Response of tropical vegetation to Paleogene warming...Carlos A. Jaramillo Abstract.—The late Paleocene-early Eocene transition was characterized by a long period of global warming

q 2002 The Paleontological Society. All rights reserved. 0094-8373/02/2802-0004/$1.00

Paleobiology, 28(2), 2002, pp. 222–243

Response of tropical vegetation to Paleogene warming

Carlos A. Jaramillo

Abstract.—The late Paleocene-early Eocene transition was characterized by a long period of globalwarming that culminated with the highest temperatures of the Cenozoic. This interval is associatedwith a significant increase in plant diversity in temperate latitudes. However, data from tropicalregions remain largely unknown. The record of pollen and spore diversity across the late Paleoceneto the early middle Eocene of eight sections in central and eastern Colombia was analyzed. Severaltechniques, including range-through method, rarefaction, bootstrap, detrended correspondenceanalysis, and Shannon index, were used to assess the significance of the observed diversity pattern.The palynofloral record indicates that the lower to middle Eocene contains a significantly higherpalynofloral diversity than the underlying upper Paleocene strata. This pattern is maintained afteraccounting for sample size, number of samples/time unit, lithofacies, and depositional systems.Eocene palynofloras have higher alpha and beta diversities and a higher equitability than Paleocenepalynofloras. This increase in diversity is the product of a gradual increase in the rate of first ap-pearances and a gradual decrease in the rate of last appearances. The early to middle Eocene in-crease in diversity, as well as the increase in spore abundance and diversity, suggests that tropical(equatorial) climate became wetter during the early to middle Eocene. This interpretation favorscauses for early Eocene warming that do not involve significant increases in greenhouse gases.Samples from strata associated with the Paleocene/Eocene thermal maximum were barren for pa-lynomorphs, and the effects of this climatic event on tropical vegetation remains uncertain.

Carlos A. Jaramillo. Department of Paleobiology, MRC 121, Smithsonian Institution, Washington D.C.20560. E-mail: [email protected]

Present address: Instituto Colombiano del Petroleo, AA 4185, Bucaramanga, Colombia

Accepted: 3 July 2001

Introduction

The late Paleocene–early Eocene interval ischaracterized by a long period of global warm-ing that culminated in the Eocene thermalmaximum, the highest ocean temperature ofthe last 65 million years. The warming has twodistinct peaks, at the Paleocene/Eocene bound-ary and in the middle to late early Eocene.

The climatic event at the Paleocene/Eoceneboundary (PETM, formally known as ‘‘late Pa-leocene thermal maximum’’ or LPTM, of Za-chos et al. 1993) is recognized by a rapid(;125–200 Kyr) worldwide warming and achange in ocean circulation (Miller et al. 1987;Kennett and Stott 1991; Pak and Miller 1992;Zachos et al. 1993). Antarctic surface anddeep-water temperatures increased from;148C to ;208C and from ;108C to ;188C re-spectively (Kennett and Stott 1991), and sub-tropical sea-surface waters warmed by at least1–48C (Thomas et al. 1999). The PETM is as-sociated with a global negative anomaly(22‰ to 23‰) of the carbon isotope recordthat is documented in both terrestrial and ma-

rine environments and records a massive re-lease of methane from sedimentary reservoirs(Norris and Rohl 1999). The PETM correlateswith a low to moderate decrease in plant di-versity in midlatitudes of North America(Wing et al. 1995; Wing 1998), and a dramaticmammalian turnover (Clyde and Gingerich1998) in North America and Europe. Tropicalplanktonic foraminifera diversified duringthis time (Kelly et al. 1998), while deep-seabenthic foraminifera suffered a major extinc-tion (Pak and Miller 1992).

The climate of the late early Eocene (or Eo-cene thermal maximum, ETM) is recognizedas having the highest mean ocean temperaturein the Tertiary (Wolfe 1978; Wolfe and Poore1982; Miller et al. 1987; Prentice and Matthews1988), greatly affecting the vegetation in bothsouthern and northern middle to high lati-tudes where a large increase in plant diversityhas been documented (Askin and Spicer 1995;Christophel 1995; Wing et al. 1995; Wing1998).

Competing models exist for explaining theETM. Increases in both greenhouse gases and

223PALEOGENE TROPICAL FORESTS

oceanic heat transport have been proposed toexplain the nature of this warm climate (Rindand Chandler 1991; Pak and Miller 1992; Sloanand Barron 1992; Sloan et al. 1995). When lev-els of atmospheric CO2 similar to preindustri-al values are used in these models, a slightcooling in terrestrial tropical environments,especially for South America, is produced(Sloan and Rea 1995; Sloan and Morrill 1998).Conversely, when CO2 levels six times higherthan preindustrial values are used (green-house model), land temperature in the Tropicsrises by 48C and soil moisture decreases(Sloan et al. 1995). Proxy data from tropical re-gions for the ETM are still scarce, but most donot show a significant increase in sea-surfacetemperature (SST), in spite of a temperatureestimation bias toward values slightly coolerthan the actual SST (Crowley and Zachos2000). Oxygen isotope values of planktonicand benthic foraminifera for the late Paleoceneto early middle Eocene indicate that tropicalsea-surface temperature increased just slight-ly (;18C), with values similar to those in theHolocene (Zachos et al. 1994). However, thedata points analyzed from tropical regions arevery few (one site for the late Paleocene, twosites for the early Eocene, and no sites for theearly middle Eocene). Bralower et al. (1995)reports temperatures slightly cooler thanthose of today during the early Eocene, fol-lowed by 3–68C cooling in the late early tomiddle Eocene.

Data from terrestrial tropical regions areunavailable to test these models and furtherimprove our understanding of these examplesof global warming. How did tropical vegeta-tion respond to climate change? Is there anycoordination between floral and climaticchange? Initial data suggest that there was asignificant change in floral composition anddiversity across the Paleocene–Eocene transi-tion in the Tropics (Jaramillo and Dilcher2000).

The effects of Quaternary climatic changesupon tropical vegetation of South Americahave been the subject of an intense debate overthe last decade (Hooghiemstra and Van derHammen 1998). The proponents of the refugiahypothesis indicate that savannas expandedand rain forest was reduced to fragments dur-

ing dry and cool glacial intervals (Hooghiem-stra and Van der Hammen 1998). On the otherhand, others have suggested stability and con-tinuity of the rain forest cover during glacia-tion periods (Haberle and Maslin 1999; Col-invaux and Oliveira 2001) with a migration ofcool-tolerant plant species toward forestedlowland areas, and centers of endemism beingareas of maximal disturbance, rather thanmaximal stability, as the refugia hypothesisstates (Bush 1994; Colinvaux et al. 2000). Itseems that the rain forest has expanded southsince the last glacial maximum (Mayle et al.2000). Other data, however, show that savannaexpanded north in southern Amazonia duringthe early and middle Holocene (Freitas et al.2001). Vegetation models indicate that changesin vegetation structure were more importantthan biome changes during the last glacialmaximum (Cowling et al. 2001). Precipitationdata are also highly controversial, with datashowing a drier (Maslin and Burns 2000), orwetter (Baker et al. 2001a,b), last glacial max-imum and Younger Dryas. Data density forthe fossil record of Neotropical forests is stilltoo low to understand the effect that the gla-cial-interglacial cycles had upon forest diver-sity, distribution, and structure.

This study analyzes the pollen and spore re-cord of eight stratigraphic sections encom-passing the Paleocene–Eocene transition incentral and eastern Colombia. An analysis ofpalynofloral diversity, composition, and struc-ture is presented and its relation to climatechange is discussed.

Methods

Eight stratigraphic sections encompassingthe late Paleocene–middle Eocene transitionwere studied in central and eastern Colombia(Fig. 1). Environments of deposition rangefrom fluvial plain to lower coastal plain-es-tuarine deposits (Jaramillo and Dilcher 2000,2001).

A biostratigraphic framework slightly mod-ified from that of Jaramillo and Dilcher (2001)was used to date each section. This frameworkwas based on the graphic correlation of paly-nomorphs. The equation of the line of corre-lation for each of the sections was used to ex-trapolate the stratigraphic position of samples

224 CARLOS A. JARAMILLO

FIGURE 1. Location of the outcrop sections (Pinalerita,Uribe, Regadera) and cores (C4, P11, Veta-1, Tibu-1, ICPE-2) analyzed in this study. Pinalerita (48549N,73819W), Re-gadera (78429N–728379W), and Uribe (78209N,738209W) out-crop sections, and Tibu-1 (8837933.780N,72841921.820W),Veta-1 (8852941.400N,72850924.630W), ICPE-2 (785948.380N,73823959.330W), C4 (582934.890N,72842911.100W), and P11(983943.4150N,7285892.4540W) cores.

into the Pinalerita section, chosen as the ref-erence section (Fig. 2). Samples with sums ofpollen/spores below 40 grains were excludedfrom the analyses. The Paleocene/Eoceneboundary is not clearly defined and lies in aninterval between meters 332.6 and 451.1 of thereference section. The early/middle Eoceneboundary is tentatively located at meter 670 ofthe reference section (Fig. 2).

