sign in sign up
<<
Home , Pliocene, Pliocene climate

Signifi cantly warmer Arctic surface temperatures during the indicated by multiple independent proxies

A.P. Ballantyne1*, D.R. Greenwood2, J.S. Sinninghe Damsté3, A.Z. Csank4, J.J. Eberle1, and N. Rybczynski5 1Department of Geological Sciences, University of Colorado, Boulder, Colorado 80309, USA 2Department of Biology, Brandon University, Brandon, Manitoba R7A 6A9, Canada 3Department of Marine Organic Biogeochemistry, Royal Institute for Sea Research (NIOZ), Texel 1797 SZ, Netherlands 4Department of Geosciences and Laboratory of Tree-Ring Research, University of Arizona, Tucson, Arizona 85721, USA 5Canadian Museum of , Ottawa, Ontario K1P 6P4, Canada

ABSTRACT making tetraether lipids an excellent Temperatures in the Arctic have increased by an astounding 1 °C in response to anthropo- for reconstructing TSTs (Weijers et al., genic forcing over the past 20 and are expected to rise further in the coming decades. 2007) (see Appendix). The analysis of bacterial The Pliocene (2.6–5.3 Ma) is of particular interest as an analog for warming because tetraether composition in peat collected from global temperatures were signifi cantly warmer than today for a sustained period of time, the Beaver Pond site indicates a MAT of –0.6 with continental confi gurations similar to present. Here, we estimate mean annual tempera- ± 5.0 °C that suggests Pliocene temperatures in ture (MAT) based upon three independent proxies from an early Pliocene peat deposit in the Arctic were considerably warmer than the the Canadian High Arctic. Our proxies, including and annual ring widths modern MAT of –19.7 °C at Eureka, Nunavut (MAT = –0.5 ± 1.9 °C), coexistence of paleovegetation (MAT = –0.4 ± 4.1 °C), and bacterial (Environment Canada, 2009) (79°N, 85°W), tetraether composition in paleosols (MAT = –0.6 ± 5.0 °C), yield estimates that are statistically yielding a temperature difference (ΔMAT) of indistinguishable. The consensus among these proxies suggests that Arctic temperatures were ~19 °C (Table 1). ° ~19 C warmer during the Pliocene than at present, while atmospheric CO2 concentrations Refi ned paleotemperature estimates from the were ~390 ppmv. These elevated Arctic Pliocene temperatures result in a greatly reduced and annual growth rings and δ18O of cellulose in asymmetrical latitudinal temperature gradient that is probably the result of increased pole- fossil wood also showed considerably warmer ward heat transport and decreased albedo. These results indicate that Arctic temperatures TSTs in the Arctic during the Pliocene, yielding Δ may be exceedingly sensitive to anthropogenic CO2 emissions. a MAT of –0.5 ± 1.9 °C and a MAT of ~19 °C (Table 1). Although a previous estimate based INTRODUCTION GSA Data Repository1) to derive independent on this approach yielded a MAT of approxi- Throughout most of the era paleotemperature estimates. mately –5 °C, additional information from oxy- (0–65.5 Ma), temperatures at Earth’s surface gen isotopes in mosses at the site allowed us to have exceeded modern temperatures (Zachos RESULTS AND DISCUSSION calculate isotopic enrichment in the cellulose of et al., 2008). This is especially true for the The bacterial tetraether membrane composi- fossil trees, thereby reducing assumptions and Arctic (Dowsett, 2007; Robinson, 2009). The tion in soils has been shown to be very sensi- increasing the precision of our MAT estimates most recent interval in which sustained global tive to environmental conditions (Weijers et al., (see the Data Repository). temperatures exceeded those of today was dur- 2007). In particular, the degree of cyclization It is conceivable that the and tetraether ing the Pliocene (2.6–5.3 Ma), when and methylation among branched tetraethers paleotemperature proxies are biased toward the global surface temperatures were between 2 is highly dependent upon pH and temperature, warm season because they are effectively the and 3 °C warmer than present (Dowsett, 2007). Although tropical surface temperatures dur- ing the Pliocene were only slightly warmer TABLE 1. PALEOTEMPERATURE PROXY ESTIMATES FOR THE ARCTIC DURING THE PLIOCENE than present (Dowsett, 2007), Arctic tempera- AND THEIR ASSOCIATED STATISTICS tures were probably much warmer (Robinson, Temperature proxies Temperature transfer function MAT ΔMAT ±SE 2009). However, the magnitude of Arctic Plio- (°C) (°C) cene warming remains poorly constrained, pri- Tetraethers MAT = (MBT – 0.122 – 0.187 × CBT)/0.020 –0.6†,§ 19.1 5.0 marily because ice core records do not extend δ18 ,† Tree ring isotopes MAT = 17.5 + 0.98 × Opre – 2.71 × RW –0.5* 19.2 1.9 this far back. Thus, researchers have had to rely Paleovegetation CLIMST –0.4*,§ 19.3 4.1 on other proxy records to reconstruct Arctic Composite N.A. –0.4 19.3 0.4 temperatures during the Pliocene. Note: Estimates of mean annual temperature (MAT), difference from modern temperature (ΔMAT), Here we use three independent proxies to and standard error (SE) for each temperature proxy are reported. We used the CLIMST algorithm based on the coexistence approach (Mosbrugger and Utescher, 1997) to estimate temperature better constrain terrestrial surface temperatures from paleovegetation. Estimates of MAT from the three independent proxies were statistically (TSTs) for the High Arctic during the Pliocene. indistinguishable. The composite estimate is based upon the joint distribution of temperature estimates We use fossil tetraether lipids derived from soil resampled using a bootstrap technique (Efron and Tibshirani, 1997). CBT—Cyclisation of branched tetraethers; MBT—Methylation of branched tetraethers; pre—precipitation; RW—annual ring-width. 18 bacteria, oxygen isotope ratios (δ O) and annual *No statistical difference between distributions of estimates (p-value = 0.79). ring widths in fossil wood, as well as the com- †No statistical difference between distributions of estimates (p-value = 0.77). § position of paleovegetation (see the Appendix) No statistical difference between distributions of estimates (p-value = 0.72). from the Beaver Pond site on (78°N, 82°W, and 378 masl; see Fig. DR1 in the 1GSA Data Repository item 2010165, supplemental information, is available online at www.geosociety.org/ pubs/ft2010.htm, or on request from [email protected] or Documents Secretary, GSA, P.O. Box 9140, *E-mail: [email protected]. Boulder, CO 80301, USA.

