Is Intelligent Life Inevitable? Paleobiological Perspective from the Neotropical Rainforest

Is intelligent life inevitable? Paleobiological perspective from the Neotropical rainforest

Andrés Cárdenas & Carlos Jaramillo Is there life outside Earth? Time (Ma)

4500 4000 3500 3000 2500 2000 1500 1000 500 *? earliest biogenic carbon * earliest bacteria * earliest photosynthetic bacteria * first green algae (eukaryotes) first metazoans * R EPORTS first skeletons * A B 10 A Barberton (3.48 Ga) 1

0.1

0.01 Crystalline Interiors 100 1E-3 Glassy Rims

10 B Ophiolites C D (1.95 - 0.1 Ga) 1

0.1

0.01

10 1E-3 C Modern Fig. 2. Tubular structures of inferred microbial origin in basaltic glass from the Troodos ophiolite 10 (<0.15 Ga) (A and B) and modern oceanic crust (C and Furnes et al. 2004D). (A) Photomicrograph of the glassy chilled margin of a pillow (sample CY-1-35). (B) Detail from (A) [indicated by arrow in (A)] showing tubular 1 structures in fresh glass. (C) Photomicrograph of a modern pillow margin (Ocean Drilling Program sample 148-896A, 11R-01, 73-75 cm) showing microbial alteration features of fresh basaltic glass 0.1 (light yellow) along fractures. (D) Detail from (C) [indicated by arrow in (C)] showing tubular structures of microbial origin protruding into the fresh glass. on January 2, 2007 0.01 Wt. % Carbonate Wt. % Carbonate Wt. % Carbonate 1E-3 Fig. 3. Backscatter electron (BSE) A B image (A) and x-ray element -30 -25 -20 -15 -10 -5 0 5 10 maps of carbon (B) and calcium 13 (C) associated with tubular δ CCarbonate structures from approximately Fig. 4. Relationship between weight % carbon- the same area shown in Fig. 1E. ate versus ␦13C for the originally glassy rims In (D), the carbon map has

(solid circles) and crystalline interiors (open www.sciencemag.org been superimposed on the BSE squares) of pillows. (A) Analyses from pillow image, showing the association lavas of the Komati, Hooggenoeg, and Krom- of carbon with the margins of berg Formations of the Onverwacht Group, the tubular features. An x-ray 20 um BGB. (B) Compilation of analyses from the element map of titanitum is Troodos ophiolite, Cyprus; the Mirdita ophio- shown in ﬁg. S1. C D lite, Albania; the Solund-Stavfjord ophiolite, Norway; and the Jormua ophiolite, Finland (21). (C) Compilation of analyses from modern oce-

anic crust (13, 14). Ga, billion years ago. Downloaded from

early microbial life and the preservation of associated biomarkers is not unexpected. Some of the deepest branches in the tree of life are populated by thermophilic microbes, and there is increasing evidence that early life may have been con- nected to volcanic environments, such as deep- sea hydrothermal vents (27). This is consistent the metamorphic overprint occurred penecon- the titaniferous tubular structures, which in some with an optimal growth temperature for thermo- temporaneously with igneous activity (15, 23). cases caused segmentation (Fig. 1F). These ob- philic microbes of 70°C, determined from the Oxygen isotope and metamorphic profiles servations imply that the delicate tubular struc- only study to investigate the depth distribution of across the BGB magmatic stratigraphy are tures existed before the precipitation of titanite microbial alteration textures in the modern indistinguishable from those of Phanerozoic and chlorite and are thus premetamorphic (Fig. oceans (11). Filamentous microfossils have been ophiolites and recent oceanic crust (25, 26). 1, C to F). The lack of later regional metamor- described from a 3235 million-year-old massive This implies that the metamorphism represents phic events that may have caused recrystalliza- sulfide deposit interpreted to have formed in ancient ocean floor hydrothermal alteration. tion has allowed preservation of this 3.48 billion- much the same way as modern black smokers The morphology of the delicate structures year-old biomarker. (28). Our study indicates that microbes colo- illustrated in Fig. 1, C to F, is inconsistent with Our data come from a geological setting that nized basaltic glass of the early oceanic crust, inorganic precipitation of titanite during seafloor has not been extensively explored in the search much in the same way as they do modern vol- metamorphism of the BGB pillow lavas. Fine- for early life on Earth. The suggestion of volca- canic glass. Well-preserved pillow lavas, which grained chlorite is observed to have overgrown nic rocks from the oceanic crust as a habitat for are a major component of Archean greenstone

580 23 APRIL 2004 VOL 304 SCIENCE www.sciencemag.org Downloaded from rstb.royalsocietypublishing.org on September 5, 2012

