Eocene continental climates and latitudinal temperature gradientsDavid R. Greenwood Department of Paleobiology, National Museum of Natural History, MRC 121, Smithsonian Institution,

Washington, D.C. 20560Scott L. Wing

ABSTRACTGlobal climate during theMesozoic and early Cenozoic is thought to have been warmer

than at present, but there is debate about winter temperatures. Paleontological data in-dicate mild temperatures even at high latitudes and in mid-latitude continental interiors,whereas computer simulations of continental paleoclimates produce winter temperaturescloser to modern levels. Foliar physiognomy and floristic composition of 23 Eocene florasfrom the interior of North America and Australia indicate cold month means generally>2 &C, even where the mean annual temperature (MAT) was <15 &C. Reconstructed Eo-cene latitudinal gradients of MAT are curvilinear but are about 0.4 &C per 1& of latitude incontinental interiors at mid-latitudes, much less than the 0.8–1.0 &C per 1& of latitudeobserved in eastern and central North America today, but similar to modern gradients inthe Southern Hemisphere mid-latitudes and on the west coast of North America. Latitu-dinal temperature gradients reconstructed here are broadly representative of Eocene cli-mates, showing that the discrepancy between proxy data and simulations will not be re-solved by regional adjustments to paleogeography or reinterpretation of individual fossilassemblages. Similar discrepancies between proxy data and general circulation modelsimulations for other time periods suggest that there is a basic flaw with the way climatemodels simulate heat transport to, or loss from, continental surfaces.

INTRODUCTIONIsotopic marine records and paleontolog-

ical evidence indicate that the early Eocenewas the warmest part of the Cenozoic (Mill-er et al., 1987) and that latitudinal sea-sur-face temperature (SST) gradients were shal-lower than at present (Zachos et al., 1994).Early general circulation models (GCMs) ofEocene continental climates predictedlower mean annual temperatures (MAT)and greater thermal seasonality for conti-nental interiors than inferred from fossil ev-idence, cold month means (CMM) beingsubstantially below freezing (Sloan and Bar-ron, 1990, 1992).More recent modeling studies have given

MAT estimates close to paleontologicallyinferred values and support shallow Eocenelatitudinal temperature gradients, but thereis still broad disagreement between proxydata and simulations of CMM values (mod-eling—Sloan and Barron, 1992; Sloan, 1994;proxy data—Hutchison, 1982; Wing andGreenwood, 1993; Markwick, 1994). Similardifferences between simulations and proxydata for Permian Gondwana have been inpart attributed to the effect of lakes and un-certainty about paleocoastlines and paleo-elevations of interior regions (Yemani,1993). Recent simulations show that the Eo-cene Green River Lake system would haveraised minimum winter temperatures in asmall downwind area (Sloan, 1994). In thispaper we summarize the geographic extentof equable climates (i.e., those with littleseasonal variation in temperature) duringthe Eocene to see if regional paleogeogra-phy can account for equable continentalinteriors.

North America and Australia were largeland areas with well-defined continental in-teriors during the Paleogene. The latitude ofNorth America has changed little since theEocene, but Australia has drifted from highto middle southern latitudes since the earlyPaleogene (Fig. 1). Topography in Australiawas subdued during Paleogene time (Quilty,1994), removing paleoaltitude as a potentialsource of confusion in reconstructing climate.

METHODS OF PALEOCLIMATICINFERENCEPlant populations persist only if all stages

of the life cycle can tolerate the local cli-mate; therefore, the distribution of plants isinfluenced strongly by climate. Plant mega-fossils, especially leaves, are generally de-posited locally and reflect local vegetationand climate (Greenwood, 1992). Paleocli-mate has been inferred from fossil plants onthe basis of the climatic preferences of near-est living relatives of taxa found in the fossilassemblage and correlations between dicotleaf physiognomy and climate parameters.We apply both approaches here.

