-
A late Miocene subtropical-dry £ora from the northernAltiplano,
Bolivia
Kathryn M. Gregory-WodzickiLamont-Doherty Earth Observatory of
Columbia University, Palisades, NY 10964-8000, USA
Received 18 October 2000; accepted 24 September 2001
Abstract
A variety of evidence suggests that the Altiplano of the Central
Andes, the second highest and largest plateau onearth, underwent
significant uplift in the late Miocene^Pliocene. The most important
datum supporting recent uplift isa collection of the 10.660 0.06 Ma
Jakokkota flora from west-central Bolivia, which implies a
paleoelevation no morethan 16000 1200 m; today the site has an
elevation of almost 4000 m. In order to test the reliability of
this estimate,the present study analyzes a new collection of the
Jakokkota flora from a lacustrine unit that is 0.2^0.5 Myr
youngerthan the previously analyzed collection from a fluvial unit.
Climate estimates based on leaf morphology for the twocollections
are statistically indistinguishable; the combined flora has a mean
annual temperature of 21.50 2.0‡C and amean annual precipitation of
5500 180 mm. The similarity of the climate estimates for the two
floras suggests thatthere was not a significant climate change
between them, nor a significant bias in the leaf morphology due to
differingtaphonomic processes. The climate estimate for the
combined flora thus presents a representative picture of the
lateMiocene climate of the northern Altiplano. If one assumes that
the climate of the tropics has not changed significantlysince the
late Miocene, as is suggested by marine isotopic data, then the
paleoclimate of the Jakokkota flora implies apaleoelevation of
11600 600 m. Thus, the Jakokkota flora supports the hypothesis of a
young age for theAltiplano. 5 2002 Elsevier Science B.V. All rights
reserved.
Keywords: Central Andes; Miocene; leaves; paleoclimatology;
Andean Orogeny
1. Introduction
The Altiplano, which is the second highest andlargest plateau on
earth and forms the heart ofthe Central Andes, is perhaps a very
young fea-ture; a variety of evidence suggests that morethan half
its modern elevation of 3700 m wascreated after the middle Miocene.
For example,
erosion rates and the deposition of evaporites sug-gest that the
Central Andes began to in£uencerainfall patterns by about 15 Ma,
and fossil £orasand erosion surface remnants suggest that the
Al-tiplano was at elevations of no more than V1500m as recently as
10 Mya (Gregory-Wodzicki,2000a).The age of the Altiplano is of
interest for sev-
eral reasons. First of all, the Altiplano has a ma-jor in£uence
on regional climate, and perhaps onglobal climate. It anchors the
location of thesouth paci¢c subtropical anticyclone,
strengthens
0031-0182 / 02 / $ ^ see front matter 5 2002 Elsevier Science
B.V. All rights reserved.PII: S 0 0 3 1 - 0 1 8 2 ( 0 1 ) 0 0 4 3 4
- 5
E-mail address: [email protected](K.M.
Gregory-Wodzicki).
PALAEO 2793 21-5-02
Palaeogeography, Palaeoclimatology, Palaeoecology 180 (2002)
331^348
www.elsevier.com/locate/palaeo
-
the austral summer anticyclone over Bolivia, andenhances
convective rainfall in the Central Andes(Meehl, 1992; Lenters and
Cook, 1995, 1997).Determining the uplift history of the Altiplanois
thus critical to understanding the evolution ofthe South American
climate and biota. Secondly,the Altiplano is tectonically a rather
enigmaticfeature: a plateau formed at a non-collisionalmargin.
De¢ning the timing of uplift can helpconstrain the dynamic
processes responsible forits formation.As of now, the most
important piece of evi-
dence for the young-Altiplano hypothesis is the10.660 0.06 Ma
Jakokkota £ora from west-cen-tral Bolivia (Fig. 1). This £ora is
the most pre-cisely dated Miocene £ora from the Central An-des, and
was deposited during the most recentphase of Andean orogeny. An
analysis of theleaf morphology of the £ora suggests that thesite,
which today has a mean annual temperature(MAT) of 8.3‡C, was
signi¢cantly warmer in thepast, with a paleoMAT of 18.6^21.00
2.5‡C. Ifone assumes that the climate of the tropics haschanged
little since the late Miocene, as suggestedby marine isotope data,
the paleotemperatureimplies a paleoelevation of 590^16100 1200
m(Gregory-Wodzicki et al., 1998; Gregory-Wod-zicki, 2000a), which
is signi¢cantly lower thanthe modern elevation of 3940 m.But does
this estimate present a representative
picture of the late Miocene climate and elevationof the
Altiplano? Collected from a £uvial horizonless than 25 cm thick,
the Jakokkota £ora is therecord of probably no more than 10 000
yr.Though short-term climate variation before theice ages was
perhaps of a smaller magnitudethan we observe during the ice ages,
it probablystill was signi¢cant; for example, marine isotopedata
suggest that in the Pliocene short-term tem-perature £uctuations
were on the order of 1.5^4‡C(King, 1996). Thus it is possible that
this collec-tion of the Jakokkota £ora may not typify the
lateMiocene paleoenvironment.This study attempts to provide a more
charac-
teristic estimate of late Miocene climate by ana-lyzing a sample
from a new horizon at the Jakok-kota locality, a lacustrine unit
found 10 m abovethe previously collected £uvial horizon.
Analysis
of the £ora of this younger horizon, here calledthe upper
Jakokkota £ora, can improve ourunderstanding of late Miocene
climate in twoways. First of all, by providing a larger sampleof
the Jakokkota £ora, it reduces the error of theclimate estimate;
several authors have shown thatthe accuracy of climate estimates
based on leafmorphology increases with the increasing numberof
species in a sample (Wolfe, 1971; Povey et al.,1994; Wilf, 1997;
Burnham et al., 2001). Sec-ondly, by sampling a di¡erent time
horizon anddepositional environment, the new collection willprovide
some measure of short-term climate var-iation and will provide some
constraints on errordue to di¡erent taphonomic processes.In this
study, the climate of the upper Jakok-
kota £ora is analyzed using the method of Wolfe(1993). First,
the leaf morphology of the upperJakokkota £ora is scored, and then
these scoresare input into models that relate leaf morphologyto MAT
and mean annual precipitation. At thismoment, it is di⁄cult to
determine which climateleaf morphology models provide the most
accu-rate estimates of climate for Bolivian paleo£oras,so the
assumption is made that the most accurateresults will be from those
models that provide themost accurate climate estimates for modern
vege-tation from Bolivia. The climate estimates for theupper
Jakokkota £ora are then compared to thosefrom a similar analysis of
the lower Jakokkota£ora. The climate of the combined £ora is
thenused to calculate the paleoelevation at which itgrew.
