, Mourer-Chauvire, Chambers, Haeseler, Wilson and Paabo1992) and various flightless insects (Trewick, Reference Trewick2000). Only a few species have levels of molecular variation that are compatible with a Gondwanan origin, such as tuatara (Rest et al. Reference Rest, Ast, Austin, Waddell, Tibbets, Hay and Mindell2003), leiopelmatid frogs (Roelants & Bossuyt, Reference Roelants and Bossuyt2005), kauri (Stöckler, Daniel & Lockhart, Reference Stöckler, Daniel and Lockhart2002) and terrestrial gastropods (McDowall, Reference McDowall2004). However, ancient lineages, on their own, say very little about the palaeobiogeographic distribution of their ancestors; there is no easy way to tell if such a lineage has not colonized relatively recently before going extinct at the source. Evidence is also accumulating that overseas dispersal is generally an important factor among Southern Hemisphere biota (Sanmartin & Ronquist, Reference Sanmartin and Ronquist2004) and that asymmetrical colonization rates between areas will often mimic geological rifting scenarios (Cook & Crisp, Reference Cook and Crisp2005). In this case, westerly wind-flows and currents that make the colonization of New Zealand from Australia more likely than movement in the other direction provide a similar pattern to that expected from a ‘Gondwana ark’. This has led McGlone (Reference McGlone2005) to comment that rather than ‘the land that time forgot’, New Zealand is the ‘flypaper of the Pacific’. Recently, Campbell & Landis (Reference Campbell and Landis2001) have suggested that the entire New Zealand terrestrial biota actually became established by accidental colonists since the Oligocene and this view has been echoed by Waters & Craw (Reference Waters and Craw2006).

3.h. New Zealand fossil record

New Zealand has a remarkably comprehensive and well-documented marine fossil record spanning the last 70 million years. Crampton et al. (Reference Crampton, Foote, Beu, Cooper, Matcham, Jones, Maxwell and Marshall2006) have established that the fossil record for post-Eocene molluscs is representative of 40–45% of the original total molluscan fauna. In contrast, the terrestrial fossil record is sparse and incomplete. Terrestrial animal fossils older than Pleistocene are particularly scarce, and fossil plants, with noted exceptions from Miocene time, are known mainly from pollen studies. Only the last 22 million years of terrestrial life are known with any modicum of detail.

In spite of the incomplete record, there remains a widely held belief that a substantial proportion of the extant biota has evolved from plants and animals that were present when New Zealand separated from Gondwanaland about 85 million years ago. Modern workers have generally maintained the continuous existence of a diverse Gondwanan terrestrial biota (Stevens, McGlone & McCulloch, Reference Stevens, McGlone and McCulloch1988; Cooper & Cooper, Reference Cooper and Cooper1995; Lee, Lee & Mortimer, Reference Lee, Lee and Mortimer2001) and reference to ‘Gondwanan biota’ is commonplace. A biotic ‘bottleneck’ within the Oligocene was proposed by Cooper & Cooper (Reference Cooper and Cooper1995), and Pole (Reference Pole2001) considered the case that the New Zealand flora represents a complete biotic turnover from the original Gondwanan biota. Both of these papers have assumed the continuous existence of a landmass, though with reduced area in the Oligocene.

Lee, Lee & Mortimer (Reference Lee, Lee and Mortimer2001) argued for a continuous Cenozoic terrestrial flora record in New Zealand. They recognize only one unit that spans the critical Oligocene to earliest Miocene period: the Gore Lignite Measures. These strata are portrayed by Lee, Lee & Mortimer (Reference Lee, Lee and Mortimer2001) to have accumulated during the interval between 16 and 31 million years ago; no breaks in this sequence are discussed. Although palynological evidence for sedimentation of the Gore Lignite Measures during this period is well documented (Pocknall, Reference Pocknall, Isaac and Lindqvist1990), no case has been made for continuous terrestrial sedimentation (e.g. Isaac & Lindqvist, Reference Isaac and Lindqvist1990) and middle Cenozoic marine beds are well known in the area (Cooper, Reference Cooper2004; see also Section 3.e above). Rather enigmatically, the palaeogeographic maps of Lee, Lee & Mortimer (Reference Lee, Lee and Mortimer2001) show the Gore area lying 100–200 km offshore at both 20 and 30 million years ago.