© 2010 Geological Society of America. For permission to copy, contact Copyright Permissions, GSA, or [email protected]. GEOLOGY,Geology, July July 2010; 2010 v. 38; no. 7; p. 603–606; doi: 10.1130/G30815.1; 3 fi gures; 1 table; Data Repository item 2010165. 603

Downloaded from http://pubs.geoscienceworld.org/gsa/geology/article-pdf/38/7/603/3538666/603.pdf by guest on 25 September 2021 result of summer productivity. Temperature esti- CMMT MAT WMMT CONCLUSIONS mates derived from tetraether composition may The three independent temperature proxies

be more representative of summer processes Alnus incana* measured from the same peat deposit converge when temperatures are higher and soils have Saxifraga oppostifola on an Arctic MAT for the Pliocene of ~0 °C that Myrica gale greater water content, promoting the facultative Empetrum nigrum corresponds with a ΔMAT of ~19 °C (Table 1). anaerobic bacteria that are hypothesized to syn- Andromeda polifola Because these independent proxy estimates of Carex diandra thesize tetraether lipids (Weijers et al., 2007). Potentilla norvegica MAT are statistically indistinguishable from Menyanthes trifoliata However, temperature estimates derived from Vaccinium vitis-idae each other (Table 1), we combined all of the tetraethers in modern soils from Svalbard, Nor- Betula nana* estimates into a joint distribution that was then Genera Betula papyrifera* way (MAT ≈ –4 °C) were within 2 °C of instru- occidentalis resampled using a bootstrap technique (see ≈ Salix spp. mental temperature records (MAT –6 °C), Picea mariana Appendix). This approach yielded a composite suggesting that tetraethers are an effective proxy Pinus albicaulis* MAT estimate of –0.4 ± 0.4 °C, and a ΔMAT Larix laricina* for reconstructing temperatures from paleosols of 19.3 °C (Table 1). This more robust tempera- at high latitudes (Peterse et al., 2009). It is also ture estimate suggests that Arctic temperatures possible that MAT estimates from isotopes −20 −10 0 10 20 were remarkably warmer during the Pliocene and annual ring widths may be biased toward Temperature (°C) (Fig. 2). In fact, these estimates are 5–10 °C summer months because this is when trees are warmer than previous proxy estimates (Bal- accruing biomass and synthesizing cellulose Figure 1. Paleotemperatures for the Pliocene lantyne et al., 2006; Elias and Matthews, 2002). from the available water. Although it is unclear Beaver Pond site inferred from paleovegeta- These temperature estimates are also consider- tion. Plotted are the inferred temperature es- how much of the water pool available for photo- timates based on the coexistence approach ably warmer than model simulations at high synthesis is derived from winter versus summer (Mosbrugger and Utescher, 1997). Black latitudes (Haywood et al., 2009). However, cli- precipitation, research on oxygen isotopes in boxes represent mean annual temperature mates at high latitude are known to be very sen- modern larch suggests that larch rely on spring (MAT) estimates for each genus identifi ed sitive to orbital parameters affecting insolation snow melt and thus integrate the isotopic signal at the site, with gray whiskers representing (Ravelo et al., 2004), and thus proxy estimates the minimum MAT and maximum MAT of the of annual precipitation (Sugimoto et al., 2002). range for each genus. The area of the box is with uncertain age constraints are not directly Furthermore, these two independent proxies relative to the precision of the temperature comparable to model simulations that typically effectively yield the same temperature estimate estimate (1/SE). The overall MAT inferred span hundreds of years. Nonetheless, the agree- (Table 1), providing greater confi dence in our as the mean of the temperature range was ment among these