Discussion

Biostratigraphy

The palynological samples show a striking variation in overall dominance and taxonomic composition. They clearly range from the (at least) mostly Paleocene PM3 to the Early Eocene MH1 assemblage zones of Raine (1984). Zone PM3 is conifer-dominated with Phyllocladidites mawsonii Cookson ex Couper (1953) usually as the dominant species. The angiosperm Tricolpites secarius McIntyre (1965) was common in some samples - its first appearance defines the lower boundary of the PM3 Zone and it extends up into the mid-late Eocene. The uppermost part of Zone PM3 is indicated by the appearance of Malvacipollis subtilis Stover in Stover and Partridge (1973) and Cupanieidites orthotheichus Cookson and Pike (1954). Spinozonocolpites prominatus ((McIntyre) Stover and Evans (1973), representing the mangrove palm, Nypa) appears at about the Paleocene-Eocene boundary in the North Island (Crouch and Visscher 2003), and thus in latest PM3 time. It is found in younger sediments elsewhere in New Zealand (Raine 1984; Pocknall 1990). The lower boundary of Zone MH1 is "provisionally taken as the lowest level of" Myricipites harrissi (Couper) Dutta and Sah (1970) "in abundance" (Raine 1984), and there is a clear increase in this taxon to outright dominance at Kakahu. Zone MH1 is also characterized by diverse Proteaceae, while gymnosperms are not usually common. At Kakahu there are pollen assemblages lying between those clearly belonging to the PM3 and MH1 Zones, which do not clearly belong in either. The samples lack both dominance by conifers and M. harrissi and therefore are here provisionally placed into an 'interzone'.

Raine (1984) regarded the Paleocene-Eocene boundary as lying within the PM3 palynological zone. The boundary between his PM3 and the overlying MH1 Zone lies within the New Zealand Waipawan Stage. The Paleocene-Eocene boundary has been placed at the boundary of the Teurian and Waipawan local stages by Hollis (2006) but has also been regarded as lying within the Waipawan Stage (Edwards 1982) or within the latest Teurian Stage (McGowran et al. 2004). This uncertainty reflects the fact that the Paleocene-Eocene boundary is apparently not prominently marked by palynological changes in New Zealand. This conclusion was supported by Crouch and Visscher (2003) who documented the palynology across the boundary in marine sediments of the North Island. They found little appreciable taxonomic change, with Dilwynites granulatus Harris (1965) dominant throughout, along with spores of Cyatheaceae, Gleicheniaceae, and bisaccate Podocarpaceae. Haloragacidites and Phyllocladidites mawsonii were uncommon, sometimes not appearing in a count at all. Proteaceae were also uncommon, forming only 1-2% of some counts. Crouch and Visscher (2003) noted that these results supported conclusions elsewhere in the world, and that the period of greatest vegetation change lies within the Early Eocene, rather than at the Paleocene-Eocene boundary. However, Raine (in Cooper 2004) regarded the lowest occurrence of Cupaneididites orthotheichus as approximating the Paleocene/Eocene boundary and used it to define the latest part of Zone PM3 as an Eocene Subzone PM3b. C. orthotheichus is present in samples Kakahu-12, 14, and 16, which are clearly zone MH1. It is also present in Kakahu-3 and 19, which fall into the interzone as recognised here, and a concerted search in sample Kakahu-2, about 18 m below Kakahu-3, failed to locate any. It is also absent in all samples within and below the coal seam at the Insulators Pit. It is possible therefore that the Paleocene-Eocene boundary at Kakahu falls within the interzone. It could further be speculated that as the interzone is a relatively transient phase between the PM3 and MH1 Zones, it may represent an equally transient or unstable climate, which was fundamental to the PM3 - MH1 boundary. Thomas et al. (2006) suggested that the Early Eocene Climate Optimum may have been a time of alternating warm and very warm conditions – which might yet hold a key to the disagreements over the precise location of major vegetation and biodiversity changes with respect to the boundary.

Pocknall (1990) summarised the palynology of the Early-Middle Eocene of New Zealand and discussed other samples, including Waipara River, Otaio Gorge, Waihao River, and Boundary Creek. The nearest quantitative comparison for the Paleocene is provided by the Mt Somers Coal Mine (Raine and Wilson 1988), which was considered to fall entirely within Zone PM3, although the upper part of the sequence was regarded as close to the upper boundary of the zone, and probably within the Waipawan zone. Mt. Somers shows a typical dominance by saccate coniferous pollen, contrasting strongly with the typically less than 10% in the succeeding MH1 Zone. Raine and Wilson (1988) did not record any Araucariacites australis, (Cookson 1947) although Dilwynites granulatus made a low-abundance appearance in their uppermost samples. Casuarinaceae was about 5% or less, and no samples showed any sign of dominance by Tricolpites secarius. The presence or absence of charcoal was not reported.

