Banded Corals: Changes in Oceanic Carbon-14 During the Little Ice Age

Radiocarbon analyses and stable isotope measurements are presented foro recent cores of banded corals from the Florida Straits. These values provide a record of variations in the ratio of carbon-14 to carbon-12 in the dissolved inorganic carbon in the surface waters of the Gulf Stream from A.D. 1642 to 1800. An increase in the carbon-14/carbon-12 ratio of 7 per mil for coral growth during the early 1700's was most likely induced by an increase in the carbon-14/carbon-12 ratio of 20 per mil in the atmospheric carbon dioxide that occurred at about 1700. The ratios of oxygen 18 to oxygen-16 in these coral bands show a small decrease of a water temperature (∼1�C) during the latter part of the Little Ice Age (1700 to 1725). These results support the hypothesis that the increase in atmospheric carbon-14 at about 1700, and possibly the temperature change as well, was caused by a decrease in solar activity (Maunder sunspot minimum).

14N. It is quickly oxidized to 14CO2 and distributed into the troposphere, oceans, and land biota. During preindustrial times (before A.D. 1900), spatial variations in the 14C content of the tropo-resembles a partial s period of about 11,500 tion is attributed to a intensity of the magnel of the earth (4). A stro decreases the flux of rays that reaches th Summary. Radiocarbon analyses and stable isotope measuremE for two recent cores of banded corals from the Florida Straits. Thes record of variations in the ratio of carbon-14 to carbon-i 2 in the d carbon in the surface waters of the Gulf Stream from A.D. 1642 to in the carbon-14/carbon-1 2 ratio of 7 per mil for coral growth durin was most likely induced by an increase in the carbon-1 4/carbon-12 in the atmospheric carbon dioxide that occurred at about 1700. Th4 18 to oxygen-16 in these coral bands show a small decrease of a (-10C) during the latter part of the Little Ice Age (1700 to 172 support the hypothesis that the increase in atmospheric carbon- 14 4 possibly the temperature change as well, was caused by a decrea (Maunder sunspot minimum). sphere were minimal (±0.2 to 0.3 percent) (1). However, the world's surface oceans exhibited large variations in "'C (±15 percent), due to a wide range of vertical mixing coefficients for near-surface waters. Regardless of these variations, the 14C concentrations in the surface ocean have always been lower than those in the atmosphere (2). This depletion is maintained by the diffusion and advection of older subsurface waters into the surface layers of the oceans.
Secular variations in the atmospheric 14C concentrations (3) have been recorded in tree rings for the past 8000 years (Fig. 1). The major trend in these data two recent episodes of unusually high atmospheric "'C concentration that occurred around A.D. 1500 (Sporer sunspot minimum) and A.D. 1700 (Maunder sunspot minimum). These periods of nic high 14C were coincident with decreased solar activity (9). Stuiver and Quay (10) ege determined the increase in the atmospheric 14C concentration that occurred around A.D. 1300 (Wolf sunspot minitruffel mum) (Fig. 2). These three periods of unusually high 14C production were coincident, although not directly correlated (11), with recorded intervals of especialine wave with a ly severe winters in Europe (3), a period years. This varia-known as the Little Ice Age. Whether gradual change in the Little Ice Age was the direct result of tic dipole moment low solar activity or whether this was a ng dipole moment coincidence has not yet been resolved. f galactic cosmic It is possible, although not probable, e earth's strato-that the observed increase in the atmospheric 14C concentrations during the Little Ice Age could also have been ants are presented caused by decreased vertical mixing in ,e values provide a the surface oceans or by large changes in lissolved inorganic the rate of CO2 exchange between the air 1800. An increase and the ocean. Neither of these changes g the early 1700's is expected during periods of cooler avratio of 20 per mil erage world temperature, such as that a ratios of oxygen-during the Little Ice Age. In order to water temperature eliminate these possibilities, however, it 25). These results is necessary to acquire "'C records for at about 1700, and this period in the oceans. I show here se in solar activity that banded, hermatypic corals can be used as recorders of "'C concentration in ocean waters during the Little Ice Age, just as tree rings are used to record 14C ower production of concentrations in the atmosphere. The determined that a atmospheric 14C record combined with f the geomagnetic the oceanic 14C record may be used to crease the intensi-determine the causal relationship and the lent on the strato-timing of the 14C variations observed in Fig. 1). these two reservoirs during earlier times. is major trend are the '4C/'2C ratio 200 years (Fig. 1). Corals as Oceanic Recorders are most likely a solar activity (6, Hermatypic corals accrete aragonite, a c-ray flux respon-crystalline form of calcium carbonate, n is modulated by with "4C/12C ratios equal to those in the naagnetic fields. As dissolved inorganic carbon (DIOC) in the ar wind increases, seawater at the time of coral ring formarays are diverted tion (12,13). As the world's surface Isphere, causing a oceans are saturated with respect to production. From aragonite, hermatypic coral skeletons do Vries (8)  Corals also record evidence within their skeletons of significant ecological conditions and changes that occurred during their lifetimes. Wells (14) studied middle Devonian fossil corals and interpreted fine ridges on the surface of the coral epitheca to be daily growth bands. He found approximately 400 ridges (days) per annum, which agrees with astronomical expectations of the deceleration of the earth's period of rotation due to tidal friction. c.