Temperature Limits of PalmsPalms cannot withstand sustained freezes

and are today limited to areas that arelargely free of frost (Fig. 2). This frost in-tolerance is due to their soft, water-rich tis-sue, their exposed growth point(s), and theirlack of physiological mechanisms for frosttolerance. Only a few degrees of cold hard-ening can be induced, and no species showsmuch seasonal fluctuation in frost sensitivity(Sakai and Larcher, 1987). The most cold-tolerant palm species (e.g., Trachycarpus for-

tunei) show differential freezing tolerance atdifferent stages of life and between differentorgans: adult palms can survive at 210 to212 8C for a few hours but are killed by sus-tained freezes; roots and seedlings are killedat21.5 to28 8C, and all palms are killed byground frost (Sakai and Larcher, 1987).Frost sensitivity of palms restricts them toclimates with MAT .10 8C, CMM .5 8C,and yearly minimum temperature. 210 8C(Fig. 2).Nearest living relative–based paleocli-

matic inference can be unreliable becauselineages evolve and because the nearest liv-ing relatives of some Tertiary taxa (e.g.,Metasequoia) no longer occupy the range ofclimatic conditions they can tolerate (Wolfe,1979). This is likely not a problem withpalms, which are morphologically and taxo-nomically diverse and yet uniformly limitedin their cold hardiness. Furthermore, Eo-cene palms represent modern tribes (e.g.,Corypheae, Nypoideae), and have leaf,stem, and root anatomy indistinguishablefrom that of modern palms (Daghlian, 1978;Schaal and Ziegler, 1992), indicating thatthe climatic tolerances of Paleogene palmsare well represented by living species. Fossilpalms are reliable ‘‘threshold’’ measures ofpaleo-MAT and CMM. Global climaticcooling in the Cenozoic may have favoredevolution of palms with greater cold toler-ance, so palm-based estimates of minimumtemperature are more likely to be too coolthan too warm.

Foliar PhysiognomyLeaf margin analysis is based on the pos-

itive correlation between MAT and the pro-portion of entire-margined woody dicot spe-cies in 34 extant East Asian mesic forests(Wolfe, 1979). Leaf margin analysis hasbeen criticized for being based on floral lists,not vegetation samples, and for ignoring therole of soil and moisture in controlling phys-iognomy (Dolph and Dilcher, 1979). How-ever, data from North American and Aus-tralian forests have shown similar thoughless-precise correlations between leaf mar-gin type and MAT (Dolph and Dilcher,1979; Greenwood, 1992).The CLAMP (climate analysis multivari-

ate program) method developed by Wolfe(1993) improves on leaf margin analysis bydescribing foliar physiognomy with multiplesize, shape, and margin states. The CLAMPdata set is derived largely from modern veg-etation in North America and Central

Geology; November 1995; v. 23; no. 11; p. 1044–1048; 3 figures; 1 table.1044

America. Wolfe used reciprocal averagingordination to study the relation of 29 leafcharacters to seven climatic parameters, in-cluding MAT, CMM, and mean annualrange of temperature (MART). We haveanalyzed the CLAMP data set using multi-ple regression to develop a set of equationsto predict MAT, CMM, and MART on thebasis of a subset of the physiognomic vari-ables (Wing and Greenwood, 1993). A sim-ilar analysis by Gregory and Chase (1992)was based on an earlier version of theCLAMP data set.

EOCENE CLIMATE ESTIMATESNorth AmericaMultiple regression analyses of North

American Eocene floras estimated MATs inthe range 9–24 8C (Table 1). Inferred ther-mal seasonality was moderate for all sites(MART 8–19 8C), and winter temperatureswere above freezing (CMM .0 8C exceptfor Arctic sites), even where MAT estimateswere 10–15 8C. CMM estimates for theBears Paw, upper Willwood, Camels Butte,Wind River, and Kisinger Lakes assem-blages are particularly important for under-standing the influence of a possible GreenRiver Lake effect. These sites are 120–650km north of the lake, and the first three pre-date major lake development in the GreenRiver basin. CMM at these sites is con-strained by palms and crocodilians to havebeen .5 8C, although the multiple regres-

sion estimates of CMM at Bears Paw andCamels Butte are lower. The Arctic florasgive CMM estimates from 22.0 to 20.8 8C,but the presence of crocodilians in the Arc-tic implies a minimum CMM of 5–7 8C(Markwick, 1994). The discrepancy betweennearest living relative and multiple regres-sion model estimates of CMM for florasnorth of about 508N paleolatitude probablyresults from deciduousness induced by lowwinter light; floras that were deciduous inresponse to light rather than winter coldmay produce physiognomy-based tempera-ture estimates that are too low (Basinger etal., 1994).The North American paleotemperature