2. Geology and age of the Jakokkota £ora
The Jakokkota £ora is located in the northernAltiplano of
Bolivia (Fig. 1), in member 6 of theMiocene Mauri Formation (Sirvas
and Torres,1966; Suarez Soruco and Diaz Martinez, 1996).Leaf
impressions are found in two di¡erent units.The lower Jakokkota
£ora, which was discoveredby Berry (1922b) and further described by
Greg-ory-Wodzicki et al. (1998), is found in a white,ash-rich
claystone to ¢ne-grained sandstone, in-terpreted to be a £uvially
reworked ash deposit(Fig. 2).
PALAEO 2793 21-5-02
K.M. Gregory-Wodzicki / Palaeogeography, Palaeoclimatology,
Palaeoecology 180 (2002) 331^348332
-
The upper Jakokkota £ora is found in an 8 mthick sequence of
reddish-tan to green laminatedsandstones, siltstones, and mudstones
that weredeposited 10 m above the lower Jakokkota £ora(Fig. 2).
Individual laminae range from 1 mm to3 cm in thickness, and grade
from siltstone tocoarse sandstone at the base to clay at the
top.The laminated bedding and graded layers suggestthat these
deposits are lacustrine.A fall ash located 3 m above the lower £ora
has
an age of 10.660 0.06 Ma, based on single-crystallaser fusion
analysis of sanidine, while a fall ashjust below the upper
Jakokkota £ora has less pre-cise ages of 11.35 0 0.69 Ma on biotite
and12.740 0.69 Ma on hornblende (Gregory-Wod-zicki et al., 1998)
(Fig. 2). A fall ash 29 m abovethe upper Jakokkota £ora has ages of
10.810 0.72
Ma (biotite) and 11.310 0.53 Ma (hornblende)(Fig. 2). The low
precision of the biotite andhornblende ages probably re£ects
alteration;these minerals alter much more quickly than sa-nidine.
Thus the 10.660 0.06 Ma age is consideredthe most accurate and
reliable age derived fromthe ash falls.The upper Jakokkota £ora is
probably not sig-
ni¢cantly younger than the lower Jakokkota £ora.The only
physical sign of a depositional hiatus, anunconformity just above
the lower Jakokkota£ora (Fig. 2), is associated with £ame
structures,which suggest soft-sediment loading and thus onlya minor
time lapse. If the biotite and hornblendeages of the two fall ashes
are compared, they sug-gest an average sedimentation rate of
around20^50 m/Ma. This would suggest that the 10 m
Fig. 1. (A) Relief map of the Central Andes (USGS 30 arc-second
DEM data as processed by the Cornell Andes Project) show-ing
Miocene fossil £oras: 1, Los Litres; 2, Chucal; 3, Potos|¤; 4,
Goterones and Boca Pupuya; 5, Jakokkota; 6, Pislepampa.(B)
Morphotectonic provinces of the Central Andes, showing the location
of the Altiplano and Eastern Cordillera (after Jordanet al., 1983).
The heavy line indicates the location of the continental
divide.
PALAEO 2793 21-5-02
K.M. Gregory-Wodzicki / Palaeogeography, Palaeoclimatology,
Palaeoecology 180 (2002) 331^348 333
-
di¡erence between the lower and upper £oraswould represent
approximately 0.2^0.5 Ma.
3. Materials and methods
3.1. Collection and description
Fossil leaves and leaf fragments were collectedfrom two
locations in the lacustrine facies, with acombined volume of
approximately 1 m2. Whendry, these sediments are characterized by
perva-sive crackle fracturing. In order to preserve thecollected
fossils, most specimens were painted onthe base and sides with a
clear plastic coating.Once in the lab, the specimens were
photographedand split into morphospecies based on
venationcharacteristics. Identi¢cations were made usingcomparisons
to modern herbarium material.The morphology of the leaves in each
morpho-
species was scored after the method of Wolfe(1993). In order to
be consistent with the scorefor the lower Jakokkota £ora, leaves
that werevery close in size to the next larger size categorywere
scored in both categories in order to com-pensate for the size
reduction observed betweencanopy and litter samples (Greenwood,
1992;Gregory and McIntosh, 1996). The morphospe-cies scores were
summed and divided by the totalnumber of morphospecies to derive a
site score.Further details about the scoring method are giv-en in
Table 1 and in Wolfe (1993).
3.2. Estimating paleoclimate
The climate of the upper Jakokkota £ora wascalculated using
models based on relationshipsobserved in modern vegetation between
leaf mor-phology and climate, such as the increase in thepercentage
entire-margined species with increasingtemperature and the increase
in leaf size with in-creasing precipitation (Fig. 3). A large
number of
Fig. 2. Generalized stratigraphic column of fossil
locality,showing the lower and upper Jakokkota £oras (leaf
sym-bols). Radiometric ages from Gregory-Wodzicki et al.
(1998).
