Effects of Rapid Global Warming at the Paleocene-Eocene Boundary on Neotropical Vegetation

See allHide authors and affiliations

Science  12 Nov 2010:
Vol. 330, Issue 6006, pp. 957-961
DOI: 10.1126/science.1193833


Temperatures in tropical regions are estimated to have increased by 3° to 5°C, compared with Late Paleocene values, during the Paleocene-Eocene Thermal Maximum (PETM, 56.3 million years ago) event. We investigated the tropical forest response to this rapid warming by evaluating the palynological record of three stratigraphic sections in eastern Colombia and western Venezuela. We observed a rapid and distinct increase in plant diversity and origination rates, with a set of new taxa, mostly angiosperms, added to the existing stock of low-diversity Paleocene flora. There is no evidence for enhanced aridity in the northern Neotropics. The tropical rainforest was able to persist under elevated temperatures and high levels of atmospheric carbon dioxide, in contrast to speculations that tropical ecosystems were severely compromised by heat stress.

The Late Paleocene-Eocene Thermal Maximum [PETM, 56.3 million years ago (Ma)], lasting only ~100,000 to 200,000 years, was one of the most abrupt global warming events of the past 65 million years (13). The PETM is associated with a large negative carbon isotope excursion recorded in carbonate and organic materials, reflecting a massive release of 13C-depleted carbon (4, 5), an ~5°C increase in mean global temperature in ~10,000 to 20,000 years (1), a rapid and transient northward migration of plants in North America (6), and a mammalian turnover in North America and Europe (7).

Efforts to understand the impact of climate change on terrestrial environments have focused on mid- to high-latitude localities, but little is known of tropical ecosystems during the PETM. Tropical temperature change is poorly constrained, but, given the magnitude of temperature change elsewhere, tropical ecosystems are thought to have suffered extensively because mean temperatures are surmised to have exceeded the ecosystems’ heat tolerance (8).

For this study, three tropical terrestrial PETM sites from Colombia and Venezuela were analyzed for pollen and spore content and stable carbon isotopic composition of organic materials (Fig. 1) (9). The analysis included 317 pollen samples, 1104 morphospecies, and 37,952 individual occurrences together with 489 carbon isotopes samples, 21 plant biomarker samples, and one radiometric age (9). Localities Mar 2X (core), Riecito Mache (outcrop), and Gonzales (outcrop and well) contain sediments that accumulated in fluvial to coastal plain settings. Biostratigraphy and carbon isotope stratigraphy, using bulk δ13C and plant wax n-alkanes, confirm the position of the PETM at the three sites (Figs. 2 and 3) (9), which is not associated with any formation boundary or major lithological or depositional environment change. The location of the PETM is further confirmed by a U-Pb radiometric dating of a felsic volcanic tuff found within the PETM in Riecito Mache section (Fig. 3) (9).

Fig. 1

Geographical location of the studied sections. Inset shows the Late Paleocene paleogeographic location of the studied sections [paleogeographic map after (27)]. Note that the northern Andes had not been uplifted yet, and most of Central America was still underwater.

Fig. 2

Changes in palynofloral diversity and composition across the PETM in the Mar 2X core. The lithological description of the core and key biostratigraphic events are shown to the left. (A) δ13C values of bulk organic matter across the PETM. Isotope data from the lowest part of the section was previously published (28). Blue line corresponds to a 20-point running mean of bulk data. (Inset) δ13C values of the C31 n-alkanes (n-C31) showing a pattern similar to those of n-C25, n-C27, and n-C29 (fig. S11). (B). First axis of a DCA, which explains 34.9% of the total variance in species composition along the stratigraphic profile. Paleocene palynofloras are distinct from Eocene palynofloras, with a gradual transition during the PETM. (C) Within-sample (rarefied) diversity at a count of 100 grains. The Paleocene (mean = 21.8) is less diverse than the PETM (30.3), and the PETM and Eocene (31.5) are very similar. (D) Pollen and spore standing diversity using the range-through method and eliminating edge effect and single-occurrence species. Note increase in diversity at the onset of the PETM. (E) Eocene-restricted taxa (blue) are more diverse than Paleocene-restricted taxa (red), whereas diversity of taxa that cross the Paleocene into the Eocene (green) does not change. Increase in diversity at the onset of the PETM is due to a large addition of taxa to the preexisting Paleocene flora. Family relative abundances are shown to the right of the figure. Families that are more abundant in the Paleocene are shown in purple; families that do not change in abundance across the Paleocene/Eocene transition, green; families that are more abundant in the Eocene, dark blue; and families that originate at the latest Paleocene, PETM, or Early Eocene, light blue. Images in the lower left correspond to distinct taxa of the sequence: Retidiporites magdalenensis (Proteaceae), Momipites africanus (Moraceae), Corsinipollenites undulatus (Onagraceae), and Rhoipites guianensis (Sterculioideae).

