Terrestrial climate evolution in the Southwest Pacific over the past 30 million years

JID:EPSL

AID:14170 /SCO

[m5G; v1.191; Prn:30/11/2016; 15:37] P.1 (1-9)

Earth and Planetary Science Letters ••• (••••) •••–•••

Contents lists available at ScienceDirect

Earth and Planetary Science Letters www.elsevier.com/locate/epsl

Terrestrial climate evolution in the Southwest Pacific over the past 30 million years Joseph G. Prebble a,∗ , Tammo Reichgelt b , Dallas C. Mildenhall a , David R. Greenwood c , J. Ian Raine a , Elizabeth M. Kennedy a , Hannu C. Seebeck a a b c

GNS Science, P.O. Box 30-368, Lower Hutt, New Zealand Lamont–Doherty Earth Observatory of Columbia University, PO Box 1000, Palisades, NY 10964-8000, USA Department of Biology, Brandon University, 270 18th Street, Brandon MB, R7A 6A9, Canada

a r t i c l e

i n f o

Article history: Received 1 July 2016 Received in revised form 21 October 2016 Accepted 4 November 2016 Available online xxxx Editor: H. Stoll Keywords: pollen paleoclimate Neogene bioclimatic analysis New Zealand

a b s t r a c t A reconstruction of terrestrial temperature and precipitation for the New Zealand landmass over the past ∼30 million years is produced using pollen data from >2000 samples lodged in the New Zealand Fossil Record Electronic Database and modern climate data of nearest living relatives. The reconstruction reveals a warming trend through the late Oligocene to early Miocene, peak warmth in the middle Miocene, and stepwise cooling through the late Neogene. Whereas the regional signal in our reconstruction includes a ∼5–10◦ northward tectonic drift, as well as an increase in high altitude biomes due to late Neogene and Pliocene uplift of the Southern Alps, the pattern mimics inferred changes in global ice extent, which suggests that global drivers played a major role in shaping local vegetation. Importantly, seasonal temperature estimates indicate low seasonality during the middle Miocene, and that subsequent Neogene cooling was largely due to cooler winters. We suggest that this may reflect increased Subantarctic influence on New Zealand vegetation as the climate cooled. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Near-drowning of New Zealand during the Oligocene (Landis et al., 2008; Strogen et al., 2014) was followed by a period of tectonic uplift that began approximately at 25 Ma with reactivation of movement along the Pacific–Australian plate boundary (Batt et al., 2004; King, 2000). The New Zealand landmass has since consisted of a series of evolving archipelagos located in the mid-latitudes of the south west Pacific (Lee et al., 2001; Mildenhall and Pocknall, 1984). Due to its relatively small land mass, New Zealand’s terrestrial climate has been closely coupled to conditions in the surrounding oceans (Pole, 2003). During the Oligocene – early Miocene, ‘New Zealand’ was located north of the proto-Subtropical Front and was surrounded by subtropical water transported from low latitudes south down the east coast of Australia (Beu, 1990; Buening et al., 1998; Hornibrook, 1992; Nelson and Cooke, 2001). Pulses of warmer, subtropical water may have reached New Zealand at times during the middle Miocene (Hornibrook, 1992). Today the Subtropical Front surrounds New Zealand’s South Island and currents from the Subantarctic trans-

*

Corresponding author. E-mail address: [email protected] (J.G. Prebble).

http://dx.doi.org/10.1016/j.epsl.2016.11.006 0012-821X/© 2016 Elsevier B.V. All rights reserved.

port cool water northward (Chiswell et al., 2015). Growth of more persistent terrestrial ice sheets in Antarctica during the latter part of the middle Miocene was accompanied by northward expansion of Subantarctic Surface water across the Campbell Plateau, which brought the Subtropical Front in close proximity to the New Zealand landmass (Field et al., 2009; Hayward et al., 2004; Nelson and Cooke, 2001); similar to the present day (Chiswell et al., 2015). Our understanding of Oligocene – Quaternary New Zealand terrestrial climate is derived mainly from Miocene macroscopic plant fossils preserved in lacustrine and lignite deposits from the South Island provinces of Southland and Otago (e.g. Pole (2014), Pole et al. (2003), Pole and Moore (2011), Reichgelt et al. (2013, 2015)), and from Miocene – Pliocene pollen assemblages recovered from a range of depositional settings, e.g. Mildenhall (1980), Mildenhall and Pocknall (1984, 1989), Pocknall and Mildenhall (1984). Other, non-biological proxies for terrestrial climate, include mineral indicators and provenance data that record the effects of climaticallydriven erosion from the Southern Alps during the late Miocene– present (Carter and Gammon, 2004; Heenan and McGlone, 2013). In this study, we compiled fossil pollen and spore data from the New Zealand Fossil Record Electronic Database (FRED) to create an estimate of New Zealand terrestrial climate evolution over the last 30 million years, using recently compiled modern range

JID:EPSL

AID:14170 /SCO

2

[m5G; v1.191; Prn:30/11/2016; 15:37] P.2 (1-9)

J.G. Prebble et al. / Earth and Planetary Science Letters ••• (••••) •••–•••

Fig. 1. Distribution of samples used in this study. A. Density plot showing paleo-latitude of samples, grouped by New Zealand stage. B. Density plot showing modern location of samples, grouped by New Zealand Series. C. Number of samples assigned to each single New Zealand stage. (For interpretation of the colours in this figure, the reader is referred to the web version of this article.)

data for 70 modern analogues to the fossil taxa. FRED (online at http://www.fred.org.nz/) is a public database that contains a range of information for fossil localities in New Zealand. FRED contains about 98000 locality records registered at regional recording centres since 1946, and includes palynomorph species lists from more than 5500 samples of Paleocene to Recent age. Long-term paleoclimatic reconstructions are often dominated by ocean-based proxies, which is particularly true in the Southern Hemisphere given the paucity of terrestrial records. The approach presented in this paper provides unique insight into terrestrial climatic evolution in the southwest Pacific. It illustrates southern mid-latitude climate response to stepwise global cooling since the Miocene Climate transition, the expansion of the Antarctic cryosphere, and Southern Ocean change since the late Oligocene. 2. Methods 2.1. Data extraction from FRED Samples with an age assignment to a single New Zealand Stage of Whaingaroan (early Oligocene) or younger, which also contained at least six of the 70 fossil pollen or spore taxa for which a modern affinity of genus level or finer has been suggested, were extracted from the FRED database. Samples from drill hole cuttings were excluded to avoid problems occasionally encountered with caving and reworking of microfossils in this type of sample. A synonymy list for each of the seventy taxa was compiled by reviewing and updating the synonymies from Raine et al. (2011), augmented