New Zealand palynologists have long been aware of a terrestrial floral turnover in the vicinity of the Oliogcene/Miocene boundary. The spore/pollen range chart of Couper (Reference Couper1960) shows this clearly, even within the limits of accurate dating at the time. Immediately after the demise of many Palaeogene taxa there was a sudden and dramatic influx of new Neogene taxa, ancestral to the present New Zealand flora. There was also a rapid increase in diversity as new ecological niches opened up. McGlone, Mildenhall & Pole (Reference McGlone, Mildenhall, Pole, Veblen, Hill and Read1996) looked in detail at the distribution of fossil Nothofagus pollen in the New Zealand Cenozoic. They show a sudden change in the types of Nothofagus pollen in the vicinity of the Oligocene/Miocene boundary, specifically the demise of Nothofagidites matauraensis, N. flemingii and N. waipawaensis and the rise of N. cranwelliae, N. falcatus and N. spinosus.

Recent morphological (Parrish et al. Reference Parrish, Daniel, Kennedy and Spicer1998) and DNA (Stöckler, Daniel & Lockhart, Reference Stöckler, Daniel and Lockhart2002) studies have provided evidence consistent with the conifer genus Agathis being continuously present in New Zealand since the Cretaceous (I. L. Daniel, unpub. Ph.D. thesis, Univ. Canterbury, 1989; Daniel, Reference Daniel2004). Although a case is clearly presented, it hinges on the absence of any extant Australian species that could have shared a common ancestor with living New Zealand kauri, Agathis australis. As mentioned above, it is also possible that suitable ancestral Agathis may have once lived in Australia but are now extinct. Furthermore, according to Pole (Reference Pole2001) there are no pre-Pleistocene fossil records of Agathis in New Zealand and no records of similar Cenozoic araucarian fossils younger than Early Miocene.

The available Cenozoic floral record for New Zealand reflects constant change (Mildenhall, Reference Mildenhall1980; Macphail, Reference Macphail1997) with continuous arrival of immigrant species, particularly from Australia. Most significant is a dramatic change in flora from Oligocene to Miocene time with almost total turnover bar a few exceptions (Couper, Reference Couper1960; Mildenhall, Reference Mildenhall1980; Macphail, Reference Macphail1997).

For the purposes of this research, one of us (JGB) has interrogated the New Zealand Fossil Record File. This is a unique national electronic database of all documented fossil localities in the New Zealand region. It has been operating since the 1950s and is administered by GNS Science and the Geological Society of New Zealand. All data entered for 4602 fossil localities (as at June 2006) of Oligocene to Early Miocene age (the local New Zealand Landon Series comprising Whaingaroan, Duntroonian and Waitakian stages; Fig. 2) have been examined and plotted.

On the basis of this exercise we can confidently conclude that we have not overlooked any known and available palaeontological evidence of terrestrial conditions on Zealandia during the time interval relevant to this study, namely Late Oligocene to Early Miocene time. As expected, this exercise established the status quo: it revealed the palaeontological basis (and bias) for interpretation of continuous land from Oligocene to Miocene time. On close scrutiny, only 84 fossil localities of Late Oligocene to Early Miocene age are interpreted as terrestrial sediments (as opposed to marine) and in all cases the age interpretation is expressed in terms of a large age range spanning much of Oligocene to Early Miocene time. On the basis of this result, we conclude that the age control afforded by known fossil evidence for continuous terrestrial conditions in Zealandia (or New Zealand) from Late Oligocene to Early Miocene time is too imprecise to permit any definitive confidence. If the 84 fossil localities mentioned above are indeed indicative of terrestrial conditions, they could all be of later Early Miocene age and relate to uplift of New Zealand, post-dating maximum flooding of Zealandia. The ages of these fossil biotas, mainly palynomorphs, are just too imprecise.