572 C. Stringer Modern human origins

mitochondrial divergence by an unknown amount of time. But they are consistent with fossil evidence of an effective separation date of the H. neanderthalensis and H. sapiens Neanderthal lineages at ca. 300 kyr and also with subsequent genetic divergence among recent humans beginning less than 200 kyr ago (Stringer 1998b). Both the morphological data and the limited amount of fossil DNA available suggest that Neanderthal–recent human differences were of the order of two or three times that found within recent humans. But even in this case, where genetic and morphological differences are clear, the data can be used to support a placing of Neanderthals and recent humans in modern either the same or different species, given the recency of common ancestry. There have also been recent claims for the recovery of ancient DNA from Australian fossils. Adcock et al. (2001) age (kyr) 600 250 175 0 reported that 10 out of 12 specimens tested from Willan- dra Lakes and Kow Swamp had yielded mitochondrial Figure 4. Schematic depiction of Neanderthal–modern sequences. One of these, from Mungo 3, was claimed to human mtDNA relationships, with approximate coalescence form an outgroup with a previously reported mitochon- dates (from data in Ovchinnikov et al. (2000) and Krings et drial nuclear insert, distinct from the other fossils and al. (2000)). The small sample of Neanderthal sequences is sufficient to suggest comparable diversityTaken and modified from Stringer 2002 to, but clear from recent human sequences. Adcock et al. (2001) distinctiveness from, recent humans. claimed, moreover, that the distinctiveness of the Mungo 3 sequence undermined genetic support for a recent African origin. In an accompanying commentary, Releth- ford (2001) used the results to support alternative multi- conclusions were extremely controversial, and were sub- regional interpretations, and to question previous jected to critical scrutiny concerning the samples, methods interpretations of Neanderthal DNA. However, Cooper et and calibration used (Templeton 1993). Although it is al. (2001) in turn criticized various aspects of the work. now evident that Cann et al. (1987) were premature in First, they observed that the claimed recovery rate for the the confidence with which they presented their results, Australian ancient DNA was exceptional compared with much more extensive analyses (e.g. Ingman et al. 2000) results from elsewhere, and that standard experimental have shown that they were fundamentally correct in protocols had not been employed, suggesting the possi- their conclusions. bility of contamination. Second, they reanalysed the data, In the past ten years, with the development and appli- using a larger number of recent Australian and African cation of PCR techniques, a wealth of sequenced data has sequences, and demonstrated that the Mungo 3 sequence been made available from autosomal (biparentally did not now form an outgroup to recent human mtDNA inherited) DNA, Y-chromosome DNA (inherited through in the most parsimonious phylogeny. Third, they observed males) and mitochondrial DNA (inherited through that even the original published phylogeny presented no females).These data have been used to compare the DNA serious challenge to Recent African Origin. Australian fos- of human populations in ever greater detail (Tishkoff et sils classed by multiregionalists as ‘robust’ and ‘gracile’, al. 2000; Kayser et al. 2001), to estimate coalescent (last purportedly derived from archaic Indonesian and Chinese common ancestral) dates for various gene systems ancestors respectively, grouped with the recent human (Ingman et al. 2000), to reconstruct ancient demographic sequences from regions such as Europe and Africa, while patterns (e.g. Rogers 2001), and to develop phylogeo- Mungo 3 was more closely related to all these than it was graphic studies to map ancient dispersal events (e.g. Rich- to the Neanderthal sequences used as an outgroup. ards & Macaulay 2000; Underhill et al. 2001). While most of these data support a recent African origin for recent 7. NEW APPROACHES TO MODERN HUMAN humans and their genetic diversity (e.g. Jorde et al. 2000; ORIGINS RESEARCH Ke et al. 2001), others may not (Zhao et al. 2000). Although the data are growing in power and resolution, In these concluding sections, I would like to draw analyses cannot yet resolve the precise time and place of together aspects of this review and also look at new our origins, nor establish whether there was only one or approaches to some remaining problems. In my opinion, perhaps several significant dispersals of H. sapiens from variants of one of the polar extremes in the debate about Africa during the later Pleistocene. modern human origins discussed at the beginning of this Some genetic data, in the form of mtDNA, are now paper—Multiregional Evolution—have been falsified, and available from Neanderthal fossils (Krings et al. 2000) and the fundamental mode of modern human origins can be these suggest a separation time of their lineage from that assumed to be that of a recent African origin. But until leading to recent humans of ca. 600 kyr (Krings et al. we have better records of late Pleistocene events in human 2000; Ovchinnikov et al. 2000; figure 4). As explained history from regions such as China and Australia, we will earlier, such estimates necessarily provide maximum ages continue to depend on genetic data to inform us whether a for evolutionary separation, since any population and strict Recent African Origin model is likely to be adequate, species separations would inevitably post-date the first rather than a variant incorporating a greater and more

Phil. Trans. R. Soc. Lond. B (2002) What is an ‘intelligent’ species?

Interstellar communication What is the likelihood of discovering life outside Earth?

What is the likelihood of discovering life on an Earth-like planet outside of the solar system?

What is the likelihood of discovering intelligent life outside of the solar system? The Drake Equation

N = Probability of finding intelligent life beyond our solar system

N = N* fp ne fl fi fc L

N = number of stars in the Milky way * fp = fraction of stars that have planets around them

ne = fraction of planets capable of sustaining life

fl = fraction of planets in ne where life evolves fi = fraction of planets where intelligent life evolves fc = fraction of fi that try to communicate L = length of time for which such civilizations release detectable signals into space Probability of life evolving = N* fp ne fl

200 billionX0.20X0.10X0.50 = 2 billion of planets with life in Milky Way Probability of finding intelligent life beyond our solar system N = (N* fp ne fl)fi fc L 2 billionX0.000000001X1X0.00001 = 0.00002 planets with intelligent life capable of communication and alive during our civilization Is intelligent life an inevitable outcome of any given evolutionary process? ...Almost any planet with life, in my view, will produce living creatures we would recognize as parallel in form and function to our own biota.

...But first, life must arise, and we have no idea how rare an event that might be. ...If complex consciousness has evolved but once...how can anyone defend the inevitability of its convergent evolution? What is a ‘Neotropical rainforest’?

