Did an impact alone kill the dinosaurs?

Transcription

1 Did an impact alone kill the dinosaurs? Shin Yabushita and Anthony Allen ask if the dinosaurs and many of the world s other inhabitants died out as a result of an impact alone, or whether interstellar events played a part in their demise. The theory that the Earth has been bombarded by small bodies in the solar system with catastrophic consequences has only come to be accepted during the past 15 years or so. However, the subject has a long history going back to Laplace. If we restrict ourselves to the 20th century, the possible collision of comets with the Earth was only seriously taken up by the book of Russel et al. (1945). The authors devoted a chapter to cometary collisions and calculated that a collision would occur once in 80 million years. They argued that nothing serious would happen to the Earth because the model of comets that they adopted was a swarm of tiny particles that would not reach the ground, even if they entered the atmosphere. Since then, of course, the icy nucleus model has been proposed and the situation has changed drastically. The late Harold Urey was interested in cometary collisions as early as When one of us (SY) visited him in California he expressed his belief that the Tsunguska event was due to a cometary collision. To the question of how tektites could be formed, his answer was that they were abrasive products. In a well known paper (Urey 1973) he put forward the idea that geological boundaries could be due to cometary impacts. He calculated the amount of energy released by the impact of a comet having the orbit of Halley and a radius of 10 km, and compared it with energies of earthquakes, heat received by the earth etc, and how hot the atmosphere would be if the energy were consumed through heating alone. He postulated that several of the geological boundaries that coincide with the ages of large craters were caused by cometary collisions. Alvarez et al. (1980) detected high concentrations of iridium in the Cretaceous/Tertiary boundary clay at Gubbio, Italy, and consequently the catastrophic roles played by minor bodies in the solar system came to be widely accepted. There are, however, certain aspects in the geological and palaeontological records that appear difficult to reconcile with the simple impact model. In this article we review some aspects of the geological records that have surfaced since the original paper of Alvarez et al. was published, and try to scrutinize the impact model. We also wish to argue that our giant molecular cloud encounter model (Yabushita and Allen 1983, 1989) may yet prove to be more viable than the simple impact model for the Cretaceous/ Tertiary boundary. Iridium enrichment In this section we review geological records relating to Ir. This is the element that first pointed to an extra-terrestrial cause of the Cretaceous/Tertiary boundary event. Elevated Ir at extinction boundaries reported by various authors are summarized in Rampino and Haggerty (1994). These values are reproduced in table 1. It may be noted that the worldwide Ir enrichment is seen only at the K/T boundary; other Ir enrichment sites are local, although at the Permian/Triassic boundary Ir is widespread. This suggests that the K/T boundary is exceptional and has led some authors to suggest that the K/T Ir may be of different origin from other Ir enrichment. The uniqueness of the K/T iridium is also supported by the Ir density profile of the core of the Pacific Ocean north to the Hawaiian Islands (Kyte and Wasson 1986). Although a high Ir enrichment is seen at the K/T boundary, no such enrichment is seen at other geological boundaries. From Kyte and Wasson s work, the excess Ir at the K/T boundary is 70 ng cm 2. If this is typical, the amount of Ir deposited all over the Earth would be g. This is the figure we will adopt in later discussions on Ir deposition at the K/T boundary. Further, as argued by Officer et al. (1987), the Ir enrichment is unlikely to have been instantaneous. Boundary craters About 150 craters are known with ages and diameters. Of these, three probably correspond to the K/T boundary event. These are given in table 2. Of these, Chicxulub is commonly regarded as the impact crater that caused the K/T event. The impactors are most likely to have been comets similar to the fragments of comet Shoemaker-Levy 9 for two reasons. One is that they were probably not independent. The other is that amino acids of extra-terrestrial origin have been detected (Zhao and Bada 1989). Astronomical catastrophe models of the Cretaceous/Tertiary boundary event are reviewed in relation to geological records. Particular attention is paid to the consistency of the simple impact model with iridium enrichment, amino acids and the oxygen pressure decrease, as evidenced by the air contained in ancient amber. It is argued that the uniqueness of the iridium enrichment at the Cretaceous/Tertiary boundary and the decrease in oxygen pressure in the atmosphere preclude the simple impact model, but favour the giant molecular cloud encounter model that is consistent with the two. Finally, circumstantial evidence supporting the interstellar modulation of extinction events is considered. 15