estimates show expected trends; MAT and

CMM decrease with increasing latitude, andinterior sites have greater MART thancoastal sites (Figs. 1 and 3). The coolest es-timates are derived from floras (Yellow-stone and Bears Paw) that are inferred to beupland on the basis of geological evidence(Fritz and Harrison, 1985; Fig. 1). Our es-timates at lower latitudes (e.g., Puryear) arealso consistent with nearest living relativeestimates from a mangrove biota found at asimilar latitude in Texas (Westgate and Gee[1990] estimated SST of 24 8C and CMM of19 8C).Although our paleoclimate estimates

show the expected trends, leaf margin anal-ysis and nearest living relative analyses ofsome of the floras used in this study havegiven MAT and CMM estimates 2–3 8Chigher than ours (MacGinitie, 1974; Hickey,1977). The multiple regression approachyields cooler estimates because it is sensitiveto the low proportions of species in Eocenefloras with emarginate apices, and becauseof deciduousness induced by light ratherthan temperature seasonality. Particularly athigh latitudes, nearest living relative–basedminimums of the CMM are probably morereliable than those based on physiognomy.

AustraliaEstimates of MART (12–22 8C) and

CMM (5–15 8C) suggest that no area of Aus-tralia (at sea level) had significant freezingtemperatures (Table 1). The MAT andCMM multiple regression estimates fromAustralian floras are consistent with previ-ous physiognomic and nearest living relativeanalyses, including the presence of palms,particularly the thermophilic mangrovepalm Nypa (Greenwood, 1994). The domi-nance of Lauraceae at many of these sitesand the presence of crocodiles throughoutthe Eocene (Willis and Molnar, 1991) arealso consistent with our MAT and CMM es-timates. Eocene palynofloras in southwest-ern Australia, and eastward along the coast,indicate widespread development of man-

Figure 1. Eocene reconstruction (from PGIS/Mac™) showing distribution of megafloras pro-viding quantitative paleoclimate estimates, and frost intolerant plants. Squares 5 palms, tri-angles 5 cycads, gingers, or tree ferns, and circles 5 no frost intolerant plants; green 5lowlands, yellow 5 uplands, red 5 higher upland areas. Numbers 1–23 are keyed to Table 1;24—Tierra del Fuego, 25—Zhangye, 26—Yilan, 27—Kamchatka, 28—Sorachi, 29—LondonClay, 30—Messel, 31—Geiseltal.

Figure 2. Modern distribu-tion of palms worldwideand detail for easternUnited States (Sabal spp.)and New Zealand (Rhopa-lostylis sapida) showingclimatic limits to palm dis-tribution. Solid red lines 5average duration (DUR.)of longest annual frost;dashed blue lines 5 coldmonth mean (CMM) tem-perature in °C. Data onpalm distribution fromMoore (1973), Little (1977),and Wardle (1991); cli-matic data from Gentilli(1971) and Bryson andHare (1974).

GEOLOGY, November 1995 1045

groves, including Nypa (Truswell, 1993).These floras imply SST .20 8C adjacent tomuch of southern Australia at paleolatitude508S.Many of the plant taxa in the Australian

megafloras are today restricted to, or mostspeciose in, high-rainfall equable climateswith a MAT of 18–22 8C (Kershaw and Nix,1988). Greenhouse experiments with near-est living relatives of key Australian Eocenegenera indicate that the modern tolerancesof these taxa have a strong physiological ba-sis (Read and Hill, 1989), and are thereforelikely to reflect Eocene tolerances. Most ofthe Australian megafloras analyzed werenear Eocene coasts, but the Nelly Creek andPoole Creek floras are indicators of paleo-climate in the interior. Leaf cuticles and pol-len from these sites are similar to those incoeval sites in basins even farther into theAustralian interior (Truswell, 1993).