In the last few decades, two very exciting fossil biotas of Early to Middle Miocene age have been discovered in New Zealand (Lee, Lee & Mortimer, Reference Lee, Lee and Mortimer2001; Worthy et al. Reference Worthy, Hand, Archer, Tennyson, Trewick and Phillips2006a, Reference Worthy, Tennyson, Archer, Musser, Hand, Jones, Douglas, McNamara and Beckb). Both are within lacustrine sequences in Otago, South Island, and have produced diverse terrestrial and freshwater vertebrate fossils (fish, lizard, bird, crocodile and mammal) and also plant fossils (terrestrial and freshwater). Both localities are poorly constrained in terms of age but they are likely to be between 20 and 15 million years old. Research on these sequences and biotas is underway. However, it is unlikely that their interpretation will relate to our understanding of the terrestrial history of Zealandia. These fossil discoveries do not in themselves constitute evidence of continuous land or any direct relationship between Gondwanaland and New Zealand terrestrial biotas.

6.a. Otago and inland Canterbury

This is the area that is most widely regarded as enduring land throughout the Cenozoic, from the break-up of Gondwanaland to the present day (Fig. 1). Portrayal as a substantial Oligocene landmass is based mainly on four factors: widely preserved remnants of the ‘peneplain’ surface, absence of Cenozoic marine strata, an Oligocene unconformity around the periphery of the area, and the perceived need for Oligocene land to support New Zealand's ‘Gondwanan’ biota. These and other factors are discussed in Section 3 (above). In brief, we maintain that (1) the visible ‘peneplain’ remnants were actually created by marine processes, (2) middle Cenozoic marine strata formerly extended into this area but were removed by Neogene terrestrial erosion, (3) the Oligocene unconformity is a submarine feature and (4) there is no biological necessity for a continuously present ‘Gondwanan’ biota.

6.b. The Fiordland island

A middle Cenozoic island is postulated in the Fiordland area (Turnbull & Uruski, Reference Turnbull and Uruski1990; King, Reference King, Cooper and Jones1998). Fiordland (Fig. 1) consists of a Palaeozoic–Mesozoic crystalline massif flanked to the east and south by Cenozoic sedimentary basins and truncated to the west by the Alpine Fault. It is a region of rapid Neogene uplift (Ward, Reference Ward1988) from which Cenozoic cover strata have been stripped by Pleistocene glacial erosion. Thick Cenozoic sedimentary sequences of the Te Anau and Waiau basins (Carter, Lindqvist & Norris, Reference Carter, Lindqvist and Norris1982; Turnbull & Uruski, Reference Turnbull and Uruski1990) lie east of the massif (Fig. 5). They commence with Eocene coal measures and shallow marine sands, passing rapidly upward into Oligocene turbidites and deep-sea mudstone, followed by Late Oligocene–Early Miocene limestone including calc-turbidites and re-deposited sandstone. Along the western side of the basin, thinner sequences lap onto the Fiordland margin. These consist mainly of shelf sandstone and limestone (Carter, Lindqvist & Norris, Reference Carter, Lindqvist and Norris1982). Thick-bedded Oligocene limestone is particularly well developed and conspicuous. Although usually resting on earlier Oligocene sandstone, the limestone unit locally rests directly upon Fiordland crystalline basement. For example, near the summit of Mt Luxmore at 1310 m, a thin fossiliferous conglomerate resting on ultramafic basement grades up into thick bioclastic limestone (Lee et al. Reference Lee, Carter, King and Cooper1983).

We interpret the basal Cenozoic contact at Mt Luxmore as an Oligocene surface of marine planation, the Waipounamu Erosion Surface. The same Oligocene bioclastic limestone, 10–25 m thick, is present elsewhere along the east side of Fiordland. There is no evidence that the present extent of outcrop relates to maximum extent of the sea. We conclude that a carbonate-floored sea must have extended much further west, covering most (if not all) of the Fiordland massif.

What then, is the evidence for a landmass in this region during the Oligocene? Although not discussed by previous workers, the case would appear to depend on the following: (1) absence of any Cenozoic strata in central Fiordland, an area of concordant summit elevations interpreted as a dissected peneplain by Andrew (Reference Andrew1906), Park (Reference Park1921), Benson (Reference Benson1935) and other early workers; (2) the presence of terrigenous clastic sediment within the Oligocene limestone and presence of clastic sediment overlying the limestone and in correlative strata to the east (Turnbull & Uruski, Reference Turnbull and Uruski1990); and (3) recognition of Oligocene shoreline facies along the margin of the Fiordland massif (Lee et al. Reference Lee, Carter, King and Cooper1983).

We will discuss these three points below.