2 These are AIB and ISOVAL, which are extremely rare on Earth. Accepting that the impactors were comets, one can calculate the diameter of the impactor responsible for the Chicxulub crater. One finds that d = 11.0 km for the measured crater diameter D = 180 km, velocity of impact V i = 28.9 km s 1 and density ρ i = 1.0 g cm 3 (Yabushita 1995). Alvarez et al. (1980) assumed that only 22% of the impactor mass remained on the Earth, the rest being lost to space by the explosion. From these figures the total mass that remained on Earth would have been g. We can estimate the likely amount of Ir contained in the cometary material retained by the Earth. Geiss (1988) found that Si abundance in Halley dust particles is 3.6%, while there is 10% Si in CI carbonaceous chondrites. It follows that the fraction of chondritic material in comets is 36%. The Ir content of CI chondrite is 592 ng/g or g per gram of CI chondrite. Thus the amount of Ir contained in a comet of mass g is g. This may be compared with g which is the excess Ir at the K/T boundary. This crude estimate gives a discrepancy of an order of magnitude between the deposited Ir and the value inferred from the impact model. Taken at face value this implies an extra source of Ir enrichment at the K/T boundary, but note that there are uncertainties in these estimates. Atmospheric oxygen at the boundary Rigby and Landis (1994) and Landis et al. (1994) propose the Pele hypothesis, which asserts that varying O 2 content in the atmosphere played a dominant role in biological evolution. Their hypothesis is based on the measurement of O 2 content in ancient ambers collected at Hell Creek and Tullock, central Montana, US. By crushing ambers collected from different sections of the geological site, they measured how the O 2 content varied from the Cretaceous to the Tertiary periods over the K/T boundary. It was argued previously (Hophenberg et al. 1988) that O 2 molecules may diffuse through amber and that air bubbles in the amber may not be pristine, but, by measuring the gas diffusivity of argon in amber, Landis and Snee (1991) concluded that the gas is genuine and represents the atmosphere of the time when the amber was formed. The results obtained by Landis et al. (1994) are reproduced here (figure 2). There are fluctuations in the O 2 content, but what is striking is its rapid decline over the K/T boundary from some 30% to 20%. Landis et al. (1994) ascribed the variation of the O 2 content to changes in the rate of degassing of the mantle and other Earth and biological processes, which are the results of superplumes (arising from instability at the core mantle boundary). In view of the shortness of the time interval where the O 2 content varied and its coincidence with the K/T boundary, where large craters, extra-terrestrial amino acids and Ir enrichment are found, it does not seem reasonable to decouple the O 2 variation from extraterrestrial phenomena. It seems more reasonable to associate the variation of the O 2 content with a process that appears likely to have occurred based on our knowledge of the solar system environment. It is amusing to note that oxygen depletion of the detected order was independently predicted for the K/T boundary (Yabushita and Allen 1989), based on encounters with interstellar molecular clouds in the galactic disc. Mass extinction at the K/T boundary The K/T boundary is one of the greatest extinction episodes in the biological history of the Earth. Some 66% of marine species disappeared at the end of the Cretaceous period (Rampino and Haggerty 1994). What is hidden by the simple statistics is that the change in fauna has specific features: larger insects disappeared while their smaller counterparts survived; large reptiles (dinosaurs) disappeared but small ones (crocodiles and lizards) survived; large marine creatures such as ammonites disappeared while smaller ones survived; mammals survived and evolved greatly afterwards. It is seen that the extinction at the K/T boundary was not uniform