Fig. 3

Changes in palynofloral diversity and composition across the PETM at Riecito Mache and Gonzales. Lithological description of the section, key biostratigraphic events, and a radiometric age of a volcanic tuff dated by the U-Pb zircon radiometric method are shown to the left. (A) δ13C values of bulk organic matter across the PETM. Blue line corresponds to a 7-point running mean of bulk data. (B) First axis of a DCA, which explains 22.2% (Riecito Mache) and 34.6% (Gonzales) of the total variance in species composition along the stratigraphic profile. (C) Pollen and spore standing diversity using the range-through method and eliminating edge effect and single-occurrence species. Note the increase in diversity at the onset of the PETM. (D) PETM-restricted taxa (blue) are more diverse than Paleocene-restricted taxa (red), whereas diversity of taxa that cross from the Paleocene into the PETM (green) does not change. Increase in diversity at the onset of the PETM is due to a large addition of taxa to the preexisting Paleocene flora. (E) Within-sample (rarefied) diversity at a count of 100 grains. The PETM (mean = 18.1) is more diverse than the Paleocene (12.5) and less diverse than the Eocene (26.2).

Plant standing diversity, inferred from pollen assemblages at Mar 2X (Fig. 2D) and Riecito Mache (Fig. 3C), shows a low diversity during the Late Paleocene, followed by a significant increase during the PETM. Low-diversity Paleocene floras and an increase in Early Eocene diversity had been previously observed in tropical South America (10, 11), but the timing of the onset of the diversity changes was not resolved. Within-sample diversity in Mar 2X (9) shows a pattern similar to that for standing diversity. Paleocene samples (mean = 23.5) have fewer species than the PETM (mean = 35.9, P < 0.00449, df = 23.9), rarefied to a grain count of 150 (maintained even at a 100-grain count, Fig. 2C), whereas no significant difference is observed between the PETM and the Eocene (mean = 38.9, P < 0.33, df = 12.04). Similar diversity patterns characterize Gonzales (Fig. 3C), but within-sample diversity could not be calculated for Riecito Mache (9). The Simpson index, which measures the probability that two individuals taken at random will not belong to the same species, also reveals a similar pattern. In Mar 2X, the Simpson index is higher for the PETM (mean = 0.90) than for the Paleocene (mean = 0.76, P < 0.0008, df = 30.4), whereas the PETM and the Eocene (mean = 0.89) show no difference (P < 0.44, df = 54.8). A higher Simpson index for the PETM indicates that it has more species per sample than the Paleocene (as indicated by the higher within-sample diversity) and more evenly distributed species abundances. A similar pattern is recorded for Gonzales, but the Simpson index is slightly higher for the Eocene. In contrast, the Simpson index of the PETM at Riecito Mache was only slightly higher than the Paleocene, but the difference is not significant (9).

Taxa were divided into three groups, those for which extinction occurs in the Late Paleocene (P taxa), those whose origination occurs in the Early Eocene, including the PETM (E taxa), and those that are found in both the Paleocene and the Eocene (PE taxa). At the Mar 2X locality, standing diversity is lower for P taxa (13.6) than for E taxa (85.7; P < 0.001, df = 91.6) (Fig. 2E). Paleocene samples have a slightly higher standing diversity of PE taxa than Eocene samples (157.6 on the Paleocene side versus 135.8 on the Eocene side; P < 0.001, df = 58.542). When relative abundances are compared, the relation is similar, with P taxa less abundant in the Paleocene flora (4.6% of the assemblage) than E taxa in the Eocene flora (23.8% of the assemblage; P < 0.001, df = 107.9). The Paleocene flora at Mar 2X is composed largely of taxa that persisted into the Early Eocene, a pattern also evident in both Riecito Mache and Gonzales (Fig. 3D). Thus, our results demonstrate that the increase in tropical floral diversity during the PETM and Early Eocene resulted from the addition of a large set of new species, mostly angiosperms, to the existing, low-diversity floras that were holdovers from the Paleocene.