estimates indicates signifi - –0.4 °C (black line). The overall cold-month estimates of appreciably warmer temperatures mean temperature (CMMT) inferred as the cant Arctic warming during the Pliocene. in the Arctic during the Pliocene. mean across all taxa was –11.6 °C (gray solid The third temperature proxy we employed line), and the overall warm-month mean temperature (WMMT) inferred as the mean at the Beaver Pond site was paleovegetation 3.0 Tree ring isotopes composition (Table 1). By utilizing a slightly across all taxa was 14.4 °C (gray dashed line). Salix spp. refers to Salix alaxensis and Paleovegetation modifi ed coexistence approach (Mosbrugger Salix arbusculoides and their combined cli- Tetraethers and Utescher, 1997), we were able to generate matic range. Species noted with an asterisk 2.0 Composite climatic ranges for 16 plant genera identifi ed were selected as best matches for at the site and a MAT estimate (see Appendix). identifi ed to genus only. Several genera identifi ed at the Beaver Pond site 1.0 occur in much warmer climates today, including Northern white cedar (Thuja occidentalis) that with nearest living relatives (Elias and Mat- Probability density has a modern distribution with a MAT range of thews, 2002; Greenwood and Wing, 1995). 0.0 –4.4 to 14.2 °C and Rough Cinquefoil (Poten- However, the paleovegetation-based CMMT −2 −1 0 1 2 tilla norvegica) that has a modern distribution of –11.6 ± 7.1 °C is considerably warmer than MAT (°C) with a MAT range of –5.1 to 14.0 °C (Fig. 1). the CMMT of –26.6 °C derived from the bee- Figure 2. Probability density functions of However, also preserved at this site are cooler- tle assemblage (Elias and Matthews, 2002). mean annual temperature (MAT) estimates climate taxa such as a dwarf shrub form of Although beetles and plants are both effectively for the Arctic during the Pliocene based on Birch (Betula nana) (MAT = –13.4 to 3.7 °C). ectothermic, and thus sensitive indicators of three independent proxies. Plotted are the Although a broad range of TSTs was inferred ambient temperature, beetles can better regu- three bootstrapped estimates of MAT de- rived from our three proxies (colored lines) from individual plant taxa present at our site, late their temperature under cold conditions by and the density function for the composite the coexistence interval among all taxa gives physiological mechanisms, such as increased estimate of MAT derived from resampling an estimated MAT of –0.4 ± 4.1 °C, yielding a respiration (Morgan and Bartholomew, 1982), the joint distribution across all three inde- ΔMAT of ~19 °C (Table 1). Applying a similar and behavioral mechanisms, such as burrowing pendent proxies (gray fi lled). coexistence approach, we also estimated a cold- (Strathdee and Bale, 1998). These mechanisms month mean temperature (CMMT) of –11.6 allow beetles to inhabit colder climates and may ± 7.1 °C and a warm-month mean temperature explain the anomalously low CMMT indicated The Pliocene is a paradox when compared (WMMT) of 14.4 ± 2.0 °C. by the beetle assemblage. In fact, Elias and Mat- to other Cenozoic warm intervals because Our WMMT estimate of 14.4 ± 2.0 °C com- thews (2002) showed a much better relation- global mean temperatures were 2–3 °C warmer pares well with a previous WMMT estimate ship between modern beetle distributions and than present (Dowsett, 2007), despite levels 2 2 of 12.4 °C based on the assemblage of fos- WMMT (r = 0.95) than CMMT (r = 0.80), of atmospheric CO2 that were only slightly sil beetles from this site (Elias and Matthews, suggesting that the seasonal range of warming higher than preindustrial levels (Fedorov et al., 2002), which is not too surprising given that inferred from the beetle assemblage (ΔMAT = 2006). Recent proxy estimates that are better