by a review of the list of spore and pollen names that appear in FRED. The synonymy list used here is included in Supplementary Table ST1. Modern affinities are after Raine et al. (2011). The search returned 2036 samples, with a mean of 10.7 modern analogue taxa per sample. The majority (1391) of these samples were of Nukumaruan or younger (Quaternary) age, with the remaining 645 samples distributed across the 13 New Zealand Stages of Oligocene–Pliocene age (Fig. 1). The smallest number of samples uniquely assigned to a single New Zealand stage was 11; which were from the Clifdenian Stage (middle Miocene, 15.9–15.1 Ma). Although any registered user is welcome to submit samples to the FRED database, 98% of the 2036 samples used in this study were analysed by researchers at GNS Science or its institutional predecessors and >75% (1564 samples) were examined by a single palynologist (D.C. Mildenhall). In the FRED database, age is assigned to each sample by the scientist who submitted the record, and is usually by reference to a Stage, or range of Stages, of the New Zealand Geological Time Scale (Cooper, 2004; Raine et al., 2015). Multiple ages have been assigned to about 10% of the samples used in this study. Multiple ages can arise when overlapping or conflicting ages are assigned by workers on different fossil groups extracted from the same sample, when a reassessment of the age determination is made on a fossil assemblage in a previously submitted sample, or where new observations are made of a fossil group that had already been recorded from that sample. In this study, we took the most recent age determination that had been made for each fossil group. Where samples

JID:EPSL

AID:14170 /SCO

[m5G; v1.191; Prn:30/11/2016; 15:37] P.3 (1-9)

J.G. Prebble et al. / Earth and Planetary Science Letters ••• (••••) •••–•••

had multiple age assignments arising from different fossil groups, we assigned the age to be the range of overlap from the multiple fossil groups in the sample. However, most of the samples contained only pollen and spore assemblages, and the age assignment is based on the pollen and spore ranges and zones described in Cooper (2004), and earlier biostratigraphic studies listed therein. In many cases, age will have been assigned considering contextual stratigraphic constraints, which are not captured in our search. The paleo latitude for each sample was calculated using Gplates (online at http://www.gplates.org/index.html), using a moving hot spot absolute reference frame (Seton et al., 2012). These are intended to document the general trends and changes in latitudinal range sampled over time (Fig. 1). A list of the FRED records examined, along with their assigned age, calculated paleo-latitudes and mean annual temperature ranges, is included in Supplementary Table ST2. A second set of samples were extracted from the FRED database with the same criteria as described above, but for samples that spanned exactly two New Zealand Stages. These samples were extracted as an independent test of the climate results obtained from the ‘single-stage’ sample suite, albeit with a lower temporal resolution. This ‘double stage’ dataset comprised only 329 samples of Whaingaroan–Duntroonian age or younger, but they were more evenly spread across the time series than the ‘single stage’ dataset, as very few ‘double stage’ samples were of Quaternary age. Results were comparable to the ‘single stage’ dataset, and are included in Supplementary Fig. SF1. 2.2. Modern environmental data Modern environmental data were compiled from geospatially referenced floral distribution data from Australia, New Zealand and New Caledonia for extant taxa with affinities to the 70 fossil plant taxa included in this study. Spatial distribution data for relevant flora were cross-correlated to high resolution regional gridded climate datasets to identify the mean annual temperature (MAT), warmest quarter temperature (ST), coldest quarter temperature (WT), and annual precipitation (AP) for each taxon. The core representation of plant environmental tolerance was estimated by using the 10th and 90th percentiles of the environmental statistics for each plant taxon. Employing the core representation of distribution data is effective when working with large occurrence datasets (Kershaw, 1997; Reichgelt et al., 2015). Modern environmental data are included in Supplementary Table ST1. The distribution and climate data was extracted from a range of sources. Australian climate data was derived from Australian Bureau of Meteorology and the ANUClim climate profile software, and occurrences from Australia’s Virtual Herbarium (www.chah.gov.au/avh/). New Zealand climate data was from the Land Environments of New Zealand (LENZ) dataset (Leathwick et al., 2002) and plant occurrences from New Zealand National Vegetation Databank (Wiser et al., 2001) at the Allan Herbarium at Landcare Research. New Caledonia climate was derived from Hijmans et al. (2005) and WorldClim software, while plant occurrence data was extracted from Herbarium of Centre IRD Nouméa (http://herbier-noumea.plantnet-project.org) and Museum National d’Histoire Naturelle: Phanerogams Herbarium Specimens. 2.3. Bioclimatic analysis Environmental estimates were made for each sample using a bioclimatic analysis (Greenwood et al., 2005; Reichgelt et al., 2013), where the climatic envelopes of the nearest living relatives of a fossil flora are used to determine the climatic range in which the majority of the flora could co-occur (Kershaw, 1997; Thompson et al., 2012). Bioclimatic Analysis differs from the Coexistence Approach (Utescher et al., 2014), in that climatic outliers in

3

Fig. 2. Illustration of the method to identify the ‘interval of environmental overlap’; in sample SW41171/f0002(108.89-), one of the 64 samples of Waipipian age. This sample contained 10 fossil pollen for which the climatic range of modern analogues were available. Note that the warmest modern extent of Phyllocladus and Pherosphaera are disregarded, as the distribution of these species is likely to be geographically, rather than climatically, controlled. For many of the samples, including this one, there is a gap between the coldest maximum extent, and warmest minimum extent (i.e. the area between red horizontal lines). Where this occurs, the second coldest maximum extent, and the second warmest minimum extent was used (area between blue lines). The mean temperature for the sample, used for ordering, is shown by the black line. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

the modern data are excluded by only retaining the 10th and 90th percentiles of the modern climatic range (Thompson et al., 2012), and that large modern occurrence datasets are cross-correlated with gridded climatic data instead of using nearest station data. Bioclimatic analysis is performed at genus level as many species in the fossil data set are extinct and family distributions are usually so widespread that they generate a range of climate variables that are too broad to be of use. An example of the approach we used to collate and analyse the data is shown in Figs. 2–3, and described below. This illustrated example uses the 64 samples from the late Pliocene (Waipipian Stage). For each sample, the interval of climatic overlap was identified, which includes the overlap of the taxa with the coldest maximum extent and taxa with the warmest minimum extent (Fig. 2). However, many of samples produce no climatic overlap as there is a gap between the coldest maximum extent and warmest minimum extent. Where this occurs, we have taken a ‘step out’ to the second coldest maximum extent and second warmest minimum extent (Fig. 2). If no overlap was encountered at that stage, the sample was discarded. Each viable sample was placed in an ordered sequence based on the mean MAT of each sample (for calculation of mean see Fig. 2) and the warmest and coldest 20% of samples identified (Fig. 3). Climatic overlap ranges for the warmest and coldest samples were summed into temperature bins of 1 ◦ C intervals and used to plot the climatic range. Stronger colour saturation in Figs. 3–4 indicate ranges where there is the greatest consensus between samples, essentially an indication of relative confidence in the climatic estimates between stages. The same method was used to derive mean annual precipitation. Thus, the wettest and driest 20% of samples are not necessarily the same subset of samples used to derive temperature.