Throughout New Zealand, the lower boundary of the MH1 appears to be within the Waipawan. In Canterbury, coal measures at Boundary Ck are dominated by Podocarpaceae, Liliaceae, and Nypa, and rare Phyllocladidites and Casuarinaceae (McIntyre 1965; Raine 1984, Pocknall 1990), are overlain by well-dated Waipawan marine sediments dominated by Casuarinaceae (Couper 1960; Pocknall 1990).

Dinoflagellate spores are common in the stratigraphically highest samples, consistent with other evidence that deposition was in a near-coastal environment, and that the area was submerged under a regional marine transgression. Sporadic pollen of the palm Nypa (Spinozonocolpites prominatus) confirms that mangrove vegetation was present. The sedimentological relations of this pollen type along with associated macrofossil material in the Early Eocene of Tasmania (Pole and Macphail 1996) support the interpretation of similar mangrove ecology of the fossils to extant Nypa. The pollen is a rare component of those Kakahu samples in which it is found but this is a widespread phenomenon discussed by Frederiksen (1985), who noted that only a few pollen grains may still be indicative of a broad belt of Nypa mangroves.

Leaves and Cuticle

Although leaf cuticle was abundant in some samples, it was always low in diversity, or at least overwhelmingly dominated by one type. It showed nothing of the diversity which the leaf impression assemblages of Pole (1997) suggested. It is possible that in some cases weathering has removed all but the most robust cuticles. However, in some samples cuticle is abundant and well-preserved, and the sediment is an organic-rich grey mud with no sign of any strong weathering. This suggests that at least in these cases, the low diversity is genuine. One of these, sample Kakahu-09, consists entirely of small Lauraceae leaves, and another, Kakahu-28, is overwhelmingly dominated by conifers. This in turn suggests that there may have been a broad ecological/spatial distinction between angiosperm and conifer-dominated vegetation within the floodbasin. There is no evidence that this was related to sedimentary facies, but perhaps is due to water-table level, or successional status. For instance the low diversity may be a function of a "backswamp" habitat – as noted in modern environments (e.g., Frye and Quinn 1979). In the Paleocene of the Clark's Fork basin of the USA, Hickey (1980) found assemblages from the backswamps to be dominated by conifers and low in diversity, contrasting with the better drained habitats nearer the river channels. However, Wing et al. (1995) in the USA found the diversity difference to be insignificant in the Bighorn Basin of comparable age.

Charcoal

Fragments of charcoal are abundant in most palynological preparations and occasionally are large enough to be seen in the field. They show the homogenized cells walls which are one of the diagnostic characters of charcoal (Jones and Chaloner 1991) (Figure 15.1). Among the relatively uncommon fragments which show structure, the most common morphology is angiosperm with scalariform pitting. Some fragments of perforation plates have more than 20 bars, which is regarded as a uncommon feature and therefore of taxonomic importance (Wheeler and Baas 1998; Carlquist 2001; see McIver 1999 fig. 9 for similar material in the Canadian Paleocene). This morphology is most typical of relatively primitive angiosperms and would be consistent with the Lauraceae, which is so prominent in the cuticle fraction. The range of morphology (Figure 15.2-15.3) within this charcoal type could reflect the difference between the perforation plate and lateral wall within a single species (e.g., Carlquist 2001; Meylan and Butterfield 1972). Smaller and more scattered pits (Figure 15.4-15.6) may represent cross-field pitting, or other species. Conifer charcoal is uncommon and represented by cells with large, uncompressed, circular, bordered pits, consistent with (but not limited to) Podocarpaceae (Figure 15.7-15.8), and smaller bordered pits which may be cross-field pits.