Other AreasAt middle to high latitudes throughout

the world there is evidence for warm cli-mates in the Eocene. Tropical and subtrop-ical elements are common in the early Eo-cene London Clay flora, including Nypa andother palms, mangroves, and zingiberaleans(Collinson, 1983). Thermophilic fauna andflora in Germany indicate MAT.20 8C andCMM .10 8C during the early and middleEocene (Schaal and Ziegler, 1992). Eocene

palms have been reported from Kamchatka(Budantsev, 1992), inland north China(Guo, 1990), northern Japan (Tanai, 1992),and Tierra del Fuego (Romero, 1993). InChina vegetation as far as 508N paleolati-tude was dominantly broad-leaved ever-green, implying a CMM .1 8C (Guo, 1990;Tanai, 1992). The paleogeographic posi-tions of these points are plotted in Figure 1.Foliar physiognomy and presence ofNypa

indicate MAT of 15–20 8C, and CMM.10 8C for New Zealand middle Eocenefloras at paleolatitudes of 458–558S (Pock-nall, 1990; Pole, 1994). Eocene floras fromthe Antarctic peninsular (Birkenmajer andZastawniak, 1986; Case, 1988) and Trans-antarctic Mountains (Stilwell et al., 1993; M.Pole, 1994, personal commun.) containAraucaria and Nothofagus, suggesting MAT.6 8C, and CMM .0 8C, at 658–758S. Pa-leobotanicalestimatesforAntarcticaarecon-sistent with SST estimates of 8–12 8C, de-rived from oxygen isotope values of shellsfrom James Ross Island (Ditchfield et al.,1994), and winter SSTs of 9 8C for the Wed-dell Sea (Stott et al., 1990). With the excep-tion of the Zhangye and Yilan sites in north-ern China and the Mount Discovery site inAntarctica, the sites listed here are within100 km of present-day coastlines, and mayhave been coastal in the Eocene. High-lat-itude warmth was clearly global in the Eo-

cene, but only North America and Australiahave produced abundant evidence for warmcontinental interiors.

Latitudinal Temperature GradientsUsing modern meteorological data, we

reconstructed latitudinal gradients of theMAT and CMM for the Northern andSouthern hemispheres, along east and westcoasts, and north-south transects throughthe continental interiors of Antarctica, Aus-tralia, Africa, and North America and SouthAmerica (Fig. 3). Modern data have beenplotted as an envelope, the combined inte-rior and coastal data trends being expressedas a polynomial regression. The MAT de-creases poleward in the Southern Hemi-sphere in an almost linear manner at a gra-dient of 0.4 8C/18 latitude (west coasts) to0.6 8C/18 latitude (interior and east coasts),to high-middle to high latitudes, where thegradient steepens. The MAT gradient alongthe North American west coast is mostly lin-ear at about 0.4 8C/18 latitude from theequator to about 508–608N, where a signif-icant steepening of the gradient to 1.2 8C/18latitude occurs. The North American inte-rior and east coast gradients of the MATand CMM are almost linear, with gradientsof 1.0 8C/18 latitude and 0.8 8C/18 latitude.The principal reason for nonlinear gradientsin both hemispheres is a significant decreasein CMM (but not warm month mean) at thepoints of deflection.Our estimates of Eocene temperatures

have gaps along the latitudinal transects, buttrends can be examined. The most dramaticobservation is the low latitudinal tempera-ture gradient in the Eocene, caused by highvalues of both MAT and CMM at middle tohigh latitudes (Fig. 3). If, as we have argued,the North American foliar physiognomic es-timates of MAT and CMM are 2–3 8C toocool, then Eocene thermal gradients for theNorthern Hemisphere at middle latitudes(polynomial regression) would have beenabout 0.4 8C/18 latitude. There is little dif-ference between the latitudinal gradient ofMAT estimated from continental floras andthe gradient estimated from d18O data(Fig. 3; Zachos et al., 1994). These are sim-ilar to the ‘‘low-gradient’’ GCM simulationsof Eocene climate (Sloan and Barron, 1992).The gradient estimated for Eocene conti-nents is similar to the gradient seen alongthe coasts of Southern Hemisphere conti-nents today at low to middle latitudes, andless than half of current gradients in the in-terior of North America. Paleobotanical ev-idence for tropical temperatures in the Eo-cene suggests warm rather than cool tropics,with MAT and CMM at 98N paleolatitudebeing within 2–3 8C of present values (Gra-ham, 1994). Paleobotanical data confirm

1046 GEOLOGY, November 1995

that the Eocene world had a much moreeven distribution of heat across latitude thanat present.