JID:EPSL

AID:14170 /SCO

4

[m5G; v1.191; Prn:30/11/2016; 15:37] P.4 (1-9)

J.G. Prebble et al. / Earth and Planetary Science Letters ••• (••••) •••–•••

Fig. 3. Illustration of the method to extract climate statistics from each Stage; the example shows the 64 samples of Waipipian age. The yellow bar shows position of sample SW41171/f0002(108.89-), from Fig. 1. Red vertical bars show samples with a gap between the coldest maximum extent and warmest minimum extent (i.e. climatic ranges that were discarded), blue bars show the range of overlap; wider bars show ‘first overlap’, narrow bars show ‘second overlap’. The warmest and coldest 20% of samples are those that lie to the far left and right of the dashed vertical lines, discarding the three right-most samples for which no climatic overlap existed for the second coldest maximum and second warmest minimum ranges. Thick blue and red horizontal lines show the mean values used for the warmest and coldest 20%. The blue and red tiles in the right hand box show how the coloured tiles in Fig. 4 represent the number of samples that occupy each 1 ◦ C interval. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

2.4. Quality of climatic estimates There are three main sources of uncertainty in these pollenbased climate records: (1) inconsistency of measurement between the samples collated, (2) biases due to the nature of samples included in each time bin, e.g. some time bins may include samples from a limited range of facies that represent a small range of paleoenvironments, and some paleolatitudes will be under sampled, and (3) errors associated with the bioclimatic analysis method itself, principally the severe difficulties in establishing that the underpinning assumptions of coexistence-type approaches (Utescher et al., 2014) are not violated (Grimm and Potts, 2016), and related data processing decisions. The samples analysed in this study were originally collected and processed to address a broad range of earth-science questions and were not specifically studied for bioclimatic analysis. Fortunately, as noted above, three quarters of the samples were analysed by a single palynologist, and 98% by workers from the same institute. This internal consistency reduces the chance for taxonomic errors that can occur between different operators. Age assignments for most of the samples are based on pollen assemblages. We have used the New Zealand stage that was nominated by the operator when the sample was submitted to FRED. Some samples are likely to have an incorrect global age assignment as the calibration of the New Zealand timescale to the international timescale continues to be developed (Raine et al., 2015). In practice, however, this is unlikely to be a problem, because the stage boundaries (as opposed to their calibration to absolute time and international timescales) have remained relatively stable since palynological investigation of New Zealand sediments began in earnest c. 65 years ago (Cooper, 2004; Raine et al., 2015). However, we acknowledge that uncertainty in the precise correlation of many terrestrial biostratigraphic events with the marine stages remains (Cooper, 2004). Reworked specimens may also affect our results, particularly as our analytical approach only considers binary (presence/absence) data, rather than species abundance. Effects due to reworked spec-

imens could be particularly pronounced in stages that contain low numbers of samples. For example, a few ‘warm’ taxa reworked up section from older strata could have a large effect on the “warmest 20%”. We also know that the New Zealand region was dominated by quite distinct depositional environments at different times through the Cenozoic. For example many early and middle Miocene samples in FRED come from peat/coal mire sediments (Mildenhall and Pocknall, 1989) whereas some of the Pliocene–Pleistocene samples are derived from open marine sediments. Finally, we expect the New Zealand fossil pollen record to have a comparable rate of omission to other settings around the globe, and know there are likely to be many important floral elements missing from the fossil pollen record. For example, fossil leaf assemblages indicate that Lauraceae dominated rainforests in the Miocene, but there is no pollen record of these flora because Lauraceae pollen does not preserve well (Mildenhall, 1980; Mildenhall et al., 2014). Modern environmental envelopes that are applied to the fossil taxa are subject to revision as new observations of plant evolution and relationships are made. Furthermore, bioclimatic analysis (as applied in this study) produces an estimate even where it is apparent that aspects of the fossil ecology are different to the modern flora. For example, increased greater species diversity, and co-occurrence of genera that are not associated today, has been reported for the New Zealand Miocene (Jordan et al., 2011; Pole et al., 2003). Finally, our data-processing decisions influence the final result. Two key settings in our processing are the “minimum number of modern analogue taxa” allowed in a sample (set to six in this study) and the “proportion of samples extracted to describe the warm and cold range” (set to 20%). We explored the effect varying both, and our chosen settings represent a balance between retaining sufficient samples through the time series and clarity in our signal. When the “minimum number of analogue taxa” is between 3 and 9 (including the 6 we report), the shape of the resulting climate curves through time are similar. As the minimum num-

JID:EPSL

AID:14170 /SCO

[m5G; v1.191; Prn:30/11/2016; 15:37] P.5 (1-9)

