Paleobotanical Evidence for High Altitudes in Nevada During the Miocene

See allHide authors and affiliations

Science  13 Jun 1997:
Vol. 276, Issue 5319, pp. 1672-1675
DOI: 10.1126/science.276.5319.1672


Leaf physiognomy provides estimates of environmental parameters, including mean annual enthalpy, which is a thermodynamic parameter of the atmosphere that varies with altitude. Analyses of 12 mid-Miocene floras from western Nevada indicate that this part of the Basin and Range Province stood ∼3 kilometers above sea level at 15 to 16 million years ago, which is 1 to 1.5 kilometers higher than its present altitude. Much, if not all, of the collapse to present-day altitudes seems to have been achieved by ∼13 million years ago. The crust in much of this area has been extended and thinned throughout the past 40 to 50 million years, and the isostatic balance of a thinning crust requires subsidence, not uplift as suggested by previous paleobotanical work.

Terrestrial plants are generally regarded as highly responsive to environmental changes, and thus fossil plants offer one of the best methods for inferring paleoenvironmental parameters (1). In one approach, the environmental tolerances of a fossil species were assumed to be the same as those of its nearest living relative. Because in this method it was assumed that plants do not evolve by adapting to different environments (2), conclusions based on this method must—to varying degrees—be doubtful. Paleobotanical data using this nearest living relative method have been interpreted to indicate that most of the Basin and Range Province of Nevada was at low altitudes (<1 km) until <5 million years ago (Ma), when uplift is inferred to have started, which resulted in the ∼1- to 1.5-km present-day mean altitudes of the basins (3). These estimates have been used as boundary conditions in numerical models of global climate during the Cenozoic, and it has been suggested that recent uplift helped initiate late Cenozoic glaciation (4). A second paleobotanical method relates the general physiognomy of plants to the environment, a relation that has generally been assumed to be valid, both today and in the past (1). In particular, gross physical aspects of leaves, including outlines, shapes, and sizes that can be readily observed on fossilized leaves, can be observed to change along present-day environmental gradients (5).

To calibrate changes in foliar physiognomy with changes in environmental parameters, leaves of at least 20 species of woody dicotyledons were collected close to meteorological recording stations (5), primarily in North America and the Caribbean region from latitudes 18°N to 62°N. Included in the sampling was vegetation ranging from desert to wet tropical forest to boreal forest. Physiognomic character states determined or measured for all species in the samples were then analyzed in a multivariate context. In the Climate-Leaf Analysis Multivariate Program (CLAMP), we used canonical correspondence analysis (6, 7), a multivariate ordination method that is widely used in ecology (8) to rank samples simultaneously relative to several environmental factors (such as temperature and precipitation values) by partial constraint of the ordination axes by supplied environmental data. Because leaf physiognomy character states have typically nonlinear relations to environmental parameters (5), canonical correspondence analysis, which was developed to analyze nonlinear relations, is preferred to methods that assume linear relations (such as multiple regression analysis).

Among the environmental parameters that can be inferred from leaf physiognomy, we concentrated on mean annual temperature, specific humidity, and enthalpy (Fig. 1). Forest et al. (9) have shown that mean annual values of moist static energy, a thermodynamically conserved variable in the atmosphere, are approximately zonally invariant. The moist static energy h of an air parcel consists of three terms that quantify the total energy content (per unit of mass) of the air parcel (the negligibly small kinetic energy is excluded):Embedded Image where cp is the specific heat capacity of moist air at constant pressure, T is the absolute temperature, Lv is the latent heat of vaporization, q is the specific humidity, g is the gravitational acceleration, Z is height, and H (= cpT + Lvq) is enthalpy. With the assumption that moist static energy is zonally invariant, the difference between two estimates of mean annual enthalpy for sites at similar paleolatitudes should yield an estimate of their difference in potential energy, gZ.

Figure 1

Plot of the canonical correspondence analysis of environmental factors (arrows), modern samples (open circles), and fossil leaf assemblages (solid circles). Axis 1 (which is shown as vertical here), which explains ∼50% of the physiognomic variation, represents temperature factors; whereas axis 2, which explains ∼20% of the physiognomic variation, represents water stress. The length of the environmental vector approximates the relative significance of the environmental factor in explaining variation; enthalpy, which varies relative to altitude, is almost as significant as mean annual temperature. Other environmental factors included in the analysis but not shown here are the cold-month and warm-month mean temperatures, growing season length, precipitation during the three consecutive driest months and three consecutive wettest months during the growing season, mean growing season and mean annual precipitation, and relative humidity, all of which had shorter vectors than T and H. The modern samples analyzed exclude the subalpine samples.

Some error is introduced because moist static energy varies with longitude; this error (σh) has been estimated (9,10) to be 4.5 kJ kg−1. Additional error is introduced because the enthalpy predicted for sites in the CLAMP database differs from observed or calculated enthalpy. In canonical correspondence analysis, the standard error for the predicted enthalpy (σH) is 4.2 kJ kg−1. When the standard error is applied to two coeval sites, the combination of the two errors produces a standard error in the estimated difference in altitude of ∼760 mEmbedded Image In a previous analysis that estimated altitude for a late Eocene (∼35 Ma) flora in Colorado, Forest et al. (9) used principal components analysis, which assumes linearity, to derive direct estimates of enthalpy and of mean annual temperature (T) and specific humidity (q) for samples then composing the CLAMP database (11). Canonical correspondence analysis corroborates that H, T, and q can be estimated from leaf physiognomy, and, although q is not linearly independent of T, their dependence is not great (Fig. 1).