Proxy Data-Model DiscrepanciesA key area of dispute between earlier

GCM simulations of Eocene climate and theterrestrial paleontological proxy record isthe latitudinal and geographic extent of sub-freezing temperatures in continental interi-ors (Sloan and Barron, 1990, 1992; Wing,1991; Wing and Greenwood, 1993). The lat-est GCM simulations for Eocene NorthAmerica (Sloan, 1994) in which a CO2 valuetwice that of present was used (Green RiverFormation Eocene Lake [LAKE] case) pre-dicted a zone of winter freezing (CMM,0 8C) covering most of the interior of thecontinent north of 408N, except in areasclose to the lake. The freezing zone was dis-placed to about 508N in the simulation usinga CO2 value six times that of present. Onlythe warmest of these scenarios begins to beconsistent with proxy data, because florasnorth of the Green River Lake to about508N contain palms, indicating CMM.5 8C. The regression line for CMM by lat-itude (Fig. 3) implies that in North AmericaCMM ,0 8C would have occurred onlynorth of about 608N and south of the ArcticOcean coast. In this region, CMM would

have been between 2 8C and 22 8C, andMAT would be 8–15 8C. This result dem-onstrates continuing disagreement betweenthe proxy record and model prediction, theextent and degree of freezing within NorthAmerica being much less severe (i.e., CMM2 to22 8C vs. 0 to210 8C) than predicted bythe model.All lines of paleontological proxy evi-

dence indicate that in the early to middleEocene North America and Australia hadequable climates lacking severe frost in bothinterior and coastal regions. Discrepanciesbetween the different lines of paleontologi-cal proxy evidence are small. MAT is esti-mated by all methods to be greater in theEocene at middle to high palaeolatitudesthan in comparable areas today (15–23 8C atmiddle latitudes, about 10 8C at sea level inhigh latitudes). CMMs are estimated tohave been .5 8C and perhaps .10 8Cthroughout all of Australia in the Eocene,and in most of North America.

CONCLUSIONSFor at least the first 10 m.y. of the Eocene,

a large part of Earth’s surface, includingcontinental interiors, had climates withmuch higher than modern winter tempera-tures. Latitudinal gradients of MAT andCMM in Eocene continental interiors were

similar to Eocene SST gradients, and topresent-day Southern Hemisphere mid-lat-itude values. Some recent simulations (Sloan,1994) have produced predictions close tothe estimates given here, but these ‘‘warm’’simulations have been achieved either bymodeling very high partial pressures of CO2(six times present), which are not supportedby proxy data or carbon cycle modeling(Cerling, 1991; Berner, 1991), or by includ-ing lakes, or both. There is geologic evi-dence for large lakes in the Green River ba-sin from the late-early through middleEocene, but the ‘‘lake effect’’ does not ex-plain warm proxy data points from far northof the Green River basin or the global extentof warm climate proxies at high latitudes(Sloan, 1994; Sloan et al., 1995).Given the widespread and long-lasting ev-

idence for equable climates in continentalinteriors and coastal regions alike, the dis-crepancy between Eocene models and datais likely to represent a major flaw in themodels (Walker and Sloan, 1992). This flawmay relate to simulated heat flow to andfrom continental surfaces, and/or to overlysimplified modeling of paleovegetationfeedbacks, as has been demonstrated instudies of the Holocene boreal forest (Foleyet al., 1994). The world of the models is stilltoo much like the present.

ACKNOWLEDGMENTSWe thank James Basinger, David Dilcher, Kirk

Johnson, Neville Pledge, Mike Pole, Lisa Sloan,Paul Willis, and Jack Wolfe for advice. CathyGreenwood and Jim McRea provided technicalassistance. This work was completed during aSmithsonian postdoctoral fellowship at the Na-tional Museum of Natural History (Greenwood),and is Evolution of Terrestrial Ecosystems con-tribution 28.