J.G. Prebble et al. / Earth and Planetary Science Letters ••• (••••) •••–•••

5

ber increases beyond 9, the smaller number of taxonomically diverse samples available causes the “warmest” and “coldest” climate curves to tend towards convergence. When the minimum analogue number is set to 15, the number of available samples across the entire time series is reduced to 323 (from 2036), although the broad climate trends are still visible. Varying the “proportion of samples extracted” has a more immediate effect on the shape of our climate curve through time: including proportions >∼30% noticeably flattens and converges both curves, as many samples in any given group have a large climatic range (Fig. 3). Conversely, reducing the proportion results in very small sample sizes in the poorly sampled Stages (e.g. Clifdenian and Otaian). Despite the sources of uncertainty outlined above, we are encouraged that the results presented below generally track the cooling/warming patterns seen in regional and global records, and note the approach allows a clear synthesis and representation of a large volume of otherwise inaccessible data. 3. Results 3.1. Mean annual temperature The pollen-based regional climate record shows average temperatures cooled through the Oligocene, warmed through the early Miocene, peaked during the middle Miocene, and cooled again through the late Miocene (Fig. 4A–B, Supplementary Table ST3). Mean annual temperatures recorded by the warmest and coldest 20% of samples in each age bin stage are generally within 1–2 ◦ C of each other through the Oligocene to middle Miocene but diverge in the late Neogene (Fig. 4A). Specifically, the warmest 20% of all samples indicate MAT dropped 1 ◦ C at the end of the middle Miocene (Lillburnian) and remained at ∼20 ◦ C during the late Miocene to early Pliocene (Waiauan to Opoitian) and dropped ∼1.5 ◦ C in the Pleistocene (Mangapanian-Haweran). In contrast, the coolest 20% of the samples from late Miocene to Recent sediments suggest MAT decreased by ∼7 ◦ C over this time. For most of the record, but particularly in the late Miocene and Pliocene, there is greater consensus between samples in the warmest 20% than in the coldest 20% (compare range of shading in Fig. 4A with Fig. 4B). Although there is less consensus in the coldest 20% of samples, there is relatively strong consensus in the middle Miocene and Quaternary (Nakumaruan–Recent) time bins, suggesting a maximum cooling of ∼7 ◦ C is a reasonable interpretation. Regardless of the potential issues associated with constraining magnitude of temperature change, both the warmest and coldest sample sets indicate the two largest cooling episodes in the record occurred: 1) during the middle-late Miocene (Lillburnian to Tongaporutuan) and 2) during the late Pliocene to early Pleistocene (Opoitian to Mangapanian). The coolest 20% of samples also suggests major temperature fluctuations occurred in the late Miocene, with a large, transient, cooling episode in the Kapitean. 3.2. Temperature seasonality A distinct change in seasonality is observed in samples of late Miocene (Waiauan and Tongaporutuan) age and younger (Fig. 4A–B), where seasonality is estimated by average summer temperature (ST) minus winter temperatures (WT). During the Oligocene and early Miocene ST − WT was ∼6 ◦ C. In the warmest 20% of samples, the mean estimate of ST remains about 23 ◦ C for the entire record. In contrast, the mean estimate of WT from the same samples dropped ∼1 ◦ C between the Lillburnian and Waiauan Stages, and a further ∼0.5–1 ◦ C over the next 7 Ma. By the late Pliocene (Waipipian), seasonality was ∼8–9 ◦ C. The largest decrease in warm-sample WT is across the Plio-Pleistocene boundary (Mangapanian and Nukumauruan-Castlecliffian), with a decrease of

Fig. 4. Pollen-based climate estimates for New Zealand. A. Temperature estimates using the warmest 20% of samples. Red tiling shows the 1 ◦ C intervals that contain the greatest proportion of estimates (see Fig. 3 for explanation), solid red line shows the mean temperature, bounding grey lines show mean summer and winter temperatures. B. Temperature estimates using the coldest 20% of samples. Blue tiling shows the 1 ◦ C intervals that contain the greatest proportion of estimates (see Fig. 3 for explanation), solid blue line shows the mean temperature, bounding grey lines show mean summer and winter temperatures. C. Mean annual precipitation estimates using the wettest 20% of samples. Green tiling shows the 250 mm/yr intervals that contain the greatest proportion of estimates, solid green line shows the mean precipitation. D. Mean annual precipitation estimates using the driest 20% of samples. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

almost 3 ◦ C. Mean ST − WT in the warmest 20% of Quaternary aged samples reached 10 ◦ C. For comparison, modern ST − WT in New Zealand is 6–8 ◦ C in northern North Island, and 10–13 ◦ C inland and eastern South Island. A similar trend of increasing seasonality

JID:EPSL

AID:14170 /SCO

6

[m5G; v1.191; Prn:30/11/2016; 15:37] P.6 (1-9)

J.G. Prebble et al. / Earth and Planetary Science Letters ••• (••••) •••–•••

during the late Miocene, is observed in the coldest 20% of samples, also driven by decreases in winter temperatures. A seasonal range of 6–7 ◦ C is observed during the early–middle Miocene expanding to ∼9 ◦ C through the late Miocene. However, as with the mean annual estimates, the seasonal estimates of the coldest 20% have poorer consensus than the warmest 20% (Supplementary Fig. SF1). 3.3. Precipitation Annual precipitation estimates for the driest 20% of samples indicate a small increase since the late Oligocene, from ∼1600 mm/yr during the late Oligocene (Waitakian) to ∼1900 mm/yr during the Quaternary (Nukumaruan–Castlecliffean) (Fig. 4D). In contrast, estimates from the wettest 20% of samples indicate a stable ∼2000 mm/yr precipitation between the early Oligocene – middle Miocene (Whaingaroan to Waiauan Stages), followed by a step-wise increase in precipitation values to ∼2500–3700 mm/yr by the Quaternary (Nukumaruan) (Fig. 4C). The increase in mean AP in the wettest 20% of samples for Stages after 12 Ma is accompanied by a reduced consensus in these samples. The late Miocene Kapitean Stage, which contained a pronounced cooling in the coldest 20% of samples, also has a pronounced increase in the range of precipitation in the wettest 20% of samples (Fig. 4C). For comparison, annual precipitation of modern day New Zealand ranges between 350 mm/yr in areas of rain shadow in central and eastern South Island to over 5000 mm/yr in western South Island (Leathwick et al., 2002). The driest 20% AP estimates over the Cenozoic are remarkably stable (Fig. 4D), whereas the maximum 20% AP estimates increase, starting at the Waiauan–Tongaporutuan boundary (Fig. 4C). Our method was not capable of resolving the very low precipitation regimes that exist in New Zealand currently, which are due to the rain shadow caused by the Southern Alps. 4. Discussion 4.1. Comparison with regional climate records The pollen-based regional climate record shows average temperatures cooled through the Oligocene, warmed through the early Miocene, peaked during the middle Miocene, and cooled again through the late Neogene. The pattern is broadly similar to the structure in two qualitative estimates of New Zealand shallow marine conditions, based mainly on shallow water foraminifera assemblages (Hornibrook, 1992), and generic diversity of molluscs (Beu, 1990) (Fig. 4B). All three records indicate late Oligocene – early Miocene warming and middle Miocene peak warmth, followed by gradual decline through the middle/late Miocene. A pronounced, short-lived cooling during the latest Miocene (Kapitean) is a feature of both the pollen and regional marine records (Fig. 5A–B). In the pollen record, it is seen in the ‘coldest 20%’ temperature estimates from the pollen samples, in both mean annual, and seasonal, estimates. Although it is not a notable feature of the warmest 20% of samples, a pronounced increase in the range of annual precipitation is also observed (Fig. 5C). One disparity between our pollen compilation and records from regional seas is that the pollen record suggests latter part of the middle Miocene (Clifdenian–Lillburnian) contained the warmest terrestrial temperatures, in contrast to the shallow-water marine records, which suggest peak warmth was slightly earlier in the Miocene (Altonian–Clifdenian) (Beu, 1990; Hornibrook, 1992) (Fig. 5A–C). Additional work is required to determine if this off set is real, or an artefact of poor correlation of terrestrial deposits with the marine stage timescale. A second departure between the shallow marine and pollen derived climatic estimates is during the Oligocene, where the pollen

