Freshwater Gastropods of Equatorial Africa: Correlations Between Shell Isotope Ratios and Environment

A stable oxygen isotopic study of modern African freshwater gastropod shells was carried out to ascertain the feasibility of the use of the stable oxygen isotope ratios of freshwater gastropods as paleoclimatic indicators. rn this study, the oxygen isotope ratios of the freshwater gastropod shell samples taken from the apertures of the individual shells collected at monthly intervals from one site were examined to ascertain their correlatability with weather, temperature, rainfall, water isotope ratios and other ambient conditions. The gastropod shell analysis was by the procedure of Abell (1985) and the water sample analysis by the method of Epstein and Mayeda (1953). The monthly analyses of both water and gastropod shells from the tributary of the River Pra at Krobo, in the Western region of Ghana, approximately 700km from the Sahel, have shown amount effect, temperature effect temperature plots and evaporative effect·. Good parallelism between ' 18 and the S o plots of the shells and the water was observed. The s18o value~ were also in agreement with vegetative cover map of Ghana and Africa in addition to their correspondence with seasonality information. Application of these results to fossil gastropod shells, however, requires additional geological and geographical data.

The National Science Foundation grant for this project is also acknowledged.
The geographical patterns of the stable oxygen isotope ratios obtained from whole shell analysis of individual African freshwater shells have already been established by Abell (1985). In this study reported here, the oxygen isotope ratios of freshwater gastropod shell samples, taken from the apertures of the individual shells collected at monthly intervals from one individual site were examined to ascertain their correlation with weather, temperature, rainfall, water isotope ratios and other ambient conditions. The gastropods and water samples studied were collected at monthl. Y intervals from a tributary of the River Pra at ·Krobo in the Western region of Ghana, approximately 700km south of the Sahel in the tropical rainforest region of that country.
How this information on modern African freshwater gastropods can be applied to interpret the oxygen isotope ratios of some fossil shells is also examined. If clear correlations exist, then we may achieve our ultimate goal of interpreting paleoclimates from shell isotope data.

BACKGROUND
In pursuance of the study of paleoclimates many techniques and substrates have been examined for their utility and reliability, many of which have been reviewed by Schneider (1984). For example, among these tools are studies of the identification of pollen grains in sediments, with the attendant problems of differential production, distribution and preservation. A second example is the use in paleontology, of the assumption that modern species of animals inhabit environments similar to the distribution of Coleoptera indication of maxinrum winter their ancestors. A third example is beetles in sediments to give a rough temperatures. Also the deposition of layers of windblown, yellowish-brown dust called loess can serve as a paleoclimatic proxy, while the oceans with their accumulation of sand, mud, gravel, plant and animal fossils also provide geological paleoclimatic indicators (Schneider 1984  It has also been found out that an increase in 6 18 0 of only 1 per mil in calcite shells corresponds to a change in water temperature of about 4.o 0 c to s.o 0 c (Craig, 1953). The short life spans of most species of freshwater gastropods and their isotopic 3 ratio sensitivity to their environment makes them unique proxies to the range of annual climatic conditions. This does not imply that much is known about the life cycle or reproductive strategies of the freshwater gastropods (Brown, 1980;L~veque, 1968;Brown, personal communication, 1984). While the gastropods which we have examined (mainly Melanoides tuberculata, Bellamya unicolor and Cleopatra ferruginea) are believed to live for one to two years, these estimates are based on aquarium-grown specimens, and need to be verified for gastropods growing in natural habitats.
Since the application of stable isotope ratio measurements in paleoclimatology, the volume of work done on freshwater mollusks with this technique has not approached that done on marine mollusks apparently due to complexity of data interpretation on freshwater systems. Despite these complexities, we would like to see if it is possible to apply this technique to the study of paleoclimates using freshwater gastropods shells. We already know that some correlations have been found to exist between isotopic ratios of modern shells and the environment in which they grow (. Abell,1985).