1. (1) Absence of sedimentary rocks from central Fiordland is consistent with documented late Neogene uplift and erosion. It does not necessarily have any bearing on earlier Neogene or Palaeogene palaeogeography. Previous interpretations of the area as a dissected Cenozoic peneplain were based on the belief that prolonged subaerial and fluvial erosion was the most likely means to produce summit concordance.

2. (2) Terrigenous clastic sediment in limestone flanking the Fiordland massif, while compatible with terrestrial erosion from a residual landmass, can also be produced by wave erosion of shoals and by wave and current re-working of pre-existing sands on submarine highs. Thus, terrigenous sand in limestone, as well as terrigenous hemipelagic mud in adjacent basinal deposits, while permissible as evidence for the existence of a persistent residual land area, is not conclusive. A compelling argument would require evidence for first-cycle origin of the sand from a non-marine source and for continuous supply from that source throughout the period of limestone deposition.

3. (3) The recognition of Oligocene shoreline deposits on Mt Luxmore only tells us when the transgressive sea advanced inland from that point. The thick overlying limestones attest to transgression proceeding much further into Fiordland.

Thus the available data do not permit a compelling case for continuous existence of an island in the Fiordland area throughout middle Cenozoic time. Conversely it is clear that Oligocene wave planation occurred along the Fiordland margin and that transgression proceeded well beyond the present outcrops of Oligocene marine sedimentary rocks.

Widespread Pleistocene submarine planation of an older sedimentary sequence is well documented offshore southwestern Fiordland (Sutherland, Barnes & Uruski, Reference Sutherland, Barnes and Uruski2006). Here, on the Puysegur Banks (100–150 m water depth), the modern flat sea-floor truncates a gently dipping Eocene–Pliocene marine clastic sequence. Sediment eroded from the flat banks during eustatic lowstands was re-deposited in deep basins, Puysegur Trench and Solander Trough, flanking the banks on the west and east.

6.c. Structural high in Northwest Nelson

The Tableland of Northwest Nelson forms a remarkable planar upland surface surrounded by rugged mountains. It is cut onto deformed Palaeozoic–Mesozoic basement and is widely regarded as a remnant of stripped peneplain created by long-continued sub-aerial erosion during Late Cretaceous–Early Cenozoic time (e.g. Cotton, Reference Cotton1916; Benson, Reference Benson1935; Suggate, Stevens & Te Punga, Reference Suggate, Stevens and Te Punga1978; Grindley, Reference Grindley1961; Nathan et al. Reference Nathan, Anderson, Cook, Herzer, Hoskins, Raine and Smale1986). The intra-montane Tableland erosion surface, with an elevation of 1100 m, is adjoined by high-relief mountain lands at 1300–1800 m. The two are separated by reverse faults of late Cenozoic age. Extensive areas of planated basement are also well known elsewhere in Northwest Nelson (e.g. Bishop, Reference Bishop1968).

In most areas the Tableland erosion surface is overlain directly by a veneer of Late Oligocene–Early Miocene limestone (Takaka Limestone). These cover strata are richly fossiliferous, deficient in terrigenous clastic detritus and contain large-scale cross-stratification. They were deposited in an offshore marine shelf environment (Grindley, Reference Grindley1980) and although mostly resting directly on basement rock, are locally underlain by thin marine sand and mud. To the south and west, Oligocene marine mudstone underlies the Takaka Limestone. Although Fleming (Reference Fleming1979) regarded the area as being fully submerged by mid-Oligocene time, both Nathan et al. (Reference Nathan, Anderson, Cook, Herzer, Hoskins, Raine and Smale1986) and Kamp (Reference Kamp1986) interpreted the Tableland region as an emergent basement high, persisting as an island until Middle to Late Oligocene submergence.

Nathan et al. (Reference Nathan, Anderson, Cook, Herzer, Hoskins, Raine and Smale1986) interpreted the West Coast region (including Northwest Nelson) as having been ‘emergent and undergoing peneplanation’ throughout Paleocene and Eocene time with the result being a regionally extensive flat to gently undulating erosion surface. In most of the area, this was followed by deposition of coal measures and eventual submergence. However, Nathan et al. (Reference Nathan, Anderson, Cook, Herzer, Hoskins, Raine and Smale1986) noted that coal measures are not present in the Tableland, an area which they regard as having been an island experiencing intense weathering and fluvial erosion until Late Oligocene time, being the last part of Northwest Nelson to become submerged (Nathan et al. Reference Nathan, Anderson, Cook, Herzer, Hoskins, Raine and Smale1986). Similarly, Kamp (Reference Kamp1986) portrayed the area as an Oligocene island within the sea covering his Challenger Rift System.