PALAEO 2793 21-5-02
K.M. Gregory-Wodzicki / Palaeogeography, Palaeoclimatology,
Palaeoecology 180 (2002) 331^348334
-
such models exist, which vary both in terms of thedata set and
statistical method used, and there hasbeen much discussion in the
literature aboutwhich provide the most accurate climate
estimates(Wolfe, 1995; Jordan, 1997; Stranks and Eng-land, 1997;
Wilf, 1997; Wilf et al., 1998, 1999;Wolfe and Uemura, 1999;
Gregory-Wodzicki,
2000b; Kowalski, 2001; Greenwood et al., in re-view; Jacobs, in
review).Choosing an appropriate predictor data set is
of primary importance, because the closer its re-lationships
between leaf morphology and climateto those of the site to be
analyzed, the more ac-curate the climate estimates will be
(Gregory-
Table 1Morphologic character state scores for the lower, upper,
and combined Jakokkota £oras
LMCS Lower J. score Upper J. score Comb(%) (%)
1. TLob 1.6 0.0 1.02. NoT 71.0 77.3 73.93. TRg 17.7 15.9 16.34.
TCl 8.1 11.4 8.75. TRnd 16.1 11.4 14.15. TAct 12.9 11.4 12.06. TCmp
0.0 0.0 0.07. Size: Nan 7.6 8.3 7.5
Le1 14.1 14.6 13.9Le2 33.3 38.9 37.8Mi1 39.1 32.6 36.1Mi2 6.0
5.6 4.7Mi3 0.0 0.0 0.0Me1 0.0 0.0 0.0Me2 0.0 0.0 0.0Me3 0.0 0.0
0.0
8. AEmg 4.0 11.8 5.69. Apex: ARnd 54.0 70.6 59.7
AAct 46.0 29.4 40.3AAtn 0.0 0.0 0.0
10. Base: BCd 1.2 2.3 1.9BRnd 28.0 68.2 42.8BAct 70.8 29.5
55.3
11. L:W ratio: LW6 1:1 0.0 2.3 1.1L:W 1^2:1 21.1 25.0 21.4L:W
2^3:1 40.6 20.5 36.2L:W 3^4:1 20.6 22.7 22.1L:W s 4:1 17.8 27.3
19.2
12. Shape: SOb 19.4 11.6 17.0SElp 64.4 79.0 69.1SOv 16.1 9.4
13.8
Numbers in LMCS (leaf morphology character state column) denote
categories. Some categories have only two character states,for
example ‘lobed’ and ‘not lobed’ are in the category ‘lobed’. For
simplicity, only one character state is usually reported.
‘teethacute’ and ‘teeth round’ are an exception, as both are
reported. Other categories, such as size, contain several character
statesand all are reported. Quanti¢cation of physiognomic score for
given leaf form: (1) if present, character state receives score of1
divided by number of character states present for form in that
category; (2) if absent, character scored 0; (3) if partly
present,scored as 0.5 divided by number of character states in
category. Form scores then added for each character state and
divided bytotal number of forms to derive physiognomic score. See
Wolfe (1993) for further details of scoring and de¢nitions.
Abbrevia-tions: TLob, teeth lobed, NoT, no teeth; TRg, teeth
regularly spaced; TCl, teeth closely spaced; TRnd, teeth round;
TAct, teethacute; TCmp, teeth compound; Le1,2, leptophyllous 1,2;
Mi1,2,3, microphyllous 1,2,3; Me1,2, mesophyllous 1,2; AEmg,
apexemarginate; ARnd, apex round; AAct, apex acute; AAtn, apex
attenuate; BCd, base cordate; BRnd, base round; BAct, baseacute;
L:W, length to width ratio; Sob, shape obovate; SElp, shape
elliptical; SOv, shape ovate.
PALAEO 2793 21-5-02
K.M. Gregory-Wodzicki / Palaeogeography, Palaeoclimatology,
Palaeoecology 180 (2002) 331^348 335
-
Fig. 3. (A) Mean annual temperature vs. percentage
entire-margined species for di¡erent data sets: Bolivia+Peru
(composed ofthe Bolivian sites from Gregory-Wodzicki, 2000b and two
samples from Bolivia and Peru from Wilf, 1997); East Asia (datafrom
caliper measurements by D.R. Greenwood of data published in Wolfe,
1979); CLAMP 3B; subalpine sites from CLAMP3A; and Africa. See
Table 2 for sources of data. Regression lines: Solid line,
Bolivia+East Asia; dashed line, CLAMP 3B.(B) Mean annual
precipitation vs. percentage species with Mesophyllous
1+Mesophyllous size 2 leaves for di¡erent data sets: Bo-livia,
CLAMP 3B mid-latitude sites (s 24‡ latitude), CLAMP 3B ^ tropical
(6 24‡ latitude), and Africa. See Table 2 for datasources.
Regression lines: Solid line, Bolivia+Africa; dashed line, CLAMP 3B
^ tropical.