REFERENCES CITEDBasinger, J. F., Greenwood, D. R., and Sweda, T.,

1994, Early Tertiary vegetation of ArcticCanada and its relevance to paleoclimatic in-terpretation, in Boulter, M. C., ed., Arcticplants and climates: 65 million years ofchange: Berlin, Springer-Verlag.

Berner, R. A., 1991, A model for atmosphericCO2 over Phanerozoic time: American Jour-nal of Science, v. 291, p. 339–376.

Birkenmajer, K., and Zastawniak, E., 1986, Plantremains of the Dufayel Island Group (earlyTertiary?), King George Island, South Shet-land Islands (West Antarctica): Acta Palaeo-botanica, v. 26, p. 33–54.

Bryson, R. A., and Hare, F. K., 1974, Climates ofNorth America (World Survey of Climatol-ogy, Volume 11): Amsterdam, Elsevier, 420 p.

Budantsev, L. Y., 1992, Early stages of formationand dispersal of the temperate flora in theboreal region: Botanical Reviews, v. 58,p. 1–48.

Case, J. A., 1988, Paleogene floras from SeymourIsland, Antarctic Peninsula, in Feldmann,R. M., andWoodburne, M. O., eds., Geologyand paleontology of Seymour Island, Antarc-tic Peninsula: Geological Society of AmericaMemoir 169, p. 523–530.

Figure 3. Reconstructedearly-middle Eocene andmodern latitudinal ther-mal gradients (polynomialregression). A: Latitudinalgradients of mean annualtemperature (MAT). B:Latitudinal gradients ofcold month mean (CMM)temperature. Modern data(dot-pattern envelope anddashed regression line)derived from terrestrialstations <200 m elevationabove sea level in north-south transects followingwest and east coasts andthrough continental interi-ors of Australia (Gentilli,1971), Africa (Griffiths,1971 ) , Nor th Amer ica(Bryson and Hare, 1974),a n d S o u t h Ame r i c a(Schwerdtfeger, 1976);polar data are from Orvig(1970). Dot-pattern circles5 inferred early-middle Eocene sea-surface temperature (SST) values (Zachos et al., 1994).CLAMP 5 Climate Analysis Multivariate Program; NLR 5 nearest living relative; LMA 5 leafmargin analysis.

GEOLOGY, November 1995 1047

Cerling, T. E., 1991, Carbon dioxide in the atmo-sphere: Evidence from Cenozoic and Meso-zoic paleosols: American Journal of Science,v. 291, p. 377–400.

Collinson, M. E., 1983, Fossil plants of the Lon-don Clay: Palaeontological Association FieldGuides to Fossils, No. 1, 121 p.

Daghlian, C. P., 1978, Coryphoid palms from theLower and Middle Eocene of southeasternNorth America: Palaeontographica, Abt. B,v. 166, p. 44–82.

Ditchfield, P. W., Marshall, J. D., and Pirrie, D.,1994, High latitude palaeotemperature vari-ation: New data from the Tithonian to Eo-cene of James Ross Island, Antarctica:Palaeogeography, Palaeoclimatology, Palaeo-ecology, v. 107, p. 79–101.

Dolph, G. E., and Dilcher, D. L., 1979, Foliarphysiognomy as an aid in determining paleo-climate: Palaeontographica, Abt. B, v. 170,p. 151–172.

Foley, J. A., Kutzbach, J. E., Coe, M. T., and Le-vis, S., 1994, Feedbacks between climate andboreal forests during the Holocene epoch:Nature, v. 371, p. 52–54.

Fritz, W. J., and Harrison, S., 1985, Early Tertiaryvolcaniclastic deposits of the northern RockyMountains, in Flores, R. M., and Kaplan,S. S., eds., Cenozoic paleogeography of thewest-central United States: Denver, Colo-rado, Rocky Mountain Section, Society ofEconomic Paleontologists and Mineralo-gists, p. 383–402.

Gentilli, J., 1971, Climates of Australia and NewZealand (World Survey of Climatology, Vol-ume 13): Amsterdam, Elsevier, 405 p.

Graham, A., 1994, Neotropical Eocene coastalfloras and 18O/16O-estimated warmer vs.cooler equatorial waters: American Journalof Botany, v. 81, p. 301–306.