but selective. That the extinction was selective has also been stressed by Officer et al. (1988). It may be remarked that insects do not have lungs; they breath through arrays of spiracles so that they are probably more susceptible to changes in oxygen pressure in the atmosphere. Reptiles have lungs that are less developed than those of mammals, so the same applies to them. No experimental result exists which supports the argument. However, in this respect we note that Rigby and Landis (1994) refer to the dinosaur Apatosaurus which had an elongated neck with considerable dead air space, and extremely small nostrils for its size. Hengst et al. (1993) estimated tidal (respired or intake) volumes at slightly over 100 l. With a dead space conservatively estimated at 30 l, ventilation was probably by rib movements alone, with maximal capability approximately four times that of tidal volume, so that nearly 30 breaths were needed to refresh the lung. Thus the animal would have been particularly disadvantaged in an O 2 -poor atmosphere. Similarly, large marine creatures such as ammonites would have required high concentrations of CaCO 3 in seawater, which in turn would have required a higher O 2 pressure in the atmosphere than exists today. We also note that the disappearance of dinosaurs may not have been instantaneous, as predicted by the impact, cold winter, loss of food chain scenario. According to Sloan et al. (1986), of 12 genera of dinosaurs alive in Montana, Alberta and Wyoming just before the K/T boundary, between 7 and 11 survived into the Palaeocene. Again, dinosaur fossils are found 1.3 m above the Ir-rich layer. Using the sedimentation rate at the site of the fossils (Sloan et al.), this would correspond to years after the impact. In this way the change and disappearance of fauna over the K/T boundary strongly indicates that it was not instantaneous, and the qualitative character of the extinctions suggests that it was associated with the decrease in the O 2 content of the atmosphere. Giant molecular cloud model The giant molecular cloud (GMC) encounter model (Yabushita and Allen 1983, 1989) for the K/T boundary event assumes that the Sun penetrated the core region of a GMC with an H 2 density of about cm 3. The total excess Ir deposited during such an encounter matches the total estimated above for K/T sediments. At these densities the hydrogen gas remains neutral as it accretes onto the Sun down to distances of less than 1 au. According to the model, as gas molecules (predominantly H 2 ) accrete onto the Sun, part of the gas is accreted onto the Earth as well. Since the cloud contains dust grains, these are injected into the atmosphere. Hydrogen molecules react with atmospheric O 2 and are precipitated out as water, removing O 2 from the carbon oxygen cycle. The model thus predicts a decrease in the solar radiation reaching the ground because of grains suspended in the atmosphere and a decrease in atmospheric O 2. In this model the amount of accreted H 2 can be inferred from the amount of Ir deposited at the K/T boundary, assuming that GMC dust grains have compositions similar to carbonaceous chondrites. The accreted H 2 would be such as to decrease the atmospheric O 2 by 20%. Because the Oort cloud of comets (Bailey et al. 1990) is disturbed by a GMC encounter, it is reasonable to assume that comets are injected into the inner solar system, some of which impact the Earth. The K/T event, according to the GMC encounter model, was brought about primarily by GMC penetration rather than these associated impacts. Conversely, if one associates the special case of the K/T boundary with the direct penetration of a GMC, then the other extinction boundaries also result from comet impacts. These may arise, as Bailey et al. suggested, from perturbations of the Oort 16

3 Interstellar gas in the Lagoon Nebula appears to twist and swirl in this false-colour image in which red represents S +, blue O 2+ and green H +. (A Caulet [ST-ECF] and NASA.) cloud by close GMC encounters, as well as from the mean galactic field. GMC versus giant comet models If the simple, isolated impact of a large asteroid does not seem to be a viable model for the K/T boundary event, one is led to consider the giant comet model of Clube and Napier (1986). In a series of papers, Clube and Napier argued that large comets which disintegrate into fragments are the cause of catastrophes in the Earth s history. Large comets are now observed (such as Chiron, say) and, since comets are seen to fragment as comet Shoemaker-Levy 9 did, the Clube Napier model has gained an observational basis. This model