Various techniques were used to analyze thepatterns of pollen and spore diversity (theword diversity is used here to denote numberof species [Rosenzweig 1995]): (1) The range-through method (Boltovskoy 1988) was usedto estimate standing diversity. This methodassumes that a taxon is present in all samplesthat lie between its first- and last-appearancedatum. It minimizes the effect that facies-re-lated fossils and differences in capture prob-ability have on standing diversity. The stand-ing diversity was calculated by excludingunique taxa (those that are present in only onesample), because they can introduce noise tothe diversity pattern (Wing 1998). (2) Sand-ers’s rarefaction, an interpolation technique,was used to estimate how many species mighthave been found within a sample if the samplehad been smaller (Raup 1975). Thus, diversity

of small and large samples can be comparedon equal terms. (3) Bootstrapping is a tech-nique that facilitates comparison of intervalswith different sample density (Gilinsky 1991).A single sample is selected at random, thenumber of species is counted on the basis ofthat sample, a second sample is selected andthe number of species is recalculated using thepooled data from both samples, a third is se-lected, and the process continues until allsamples are included (Colwell and Codding-ton 1994). The whole process was repeated 50times, and mean and standard deviation werecalculated for each sampling level. Both em-pirical diversity data and an estimator of spe-cies diversity (Chao2) were used. SChao2 5Sobs 1 {[Q1

2/2(Q2 1 1)] 2 [Q1Q2/2(Q2 1 1)2]};SChao2 5 estimated number of species, Sobs 5total number of species observed in all sam-ples pooled, Q1 5 uniques, the number of spe-cies that occur in exactly one sample, Q2 5doubles, the number of species that occur inexactly two samples (Colwell 2000). This es-timator is very responsive to low sample num-bers and highly diverse data sets (Colwell andCoddington 1994). (4) The Shannon index (H5 2Spi log10 pi, pi 5 proportion of individualsthat belong to species i) is a measure of un-certainty of a selection process (Hayek andBuzas 1997; Zar 1999). A maximum value of Hoccurs when species are equally abundant ina sample, therefore the uncertainty of know-ing which species will be observed next wouldbe the highest (Hayek and Buzas 1997). H is afunction of the number of species and theabundance distribution of individuals withinthose species (Hayek and Buzas 1997). (5)Evenness (J 5 H/Hmax) is a measure of theabundance distribution of a population (Hay-ek and Buzas 1997). The highest possibleevenness (J 5 1) occurs when all species havethe same number of individuals. A low valueindicates that most of the assemblage is dom-inated by only a few species. (6) The overallvariation in floral composition throughout thesection was observed by using detrended cor-respondence analysis, or DCA (Hill andGauch 1980). DCA summarizes variation inthe composition of the assemblages in a smallnumber of dimensions (Wing 1998), and it isan appropriate ordination for nonlinear data


FIGURE 2. Lines of correlation for each of the sections compared with the reference section (Pinalerita). The Pin-alerita has also been added to the graphic to allow the visual comparison of the accumulation rates of each sectionversus the reference section. Note that the Pinalerita line is not a correlation line; it is only a plot of the sectionagainst itself. m.E. 5 middle Eocene; l. Eocene 5 lower Eocene.

(Pielou 1984). The ordination was performedon the raw abundance data. Rare taxa weredown-weighted using the algorithm of Orloci(Kovach 1998). (7) The Jaccard’s Coefficientand the modified Morisita-Horn similarity in-dex were used to assess the similarity amongsamples (Magurran 1988). Jaccard’s index isbased on presence-absence data, whereasMorisita-Horn uses abundance data. Morisita-Horn is sensitive to very abundant species;therefore, abundance data were transformedbefore the analysis by taking the square rootof each value. (8) Rates of first and last ap-pearances were calculated using a four-stepprocedure. First, stratigraphic meters of thecomposite section were converted to time.Linear accumulation rates were assumed be-tween the chronostratigraphic boundariesused for this study. Dating of the boundariesfollows that of Berggren et al. (1995) and Nor-ris and Rohl (1999). Second, first- and last-ap-pearance datums (FADs and LADs respec-tively) of each species were sorted from oldestto youngest. Only FADs and LADs derivedfrom this study were used, and uniques wereexcluded from analysis. Third, it was neces-

sary to estimate the edge effect (Foote 2000) inthe number of FADs and LADs throughout thereference section. The edge effect is an exag-gerated high concentration of FADs at the be-ginning of a section and high concentration ofLADs at the end of a section (Boltovskoy1988). A piecewise regression was performedto estimate the edge effect. This type of re-gression assumes that there are two differentregression functions to the same data (SPSS1999). The regression attempts a two-segmentfit of the data. The breakpoint is the intersec-tion of the two fitted regression lines. The re-gression iteratively tries all possible positionsof the breakpoint and chooses the one thatproduces the lowest residual sum of squares(Yeager and Ultsch 1989). The model to fit fol-lows the algorithm described by Dugglebyand Ward (1991) for a two-segment linear re-gression.

y 5 y 1 [(m 1 m )(x 2 x )T L R T

2 (m 2 m ) zx 2 x z]/2,L R T

where y 5 FAD or LAD, x 5 species, xT 5breakpoint species, yT 5 breakpoint FAD or


FIGURE 3. Environments of accumulation of each sam-ple analyzed. Samples from different sections were pro-jected into the reference section using the equations ofcorrelation shown in Appendix. Labels to the left of eachcircle indicate the section from which the sample wastaken. BL 5 fluvial plain, bed-load dominated channels;ML 5 fluvial to coastal plain, mixed-load dominatedchannels; SLu 5 upper coastal plain, suspension-loaddominated channels; SLl 5 lower coastal plain; E 5 es-tuarine; PIN 5 Pinalerita; UR 5 Uribe; V 5 Veta-1; T 5Tibu-1; RE 5 Regadera; ICPE 5 ICPE-2; P11 5 core P11;C4 5 core C4. Units of composite section are given instratigraphic meters of the reference section (Pinalerita).

LAD, mL 5 slope left of breakpoint, and mR 5slope right of breakpoint. Fourth, regressionfits were performed on the FAD and LAD dataexcluding those produced by the edge effect.The best fit regression model was selected us-ing the sum of squares and the minimum val-ue of the following equation:

best model 5 SSQ(m)/n 2 2m,

where SSQ is the residual sum of squares, mis the number of parameters of the model andn the number of data points (Hillborn andMangel 1997).

Parametric statistics (ANOVA and t-test,equal versus unequal variance for t-test wasdetermined using F-test) were used to test hy-potheses on the distribution of H, J, S (numberof species), N (number of grains counted persample), Jaccard, and Morisita-Horn similari-ty values. If H is calculated for a number ofsamples, the indices themselves will be nor-mally distributed (Taylor 1978). This propertymakes it possible to use parametric statistics(Magurran 1988). Nonparametric statisticswere used to test hypotheses about DCAscores. Analyses were performed in MVSP 3.0(Kovach 1998), EstimateS version 6 (Colwell2000), BioDiversity Professional Beta version(McAleece et al. 1997), Systat 9, and Statview5.0.

Results

Seventy-seven samples containing 735 spe-cies were used for this study (Appendix in Pa-leobiology data repository contains rangecharts for each section and the composite sec-tion). Samples from seven different sectionsand diverse depositional environments (Fig.3) were extrapolated into Pinalerita, the cho-sen reference section, using the equations ofcorrelation (Fig. 2; equations of correlation areincluded in Appendix). There is an overalltrend of regression of the coast from the upperPaleocene to the lower Eocene and then trans-gression from the upper lower Eocene to thelower middle Eocene (Fig. 3). A pollen recov-ery gap of approximately 118.5 m is presentacross the Paleocene/Eocene boundary (Fig.3). In this gap, 45 samples were analyzed withvery poor recovery of pollen, spores, and dis-persed organic matter. All samples from this


FIGURE 4. Standing diversity of range-through pollenand spores. Taxa with occurrences in only one samplewere excluded. Mean diversity per sample of Paleocenestrata is significantly lower than in Eocene strata (t-test:p , 0.0001). Units of composite section are given instratigraphic meters of the reference section (Pinalerita).

interval in all of the sections have similar fa-cies—light olive-gray to dark gray mudstones,massive, and intensely mottled (root biotur-bation)—suggesting incipient development ofpaleosols.

The standing diversity results, incorporat-ing range-through taxa (Tables 1, 2, Fig. 4), in-dicate that mean diversity per Paleocene sam-ple is 68.7 pollen/spore species whereas Eo-cene diversity is 103.4 (Table 1). The numberof grains counted per sample directly affectsthe diversity of a given sample. To take intoaccount this factor, rarefaction analyses wereconducted at three different cutoff countingpoints or ‘‘knots’’ (Table 2, Fig. 5). All knots

analyzed (101, 231, 291) indicate that mean di-versity per sample is significantly higher inthe Eocene than in the Paleocene (Table 1).Samples in which the count did not reach thedefined cutoff point were not included in thecalculation of the mean.

The type of lithology can control the abun-dance of pollen and spores recovered in asample; i.e. oxidized lithologies usually pre-serve fewer pollen grains/spores. Therefore,the recovery of a sample can affect its diver-sity. A t-test was performed to test the prob-ability that differences in Paleocene versus Eo-cene diversity were due to differences in sam-ple recovery (Ho: pollen/spore abundance persample in Paleocene 5 pollen/spore abun-dance per sample in Eocene; t-test: p , 0.91).The test indicates that samples from both Pa-leocene and Eocene strata did not yield sig-nificantly different recoveries of pollen andspores.

Depositional environments also can affectsample diversity (Holland 1995). The Shannonindex (H) and rarefied number of species/sample (S) were used to test for this factor. Atwo-way ANOVA was performed to evaluatethe influence of the environments of deposi-tion and age on Shannon index (lithology andenvironments of deposition for each sampleare given in Table 2). Samples from the BL en-vironment (fluvial plain, bed-dominatedchannels) were not used in the analysis be-cause of the low sample number (n 5 1 in Pa-leocene, 3 in Eocene). The results of the anal-ysis indicate that age had a significant effecton H (Shannon index) (p , 0.0011), but thatneither environment nor the interaction be-tween age and environment had a significanteffect (p , 0.39 and p , 0.58, respectively). Aone-way ANOVA was performed to evaluatethe differences in H among depositional en-vironments within each epoch. For Paleocenestrata, differences in H among depositionalenvironments are not significant (ML: 0.72,SD 5 0.23; SLu: 0.98, SD 5 0.17; SLi: 0.87, SD5 0.29; and E: 1.14, SD 5 0.11; p , 0.18, en-vironment labels are given in Figure 3). ForEocene strata, differences are also not signifi-cant (ML: 1.10, SD 5 0.35; SLu: 1.18, SD 50.38; SLi 1.25, SD 5 0.23; and E 1.19, SD 50.15; p , 0.55). Finally, the mean diversity per


TABLE 1. Summary of results and significance tests. SD 5 Standard deviation; ML 5 fluvial to coastal plain, mixed-load dominated channels; SLu 5 upper coastal plain, suspension-load dominated channels; SL1 5 lower coastalplain; P 5 Paleocene; E 5 Eocene.