both methods are based on correspondence 10.0 to 14.8 °C) is probably biased low. constrained indicate Pliocene atmospheric CO2

604 GEOLOGY, July 2010

Downloaded from http://pubs.geoscienceworld.org/gsa/geology/article-pdf/38/7/603/3538666/603.pdf by guest on 25 September 2021 levels of ~390 ppmv (Pagani et al., 2010), which (see Fig. 3B; Table DR2 in the Data Reposi- the modeled effects of clouds on Arctic climate are comparable to today’s levels (~385 ppmv). tory), resulting in an “Arctic tail” in the latitu- are highly dependent upon the physical proper- However, if we place our estimates of Pliocene dinal temperature distribution (Fig. 3B). This ties and seasonal distribution of clouds. A recent Arctic TSTs in a global context, most of the tail is even more pronounced during the , study by Abbot and Tziperman (2008) shows temperature response to climate forcing is due when temperatures in the Arctic were almost that deep convective clouds can produce signifi - to increased temperatures at high latitudes. This 35 °C warmer than present (Fig. 3B). There- cant winter warming under ice-free conditions. increased at high latitudes fore, the decreased temperature gradient can be In contrast, reduced summer cloud cover may has resulted in a greatly reduced latitudinal tem- explained in part by increased poleward heat lead to warmer, more equable climates (Kump perature gradient (Fig. 3A). Several mechanisms transport, but other physical mechanisms must and Pollard, 2008). Although these studies have been hypothesized to explain this reduced be invoked to explain the observed asymmetry clearly illustrate the importance of cloud cover temperature gradient, including increased pole- in the latitudinal temperature gradient. on the radiative balance of the Arctic, variations ward heat transport, decreased ice albedo, and This pronounced Arctic tail in the latitudinal in the seasonal sign and magnitude of the radia- changes in cloud cover (Fedorov et al., 2006). temperature gradient can be explained in part tive forcing associated with clouds must be fur- Most of the observed decline in the latitudinal by a greater ice-albedo feedback in the Arctic ther investigated. temperature gradient during the Pliocene can be compared to . Ice has a much higher Therefore the reduced temperature gradi- explained by increased poleward heat transport. albedo (~0.7) (Lindsay and Rothrock, 1994) rel- ent can be explained in part by an increase in Model experiments suggest that a 15% increase ative to the ocean (~0.07) (Payne, 1972) than it meridional heat transport by either the oceans in the poleward transport of sensible heat by does relative to forest (~0.1–0.4) (Betts and Ball, or the atmosphere. However, the mechanism of ocean circulation is suffi cient to explain the 1997). Therefore removal of ice from the ocean increased poleward heat transport cannot be the reduced latitudinal gradient in sea surface tem- decreases surface albedo by a factor of ten, com- only physical mechanism driving the reduced peratures (SSTs) observed during the Pliocene pared with removal of ice from the land, which temperature gradient because it is in fact the sur- (Dowsett et al., 1992). Model simulations of the decreases surface albedo by a factor of only two face temperature gradient that ultimately drives atmosphere also indicate an increase in the pole- to seven. Given that the northern polar region is the fl ux of heat poleward. Thus other mecha- ward transport of latent heat as precipitation, dominated by water, whereas the southern polar nisms such as ice-albedo feedbacks and changes mainly at tropical latitudes (Haywood et al., region is dominated by land, one would expect in cloud cover must be invoked to explain both 2009). However, Earth’s surface energy balance a greater temperature response to changes in ice the reduced temperature gradient and its asym- dictates that net poleward heat transport should extent in the Arctic than Antarctica. However, metry. The interactive feedback between clouds be symmetrical in both hemispheres. Accord- the exact timing and extent of formation and ice is probably the phenomenon that best ing to our temperature estimates, the Arctic was and continental glaciation in the Arctic during explains the amplifi cation of Arctic tempera- ~19 °C warmer during the Pliocene than today, the Pliocene remains uncertain (Zachos et al., tures because as sea ice is removed there is a whereas Antarctica was only ~13 °C warmer 2008). Although there is limited evidence of dramatic decrease in albedo and an increase in ice-rafted debris in the Arctic from the moisture source necessary for cloud formation. and into the Eocene, suggesting some continen- The Arctic is clearly a bellwether for mod- tal glaciation (Stickley et al., 2009; St. John and ern climate change. Arctic temperatures have 30 A Krissek, 2002), empirical evidence suggests increased more rapidly in response to anthro- that widespread Northern Hemisphere glacia- pogenic greenhouse forcing than global tem-