PALAEO 2793 21-5-02
K.M. Gregory-Wodzicki / Palaeogeography, Palaeoclimatology,
Palaeoecology 180 (2002) 331^348336
-
Wodzicki, 2000b; Jacobs, in review). The maindata sets that have
been collected are listed inTable 2. The most extensive is the
Climate LeafAnalysis Multivariate Program (CLAMP) dataset of Wolfe
(1993, 1995), which includes 173 sitesmostly from North America and
Asia, scored for31 di¡erent leaf morphology character states.Other
data sets contain sites from equatorial Afri-ca, Ecuador, East
Asia, New Zealand/Australia,and Bolivia.Several studies suggest
that the relationships
between leaf morphology and climate are not uni-versal, but vary
from region to region (Wolfe,1993; Stranks and England, 1997;
Jacobs, in re-view). For example, subalpine vegetation appearsto
have a di¡erent relationship between leaf mar-gin and temperature
than vegetation from warmerclimates (Fig. 3). The three major leaf
morphol-ogy domains identi¢ed so far are 1: North Amer-ica,
Caribbean, Japan, Bolivia, 2: Australia andNew Zealand, and 3:
subalpine zones (Wolfe,1993; Kennedy, 1998;
Gregory-Wodzicki,2000b). Jacobs (in review) and Kowalski (2001)
suggest that the ¢rst domain should be furtherdivided into
tropical and mid-latitude subsets;they ¢nd that the CLAMP data set,
which hassites predominately from mid-latitudes, is inap-propriate
for estimation of temperature and pre-cipitation for sites from
equatorial Africa andequatorial South America.Choosing an
appropriate data set for the Ja-
kokkota £ora is complicated by the fact that wedo not yet
understand why there are di¡erentcorrelations between climate and
leaf morphologyin these di¡erent domains; the di¡erences ob-served
could be due to di¡erences in the climate,environmental conditions,
£oristic composition,or scoring style. For example, the
di¡erencesthat Kowalski (2001) observes between CLAMPsamples and
her samples from Ecuador could bedue to the in£uence of cold season
temperatureson the higher latitude CLAMP sites (Jacobs, inreview),
or could be due to di¡erent samplingstrategies; Kowalski (2001)
sampled herbariumspecimens that were mostly from trees s 5
cmdiameter at breast height, while CLAMP samples
Table 2Leaf morphology data sets in the literature
Data set N Character states Sampling strategy Source
CLAMP 3Aa 173 31 of Wolfe CLAMP Wolfe, 1993, 1995CLAMP 3Ba 144
31 of Wolfe CLAMP Wolfe, 1993, 1995Western Hemisphere, Africa 50
MlnA primarily data from £oral
manuals, primarily all woodydicots
Wilf et al., 1998
East Asia 34 NoT ^ Wolfe, 1979Equatorial Africa 30 15 of Jacobs,
31 of Wolfe herbarium samples, primarily
all woody dicots, thoughsome samples lack lianas
Jacobs, 1999; Jacobs, inreview
New Zealand 30 31 of Wolfe CLAMP Stranks and England,
1997;Kennedy, 1998
Ecuador 30 29 of Wolfe herbarium samples, primarilyonly trees s
5 cm dbh
Kowalski, 2001
SE Australia, New Zealand 13 31 of Wolfe CLAMP Jordan,
1997Bolivia 12 31 of Wolfe CLAMP Gregory-Wodzicki, 2000bAustralia 8
NoT leaf litter Greenwood, 1992Western Hemisphere 7 NoT live
foliage, all woody dicots Wilf, 1997
Abbreviations: N, number of sites; MlnA, mean of the natural log
leaf area; NoT, no teeth (entire-margined).a CLAMP refers to the
Climate Leaf Analysis Multivariate Program data set of Wolfe (1993,
1995). CLAMP 3A refers to the
most recent version of the data set, which has percent
occurrence data for 31 di¡erent leaf morphology character states
from 173sites. Most CLAMP 3A sites are from North America and
Japan, with some sites from Caribbean and South Paci¢c
islands.CLAMP 3B is a subset of CLAMP 3A that excludes 29 subalpine
sites; these sites are known to be outliers (Wolfe, 1993). Inthe
CLAMP sampling strategy, live foliage from at least 30 species of
woody dicotyledons is collected from an area of 1^5 hec-tares,
generally in a riparian zone. See Wolfe (1993) for more
details.
PALAEO 2793 21-5-02
K.M. Gregory-Wodzicki / Palaeogeography, Palaeoclimatology,
Palaeoecology 180 (2002) 331^348 337
-
are from living foliage from all woody dicotyle-dons from a
limited area. For now, it is probablybest to compare the fossil
Bolivian £oras withdata sets that provide the most accurate
climateestimates for a collection of 12 modern foliagesamples from
Bolivia, namely the Bolivia andequatorial Africa data sets
(Gregory-Wodzicki,2000b; Jacobs, in review).After one has chosen an
appropriate predictor
data set for a fossil £ora, then one must choose atype of
statistical analysis. Methods that havebeen proposed include:
linear regression analysis(Wolfe, 1971; Wilf, 1997; Wilf et al.,
1998), multi-variate regression analysis (Wing and Greenwood,1993;
Gregory and McIntosh, 1996; Wiemann etal., 1998; Jacobs, 1999),
canonical correspon-dence analysis (Wolfe, 1995), and
nearest-neigh-bor analysis (Stranks and England, 1997). A studyby
Gregory-Wodzicki (2000b) suggests that multi-ple regression
analysis tends to produce the mostaccurate estimates for small data
sets with a nar-row range of environmental variation that
havesimilar relationships to the £ora to be analyzed,and linear
regression or canonical correspondenceanalyses produce the most
accurate estimateswhen using the larger and more varied CLAMPdata
set. If a similar predictor data set is notavailable, then
nearest-neighbor analysis can stillproduce accurate paleoclimate
estimates.Wilf (1997) advocates linear regression analysis,
arguing that the percent entire-margined speciescharacter state
explains most of the variation intemperature, and the size
character state explainsmost of the variation in precipitation,
thus theadditional character states in the CLAMP dataset only add
noise. While it is indeed true that
these character states explain a large part of thevariation,
studies by both Gregory-Wodzicki(2000b) and Jacobs (1999; in
review) show thatadditional character states can improve
results.More study is needed on which character statesprovide
useful information.For the paleoclimatic analysis of the
Jakokkota
£oras, this study uses the database/statisticalmethod
combinations that produced the most ac-curate estimates for 12
modern foliage samplesfrom Bolivia (Gregory-Wodzicki, 2000b).
Thesemodels are listed in Table 3, and include a multi-ple
regression analysis based on the Bolivia data-base for MAT, a
multiple regression analysisbased on the Bolivia database plus the
Africa da-tabase of Jacobs (in review) for mean annual
pre-cipitation, and a linear regression based on theBolivian
database for mean growing season pre-cipitation.
3.3. Estimating paleoelevation from paleoclimate
Because of the cooling of the earth’s atmo-sphere with
increasing elevation, one way to esti-mate the paleoelevation of a
fossil £ora locality isto compare its mean annual temperature
(MAT)to the MAT at sea level, and then apply a terres-trial lapse
rate. Another approach uses variationsin moist enthalpy, a climatic
variable that is afunction of temperature and relative and
speci¢chumidity (Forest et al., 1995, 1999). However, thislatter
method has only been veri¢ed for NorthAmerica, so this study will
use the temperature-based paleoaltimeter.To estimate the
paleoelevation of the Jakokko-
ta £ora using this method, one needs to correct
Table 3Equations used to estimate the paleoclimate of the
Jakokkota £ora
Method Data set Equation F SE r2
Mean annual temperature (MAT)MRA Bolivia
0.306(NoT)30.15(Mi1)+4.42 239 0.7 98Mean annual precipitation
(MAP)LRA Bolivia+Africa exp(6.302+1.354(Me1+2)) 188 180 82Mean
growing season precipitation (MGSP)LRA Bolivia 3.79(Me1)+58.2 22
160 66
Models listed produced the most accurate estimates for modern
Bolivian £oras of Gregory-Wodzicki (2000b). Method: LRA, lin-ear
regression analysis; MRA, multiple regression analysis. Equation:
See Table 1 for abbreviations of character states.