for K/T catastrophe proposes that these fragments injected a large amount of dust grains into the Earth s atmosphere so that the terminal event was not instantaneous; the timescale would be between 10 4 and 10 6 years. Because there are three craters with ages close to that of the K/T boundary, the Clube Napier model is consistent in this respect. The model does not, however, account for O 2 depletion, and may therefore not work for the K/T boundary, at least. Furthermore, as seen above, the Ir signature at K/T is exceptional among all of the geological boundaries. The K/T Ir enrichment is worldwide and is not consistent with the single boundary crater. Estimates imply an order of magnitude more dust accretion than expected from the Chicxulub impactor. Accepting this would require the accretion of dust grains released by the disintegrating comet or many smaller associated impactors. This is reasonable and expected in the Clube Napier model, but then one must ask why the K/T boundary is exceptional compared with other boundaries, if they are similarly attributed to disintegrating comets? Likelihood of a GMC encounter For the GMC model to be viable it is important to know whether it was likely that the Sun encountered a GMC 65 million years ago. The solar motion within the galaxy can be approximated as a near-circular orbit in the disc with 17

4 a period of 200 million years, with small superposed oscillations normal to the disc plane. The oscillatory motion below and above the galactic mid-plane has a period close to 60 million years, so it crosses the mid-plane roughly every 30 million years. Periodicities in terrestrial records suggest a half period nearer to 27 million years according to Clube and Napier (1996). For comparison with periodicity in extinction events one requires a measure of the collision rate as a function of time. Encounters result from the differential motion of the molecular clouds and the Sun, and, neglecting the eccentricity of the orbit in the plane (which results in small radial and azimuthal velocity dispersions), the motion normal to the plane dominates the dynamics of encounters. The collision rate per unit time is proportional to the local density ρ(z), and the normal velocity v(z), where z is the distance above or below the midplane of the galaxy. The Sun s excursion above the galactic mid-plane is small compared with the scale height of the disc mass, so the disc may be approximated by a uniform slab. The z motion is then simply that of a harmonic oscillator with displacement z = z max sin(kt) and velocity v(z) cos(kt), taking time to start at a mid-plane crossing. The density of molecular clouds as a function of z is normally taken to be an exponential with scale height h, so that the rate of encounters between gas clouds and the solar system R(t) is given by R(t) ρ 0 exp zmax sin(kt) /h (v 0 cos(kt)+ v GMC ) where v 0 is the vertical component of the Sun s velocity at the galactic mid-plane and v GMC is the local velocity dispersion for giant molecular clouds. The best estimate for the constant k, from Clube and Napier (1996), is k =(π/27.5) per million years. The amplitude of the solar z-motion is Table 1: Values and location of iridium enrichment at selected boundaries in the geological record z max 80pc, and the scale height for GMCs is 65 85pc from CO emission (Thaddeus 1986). The encounter probability is thus weakly modulated by the Sun s position above the mid-plane, the dominant contribution resulting from the relative velocities of clouds and the solar system. Near the mid-plane the encounter likelihood is dominated by the Sun s motion normal to the plane, while at high z the encounters result entirely from relative motions of GMCs. Adopting the best-fit period from terrestrial records would place the Sun s position 65 million years ago at its maximum elevation above the disc mid-plane, while adopting the longest period consistent with the data would place it The best fit would place the Sun at its maximum elevation above the disc mid-plane 65 million years ago. near the mid-plane. This uncertainty affects both GMC encounters and cometary impact models, and is alleviated by a suggestion that the amplitude may in the past have been greater and at different phase (Bailey et al. 1990), although it does appear to meet some difficulties (Wolfendale and Wilkinson 1989). There is also reason to think that really large GMCs may have scale heights less than for ordinary GMCs. According to Scoville and Sanders (1986), the mass distribution of GMCs is given by N(m) m 1.58, where N(m) gives the number of GMCs with mass between m and