Obviously, an understanding of how the oxygen isotope ratios of freshwater gastropod shells .vary with . the environment in which they grew could open wide the doors leading to the application of the isotopic ratio measurements to continental paleoclimatology. This was first suggested forty years ago when Harold Urey (1947) recognized that naturally occurring carbonates might be employed in calculating paleotemperatures, with subsequent verification by McCrea (1950).
Since then, various workers (Epstein et al., 1951Mook, 1971Mook, , 4 1977aurchardt,1977;Keith et al., 1960Keith et al., , 1964Keith et al., , 1965 (Mix and Pisias, 1988) The successes thus far have been in the marine environment. By carrying out oxygen isotoP7 study on Pismo Clam (Tivela stultorem) off the coast of California, it has been demonstrated by Lee (1979) that seasonal temperatures are recorded in the growth bands of both modern and fossil clams. Micro-sampling technique along a section cut from the Pismo clam (Tivela stultorem) was effective in this study.
Similarly, Jones et al. (1983) and Arthur et al. (1983)  Similarly Schifano (1983) and that the growth habit of coast of Sicily correlated In yet another recent marine experiment, Grossman (1987) (Mix and Pisias, 1988). Thus, analyses of the shells of marine mollusks may be temperatures with minimal expected effects to give information on marine from ice volume changes or evaporative alterations in the oxygen isotope ratios of the ocean.
The latter effects can of ten be ignored in marine studies over short time spans when it is assumed that the ocean is a constant, unchanging reservoir of oxygen isotopes.
The disparity in the amount of work done on freshwater mollusks compared to that of marine mollusks since the work of Epstein et al. (1951) is attributable to the complexity of data interpretation on freshwater systems. The two major effects that hinder quantitative 6 interpretation of oxygen isotope ratios in freshwater mollusks are (1) variation in the s 18 o values of the rainfall, which supplies the water bodies in question, with time, topography, and latitude (Mook, 1970) and (2) Alterations of the s 18 o values of the original precipitation with time because of evaporative losses, which may be associated with temperature, humidity and residence time (Craig, 1961). Despite these complexities we believe it is possible to correlate oxygen isotope ratios of modern African freshwater gastropods with their environment at least semi-quantitatively and to use this information can be utilized in studying past African climates.
Why African climates ? The geographical limitations of this study were dictated by both practical and theoretical reasons: (1) The African continent provides a wide range of modern climates for comparisons with past climates.
(2) The latitudinal range of the continental land mass, from approximately 30°N to 30°s gives a useful range of climatic conditions, without encountering extremely low temperatures where molluscs may be dormant over large portions of the year.
(3) Initial surveys (Abell, 1985) were made possible by the availability of good museum collections of gastropod shells collected, at least in part, because of public health problems associated with freshwater gastropods.
(4) Finally, it was hoped that one consequence of this study would be application to paleoclimatic influences on the development of early man, whose original home was Africa. 7 past African climates are of interest because we would like to know if the expansion of the Sahara desert in Africa is a cyclical process or not. Such knowledge would help in the formulation of Land use policies on expectations of probable changes rather than merely hoping for the best.
Knowledge of ancient African climatic variations could provide some answers to some questions raised in the evolutionary process and in geology. The geographical conditions of past water basins could be deciphered, and the validity of the assumption linking environmental change to the process of evolutionary change could be confirmed. This assumption was engendered by the fact that the ancient environment had been profoundly different from that of the present at various times, not only in its physiography but also in its ephemeral aspects such as climate and vegetation, a view recognized by early natural scientists like oa Vinci, Lyell and Darwin.