The floras included in our analyses occur in an area bounded by the Pacific Ocean and almost 118°W longitude and by 36°N and 42°N latitude, except for one flora at ∼45°N (Fig. 2). As shown in Table 1, the estimates of enthalpy from CLAMP for assemblages from rocks deposited 15 to 16 Ma in the Basin and Range of western Nevada imply that paleoaltitudes were 2.9 to 3.2 km. The Fallon and Chloropagus assemblages imply that paleoaltitudes were close to present altitudes and thus that the collapse might have occurred between 14 and 15 Ma; the size of these assemblages is, however, small, and thus interpretations are uncertain (12). The Chalk Hills and Aldrich Station assemblages indicate that by ∼12.5 to 13 Ma, western Nevada stood at about its present altitude (Fig.3). The trend to increasing temperature shown by the Nevada mid-Miocene floras is counter to the observations from oxygen isotope data in the marine record that global climates generally cooled between 15 and 10 Ma (13). These data also show some moderate reversals in declining temperature during the 13 to 15 Ma interval, so we offer alternative estimates of heights for the Chalk Hills and Aldrich Station, one set using the 10.5- to 11.0-Ma Neroly leaf assemblage, which formed near sea level, and the second set using the 13.1-Ma Molalla assemblage near Portland, Oregon, which was also at a low altitude. Either set places the Miocene altitudes of the Chalk Hills and Aldrich Station at no more than 400 m above or 600 m below their present altitudes.

Figure 2

Map of part of California and Nevada showing the present-day topography and the Miocene fossil sites (+) that produced the collections of leaves analyzed in this report. Numbers coordinate with those in parentheses after the assemblage names in Table 1. Not shown is Molalla, which is about 50 km southeast of Portland, Oregon, on the eastern side of the Willamette Valley.

Figure 3

Paleoaltitudinal estimates for middle Miocene leaf assemblages of western Nevada (circles) versus time. The altitude estimates for assemblages from ∼15 to 16 Ma are consistently higher than present-day altitudes (denoted by squares), whereas late middle Miocene (12 to 14 Ma) assemblages have estimates that are close to present-day altitudes.

Table 1

Environmental estimates for some California, Oregon, and Nevada Miocene fossil leaf assemblages (25). Ages given as decimals are radiometric ages (26), except for Temblor and Neroly, whose ages are based on a marine time scale. Alternative 1 uses Neroly as the low-altitude time equivalent. Alternative 2 uses Molalla in Willamette Valley as the low-altitude time equivalent. No., number; Paleolat., paleolatitude; Alt., altitude; Est., estimated. T, mean annual temperature; standard error, 1.3°C. H, enthalpy; standard error, 4.2 kJ kg−1. The numbers in parentheses refer to locations in Fig. 2. Dashes indicate that altitude was at (or near) sea level.

View this table:

The Weaverville assemblage (∼14 to 18 Ma) is based on localities from three separate basins of low altitude, which are assumed to have once been in the same depositional basin (14). The CLAMP analyses suggest that the altitude of the basin has changed little, if any, since deposition. Similarly, the Mohawk, Webber Lake (both ∼14 to 18 Ma), and Table Mountain (10.5 Ma) assemblages from the Sierra Nevada of northern California all have paleoaltitudinal estimates within 700 m of the present-day altitudes at the fossil localities. Since 14 to 18 Ma, the Sierra Nevada may have been no more than the western flank of the high plateau that constituted western Nevada; our data and interpretations are consistent with the model of uplift for the Sierra Nevada proposed by Small and Anderson (15).

The estimates of mid-Miocene altitudes of 2.9 to 3.2 km are somewhat lower than the estimates of 3 to 4 km (16), based on estimates of T derived from CLAMP and present-day terrestrial lapse rates [(TLR) which is the gradient of surface temperatures dTs/dZ measured at different surface heights, Z]. The degree to which modern TLR applies to the past is not clear (9, 10). Our independent estimates of T and elevations imply that the mid-Miocene TLR was 2.0° to 3.3°C km−1, which is about half that of the average modern worldwide TLR of 5.5°C km−1 (3) but is comparable to the present-day empirically estimated value for much of mid-latitude western North America (17).

Paleobotanical evidence supports the hypothesis that Mesozoic thrust faulting and crustal thickening built a high terrain in what is now the Basin and Range Province (18, 19). By analogy with the Andes and the Tibetan Plateau, altitudes in Nevada could have been even higher than ∼3 km before 16 Ma. Our results imply decreasing altitudes at about the time that most Basin and Range–style faulting began in the Great Basin. Thus, late Cenozoic uplift of Nevada (3) does not appear to have occurred. The drop in elevations reported here concurs with similar paleobotanical arguments for high latest Eocene elevations in the Rocky Mountains of Colorado based on nearest living relative and general physiognomic arguments (20) and on multivariate analysis of leaf physiognomy (21). Molnar and England (22) also have suggested that many previous estimates of paleoaltitudes for the early and middle Tertiary of western North America were much too low. Geophysical observations combined with theoretical considerations of the region to the south of our study area suggested high altitudes at ∼20 Ma and a subsequent collapse (23); an increasing body of data and interpretations argue for high altitudes during the Tertiary in much of western North America (24). The contention that late Cenozoic uplift of mountain ranges and plateaus throughout the world was a trigger for the onset of the Ice Age (5) should be reevaluated.

  • * To whom correspondence should be addressed. E-mail: jwolfe{at}


View Abstract

Stay Connected to Science

Editor's Blog

Navigate This Article