High MAP > 1.8 m/yr High MAT > 18 ∘C Small seasonal variation T < 7 ∘C

Angiosperms > 90% (two leaf types) [SI Appendix (see Table S10)]. Angiosperms so far identified to extant taxa below the family level include the Neotropical aroid Montrichardia (18) and a member of the palm subtribe Euterpeinae (19). The four, rare ferns belong to dis- 58 Matinctive genera: the pantropical, floating-aquatic Salvinia; Stenochlaena,acommonclimbingferninOldWorldtropical A swamps; the cosmopolitan Lygodium;andAcrostichum,apan- tropical fern common in freshwater and mangrove swamps. To compare the familial/ordinal taxonomic composition of the Cerrejo´n megaflora with that of extant lowland Neotropical B CD E F forests, we selected 72 ofL Gentry’s 0.1 hectare transects (20) as acomparativemoderndataset[Fig.1andSI Appendix (see Table

S5)]. The small size of the modernFig. 2. Cerrejo´ plots n megafossils is identifiable a good to var- match for the ious higher taxa. (Scale bar: 5 cm, except where spatial scale of the parautochthonousnoted in parentheses after taxon.) Cerrejo (A) Coniferalesń fossil assem- (CJ60, 3 cm); (B) Stenochlaena (CJ81); (C and D) GHblagesI (21)J K and the minimumundetermined stem flowers diameter (CJ77, CJ78); (E of) fruit 2.5 of cm at 1.5 m Fabaceae (CJ100); (F) winged seed of Fabaceae above thePQ ground also simulates(CJ79); (G) Arecaceae,the dominance possibly Nypa (CJ68, ham- of woody plant mer Ϸ30 cm); (H) Zingiberales (CJ65, 10 cm); (I–K) remains in leaf litter and fossilleaves and leaf leaflets deposits of Fabaceae (CJ76, (22). CJ1, and We excluded CJ19); (L) Malvales with hole feeding (CJ11); (M) Gentry’s sites from higherLauraceae elevations (CJ22); (N) closeup and of M showing different lam- climatic inar resin glands (1 mm); (O) Malvaceae (CJ26); (P) Euphorbiaceae (CJ24); (Q) Sapotaceae with mar- zones. For each modern site,gin and we hole feeding tabulated (CJ8). See SI Appendix the(Ta- proportions of ble S10 and Figs. S2–S98) for descriptions and MN speciesO and individuals belongingphotographs of to all morphotypes. each family/order. For the fossils, weWing et al. 2009 tabulated the proportion of leaf morphotypes in each pairs of modern sites [SI Appendix (see Table S5)]. Calculating Table S5)]. Thus, in terms of the rank-order diversity and Spearman coefficients using rank orderfamily/order diversity (species or leaf forabundance the whole of plant flora families, and the Cerrejo theń numberflora is within of the leaves in each morphotypes per family) showed even higher congruence be- range of living Neotropical lowland forest plots. Legumes and tween Cerrejoń and the modern sites.family/order The mean Spearman forpalms random especially, samples the modern forest collected dominants in the at region two (14), sites, 0307 and coefficient of Cerrejoń with the modern sites wasSI 0.30 Appendix (best were already established as diverse and abundant families during guess) or 0.41 (conservative), whereas0705 the mean [ Spearman the late(see Paleocene. Tables The diversity S2, S5,and abundance and S10)]. of legume coefficient among the modern sites was 0.38We [SI Appendixquantified(see megafossils the similarities at Cerrejoń is particularly between intriguing, all because pairs the of living and Cerrejón legumes are among the earliest reliable records of the Fig. 1. Location of Cerrejo´ n fossil flora and comparative modern samples. fossil samples byfamily converting (23, 24) (all from proportional South America), which diversity is thought to and propor- (Upper) Cerrejo´ n fossil flora location (ૺ). (Lower) Comparative modernSpecies sam- per family have diversified rapidly in the Paleocene (25). tional abundanceWe values estimated the to temperature rank under orders which the and Cerrejoń calculating flora Spear- ples are indicated by: ᮀ, Quaternary pollen cores; I, modern leaf litter grew using the positive correlation observed in living floras man’s rank orderbetween correlation mean annual temperature coefficient (MAT) and [Fig. the proportion 3 and SI Appendix samples; F, modern forest transects. See SI Appendix (Tables S1–S4) for details of dicot species with untoothed leaves (P)(26,27).Thirty-fiveof (see Table S5)].45 We dicot leaf used types both from Cerrejo theń are conservative untoothed (P ϭ 0.78), and best guess on comparative samples. Base map of Cerrejo´ n region courtesy of National yielding estimates of MAT ranging from 24 °C (28) to 31 °C (29) Aeronautics and Space Administration Jet Propulsion Laboratory, CaliforniaFrequency taxonomic assignmentsdepending on the for regression the used.fossil The 95% flora. confidence In intervals living forest plots, 0600 of the MAT estimates are 2.1–2.7 °C using the method of Miller Institute of Technology. -0.5 0.0leaf 0.5 abundance1.0 et and al. (28). Like stem other wetland basal fossil area assemblages, per Cerrejo speciesń leaves are strongly Rank Order Correlation Coefficient likely overrepresent toothed leaf types in the regional vegetation

positively correlatedand therefore (22), underestimate so MAT we (30, compared 31). We thus favor the an rankECOLOGY order Individuals/leaves per family MAT estimate of Ն28 °C, which is consistent with oxygen abundance of leavesisotopic estimates (fossil of late samples) Paleocene tropical with sea stems surface tem- (living forests) perature from pristine foraminiferal tests (28 °C at 58.5 Ma, orders, the most diverse and abundant being: Araceae1000 (six to per family. The mean29.5 °C at 56.6Spearman Ma) (32), and correlation the MAT estimate of coefficient 32–33 °C between seven leaf types), Arecaceae (two leaf types, three fruit types), the Cerrejo´n andderived all from extant the exceptionally sites large-bodiedwas 0.20 fossil (best snakes recov- guess taxonomy)