C) 10 tion did not occur until 2.75 Ma (Ravelo et al., peratures (ACIA, 2004). Our independent proxy ° 2004), which is substantiated by recent Pliocene estimates indicate that Arctic temperatures dur- Pliocene −10 paleotemperature SST estimates near Svalbard ing the Pliocene were considerably warmer than MAT ( Eocene between 10 and 18 °C (Robinson, 2009). This previous estimates derived from empirical prox- Modern is corroborated by model simulations indi- ies (Ballantyne et al., 2006; Elias and Matthews, −30 cating that atmospheric CO2 levels must fall 2002) and climate model simulations (Haywood 30 B Pliocene below 280 ppmv to promote widespread con- et al., 2009), despite estimates of Pliocene atmo-

C) Eocene tinental glaciation of the Northern Hemisphere spheric CO levels that are comparable to today ° 2 20 (DeConto et al., 2008) and 250 ppmv to pro- (Pagani et al., 2010). This indicates that climate 10 mote major continental glaciation on Greenland models do not incorporate the full array of MAT (

Δ (Lunt et al., 2008), both of which are well below atmospheric, biospheric, and cryospheric feed- 0 recent estimates of Pliocene atmospheric CO2 back mechanisms necessary to simulate Arctic −90 −60 −30 0 30 60 90 estimates of ~390 ppmv (Pagani et al., 2010). climate. Regardless of the feedback mechanism Latitude Our MAT estimates near 0 °C, in combination responsible for amplifi ed Arctic temperatures, Figure 3. Pliocene temperature gradient in with well-documented widespread forest eco- our results indicate that a signifi cant increase a global Cenozoic context. A: Gradients of systems (Matthews and Ovenden, 1990), pro- in Arctic temperatures may be imminent in

