A latest middle Miocene subtropical flora from the northern Altiplano, Bolivia: evidence for a young age of the Central Andean Plateau
___________________
Kathryn M. Gregory-Wodzicki
Luis Felipe Hinojosa
Lamont-Doherty Earth Observatory of Columbia University, Palisades, NY 10964-8000 USA. gregory@ldeo.columbia.edu
Laboratorio de Palinología, Facultad de Ciencias, Universidad de Chile, Casilla 653, Santiago, Chile
ABSTRACT
Combined flora: MAT = 23.8 °C, MAP = 600 mm. Modern values for the same site are MAT = 8.3 and MGSP = 300 mm.
RESUMEN
INTRODUCTION
An interesting problem receiving attention in recent years is the nature of the interaction between the uplift of the earth surface and global climate change (i.e. Ruddiman, 1997 and references therein). Most study has focused on the uplift of the Himalayan Plateau and Rocky Mountains and their possible contribution to Cenozoic cooling via affects on atmospheric circulation or weathering rates. The Andes are thought to have less influence on global climate, because even though they extend 8000 km and have elevations up to almost 7000 m, they are fairly narrow; they extend only 300 km their widest point in the Central Andean Plateau of Bolivia (Lenters et al., 1995).
Yet the Andes undoubtedly have a dramatic affect on the climate of South America. First of all, the high, north-south running barrier formed by the Cordillera blocks zonal airflow. One effect of this influence is to block the eastward drift of the southern Pacific subtropical anticyclone, thus stabilizing its location. This circulation creates a descending current of dry air along the western coast of South America, and drives the Humbolt current, a cold current which flows from south to north along the coast.
Secondly, the Andes create one of the largest horizontal precipitation gradients on earth. The Andes force moist air masses from the Amazon to rise; as the air masses rise they cool and condensation occurs along the eastern slopes. The area leeward of this zone becomes increasingly more arid from east to west. The combination of the rain shadow and cold coastal current make the central coast of South America one of the driest places on earth, while the eastern Andes of Columbia are one of the wettest places on earth due to orographic rainfall (Trewartha, 1981)
Thirdly, the elevated topography, most notably the Central Andean Plateau of Bolivia, changes patterns of seasonal heating. In the winter, the plateau is much colder than the surrounding land, while in the summer, solar heating is intense. The rising of the heated air draws in moist air from the Amazon in a quasi-monsoon circulation.
Because of these important influences on atmospheric circulation, a knowledge of when the Andes became a significant barrier is critical to our understanding of the climatic, biologic, and geologic history of South America. Most geologic evidence suggests that the Andes are quite young, with most of the elevation created in the late Miocene to present (Gregory-Wodzicki, 2000), though many of the indicators are qualitative. As of now, some of the most quantitative estimates are derived from paleobotanical techniques, which use the fact that plants are sensitive recorders of climate and other environmental conditions. Unfortunately, very few floras have been studied from the Central Andean Pleateau with a view to reconstructing paleoelevation.
The 10.7 Ma Jakokkota flora from northwestern Bolivia is an interesting flora to investigate for several reasons. First of all, it grew in the latest middle Miocene, and can thus constrain elevation at a critical time. Also, it grew in the Altiplano, a 3700 m high, internally drained plateau which forms the heart of the Central Andean Plateau. Unlike the eastern slope of the plateau, the Altiplano has undergone very little erosion since the Miocene. Thus any elevation changes are mostly due to plate tectonic processes, rather than isostatic rebound from erosion.
The Jakokkota flora contains two distinct fossiliferous layers: a fluvial deposit, and 9 m above a lacustrine deposit. The lower Jakokkota flora has been studied previously (Berry, 1922; Gregory-Wodzicki et al., 1998), but no collection exists of the upper Jakokkota flora.
In this study, we discuss a new collection of the upper Jakokkota flora. which can 1) provide a more statistically robust sample 2) give an idea of climate variation, 3) give an idea of variation due to depositional environment. The leaf morphology is compared to that of the previously studied lower Jakokkota flora, and the climate and elevation are estimated.