'E qa The most conspicuous records contained within coral skeletons are annual density bands. These growth bands were first conclusively demonstrated as annual by Knutson et al. (15). The annual growth bands are primary skeletal characteristics that are exhibited as seasonal variations in the bulk density of the secreted skeleton (16). Many investigators have demonstrated the annual nature of density band pairs in hermatypic corals, using techniques such as alizarin staining, x-radiography, densiometry, autoradiography, and direct field observations (15,17,18).
There is a partial void in both of these cores (around A.D. 1800) that represents 000 BC. to A.D. the absence of no more than 5 years of nents obtained by coral growth (27).
to be the result of Hudson et al. (18) used alizarin-stainbest-fit sine wave ing techniques and x-radiography to deges caused by the pict annual density bands in Montastrea annularis from the Florida Straits. These investigators noticed that dense aragonitic bands accreted during the warm summer months of July through September and that thicker, less dense bands accreted during the cooler months of October through June. The most con-X f. t+vincing evidence that these bands are 't' t mannual is the presence of stress bands in 4*6** the coral record. These bands form during unusually severe winters and are thicker than the annual dense bands of summer. These events, referred to locally as cold fronts, are the result of cold weather that originates in the northwest and passes over the reef in a southeast-

Procedure
Coral slabs (4 millimeters thick) were cut along the vertical growth axes of both cores. Upon x-radiographic analyses (18) of the slabs, it was determined that TRI grew from A.D. 1642 to 1975 ( Fig. 4) and TRII grew from A.D. 1692 to 1978. Subjection of these samples to xray diffraction analyses revealed pure aragonite and showed no traces of calcite. Each core was sectioned into samples that consisted of one to ten consecutive years of growth. Radiocarbon analyses were carried out at the La Jolla Radiocarbon Laboratory (28) on a total of 135 samples from both cores. Druffel and Linick (12) reported 65 of the results from TRI for the period 1800 to 1974. The 14C measurements were carried out by standard gas proportional counting techniques (12).
All measurements were corrected for isotope fractionation (to a &'3C relative to PDB-1 = -25.0 per mil) (29) (Fig. 5a). There was a deviation from this average (to -43 ± 3 per mil) in a coral sample that grew from 1656 to 1660. The 14C measurements of coral that grew after 1706 were significantly higher than -49 per mil. Of the 16 A14C results for coral that grew from 1709 to 1740, almost 90 percent were higher than -49 per mil. A least-squares analysis of these 16 measurements reveals a significant slope; the fit of these data is shown in Fig. 5a. The 26°N 25°N 14C concentrations increased rapidly from -49 to -42 per mil in a relatively short period (1706 to 1712). The values decreased slowly during subsequent years (1712 to 1750), from -42 to -49 per mil. The trends that are apparent in Fig. 5a are also shown to be statistically significant on the basis of a spline function (third-order polynomial) fitted to the "14C data from Florida and Belize (19) corals (Fig. 6). Thus, it is apparent from these data that an increase in "14C of 7 per mil had occurred in the surface waters of the Florida Current (Gulf Stream system) during the early 1700's. During X' _' Stream System, is located 4 kilometers south of the collection site. The major part of the reefwater is supplied by the Florida Current. 15 the second half of the 18th century the 14C concentrations remained unchanged; the average of 19 A'4C values is -49 ± 3 per mil.