Analysis

Paleocene

Mean (SD)

Eocene

Mean (SD) p ,

Standing diversityRarefied diversity, 101 knotRarefied diversity, 231 knotRarefied diversity, 291 knotML rarefied diversity, 101 knotSLu rarefied diversity, 101 knotSL1 rarefied diversity, 101 knotShannon indexEvennessRarefied diversity, 101 knot, Pinalerita areaRarefied diversity, 101 knot, Catatumbo areaRarefied diversity, 101 knot, Uribe areaDCA first axisDCA first axis for SLu and SL1 environmentsJaccard’s coefficientJaccard’s coefficient for SLu environmentMorisita-Horn similarity index [including P vs.

E: 0.25 (SD: 0.13)]Morisita-Horn similarity index for SLuBombacacidites standing diversityPolypodiisporites standing diversityTetracolporopollenites standing diversityProxapertites relative abundance (in %)Fern spores standing diversityFern spores relative abundance (in %)Fern spores relative diversity (in %)

68.718.524.726.710.118.718.30.920.66

18.917.520.81.241.080.200.210.45

0.484.11.71.8

31.811.924.417.2

(13.1)(6.6)(8.8)

(10.2)(0.7)(5.3)(6.9)(0.25)(0.14)(4.5)(6.8)(9.1)(0.56)(0.56)(0.08)(0.03)(0.19)

(0.16)(1.0)(1.3)(0.2)

(24.2)(3.2)

(19.3)(2.4)

103.426.139.641.324.128.329.61.140.73

28.726.418.53.033.140.170.200.41

0.387.23.12.20.8

17.441.918.1

(23.6)(9.7)

(14.7)(16.1)(10.1)(10.6)(11.9)

(0.3)(0.15)

(10.2)(8.8)(6.8)(0.53)(0.52)(0.09)(0.1)(0.17)

(0.24)(1.9)(0.7)(0.9)(1.1)(2.2)

(20.1)(6.8)

0.0001, t-test0.001, t-test0.0004, t-test0.0053, t-test0.0009, t-test0.03, t-test0.01, t-test0.001, t-test0.026, t-test0.004, t-test0.007, t-test0.65, t-test0.001, Mann-Whitney0.0001, Mann-Whitney0.00005, t-test0.59, t-test0.0001, one-way ANOVA

0.084, t-test0.001, t-test0.001, t-test0.04, t-test0.001, t-test0.0001, t-test0.0003, t-test0.81, t-test

epoch using the values of rarefaction at 101knot (that is, the number of species of a samplein which 101 grains were counted) was com-pared for each depositional environment. Pa-leocene environments ML (fluvial plain,mixed-dominated channels), SLu (uppercoastal plain), and SLl (lower coastal plain)are significantly less diverse than their Eocenecounterparts (Table 1).

Sample density can also affect the observeddiversity of a given interval (Holland 1995). Ahigher sample density increases the probabil-ity of finding new species, and therefore over-all diversity also increases. To account for thisfactor, empirical and estimated (Chao2) spe-cies-accumulation curves were calculated bybootstrapping (Fig. 6). The empirical species-accumulation curve for all 35 Paleocene sam-ples pooled together indicates 308 species,and for Eocene strata, 422.6 species (SD 514.5) at 35 pooled samples. The Chao2 esti-mator of species diversity is lower for the Pa-leocene than for the Eocene at any givenpooled sample level (paired t-test: p , 0.004).

Shannon index (H) takes into account bothnumber of species per sample and abundancedistribution (Table 2, Fig. 7). Mean H for thePaleocene is significantly lower than for theEocene (Table 1). The mean Evenness (J) forthe Paleocene is also significantly lower thanfor the Eocene (Table 1, Fig. 7).

Sections also were compared geographical-ly. They were grouped according to theirproximity (Fig. 1). The Pinalerita section andC4 core were grouped in one area (Pinalerita);the Uribe section and ICPE-2 core in a secondarea (Uribe); and the Regadera section alongwith Veta-1, Tibu-1, and P11 cores in a thirdarea (Catatumbo). Rarefied number of speciesat 101 knot was used to calculate mean diver-sity per epoch in each area. Mean diversity persample in Pinalerita and Catatumbo is signif-icantly higher in Eocene than in Paleocenestrata (Table 1). The Uribe area does not showany significant difference (Table 1). Bootstrapanalyses of these three areas were performedusing the Shannon index and an empiricalspecies-accumulation curve (Fig. 8). Both


graphics show that in Catatumbo and Pinal-erita, Shannon index and pooled sample di-versity of the Eocene are higher than those ofthe Paleocene, whereas in Uribe they do notsignificantly change.

The overall floral composition was analyzedby using detrended correspondence analysisor DCA (Table 2, Fig. 9). The first axis explains15.84% of variation. Mean first-axis score forthe Paleocene is significantly different than forthe Eocene (Table 1). To take into account theunbalanced sampling across Paleocene versusEocene depositional environments, DCAscores on the first axis were calculated only forSLu and SLl environments. The difference inmean DCA score of the first axis for Paleoceneversus Eocene strata is still significant (Table1).

The Jaccard and Morisita-Horn similarityindices between all pairs of samples also werecalculated. Both indices indicate a significant-ly greater similarity among Paleocene samplesthan among Eocene samples, and a significantdifference between Paleocene and Eocenesamples (Table 1). Because there is an unbal-anced sampling distribution of depositionalenvironments in Paleocene versus Eocenestrata, although without missing cells (BL: 1sample in Paleocene, 3 in Eocene; ML: 3 sam-ples in Paleocene, 17 in Eocene; SLl: 17 sam-ples in Paleocene, 8 samples in Eocene; SLu: 12samples in Paleocene, 7 samples in Eocene; E:2 samples in Paleocene, 7 samples in Eocene;see Figure 3 for explanation of abbreviations),the analysis was repeated using only samplesfrom the SLu environment. This environmenthad the most equal sampling distribution. Jac-card and Morisita-Horn indices for the SLuenvironment are still higher in the Paleocene,but differences are not significant (Table 1).

Species were separated into three catego-ries: those with stratigraphic ranges restrictedto the Paleocene, those restricted to the Eo-cene, and those with occurrences in both Pa-leocene and Eocene strata (Fig. 10). Uniquespecies were excluded from this analysis. Atotal of 63 species are present in both Paleo-cene and Eocene strata, 56 species are only inPaleocene strata, and 140 species are only inEocene strata. The mean standing diversityper sample for Paleocene-restricted species is

27.8 (SD 5 7.3), for Eocene-restricted speciesis 63.2 (SD 5 25.1), and for taxa present inboth epochs is 40.4 (SD 5 13.5).

Patterns of first-appearance datums andlast-appearance datums were examined (Ta-ble 2, Fig. 11). A large degree of uncertaintystill exists in the stratigraphic position of thePaleocene/Eocene boundary and early/mid-dle Eocene boundary in the region (Jaramilloand Dilcher 2000, 2001). Tentatively, the early/late Paleocene boundary was located at 0 mand dated 60.9 Ma (Berggren et al. 1995), thePaleocene/Eocene boundary was located at430 m and dated 54.93 Ma (Norris and Rohl1999), and the early/middle Eocene boundarywas located at 670 m and dated 49 Ma (Berg-gren et al. 1995). Using this information, I cal-culated rates of rock accumulation and alsothe age of each sample, assuming constant rateof accumulation between the dating points.

The piecewise regression indicates that theedge effect for FADs ends at species 79, andthe edge effect for LADs ends at species 83(Fig. 11A). The sterile gap in pollen and sporerecovery near the Paleocene/Eocene bound-ary presents a problem and first must be con-sidered. This gap produces an edge effect thatincreases the number of LADS before the gapand increases number of FADS after the gap.The slope of the linear regression on PaleoceneFAD data (93.374 Kyr/species) calculated bythe piecewise regression analysis to the rightof the breakpoint was used to estimate thenumber of FADS that should have occurredduring the gap (a total of 20 FADS). Thus, thefirst 20 FADS after the gap are attributed tothe edge effect. To estimate the number ofLADS due to the gap, a regression was fittedon LADS only from Paleocene LADS (exclud-ing edge-effect datums). The best fit was poly-nomial (r2 5 0.983; y 5 23.093 1 77.398 x 1*1.14 x2). Using this equation, an estimated 19*LADS should have occurred during the gap.Therefore, the last 19 LADS before the begin-ning of the gap are attributed to the edge ef-fect.

Regression analyses were performed onFADs and LADs, and data attributed to edgeeffects (a section’s upper and lower limits, andthe internal gap) were discarded, as discussedabove. The best-fit model for both FAD and


TA

BL

E2.

Div

ersi

tyd

ata.

M5

sam

ple

(in

stra

tig

rap

hic

met

ers

from

Pin

aler

ita

sect

ion

);O

5so

urc

ese

ctio

n(U

R5

Uri

be,

PIN

5P

inal

erit

a,R

E5

Reg

ader

a,T

5T

ibu

-1,

C4

5co

reC

4,V

5V

eta-

1,IC

PE

5IC

PE

-2;

P11

5co

reP

11);

M1

5st

rati

gra

ph

icp

osit

ion

ofsa

mp

lein

sou

rce

sect

ion

;L

5li

thol

ogy

(sh

5sh

ale,

ct5

clay

ston

e,m

d5

mu

dst

one,

st5

silt

ston

e,fs

d5

fin

esa

nd

ston

e,v

fsd

5ve

ryfi

ne

san

dst

one)

;E

5en

vir

onm

ent

ofac

cum

ula

tion

(BL

5fl

uv

ial

pla

in,

bed

-loa

dd

omin

ated

chan

nel

s;SL

u5

up

per

coas

tal

pla

in,

susp

ensi

on-l

oad

dom

inat

edch

ann

els;

ML

5fl

uv

ial

toco

asta

lp

lain

,m

ixed

-loa

dd

omin

ated

chan

nel

s;SL

15

low

erco

asta

lp

lain

;E

5es

-tu

arin

e);N

5g

rain

sco

un

t;S

5sp

ecie

sco

un

t;U

5n

um

ber

ofu

niq

ues

(sp

ecie

sp

rese

nt

only

ina

sin

gle

sam

ple

);R

101

5ra

refa

ctio

n10

1g

rain

scu

toff

;R23

15

rare

fact

ion

231

gra

ins

cuto

ff;R

291

5ra

refa

ctio

n29

1g

rain

scu

toff

;H5

Shan

non

ind

ex;J

5E

ven

nes

s;D

C5

firs

tax

isD

CA

;St

5st

and

ing

div

ersi

tyaf

ter

ran

ge-

thro

ugh

(exc

lud

ing

sin

gle

s);F

5n

um

ber

offi

rst

app

eara

nce

s;L

5n

um

ber

ofla

stap

pea

ran

ces.