GEOLOGY AND AGE OF THE JAKOKKOTA FLORA
The Central Andean Plateau is composed of three morphotectonic elements, which from west to east are a magmatic arc (Western Cordillera), a 250-km-wide, 3700 m high plateau with internal drainage (Altiplano) and a Miocene thrust belt (Eastern Cordillera). The Jakokkota flora is located in the northern Altiplano of Bolivia, about 100 km SW of the capital, La Paz, and close to the border of Peru (Figure 1).
The Jakokkota flora outcrops in two distinct layers in member 6 of the Miocene Mauri Formation (Sirvas and Torres, 1966; GEOBOL, 1994; Suarez and Diaz, 1996). The lower Jakokkota flora, which was discovered by Berry (1922) and further described by Gregory-Wodzicki et al. (1998) is found in a white, ash-rich claystone to fine-grained sandstone, interpreted to be fluvial in nature (Figure 2 - Strat Column).
The upper Jakokkota flora, which will be described for the first time in this study, is found in a 8 m thick sequence of reddish-tan to green laminated sandstones and siltstones which are found 9 m above the unit which contains the lower Jakkokota flora (Figure 2). Individual laminae range from 1mm to 3 cm in thickness, and grade from siltstone to coarse sandstone at the base to clay at the top. The laminated bedding and graded layers suggest that these deposits are lacustrine.
A fall ash located 3m above the lower flora has an age of 10.66 ± 0.06 Ma, based on single-crystal laser fusion analysis of sanidine, while a fall ash just below the upper Jakkokota flora had less precise ages of 11.35 ± 0.69 Ma on biotite and 12.74 ± 0.69 Ma on hornblend. A fall ash 28 m above the upper Jakokkota flora has ages of 10.81 ± 0.72 Ma (biotite) and 11.31 ± 0.53 Ma (hornblend). The low precision of the biotite and hornblend ages probably reflects alteration; these minerals alter much more quickly than sanidine. Thus the 10.66 ± 0.06 Ma age is considered the most accurate and reliable age derived from the ash falls.
The upper Jakokkota flora is probably not significantly younger than the lower Jakokkota flora. There are no physical signs of a depositional hiatus, such as a paleosol or unconformity, and the biotite and hornblend ages from just below the flora are similar to the sanidine-based age from 6m below the flora. If we compare the biotite and hornblende ages of the two fall ashes, they suggest an average sedimentation rate around 20-50 m/Ma. This would suggest that the 9m difference between the lower and upper floras would represent about 0.2 - 0.5 Ma.
MATERIALS AND METHODS
Fossil leaves and leaf fragments were collected from 2 locations in the lacustrine facies, with a combined volume of approximately 1m2. When dry, these sediments are characterized by pervasive crackle fracturing. In order to preserve the collected fossils, most specimens were coated on the base and sides with plasticine. Once in the lab, the specimens were photographed and split into morphospecies based on venation characteristics. The morphospecies with distinctive venation were identified using comparisons to modern herbarium material.
The morphology of the leaves in each morphospecies was scored after the method of Wolfe (1993). In order to be consistent with the score for the lower Jakokkota flora, forms which had leaves which were very close in size to the next larger size category were scored in both categories in order to compensate for the size reduction observed between canopy and litter samples (Greenwood, 1992; Gregory and McIntosh, 1996). The form scores were summed and divided by the number of forms to derive a site score. For more details on the scoring method, see Wolfe (1993) and Gregory-Wodzicki (2000).
COMPOSITION AND MORPHOLOGY
Floristic Composition
The upper Jakokkota sample was split into 24 morphospecies (Fig X). Those that could be identified are listed below.
ANACARDIACEAE
Schinus sp. 1
Schinus sp. 2
BERBERIDACEAE
Berberis sp.