As an outgrowth of the recognition of the process of transformation from past to present environment, this assumption is held by many scientists in this field, e.g. Dobzhansky, (1962);Levins, (1953) ;Lewin, (1984);Matthew, (1915Matthew, ( , 1939Pearson, (1978) ;Simpson, (1953); and Vrba, (1980and Vrba, ( , 1984. Of particular concern to anthropologists are the environmental changes that occurred in Africa during the latter half of the Cenozoic, and the role of this environmental change in the evolution of the higher primates including the hominids (Brain, 1981a(Brain, , 1981bButzer, 1977Butzer, , 1978Livingston, 1971 ). Laporte and Zihlrnan (1983) observe that " ...•• the environmental setting is a major driving force in hominoid evolution," and Brain (cited in Lewin (1984)) argues that 8 " had it not been for temperature-based environmental changes in ....
habitats of early hominids, we would still be secure in some hospitable forest, and would still be in the trees." However, while major trends in worldwide Cenozoic climate are well understood, details are scarce and the chronology, particularly in the first half of the Tertiary, is discontinuous, permitting anthropologists to invoke environmental change at nearly any juncture in a particular evolutionary scenario. As Butzer and Cooke, (1982) have observed, it would be useful to supplement essentially deductive approaches with "critical regional studies emphasizing empirical data that can be set into tightly controlled radiometric or stratigraphic frameworks." To find an appropriate tool which can be used to decipher seasonality in past climates, that we have undertaken to examine the correlation of s 18 o of freshwater gastropod shells and the water in which they lived over a period of one year. Seasonality, which has been defined as marked cyclic annual alteration of temperature and rainfall, is presently being discussed for its role in the interplay between climate and the evolution of the hominoids, hominids and any organism. According to current climatic theory, global weather patterns exist because dif~erent parts of the globe receive and absorb varying amounts of solar insolation; upper air currents and ocean currents are mechanisms for restoring heat balance (Budyko, 1978 (increasing S 0 as evapora ion increases will tend to cancel each other. As it turns out, there is a good correlation between oxygen isotope ratios and general climatic conditions and some of the factors which influence the precipitation that eventually determines the inmiediate environment of the freshwater gastropods have been examined by Yurtsever (1975) and Dansgaard (1964 (Dansgaard, 1954) to increase with high latitude and (Epstein,1956) to increase with altitude, giving what Dansgaard refers to as "latitude effects" and "altitude effects". I will utilize these four named effects in the following discussion, although it is obvious that they are not truly independent effects.
Fractionation may occur in the evaporation of liquid precipitation or in the condensation of water vapor. During any of these processes, it is the more volatile H 2 o 16 which is readily exchanged as opposed to the heavy isotope counterpart. The fractionation of liquid precipitation in nature may be induced by biological processes or be engendered by mere exchange with other environmental materials.
Irrespective of how the fractionation is brought about, it is ultimately dependent upon temperature and the rate of exchange.
rn this work only simple equilibrium processes are considered.
This is because the equilibrium and kinetic effects of the fractionation process of liquid precipitation are not fully understood yet.
The environment in which this process occurs must be considered in any plausible explanation offered. Full consideration is therefore given to the environment in which there is an isotopic exchange between an evaporated water molecule and at the surrounding vapor. (Dansgaard, 1953;Friedman and Machta, 1962;Craig et al. 1963;Erickson, 1964). At higher latitudes where evaporation from falling raindrops is minimal, the amount effect becomes less pronounced. Direct evidence for the enrichment of falling drops through evaporation have been reported by Dansgaard (1953Dansgaard ( , 1961 and by Ehhalt et al.(1963).
The gross mean 6 18 0 over a year can be represented by a simple 12 mean over 12 months: The third genus, Cleopatra, is widely distributed, usually as C. ferruginea, but is less common than the other two. See Brown, (1980) for distributions of these species throughout Africa.
The procedure involved collection on a monthly basis of water samples, freshwater gastropods and data on prevailing climatic conditions during the 12 month collection period from December 1985 to November 1986 (Table 1). The gastropods were collected at shallow (1 to 2 meters) depths, and were thus growing in water that was wellmixed, not deep or stagnant. Approximately 10 gastropods were collected at the bottom of the river (usually in a cluster), allowed to die and then dried before shipping to the laboratory. The water sample was collected in a 25rnl septum capped vial, which was lowered down into the river at the proximity of the gastropods and then opened to fill up. It was recapped, labeled and kept sealed for analysis in the laboratory. Water temperatures were also measured monthly over the same period at the collection site.