400 ered from the Cerrejo´n Formation (33). After deposition of the Fabaceae (five to seven leaf types, seven fruit types), LauraceaeFrequency or 0.22 (conservativeCerrejo´n flora taxonomy). in the late Paleocene, These global and values tropical tem- did not differ

0 peratures rose several degrees into the early Eocene (32, 34), (two leaf types), Malvaceae (two to four leaf types), Menisper--1.0-0.5 0.0significantly 0.5 1.0 fromalmost the certainly mean subjecting Spearman Neotropical rainforests coefficient to MATs of 0.35 ob- Rank Order Correlation Coefficient higher than 30 °C. maceae [four leaf types (17), one fruit type], and Zingiberales tained by comparingWe estimated rank precipitation order abundancefor Cerrejoń using correlations of families of among all Fig. 3. Similarity of the Cerrejo´ n megaflora to living Neotropical forests. leaf size and precipitation observed in living vegetation (35, 36). Histograms are the frequency distributions of Spearman rank order correla- More than 50% of the dicot leaves in the Cerrejoń flora are tion coefficients derived from all pairwise comparisons of 72 living forest sites 2 [SI Appendix (see Table S5)]. Arrows show the mean similarity of the Cerrejoń mesophyll or larger in size (Ͼ1,900 mm ), which is striking given flora to all living sites (red, conservative identifications of fossils; green, the preservational biases against large leaves (37). We infer Table 1. Diversity and evenness of Cerrejón leaf andbest-guess pollen identifications). floras Black lines compared represent 95% confidence with intervals. Paleogenemean annual precipitation fossils and (MAP) Quaternary–Recent of 3,240 mm (range 2,260– tropical See SI Appendix (Table S5) for additional information. 4,640 mm), which is probably an underestimate (35). Surpris- samples Wing et al. PNAS ͉ November 3, 2009 ͉ vol. 106 ͉ no. 44 ͉ 18629 Leaves Pollen

Modern tropical litter Paleogene temperate sites Tropical Quaternary cores

S. S. S. N. Lagoa Rio W. WY WY WY WY Castle Palacio de das Cerrejo´n BCI Negro Bijou Ti Cf Eoc Eoc Rock Loros Cerrejón Patas Monica Piusbi

No. sites 16 3 7 4 13 10 5 22 5 2 74 49 17 57 No. specimens 2,191 672 1,048 1,548 8,695 4,216 3,295 6,633 1,015 2,298 10,925 18,750 5,777 19,006

Stotal 65 79 110 26 37 16 32 68 90 33 148 169 146 187

Mean S100 18 24 21 9 6 4 7 9 30 18 17 24 31 27

Mean Salpha 23 32 26 12 10 5 11 23 48 28 21 41 54 50

Sbeta 34 47 84 14 27 11 21 56 42 5 127 128 92 137

Proportion Salpha 0.40 0.40 0.23 0.47 0.26 0.32 0.34 0.18 0.46 0.85 0.14 0.24 0.37 0.27

Proportion Sbeta 0.60 0.60 0.77 0.53 0.74 0.68 0.66 0.82 0.54 0.15 0.86 0.76 0.63 0.73

Dtotal 0.94 0.95 0.95 0.85 0.81 0.55 0.80 0.87 0.95 0.90 0.85 0.92 0.88 0.88 Mean PIE 0.83 0.86 0.88 0.76 0.60 0.25 0.46 0.73 0.91 0.88 0.75 0.89 0.86 0.87

Mean Dalpha 0.82 0.85 0.85 0.76 0.60 0.27 0.45 0.73 0.91 0.88 0.74 0.88 0.86 0.86

Dbeta 0.13 0.09 0.10 0.09 0.21 0.28 0.34 0.14 0.04 0.02 0.11 0.04 0.02 0.02

Proportion Dalpha 0.87 0.90 0.89 0.90 0.74 0.48 0.57 0.84 0.96 0.98 0.87 0.96 0.98 0.98

Proportion Dbeta 0.13 0.10 0.11 0.10 0.26 0.52 0.43 0.16 0.04 0.02 0.13 0.04 0.02 0.02

Information on sites is given in SI Appendix (Tables S1–S4); details of analysis are given in SI Appendix (Table S6). Stotal, number of species in all samples; Salpha, number of species within samples; S100, number of species in sample rareﬁed to 100 specimens; Sbeta, among sample diversity calculated using formula of Lande (38); Dtotal, Simpson’s diversity index for all samples; Dalpha, Simpson’s diversity index within samples; Dbeta, among sample Simpson’s diversity index calculated using the formula of Lande (38); PIE, Hurlbert’s probablility of interspeciﬁc encounter.