The abundant presence of charcoal in many of the Kakahu samples was surprising and has ecological and possible climatic implications. Fire in New Zealand has been a relatively unusual event. For instance, charcoal is abundant in the Cretaceous of Horse Range and Kaitangata (Pole, pers. obs.), although it seems to be mainly detrital fragments of a few millimeters dimension, rather than the ubiquitous microscopic size range found at Kakahu. It is clear that fire was an important environmental factor influencing the Kakahu vegetation. Fire then seems to have been absent throughout much of Eocene times (Pole, unpublished data). New Zealand's forests in the Oligocene-early Miocene did not burn, as charcoal is essentially absent, although burning did resume later in the Miocene (Mildenhall 1989), probably near the Middle Miocene boundary (Pole and Douglas 1998; Pole 2003). New Zealand forests today often have a high gymnosperm component, but they are ever-wet, and although they will burn with anthropogenic intervention, natural fires as a result of lightning strikes, are rare and localized (McGlone 1988).

Where structural elements of the charcoal allow identification it can be seen to be overwhelmingly angiosperm. Most identifiable fragments are angiosperm vessels with many scalariform perforations, which are taxonomically uncommon. Conifer charcoal is much rarer. Thus it appears to have been mainly angiosperm vegetation which was burning. The burning was a feature of the landscape for a considerable period of time as the charcoal extends across at least two palynological zones and the putative interzone. There is no clear pattern to the amounts of charcoal. Sample Kakahu-17, in which no charcoal was recorded, is stratigraphically close to samples Kakahu-15, which has more charcoal than pollen, and 16, in which it is common. Each of the three palynological zones has a sample in which the amounts of charcoal are very high (Kakahu-1, 8, and 24). The diverse, angiosperm-dominated impression assemblages of Kakahu-3, and 28 both have charcoal, although in relatively small amounts, as does the conifer-dominated cuticle assemblage of Kakahu-28. In general, a persistent fire regime appears to have been present, at least for some component of the Kakahu ecosystem. Further study might be able to identify which parts of the landscape were burning. After much debate it is now clear that peat-forming mires (i.e., the precursors of coal) have a long history of fire (Scott 1989, 2000; Scott and Jones 1994). They are inherently flammable and are prone to drying-out. It might be speculated that the Kalahu peat-mires burnt frequently, while the vegetation on the levees and back-swamps was relatively fire-free.

Prominent amounts of charcoal across the P-E boundary have been reported in England by Collinson et al. (2003, 2007), who interpreted it in terms of repeated fires in a flowering plant dominated vegetation. Similar material has been recorded elsewhere, for instance Crosdale et al. (2002) reported large amounts of inertinite (fire-induced) in more broadly dated Paleocene coals from far eastern Russia. It is intriguing to think that these fires may have been one of the environmental responses to the events of the Paleocene-Eocene Thermal Maximum. Based on a study of the global carbon and sulfur cycles, Kurtz et al. (2003) have suggested that the prominent negative carbon isotopic excursion near the Paleocene–Eocene Thermal Maximum may result from a "global conflagration," essentially the sustained burning of accumulated Paleocene peat, which may have been triggered by a change to a drier climate. However, neither Collinson et al. (2003, 2007) nor this study found burnt Paleocene peat, but as those authors pointed out, widespread burning of standing vegetation may still be implicated in the changes across the P-E boundary.

The fossil record suggests that New Zealand's vegetation in the Paleogene had a significant component of what would be regarded as 'sclerophyll' in the Australian sense, and probably 'pyrophytic' as well. This may also have encouraged the ignition of fires in an essentially ever-wet climate. It is clear from the charcoal record that both woody gymnosperms and angiosperms were burning. However, burning did cease later in the Cenozoic, when many of these sclerophyll elements were still present.

There are clear peaks of charcoal in the mid-part of the Kakahu sequence, although more samples would be needed to understand if this actually reflected a more fire-ridden landscape at the time. There is no indication of a relationship with overall floristic content. Perhaps more important is the observation that charcoal is present throughout the sequence.

Climate

The Paleocene-Eocene has long been considered to represent an important transition in vegetation, associated with the rise to the peak in global warmth in the Early Eocene. However, the boundary is often not obviously marked and the biotic changes associated with it are mostly gradual. The abrupt, but transient Paleocene-Eocene Thermal Maximum (PETM) or Initial-Eocene Thermal Maximum (IETM, Schmitz et al. 2001) is now regarded as marking the P-E boundary at about 55 Ma (it was earlier thought to lie within the late Paleocene and was then termed the "Late Paleocene Thermal maximum," or LPTM; Zachos et al. 1993). The actual peak of global warmth is thought to have been about 52-53 Ma, about 2-3 Ma into the Eocene (Wing 1987).