PALAEO 2793 21-5-02
K.M. Gregory-Wodzicki / Palaeogeography, Palaeoclimatology,
Palaeoecology 180 (2002) 331^348338
-
the sea level MAT for factors such as global cli-mate change and
continental drift that could havecaused the temperature to change
since the lateMiocene. One way to do this is to calculate sealevel
MAT from a coastal paleo£ora that is coevalto the Jakokkota £ora,
after the method of Ax-elrod and Bailey (1976), Meyer (1992) and
Wolfe(1992):
Zp ¼MATpc3MATpi
Qþ Sp ð1Þ
where Zp = paleoelevation (m); MATpc = paleomean annual
temperature at sea level (‡C);MATpi = paleoMAT from a coeval inland
site(‡C); Q= ‘the empirical relationship betweenmean annual
temperature at the surface and alti-tude’ (Forest et al., 1995),
equivalent to the ter-restrial lapse rate of Wolfe (1992) and
Meyer(1992) and Sp = paleo sea level relative to modernsea level
(m). However, this method cannot beused for the Jakokkota £ora,
because there areno early late Miocene sea level £oras from
theCentral Andes that have been analyzed in termsof their
quantitative paleoclimate.
Alternatively, one corrects the sea level MATby adjusting the
modern sea level MAT for anychanges in temperature since the late
Mioceneafter the following equation:
Zp ¼ðMATmc3vMATgc3vMATcdÞ3MATpi
Qþ Sp
ð2Þ
where MATmc = the modern sea level MAT, ob-served or projected
(‡C); and vMATgc,vMATcd = the change in MAT for the sea levelsite
since deposition of the fossil £ora due to glob-al climate change
and latitudinal continental drift,respectively
(‡C).Gregory-Wodzicki et al. (1998) modi¢ed Eq. 1
in a slightly di¡erent way; instead of correctingthe modern sea
level MAT for post-Jakokkotaclimate change, they corrected the
modern MATat the Jakokkota site. But the above equation iseasier to
use, because values for Q, the terrestriallapse rate, are typically
calculated in reference tocoastal or low-elevation sites (see
below).The global climate change term (vMATgc) for
the late Miocene to the present is apparently rel-
Fig. 4. Elevation vs. mean annual temperature (MAT) for modern
climate stations in the Altiplano region of the Central
Andes(15^24‡S), projected to the latitude of the Jakokkota £ora
(17‡10P). Circles represent climate stations from the Altiplano,
EasternCordillera, and eastern lowlands, while triangles represent
sites on the west coast. The projected sea level temperature is
calcu-lated by solving the regression equation for an elevation of
0 m. MATs calculated from climate data from Vose et al., 1992.
PALAEO 2793 21-5-02
K.M. Gregory-Wodzicki / Palaeogeography, Palaeoclimatology,
Palaeoecology 180 (2002) 331^348 339
-
atively minor; marine isotope data for tropicallatitudes suggest
that there has been either a cool-ing or warming on the order of
1‡C in sea surfacetemperatures since the late Miocene (Savin,
1977;Savin et al., 1985; Savin and Woodru¡, 1990). Asfor the
continental drift term (vMATcd), platetectonic reconstructions of
Smith et al. (1981) sug-gest that the Central Andean area was V2‡
lat-itude further south 10 Mya (Smith et al., 1981).Marine stable
isotope data suggest that the lateMiocene latitudinal gradient was
about 3/4 of themodern-day gradient (Loutit et al., 1983). Thus
ifwe take the modern temperature gradient in theeastern lowlands of
Bolivia and Argentina of0.44‡C/‡ latitude (Gregory-Wodzicki et
al.,1998), and reduce this by 3/4 to 0.33‡C/‡ latitudeto simulate
the Miocene, the area would havewarmed 0.7‡C since the Jakokkota
£ora was de-posited 10 Ma. According to the eustatic curvesof Haq
et al. (1987), sea level at 10.7 Ma (Sp) wasabout 50 m higher than
today.Meyer (1992) and Wolfe (1992) have compiled
terrestrial lapse rates, with the intent of providingan average
value for Q for use in Eq. 1. Meyer(1992) found a mean value of 5.9
0 1.1‡C/km forclimate stations from 39 areas of high topo-graphic
relief from around the world, while Wolfe(1992) found a
signi¢cantly lower average value of
3.0‡C/km for hundreds of climate stations fromthe western
US.This large di¡erence is due to the di¡erent
methodologies used in these compilations for cal-culating
terrestrial lapse rate. Meyer (1992) per-formed linear regressions
between stations in arestricted geographical area, preferably an
areawith high relief, while Wolfe (1992) comparedthe MAT of each
inland climate station to theMAT of the west coast of the US at
that samelatitude. MATs along the west coast are lowerthan would be
expected for their latitude becauseof the upwelling of cold water
o¡shore. Thus,comparing these unusually cool coastal sites tothe
inland sites produced lower terrestrial lapserates than the linear
regressions of Meyer (1992).Meyer (1992) observed a standard error
for his
database of 1.1‡C/km. A large part of this varia-tion is due to
topography. In a column of free air,the average moist adiabatic
lapse rate is 6.0‡C/km. For an isolated peak, surface air mixes
withthe surrounding air, and thus temperatures for agiven elevation
tend to be fairly similar to those inthe column of free air, though
with variation dueto albedo, local topography, and the nature
andsource of the air mass (Meyer, 1992). However,the temperature
for a given elevation from abroad area of high elevation tends to
be higher
Table 4Terrestrial lapse rates for di¡erent physiographies
Region Elevation range TLR Source(m) (‡C/km)
Sea level vs. elevated base levelTexas^New Mexico 2^2652 5.3
Axelrod and Bailey, 1976Western US 2^3460 3.0a Wolfe, 1993Eastern
Mexico NR 4.7 Meyer, 1992Eastern Bolivia^Altiplano 135^4071 4.4
this studyBrazil 40^2195 5.5 Axelrod and Bailey, 1976Tibetan
Plateau SL^4000 1.4^2.9b Wolfe, 1993Elevated base level vs.