18628 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0905130106 Wing et al. Carbonemys cofrinii

Titanoboa cerrejonensis TIME ‘Post-extinction interval’ ‘M. Extinction interval’

‘Background’ RESEARCH ARTICLES

Fig. 2. Changes in palynofloral diversity and composition during the early to middle Ce- nozoic (15). (A)Pollen and spore standing diversity calculated by using the range-through method (15)andelim- inating single-occurrence species. A Holocene com- posite palynological sample (15)of321species is drawn as a bench- mark. Notice that diversity steadily increases during the early Eocene and gradually decreases during the late middle Eocene to early Oligo- cene, with a large drop at the Eocene-Oligocene boundary. (B)Firstaxis of a detrended correspondence analysis (15) that explains 45.9% of the total variance in species composition along the stratigraphic profile. Paleocene palynofloras are showing three distinct palynofloras: Paleocene, Eocene, and Oligocene to clearly different from Eocene to Miocene palynofloras. Characteristic early Miocene. (D) Global oxygen isotope curve for the Cenozoic (27). on December 13, 2007 pollen species of each flora are shown: Proxapertites cursus for the Raw data were smoothed using a five-point running average. There is Paleocene, Nothofagidites huertasii and Echitriporites trianguliformis correspondence between the general trend of the diversity curve and the orbicularis for the Eocene, and Jandufouria seamrogiformis for the global oxygen isotope curve that is a proxy for average global tem- Oligocene to early Miocene. (C) Agglomerative cluster analysis (15), perature (27). Jaramillo et al. 2006

Origination and extinction rates. Rates of the early Eocene thermal maximum is par- proxies, based on plant stomatal indices, in- origination and extinction were calculated by alleled, although slightly offset, by an indicate that atmospheric CO2 concentrations using the per capita rates of Foote (15, 20)(Fig. crease in floral diversity. The subsequent long were near present-day levels during the Eocene www.sciencemag.org 3). The rate of extinction is stable through drop in temperature (27)betweenthelate (35). This conflicting evidence precludes a bet- time with an increase over background levels middle Eocene and the early Oligocene is ter understanding of the role of CO2 on the during the Eocene-Oligocene transition (Fig. also paralleled by a similar drop in diversity, pattern shown here. 3A). Although the rate of origination grad- with a larger drop in both temperature and Species-area effect. An alternative expla- ually decreased over time (Fig. 3B), an in- diversity at the Eocene-Oligocene boundary nation to climate change in the Neotropics crease over background levels occurred during (Fig. 2). This correspondence between diver- driving diversification and extinction is a the early Eocene (Fig. 3B). A high rate of sity patterns and global temperature sug- species-area effect. This idea has been pro- both origination and extinction is apparent at gests a causal relationship. However, climate posed before (36), but it has received little Downloaded from the late Paleocene (Fig. 3, A and B). In fact, change at the tropics per se may not explain attention (37). During the early and middle many of the species that originated at the be- the differences seen here, because there is no Eocene, there was a major global warming ginning of the late Paleocene became extinct strong evidence indicating that climate in event that allowed tropical lineages to ex- by the end of the Paleocene. There is also a the lowland tropics changed significantly pand well into the modern temperate areas major floral turnover at the end of the late during the Paleogene. The few available data (24, 25, 27). High-diversity forests existed in Paleocene (Fig. 2, B and C). This interval of Cenozoic climate in the lowlands of the the early Eocene of northern Patagonia needs further investigation to establish Neotropics indicate that temperature and pre- (12, 38), which was located near the southern whether this extinction was gradual, or if it, cipitation may have been similar to modern tip of the tropical belt during the Eocene in fact, dates from the short-lived Paleocene- values (28–32). However, these results are (12). This increase in the area with tropical- Eocene thermal maximum and represents a controversial, because the marine record seems like climate could be the main factor en- rapid turnover as seen in North American to show a warming of the Tropics (33). The hancing the increase in local diversity in mammals (26). floral record also mimics recently estimated at- the Neotropics during the Eocene. Larger

Standing diversity and global tempera- mospheric CO2 concentrations, which appear to regions can support more species, which ture. There is a correspondence between the be coupled with global temperature during the enhance both regional and local diversity global temperature curve for the Cenozoic Paleogene (34). A gradual decrease in the (2, 35, 39) by reducing the risk of extinction (27) and the diversity pattern shown here partial pressure of atmospheric carbon dioxide and increasing niche opportunities (2). In

[Fig. 2, A and D, first-differencing correla- (pCO2) from the middle Eocene to the late contrast, a cooling event in the late Eocene tion of diversity versus d18O, Spearman rho Oligocene was identified from stable carbon to early Oligocene reduced tropical areas of –0.508; P G 0.023 (15)]. The increasing isotope values of di-unsaturated alkenones drastically and, thus, drove local extinction temperature trend from the early Paleocene to from deep-sea cores (34). However, the pCO2 in the Neotropics. A recent analysis of biome

www.sciencemag.org SCIENCE VOL 311 31 MARCH 2006 1895 REPORTS

Riecito Mache A Gonzales B A E D m. ft. m. m. BCD

C. scabrata, R. guianensis

1300 26.2

M. vanwijhei M. Early Eocene

C. C. scabrata,R. irregularis triangulatus R.

Early Eocene PETM 18.1

PETM -250 1200 1600

56.09 +/- 0.03 Ma 12.5 B. annae -1000 -600 1200 800 R. magdalenensis R.