The increase in 14C in oceanic DIOC during the early 1700's was most likely the result of increased 14C concentrations in the atmospheric CO2 during that time. Figure 2 shows two 14C maxima that were observed in tree rings from the early 16th and 18th centuries (10). These maxima, which coincide with the Little Ice Age, are almost certainly the result of decreased solar activity (7). As CO2 is exchanged between the atmosphere and the surface ocean, an increase in the atmospheric 14CO2 concentration would also appear in the DIOC of the surface ocean. There is a lag time of three to four decades between the onset of the increase in the atmosphere and that in the ocean (Fig. 7). Part of this delay can be attributed to the long residence time (10 to 15 years) for '4CO2 in the atmosphere (2,12). The 14C increase in the atmosphere was about 20 per mil by the beginning of the 18th century, whereas the increase observed in corals during this time was only 7 per mil (Fig. 7). The attenuation of the 14C peak in the surface ocean was caused by vertical exchange of older subsurface waters, which contain less 14C, with surface waters (2).
The isolation of subsurface waters from the atmosphere for extended periods of time causes 14C to be lost as a result of in situ radioactive decay (2).
It is possible that the increase of A14C in the surface waters of the Florida Straits during the Little Ice Age could also have been caused by a decrease in the rate of vertical exchange in the upper few hundred meters of the water column; this could have been caused by warmer surface water temperatures, which would not have been expected during an ice age. However, the 8180 values from these banded corals (Fig. 8) indicate that the average temperature in Gulf Stream surface waters was slightly lower during the latter stage of the Little Ice Age (about 1700). A least-squares analysis of the 8180 values reveals an overall decrease of 0.2 per mil from 1700 to 1790 (dashed line, Fig. 8), which corresponds to a warming of about 1°C (24,25) from the latter part of the Little Ice Age to the end of the 18th century. A closer look at these data reveals a maximum 8180 value of -4.0 per mil around 1725 and a sharp decrease to -4.4 per mil in the 1740's, which represents a rise in temperature of 2°C over this time period. There also appears to be a somewhat smaller maximum (-4.1 per mil) around 1760. These maxima (solid lines, Fig. 8) represent periods when the surrounding seawater was 10 to 2°C cooler than during the earlier periods. Cooler surface water temperatures during the early 1700's imply that there was an increase in vertical mixing between surface and subsurface waters, not decreased vertical mixing as would have been the case had changes in seawater temperature been the cause of the 14C increase in the 1700's. An increase of the '4C/12C in the atmosphere due to the higher solubility  (2), a subtropical gyre in the North Atlantic that circulates in a clockwise (anticyclonic) direction. The circulation in the surface mixed layer is wind-driven (Ekman transport), and that in the deep water is a result of geostrophic forces. Ekman transport is convergent in a subtropical gyre, which forces water downward from the mixed layer. There is a complex process at work that selects only late winter water for actual net downward pumping, called Ekman pumping, into the geostrophic regime below (30). As the surface waters in the Gulf Stream were 1°to 2°C cooler during the early 1700's, it is probable that Sargasso Sea surface waters were also cooler during this period. It is likely that cooler surface water temperatures during the latter part of the Little Ice Age enhanced the downward penetration of waters in the Sargasso Sea during the late winter and perhaps induced prolonged convection that extended from early winter to spring.
Results for Subsequent Coral Growth, A.D. 1800 to 1952 The 14C concentrations in Florida coral that grew subsequent to 1800 also appear to reflect 14C changes in the atmosphere. The A"'C measurements of Florida coral that grew during the 19th century (12) are shown in Fig. 5b. From 1800 to 1820, A14C values remained unchanged from the previous 50-year period (-49 ± 5 per mil). However, further 14C analyses since the publication of (12) reveal a significant decrease from about 1815 to 1865.
This decrease of 4 to 5 per mil in the A14C of the surface ocean during the early and mid-1800's was coincident with a decrease of 9 per mil observed in the atmosphere (10) (Figs. 2 and 7). As was the case for the data from the early 1700's, the magnitude of the 14C variation in the atmosphere was at least twice that in the surface ocean ( Table 2). It is probable that the 4 to 5 per mil decrease observed in ocean waters was caused by declining atmospheric '4CO2 concentrations. It is unlikely that this decrease was caused by enhanced vertical mixing in the upper layers of the ocean in view of preliminary b180 measurements which show a slight increase in the average surface water temperature from 1800 to 1900 (31).