MO

M1

LE

NS

UR

101

R23

1R

291

HJ

DC

StF

L

794.

478

4.8

775.

676

6.8

758.

675

3.9

752.

674

7.0

738.

373

6.2

UR

PIN

PIN

PIN

PIN UR

PIN RE

PIN RE

1792 784.

877

5.6

766.

875

8.6

1618 752.

630

6.3

738.

329

1.4

sh ct md

ct ct md

md

md

ct tsd

BL

SLu

SLu

SLu

SLl

ML

SLl

SLl

SLl

SLl

121

410

379

250

269

111

339

352

315

149

12 66 51 48 43 22 38 23 62 28

0 6 1 6 0 3 2 1 8 3

10.9

836

.78

28.1

133

.62

27.5

420

.823

.55

13.9

137

.31

24.0

9

52.8

241

.92

46.8

740

.22

32.8

719

.05

54.9

4

57.8

146

.16

35.8

321

.06

60.1

8

0.67

91.

511

1.25

1.41

31.

213

0.86

41.

132

0.91

11.

491.

191

0.67

90.

831

0.73

20.

840.

743

0.64

30.

717

0.66

90.

831

0.82

3

2.41

23.

455

3.76

53.

439

3.7

2.59

23.

689

3.79

3.37

83.

692

12 67 83 86 91 92 94 96 102

101

0 0 1 1 2 1 1 0 1 3

55 16 4 6 3 3 3 6 073

3.7

724.

971

8.8

716.

871

5.6

715.

471

4.0

710.

870

4.6

702.

9

PIN

PIN T T

PIN UR T

PIN UR

PIN

733.

772

4.9

434.

543

7.4

715.

614

53 441.

471

0.8

1406

.770

2.9

ct md

st md

ct md

st ct md

ct

E E SLl

SLu

E ML E E ML E

410

199

151

295

425

127

336

379 52 387

72 32 34 57 65 30 56 38 18 40

6 2 10 8 10 4 11 4 0 1

32.7

923

.47

27.1

332

.88

30.1

926

.87

32.0

821

.19

22.2

4

53.0

1

50.3

747

.23

46.9

631

.12

31.8

3

60.1

2

56.6

153

.23

52.2

934

.22

35.2

3

1.34

81.

118

1.10

91.

394

1.16

51.

083

1.39

50.

975

1.05

61.

1

0.72

60.

743

0.72

40.

794

0.64

20.

733

0.79

80.

617

0.84

10.

686

3.39

63.

412

3.09

83.

368

3.41

72.

975

3.23

73.

312

2.72

93.

504

107

104

108

120

122

121

128

129

131

133

5 0 1 3 7 1 2 0 0 3

9 2 4 13 5 6 8 3 2 269

8.0

697.

669

6.6

693.

268

9.4

660.

664

0.3

636.

663

1.3

630.

5

T UR

PIN T

PIN

PIN UR

PIN UR

RE

464.

113

76.5

696.

647

0.9

689.

466

0.6

1130

.763

6.6

1092 145.

2

fsd

ct md

ct md

ct md

vfs

dfs

dco

al

ML

ML E ML

SLl

SLl

ML

BL

ML

ML

259

116

355

318

113

212 81 112

145

165

50 20 39 63 58 44 24 32 30 16

14 3 1 14 6 3 5 4 14 1

32.4

118

.79

23.6

531

.38

54.2

828

.87

30.1

23.8

812

.46

47.4

5

32.8

851

.38

36 59.4

8

1.39

0.91

91.

233

1.35

71.

652

1.26

31.

103

1.21

0.82

90.

44

0.81

80.

707

0.77

50.

754

0.93

70.

768

0.8

0.80

40.

561

0.36

6

2.97

42.

093.

353.

334

3.39

23.

407

2.19

2.82

62.

012

3.47

7

134

134

136

138

131

116

108

107

104

102

3 0 2 9 16 11 2 4 2 0

4 3 2 4 2 1 3 1 1 061

7.7

595.

458

7.4

569.

056

6.0

RE

RE

RE

C4

RE

127.

510

0.5

9646

19.5

84

fsd

md

vfs

dm

dm

d

ML

ML

ML

SLu

ML

54 58 72 76 93

27 31 35 18 40

3 5 15 5 18

1.31

11.

405

1.44

80.

894

1.45

8

0.91

60.

942

0.93

80.

712

0.91

3.33

12.

807

2.61

2.67

62.

641

104

104

102

100

100

3 2 2 0 3

2 3 0 0 0


TA

BL

E2.

Con

tin

ued

.

MO

M1

LE

NS

UR

101

R23

1R

291

HJ

DC

StF

L

564.

850

8.8

496.

748

1.4

475.

2

C4 T V V P

IN

4621

.373

3.1

1882

.519

15.4

475.

2

md

ct md

fsd

st

SLu

ML

ML

ML

ML

75 236

301

174

409

35 69 41 42 13

11 20 13 20 1

41.8

821

.57

29.5

55.

37

68.1

735

.3 8.92

40.2

3

10.3

6

1.33

81.

558

0.77

61.

243

0.38

1

0.86

70.

847

0.48

10.

766

0.34

2

2.95

2.64

73.

301

2.65

92.

135

98 100 88 82 80

5 14 8 3 3

1 7 2 2 147

1.8

451.

133

2.6

311.

429

8.3

266.

424

9.7

236.

421

8.2

214.

2

RE

ICP

EP

11P

IN UR

PIN P11 V P11

PIN

31.2 5.84

3.4

311.

458

826

6.4

51.2

822

82.6

69.4

521

4.2

fsd

md

md

st md

md

st st st vfs

d

BL

SLu

BL

ML

SLl

SLu

SLu

SLu

ML

SLu

391

416

179 91 313

254

298

135

306

318

30 20 40 23 38 35 56 23 20 30

2 4 10 2 11 4 23 5 6 3

17.4

69.

9629

.3

20.3

222

.98

30.7

319

.65

9.59

21.2

3

24.5

14.7

32.6

533

.59

48.9

8

16.4

527

.62

26.8

316

.51

36.7

55.3

1

19.3

129

.36

0.89

60.

453

1.16

0.97

80.

973

1.14

11.

207

0.97

10.

631

1.15

0.60

60.

348

0.72

40.

718

0.61

60.

739

0.69

0.71

30.

485

0.77

8

2.08

91.

988

2.01

41.

774

2.24

41.

603

1.75

90.

688

1.67

91.

325

77 69 69 74 76 82 82 79 79 82

10 6 1 2 3 2 5 1 0 2

0 2 6 6 4 9 2 2 1 321

4.0

196.

215

4.8

153.

814

9.5

136.

313

1.4

118.

311

7.6

105.

3

P11

PIN

PIN P11

P11

P11

PIN UR

P11

PIN

71.9

196.

215

4.8

106.

710

9.15

116.

813

1.4

214.

512

7.6

105.

3

st st vfs

dm

dm

dm

dm

dct m

dct

ML

SLu

SLu

SLu

SLu

SLl

SLl

SLu

SLu

SLl

319

387

100

326

174

481 89 400

303

500

21 32 20 29 16 20 31 22 26 23

5 1 1 3 8 6 2 4 8 3

10.5

518

.6

15.4

312

.01

8.45

14.4

15.6

712

.47

17.4

227

.23

24.1

6

13.3

7

18.5

722

.92

17.4

7

19.9

229

.59

27.3

3

15.1

9

19.9

225

.52

19

0.55

10.

898

0.98

80.

866

0.71

20.

579

1.32

20.

919

0.75

90.

731

0.41

70.

596

0.75

90.

592

0.59

10.

445

0.88

60.

685

0.53

60.

537

1.36

11.

782

1.19

51.

012

1.54

51.

157

1.47

11.

367

1.73

41.

463

80 78 75 75 73 76 77 75 75 72

4 3 1 2 0 0 3 2 3 0

0 2 0 1 0 3 1 1 2 0

81.0

80.9

79.2

66.7

62.0

53.7

46.8

43.9

37.0

30.9

PIN P11

PIN P11 UR

P11

PIN UR

PIN UR

81 148.

879

.215

7 97.5

164.

546

.860

.137 33

.1

md

md

ct md

md

md

ct vfs

dct v

fsd

SLl

SLl

SLl

SLl

SLl

SLl

SLu

E SLl

E

80 61 336

415

106

329 75 49 170

123

17 10 20 35 35 21 27 24 24 28

0 5 3 10 12 6 2 6 3 8

13.8

515

.61

34 14.0

1

18.8

824

.59

17.6

924

.93

18.6

8

19.0

428

.52

20.1

6

1.01

80.

369

0.94

50.

923

1.24

80.

623

1.21

81.

220.

948

1.06

6

0.82

70.

369

0.72

60.

598

0.80

80.

471

0.85

10.

886

0.68

70.

737

1.33

60.

009

1.16

22.

138

1.13

90.

142

1.10

11.

463

1.02

1.06

4

73 73 75 74 72 72 72 68 64 61

1 0 1 4 2 0 5 5 3 2

1 1 2 0 2 2 0 1 1 030

.830

.615

.711

.8 5.4

21.

02

27.3

P11

PIN UR

P11

PIN P11

P11

177.

730

.6 1.5

188.

75.

419

6.1

211.