We assessed changes in floral composition by using a detrended correspondence analysis (DCA) and agglomerative clustering (9). The DCA shows a gradual change in flora during the PETM in the three studied sites (Figs. 2B and 3B) that is confirmed by the clustering analysis (9) (fig. S13), indicating that there was a significant change in the overall plant assemblage across the PETM. Even though we still do not know the affinities of ~85% of the flora, we were able to compare the remaining 15% (9). Diversity remained unchanged from the Paleocene into the Early Eocene for many families, including Polypodiaceae (ferns), Podocarpaceae (gymnosperms), Onagraceae, Ctenolophonaceae, Annonaceae, Moraceae, Rhizophoraceae, and Ulmaceae. However, diversity decreased in Proteaceae and increased in Arecaceae (palms), Bombacoideae, Fabaceae, Araceae, Poaceae, and Convolvulaceae. New families also appeared (i) during the uppermost Paleocene, including Myrtaceae, Sapotaceae, and Passifloraceae; (ii) within the PETM, including Sterculioideae, Euphorbiaceae, and Pellicieraceae; and (iii) during the Early Eocene, including Olacaceae and Ericaceae. Most of these originations, such as Sapotaceae, Passifloraceae, Ericaceae, Sterculioideae, Euphorbiaceae, and Pellicieraceae, represent the oldest pollen records for each family in the neotropics (12). Relative abundances show a similar pattern (Fig. 2).

The rate of extinction increased slightly during the PETM at a rate comparable to that during intervals within the Early Eocene (Fig. 4), with the gradual extinction of a small proportion (~5%) of Paleocene flora. However, extinction does not appear different from the background extinction rates of the entire sequence (Fig. 4). In contrast, per capita rate of origination increased significantly within the PETM interval (Fig. 4). Although the rates of origination continued to be high into the Early Eocene (10), the increase in rate began within the PETM.

Fig. 4

Per capita rates of origination and extinction per 200,000 years (the time span of the PETM) in Mar 2X. The per capita extinction rate (red) shows a slight increase in extinction during the PETM that is comparable to other intervals in the Eocene. The per capita origination rate (black) shows a spike in origination during the PETM.

Overall diversity and composition analysis suggest that the onset of the PETM is concomitant with an increase in diversity produced by the addition of many taxa (with some representing new families) to the stock of preexisting Paleocene taxa. This change in diversity was permanent and not transient, as documented for temperate North America (6). Interestingly, phylogenetic molecular studies of extant ephypitic ferns (which are mostly restricted to tropical rainforest canopies) and orchids indicate a major radiation at the onset of the Eocene (13, 14).

Temperature and precipitation are important factors affecting plant communities. Estimates of PETM mean annual temperature (MAT) from tropical latitudes are scarce. By using both published literature and TEX86 values from a nearby marine core [P2 core (Fig. 1 and fig. S14) (9)], we estimate that tropical temperatures during the PETM increased by ~3°C in the northern Neotropics and that mean temperatures were between 31° and 34°C (±2°C) during the peak of global warmth [see (9) for a detailed explanation]. To reconstruct regional hydrology, we classified plant families according to the rainfall preferences using Gentry’s neotropical plant data set (15) into dry- versus wet-preferred habitats (9) (table S9). Both Paleocene and Eocene are dominated by families indicating wet habitats, with no significant difference across the Paleocene-Eocene (Paleocene = 64%, Eocene = 61%, P < 0.49, df = 32.5), and low abundance of dry elements (e.g., Poaceae) that represents <2% of the assemblage (Paleocene = 0.7%, Eocene = 2%), all suggesting that no increase in aridity occurred across the PETM. This conclusion is further supported by stable carbon and hydrogen isotope (D/H) compositions of higher-plant–derived n-alkanes at site Mar 2X. A negative carbon isotope excursion (CIE) is evident in all n-alkanes measured and ranges between 2 and 3 per mil (‰), roughly matching bulk organic 13C/12C (δ13C) trends (Fig. 1 and fig S11) (9). At the onset of the PETM, n-alkane D/H compositions become more D-depleted (~35‰) and do not return to pre-PETM values within the measured section (9) (fig. S12). Changes in the D/H composition of plant lipids are related to changes in biosynthetic fractionation (related to a plant’s specific water-use efficiency), meteoric water D/H composition, amount of precipitation, and moisture recycling in forest ecosystems. Given the setting of these sites and the observed increase in tropical rainforest diversity, the negative shift in the hydrogen isotope value (δD) of n-alkanes likely reflects the predominance of D-depleted precipitation related to increased precipitation during rainout events (e.g., amount effects).