The processing of the freshwater gastropod shells for the purpose of evolving their co 2 content into ampoules for stable oxygen isotope ratio determination was 1?Y the procedure of Abell (1985). This involved longitudinal sectioning of a whole shell with a low speed diamond wafering saw in such a way that if a shell has 6 to 8 complete turns in the growth spiral, sampling from opposite edges of each turn (180°) in the sectioned shell would give 12 to 16 samples.
Since adult Melanoides individuals generally have robust shells, long sequences of samples can be obtained along the growth spiral. The  (Grossman, 1962).
Monthly water samples were analyzed for the oxygen isotope ratios according to the procedure of Epstein and Mayeda (1953). This With the water samples having pH of 6 or lower, there was no need to adjust pH to facilitate rapid equilibration (Mills and Urey, (1940). The water ~ample was inserted into a 25cc round bottom flask provided with ground glass joints and a stopcock so as to be connectable to a vacuum manifold. The water was frozen in a liquid nitrogen-2-propanol slush bath and the air quickly pumped away. The ice was melted and warmed to room temperature to release gas which was trapped during the initial freezing. The water was then refrozen and Plllllped for a minute to remove the remaining non-condensible gases.
The ice was melted again and conunercial cylinder carbon dioxide, 99.8% purity, was introduced into the flask to a pressure of about 73.5cm Hg.
After equilibration at 31°c for 7 days with frequent shaking, an approximately 3cc aliquot of the carbon dioxide was withdrawn on a vacuum line by freezing into ampoules and analyzed on the v. G.
Micromass 602-D mass spectrometer using the method described by McKinney et al (1950).
All mass spectrometric analyses are reported relative to the PDB carbon dioxide standard gas. formula: The results were calculated from the  (Beadle,1974 Yurtsever (1975) has shown that spacial variations observed in the mean oxygen isotopic composition of precipitation will be influenced mainly by temperature. This correlation is not considerably improved by the other parameters. Yurtsever (1975) performed The results of the multiple regression analyses derived from data collected from the 91 network stations have been adapted in Table 2 as follows: 24    as amount effect or evaporative effect may become important in determining the spatial isotope variations when considering regional For example, in Figure 5, the more negative mean 8 18 0 values in data.
one group of stations which is apparently due to the amount effect deviate from the general 8 18 0-temperature relation, having significantly higher mean precipitation values. Figure 6  altitudes s ou observed that precipitation at higher altitudes is more depleted in heavy isotopes than that at lower altitudes.
In effect, it can be observed that the four major parameters considered by Yurtsever (1975) and Dansgaard (1964) may be important but are not truly independent as three of the parameters are to a large degree controlled by temperature. It is also known that the final composition of liquid precipitation on the ground would be different from the initial composition of a liquid precipitation in the clouds.

FAC'IORS THAT INFLUENCE OXYGEN ISOTOPE RATIO IN SURFACE WATER
Surface water is believed to be influenced by factors which are similar but are over and above those that influence precipitation in the cloud. Prominent among the factors that alter the oxygen isotope ratio in the post-precipitation phase are temperature and evaporation.
The alteration of the oxygen isotope ratio may be caused by evaporation and exchange with environmental vapor. In dry air, relatively high evaporatio~ causes the preferential escape of H 2 o 16 (light water) creating a non-equilibrium conditions (Ephalt et al., 1963) resulting in heavy water being left behind. Dynamic equilibrium exchange between the precipitation and the environment which occurs in humid environment has been observed by Friedman and Machta (1962).