Our interpretation is in broad agreement with the above workers but differs in that we see no compelling evidence to indicate that the erosion that produced regional flattening of the basement rock surface in the West Coast–Northwest Nelson area was caused by fluvial or other subaerial processes. We note that in areas south of the Tableland (e.g. Buller region), where the unconformity is overlain by non-marine strata, no planar Cenozoic erosion surface has been documented. In fact, Nathan (Reference Nathan1996, p. 28) recorded a ‘local relief of up to 50 m beneath the Buller Coalfield’. Alteration of basement rock underlying the erosion surface in the Northwest Nelson–Buller region has been interpreted as being due to chemical weathering and cited as evidence supporting peneplanation (Nathan et al. Reference Nathan, Anderson, Cook, Herzer, Hoskins, Raine and Smale1986). Elsewhere in the South Island, similar alteration effects have formed along the Waipounamu Erosion Surface unconformity by groundwater alteration following deposition of the Cenozoic cover strata. In contrast, in areas such as the Tableland where the unconformity surface is of conspicuously planar nature, the basal sediments are marine. We conclude that while a ‘mature’ regional landscape formed by Cretaceous–Eocene subaerial processes, the planar surface of the Tablelands was formed by marine erosion bevelling an earlier landscape.

6.d. The Southeast Nelson–North Canterbury island

An Oligocene–Early Miocene landmass approximately 150 × 150 km (Fig. 1) is portrayed in the Southeast Nelson–North Canterbury area on maps of Fleming (Reference Fleming1962), Stevens (Reference Stevens1985) and others. Lying along the southeastern side of the Alpine Fault and within the Marlborough fault zone, this is an area of rapid tectonic uplift and erosion (Wellman, Reference Wellman1979). We are not aware of any evidence to suggest Oligocene land existed in this region, nor are there any published discussions justifying its existence. The only rocks exposed within the area of the putative island are Mesozoic greywacke and schist; any Oligocene cover strata or any remnant erosion surfaces that may have once been present have been removed by erosion during the past 10 million years.

The closest areas of middle Cenozoic strata are found in fault angle depressions along the Clarence Valley (Fig. 1), 20 km southeast of the proposed landmass. Here marine limestone and marl of Late Eocene, Oligocene and Early Miocene age are well exposed. Detailed stratigraphic and palaeontological studies (Reay, Reference Reay1993) indicate that these strata were deposited at outer shelf or greater depths. No correlative shallow-marine or non-marine rocks are recognized. Thus, whereas we cannot eliminate the possibility that land may have been continuously present, there is no evidence to suggest an Oligocene landmass anywhere within this area.

Why then has this island become part of the established New Zealand palaeogeographic surface? Perhaps it is because the Marshall Paraconformity, a sea-floor erosion surface (Carter & Landis, Reference Carter and Landis1972) that is present in the Clarence section, had apparently been interpreted as evidence for subaerial erosion by Fleming (Reference Fleming1962; see also Section 3.a above), combined with a perceived need for land at this time (see Section 3 above, especially points a, b and g). This ‘absence of evidence’ for Oligocene marine conditions in this area cannot be construed as ‘evidence of absence’.

6.e. North Island peninsulas and islands

Two North Island Oligocene land areas, bounding the North Wanganui Basin (Fig. 1) to the east and west have long been postulated. Although shown as islands by Fleming (Reference Fleming1962, Reference Fleming1979), McQuillan (Reference McQuillan1977), Stevens (Reference Stevens1985) and Nelson (Reference Nelson1978), more recent workers have tended to attach these ridges to a larger landmass to the south (Kamp, Reference Kamp1986; King, Reference King2000). Thus two northward-pointing peninsulas are postulated (King et al. Reference King, Naish, Browne, Field and Edbrooke1999; Lee, Lee & Mortimer, Reference Lee, Lee and Mortimer2001) for the period 20–30 Ma. In contrast, an early summary by Suggate, Stevens & Te Punga (Reference Suggate, Stevens and Te Punga1978) states: ‘Land had probably disappeared by Late Oligocene (Waitakian), and flaggy argillaceous limestone of that age is widespread’. Thus Suggate, Stevens & Te Punga (Reference Suggate, Stevens and Te Punga1978) show eastern and western ridges in the Whaingaroan Stage but complete immersion for the latest Oligocene Duntroonian Stage. Morgans et al. (Reference Morgans, Edwards, Scott, Graham, Kamp, Mumme, Wilson and Wilson1999), for the Waitakian Stage, show only two small islands lying east of the Wanganui Basin with no land to the west.