mean annual temperature (MAT) are plotted vide independent lines of evidence suggesting response to current atmospheric CO2 levels. for modern (gray line), Pliocene (black solid limited glacial extent in the Arctic during the line), and early Eocene (black dashed line) by Pliocene. Thus the differential magnitude of ACKNOWLEDGMENTS latitude. The three independent temperature ice-albedo feedbacks between the Arctic and Ballantyne and Eberle were supported by NSF, estimates from this study are plotted as over- Greenwood and Rybczynski were supported by the lying fi lled black circles with standard error Antarctica may help to explain apparent asym- Natural Sciences and Engineering Research Council bars. Previous Pliocene MAT estimates from metries in the latitudinal temperature gradient. of Canada, and Sinninghe Damsté was supported by the Arctic (×) were not included when fi tting Clouds have also been invoked as a physi- Netherlands Organization for Scientifi c Research the spline (see Table DR2 [see footnote 1]). cal mechanism to explain the amplifi cation of and the European Research Council. This manu- B: Difference (ΔMAT) between modern MAT script benefi ted from comments by C.R. Harington. and Eocene (black dashed) and the Pliocene Arctic temperatures during warmer This research was authorized by Nunavut’s Depart- (black solid) are plotted for comparison. periods (Kump and Pollard, 2008). However, ment of Culture, Language, Elders and Youth, and