Greenwood, D. R., 1992, Taphonomic constraintson foliar physiognomic interpretations ofLate Cretaceous and Tertiary palaeocli-mates: Review of Palaeobotany and Palynol-ogy, v. 71, p. 149–190.

Greenwood, D. R., 1994, Palaeobotanical evi-dence for Tertiary climates, inHill, R. S., ed.,History of the Australian vegetation, Creta-ceous to Recent: Cambridge, United King-dom, Cambridge University Press, p. 44–59.

Gregory, K. M., and Chase, C. G., 1992, Tectonicsignificance of paleobotanically estimatedclimate and altitude of the late Eocene ero-sion surface, Colorado: Geology, v. 20,p. 581–585.

Griffiths, J. F., 1971, Climates of Africa (WorldSurvey of Climatology, Volume 10): Amster-dam, Elsevier, 604 p.

Guo, S.-X., 1990, A brief review on megafloralsuccessions and climatic changes of the Cre-taceous and early Tertiary in China, in Kno-bloch, E., and Kvacek, Z., eds., Paleofloristicand paleoclimatic changes in the Cretaceousand Tertiary: Prague, Geological Survey,p. 23–38.

Hickey, L. J., 1977, Stratigraphy and paleobotanyof the Golden Valley Formation (early Ter-tiary) of western North Dakota: GeologicalSociety of America Memoir 150, 181 p.

Hutchison, J. H., 1982, Turtle, crocodilian, andchampsosaur diversity changes in the Ceno-zoic of the north-central region of westernUnited States: Palaeogeography, Palaeocli-matology, Palaeoecology, v. 37, p. 149–164.

Kershaw, A. P., and Nix, H. A., 1988, Quantitativepalaeoclimatic estimates from pollen datausing bioclimatic profiles of extant taxa:Journal of Biogeography, v. 15, p. 589–602.

Little, E. L., 1977, Atlas of United States trees,Volume 4. Minor eastern hardwoods: U.S.Department of Agriculture Forest ServiceMiscellaneous Publication 1342, 230 p.

MacGinitie, H. D., 1974, An early middle Eoceneflora from theYellowstone-Absaroka volcan-ic province, northwestern Wind River Basin,Wyoming: University of California Publica-tions in Geological Sciences, v. 108, 103 p.

Markwick, P., 1994, ‘‘Equability,’’ continentality,and Tertiary ‘‘climate’’: The crocodilian per-spective: Geology, v. 22, p. 613–616.

Miller, K. G., Fairbanks, R. G., and Mountain,G. S., 1987, Tertiary oxygen isotope synthe-sis, sea level history, and continental marginerosion: Paleoceanography, v. 2, p. 1–19.

Moore, M. E., 1973, Major groups of palms andtheir distribution: Gentes Herbarium, v. 11,p. 27–141.

Orvig, S., 1970, Climates of polar regions (WorldSurvey of Climatology, Volume 14): Amster-dam, Elsevier, 370 p.

Pocknall, D. T., 1990, Palynological evidence forthe early to middle Eocene vegetation andclimate history of New Zealand: Review ofPalaeobotany and Palynology, v. 65, p. 57–69.

Pole, M., 1994, An Eocene macroflora from theTaratu Formation at Livingstone, NorthOtago, New Zealand: Australian Journal ofBotany, v. 42, p. 341–367.

Quilty, P. G., 1994, The background: 144 millionyears of Australian palaeoclimate and palaeo-geography, in Hill, R. S., ed., History of theAustralian vegetation, Cretaceous to recent:Cambridge, United Kingdom, CambridgeUniversity Press, p. 14–43.

Read, J., and Hill, R. S., 1989, The response ofsome Australian temperate rain forest treespecies to freezing temperatures and its bio-geographical significance: Journal of Bioge-ography, v. 16, p. 21–27.

Romero, E. J., 1993, South American paleofloras,in Goldblatt, P., ed., Biological relationshipsbetween Africa and South America: NewHaven, Connecticut, Yale University Press,p. 62–85.

Sakai, A., and Larcher, W., 1987, Frost survival ofplants: Responses and adaptation to freezingstress: Berlin, Springer-Verlag, p. 1–321.