The case for Oligocene landmasses in central and western North Island rests mainly on two factors: (1) isopach maps showing thinning of Wanganui Basin strata toward the west and east and (2) presence of terrigenous sediment of Oligocene age in basinal sequences. In addition, a perceived need for middle Cenozoic terrestrial habitat (‘Moa's Ark’) may be an unacknowledged factor. Oligocene isopachs constructed by McQuillan (Reference McQuillan1977), Nelson (Reference Nelson1978), Suggate, Stevens & Te Punga (Reference Suggate, Stevens and Te Punga1978) and others clearly demonstrate the existence of a N–S-trending trough (North Wanganui Basin) thinning toward structural highs on either side. However, inferred thinning to zero thickness and the presence of persistent land areas is based on extrapolation. Although areas of emergence might have been argued from sequences of contemporaneous non-marine sediment or even persistent supply of local first-cycle terrigenous sediment to the Wanganui Basin, neither of these has been recognized.

Emergence of land to the west (the Herangi High, Fig. 1) is implied by terrigenous conglomerate lenses within Oligocene limestone along the west side of the North Wanganui Basin (Nelson, Reference Nelson1978). These conglomerates, derived from Mesozoic source rocks, comprise minor (<1%) but distinctive beds scattered through a predominantly limestone sequence. The Oligocene limestone is biogenic and inferred to have been deposited under shelf conditions (Nelson, Reference Nelson1978). Westernmost outcrops of the limestone sequence are commonly more than 100 m thick, characterized by shelf fossils, and lacking in evidence for a continuously present adjacent landmass (e.g. terrigenous sand dominated sequences, terrestrial plant detritus). Furthermore, at many localities the marine sequence, including limestone, rests directly on eroded Mesozoic basement. Apart from a small area at the southeastern end of this putative Oligocene island, the proposed landmass lies off the west coast of the present North Island (Fig. 1). Although the location and extent of the island have not been discussed, we note that it coincides approximately with a chain of seamounts lying parallel to the coastline. Recent work (e.g. King & Thrasher, Reference King and Thrasher1996; Hayward et al. Reference Hayward, Black, Smith and Balance2001) indicates that these are Miocene sea-floor volcanoes.

An equally likely scenario would suggest a much smaller Oligocene submarine ridge (Herangi High) flanked by carbonate seas. Brief episodes of sea-level fall or tectonic uplift probably created shoals or small islands from which conglomerates may have been derived. However, there is no evidence to indicate a substantial Herangi land area that persisted continuously throughout Late Oligocene and into Early Miocene time.

Despite the great predominance of carbonate-rich sediments in Late Oligocene–earliest Miocene strata of the North Wanganui Basin, most of these limestones also contain significant terrigenous mud and silt (Nelson, Reference Nelson1977, Reference Nelson1978). Terrigenous mud is particularly abundant in limestones in the central part of the basin. The cleanest limestone formation of the Te Kuiti Group is also the youngest: the earliest Miocene (Waitakian) Otorohanga Limestone. Nelson (Reference Nelson1977) described the Otorohanga as a pure limestone in which terrigenous sediment seldom exceeds 10% suspension deposit, so there appears to be no evidence for an adjacent landmass at the time of its deposition.

Thus a source of fine-grained terrigenous sediment was available during much of Oligocene time in the Waikato area. Less clear, however, is whether this source was proximal to the basin, whether it was of terrestrial (subaerial) origin, and whether it existed continuously into Early Miocene time. Reworking of marine mud by waves, bottom cements, and turbidity currents is well known, as is long-distance re-distribution. Origin of the Te Kuiti Group muddy terrigenous sediment remains unclear.