F. perforatus, B. annae B. perforatus, F. Late Paleocene Late -1400 -450 -350 Late Paleocene

1100 40 100 0 100 -29 -27 -25 -23 400 # of sp. # of sp. -29 -26 0.0 2.0 δ C(‰) δ13C(‰) DCA first axis -550 Fig. 3. Changes in palynofloral diversity and composition across the PETM -1800 10 40 at Riecito Mache and Gonzales. Lithological description of the section, key # of sp. 0 100 # of sp. biostratigraphic events, and a radiometric age of a volcanic tuff dated by the 0.0 2.0 U-Pb zircon radiometric method are shown to the left. (A) d13C values of DCA first axis 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% restricted taxa (red), whereas diversity of taxa that cross from the Paleocene on May 4, 2012 (Riecito Mache) and 34.6% (Gonzales) of the total variance in species into the PETM (green) does not change. Increase in diversity at the onset of composition along the stratigraphic profile. (C) Pollen and spore standing the PETM is due to a large addition of taxa to the preexisting Paleocene diversity using the range-through method and eliminating edge effect and flora. (E) Within-sample (rarefied) diversity at a count of 100 grains. The single-occurrence species. Note the increase in diversity at the onset of the PETM (mean = 18.1) is more diverse than the Paleocene (12.5) and less PETM. (D) PETM-restricted taxa (blue) are more diverse than Paleocene- diverse than the Eocene (26.2).

m. conditions for tropical ecosystems (8, 21), although References and Notes extinction 1. J. C. Zachos et al., Science 302,1551(2003); origination 31° to 34°C is still within the maximum tolerance

2150 10.1126/science.1090110. Early Eocene Early of leaf temperature of some tropical plants (24). 2. T. Westerhold, U. Röhl, H. K. McCarren, J. C. Zachos, www.sciencemag.org We recognize that further studies toward the Earth Planet. Sci. Lett. 287,412(2009). center of the South American continent need to 3. J. P. Kennett, L. D. Stott, Nature 353,225(1991). be performed in order to understand the effects of 4. M. Pagani, K. Caldeira, D. Archer, J. C. Zachos, Science 314,1556(2006). 2250 warming in more continental tropical settings. 5. R. E. Zeebe, J. C. Zachos, G. R. Dickens, Nat. Geosci. 2,

PETM However, at our sites in northern South America, 576 (2009). tropical forests were maintained during the warmth 6. S. L. Wing et al., Science 310,993(2005). L.Paleocene of the PETM (~31° to 34°C). Greenhouse exper- 7. W. C. Clyde, P. D. Gingerich, Geology 26,1011(1998). 8. M. Huber, Science 321,353(2008). iments have shown that high levels of CO2 together Downloaded from 2350 9. Materials and methods are available as supporting with high levels of soil moisture improve the per- 0.0 0.2 0.4 0.6 material on Science Online. formance of plants under high temperatures (25), 10. C. Jaramillo, M. J. Rueda, G. Mora, Science 311,1893(2006). per-capita rate and it is possible that higher Paleogene CO2 levels 11. S. L. Wing et al., Proc. Natl. Acad. Sci. U.S.A. 106, (26)contributedtotheirsuccess.Higherprecipita- 18627 (2009). Fig. 4. Per capita rates of origination and ex- 12. J. Muller, Bot. Rev. 47,1(1981). tinction per 200,000 years (the time span of the tion amounts could have been as important as high 13. S. R. Ramírez, B. Gravendeel, R. B. Singer, C. R. Marshall, PETM) in Mar 2X. The per capita extinction rate CO2.Precipitationreconstructionfromanearby N. E. Pierce, Nature 448,1042(2007). (red) shows a slight increase in extinction duringJaramillo et al. 2010Late Paleocene site, Cerrejon, indicate high pre- 14. E. Schuettpelz, K. M. Pryer, Proc. Natl. Acad. Sci. U.S.A. the PETM that is comparable to other intervals in cipitation regimes: about 3.2 m of rain per year 106,11200(2009). 15. S. W. Punyasena, G. Eshel, J. C. McElwain, J. Biogeogr. the Eocene. The per capita origination rate (black) (11). Our data, including a –35% shift in leaf-wax 35,117(2008). shows a spike in origination during the PETM. dDvalues,noincreaseinplantabundanceofdry 16. A. F. Diefendorf, K. E. Mueller, S. L. Wing, P. L. Koch, indicators (e.g., Poaceae), absence of a large plant K. H. Freeman, Proc. Natl. Acad. Sci. U.S.A. 107, turnover was the cause for the relatively 13C- extinction, high plant diversity, and high abun- 5738 (2010). 17. G. D. Farquhar, J. R. Ehleringer, K. T. Hubick, Annu. Rev. enriched CIE. More likely, a smaller CIE from dance of families typical of wet tropical rainforests Plant Physiol. Plant Mol. Biol. 40,503(1989). this study could simply be the result of incomplete (such as Annonaceae, Passifloraceae, Sapotaceae, 18. N. J. van der Merwe, E. Medina, J. Archaeol. Sci. 18, recovery of the PETM maxima. Araceae, and Arecaceae), suggest that precipita- 249 (1991). Today, most tropical rainforests are found at tion in the northern Neotropics during the PETM 19. N. Stoskopf, Understanding Crop Production (Prentice-Hall, Upper Saddle River, NJ, 1981). MAT below 27.5°C. Many have argued that trop- was either similar to Paleocene levels (3.2 m/year) 20. S. L. Bassow, K. D. McConnaughay, F. A. Bazzaz, Ecol. ical communities live neartheirclimaticoptimum or higher. Indeed, it is possible that rainforest fam- Appl. 4,593(1994). (19) and that higher temperatures could be ilies in general, which have been present in the 21. M. Huber, Nature 457,669(2009). deleterious to the health of tropical ecosystems Neotropics since the Paleocene (11), have the ge- 22. S. L. Lewis, Y. Malhi, O. L. Phillips, Philos. Trans. R. Soc. B 359,437(2004). (8, 19–23). Indeed, tropical warming during the netic variability to cope with high temperatures, 23. J. J. Tewksbury, R. B. Huey, C. A. Deutsch, Science 320, PETM is surmised to have produced intolerable CO2,andrainfall(25). 1296 (2008).