Schaal, S., and Ziegler, W., eds., 1992, Messel. Aninsight into the history of life and of theEarth: Oxford, United Kingdom, ClarendonPress, 322 p.

Schwerdtfeger, F., 1976, Climates of South Amer-ica (World Survey of Climatology, Volume12): Amsterdam, Elsevier, 367 p.

Sloan, L. C., 1994, Equable climates during theearly Eocene: Significance of regional paleo-geography for North American climate: Ge-ology, v. 22, p. 881–884.

Sloan, L. C., and Barron, E. J., 1990, ‘‘Equable’’climates during Earth history?: Geology,v. 18, p. 489–492.

Sloan, L. C., and Barron, E. J., 1992, A compar-ison of Eocene climate model results toquantified paleoclimatic interpretations:Palaeogeography, Palaeoclimatology, Palaeo-ecology, v. 93, p. 183–202.

Sloan, L. C., Walker, J. C. G., and Moore, T. C.,Jr., 1995, Possible role of oceanic heat trans-port in early Eocene climate: Paleoceanog-raphy, v. 10, p. 347–356.

Stilwell, J. D., Levy, R. H., and Harwood, D. M.,1993, Preliminary paleontological investiga-tion of Tertiary glacial erratics from the Mc-Murdo Sound region, East Antarctica: Ant-arctic Journal of the United States, v. 28,p. 16–19.

Stott, L. D., Kennett, J. P., Shackleton, N. J., andCorfield, R. M., 1990, The evolution of Ant-arctic surface waters during the Paleogene:Inferences from the stable isotopic composi-tion of planktonic foraminifers, ODP Leg113, in Barker, P. F., Kennett, J. P., et al.,Proceedings of the Ocean Drilling Program,Scientific results, Volume 113: College Sta-tion, Texas, Ocean Drilling Program,p. 849–863.

Tanai, T., 1992, Tertiary vegetational history ofEast Asia: Mizunami Fossil Museum Bulle-tin, v. 19, p. 125–163.

Truswell, E. M., 1993, Vegetation changes in theAustralian Tertiary in response to climaticand phytogeographic forcing factors: Austra-lian Systematic Botany, v. 6, p. 533–558.

Walker, J. C. G., and Sloan, L. C., 1992, Some-thing is wrong with climate theory: Geotimes,v. 37, no. 6, p. 16–18.

Wardle, P., 1991, Vegetation of New Zealand:New York, Cambridge University Press,672 p.

Westgate, J. W., and Gee, C. T., 1990, Paleoecol-ogy of a middle Eocene mangrove biota (ver-tebrates, plants, and invertebrates) fromsouthwest Texas: Palaeogeography, Palaeo-climatology, Palaeoecology, v. 78, p. 163–177.

Willis, P. M. A., and Molnar, R. E., 1991, A newmiddle Tertiary crocodile from Lake Palan-karinna, South Australia: South AustralianMuseum Records, v. 2, p. 39–55.

Wing, S. L., 1991, ‘‘Equable’’ climates duringEarth history?: Comment: Geology, v. 19,p. 539–540.

Wing, S. L., and Greenwood, D. R., 1993, Fossilsand fossil climates: The case for equable con-tinental interiors in the Eocene: Royal Soci-ety of London Philosophical Transactions,ser. B, v. 341, p. 243–252.

Wolfe, J. A., 1979, Temperature parameters ofhumid to mesic forests of eastern Asia andrelation to forests of other regions of theNorthern Hemisphere and Australasia: U.S.Geological Survey Professional Paper 1106,37 p.

Wolfe, J. A., 1993, Amethod of obtaining climaticparameters from leaf assemblages: U.S. Ge-ological Survey Bulletin 2040, p. 1–71.

Yemane, K., 1993, Contribution of Late Permianpalaeogeography in maintaining a temperateclimate in Gondwana: Nature, v. 361,p. 51–54.

Zachos, J. C., Stott, L. D., and Lohmann, K. C.,1994, Evolution of early Cenozoic marinetemperatures: Paleoceanography, v. 9,p. 353–387.

Manuscript received April 6, 1995Revised manuscript received July 19, 1995Manuscript accepted July 28, 1995

1048 GEOLOGY, November 1995Printed in U.S.A.