LEGUMINOSAE undet. (4 morphospecies)
MYRTACEAE undet. (2 morphospecies)
RHAMNACEAE
Zizyphus sp.
ROSACEAE
Polylepis sp.
FELIPE - PUEDES AÑADIR ALGO MAS?
The most abundant form was Schinus sp. 1, which made up almost half of the collected specimens (Table 1). This same form was also quite common in the lower Jakokkota flora, making up almost a tenth of the fossil forms. Another probable Schinus, this one toothed, was less common. Today, there are at least 27 species of Schinus in South America, with the eight species in Bolivia found from 1000 - 3300 mm, from dry to mesic environments (Kileen et al., 1993).
A species of Berberis is present in the upper Jakokkota flora, although it appears to be a different morphospecies than the Berberis encountered in the lower Jakokkota flora. Today, there are 28 species in Bolivia, which are found in montane forest and cloud forest, ranging from 1700 - 3800 m (Kileen et al., 1993).
Four legumes are present, as evidenced by the striated pulvinuses on the fossil leaves. Legumes were also common in the lower Jakokkota flora, though only one form, form 7, appears to be similar. Two Myrtaceae are present; this is an important family in subtropical environments in South America.
Zizyphus sp. was another common form. Today, there are four species in Bolivia; they range from 250 - 850 m, and are found both in xeric and humid forests.
The specimens of Polylepis sp. were very poorly preserved, but the teeth were identical to better preserved specimens in the lower Jakokkota flora, identified as Polylepis. Today this genera is found from elevations between 1700 and 5200 m, in xeric to humid environments.
In total, seven of the morphospecies in the upper Jakokkota flora were also encountered in the lower Jakokkota flora. These include: Polylepis sp., Schinus sp. 1, Zizyphus sp., Form 6 (Legume 2), Form 8, Form 18, Form 25, and Form 48.
Leaf Morphology
The score for theupper Jakokkota flora is given in Table 2, along with the score for the 31 morphospecies of the lower Jakokkota flora. This latter score is slightly modified from that given in Gregory et al (1998); one fragmentary specimen which was not considered a separate morphospecies was recalssified based on the discovery of numerous similar material in the upper Jakokkota flora. The combined score represents the combination of the two floras, and has 48 morphospecies.
The scores for margin characteristics vary somewhat; the percentage of entire-margined species is about 6% higher in the upper Jakkokota flora than the lower. In general, the size distribution of the two floras is similar; both have mostly small leaves, with none larger than the microphyllous 2 category, though the average size of the upper Jakokkota flora is smaller. The distribution of length to width ratios is different, with the upper Jakkokota flora having more species in the equant and elongate categories than the lower Jakokkota flora, which is dominated by leaves in the L:W 2-3:1 category.
The most notable differences occur in the apex and base categories. The lower Jakokkota flora is dominated by species with acute bases, and has subequal numbers of species with acute and rounded apices. On the other hand, the upper Jakokkota flora is dominated by species with round bases and round apices.
CLIMATE ANALYSIS
Climate models
The climate of the upper Jakokkota flora was calculated using models based on relationships observed in modern vegetation between leaf morphology and climate, such as the increase in the percentage entire-margined species with increasing temperature and the increase in leaf size with increasing precipitation. A large number of such models exist, which vary both in terms of the data set and statistical method used, and there has been much discussion in the literature about which provide the most accurate climate estimates (Wolfe, 1995; Stranks and England, 1997; Wilf, 1997, Wilf et al., 1998, 1999, Wolfe and Uemura, 1999; Gregory-Wodzicki, 2000; Jacobs, in prepartion; Greenwood et al., in preparation).
Choosing an appropriate predictor data set is of primary importance, because the closer its relationships between leaf morphology and climate to those of the site to be analyzed, the more accurate the climate estimates will be. The main datasets which have been collected are listed in Table 3. The most extensive is the Climate - Leaf Analysis Multivariate Program (CLAMP) dataset of Wolfe (1993, 1995), which includes 173 sites mostly from North America and Asia, scored for 31 different leaf morphology character states. Other databases contain sites from Africa, East Asia, New Zealand/Australia, and Bolivia.