3

md

md

ct md

md

sh md

SLl

SLl

SLl

SLl

SLu

SLl

SLl

254

114

278

315

172 55 281

28 28 16 37 20 6 35

3 6 3 7 4 1 3

20.3

926

.13

10.6

521

.18

17.0

3

22.0

7

27.1

7

15.0

432

.11

32.4

1

35.7

4

1.07

61.

104

0.50

90.

975

0.95

90.

405

1.12

3

0.74

40.

763

0.42

30.

622

0.73

70.

520.

727

1.17

41.

286

0.13

70.

634

1.53

60 0.

953

60 56 50 46 40 32 32

5 6 5 6 8 0

1 1 0 1 0 0 0


FIGURE 5. Rarefied number of species at two differentknot points, 101 and 291 grains. In both cases the meannumber of species for the Eocene samples is significant-ly higher than for the Paleocene (t-test: p , 0.001, and p, 0.0053 respectively). Units of composite section aregiven in stratigraphic meters of the reference section(Pinalerita).

FIGURE 6. Bootstrap of empirical and estimated (Chao2)species 5 accumulation curves for the Eocene and Pa-leocene group of samples. One standard deviation isshown in the error bar. Standard deviations for the em-pirical curve are too small to show on the scale of thegraphic. Pooled Paleocene samples accumulate speciesmore slowly than Eocene samples. Paleocene Chao2 es-timator of species diversity is significantly lower thanEocene Chao2 at any pooled sample level (paired t-test:p , 0.004).

LAD data sets was polynomial (Fig. 11B; r2 50.99 and r2 5 0.99 respectively).

Five morphologically cohesive groups cor-responding to higher taxa were analyzed sep-arately, species occurring in only one samplewere excluded from analysis. Standing diver-sity of Bombacacidites, Polypodiisporites, and Te-tracolporopollenites is significantly higher in theEocene (Table 1). The relative abundance ofthe most important taxon in Paleocene sam-ples, Proxapertites, exhibits a significant de-crease from the Paleocene to the Eocene (Table1, Fig. 12). Fern spores are significantly moreabundant and diverse in Eocene samples thanin Paleocene samples (Table 1, Fig. 12). How-

ever, the proportion of fern spores relative tothe total standing diversity per sample did notchange appreciably, from 17.2% in Paleoceneto 18.1% in Eocene strata (Table 1).

Discussion

Tropical pollen and spores provide a recordof many canopy trees, shrubs, and understoryherbs that is useful for reconstructing localand regional tropical vegetational patternsand community dynamics in Quaternary andpre-Quaternary sediments (Burnham andGraham 1999; Colinvaux et al. 1999). Pollenand spores also provide a measure of mini-mum taxic diversity because of their low tax-onomic resolution; e.g., some modern familiessuch as Poaceae produce morphologically uni-form pollen generally indistinguishable at thegenus or species level (Traverse 1988; Graham1999). The vegetation change recorded by pol-len and spores can be masked by several fac-tors, mainly sampling size, sampling density,lithologies, and depositional environments(Farley 1989, 1990; Holland 1995). Therefore,


FIGURE 7. Change through time in Shannon index (H) and Evenness of abundance distribution (J) within samples.Mean Paleocene H is lower that mean Eocene H (t-test: p , 0.001), mean Paleocene J is lower than mean Eocene J(t-test: p , 0.026). Units of composite section are given in stratigraphic meters of the reference section (Pinalerita).Vertical line shows mean; horizontal box shows one standard deviation.

an analysis of pollen and spore diversity musttake into account these factors before any in-ference can be made from the diversity pat-tern observed.

Rarefaction and bootstrap results indicatethat, regardless of the number of pollen grainsand spores counted per sample or the numberof samples studied in each epoch, there is astatistically significant difference in diversity

between late Paleocene and early to middleEocene strata (Table 1, Figs. 5, 6). ANOVA re-sults also indicate that variations in lithofaciesor depositional environments are not respon-sible for the differences in diversity betweenPaleocene and Eocene samples. Pollen andspore recovery rates for Eocene strata are notsignificantly different from Paleocene strata,and palynofloras from Eocene depositional


FIGURE 8. Bootstrap of Shannon index and an empirical species 5 accumulation curve for the Eocene and Paleocenesamples that were grouped by geographic area. Both graphics show that Eocene diversity is higher than Paleocenediversity in Catatumbo and Pinalerita, whereas diversity does not significantly change in Uribe (see discussion).Pin 5 Pinalerita 1 C4; Uribe 5 Uribe 1 ICPE-2; Catatumbo 5 P11 1 Tibu-1 1 Veta-1 cores.

environments are more diverse than their Pa-leocene counterparts.

Alpha Diversity. A primary element inmeasuring diversity is the alpha diversity, ornumber of species within samples (Hayek andBuzas 1997). Every parameter used to mea-sure the alpha diversity indicates that Eocenesamples are more diverse on average than Pa-leocene samples. The mean number of speciesper sample measured using standing diversi-ty data shows that the late Paleocene is less di-verse (68 species/sample) than the early–mid-dle Eocene (103 species/sample; Fig. 4). Rar-efaction data that take into account the ob-served number of individuals per sample alsoshow that Eocene samples are significantlymore diverse than Paleocene samples (Fig. 5).The Shannon and Evenness indices (Fig. 7) in-dicate that Eocene strata not only have on av-erage a significantly higher number of species

per sample, but also that the abundance dis-tribution of Eocene taxa is more equitable.

The abundance distribution of Eocene pa-lynofloras seem to be similar to pollen recordsof modern tropical rain forest, where the as-semblages are composed of many taxa thatappear in low numbers (Colinvaux et al. 1999).I used a 1.9-m pollen record of the last 42 Kyrfrom Lake Patas, lowland Amazon rain forest(Colinvaux et al. 1996), to compare the diver-sity of modern tropical rain forest with the Eo-cene record. The mean diversity for the Patacore was 37.1 pollen types/sample (SD 5 14.3,number of samples 5 49; grains counted persample ;300). This diversity value is verysimilar to the Eocene mean diversity of 41.3(SD 5 16.1; rarified grains counted 5 291).The comparison of diversity per sample be-tween the Eocene and the Quaternary recordsis realistic because the time represented in a


FIGURE 9. Changes in palynofloral composition as mea-sured by the Detrended Correspondence Analysis(DCA). The first axis explains 15.84% of variation. Thereis a clear separation of Paleocene from Eocene palyno-floras. Units of composite section are given in strati-graphic meters of the reference section (Pinalerita).

FIGURE 10. Standing diversity of range-through spo-romorphs divided into those restricted to Paleocene,those restricted to Eocene, and those that occur in bothPaleocene and Eocene strata. Taxa with occurrences inonly one sample were excluded. Units of composite sec-tion are given in stratigraphic meters of the referencesection (Pinalerita).

sample from the Eocene record is within thesame order of magnitude as a sample from thePatas record. A sample (;1 stratigraphic cm ofrock) in the Eocene would correspond to ;250years (given the accumulation rate of ;1 m/25 Kyr found for the Eocene in this study),whereas a Quaternary sample would corre-spond to ;234 years before compaction inLake Patas.

Beta Diversity. Beta diversity, or diversityamong samples (Whittaker 1972), measureshow different sites or habitats are in terms ofthe abundance and/or presence/absence ofspecies (Magurran 1998). The terminal slopeof the empirical species-accumulation curvescan give an idea of the degree of heterogeneityof a region, a proxy for beta diversity. If the

beta diversity is low, the bootstrapped spe-cies-accumulation curve becomes flatter as thepooled number of species increase. This flat-tening is produced because, as the pooling ofsamples progresses, fewer new species areadded to the total number of species. The em-pirical species-accumulation curves show thatthe Eocene curve flattens less than the Paleo-cene curve (Fig. 6). This suggests that the Eo-cene has a greater between-sample heteroge-neity than the Paleocene. This is also con-firmed by the Chao2 curves (Fig. 6). Chao2 at-tempts to estimate the number of species of aregion mainly from the number of uniques.


FIGURE 11. Stratigraphic first-appearance datums (FADs) and last-appearance datums (LADs) from the late Paleoceneto the middle Eocene in the central Colombian sections through the study interval. A, Piecewise regression to deter-mine edge effect in both LAD and FAD record. Edge effect for FADs ends at species 79 and edge effect for LADs endsat species 83. B, Regression analysis on FAD and LAD. Data attributed to edge effects were excluded from analysis.The best fit was polynomial. Note the increase in FAD rate and the slightly declining LAD rate during the Eocene.

Chao2 curves for the Eocene indicate a higherexpected number of species, demonstratingthat Eocene samples have a higher proportionof uniques and suggesting a greater between-sample heterogeneity. Both empirical andChao2 curves do not significantly flatten, sug-gesting severe undersampling (Fig. 6). Thisundersampling is probably because the sam-ples come from a large area (;2.5 3 105 km2),a range of depositional environments, and along time period. This array of factors increas-es the heterogeneity of the samples and there-fore produces an increase in the number ofrare species that ultimately controls the shapeof the species-accumulation curve (Colwell

and Coddington 1994). Analysis of sectionsgrouped in smaller geographic areas, ratherthan all sections pooled together, allows ex-amining the spatial pattern of this increase indiversity. Rarefaction, empirical accumulationcurves, and bootstrapped Shannon index (H)indicate that diversity significantly increasedfrom the Paleocene to the Eocene in the Pin-alerita and Catatumbo areas (Fig. 8). The di-versity in the Uribe area does not significantlychange (Fig. 8), probably owing to the effectof small sample size and depositional envi-ronments on diversity patterns. In the Uribearea 83% of Paleocene samples (five out of six)come from lower coastal plain (SLl) and es-


FIGURE 12. Changes in relative abundance and standing diversity (excluding uniques) of several monophyleticgroups. The standing diversity of Bombacacidites (pollen from Bombacaceae, a tropical woody family) during theEocene is significantly higher than during the Paleocene. Relative abundance of Proxapertites (an extinct pollen typefrom Palmae, dominant element of coastal plain Paleocene environments) significantly decreases during the Eocene.The relative abundance and diversity of fern spores are significantly higher in Eocene samples than in Paleocenesamples. Units of composite section are given in stratigraphic meters of the ref erence section (Pinalerita).

tuarine environments. In contrast, 87.5% ofEocene samples (six out of seven) are from flu-vial (BL) and fluvial to coastal (ML) environ-ments (Table 2). The mean H for all samplesfrom ML Eocene environments is 1.10, signif-icantly higher than the H value for SLl Paleo-cene environments of 0.98 (p , 0.05). Further-more, Paleocene environments are always lessdiverse than their Eocene counterparts, as in-

dicated by t-tests and ANOVA results. Thisshows the importance of analyzing more thanone section when studying diversity patternsthrough time.