Should amount effect become important in the oxygen isotope ratios of a given precipitation, it would be expected that the environment with a greater amount of precipitation in a given month, would be more depleted in oxygen-18 which would also be incorporated into the shell of gastropods that lived in that environment. Despite the different effects that can influence the oxygen isotope ratios in precipitation and in shell incorporation, it is still possible to correlate oxygen isotope ratios of a given body of water with the oxygen isotope ratios of the freshwater gastropods that inhabit it and even calculate the temperature at which the gastropod shells were formed (Epstein and Mayeda, 1953). The isotope temperature scale formulated for this calculation is : where T is the temperature in °c, A is the S value for the H 2 o in which the caco 3 was precipitated and S is the oxygen isotope ratio in the gastropod shell. Any considerable deviation from the above temperature formula would mean considerable fractionation or influence of the precipitation by other factors other than temperature.
The inability of freshwater gastropods shells to record the original s 18 o of the rainfall as predicted by latitudinal influence, (as shown in Figure 7) is due to other effects over and above those discussed by Yurtsever (1975) and Dansgaard (1964) on postprecipitation isotope ratios. The main parameters that affect postprecipitaion isotope ratio are amount, temperature and evaporation which Abell (1985) has rightly linked up with vegetation cover. As a 32 FIGURE 7 EXPECTED LATITUDINAL EFFECTS ON OXYGEN ISOTOPE RATIOS, IGNORING LAND MASS -INFLUENCES (ADAPTED FROM ABELL, 1985) 33 result of regional patterns in s1ao of gastropods it is possible to sort out the major effects on post-precipitation isotope ratio. This is possible because some of the influential parameters on postprecipitation isotope ratio may be emphasized or discarded. This does not necessarily mean the same pattern of emphasis or discarding would apply for every site, but that some rules apparently apply. over much of Africa, the primary influence determining the isotope ratios in gastropod shells is the latitude. Rainfall is fractionated as air masses move and this is clearly manifested in the latitudinal effect on oxygen isotope ratios in precipitation as noted by Dansgaard (1964) and Yurtsever (1975).
Gastropods, sampling the oxygen isotope ratios in that rainwater in their host body of water, will generally provide a faithful record of that isotope ratio, particulary if the Saharan lakes will be maintained by seasonal rainfall in their source areas, but will be subject to continuous high evaporation rates. Lake Turkana, for example, fed largely (80%) by seasonal rainfall in the highlands of Ethiopia, has a climate of relatively unchanging mnPrature, but te"'t'-the lake 18 level, and the 6 O of the gastropod shells oscillates with this seasonal imput. In all these arid locations, the average value of evaporative conditions. 36 l e at both Lake Malawi and Lake Victoria (Winam Gulf) there is exa111P ' · t ure change with the seasons, but 6 18 0 of the shells nU.nimal tempera varies considerably (Abell, unpublished data). There will be exceptions to these generalities, but they make a starting point.
In an attempt to apply these interpretations to paleoclimates we lllllSt not lose sight of the importance of the correlation between amount effect or rainfall and vegetation. The amount effect which according to Dansgaard (1964) is engendered by the deep cooling of air in heavy frequent rainfall, with minimum possible post-precipitation enrichments through evaporation, has been found to correlate well with vegetation cover (Tucker et al., 1985, Abell, 1985. With respect to the African climate, it has been found that areas with enough rainfall, and rainfall sufficiently well distributed throughout the year to ensure permanent vegetation cover (Figure 9 Tables 3 and 4 summarize the oxygen isotope ratio information on the freshwater gastropod shells and the water samples respectively.
In Table 1 Table 4 which is also plotted in It can be observed that the shell isotope ratios lag behind water isotope ratios by about 1 to 2 months. 39      Table 1 are also shown on the bar graph plotted in Figure   12 . The maximum, minimum, temperature range and average temperatures extracted from Table 1 are sununarized in Table 5. The average temperature recorded in Table 5 was determined by the simple mean over 12 months period. Average sUimner and winter temperatures were obtained over 6 months period with sUimner in the Northern hemisphere defind by the months between May and October and winter months from November to April. Ambient temperature difference over the period of collection is twice that of the water temperature difference over the same period. The average sunnner temperature is lower than the average winter temperature.