isolated peaksArizona (eastern) 670^2438 7.6 Meyer, 1992Arizona
(Grand Canyon) 759^2560 6.8 Meyer, 1992New Mexico (northern)
1705^2484 7.8 Meyer, 1992New Mexico (southern) 1277^2652 7.9 Meyer,
1992Utah 1423^2652 6.8 Meyer, 1992Colorado (southwestern) 1958^2841
8.1 Meyer, 1992
TLR, terrestrial lapse rate.a This value is low because inland
sites were compared to coastal sites a¡ected by upwelling.b These
sites are from river valleys, which tend to have lower lapse rates
than more open areas.
PALAEO 2793 21-5-02
K.M. Gregory-Wodzicki / Palaeogeography, Palaeoclimatology,
Palaeoecology 180 (2002) 331^348340
-
than that in a column of free air, because theelevated surface
acts as a heat source.Thus, if lowland sites are compared with
up-
land sites from an elevated base level, then lapserates tend to
be low. For example, comparing theeastern lowlands of Bolivia with
the Altiplanosuggests a lapse rate of 4.4‡C/km, and
comparinglowland sites in eastern Mexico with sites fromthe
elevated interior gives a lapse rate of 4.7‡C/km (Meyer, 1992)
(Fig. 4, Table 4). On the otherhand, comparing lowland sites from
an elevatedbase level with highland sites from isolated peaksgives
high terrestrial lapse rates. For example,lapse rates for the
Colorado Plateau, RockyMountain area of the Western US have
lapserates between 6.8 and 8.1‡C/km (Table 4).In terms of choosing
a terrestrial lapse rate for
Eq. 2, it is probably most appropriate to use themodern lapse
rate of 4.4‡C/km between the east-ern lowlands of Bolivia and the
Altiplano. Theconstant of this regression represents the MATat 0 m
elevation, or the ‘projected sea level tem-perature’ of Meyer,
1992, and can be used to rep-resent sea level MAT in Eq. 2.Using
the average lapse rate of Meyer of
5.9 0 1.1‡C/km would not be appropriate, as thisvalue is
calculated for lowland areas of elevatedbase level compared to
highland areas from iso-lated peaks. The Jakokkota £ora did not
grow onan isolated peak; it grew in an area of low
relief.Therefore, the terrestrial lapse rate for this regionhas
probably been lower than Meyer’s averagevalue ever since the
Altiplano was elevated abovesea level. Wolfe (1994) observes that
500^1000 mof elevation is enough to create the elevated baselevel
e¡ect.Another approach for choosing a lapse rate for
Eq. 2 would be to use the modern terrestrial lapserate between
the west coast of South America andthe Altiplano. However, like the
sites from thewest coast of the US, sites along the west coastof
South America are unusually cool for theirlatitude because of the
cold-water Humboldt cur-rent (Fig. 4). Using this comparison would
neces-sitate correcting for the evolution of ocean circu-lation in
the southeastern Paci¢c. Thus, itprobably involves less error to
use the terrestriallapse rate for the eastern slope of the
Andes.
It is probably reasonable to assume that lapserates have not
changed signi¢cantly through time,at least since the Altiplano
attained between 500and 1000 m of elevation. For example, Wolfe
etal. (1997) analyzed the paleoelevation of 14 mid-dle and late
Miocene £oras from California andNevada using the enthalpy method
of Forest et al.(1995, 1999), which does not rely on lapse rate,and
found an average error of 0.66‡C/km betweenmodern lapse rates and
Miocene lapse rates. Thiserror is surprisingly low considering that
duringthis period the Great Basin collapsed to abouthalf its former
elevation.Formal errors for the elevation calculation are
a combination of the errors for the MAT estimatefor the
Jakokkota £ora, the estimates of MATchange due to global climate
change and conti-nental drift, and the terrestrial lapse rate.
Forthe present day, the lapse rate term probablyhas an error of
around 0.5‡C/km, based on thestandard deviation for the lapse rates
from areasof elevated base level in Table 4. However, theerror of
0.7‡C/km observed by Wolfe et al.(1997) for the Miocene of the
western US is prob-ably more appropriate when using this
equationfor the Miocene.
4. Results
4.1. Floristic composition and leaf morphology
The fossil leaf impressions from the upper Ja-kokkota £ora were
split into 24 morphospecies(Fig. 5). The most abundant form was
Anacardia-ceae sp. 1, which made up almost half of thecollected
specimens (Table 5). This same formwas also quite common in the
lower Jakokkota£ora, making up almost a 10th of the fossil forms.A
second species of Anacardiaceae, this onetoothed, was less common.
Today, this family ismostly found in the tropics and subtropics
(Kil-leen et al., 1993).A species of Berberis is present in the
upper
Jakokkota £ora, although it appears to be a dif-ferent
morphospecies than the Berberis encoun-tered in the lower Jakokkota
£ora. Today, thereare 28 species in Bolivia, which are found in
mon-
PALAEO 2793 21-5-02
K.M. Gregory-Wodzicki / Palaeogeography, Palaeoclimatology,
Palaeoecology 180 (2002) 331^348 341
-
tane forest and cloud forest, ranging from 1700 to3800 m
(Killeen et al., 1993).At least ¢ve legumes are present, as
evidenced
by the striated pulvinuses on the fossil leaves. Le-gumes were
also common in the lower Jakokkota£ora, though only one form, form
6, appears to beshared by both £oras. This family is common inthe
dry subtropical forests of eastern Bolivia. OneMyrtaceae is
present; this is an important familyin subtropical environments in
South America.