m + dm. On the other hand, if particles grow by sticking together, the resulting distribution would be N(m) m 1.65 and m 1.33 for the case where the total number of unit (original) mass particles present remains constant and where 1 the total mass of the system remains constant, respectively (Daniels and Hughes 1981). That the power index of GMC mass distribution lies between the two theoretical values indicates that large GMCs form by the merger of smaller clouds. Since the collision of GMCs is inelastic (energy lost to internal heating), the kinetic energy of a GMC per unit mass is constantly lost. This results in large GMCs being confined to the bottom of the gravitational potential, which is the mid-plane. This argument follows from the mechanism for cloud growth and does not depend on dynamical friction. There clearly remains considerable uncertainty in both the amplitude of modulation of cloud encounter rates and the phase of this modulation relative to terrestrial records. Adopting the worst case for the GMC encounter models, the scale height for the molecular clouds is comparable to the solar excursion distance above and below the mid-plane, thus v 0 v GMC. Local cloud densities contribute a factor of order 1/e between the midplane and maximum excursion, while velocity effects add another factor of order 2. The amplitude of modulation in the encounter probability is therefore of order 5 or more. Other evidence Lastly, we wish to refer to a geological finding that may be taken as circumstantial evidence of an interstellar connection for extinction events. Miono (1995) and Miono et al. (1993) found and collected microspherules from the bedded cherts of Kiso river, Aichi, and Sasayama, in Hyogo, Japan. These spherules could be collected by magnets. The spherules are hollow (figure 1) and are µm in diameter. Inside, one can see needle-like structures. Elemental abundance analyses show that they are similar to magnetic components of chondrites (figure 3). These spherules are abundant at the Palaeo- Geological boundary Elevated Ir (ppt) Site Pliocene 5000 southern Pacific middle Miocene 152 southern Pacific late Eocene 4000 widespread Cretaceous/Tertiary 1000s worldwide Cenomanian/Turconian 560 western US, Colombia, Europe Jurassic/Cretaceous 7800 central Siberia Callovian/Oxfordian 1000 Spain, Poland Triassic/Jurassic 400 Europe Permian/Triassic 600 various parts of China, Pakistan, Italy, Austria and India Mississippian/Pennsylvanian 380 Texas, US Frasnian/Famennian 230 China 300 Austria Europe Precambrian/Cambrian China 1 SEM image of a microspherule collected from Sasayama, Kyoto, Japan. It is hollow and contains needle-like structures. (Photograph provided by S Miono.) 18

5 zoic/mesozoic boundary (Permian/Triassic boundary), where one of the major mass extinctions took place. Another bedded chert at Guangxi, China, is associated with the P/T boundary. Unfortunately, no bedded cherts corresponding to the K/T boundary are known. That the P/T boundary has some similarities to the K/T boundary is indicated by the large number of sites where Ir enrichment has been found (table 1). We may note that next to the K/T boundary, the P/T boundary Ir is widely distributed. In terms of mass extinction, nearly 90% of fauna disappeared at this boundary, so it stands out along with the K/T event as exceptional. It was noted earlier that the encounter frequency for the observed GMC number density is lower than the frequency of galactic midplane crossings. If both the P/T and K/T boundaries are associated with encounters with interstellar dust and gas clouds, then a population of dense clouds with small scale height, or a local enhancement of dense clouds, such as might be associated with the back of a spiral arm density wave, needs to be postulated. This latter possibility is particularly relevant in assessing the origin of possible interstellar debris at the P/T boundary. Direct penetration of the core of a GMC is not postulated for the P/T boundary, and O 2 depletion has not been associated with this boundary. Nevertheless, material that possibly has interstellar origin provides circumstantial evidence for the GMC encounter hypothesis. Particles collected from the Antarctic that probably represent interplanetary material are not hollow (Zbik and Gostin 1995) and are somewhat larger (100 µm in diameter). Miono and his colleagues believe that these objects are of extra-solar system origin and probably originate from GMCs. Isotope analyses of carbon and oxygen by a method such as secondary ion mass spectrometry (SIMS) might provide a clue 2 3 K/T 40 boundary atmospheric