Studies by Wolfe (1993) and Stranks and England (1997) suggest that the relationships between leaf morphology and climate are not universal, but vary somewhat from region to region. For example, floras from subalpine zones tend to have a higher percentage of entire-margined species than one would predict based on the temperature. The three major leaf-morphology domains identified so far are 1: North America, Carribean, Japan, Bolivia, 2: Australia and New Zealand, and 3: Subalpine zones (Wolfe, 1993; Kennedy, 1998; Gregory-Wodzicki, 2000). Choosing an appropriate data set for comparison with a fossil flora is complicated by the fact that we do not understand why there are different correlations between climate and leaf morphology in these different domains; the differences observed could be due to differences in the climate, environmental conditions, floristic composition, or scoring style. For now, it is probably best to compare fossil floras with modern floras from the same leaf morphology domain. Thus in this study, the Bolivian fossil floras are compared with the Bolivian and CLAMP datasets. However, more work is needed to improve this simplistic strategy.
Some workers have criticized the sampling strategy used in the CLAMP database, arguing that the samples have too few species and are collected from areas that are too small, and as a result they are biased against large, entire-margined leaves. In collecting CLAMP samples, Wolfe (1993, 1995) attempted to mimic fossil floras from a single quarry by collecting foliage from around 30 species of woody dicotyledon from an area of 1-5 hectares, usually in a riparian zone, that is within 5 km of a climate station. Several workers have shown that samples with less than 20 species have large sampling errors for the percent entire-margined species character state; most studies suggest that at least 30 species are needed to obtain climate estimates with an accuracy of ± 2°C (Wolfe, 1971, Povey, Wilf, 1997,. Burnham et al., in preparation). Thus CLAMP samples would seem to be of a sufficient size, if somewhat on the small side, to obtain accurate climate estimates, especially for lower-diversity vegetation such as that from the Sonoran Desert.
After one has chosen an appropriate predictor data set for a fossil flora, then one must choose a type of statistical analysis. Methods which have been proposed include: linear regression analysis, multivariate regression analysis, canonical correspondence analysis, and nearest neighbor analysis (Table 4). A study by Gregory-Wodzicki (2000) suggests that multiple regression analysis tends to produce the most accurate estimates for small data sets with a narrow range of environmental variation that have similar relationships to the flora to be analyzed, and linear regression or canonical correspondence analysis produce the most accurate estimates when using the larger and more varied CLAMP data set. If a similar predictor data set is not available, then nearest-neighbor analysis can still produce accurate paleoclimate estimates.
Wilf (1997) advocates linear regression analysis, arguing that the percent-entire margined species character state explains most of the variation in temperature, and the size character states explains most of the variation in precipitation, thus the additional character states in the CLAMP data set only add noise. While it is indeed true that these character states explain a large part most of the variation, studies by both Gregory-Wodzicki (2000) and Jacobs (in preparation) show that additional character states can improve results. More study is needed on which character states provide useful information.
As the question of which data set and statistical method produce the most accurate results has not yet been resolved, the approach taken in this study is to use all the proposed analysis techniques, and to favor those which provided accurate results for a collection of modern Bolivian floras (Gregory-Wodzicki, 2000). This assumption that the fossil flora will have relationships most similar to the modern Bolivian floras is supported by the fact that the Jakokkota flora is close to the modern Bolivian floras in physiognomic space.
Errors for temperature analysis were calculated using the formula of Wilf (1997), which calculates error as a function of sample size, while errors for precipitation analysis were estimated from the method.
Climate analysis
Results from the climate analysis are shown in Table 5. Mean annual temperature (MAT) estimates for the combined flora range from 20.6 to 23.8 ± 1.9 °C, which is considerably warmer than the modern MAT of 8.3 °C. Our best estimate is that 23.8 °C is closest to the actual temperature for the combined flora, because in the analysis of the modern Bolivian flroas, all models except linear regression analysis tended to underestimate the temperature of subtropical floras.