Zizyphus sp. was another common form. To-day, there are four
species in Bolivia; they range
from 250 to 850 m, and are found both in xericand humid
forests.The specimens of Polylepis sp. were very poorly
preserved, but the teeth were identical to betterpreserved
specimens in the lower Jakokkota £ora.Today this genus is found
from elevations be-tween 1700 and 5200 m, in xeric to humid
envi-ronments. In total, seven of the 24 morphospeciesin the upper
Jakokkota £ora were also encoun-tered in the lower Jakokkota £ora
(Table 5).The leaf morphologic score for the upper Ja-
kokkota £ora is given in Table 1, along with the
Fig. 5. Leaf morphospecies from the upper Jakokkota £ora: (a)
Polylepis sp. ; (b) Form 18; (c) Leguminosae (Form 6); (d)
Legu-minosae (Form 62); (e), (f) Zizyphus sp.; (g) Form 8; (h)
Anacardiaceae sp. 1; (i) Form 25; (j) Form 48; (k) Form 51; (l)
Form54; (m) Form 55; (n) Form 58; (o) Leguminosae (Form 59); (p)
Anacardiaceae sp. 2; (q), (r) Leguminosae (Form 57); (s) Form53;
(t) Form 52; (u) Leguminosae (Form 61); (v) Myrtaceae; (w) Form 68;
(x) Form 64; (y), (z) Berberis sp. ; (aa) Leguminosae(Form 60).
PALAEO 2793 21-5-02
K.M. Gregory-Wodzicki / Palaeogeography, Palaeoclimatology,
Palaeoecology 180 (2002) 331^348342
-
score for the 32 morphospecies of the lower Ja-kokkota £ora and
the 48 morphospecies of thecombined £ora. The score of the lower
Jakokkota£ora is slightly modi¢ed from that given in
Greg-ory-Wodzicki et al. (1998); one fragmentaryspecimen that was
not considered a separate mor-phospecies was reclassi¢ed based on
the discoveryof numerous similar material in the upper Jakok-
kota £ora. In morphologic space, the fossil £orasare most
similar to modern microphyllous scrubfrom the dry interandean
valleys of Bolivia.The scores for margin characteristics vary
somewhat between the two levels ; the percentageof
entire-margined species is about 6% higher inthe upper Jakokkota
£ora than in the lower. Ingeneral, the size distribution of the two
£oras issimilar; both have mostly small leaves, with nonelarger
than the microphyllous 2 category, thoughthe average size of the
upper Jakokkota £ora isslightly smaller. The distribution of length
towidth ratios is di¡erent, with the upper Jakokkota£ora having
more species in the equant and elon-gate categories than the lower
Jakokkota £ora,which is dominated by leaves in the L:W
2^3:1category.The most notable di¡erences occur in the apex
and base categories. The lower Jakokkota £ora isdominated by
species with acute bases, and hassubequal numbers of species with
acute androunded apices. On the other hand, the upperJakokkota £ora
is dominated by species withround bases and round apices.
4.2. Climate analysis
When the leaf morphologic scores for the Ja-kokkota £oras are
plugged into the regressionmodels in Table 3, the results
consistently suggesta subtropical-dry climate (Table 6). MAT
esti-mates range from 20.1 0 0.7‡C for the lower £orato 23.0 0
0.7‡C for the upper £ora. The combined£ora has a MAT of 21.5 0
0.7‡C. These MATs areconsiderably warmer than the modern MAT
of8.3‡C for the Charan‹a station, which is 60 kmto the southwest of
Cerro Jakokkota (Vose etal., 1992).The formal error given for these
estimates is
probably too low, due to the small size of the
Table 6Climate estimates for the lower, upper, and combined
Jakokkota £oras
Climate variable Lower Upper Combined
MAT (‡C) 20.10 2.5 23.00 2.7 21.50 2.0MAP (mm) 5500 180 5500 180
5500 180MGSP (mm) 5800 160 5800 160 5800 160
Estimates calculated using the equations listed in Table 3.
Table 5Floristic composition of the upper Jakokkota £ora
Form # sp. %
Anacardiaceaesp. 1a 313 47.1sp. 2 3 0.5BerberidaceaeBerberis sp.
29 4.4LeguminosaeLegume (6)a 77 11.6Legume (57) 49 7.4Legume (59)
31 4.7Legume (60) 7 1.1Legume (62) 18 2.7Myrtaceae undet. 1
0.2RhamnaceaeZizyphus sp.a 33 5.0RosaceaePolylepis sp.a 20
3.0Incertae cedisForm 8a 2 0.3Form 18a 3 0.5Form 25a 18 2.7Form 48a
5 0.8Form 51 7 1.1Form 52 15 2.3Form 53 3 0.5Form 54 13 2.0Form 55
1 0.2Form 58 1 0.2Form 61 5 0.8Form 64 9 1.4Form 68 1 0.2
sp., number of specimens.a Also present in lower Jakokkota
£ora.
PALAEO 2793 21-5-02
K.M. Gregory-Wodzicki / Palaeogeography, Palaeoclimatology,
Palaeoecology 180 (2002) 331^348 343
-
modern Bolivia data set; as more samples areadded to the
database, the error will probablyincrease. Wilf (1997) shows that
for samples ofless than about 75 species, the error associatedwith
using a small sample of fossil morphospeciesto represent a large
£ora is greater than the aver-age formal error, and is thus a
better approxima-tion of the actual error. Using the equation
ofWilf (1997), the sampling error for the lower Ja-kokkota £ora
with 32 species is 2.5‡C, for theupper Jakokkota £ora with 24
species is 2.7‡C,and for the combined £ora with 48 species is2.0‡C
(Table 6). Given these errors, the MATsestimated for the di¡erent
£oras are statisticallyindistinguishable.Estimates of the mean
annual precipitation
(MAP) and mean growing season precipitationsuggest a xeric
climate. MAP is estimated as5500 180 mm and the mean growing season
pre-cipitation as 5800 160 mm for the upper, lower,and combined
£oras (Table 6). The estimates forthese variables are the same for
all of the £orasbecause the linear models rely on the mesophyl-lous
1 and 2 leaf size categories, and none of the£oras has leaves of
these sizes. The estimates ofmean growing season precipitation are
probably agood approximation of the MAP, because thegrowing season
for this variable is de¢ned byWolfe (1993) as the number of months
with amean monthly temperature s 10‡C. The warmpaleotemperatures of
the Jakokkota £ora suggestthat the growing season, so de¢ned, was
year-round. Today, the MAP of the Charan‹a stationis around 300 mm
and the mean growing seasonprecipitation is 150 mm, with a marked
dry sea-son during the winter (Vose et al., 1992).