A greater Eocene heterogeneity is also sup-ported by the Jaccard and Morisita-Horn sim-ilarity indices that shows higher similaritywithin epochs than among them, and that Eo-cene samples are more heterogeneous than


Paleocene samples (Table 1). However, wheth-er the effects of unequal sampling of the dif-ferent depositional environments resulted inhigher Eocene heterogeneity is still not clear.When the indices in the environment with themost similar sampling density (SLu) werecompared, the Eocene samples were still lesssimilar to each other than the Paleocene sam-ples, but the difference were not statisticallysignificant (p , 0.59 using Jaccard and p ,0.084 using Morisita-Horn, Table 1). A moreequal balance in sampling distribution amongdepositional environments is still needed totest if Eocene beta diversity was indeed higherthan Paleocene beta diversity, as the data sug-gests.

Overall, Eocene strata have higher alphaand beta diversities than Paleocene strata, andare comparable to many modern Neotropicalrain forests (Gentry 1988). Not only do indi-vidual pollen samples have a larger number ofspecies, but also the change in palynofloralcomposition between samples is higher. Thisraises the possibility that tropical Eocene flo-ras were the first forests with diversities com-parable to modern tropical rain forests.

Floral Change. A floral change through thelate Paleocene to early–middle Eocene intervalis evident from the detrended correspondenceanalysis (Fig. 9). It is clear that Paleocene flo-ras are more similar to each other than to Eo-cene palynofloras as indicated by the meanDCA score on the first axis (Table 1). Was thischange gradual or fast? Rates of first- and last-appearance datums (FADs and LADs respec-tively) indicate that there is a gradual increasein the rate of FADs and a gradual decrease inthe rate of LADs from the late Paleocene to theearly–middle Eocene (Fig. 11B). This supportsviews of high tropical diversity produced byhigh rates of origination and low rates of ex-tinction (Sepkoski 1998). It is important tokeep in mind, however, that more detailed in-terpretations of these rates may be warrantedbecause the stratigraphic position of the chro-nological boundaries (Paleocene/Eocene andearly/middle Eocene boundaries) are stillpoorly constrained (Jaramillo and Dilcher2000, 2001).

The rates of FADs and LADs suggests thatEocene palynofloras are significantly more di-

verse than Paleocene floras because of a fasteraddition of new taxa coupled with a decreasein the number of extinctions. This increase inFAD and decrease in LAD rates seems to be aphenomenon present throughout the earlyand early middle Eocene rather than discretepulses of new occurrences. A higher rate ofFADs could be the product of higher origina-tion and/or higher immigration rates. Rull(1999) showed a gradual and significant pa-lynofloral change from late Paleocene to earlyEocene strata in Venezuela and also suggestedthat the early Eocene could be a time of highorigination rate. This phenomenon seems tobe a regional pattern for northern SouthAmerica and has been noted previously byother authors (Gonzalez 1967; Romero 1993;Van der Hammen and Hooghiemstra 2000).Romero (1993), recognizing the high turnoverin taxa, considers the Eocene floras in tropicalSouth America to be significantly differentfrom his late Cretaceous–Paleocene Palmaefloristic province. He classifies the Eocene flo-ras in a new floristic province called ‘‘Neo-tropical,’’ with more than 45 angiosperm fam-ilies having first occurrences within Eocene.

Most of the species appearing in the Eocenehave been reported consistently only in north-ern South America (Germeraad et al. 1968; Re-gali et al. 1974; Colmenares and Teran 1993;Rull 1999; Morley 2000), suggesting that alarge proportion of the increase in diversitymay be the product of in situ evolution ratherthan immigration from other latitudes; ex-amples are the monophyletic taxa Bombacaci-dites (similar to that of extant Bombax and rel-atives within Bombacaceae) and Tetracolporo-pollenites (Sapotaceae), which are significantlymore diverse in Eocene times (Table 1, Fig. 12).There is a small percentage of taxa, however,that clearly are immigrants from northern lat-itudes, such as the fern spore Cicatricosisporitesdorogensis and the angiosperm Corsinipollenites(Onagraceae). An overall comparison of taxabetween northern South America and the U.S.Gulf Coast indicates an increase in palyno-floral shared taxa from 0.7% in the Paleoceneto 5.2% in the Eocene (Jaramillo and Dilcher2001). Nonetheless, these shared taxa repre-sent a small proportion (,6%) of the total in-crease in diversity, even if we assume that the


direction of migration of the shared taxa wasfrom north to south and therefore added tothe diversity in tropical latitudes.

Not only was a new wave of taxa present inthe Eocene, but the dominant taxa changed aswell. Paleocene coastal plain deposits arestrongly dominated by only a few taxa: Prox-apertites (Palmae), Bombacacidites annae (Bom-bacaceae), and Retidiporites magdalenensis (Pro-teaceae). In contrast, Eocene coastal plains aredominated by many more taxa: several types ofspores, Lanagiopollis crassa (Pellicieraceae), Per-isyncolporites pokornyi (Malphigiaceae), Spini-zonocolpites grandis (Palmae), Brevitricolpitesgroup, and others. Eocene assemblages are in-deed more equitable, as indicated by the Even-ness index (Fig. 7).

Climate Change. Analyses of extant Neo-tropical rain forests have shown that there isa positive linear correlation between rainfalland number of plant species (Gentry 1982,1986, 1988). In addition, a significantly largerabundance and diversity of ferns exists intropical forests that have higher annual rain-fall (Moran 1998). These two characteristics,an increase in overall plant diversity and anincrease in fern spore abundance and diver-sity (Figs. 4, 12), were observed in the early tomiddle Eocene in this study. They suggestthat the early Eocene warming in the Tropicswas associated with an increase in effectiverainfall. Increased rainfall in the South Amer-ican Tropics has been predicted by climaticmodels of early Eocene warming that use con-centrations of CO2 similar to or twice prein-dustrial values (Sloan and Rea 1995; Huberand Sloan 1999; Bice et al. 2000). In these mod-els, the net precipitation beneath the intertrop-ical convergence zone doubles and tropicaltemperature is only slightly warmer than atpresent (Bice et al. 2000). In contrast, the datahere do not support other models that use sig-nificant volumes of greenhouse gases (;sixtimes CO2), and increase tropical land tem-peratures by ;48C without increasing totalrainfall, which would thus decrease soil mois-ture and effective rainfall (Sloan and Rea1995). Sedimentological or isotopic data toconfirm this hypothesis of increased rainfallare still lacking for tropical terrestrial ecosys-tems.

The hypothesis proposed here of increasedrainfall during the ETM correlated to an in-crease in diversity could be affected by the po-sition of the early/middle Eocene boundary.Its exact position is tentative because there isstill not a first-order calibration of the paly-nological record with the geological timescalefor northern South America. According to thepresent position of the boundary (Fig. 2), thehighest peak of diversity seems to be duringthe early middle Eocene rather then duringthe late early Eocene (Figs. 4, 11), even thoughthe increase in diversity begins in the late ear-ly Eocene (Fig. 4). If the position of the bound-ary is correct, perhaps the effects of the ETMevent lasted longer in tropical latitudes, as val-ues of d18O returned to Paleocene levels dur-ing the lower middle Eocene, ;47.5 Ma (Za-chos et al. 2001).

An alternative explanation would be thatthe increase in the rate of first appearances(originations) coupled with a slight decreasein the rate of last appearances (extinctions),starting in the late early Eocene, did not endwith the termination of the ETM. In that case,the internal dynamics of the diversificationprocess would become more dominant thanthe causes that started the process in the firstplace. Origination usually increases in the ini-tial stages of a radiation process until mostavailable niches are filled, and then decreasesas crowding becomes significantly increased(Sepkoski 1978; McKinney 1998). As morespecies are generated during the radiation, theniche space occupied by each species, as wellas its population density, contracts (Ricklefsand Schluter 1993). Therefore, it would be ex-pected that following this fast initial increasein diversity, the rate of extinction should in-crease as more species are added to the region(Sepkoski 1978), or the rate of originationshould slow, approximating the extinctionrate (McKinney 1998). Unfortunately, rates ofLADs for the early middle Eocene could notbe distinguished from those produced by theedge effect (Fig. 11A,B), and FAD rates forstrata younger than 46.5 Ma were beyond thisstudy.

A third hypothesis would state that theETM and the increase in diversity are not re-lated. However, a mechanism to increase orig-


ination rates would still be necessary. It wouldbe a regional phenomenon affecting a wholerange of clades, as this study has shown.

An overall pattern of increase in plant di-versity during the early to middle Eocene wasobserved in the Bighorn basin in Wyoming(Wing 1998; Wilf 2000). Wing (1998) suggest-ed that warming in the early Eocene increasedplant inter- and intracontinental migration,producing the pattern observed. Thus, al-though the pattern of increased diversity issimilar in both temperate and tropical areas,the causes are different—migration and in-creased temperature in mid- to high latitudes,and in situ origination and increased rainfallin tropical latitudes.