Whether the upper or lower flora represents a warmer paleoclimate depends on the analysis; the regression models suggest the upper flora was 2-3 °C warmer, while the canonical correspondence and nearest neighbor analyses suggest it was a degree or so cooler. These differences are within the error of the method, so the analysis suggests that the difference in the climate of the two floras is statistically indistinguishable.
The various models derive consistent estimates of the mean growing precipitation of the combined flora of around 600 mm. This figure is likely similar to the mean annual precipitation, because for this calculation is defined as the number of months with a mean monthly temperature > 10 °C. The warm paleoetemperature suggests that the growing season so defined probably was year-round. Today, the mean annual precipitation for the Jakokkota site is around 300 mm.
The upper flora looks drier than the lower flora using multivariate analysis, but is similar according to the linear regression models.
Discussion of Climate Analysis
Estimates of temperature and precipitation for the two Jakokkota floras are very similar, even though the upper flora could be as much as a half a million years younger. The consistency of the estimates suggests that the climate estimate for the combined Jakokkota flora presents a representative picture of latest middle Miocene climate for the northern Altiplano.
Some small differences do exist between the two floras. The upper flora has a higher percentage of entire-margined species, which translated into a higher paleotemperature, at least according to linear regression analysis. However, both canonical correspondence and nearest neighbor analysis suggested that it had a lower paleotemperature, due to morphological differences discussed in leaf size, apex, base, or length to width ratio. In order to see which factors most influenced the temperature analysis the most, we reran canonical correspondence analysis, each time changing the scores for the upper flora in one of the categories so that they were equal to those for the lower flora (Table 6). This analysis shows that the large number of round bases and the more equant and elongate leaves in the upper flora were most responsible for the cooler temperature estimates. The smaller amounts of precipitation estimated for the upper Jakokkota flora is most likely because of the smaller average leaf size and the higher percentage of emarginate leaves.
There are a number of factors besides climate change which could explain the differences in leaf morphology between the two floras. For example, some variation could be due to sampling error. As discussed previously, vegetation samples with only 10-20 species generally have large errors for the percent entire-margined species when compared to larger samples. The upper Jakokkota flora is on the small side, with only 24 species, and only 22 of these with margin information. Thus errors could be large. Presumably, errors would also be high for the other leaf morphology states, especially apex type, with only 17 species, and base and length to width ratio, with 22 species.
Another factor which could cause the morphologic scores of the two floras to vary is taphonomy. The two different levels of the Jakokkota flora represent different paleoenvironments; the lower Jakokkota flora was deposited in a low-energy stream, while the upper Jakokkota flora was deposited in a stratified lake. Studies by Greenwood (1992) and Roth and Dilcher (1978) show that depositional processes have an important affect on leaf size; they found that average leaf size drops both with increasing distance from the shore in lake deposits, and with increasing amounts of transportation in stream deposits. Less is known about the effects of taphonomic processes on other leaf morphology character states. Burnham et al. show that the percent entire-margined species varies for riparian versus terra firme vegetation, but both Jakokkota floras represent riparian vegetation, so this is unlikely to be a factor.
Given these sources of error, and the fact that the differences in climate estimates are within the error of the statistic methods, it would be an overinterpretation to suggest a climate change occurred between the two floras.
Elevation Analysis
Elevation Calculations
West Coast Comparison: At present, the elevation of fossil floras can be estimated using the enthalpy-based paleoaltimeter of Forest et al. (1995, 1999) or the MAT-based paleoaltimeter of Wolfe (1992) and Meyer (1992). The enthalpy-based paleoaltimeter cannot be applied to Bolivian floras because the atmospheric dynamics have not yet been worked out.