Fig. 5. Summary pollen-based temperature and precipitation estimates for New Zealand compared to regional and global climate drivers and proxies. A. Green = mean summer temperature, upper bound of polygon = mean of warmest 20% of samples, lower bound = coldest 20%. Pink = mean annual temperature, Blue = mean winter temperature. Orange targets show climatic estimates from a compilation of leaf morphology studies (CLAMP) from New Zealand (Kennedy et al., 2008; Reichgelt et al., 2013, 2015, 2016). B. Long dashed lines show relative (unquantified) New Zealand climatic variation from shallow water foraminifera (Hornibrook, 1992), short dashed lines show relative New Zealand climatic variation inferred from mollusc diversity (Beu, 1990). C. Mean annual precipitation, upper bound of polygon = mean of wettest 20% of samples, lower bound = driest 20%. D. Summary of South Island New Zealand compressional tectonics. E. Paleo latitude of each of the warmest (red) and coldest (blue) 20% of samples for each Stage. F. Global benthic oxygen isotope compilation (Zachos et al., 2008). G. Estimated sea level variation due to ice volume (Cramer et al., 2011). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

JID:EPSL

AID:14170 /SCO

[m5G; v1.191; Prn:30/11/2016; 15:37] P.7 (1-9)

J.G. Prebble et al. / Earth and Planetary Science Letters ••• (••••) •••–•••

shows a muted cooling trend between Whaingaroan and Waitakian time (Fig. 4A–B, Fig. 5A), while warming is shown in the marine sequences, although the record of Beu (1990) does not extend into the early Oligocene. Additional temperature proxies from the region might assist to resolve this discrepancy. Our estimates of middle Miocene temperature are similar to other estimates that have applied bioclimatic analysis to subsets of the data set we have used in this study, for example from Central Otago (Pole, 2014; Reichgelt et al., 2015, 2013). These studies indicate that peak MATs were 18–21 ◦ C during the middle Miocene. These temperatures were 2–5 ◦ C higher than those of the warmest latitudes of modern New Zealand (35–37◦ S), and are derived from samples deposited at paleo latitudes of 47–49◦ S (Fig. 5E). Estimates of MAT using leaf morphology (CLAMP), mostly from a series of middle Miocene locations in Central Otago, are also consistent with our estimates (Fig. 5A). A notable exception is “study 9” from a stratigraphic section of Waiauan age, representing a macrofloral assemblage that may have grown in a heavily shaded environment (Reichgelt et al., 2016). Modern NZ mean annual temperature range is 10–16 ◦ C. The Quaternary MAT estimates from the coldest 20% of samples lie in the middle of this range, while the warmest 20% of samples of Quaternary age are significantly warmer. Given that late Quaternary glacial variability includes temperatures considerably cooler than the modern range (Newnham et al., 1999) the ‘warmest 20%’ Quaternary temperature estimates appear generally too warm. This might partly be because the Bioclimatic Analysis as applied here does not have the ability to identify the late Quaternary grassdominated glacial assemblages as cold, due to a lack of grass analogues. However, this issue is unlikely to bias the Neogene record, as grass dominated assemblages were not a significant feature of pre-Pliocene pollen assemblages from New Zealand (Mildenhall, 1980). Many modern NZ genera have representatives in New Caledonia or Australia and therefore have wide modern temperature ranges. The more plausible correlation with the coldest 20% of estimates during the Quaternary, and the better agreement of these estimates with the CLAMP estimates, suggests this cooler curve is the more appropriate one to explore variability following the first cooling events of the Miocene. 4.2. Correlations and drivers Our record of mid latitude terrestrial climate will reflect a combination of global and regional influences on the vegetation communities from which it is derived. We expect the key global influences on regional vegetation to be climatic and circulation changes arising from changes in ice extent and sea level, and far-field tectonic events. These global events will be filtered through regional influences that include paleo latitude, local tectonic influences on ecological niche diversity, and changing local topographical and bathymetric influences on atmospheric and oceanic circulation. Overall, there are many similarities between our record and those that document evolution of global climate since the Oligocene. For example, broad scale changes in sea level attributable to global ice extent, derived from global benthic isotope and Mg/Ca estimates (Cramer et al., 2011) (Fig. 5G), are mirrored as temperature variability in the mid latitude pollen record. These include sea level maximum in the middle Miocene, followed by gradual decline through the late Miocene, a minimum in the late Miocene (Kapitean), and middle Pliocene (late Opoitian) warmth. The unprocessed global compilation of benthic oxygen isotopes, thus recording a combination of ice volume and deep sea temperature (Fig. 5F), also shows a warm middle Miocene, progressively cooling until the cold, high temporal variability of the Quaternary deep ocean (Zachos et al., 2008). During the time covered by our record, the New Zealand landmass migrated northwards 5–10◦

7

(Seton et al., 2012), at the same time as increasing the sampled meridional range from ∼5◦ during the Oligocene to ∼13◦ during the Quaternary (Figs. 1 and 5E). The likely effect of the northwards drift is providing a relatively muted cooling signal through the Miocene. The increase in meridional range through time may also have increased the temperature range of the climate estimates, although the separation by latitude of cold and warm samples that might be expected if this were true, is not observed (Fig. 5E). 4.3. Middle Miocene Climatic Transition Following the variability of the Antarctic ice sheets during the Miocene Climatic Optimum (Levy et al., 2016), ice growth and cooling was step-wise, with an abrupt expansion at 13.8 Ma (Lear et al., 2015). Although abrupt environmental changes attributed to this 13.8 Ma Miocene Climatic Transition have been noted in South Island sections from the New Zealand (Pole, 2014), this event is somewhat cryptic in our composite pollen record. Although a cooling in winter temperature is observed in the warmest 20% of samples between the Lillburnian and Waiauan stages, the cooling of mean annual temperature is not observed in the coldest 20% of samples until the following stage boundary, between the Waiauan and Tongaporutuan Stages. Our results do, however, suggest a slight increase in seasonality associated with the Miocene Climatic Transition in both the warmest and coldest 20% of samples, from ∼5.5–6.6 ◦ C during the early and early-middle Miocene (Otaian–Clifdenian), and 6.6–7 ◦ C by the end of the late Miocene (Waiauan). One regional expression of middle Miocene cooling was likely northward encroachment of the Subtropical Front (STF), and Subantarctic Water, onto the southern part of the New Zealand continent (Hayward et al., 2004; Nelson and Cooke, 2001). Although the timing of STF movement is not yet well constrained, the cooler winter temperatures suggested by the pollen record may be a reflection of increased exposure to Subantarctic weather systems. 4.4. Late Miocene influence of tectonics The pronounced cooling during the earliest part of the late Miocene (between the Waiauan and Tongaporutuan Stages) in the coldest 20% of samples (Fig. 4B), and the resulting separation between the temperature estimates for the warmest and coldest 20% of samples, is a feature that endures for the reminder of the record. One possible reason for the greater range of temperature following middle Miocene is greater temporal variability, and the range represents distinct glacial and interglacial assemblages. However, stable isotopes show no increase in deep ocean temperature variability until the Pliocene (Fig. 4F). Furthermore, the qualitative estimates of regional marine conditions also do not change markedly during this time (Fig. 5B). This cooling that appears in the pollen record did not extend to the local marine environment. The cooling also significantly predates the increase in sampled latitudinal range, which does not occur until the Pliocene (Fig. 5E). During the late Miocene, the samples that contribute to the warmest and coldest 20% datasets are from a latitudinal range of less than 4◦ . The increase in latitudinal range to 10–13◦ during the Pliocene– Quaternary is also likely to contribute to the difference in temperatures between the warmest and coldest datasets. Late Miocene (Tongaporutuan) cooling in the coldest 20% of samples does coincide with the earliest onset of compressional plate motion between the Pacific and Australian plates (Cande and Stock, 2004), and increased uplift in the South Island (Batt et al., 2004). The cooling in the coldest 20% of samples plausibly reflects a pollen source from the new range environments arising from this tectonic event, which became more pronounced through the late