960 12 NOVEMBER 2010 VOL 330 SCIENCE www.sciencemag.org How temperature has influenced the long-term evolutionary process on the Neotropical rainforest? Species-area effect A. 100 G G G G G G G

G 96 Paleocene Eocene

ature C. r G 94 PALEOCENE EOCENE

10 N 10 N

92 Cerrejon GG Mar2x Regadera G Mar2x G G Regadera BA3 GGG BH15 90 G Medina BA3 GG Medina 5 S 5 S

Land area with tempe greater than 18 °C (%) 88

G 86

84 20 S 20 S 0 2 4 6 8 10 Increase in temperature, Paleocene gradient (°C) Salta G B. 100 35 S 35 S

95 Temperature °C PL G Bariloche G G Chubut G 35

ature G G LH r 50 S 30 50 S 90 G G 25 G 20 G Rio Turbio

G 15

85 G 65 S 65 S

G 80 W 70 W 60 W 50 W 40 W 30 W 20 W 10 W 80 W 70 W 60 W 50 W 40 W 30 W 20 W Land area with tempe greater than 18 °C (%) G 80

G 75 0 2 4 6 8 10 Increase in temperature, Modern gradient (°C) DownloadedEnergy-supply hypothesis from rspb.royalsocietypublishing.org on September 3, 2012

2198 T. J. Davies and others Environmental energy and evolutionary rates

0 species richness molecular evolution

0.19 energy 0.16

Figure 1. Relationships among species richness, substitution rates at third positions, and environmental energy. The black arrows indicate the direct relationships we found between energy and both species richness and molecular rates. Partial r2 for energy as the explanatory variable in the simplified models including area are shown (table 1), calculated as the proportional increase in residual deviance incurred through removing the respective terms from the model. The grey arrow indicates the relationship between species richness and molecular rates proposed by the faster evolution hypothesis but not supported by our analyses. The faster evolution pathway (dashed arrow) has zero weight, despite the strong relationship between energy and Taken and modified from Davies et al. 2004 molecular rates. and species richness is mediated by evolutionary rates. relationship between molecular rates and environmental When substitution rate and energy variables were added energy, the first step in this process. Until now a link simultaneously as explanatory variables for species between energy and molecular rates has not been demon- richness, substitution rate dropped out of the simplified strated in any group (Bromham & Cardillo 2003). The model (again yielding model 1; see table 1). This is the result reflects general rates of molecular evolution, in outcome predicted if the main effect of energy on species both nuclear and plastid regions and protein-coding and richness is direct, rather than via an intermediate effect on ribosomal genes. Because both synonymous and non- molecular rates (figure 1). The relationship between synonymous substitutions are affected, the underlying species richness and substitution rate appears to be an mechanisms must affect both neutral and functional artefact of both variables being correlated with energy changes. Shorter generation times and/or faster mutation measures. rates in high-energy environments would provide a simple, To check the generality and possible mechanism of these general explanation for the observed patterns (Rohde findings, we repeated all analyses for substitution rates 1992; Barraclough & Savolainen 2001). derived from the entire DNA data, second positions (sites The direct measures of environmental energy that best at which most substitutions lead to amino acid changes) explained mutation rates were UV and temperature. The and 18S rDNA in turn. Substitution rates of the different mutagenic effects of UV radiation are well documented DNA partitions were found to correlate with one another and have been implicated in influencing mutation rates and with species richness by Barraclough & Savolainen among lineages of marine protists (Pawlowski et al. 1997). (2001). In all cases, substitution rates correlated strongly Temperature might affect mutation rates, either through with measures of environmental energy (electronic decreasing development times or increasing metabolic rate Appendix A, table 4). In no cases did substitution rates and the production of DNA-damaging metabolites (Allen displace the energy variables as explanatory variables of et al. 2002). However, latitude was identified as the single species richness. The only partition retained in the models most important predictor and, therefore, the exact was 18S rDNA, but this displayed a negative relationship relationship between the different components of environ- with species richness (opposite to the predictions of the mental energy and molecular rates remains unclear. faster evolution hypothesis) and was not robust to the Further analysis at finer spatial and taxonomic scales might sensitivity analyses (see electronic Appendix A). provide greater resolution required to differentiate between 4. DISCUSSION alternate measures. Our analyses investigated the relationship between species Our results also confirm previous studies that species richness and energy in flowering plants and its possible richness correlates with molecular evolutionary rates in evolutionary causes. Area explained most variation in angiosperms, consistent with the second step of the faster species richness regardless of the other parameters within evolution theory. However, despite finding evidence for the models. Geographical range may put an upper limit on both faster molecular evolution and higher species richness diversification of a clade (Owens et al. 1999; Ricklefs in higher energy environments, there is no evidence that 2003), whereas a minimum range size may be required for faster evolution explains the relationship between energy speciation to occur (Rosenzweig 1992; Losos & Schluter and species richness. The main effects of environmental 2000). After controlling for area, all measures of environ- energy on both species richness and molecular rates are mental energy were strong predictors of species richness, direct (figure 1). The relationship between species richness providing broad support for species–energy theory and and molecular rates, needed for the faster evolution confirming recent macroecological studies at large spatial hypothesis, is lost or extremely weak when environmental scales (Francis & Currie 2003). We then performed a series variables are added to the model. of tests to distinguish the two main theories for explaining The faster evolution hypothesis relies on the effects of species–energy relationships: the faster evolution and environmental energy on generation time, mutation, or biomass theories. other factors affecting general evolutionary rates; therefore, The faster evolution hypothesis proposes that environ- we would have detected any strong effect with our mental energy speeds up evolutionary rates and thereby measures of molecular rates. That such an effect is lacking speeds up speciation rates. Our results demonstrate a leads us to reject the faster evolution hypothesis for flowering