An Oligocene–Early Miocene landmass east of the Wanganui Basin in the region of present central North Island has also been postulated by many workers. It is shown variously as an Oligocene island approximately 400 km long and 150 km wide flanking the east side of the Wanganui Basin (Fleming, Reference Fleming1979) and as a 150 km wide portion of a large landmass extending continuously from Auckland to southern South Island (Kamp, Reference Kamp1986). More recently, King (Reference King2000) and Lee, Lee & Mortimer (Reference Lee, Lee and Mortimer2001) have portrayed a somewhat reduced Oligocene New Zealand land area with a peninsula about 150 km long and 50 km wide along the east side of the North Wanganui Basin.

Recognition of this Late Oligocene–earliest Miocene landmass is not based on firm evidence. The area east of the Wanganui Basin presently consists of pre-Cenozoic basement with Pleistocene volcanic rocks and cover-strata; no middle Cenozoic strata are known here. Along the western side of the Rangitoto Range (Fig. 1), Oligocene coal measures (Edbrooke, Sykes & Pocknall, Reference Edbrooke, Sykes and Pocknall1994) rest unconformably on Mesozoic greywacke basement. They are overlain by later Oligocene and Early Miocene marine mudstone and sandstone. No evidence has been presented to show that these younger strata represent the margin of a marine basin or that land existed continuously anywhere in the central North Island. It is entirely possible that the sea covered this entire area during Late Oligocene and Early Miocene time when limestone was widely deposited throughout large areas of the North Island. Indeed, Suggate, Stevens & Te Punga (Reference Suggate, Stevens and Te Punga1978) interpreted the area as inundated by Oligocene seas.

Thus the case for continuous existence of a central North Island Cenozoic land area is not established. Certainly a source for siliciclastic sediment in the Te Kuiti Group limestone is required, but there is no reason to suggest that this source existed continuously as flanking subaerial landmasses throughout the middle Cenozoic.

6.f. The Auckland–Coromandel region of the North Island

In the Auckland–Coromandel Peninsula region (Fig. 1), Neogene erosion and volcanism have removed or buried much of an inferred widespread earlier Cenozoic cover sequence. However, isolated remnants record a regional transgression beginning in the Eocene and peaking in the earliest Miocene (Nelson, Reference Nelson1978; Dix & Nelson, Reference Dix and Nelson2004).

Originally shown as part of a large Oligocene island (Fleming, Reference Fleming1962; Fig. 1), the region around Auckland has been more recently interpreted as the distal end of a northward-pointing peninsula (Kamp, Reference Kamp1986; King, Reference King, Cooper and Jones1998; see Section 6.d) and as a smaller island or group of island ridges (Isaac et al. Reference Isaac, Herzer, Brook and Hayward1994). Here, Early to middle Oligocene marine sediments were deposited on eroded greywacke basement within erosional lows and local graben. Youngest Oligocene and basal Miocene sediments are not recognized in this area but at some localities, basement is unconformably overlain by freshwater to shallow marine sediments of basal Waitemata Group (Early Miocene; 22–21 Ma). Locally, the greywacke basement shows some 50 m of topographic relief (B. W. Hayward, pers. comm. 2007). These relations, plus the absence of Late Oligocene sediments (both marine and non-marine) from the same area, may be evidence for land at this time (e.g. Isaac et al. Reference Isaac, Herzer, Brook and Hayward1994).

However, it is also possible that marine conditions existed during Late Oligocene time (Duntroonian–Waitakian) but did not leave any preserved record. Evidence for bottom-scouring currents at this time is widespread elsewhere in the New Zealand area (Carter & Landis, Reference Carter and Landis1972; Fulthorpe et al. Reference Fulthorpe, Carter, Miller and Wilson1996), and there are good examples of regions being submerged throughout the Oligocene and Early Miocene but leaving little or no surviving sedimentary record. Overlying the basal Waitemata Group, younger Waitemata sediments are bathyal–abyssal turbidites that imply rapid subsidence of the area between 22 and 19 Ma (Isaac et al. Reference Isaac, Herzer, Brook and Hayward1994). Thus any original Late Oligocene–Early Miocene (27–22 Ma) marine sequence may be represented by a non-depositional hiatus (paraconformity) or have been eroded away prior to deposition of the basal Waitemata Group beds (22–21 Ma).