The presence of a Paleocene-Eocene boundary section in New Zealand, containing plant fossils, is potentially important in helping to understand what happened over this period. At Kakahu, the foliar physiognomy of the leaf impressions described by Pole (1997) may help quantify the paleoclimate in the latest Paleocene (locality J38/f77) and in the interzone, close to the Paleocene-Eocene boundary (J38/f058). Since that publication, Greenwood et al. (2004) have published leaf margin-mean annual temperature (MAT) regressions based on species inventories for the rainforests of eastern Australia. They suggested that they be used in addition to those of Wolfe (1979), Greenwood (1992), Wing and Greenwood (1993) and Wilf (1997). The use of several equations would have the effect of broadening the range of MAT estimates. On this basis, the leaf assemblage from J38/f058 (12 species, 33% entire-margined) suggests a MAT of 7-9 °C (Greenwood et al. 2004 equation) or 11-12 °C (Wilf 1997; Greenwood 1992; Wolfe 1979; Wing and Greenwood 1993 equations). However, the low number of species it is based upon means the sampling error (Wilf 1997) is in the order of 4 °C. Similarly, the assemblage from J38/f077 (10 species, 30% entire-margined) suggests a MAT of 6-8 °C or 11 °C and a comparable sampling error. The MAT estimates from both assemblages are not statistically distinct. Both assemblages are well below the 20, or even 30 species generally regarded as necessary for reasonable accuracy. Combined, the two assemblages contain 18 species, with 22-28% entire margins, giving a MAT of 4-8 °C or 8-11 °C and a sampling error nearer to 3 °C. The overall suggestion is thus of cool to mild temperatures (1-15 °C). Despite this estimate being so broad, it is still distinctly cooler than the 20-25 °C estimated for the Late Paleocene-Early Eocene by Hornibrook (1992).

The average leaf length is approximately 76 +-5 mm, which is close to the boundary between the microphyll and notophyll leaf size classes (Webb 1959). According to both Greenwood (1992) and Carpenter et al.'s (1994) data for eastern Australian rainforest leaf litter, this correlates with mean annual temperature of about 17 °C (although with a broad range of accuracy of around 12-20 °C). Thus there is overlap with the estimates from leaf margin proportions, but it also suggests caution. Leaf litter size may be suggesting warmer temperatures than inventory-based leaf margin data.

There are no clear features which allow identification of the paleosols in the manner of Retallack (2001) (the soils may be too immature), and the chemical and microscopic features which Retallack (1997) indicated had become necessary to classify paleosols into the current US Soil taxonomy are beyond the scope of this project. However, the grey muds in the lower part of the section (including samples Kakahu-08 and Kakahu-28) are best interpreted as gleyed, reducing, anoxic soils of the poorly-drained backswamps. The coals are the relatively raised peats of the interfluve areas. The fissile mudstones with the leaf impression assemblages of Pole (1997) are associated with fluvial channels and are considered to preserve vegetation growing on the relatively richest soils of the region situated on the levees. This is unlikely to be such an extremely poor situation that would significantly alter the foliar physiognomy, such as reducing leaf size. In one of the more profound conclusions for paleobotany Burnham (1994) found that the proportion of a species leaf area in litter is proportional to its trunk basal area in the source forest. With a total of 18 leaf taxa distinguished from a collection of about 45 specimens, the levee vegetation of Pole (1997) was certainly diverse.

Further estimates of climate are suggested by key taxa. Extant Nypa is restricted to truly tropical locations, and the presence of Nypa in the New Zealand fossil record is generally taken to indicate conditions of peak warmth in the overall Cenozoic succession. However, temperatures were clearly less than truly tropical. The record of Nypa from Tasmania (Pole and Macphail 1996) emphasises that the Spinozonocolpites prominatus (McIntyre 1965) Stover and Evans (1973) pollen morphology, in addition to the living species, covers at least one extinct species, Nypa australis, and that this had a cooler, temperature tolerance to the extant species, Nypa fruticans (Pole and Macphail 1996).