oxygen time (million years ago) to the origin of the microspherules. However, isotope ratio measurements of carbon and oxygen are unlikely to yield original isotope abundances. These particles must have been heated during the atmosphere penetration, so their C or O composition may not be pristine. Discussion One of the objectives of this article has been to discuss the consistency or otherwise between the geological records relating to the K/T boundary event and the impact model, currently favoured by planetary scientists and many geologists. We have argued that the K/T boundary is unique among geological boundaries and that there are inconsistencies between the simple impact model and the records. The problem is a difficult one, because some of the results that are considered well established in one field may not be so for those working in another field. Nevertheless, we believe that the GMC model is consistent with the geological records so far known, and yet it is a process that is well established in astronomy and one that must have occurred in the past the Sun cannot avoid entering a GMC. What would happen if the Sun penetrated a GMC is a problem that deserves discussion, and one that could be clarified theoretically. Ir enrichment, a decrease in the atmospheric O 2 pressure, and the selectivity of the mass extinctions at the K/T boundary are all consistent with this model. The modulation of biodiversity may be associated with solar system excursions normal to the galactic plane, where extinction events correlate with galactic mid-plane crossing. Nevertheless, it is premature to associate extinction events with impacts alone and associated encounters with gaseous galactic debris during galactic mid-plane crossings agree well with geological data. Ti/Fe Cr/Fe S Yabushita is based at the Department of Applied Mathematics and Physics, Kyoto University, Kyoto 606, Japan, and A J Allen works at the School of Mathematical Sciences, Queen Mary and Westfield College, University of London, UK. The authors thank S Miono for providing pictures of microspherules and for discussions, and G P Landis for useful communications. They also thank Bill Napier for detailed and helpful comments. This work was supported by grant no of the Ministry of Education, Science and Culture, Japan. References Alvarez L et al Science Bailey M E et al The Origin of Comets (Pergamon). Clube S V M and Napier W M 1986 in The Galaxy and the Solar System ed. R Smoluchowski et al. (Univ. Arizona Press). Clube S V M and Napier W M 1996 QJ RAS. Daniels P A and Hughes D W 1981 MNRAS Geiss J 1988 Modern Astronomy 1 1. Hengst R A et al Geol. Soc. Am. Prog. with abstract Hophenberg H B et al Science Kyte F T and Wasson J T 1986 Science Landis G P and Snee L W 1991 Palaeogeog, Palaeoclim, Palaeoecol Landis G P et al AAAS Annual Meeting, San Francisco. Miono S et al Nuclear Instruments and Methods in Physics Research B Miono S 1995 Il Nuovo Cimento 18c 9. Officer C B et al Nature Rampino M and Haggerty B M 1994 in Hazards due to Comets and Asteroids ed. T Gehrels (Univ. Arizona Press). Rigby Jr, J K and Landis G P 1994 AAAS Annual Meeting, San Francisco. Russel H N et al Astronomy vol. 1 (Ginn & Co, Boston). Scoville N Z and Sanders D B 1986 in The Galaxy and the Solar System ed. R Smoluchowski et al. (Univ. Arizona Press). Sloan R E et al Science 629. Thaddeus P 1986 in The Galaxy and the Solar System ed. R Smoluchowski et al. Urey H C 1973 Nature Wolfendale A W and Wilkinson D A 1989 in Catastrophes and Evolution ed. S V M Clube (Oxford Univ. Press). Yabushita S 1995 Observatory Yabushita S and Allen A J 1983 Observatory Yabushita S and Allen A J 1989 MNRAS Zbik M and Gostin V A 1995 in Proc. Nat. Inst. Polar Res. (Tokyo Symposium Antarct. Meteorites) Zhao M and Bada J L 1989 Nature Table 2: Impact craters of about K/T boundary age Name Diameter (km) Age (million years) Chicxulub ± 0.05 Manson ± 1 Kamensk ± Variation of the O 2 content of the atmosphere contained in ancient amber. (Reproduced from Landis et al ) 3 Chemical abundance analyses of microspherules collected by S Miono et al. (1993). The analyses indicate that the spherules are similar to chondrites. = spherule; = magnetic components of chert and volcanic ash; = magnetic components of meteorites. Volcanic ash was collected from the Osaka area and Kagoshima prefecture. The magnetic component of chert was collected from an Inuyama sample. Spherules were collected from Inuyama and Yokonami samples. The magnetic component of 10 Antarctic meteorites, three carbonaceous chondrites and one meteorite were analysed for an extra-terrestrial sample. 19