A recent decrease in the 14C concentration from A.D. 1900 to 1952, known as the Suess effect, has been noticed both in the atmosphere (32) and in the surface waters of the Gulf Stream (12,19) and in the Peru Current (20). Concentrations of '4C in the atmosphere decreased 20 to 25 per mil as a result of a dilution with 14Cfree CO2 that results from the burning of fossil fuels (33,34). The change in the surface waters of the Gulf Stream during this period (-12 per mil) (Fig. 6) was about half that in the atmosphere. Table 2 lists data on the three variations in 14C that were observed in the atmosphere and in the ocean from A.D. 1642 to 1952. The ratio of the observed variations (atmosphere/ocean) ranges from 1.7 to 2.8. The ratio predicted by the box-diffusion model of Oeschger et al. (35) for the Suess effect, a perturbation of the 'CI'2C that was introduced correlation between these stress bands and the cold winters provides conclusive evidence that the density banding in this coral is indeed annual. Fig. 5 (right). coral that grew from 1710 to 1750;' a least-squares analysis of these data revealed a significant decrease from -42 per mil in 1710 to -49 per mil by 1750. This rise of 7 per mil during the first half of the 18th century is attributed to increased 14C concentrations in atmospheric CO2 (+20 per mil) during that time (Fig. 2). (b) The A"'C measurements for coral (TRI and TRII) that grew from A.D. 1760 to 1930. There is a noticeable decrease (-4 to -5 per mnil) from 1820 to 1870. This decrease is attributed to decreased "'C concentrations in atmospheric CO2 (-9 per mil) during that time (Fig. 2).  (10,34). The A'4C trend for corals is a smoothed sp curve fitted to data from Montasfrea annularis collected from Belize (Gulf of Honduras) and the Florida Straits, both in the Gulf Stream system (see Fig. 6).   18 initially into the atmosphere, is 2.5. The ratios listed in Table 2  The activities, values, and behavior of an individual that are acquired through instruction or imitation will be termed "cultural." Such phenomena are not exclusively human (1) but are tmost highly H. Cheh, S. M. Dornbusch selection has produced the complexity and diversity of living systems is the cornerstone of interpretation in the biological sciences. Observed genetic variation is the result of interactions between Summary. Cultural phenomena may show considerable stability over time and space. Transmission mechanisms responsible for their maintenance are worthy of theoretical and empirical inquiry; they are complex and each possible pathway has different effects on evolutionary stability of traits, as can be shown theoretically. A survey designed to evaluate the importance of some components of cultural transmission on a variety of traits showed that religion and politics are mostly determined in the family, a mode of transmission which guarantees high evolutionary stability and maintenance of high variation between and within groups. developed in our species. In attempting to construct a quantitative theory for the evolution of cultural traits we have found many concepts from the quantitative theory of biological evolution to be useful (2). It has oftenl been suggested (3), though not widely appreciated, that the evolution of cultural phenomena can be viewed in a conceptual framework similar to that of biotogical evolution, but so far most analyses have been purely qualitative.
That the continuing process of evolution by random mutation and natural SCIENCE, VOL. 218, 1 OCTOBER 1982 the rules of genetic transmission, mutation, natural selection, and sampling, due to the finiteness of natural populations. Each of these phenomena can, in principle, be measured, and together they allow statistical prediction of the evolutionary trajectories of the genotypes in the population.
The cultural analog of mutation includes innovation as well as random change in the expression of traits (2). In fact, Galton (4), in explaining biological mnutations ("sports" in domesticated plants and animals) compared them to technological innovations. Our concern here is not with the comparison of mutation and selection in biological and cultural situations, but with another ingredient in the process of evolution-transmigsion. Although well studied and quantified in biology, transmission is poorly understood in its cultural context. The study of quantitative aspects of cultural transmission can, we believe, create a foundation for the study of cultural evolution and, in the quantitative theoretical development upon which we have embarked, modeling of cultural transmission has a central place (2). To date quantitative studies of cultural transmission have been limited, although there already exist theories, such as mathematical epidenliology (5), which could augment the study of diffusion of innovatiorts (6). In this article we suggest some of the possible applications of our general theory, in an empirical investigation of quantitative aspects of our general theory.

Models of Transmission
Cultural transmission is the process of acqiuisition of behaviors, attitudes, or technologies through imprinting, conditioning, imitation, active teaching and learning, or combinations of these. A quantitative theory of the evolutioh of a culturally transmitted trait requires mod-