If we assume that our compound-specific δ13C record captures the maximum extent of the CIE within the body of the PETM, a much smaller CIE is observed at Mar 2X relative to the ~4.6‰ excursion recorded globally (16). In addition to the atmospheric carbon isotopic composition, environmental parameters in tropical rainforests that influence n-alkane δ13C values include water stress (17), ecosystem changes (related to changes in net isotope fractionation) (16), and canopy effects (18). It is possible that the smaller CIE recorded in this study resulted from ecosystem turnover and the prevalence of flora with higher water-use efficiencies. Alternatively, temporal changes in the carbon isotopic composition could be attributable to changes in mean annual precipitation (MAP), given that relations between total carbon isotope fractionation during carbon fixation and MAP have been observed today (16). A smaller CIE could imply that MAP decreased in the tropics, an interpretation unsupported by floral assemblages or by the δD record, which displays a negative shift (~–35‰) during the onset of the PETM. However, we cannot exclude the possibility that changes in moisture recycling also influenced the D/H record. Indeed, it is difficult to explain the totality of the carbon isotope record in light of our understanding of the PETM elsewhere. For example, carbon isotope compositions return to pre-event values by the end of the PETM, albeit slightly more negative as observed globally. However, the recovery phase at our localities is not accompanied by changes in vegetation, inconsistent with our assumption that ecosystem turnover was the cause for the relatively 13C-enriched CIE. More likely, a smaller CIE from this study could simply be the result of incomplete recovery of the PETM maxima.

Today, most tropical rainforests are found at MAT below 27.5°C. Many have argued that tropical communities live near their climatic optimum (19) and that higher temperatures could be deleterious to the health of tropical ecosystems (8, 1923). Indeed, tropical warming during the PETM is surmised to have produced intolerable conditions for tropical ecosystems (8, 21), although 31° to 34°C is still within the maximum tolerance of leaf temperature of some tropical plants (24). We recognize that further studies toward the center of the South American continent need to be performed in order to understand the effects of warming in more continental tropical settings. However, at our sites in northern South America, tropical forests were maintained during the warmth of the PETM (~31° to 34°C). Greenhouse experiments have shown that high levels of CO2 together with high levels of soil moisture improve the performance of plants under high temperatures (25), and it is possible that higher Paleogene CO2 levels (26) contributed to their success. Higher precipitation amounts could have been as important as high CO2. Precipitation reconstruction from a nearby Late Paleocene site, Cerrejon, indicate high precipitation regimes: about 3.2 m of rain per year (11). Our data, including a –35% shift in leaf-wax δD values, no increase in plant abundance of dry indicators (e.g., Poaceae), absence of a large plant extinction, high plant diversity, and high abundance of families typical of wet tropical rainforests (such as Annonaceae, Passifloraceae, Sapotaceae, Araceae, and Arecaceae), suggest that precipitation in the northern Neotropics during the PETM was either similar to Paleocene levels (3.2 m/year) or higher. Indeed, it is possible that rainforest families in general, which have been present in the Neotropics since the Paleocene (11), have the genetic variability to cope with high temperatures, CO2, and rainfall (25).

Supporting Online Material

Material and Methods

Figs. S1 to S14

Tables S1 to S10

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. Supported by Banco de la República, National Geographic, Smithsonian Women’s Club, Instituto de Colombiano de Petroleo (ICP)–Ecopetrol SA, and Smithsonian. M.P. was supported by NSF EAR-0628358 and ATM-0902882. S.S. was supported by a VISI grant from Netherlands Organisation for Scientific Research. Thanks to PDVSA for access to Mar 2X and the Instituto de Patrimonío Cultural of Venezuela for allowing us to sample in Riecito Mache. Thanks to Agencia Nacional de Hidrocarburos, A. Pardo, J. Sanchez, C. Guerrero, M. Carvalho, and the Colombian Armed Forces for logistic support. Thanks to the biostratigraphic team at ICP. Patrice Brenac did preliminary palynological analysis of Gonzales. N. Atkins provided editing support. A. Mets (NIOZ) is thanked for analytical assistance. K. Winter, H. Muller-Landau, J. Wright, and S. Punyasena and three anonymous reviewers provided comments on the manuscript. Special thanks to M. I. Barreto for continuous support and sources of ideas.
View Abstract

Stay Connected to Science

Navigate This Article