Thus in this study we will use the MAT-based paleoaltimeter. To estimate the paleoelevation, , of a fossil flora, one compares the paleoMAT of the flora with the paleoMAT of a coeval flora which grew near sea level then applies a "terrestrial lapse rate" after the equation of Axelrod and Bailey (1976):
(1)
where = mean annual temperature at sea level (°C); = MAT from a coeval inland site (°C); g, = "the empirical relationship between mean annual temperature at the surface and altitude" (Forest et al., 1995), equivalent to the terrestrial lapse rate of Wolfe (1992) and Meyer (1986;1992) which is and = ancient sea level relative to modern sea level (m).
According to the compilation of Hinojosa and Villagran (1997), the only coevel sea level floras from southern South America that have been studied in terms of foliar physiognomy are the Miocene Goterones and Boca Pupuya floras from the Navidad Formation of Chile (Figure 1). The age of these floras is based on foraminifera, and is between 19 and 10 Ma.
The paleoclimate of these floras is estimated to be... FELIPE - PUEDES ESCRIBIR ALGO SOBRE EL PALEOCLIMATE DE LAS FLORAS DE MATANZAS-GOTERONES Y BOCA PUPUYA PARA ESTA SECCION?
Today, the locality of the Goterones-Matanzas and Boca Pupuya floras has a MAT of around 13.3 °C, based on a linear regression of coastal MATs. Thus, the paleotemperature of the Goterones-Matanzas flora is slightly warmer than today, while the subtropical Boca Pupuya flora is significantly warmer. These paleotemperatures give some additional clues to the age of the floras.
In the present climate regime, the cold waters of the Humbolt current, which flow north along the west coast of South America, reduce temperatures along about a 30 km wide strip of the coast and the lower western slope of the Andes up to about 500 to 900 m (Johnson, 1976; Prohaska, 1973; Miller, 1976) (Fig X). Studies of planktonic foraminiferas from Peruvian transgressive sequences suggest that the Humbolt current began at least by the early Miocene and intensified around 14 Ma (Tsuchi, 1992, Ibaraki, 1997; Dunbar, 1990).
The fact that the paleotemperature of the Goterones Matanzas flora is similar to present temperatures at the same site suggests that it post-dated the intensification of the Humboldt current in the middle Miocene, and is thus dated between 10 and 14 Ma. The subtropical Boca Pupuya flora is probably dated around 16 Ma, during a climatic optimum (Barron and Baldauf). Some authors see evidence for a warm event from around 11 to 12 Ma, thus the Boca Pupuya could also data from this period, though it is difficult to reconcile such warm temperatures with the presence of strong upwelling. Thus, we suggest that the Goterones-Matanzas flora is more appropriate for comparison with the Jakokkota flora. If we assume that the modern gradient along the coast existed in the Miocene, then the coast at the latitude of Jakokkota would have had a paleotemperature of (X + 5.5°C).
Note that this paleotemperature is very similar to the paleotemperature estimated for the Jakokkota flora. However, this does not mean that the paleoelevation of the flora was at sea level. As discussed previously, because of the influence of the Humbolt current, there is a temperature inversion. The fact that the temperatures are similar suggests that the Jakokkota flora was near or slightly above the elevation of the top of the temperature inversion. Today, the average height of the inversion depends on X, Y, and varies between 500 m to 900 m at the Chile/Peru border. Thus, assuming that the temperature structure along the coast was similar to today, this method would suggest that the paleoelevation of the Jakokkota flora was around 500-900 m (Table 7).
East Coast Comparison FELIPE - HAY FLORAS DEL MIOCENO MEDIO DE BRAZIL QUE PODEMOS USAR PARA UNA COMPARACION?
Same-site Comparison: Another way to calculate the elevation is using the equation of Gregory-Wodzicki et al. (1998) that compares the paleotemperature and modern temperature measured at the same site. The problem with this approach is that the temperature difference between the modern and ancient climate is a combination of climate change due to uplift, global climate change, latitudinal continental drift, and changes in paleogeography. Thus, one is presented with the difficult task of extricating the various climate signals. In this method, one can modify eq. 1 to derive:
(3)
where Zm = the modern elevation; MATgc = the change in MAT since deposition of the fossil flora due to global climate change; MATcd = the change in MAT since the deposition of the fossil flora due to latitudinal continental drift; MATpg = the change in MAT since the deposition of the fossil flora due to changes in paleogeography; and MATm = the modern MAT.