By performing regression analysis on data collected from the tributary of the River Pra at Krobo, the regression equations were obtained with the correlation coefficients. Table 6 sununarizes the correlation between water temperature, air temperature, precipitation and the oxygen isotope ratio values of both the water and the shell.  48 Figure 13 while that of the shell 8 18 0 and water temperature with the one month slippage is shown in Figure 14.
The correlation of precipitation with shell oxygen isotope ratios, 0.66, and with water oxygen isotope ratios, 0.42, are good however the shell s 18 o values correlate better with precipitation than the water 8 18 0 values. Figure 15 shows The correlation plot of the air temperature and shell oxygen isotope ratios is shown in Figure 16.
The correlation between the shell oxygen isotope ratio values and the water oxygen isotope ratios, 0.68, is also listed in Table 6. is a good indicator as to the possibility of applying oxygen isotope ratios in paleoclimatology, is in accordance with an earlier observation by    Table 7 and also plotted in Figure 17 ( Griffiths, 1972). The data given are from stations very close to Krobo (refer to Figure 18 (1967)). In areas of Africa especially West and Central Africa where rainfall is high and well spread out over a period of one year, such that there is permanent vegetative cover, 8 18 0 values in gastropod shells have been observed to be between -3.5 to -0.9 per mil (Abell, 1985). our values fall close to this range with the average monthly 8 18 0 for the one year period being -3.51 for the water samples values of -0.52 from the water and -0.54 from the shell. This is in consonance with the observation of Tucker et al., (1985) and Abell (1985) where latitudinal effect is overwhelmed by amount effect which in turn correlates well with vegetation cover map of Ghana. (Figure 3) The seasonality information sununarized in Table 8 is to assist in the separation of amount effect from temperature effect in the tropics.   The maxinrum and mininrum s 18 o values recorded by both shell and water correspond very well with the months for which maxinrum and minimum temperatures were recorded. The average ambient temperature was 3.o 0 c higher. The water temperature fluctuation was 2.5°c while the air temperature fluctuated by 5.0°c during the collection period.
The average summer temperature is lower than the average winter temperature. A negative difference is obtained when the average winter temperture is subtracted from the average summer temperature.
The conversion of this te~rature difference to 5 18 0 gives a negative s 18 o value which substantiates the amount effect obtained by the difference between 5s -Sw.
Amount effect is further substantiated by the correlation between precipitation and 5 18 0 of both the water (0.42) and the shells (0.66).
By these correlation values it is observed that approximately 18% of the variations in the water s 18 o values at the River Pra is due to amount effect while approximately 44% of the variations in the shell 6 18 0 values is attributable to amount effect (Table 6). Significant 18 18 . correlations exist between shell S O and water S O air temperature and precipitation, however, the correlation between shell 8 18 0 and water temperature is even better (0.74). This is understandable because the microenvironment of the gastropod is greatly influenced by the water temperature. Application of these results to fossil samples must be carried out with caution as several other factors come into play and must be taken into consideration in the interpretation of 8 18 0 ratios of fossil gastropod shells. Some of the factors that must be considered include altitude, latitude, reliability of dating, longevity of gastropods and knowledge of the hydrolog~cal regime from which the gastropod shell was collected.
For example, the ancient site from where the shell is collected should be well 9ocumented and knowledge of the longevity of the individual species of gastropods must be known. It is essential to know the age of the shell in order to correlate it with other historical environmental information. Some fossil shells undergo diagenesis or recrystallization from aragonite to calcite, a process engendered by local weathering conditions. Whether a shell is aragonitic or calcitic can be detected by X-ray diffraction (XRD). In 63 order to extract seasonality information from a fossil shell, a sequential sampling along the growth spirals of the individual shells is necessary, while the average oxygen isotope ratios of whole shells will be useful in the characterization of major regional climatic trends ..
Apart from gastropod shell information, some environmental information which can be acquired by other geological techniques has to be provided before interpretation of applying oxygen paleoclimates.
isotope ratio values to the Among these altitudinal, and hydrological information. 6. Information on the general character of the body of water is needed to apply shell oxygen isotope ratios as proxies in paleoclimatology. 66