JID:EPSL

AID:14170 /SCO

8

[m5G; v1.191; Prn:30/11/2016; 15:37] P.8 (1-9)

J.G. Prebble et al. / Earth and Planetary Science Letters ••• (••••) •••–•••

Miocene and Pliocene (Heenan and McGlone, 2013; Mildenhall, 1980; Mildenhall and Pocknall, 1984). The coincidence of the first large separation between precipitation estimates from the warmest and driest samples in the record (Fig. 5C) would also be consistent with increased relief, but could also be expected in a generally cooling climate. It is probable that annual precipitation was more uniform over the lowlying New Zealand archipelago prior to the uplift of the Southern Alps (Reichgelt et al., 2015). The increased precipitation ranges following South Island uplift in the late Miocene could reflect increased contrast between windward and leeward sides of the range (Mildenhall and Pocknall, 1984). However, reasons for failure to record a hypothesised decreasing minimum AP are discussed below, and the spatial distribution of our samples during the middle Miocene (Fig. 1), is likely insufficient to capture such a precipitation gradient. Within our record, several taxa that are linked to dry environments, such as Acacia and possibly Eucalyptus, disappeared from the New Zealand flora (Lee et al., 2001; Mildenhall, 1980). This local extinction could explain the increase in the maximum 20% AP estimates, as the disappearance of exclusively dry floral elements would increase the relative contribution of genera with a wide AP range, or exclusively wet-environment floral elements. Although regions in New Zealand today, such as central South Island, support semi-arid environments, the plant genera that occupy these regions are not exclusive to dry habitats. This influences the result of the minimum 20% AP estimate, because most modern representatives of these genera will not have been recorded in dry habitats, as dry habitats represent a marginal proportion of the New Zealand landmass. Bioclimatic Analysis is designed to provide the most likely environment for a given floral assemblage. Therefore, if most occurrences for every genus in the assemblage are in wet environments, the most likely co-existence envelope will also be in a wet environment. 4.5. Latest Miocene cooling A pronounced cooling and increase in variability of annual precipitation is observed during the latest Miocene (Kapitean) in the pollen record, and is associated with a cooling in both regional marine summary records (Fig. 5B–D). Records of marine cooling around New Zealand have long been correlated to the ice advance and sea level fall associated with the Mediterranean Sea desiccation during the Messinian (Ohneiser et al., 2015; Roberts et al., 1994). The limited terrestrial macro fossil evidence available from New Zealand also suggests Kapitean climate was cooler than present, possibly with more variable conditions (Pole and Moore, 2011); observations supported by our composite pollen record. 4.6. Pliocene increase in seasonality Prior to the Quaternary, the largest decrease in warm-sample winter temperatures was observed between the early and late Pliocene (Opoitian to Waipipian stages). The seasonality of 7–8.5 ◦ C during the early Pliocene increased to 8.8–9.5 ◦ C during the Waipipian, and remained ∼8.5 ◦ C for the remainder of the Pliocene and Quaternary. This winter cooling coincides with late Pliocene expansion of Antarctic Ice (McKay et al., 2012; Patterson et al., 2014), and associated intensification of westerly winds and northward migration of Southern Ocean fronts (McKay et al., 2012). An illustration in the importance of Southern Hemispheric drivers in our mid latitude record is that this change in seasonality predates the onset of significant Northern Hemisphere glaciation by >300 ka (Maslin et al., 1998).

5. Conclusions The estimates of terrestrial temperature of the New Zealand landmass since the Oligocene presented in this paper represent a concerted attempt to employ a large pollen dataset, derived from samples which were studied for a range of purposes and by a range of people over the last c. 65 years. Despite the uncertainty associated with a compilation of this type, our results are in agreement with ocean-based Cenozoic paleo-climate reconstructions. This record shows warming through the late Oligocene to the early Miocene, with peak warmth during the middle Miocene. Late Miocene and Pliocene cooling following middle Miocene warmth appears to have been largely driven by reduction in winter temperature, accompanied by an increase in seasonality. The timing of these mid latitude cooling events coincide with evolution of global climate since the Oligocene, and are influenced by regional tectonic uplift since the late Miocene. Acknowledgements JGP, DCM, JIR and EMK were supported by the GNS core funded Global Change through Time Programme, funded by the New Zealand Government, JGP and HS also received support from the GNS core funded Strategic Development Programme. DRG was supported by NSERC (2016-04337). The authors gratefully acknowledge the use of data from the New Zealand Fossil Record Electronic Database. We thank the following colleagues at GNS Science: Richard Thomas for FRED database management, David Harte for discussions on data manipulation, Martin Crundwell for discussion on regional Miocene marine records, Chris Hollis and Richard Levy for comments on an earlier draft. We thank Bob Spicer and an anonymous reviewer for thoughtful review comments that improved this manuscript. Appendix A. Supplementary material Supplementary material related to this article can be found online at http://dx.doi.org/10.1016/j.epsl.2016.11.006. References Batt, G.E., Baldwin, S.L., Cottam, M.A., Fitzgerald, P.G., Brandon, M.T., Spell, T.L., 2004. Cenozoic plate boundary evolution in the South Island of New Zealand: new thermochronological constraints. Tectonics 23 (TC4001), 4001–4017. Beu, A.G., 1990. Molluscan generic diversity of New Zealand Neogene stages: extinction and biostratigraphic events. Palaeogeogr. Palaeoclimatol. Palaeoecol. 77, 279–288. Buening, N., Carlson, S.J., Spero, H.J., Lee, D.E., 1998. Evidence for the Early Oligocene formation of a proto-Subtropical Convergence from oxygen isotope records of New Zealand Paleogene brachiopods. Palaeogeogr. Palaeoclimatol. Palaeoecol. 138, 43–68. Cande, S.C., Stock, J.M., 2004. Pacific–Antarctic–Australia motion and the formation of the Macquarie Plate. Geophys. J. Int. 157, 399–414. Carter, R.M., Gammon, P., 2004. New Zealand maritime glaciation: millennial-scale southern climate change since 3.9 Ma. Science 304, 1659–1662. Chiswell, S.M., Bostock, H.C., Sutton, P.J.H., Williams, M.J.M., 2015. Physical oceanography of the deep seas around New Zealand: a review. N.Z. J. Mar. Freshw. Res. 49, 286–317. Cooper, R.A., 2004. The New Zealand Geological Timescale. Institute of Geological and Nuclear Sciences, pp. 1–284. Monograph. Cramer, B.S., Miller, K.G., Barrett, P.J., Wright, J.D., 2011. Late Cretaceous–Neogene trends in deep ocean temperature and continental ice volume: reconciling records of benthic foraminiferal geochemistry (δ 18 O and Mg/Ca) with sea level history. J. Geophys. Res., Oceans 116 (12). http://dx.doi.org/10.1029/ 2011jc007255. Field, B.D., Crundwell, M.P., Lyon, G.L., Mildenhall, D.C., Morgans, H.E.G., Ohneiser, C., Wilson, G.S., Kennett, J.P., Chanier, F., 2009. Middle Miocene paleoclimate change at Bryce Burn, southern New Zealand. N.Z. J. Geol. Geophys. 52, 321–333. Greenwood, D.R., Archibald, S.B., Mathewes, R.W., Moss, P.T., 2005. Fossil biotas from the Okanagan Highlands, southern British Columbia and northeastern Washington State: climates and ecosystems across an Eocene landscape. Can. J. Earth Sci. 42, 167–185.