Despite the evidence for burning, rainfall was evidently high at times, based on the presence of coal. A key taxon also supporting this is the pollen type Phylocladidites mawsonii, which was often important in the early Cenozoic of the Australasian region. Based on its highly distinctive morphology, it has been linked to the extant conifer Lagarostrobos (Playford and Dettmann 1979), which is currently restricted to the wettest habitats of Tasmania. However, increasing macrofossil data are suggesting that this pollen type may have been produced by other, extinct genera (Pole 1998a). Nevertheless, in Australia, P. mawsonii contracts southwards over the Cenozoic as the bulk of the country dried out (Kershaw 1988). So, whatever its botanical affinities, it still appears to be an indicator of cool, very wet conditions. The appearance pf Cupaneidites is usually taken as the onset of relatively warm conditions in New Zealand. Muller and Leenhouts (1976) regard this pollen type as representing their pollen group B of tribe Cupanieae in the Sapindaceae, including genera currently extending from the tropics to subtropics in Australia.

It is possible that climate thresholds, which resulted in major vegetation change, occurred at different places in different latitudes. In New Zealand, the switch to a Haloragacidites harrisii-dominated pollen-rain is dramatic (Raine 1984; Pocknall 1990). In Australia expansion of H. harrisii can be detected, but it is no-where near as obvious, at least not enough for it to warrant being used there as a palynological zone boundary (Harris 1965). In North America and Europe there was a general shift to warmer types of vegetation, and deciduous vegetation was restricted to >70° N, or to the continental interiors. In other places deciduous was replaced by evergreen. Recent work by Wing et al. (1995) focusing on the Paleocene-Eocene boundary has highlighted some interesting phenomena, for instance that mammal diversity increased, while plant diversity dropped. They argued that although global climate warmed, in theory promoting higher diversity, because of the rate of climate change, the initial result may have been a drop in biodiversity.

In New Zealand the shift from conifer-dominated vegetation to one dominated by angiosperms, and specifically by Casuarinaceae, probably reflects warming and drying into the Early Eocene. This is supported by the large amounts of charcoal, which indicate that fire was a major component of the ecology. The initial conifer-dominated vegetation was probably a closed forest, but the nature of what followed it is much less certain. If the pollen species Halorgacidies harrissii does represent Casuarinaceae, then its dominance in Zone MH1 suggests dry woodland. This is unlikely, given that elsewhere in New Zealand it dominates in samples from coal and thus grew in a wet climate. The Casuarinaceae is predominantly a family of dry vegetation, but it does include Gymnostoma, which is sometimes referred to as the 'rainforest casuarina'. Gymnostoma is the most primitive genus (Johnson and Wilson 1989) and both Australian, New Zealand and South American macrofossil evidence supports that at least some of the Halorgacidies harrissii pollen was produced by this genus (the pollen produced by the different genera is practically indistinguishable). However, it typically grows as a small component of rainforest. The dominance of this pollen type throughout New Zealand for a period of several million years suggests that the parent plant was more than a minor component. In recent pollen assemblages in Australia today, those that are dominated by Casuarinaceae pollen, were deposited under Casuarinaceae-dominated vegetation. This raises a question, does this pollen assemblage represent a unique vegetation dominated by Gymmostoma, or was the pollen produced by Casuarinaceae at all? While the affinities of Halorgacidies harrissii are generally accepted as Casuarinaceae, other possibilities exist, such as Canacomyrica, of the Myricaceae (listed as pollen from the New Zealand Cenozoic by Mildenhall 1980) which, like the Casuarinaceae, is an essentially nitrogen-fixing pioneer. Some fossil leaves from Kakahu (Dryandroides comptoniaefolia von Ettingshausen 1887 = Parataxon TARA-34 Pole 1997) are remarkably similar to extant Comptonia peregrine (L.) Coulter (1894), also of the Myricaceae, and it is unfortunate that no cuticle has been obtained from the fossils. That the fossils have such a delicate cuticle is suggestive that they are not Proteaceae (see further discussion of this taxon in Pole 1998b), in which it is typically robust.

A palynological sequence of comparable age to Kakahu is present in South Australia (Harris 1965). It also shows a shift away from conifer dominance and a rise in Casuarinaceae, which is thought to correlate with that in New Zealand (Raine 1984). However, there are significant differences as Casuarinaceae never reaches the overwhelmingly dominant amounts that it does all over New Zealand, and the conifers still remain important. This suggests that South Australia remained wetter than New Zealand, possibly because of its more southerly situation. In Tasmania the Paleocene-Eocene boundary is not known, but Early Eocene sediments at Regatta Point preserve a conifer-dominant vegetation, which grew inland of a belt of Nypa-mangroves (Pole and Macphail 1996; Pole 2007). Thus temperatures were warm, but the abundance and diversity of conifers suggests it was still very wet, and there is no sign of charcoal.