GEOLOGY, July 2010 605

Downloaded from http://pubs.geoscienceworld.org/gsa/geology/article-pdf/38/7/603/3538666/603.pdf by guest on 25 September 2021 the Qikiqtani Inuit Association. This is contribution ACIA, 2004, Impacts of a warming Arctic—Arctic Arctic/Subarctic North America: A review of 00310 to the Polar Continental Shelf Program and climate impact assessment: Cambridge, UK, data: Arctic, v. 43, p. 364–392. the International Polar . Cambridge University Press, 144 p. Morgan, K.R., and Bartholomew, G.A., 1982, Ballantyne, A.P., Rybczynski, N., Baker, P.A., Har- Homeothermic response to reduced ambi- APPENDIX ington, C.R., and White, D., 2006, Pliocene ent temperature in a scarab beetle: Science, Arctic temperature constraints from the growth v. 216, p. 1409–1410, doi: 10.1126/science.216 Site Description rings and isotopic composition of fossil larch: .4553.1409. The Beaver Pond site is an exceptionally well- Palaeogeography, Palaeoclimatology, Palaeo- Mosbrugger, V., and Utescher, T., 1997, The coex- preserved, organic-rich peat layer with many pieces ecology, v. 242, p. 188–200, doi: 10.1016/j istence approach—A method for quantitative of in situ wood (Ballantyne et al., 2006). The peat .palaeo.2006.05.016. reconstructions of terrestrial palaeo- layer is embedded in a sandy deposit capped by a Betts, A.K., and Ball, J.H., 1997, Albedo over the climate data using plant fossils: Palaeogeogra- glacial deposit. A diverse assemblage of bryophytes, forest: Journal of Geophysical Research, v. 102, phy, Palaeoclimatology, Palaeoecology, v. 134, vascular plants, , and vertebrates are p. 28,901–28,909, doi: 10.1029/96JD03876. p. 61–86, doi: 10.1016/S0031-0182(96)00154-X. represented by fossils found at this site (Fig. DR1 DeConto, R.M., Pollard, D., Wilson, P.A., Palike, H., Pagani, M., Liu, Z., LaRiviere, J., and Ravelo, A.C., in the Data Repository). The mammalian taxa at this Lear, C.H., and Pagani, M., 2008, Thresholds 2010, High Earth- climate sensitivity site correspond with taxa from North America and for Cenozoic bipolar glaciation: Nature, v. 455, determined from Pliocene con- China that have been dated to be 4 to 5 m.y. old (Ted- p. 652–656, doi: 10.1038/nature07337. centrations: Nature Geoscience, v. 3, p. 27–30, ford and Harington, 2003). Dowsett, H.J., 2007, The PRISM paleoclimate re- doi: 10.1038/ngeo724. construction and Pliocene sea-surface tempera- Payne, R.E., 1972, Albedo of the sea surface: Journal ture, in Williams, M., et al., eds., Deep-time of the Atmospheric Sciences, v. 29, p. 959– Paleoclimate Proxies perspectives on climate change: Marrying the 970, doi: 10.1175/1520-0469(1972)029<0959 Temperatures were fi rst reconstructed using a signal from computer models and biological :AOTSS>2.0.CO;2. novel approach based on branched glycerol dialkyl proxies: Micropaleontological Society Special Peterse, F., Kim, J.H., Schouten, S., Kristensen, D.K., glycerol tetraethers present in the membranes of Publication 2, p. 459–480. Koç, N., and Sinninghe Damsté, J.S., 2009, anaerobic soil bacteria abundant in peat bogs (Wei- Dowsett, H.J., Cronin, T.M., Poore, R.Z., Thomp- Constraints on the application of the MBT/ jers et al., 2007). Tetraethers were extracted and son, R.S., Whatley, R.C., and Wood, A.M., CBT palaeothermometer at high latitude envi- analyzed as described by Weijers et al. (2007). The 1992, Micropaleontological evidence for in- ronments (Svalbard, Norway): Organic Geo- conservative total standard error associated with the creased meridional heat transport in the North chemistry, v. 40, p. 692–699, doi: 10.1016/j. MAT estimate based on this transfer function has Ocean during the Pliocene: Science, orggeochem.2009.03.004. been reported as 5.0 °C (Data Repository). v. 258, p. 1133–1135, doi: 10.1126/science.258 Ravelo, A.C., Andreasen, D.H., Lyle, M., Lyle, O.L., The second proxy we used to reconstruct Pliocene .5085.1133. and Wara, M.W., 2004, Regional climate shifts temperatures builds upon the multivariate approach Efron, B., and Tibshirani, R.J., 1997, An introduc- caused by gradual global cooling in the Plio- δ18 of measuring oxygen isotopes ( O) and annual tion to the bootstrap: New York, Chapman and cene epoch: Nature, v. 429, p. 263–267, doi: growth rings in fossil larch to estimate MAT (Ballan- Hall, 450 p. 10.1038/nature02567. tyne et al., 2006). We improved upon this approach Elias, S.A., and Matthews, J.V., Jr., 2002, Arctic North Robinson, M.M., 2009, New quantitative evidence of δ18 by measuring O values in the cellulose of mosses American seasonal temperatures from the latest extreme warmth in the Pliocene Arctic: Stratig- δ18 to infer the O signature of the source water using Miocene to the Early , based on mu- raphy, v. 6, p. 265–275. a Finnigan MAT Delta Plus XL mass spectrometer tual climatic range analysis of fossil beetle as- Stickley, C.E., St. John, K., Koc, N., Jordan, R.W., (Data Repository). semblages: Canadian Journal of Earth Sciences, Passchier, S., Pearce, R.B., and Kearns, L.E., The third temperature proxy was based on a com- v. 39, p. 911–920, doi: 10.1139/e01-096. 