Considerations about the effects of the Pa-leocene/Eocene thermal maximum (PETM)on the palynofloral record are still uncertainbecause this event has not been precisely iden-tified in any of the sections. Furthermore, the118.5-m stratigraphic interval where thePETM is thought to occur is barren for pollenand spores, despite extensive sampling in foursections in different parts of the basin. It is un-certain if there is any connection between theabsence of pollen and spores and the PETMclimate change. However, the widespread sim-ilarity of fine-grained, variegated, oxidizedlithofacies with poor pollen and spore recov-ery in this interval across the entire study areasuggests a regional phenomenon that pro-duced a greater fluctuation of the water table,presumably an indication of greater season-ality of precipitation (Bown and Kraus 1981).This interpretation agrees with greenhouseclimate modeling for the PETM that predictsthat tropical South America was drier in Jan-uary and wetter in July, increasing seasonality(Huber and Sloan 1999). A similar phenome-non has been observed in continental depositsaccumulated during the PETM in Wyoming(Bown and Kraus 1981; Wing and Harrington2001), where pollen and spores were not pre-served in the interval containing the PETM(Bown and Kraus 1981; Wing and Harrington2001). A major problem with this interpreta-tion is that given the calculated rates of accu-mulation for the Paleocene, the PETM wouldbe restricted to ;15 m, much less than the re-corded barren interval of 118.5 m. A possible

solution is that leaching during the PETM wasintense enough to affect many meters of sub-strate. If true, the PETM should be located inthe top ;16 m of the barren interval. Solvingthis dilemma requires finding the precise lo-cation of the PETM using carbon isotope stra-tigraphy, coupled with a more refined calibra-tion of the biostratigraphic framework thanthat used for this study.

Conclusions

Analysis of pollen and spore diversity ineight sections in central Colombia indicatesthat the lower to middle Eocene strata containa significantly higher palynofloral diversitythan the associated upper Paleocene strata.This pattern is maintained after sample size,number of samples/time unit, lithofacies, anddepositional systems are accounted for. Eocenepalynofloras have a higher number of speciesper given sampling unit, a higher equitability,and a larger among-sample heterogeneity thanPaleocene palynofloras. This increase in diver-sity is the product of a gradual increase in therate of first occurrences and a gradual decreasein the rate of last occurrences.

The early to middle Eocene increase in di-versity, and the increase in spore abundanceand diversity, suggests a wetter climate due toincreased effective annual rainfall. This inter-pretation favors climate models that do not usesignificant increases of greenhouse gases topromote the early Eocene warming. The cor-relation of the palynofloral pattern with cli-matic conditions during the latest Paleocenethermal maximum is still uncertain. Strata as-sociated with the PETM were barren for paly-nomorphs. Lithofacies suggests that the PETMaffected the preservation of pollen and sporesbecause of greater seasonality in rainfall.

Acknowledgments

This study was funded by the SmithsonianInstitution Fellowships and Grants Program,the Petroleum Research Fund (grant number34676 to Francisca Oboh-Ikuenobe, adminis-tered by the American Chemical Society); Col-ciencias, the Fundacion para la promocion dela investigacion y la tecnologıa Banco de la Re-publica; and the Corporacion Geologica Ares.Reviews by R. Lupia and P. Wilf have im-


proved the article considerably. I am gratefulto S. Wing for highly valuable discussions re-lating to data analysis, plant dynamics, andpaleobiology. Thanks to Instituto Colombianodel Petroleo and to J. Cuellar for permission tosample cores C4, Tibu-1, Veta-1, ICPE-2, andP11. Thanks to G. Bayona and J. Roncancio fortheir help during fieldwork, to L. Hayek forstatistical advice, and to B. Kowalski. The Cor-poracion Geologica Ares provided valuablelogistic support. I am also grateful to all thepeople who helped during the field season inthe towns of Sabanalarga, Uribe-Uribe, andCucuta. Special thanks go to M. I. Barreto forher continuous support and source of ideas.

Literature CitedAskin, R. A., and R. A. Spicer. 1995. The Late Cretaceous and

Cenozoic history of vegetation and climate at northern andsouthern high latitudes: a comparison. Pp. 156–173 in Nation-al Research Council, eds. Effects of past global change on life.National Academy Press, Washington, D.C.

Baker, P. A., C. A. Rigsby, G. O. Seltzer, S. C. Fritz, T. K. Low-ensteink, N. P. Bacher, and C. Veliz. 2001a. Tropical climatesat millennial and orbital timescales on the Bolivian Altiplano.Nature 409:698–701.

Baker, P. A., G. O. Seltzer, S. C. Fritz, R. B. Dunbar, M. J. Grove,P. M. Tapia, S. L. Gross, H. D. Rowe, and J. P. Broda. 2001b.The history of South American tropical precipitation for thepast 25,000 years. Science 291:640–643.

Berggren, W. A., D. V. Kent, C. C. Swisher II, and M. Aubry.1995. A revised Cenozoic geochronology and chronostratig-raphy. Pp. 129–212 in W. A. Berggren, D. V. Kent, M. P. Aubry,and J. Hardenbol, eds. Geochronology time scales and globalstratigraphic correlation. SEPM, Tulsa.

Bice, K. L., L. C. Sloan, and E. J. Barron. 2000. Comparison ofearly Eocene isotopic paleotemperatures and the three-di-mensional OGCM temperature field: the potential for use ofmodel-derived surface water d18O. Pp. 79–131 in B. T. Huber,K. G. MacLeod, and L. W. Scott, eds. Warm climates in earthhistory. Cambridge University Press, Cambridge.

Boltovskoy, D. 1988. The range-through method and first-lastappearance data in paleontological surveys. Journal of Pale-ontology 62:157–159.

Bown, T. M., and M. J. Kraus. 1981. Lower Eocene alluvial pa-leosols (Willwood Formation northwestern Wyoming, USA)and their significance for paleoecology, paleoclimatology, andbasin analysis. Palaeogeography, Palaeoclimatology, Palaeoe-cology 34:1–30.

Bralower, T. J., J. C. Zachos, E. Thomas, M. Parrow, K. Paull, D.C. Kelly, I. Premoli Silva, W. V. Sliter, and K. C. Lohmann.1995. Late Paleocene to Eocene paleoceanography of theequatorial Pacific Ocean: stable isotopes recorded at OceanDrilling Program Site 865, Allison Guyot. Paleoceanography10:841–865.

Burnham, R. J., and A. Graham. 1999. The history of neotropicalvegetation: new developments and status. Annals of the Mis-souri Botanical Garden 86:546–589.

Bush, M. B. 1994. Amazonian speciation: a necessarily complexmodel. Journal of Biogeography 21:5–17.

Christophel, D. C. 1995. The Impact of climatic changes on thedevelopment of the Australian flora. Pp. 156–173 in National

Research Council, eds. Effects of past global change on life.National Academy Press, Washington, D.C.

Clyde, W. C., and P. D. Gingerich. 1998. Mammalian communityresponse to the latest Paleocene thermal maximum: an iso-taphonomic study in the northern Bighorn Basin, Wyoming.Geology 26:1011–1014.

Colinvaux, P. A., and P. E. de Oliveira. 2001. Amazon plant di-versity and climate through the Cenozoic. Palaeogeography,Palaeoclimatology, Palaeoecology 166:51–63.

Colinvaux, P. A., P. E. de Oliveira, J. E. Moreno, M. C. Miller, andM. B. Bush. 1996. A long pollen record from lowland Ama-zonia: forest and cooling in glacial times. Science 274:85–88.

Colinvaux, P., P. E. de Oliveira, and J. E. Moreno. 1999. Amazonpollen manual and atlas. Hardwood Academic, Amsteldijk.

Colinvaux, P. A., P. E. de Oliveira, and M. B. Bush. 2000. Ama-zonian and neotropical plant communities on glacial time-scales: the failure of the aridity and refuge hypothesis. Qua-ternary Science Reviews 19:141–169.

Colmenares, O. A., and L. Teran. 1993. A biostratigraphic studyof Paleogene sequences in southwestern Venezuela. Palynol-ogy 17:67–89.

Colwell, R. K. 2000. EstimateS, Version 6. 0b1. http://vice-roy.eeb.uconn.edu/Estimates6/.

Colwell, R. K., and J. A. Coddington. 1994. Estimating terrestrialbiodiversity through extrapolation. Philosophical Transac-tions of the Royal Society of London B 345:101–118.

Cowling, S. A., M. A. Maslin, and M. T. Sykes. 2001. Paleove-getation simulations of lowland Amazonia and implicationsfor Neotropical allopatry and speciation. Quaternary Re-search 55:140–149.

Crowley, T. J., and J. C. Zachos. 2000. Comparison of zonal tem-perature profiles for past warm past periods. Pp. 50–76 in B.T. Huber, K. G. MacLeod, and S. L. Wing, eds. Warm climatesin earth history. Cambridge University Press, Cambridge.

Duggleby, R. G., and L. C. Ward. 1991. Analysis of physiologicaldata characterized by two regimes separated by an abrupttransition. Physiological Zoology 64:885–889.

Farley, M. 1989. Palynological facies fossils in nonmarine envi-ronments in the Paleogene of the Bighorn Basin. Palaios 4:565–573.

———. 1990. Vegetation distribution across the early Eocene de-positional landscape from palynological analysis. Palaeo-geography, Palaeoclimatology, Palaeoecology 79:11–27.

Foote, M. 2000. Origination and extinction components of tax-onomic diversity: general problems. Pp. 74–102 in D. H. Er-win, and S. L. Wing, eds. Deep time: Paleobiology’s perspec-tive. Paleobiology 26(Suppl. to No. 4):74–102.

Freitas, H. A. de, L. C. Ruiz, R. Aravena, S. E. Marques, A. S.Ribeiro, and R. Boulet. 2001. Late Quaternary vegetation dy-namics in the southern Amazon basin inferred from carbonisotopes in soil organic matter. Quaternary Research 55:39–46.

Gentry, A. H. 1982. Patterns of neotropical plant species diver-sity. Evolutionary Biology 15:1–84.

———. 1986. Species richness and floristic composition of theChoco region plant communities. Caldasia 15:71–91.

———. 1988. Changes in plant community diversity and floris-tic composition on environmental and geographical gradi-ents. Annals of the Missouri Botanical Garden 75:1–34.

Germeraad, J. H., C. A. Hopping, and J. Muller. 1968. Palynologyof Tertiary sediments from tropical areas. Review of Palaeo-botany and Palynology 6:189–348.

Gilinsky, N. L. 1991. Bootstrapping and the fossil record. In N.L. Gilinsky and P. W. Signor, eds. Analytical paleobiology.Short Courses in Paleontology 4:185–206. Paleontological So-ciety, Knoxville.