Study of remnant outliers of Oligocene strata in the Coromandel Peninsula (directly east of Auckland; Fig. 1) shows a basal unconformity overlain by fan-delta sediments of Early to middle Oligocene age that are in turn overlain by younger Oligocene shallow marine limestone (Dix & Nelson, Reference Dix and Nelson2004). A prominent intra-formational erosion surface separates the lower shelf clastic sediment and limestone from overlying deep-water Oligocene–Early Miocene limestone (Dix & Nelson, Reference Dix and Nelson2004). This surface, interpreted as a sequence boundary by Dix & Nelson (Reference Dix and Nelson2004), separates a highly variable, carbonate-dominated, transgressive basal marine sequence from overlying more uniform, and slightly less steeply dipping, deep sea carbonate sequence. Limestone-forming conditions persisted from Late Oligocene (24 Ma) to Early Miocene time (21 Ma).

6.g. Putative landmasses north of New Zealand

Herzer (Reference Herzer, Cooper and Jones1998, Reference Herzer and Thomas2003) and Lee, Lee & Mortimer (Reference Lee, Lee and Mortimer2001) have proposed that land bridges and stepping-stone islands existed north of New Zealand during the Oligocene and Early Miocene. Herzer (Reference Herzer, Cooper and Jones1998, p. 47) maintains that successive uplifts of the sea-floor favoured ‘one-way north-to-south’ migration pathways enabling species to colonize New Zealand from the New Caledonia region. In contrast, Lee, Lee & Mortimer (Reference Lee, Lee and Mortimer2001, p. 349) argued that the ‘emergent Norfolk Ridge provided a near-continuous land connection between New Zealand and New Caledonia’, permitting species migration south to north during this same time.

Evidence for middle Cenozoic land areas to the north of New Zealand includes firstly the recognition of rounded pebbles and a shoal fauna collected in dredge hauls from submarine highs, and secondly in the recognition of extinct volcanic arcs which were active at that time (Mortimer et al. Reference Mortimer, Herzer, Gans, Parkinson and Sewart1998; Herzer & Mascle, Reference Herzer and Mascle1996; Herzer, Reference Herzer and Thomas2003; Meffre, Crawford & Quilty, Reference Meffre, Crawford and Quilty2006). Thus Lee, Lee & Mortimer (Reference Lee, Lee and Mortimer2001) showed several ‘probable to possible’ islands and one ‘certain to probable’ large island (600 × 300 km) south of New Caledonia, while Herzer (Reference Herzer, Cooper and Jones1998; see also King, Reference King, Cooper and Jones1998) showed islands hundreds of kilometres long and up to 80 km wide during the Oligocene. Meffre, Crawford & Quilty (Reference Meffre, Crawford and Quilty2006) have provided further evidence of an island located in the south Norfolk Basin. All of these papers regard these islands as being largely submerged by Late Miocene time.

We agree that shallow marine conditions existed locally along the Norfolk Ridge during the middle Cenozoic and that some volcanic islands were probably present. However, in the papers referred to above, we find no compelling evidence requiring persistent large islands (land bridges or stepping-stones) for significant periods of time. New Caledonia was totally submerged in the Eocene but may have become emergent in the Late Oligocene. The evidence for this is the presence of inferred laterite-derived soil detritus in Oligocene marine sediments (Paris, Reference Paris1981; P. Maurizot, pers. comm. 2005). Nevertheless, the existence of a substantive and continuous middle Cenozoic landmass at New Caledonia, the postulated source and sink for New Zealand biota, must be questioned. There are neither offshore nor on-land data (e.g. Aitchison et al. Reference Aitchison, Clark, Meffre and Cluzel1995) to suggest the existence of an island at New Caledonia prior to Late Oligocene time. Like New Zealand itself, the existence of Oligocene land areas to the north appears to be rooted in the perceived need for biological sources and sinks rather than firm geological evidence.

6.h. Volcanic islands

Several occurrences of Oligocene volcanic rocks are recognized in the South Island (Suggate, Stevens & Te Punga, Reference Suggate, Stevens and Te Punga1978). All described pyroclastic and effusive rocks of Oligocene age are submarine in origin. However, it is possible that some of these volcanoes created hitherto unrecognized islands that may have provided short-lived refugia for terrestrial organisms.