2009, Evidence for middle Eocene Arctic sea parison of plant taxa present at the site (Matthews Environment Canada, 2009, National Climate Data ice from and ice-rafted debris: Nature, and Ovenden, 1990) with nearest living relatives and and Information Archive: Canadian Weather v. 460, p. 376–379, doi: 10.1038/nature08163. their climatic ranges. Several databases of modern Offi ce: www.climate.weatheroffi ce.ec.gc.ca. St. John, K.E.K., and Krissek, L.A., 2002, The late taxa were queried for nearest living relatives and Fedorov, A.V., Dekens, P.S., McCarthy, M., Ravelo, Miocene to Pleistocene ice-rafting of their climatic ranges (see the Data Repository). We A.C., deMenocal, P.B., Barreiro, M., Pacanow- southeast Greenland: Boreas, v. 31, p. 28–35, then used the coexistence approach of Mosbrugger ski, R.C., and Philander, S.G., 2006, The Plio- doi: 10.1080/03009480210651. and Utescher (1997), whereby a coexistence interval cene paradox (mechanisms for a permanent Strathdee, A.T., and Bale, J.S., 1998, Life on the edge: is calculated as the warmest minimum and coolest El Niño): Science, v. 312, p. 1485–1489, doi: Insect ecology in Arctic environments: Annual maximum MAT values for all nearest living relatives. 10.1126/science.1122666. Review of Entomology, v. 43, p. 85–106. We varied their method by expressing the MAT as Greenwood, D.R., and Wing, S.L., 1995, Eocene Sugimoto, A., Yanagisawa, N., Naito, D., Fujita, the mean of the temperature range and expressing the continental climates and latitudinal temperature N., and Maximov, T.C., 2002, Importance error as the difference between MAT and the range gradients: Geology, v. 23, p. 1044–1048, doi: of permafrost as a source of water for plants of temperatures. The cold-month mean temperature 10.1130/0091-7613(1995)023<1044:ECCALT in east Siberian taiga: Ecological Research, (CMMT) and the warm-month mean temperature >2.3.CO;2. v. 17, p. 493–503, doi: 10.1046/j.1440-1703 (WMMT) were calculated in the same manner. Haywood, A.M., Chandler, M.A., Valdes, P.J., Salz- .2002.00506.x. mann, U., Lunt, D.J., and Dowsett, H.J., 2009, Tedford, R.H., and Harington, C.R., 2003, An Arc- Statistical Analysis Comparison of mid- predic- tic mammal fauna from the Early Pliocene of The distributions of MAT estimates derived for tions produced by the HadAM3 and GCMAM3 North America: Nature, v. 425, p. 388–390, each proxy were evaluated using a two-tailed Stu- General Circulation Models: Global and Plan- doi: 10.1038/nature01892. dent’s t-test. To better approximate the distributions etary Change, v. 66, p. 208–224, doi: 10.1016/j Weijers, J.W.H., Schouten, S., van den Donker, J.C., of MAT derived from the different proxies, we used .gloplacha.2008.12.014. Hopmans, E.C., and Sinninghe Damsté, J.S., a bootstrap technique (Efron and Tibshirani, 1997), Kump, L., and Pollard, D., 2008, Amplifi cation of 2007, Environmental controls on bacterial tet- whereby distributions were resampled with sample Cretaceous warmth by biological cloud feed- raether membrane lipid distribution in soils: replacement 100,000 times and individual estimates backs: Science, v. 320, p. 195, doi: 10.1126/ Geochimica et Cosmochimica Acta, v. 71, were weighted by the inverse of their standard error science.1153883. p. 703–713, doi: 10.1016/j.gca.2006.10.003. (1/SE). Lastly, a composite distribution of Arctic Plio- Lindsay, R.W., and Rothrock, D.A., 1994, Arctic Zachos, J.C., Dickens, G.R., and Zeebe, R.E., 2008, cene temperature estimates was derived by combining sea ice albedo from AVHRR: Journal of Cli- An early Cenozoic perspective on greenhouse individual MAT estimates into a joint distribution and mate, v. 7, p. 1737–1749, doi: 10.1175/1520 warming and carbon-cycle dynamics: Nature, resampling according to the bootstrap technique. All -0442(1994)007<1737:ASIAFA>2.0.CO;2. v. 451, p. 279–283, doi: 10.1038/nature06588. statistical analyses were performed using R. Lunt, D.J., Foster, G.L., Haywood, A.M., and Stone, E.J., 2008, Late Pliocene Greenland glacia- REFERENCES CITED tion controlled by a decline in atmospheric Manuscript received 27 October 2009

Abbot, D.S., and Tziperman, E., 2008, Sea ice, high- CO2 levels: Nature, v. 454, p. 1102–1105, doi: Revised manuscript received 29 January 2010 latitude convection, and equable climates: Geo- 10.1038/nature07223. Manuscript accepted 2 February 2010 physical Research Letters, v. 35, L03702, doi: Matthews, J.V., Jr., and Ovenden, L.E., 1990, Late 10.1029/2007GL032286. Tertiary plant macrofossils from localities in Printed in USA

606 GEOLOGY, July 2010

Downloaded from http://pubs.geoscienceworld.org/gsa/geology/article-pdf/38/7/603/3538666/603.pdf by guest on 25 September 2021