JID:EPSL

AID:14170 /SCO

[m5G; v1.191; Prn:30/11/2016; 15:37] P.9 (1-9)

J.G. Prebble et al. / Earth and Planetary Science Letters ••• (••••) •••–•••

Grimm, G.W., Potts, A.J., 2016. Fallacies and fantasies: the theoretical underpinnings of the Coexistence Approach for palaeoclimate reconstruction. Clim. Past 12, 611–622. Hayward, B.W., Grenfell, H.R., Carter, R., Hayward, J.J., 2004. Benthic foraminiferal proxy evidence for the Neogene palaeoceanographic history of the Southwest Pacific, east of New Zealand. Mar. Geol. 205, 147–184. Heenan, P.B., McGlone, M.S., 2013. Evolution of New Zealand alpine and openhabitat plant species during the late Cenozoic. N.Z. J. Ecol. 37, 105–113. Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G., Jarvis, A., 2005. Very high resolution interpolated climate surfaces for global land areas. Int. J. Climatol. 25, 1965–1978. Hornibrook, N.d.B., 1992. New Zealand Cenozoic marine paleoclimates: a review based on the distribution of some shallow water and terrestrial biota. In: Tsuchi, R., Ingle, J.C. (Eds.), Pacific Neogene: Environment, Evolution and Events. University of Tokyo Press, Tokyo, pp. 83–106. Jordan, G.J., Carpenter, R.J., Bannister, J.M., Lee, D.E., Mildenhall, D.C., Hill, R.S., 2011. High conifer diversity in Oligo-Miocene New Zealand. Aust. Syst. Bot. 24, 121–136. Kennedy, E.M., Alloway, B.V., Mildenhall, D.C., Cochran, U., Pillans, B., 2008. An integrated terrestrial paleoenvironmental record from the Mid-Pleistocene transition, eastern North Island, New Zealand. Quat. Int. 178, 146–166. Kershaw, A.P., 1997. A bioclimatic analysis of early to Middle Miocene brown coal floras, Latrobe Valley, south-eastern Australia. Aust. J. Bot. 45, 373–387. King, P.R., 2000. Tectonic reconstructions of New Zealand: 40 Ma to the present. N.Z. J. Geol. Geophys. 43, 611–638. Landis, C.A., Campbell, H.J., Begg, J.G., Mildenhall, D.C., Paterson, A.M., Trewick, S.A., 2008. The Waipounamu Erosion Surface: questioning the antiquity of the New Zealand land surface and terrestrial fauna and flora. Geol. Mag. 145, 173–197. Lear, C.H., Coxall, H.K., Foster, G.L., Lunt, D.J., Mawbey, E.M., Rosenthal, Y., Sosdian, S.M., Thomas, E., Wilson, P.A., 2015. Neogene ice volume and ocean temperatures: insights from infaunal foraminiferal Mg/Ca paleothermometry. Paleoceanography 30, 1437–1454. Leathwick, J., Morgan, F., Wilson, G., Rutledge, D., McLeod, M., Johnston, K., 2002. Land Environments of New Zealand: A Technical Guide. Ministry for the Environment, New Zealand Government, Wellington. Lee, D.E., Lee, W.G., Mortimer, N., 2001. Where and why have all the flowers gone? Depletion and turnover in the New Zealand Cenozoic angiosperm flora in relation to palaeogeography and climate. Aust. J. Bot. 49, 341–356. Levy, R., Harwood, D., Florindo, F., Sangiorgi, F., Tripati, R., Von Eynatten, H., Gasson, E., Kuhn, G., Tripati, A., DeConto, R., Fielding, C., Field, B., Golledge, N., McKay, R., Naish, T., Olney, M., Pollard, D., Schouten, S., Talarico, F., Warny, S., Willmott, V., Acton, G., Panter, K., Paulsen, T., Taviani, M., 2016. Antarctic ice sheet sensitivity to atmospheric CO2 variations in the early to mid-Miocene. Proc. Natl. Acad. Sci. USA 113, 3453–3458. Maslin, M.A., Li, X.S., Loutre, M.F., Berger, A., 1998. The contribution of orbital forcing to the progressive intensification of Northern Hemisphere glaciation. Quat. Sci. Rev. 17, 411–426. McKay, R., Naish, T., Carter, L., Riesselman, C., Dunbar, R., Sjunneskog, C., Winter, D., Sangiorgi, F., Warren, C., Pagani, M., Schouten, S., Willmott, V., Levy, R., DeConto, R., Powell, R.D., 2012. Antarctic and Southern Ocean influences on Late Pliocene global cooling. Proc. Natl. Acad. Sci. USA 109, 6423–6428. Mildenhall, D.C., 1980. New Zealand Late Cretaceous and Cenozoic plant biogeography: a contribution. Palaeogeogr. Palaeoclimatol. Palaeoecol. 31, 197–233. Mildenhall, D.C., Kennedy, E.M., Lee, D.E., Kaulfuss, U., Bannister, J.M., Fox, B., Conran, J.G., 2014. Palynology of the Early Miocene Foulden Maar, Otago, New Zealand: diversity following destruction. Rev. Palaeobot. Palynol. 204, 27–42. Mildenhall, D.C., Pocknall, D.T., 1984. Palaeobotanical evidence for changes in Miocene and Pliocene climates in New Zealand. In: Late Cainozoic Palaeoclimates of the Southern Hemisphere, Proc. SASQUA Symposium. Swaziland, 1983, pp. 159–171. Mildenhall, D.C., Pocknall, D.T., 1989. Miocene–Pleistocene Spores and Pollen from Central Otago, South Island, New Zealand. New Zealand Geological Survey, Lower Hutt.