Proc. R. Soc. Lond. B (2004) Species-area effect X

Energy-supply hypothesis X A

B CD E F L

Fig. 2. Cerrejo´ n megafossils identifiable to various higher taxa. (Scale bar: 5 cm, except where Biotic interactions noted in parentheses after taxon.) (A) Coniferales (CJ60, 3 cm); (B) Stenochlaena (CJ81); (C and D) hypothesis GHI J K undetermined flowers (CJ77, CJ78); (E) fruit of Fabaceae (CJ100); (F) winged seed of Fabaceae PQ(CJ79); (G) Arecaceae, possibly Nypa (CJ68, ham- mer Ϸ30 cm); (H) Zingiberales (CJ65, 10 cm); (I–K) leaves and leaflets of Fabaceae (CJ76, CJ1, and = CJ19); (L) Malvales with hole feeding (CJ11); (M) Lauraceae (CJ22); (N) closeup of M showing lam- inar resin glands (1 mm); (O) Malvaceae (CJ26); (P) Euphorbiaceae (CJ24); (Q) Sapotaceae with margin and hole feeding (CJ8). See SI Appendix (Ta- ble S10 and Figs. S2–S98) for descriptions and MN O photographs of all morphotypes. pairs of modern sites [SI Appendix (see Table S5)]. Calculating Table S5)]. Thus, in terms of the rank-order diversity and Spearman coefficients using rank order diversity (species or leaf abundance of plant families, the Cerrejoń flora is within the morphotypes per family) showed even higher congruence be- range of living Neotropical lowland forest plots. Legumes and tween Cerrejoń and the modern sites. The mean Spearman palms especially, the modern forest dominants in the region (14), coefficient of Cerrejoń with the modern sites was 0.30 (best were already established as diverse and abundant families during guess) or 0.41 (conservative), whereas the mean Spearman the late Paleocene. The diversity and abundance of legume coefficient among the modern sites was 0.38 [SI Appendix (see megafossils at Cerrejoń is particularly intriguing, because the Cerrejón legumes are among the earliest reliable records of the family (23, 24) (all from South America), which is thought to Species per family have diversified rapidly in the Paleocene (25). We estimated the temperature under which the Cerrejoń flora grew using the positive correlation observed in living floras between mean annual temperature (MAT) and the proportion of dicot species with untoothed leaves (P)(26,27).Thirty-fiveof 45 dicot leaf types from Cerrejoń are untoothed (P ϭ 0.78), yielding estimates of MAT ranging from 24 °C (28) to 31 °C (29) Frequency depending on the regression used. The 95% confidence intervals

0600 of the MAT estimates are 2.1–2.7 °C using the method of Miller -0.5 0.0 0.5 1.0 et al. (28). Like other wetland fossil assemblages, Cerrejo´n leaves Rank Order Correlation Coefficient likely overrepresent toothed leaf types in the regional vegetation

and therefore underestimate MAT (30, 31). We thus favor an ECOLOGY Individuals/leaves per family MAT estimate of Ն28 °C, which is consistent with oxygen isotopic estimates of late Paleocene tropical sea surface temperature from pristine foraminiferal tests (28 °C at 58.5 Ma, 1000 29.5 °C at 56.6 Ma) (32), and the MAT estimate of 32–33 °C derived from the exceptionally large-bodied fossil snakes recov-

400 ered from the Cerrejo´n Formation (33). After deposition of the Frequency Cerrejo´n flora in the late Paleocene, global and tropical tem-

0 peratures rose several degrees into the early Eocene (32, 34), -1.0-0.5 0.0 0.5 1.0 almost certainly subjecting Neotropical rainforests to MATs Rank Order Correlation Coefficient higher than 30 °C. We estimated precipitation for Cerrejoń using correlations of Fig. 3. Similarity of the Cerrejo´ n megaflora to living Neotropical forests. leaf size and precipitation observed in living vegetation (35, 36). Histograms are the frequency distributions of Spearman rank order correla- More than 50% of the dicot leaves in the Cerrejoń flora are tion coefficients derived from all pairwise comparisons of 72 living forest sites 2 [SI Appendix (see Table S5)]. Arrows show the mean similarity of the Cerrejoń mesophyll or larger in size (Ͼ1,900 mm ), which is striking given flora to all living sites (red, conservative identifications of fossils; green, the preservational biases against large leaves (37). We infer best-guess identifications). Black lines represent 95% confidence intervals. mean annual precipitation (MAP) of 3,240 mm (range 2,260– See SI Appendix (Table S5) for additional information. 4,640 mm), which is probably an underestimate (35). Surpris-

Wing et al. PNAS ͉ November 3, 2009 ͉ vol. 106 ͉ no. 44 ͉ 18629

+ + BI δT strength Is intelligent life an inevitable outcome of any given evolutionary process?