An important factor to subtract out is MATgc, or global climate change since the deposition of the floras. In terms of temperature, the late Miocene ocean appears to have been similar to the modern ocean. Oxygen isotope data from planktic foraminifera from DSDP sites in the western equatorial pacific suggest that low-latitude surface waters cooled approximately 0-1°C since the Jakokkota flora was deposited (Savin et al., 1975; Savin, 1977); the generalized curve based on these sites and additional sites from the north pacific suggest closer to no change in tropical temperature.
In terms of the MATcd, or the change in temperature due to latitudinal continental drift, the area of the Bolivian Altiplano was ~2° latitude further south 10 Mya, and ~3.5 °latitude further south 20 Mya (Smith et al., 1981). Stable isotope data suggests that the late Miocene latitudinal gradient was about 3/4 of the modern-day gradient (Loutit et al., 1983). Thus if we take the modern temperature gradient in the eastern lowlands of Bolivia and Argentina of 0.44 °C/° latitude (Fig. 10), and reduce this by 3/4 to 0.33 °C °latitude-1 to simulate the Miocene, the area would have warmed 0.7 °C since the Jakokkota flora was deposited 10 Mya.
The most difficult factor to subtract out is MATpg , or the climate change due to local paleographic changes. This factor does not include the decrease in temperature associated with increased elevation, which is evaluated in eq. 1, but includes other effects due to the creation of topography or relief. For example, the creation of topography can contribute to orographic rainfall and latent heating, and can obstruct large-scale air flow. The MATpg term is especially difficult to evaluate because one needs information about paleoelevation, the very variable that one is attempting to estimate. For now, no correction is made for this factor.
A value of 0.43 ± 0.11 °C 100 m-1 is used for g, based on modern lapse rates observed for the area. This "terrestrial lapse rate" is lower than those observed by Meyer (1986, 1992) and Axelrod and Bailey (1976) because of the effect of elevated base level. A large elevated land surface is heated more than a column of air at the same elevation (Parrish and Barron, 1986) and thus terrestrial lapse rates are reduced relative to those in a free-air column. Independents studies suggest that the Altiplano had some elevation at this time, so it is more appropriate to use this value than the low-base level values.
This calculation suggests that the Jakokkota flora had a paleoelevation of 100-1150 m.
Discussion of Elevation Analysis
Both methods of estimating paleoelevation suggest that the Jakokkota flora had an elevation of no more than around 1150 m. This estimate is consistent with the precipitation estimates. At an elevation of 1000 m, one would expect some rainshadow effect, though probably less than the modern rainshadow.
CONCLUSIONS
1. A sample of 24 morphospecies from the upper Jakokkota flora has similar morphology to a lower level. This flora is probably between 0.2 - 0.5 Ma younger than the lower level, which is dated as 10.7 Ma.
2. The climate of the combined flora is estimated to be 23.8°C and 600 mm.
3. This translates to a paleoelevation of around 1000 m. This estimate is consistent with other estimates, and suggests that uplift of the Central Andean Plateau was young.
ACKNOWLEDGMENTS
The author was supported by U.S. National Science Foundation grant EAR-99-09114 . Many thanks to Jaime Argollo for assistance with relocating the Jakokkota site and for use of laboratory facilities at the Universidad Mayor de San Andres. Adelide Auza provided assistance in the field. This is Lamont-Doherty Earth Observatory contribution number XXXX.
REFERENCES
FIGURE CAPTIONS
Figure 1. Location map
Figure 2. Stratigraphic Section
Figure 3. Temperature of climate stations from the Central Andes
Figure 4. Leaf forms from the upper Jakokkota flora.