Gonzalez, A. E. 1967. A palynologic study on the upper Los


Cuervos and Mirador formations (lower and middle Eocene),Tibu Area, Colombia. E. J. Brill, Leiden.

Graham, A. 1999. Late Cretaceous and Cenozoic history ofNorth American vegetation. Oxford University Press, NewYork.

Haberle, S. G., and M. A. Maslin. 1999. Late Quaternary vege-tation and climate change in the Amazon Basin based on a50,000 year pollen record from the Amazon Fan, ODP Site 932.Quaternary Research 51:27–38.

Hayek, L. C., and M. A. Buzas. 1997. Surveying natural popu-lations. Columbia University Press, New York.

Hill, M. O., and H. G. Gauch. 1980. Detrended correspondenceanalysis, an improved ordination technique. Vegetatio 42:47–58.

Hillborn, R., and M. Mangel. 1997. The ecological detective.Princeton University Press, Princeton, N.J.

Holland, S. M. 1995. The stratigraphic distribution of fossils. Pa-leobiology 21:92–109.

Hooghiemstra, H., and T. Van der Hammen. 1998. Neogene andQuaternary development of the neotropical rain forest: theforest refugia hypothesis, and a literature overview. Earth-Science Reviews 44:147–183.

Huber, M., and L. C. Sloan. 1999. Warm climate transitions: ageneral circulation modeling study of the Late PaleoceneThermal maximum (;56 Ma). Journal of Geophysical Re-search 104:16633–16655.

Jaramillo, C., and D. L. Dilcher. 2000. Microfloral diversity pat-terns of the late Paleocene–Eocene interval in Colombia,northern South America. Geology 28:815–818.

———. 2001. Middle Paleogene palynology of central Colom-bia, South America: a study of pollen and spores from tropicallatitudes. Palaeontographica, Abteilung B 258:87–213.

Kelly, D. C., T. J. Bralower, and J. C. Zachos. 1998. Evolutionaryconsequences of the latest Paleocene thermal maximum fortropical planktonic foraminifera. Palaeogeography, Palaeocli-matology, Palaeoecology 141:139–161.

Kennett, J. P., and L. D. Stott. 1991. Abrupt deep-sea warming,palaeoceanographic changes and benthic extinctions at theend of the Paleocene. Nature 353:225–229.

Kovach, W. L. 1998. MVSP-A multivariate statistical package forWindows, Version 3.0. Kovach Computing Services, Pen-traeth, Wales.

Magurran, A. E. 1988. Ecological diversity and its measure-ment. Princeton University Press, Princeton, N.J.

Maslin, M. A., and S. J. Burns. 2000. Reconstruction of the Am-azon basin effective moisture availability over the past 14,000years. Science 290:2285–2287.

Mayle, F. E., R. Burbridge, and T. J. Killeen. 2000. Millenial-scaledynamics of southern Amazonian rain forests. Science 290:2291–2294.

McAleece, N., P. J. D. Lambshead, G. L. Paterson, and J. G. Gage.1997. Biodiversity Professional Beta Version. http://www.nhm.ac.uk/zoology/bdpro.

McKinney, M. L. 1998. Biodiversity dynamics: niche preemptionand saturation in diversity equilibria. Pp. 1–16 in M. L. Mc-Kinney and J. A. Drake, eds. Biodiversity dynamics: turnoverof populations, taxa, and communities. Columbia UniversityPress, New York.

Miller, K. G., R. G. Fairbanks, and G. S. Mountain. 1987. Tertiaryoxygen isotope synthesis, sea level history, and continentalmargin erosion. Paleoceanography 2:1–19.

Moran, R. C. 1998. The genera of neotropical ferns, a guide forstudents. Organization for Tropical Studies, San Jose, CostaRica.

Morley, R. J. 2000. Origin and evolution of tropical rain forests.Wiley, New York.

Norris, R. D., and U. Rohl. 1999. Carbon cycling and chronology

of climate warming during the Paleocene/Eocene transition.Nature 401:775–778.

Pak, D. K., and K. G. Miller. 1992. Paleocene to Eocene benthicforaminiferal isotopes and assemblages: implications fordeep water circulation. Paleoceanography 7:405–422.

Pielou, E. C. 1984. The interpretation of ecological data. Wiley,New York.

Prentice, M. L., and R. K. Matthews. 1988. Cenozoic ice-volumehistory: development of a composite oxygen isotope record.Geology 16:963–966.

Raup, D. M. 1975. Taxonomic diversity estimation using rare-faction. Paleobiology 1:333–342.

Regali, M., N. Uesugui, and A. Santos. 1974. Palinologia dos se-dimentos Meso-Cenozoicos do Brasil. Boletim Tecnico da Pe-trobras 17:177–191.

Ricklefs, R. E., and D. Schluter. 1993. Species diversity: regionaland historical influences. Pp. 350–363 in R. E. Ricklefs and D.Schluter, eds. Species diversity in ecological communities:historical and geographical perspectives. University of Chi-cago Press, Chicago.

Rind, D., and M. Chandler. 1991. Increased ocean heat trans-ports and warmer climate. Journal of Geophysical Research96:7437–7461.

Romero, E. J. 1993. South American paleofloras. Pp. 62–85 in P.Goldblatt, ed. Biological relationships between Africa andSouth America. Yale University Press, New Haven, Conn.

Rosenzweig, M. L. 1995. Species diversity in space and time.Cambridge University Press, Cambridge.

Rull, V. 1999. Palaeofloristic and palaeovegetational changesacross the Paleocene/Eocene boundary in northern SouthAmerica. Review of Palaeobotany and Palynology 107:83–95.

Sepkoski, J. J.,Jr. 1978. A kinetic model of Phanerozoic taxonom-ic diversity. I. Analysis of marine orders. Paleobiology 4:223–251.

———. 1998. Rates of speciation in the fossil record. Philosoph-ical Transactions of the Royal Society of London B 353:315–326.

Sloan, L. C., and E. J. Barron. 1992. A comparison of Eocene cli-mate model results to quantified paleoclimatic interpreta-tions. Palaeogeography, Palaeoclimatology, Palaeoecology 93:183–202.

Sloan, L. C., and C. Morrill. 1998. Orbital forcing and Eocenecontinental temperatures. Palaeogeography, Palaeoclimatol-ogy, Palaeoecology 144:21–35.

Sloan, L. C., and D. K. Rea. 1995. Atmospheric carbon dioxideand early Eocene climate: a general circulation modeling sen-sitive study. Palaeogeography, Palaeoclimatology, Palaeoe-cology 119:275–292.

Sloan, L. C., J. C. Walker, and T. C. Moore. 1995. Possible role ofoceanic heat transport in early Eocene climate. Paleoceanog-raphy 10:347–356.

SPSS. 1999. Systat 9, statistics II. SPSS, Chicago.Taylor, L. R. 1978. Bates, Williams—Hutchinson—a variety of

diversities. Pp. 1–18 in L. A. Mound and N. Warloff, eds. Di-versity of insect faunas (Symposium of the Royal Entomolog-ical Society of London No. 9). Blackwell Scientific, Oxford.

Thomas, D. J., T. J. Bralower, and J. C. Zachos. 1999. New evi-dence for subtropical warming during the late Paleocene ther-mal maximum: stable isotope from Deep Sea Drilling ProjectSite 527, Walvis Ridge. Paleoceanography 14:561–570.

Traverse, A. 1988. Paleopalynology. Academic Press, New York.Van der Hammen, T., and H. Hooghiemstra. 2000. Neogene and

Quaternary history of vegetation, climate, and plant diversityin Amazonia. Quaternary Science Reviews 19:725–742.

Whittaker, R. H. 1972. Evolution and measurement of speciesdiversity. Taxon 21:213–251.

Wilf, P. 2000. Late Paleocene–early Eocene climate changes in


southwestern Wyoming: paleobotanical analysis. GeologicalSociety of America Bulletin 112:292–307.

Wing, S. L. 1998. Late Paleocene-early Eocene floral and climaticchange in the Bighorn Basin, Wyoming. Pp. 371–391 in W.Berggren, M. P. Aubry, and S. Lucas, eds. Late Paleocene–ear-ly Eocene biotic and climatic events. Columbia UniversityPress, New York.

Wing, S. L., and G. J. Harrington. 2001. Floral response to rapidwarming at the Paleocene/Eocene boundary and implicationsfor concurrent faunal changes. Paleobiology 27:539–563.

Wing, S. L., J. Alroy, and L. J. Hickey. 1995. Plant and mammaldiversity in the Paleocene to early Eocene of the Bighorn Ba-sin. Palaeogeography, Palaeoclimatology, Palaeoecology 115:117–155.

Wolfe, J. A. 1978. A paleobotanical interpretation of tertiary cli-mates in the Northern Hemisphere. American Scientist 66:694–703.

Wolfe, J. A., and R. Z. Poore. 1982. Tertiary marine and non-

marine climatic trends. Pp. 154–158 in National ResearchCouncil, eds. Climate in earth history. National AcademyPress, Washington, D.C.

Yeager, D. P., and G. R. Ultsch. 1989. Physiological regulationand conformation: a BASIC program for the determination ofcritical points. Physiological Zoology 62:888–907.

Zachos, J. C., K. C. Lohmann, J. C. Walker, and S. W. Wide. 1993.Abrupt climate change and transient climates during the Pa-leogene: a marine perspective. Journal of Geology 101:191–213.

Zachos, J. C., L. D. Stott, and K. C. Lohman. 1994. Evolution ofearly Cenozoic marine temperatures. Paleoceanography 9:353–387.

Zachos, J., M. Pagani, L. Sloan, E. Thomas, and K. Billups. 2001.Trends, rhythms, and aberrations in global climate 65 Ma topresent. Science 292:686–693.

Zar, J. H. 1999. Biostatistical analysis. Prentice-Hall, EnglewoodCliffs, N.J.

Response of tropical vegetation to Paleogene warming...Carlos A. Jaramillo Abstract.—The late Paleocene-early Eocene transition was characterized by a long period of global warming

Documents