9

Nelson, C.S., Cooke, P.J., 2001. History of oceanic front development in the New Zealand sector of the Southern Ocean during the Cenozoic – a synthesis. N.Z. J. Geol. Geophys. 44, 535–553. Newnham, R.M., Lowe, D.J., Williams, P.W., 1999. Quaternary environmental change in New Zealand: a review. Prog. Phys. Geogr. 23, 567–610. Ohneiser, C., Florindo, F., Stocchi, P., Roberts, A.P., Deconto, R.M., Pollard, D., 2015. Antarctic glacio-eustatic contributions to late Miocene Mediterranean desiccation and reflooding. Nat. Commun. 6, 9765. Patterson, M.O., McKay, R., Naish, T., Escutia, C., Jimenez-Espejo, F.J., Raymo, M.E., Meyers, S.R., Tauxe, L., Brinkhuis, H., Klaus, A., Fehr, A., Bendle, J.A.P., Bijl, P.K., Bohaty, S.M., Carr, S.A., Dunbar, R.B., Flores, J.A., Gonzalez, J.J., Hayden, T.G., Iwai, M., Katsuki, K., Kong, G.S., Nakai, M., Olney, M.P., Passchier, S., Pekar, S.F., Pross, J., Riesselman, C.R., Röhl, U., Sakai, T., Shrivastava, P.K., Stickley, C.E., Sugasaki, S., Tuo, S., Van De Flierdt, T., Welsh, K., Williams, T., Yamane, M., 2014. Orbital forcing of the East Antarctic ice sheet during the Pliocene and Early Pleistocene. Nat. Geosci. 7, 841–847. Pocknall, D.T., Mildenhall, D.C., 1984. Late Oligocene – early Miocene spores and pollen from Southland, New Zealand. N.Z. Geol. Surv. Paleontol. Bull. 51. Pole, M., 2003. New Zealand climate in the Neogene and implications for global atmospheric circulation. Palaeogeogr. Palaeoclimatol. Palaeoecol. 193, 269–284. Pole, M., 2014. The Miocene climate in New Zealand: estimates from paleobotanical data. Palaeontol. Electronica 17 (27A), 79. Pole, M., Douglas, B., Mason, G., 2003. The terrestrial Miocene biota of southern New Zealand. J. R. Soc. N.Z. 33, 415–426. Pole, M., Moore, P.R., 2011. A late Miocene leaf assemblage from Coromandel Peninsula, New Zealand, and its climatic implications. Alcheringa 35, 104–121. Raine, J.I., Beu, A.G., Boyes, A.F., Campbell, H.J., Cooper, R.A., Crampton, J.S., Crundwell, M.P., Hollis, C.J., Morgans, H.E.G., Mortimer, N., 2015. New Zealand geological timescale NZGT 2015/1. N.Z. J. Geol. Geophys. 58, 398–403. Raine, J.I., Mildenhall, D.C., Kennedy, E.M., 2011. New Zealand Fossil Spores and Pollen: An Illustrated Catalogue, 4th edition. GNS Science Miscellaneous Series, vol. 4. http://data.gns.cri.nz/sporepollen/index.htm. Reichgelt, T., Kennedy, E.M., Conran, J.G., Mildenhall, D.C., Lee, D.E., 2015. The early Miocene paleolake Manuherikia: vegetation heterogeneity and warm-temperate to subtropical climate in southern New Zealand. J. Paleolimnol. 53, 349–365. Reichgelt, T., Kennedy, E.M., Jones, W.A., Jones, D.T., Lee, D.E., 2016. Contrasting palaeoenvironments of the mid/late Miocene Dunedin Volcano, southern New Zealand: climate or topography? Palaeogeogr. Palaeoclimatol. Palaeoecol. 441, 696–703. Reichgelt, T., Kennedy, E.M., Mildenhall, D.C., Conran, J.G., Greenwood, D.R., Lee, D.E., 2013. Quantitative palaeoclimate estimates for Early Miocene southern New Zealand: evidence from Foulden Maar. Palaeogeogr. Palaeoclimatol. Palaeoecol. 378, 36–44. Roberts, A.P., Turner, G.M., Vella, P.P., 1994. Magnetostratigraphic chronology of late Miocene to early Pliocene biostratigraphic and oceanographic events in New Zealand. Geol. Soc. Am. Bull. 106, 665–683. Seton, M., Müller, R.D., Zahirovic, S., Gaina, C., Torsvik, T., Shephard, G., Talsma, A., Gurnis, M., Turner, M., Maus, S., Chandler, M., 2012. Global continental and ocean basin reconstructions since 200 Ma. Earth-Sci. Rev. 113, 212–270. Strogen, D.P., Bland, K.J., Nicol, A., King, P.R., 2014. Paleogeography of the Taranaki Basin region during the latest Eocene–Early Miocene and implications for the total drowning of Zealandia. N.Z. J. Geol. Geophys. 57, 110–127. Thompson, R.S., Anderson, K.H., Pelltier, R.T., Strickland, L.E., Bartlein, P.J., Shafer, S.L., 2012. Quantitative estimation of climatic parameters from vegetation data in North America by the mutual climatic range technique. Quat. Sci. Rev. 51, 18–39. Utescher, T., Bruch, A.A., Erdei, B., Franc¸ois, L., Ivanov, D., Jacques, F.M.B., Kern, A.K., Liu, Y.S.C., Mosbrugger, V., Spicer, R.A., 2014. The coexistence approach – theoretical background and practical considerations of using plant fossils for climate quantification. Palaeogeogr. Palaeoclimatol. Palaeoecol. 410, 58–73. Wiser, S.K., Bellingham, P.J., Burrows, L.E., 2001. Managing biodiversity information: development of New Zealand’s National Vegetation Survey databank. N.Z. J. Ecol. 25, 1–17. Zachos, J.C., Dickens, G.R., Zeebe, R.E., 2008. An early Cenozoic perspective on greenhouse warming and carbon-cycle dynamics. Nature 451, 279–283.