The assumption that the Jakokkota £ora wouldhave had
relationships between leaf morphologyand climate most similar to
the modern Bolivian£oras is supported by the fact that the
nearestneighbors to the Jakokkota £oras in morphologicspace, as
calculated by canonical correspondenceanalysis, are Bolivian
samples.
4.3. Elevation analysis
When the MAT of the combined £ora alongwith the values for
global climate change, conti-nental drift, sea level change, and
terrestrial lapserate discussed above are plugged into Eq. 2,
apaleoelevation of 11600 600 m is calculated (Ta-ble 7). This
paleoelevation is considerably lowerthan the modern elevation of
the Jakokkota siteof 3940 m. Because this paleoelevation is
abovethe 500^1000 m needed to produce an elevatedbase level e¡ect,
it is probably reasonable to usethe modern terrestrial lapse rate
in the calcula-tion.This paleoelevation estimate for the
combined
Jakokkota £ora is comparable to the paleoeleva-tion estimated by
Gregory-Wodzicki et al. (1998)for the lower Jakokkota £ora of
590^16100 1200m. The error is less for the new estimate for
thecombined £ora because of more accurate esti-mates of MAT and
terrestrial lapse rate.
5. Discussion
5.1. Implications of climate analysis
The climate estimates for the upper and lower
Table 7Elevation estimate for the combined Jakokkota £ora
Factor Value
Modern MAT at coast (projected) (‡C) 27.1MAT change due to
global climate change (‡C) 00 1MAT change due to continental drift
(‡C) 0.7Jakokkota paleoMAT (‡C) 21.50 2.0Late Miocene sea level (m)
50Terrestrial lapse rate (‡C/km) 4.4 0 0.7Paleoelevation (m) 11600
600a
a Error calculated using the quotient variance equation.
PALAEO 2793 21-5-02
K.M. Gregory-Wodzicki / Palaeogeography, Palaeoclimatology,
Palaeoecology 180 (2002) 331^348344
-
Jakokkota £oras are statistically indistinguish-able, and
suggest that both £oras grew under asubtropical-dry climate. The
small di¡erences inthese estimates are due to the morphological
var-iation between the two £oras. Namely, the upper£ora has a
higher percentage of entire-marginedspecies, which translates into
a higher paleotem-perature.There are two factors besides climate
change
that could explain the di¡erences in leaf morphol-ogy observed
between the two £oras. Firstly, somevariation could be due to
sampling error. Severalstudies show that the smaller the vegetation
sam-ple, the greater the error of the temperature esti-mate (Wolfe,
1971; Povey et al., 1994; Wilf, 1997;Burnham et al., 2001); most
authors suggest thatat least 20^30 species are needed to obtain
accu-rate climate estimates. The upper Jakokkota £orais on the
small side, with only 24 species, and only22 of these have margin
information. Thus errorsin the morphologic score could be large.
Note,however, that the combined £ora with 48 speciesis a
statistically robust sample.Another factor that could cause the
morpho-
logic scores of the two £oras to vary is taphono-my. The two
di¡erent levels of the Jakokkota£ora represent di¡erent
paleoenvironments; thelower Jakokkota £ora was deposited in a
low-energy stream, while the upper Jakokkota £orawas deposited in a
strati¢ed lake. Studies byGreenwood (1992) and Roth and Dilcher
(1978)show that depositional processes have an impor-
tant a¡ect on leaf size; they found that averageleaf size drops
both with increasing distancefrom the shore in lake deposits, and
with increas-ing amounts of transportation in stream depos-its.Less
is known about the e¡ects of taphonomic
processes on other leaf morphology characterstates. Burnham et
al. (2001) show that the per-cent entire-margined species tends to
be lower inriparian versus terra ¢rme vegetation, but bothJakokkota
£oras represent riparian vegetation,so this is unlikely to be a
factor.Given these sources of error, and the fact that
the di¡erences in climate estimates between thetwo £oras are
within the sampling error, it wouldbe an overinterpretation to
suggest that a signi¢-cant climate change occurred between upper
andlower Jakokkota time. Thus, the climate estimatefor the combined
£ora of MAT=21.5 0 2.0‡C andMAP=5500 180 mm should present a
represen-tative picture of latest middle Miocene climate forthe
northern Altiplano.This temperature estimate is similar to
temper-
ature estimates for other Miocene £oras from theCentral Andes
(Table 8), which, with the excep-tion of the Chucal £ora of
northwestern Chile andGoterones £ora of coastal Chile, are all
estimatedto have been subtropical or tropical. Together,these £oras
suggest that a large portion of whatis now the Central Andes was
covered by subtrop-ical-dry forest. Today, the Altiplano is covered
bypuna, that is, alpine tundra.
Table 8Paleoclimate and paleoelevation estimates for Miocene
fossil £oras from the Central Andes
Flora Lat. Long. Age Method MAT MAP PaleoZ Refs(Ma) (‡C) (mm)
(m)
1. Los Litres 33.30 70.55 21.2^26.6 M subtrop. 700^800 ^ 12.
Chucal 18.9 68.9 25^19 A temperate dry 10000 1500 2, 33. Potos|¤
19.61 65.74 20.8^13.8 M 21.60 2.1 5000 400 0^13200 1200 4, 54.
Goterones 33.96 71.87 19^10 M 15.50 2.0 1500+ 0 1, 6, 74. Boca
Pupuya 33.96 71.87 19^10 M 21.70 2.0 humid 0 1, 6, 75. Jakokkota
17.17 69.17 10.660 0.06 M 21.50 2.5 5500 180 550^16000 1200 4, 8,
96. Pislepampa 17.18 66.03 7^6 A, M V20 1000^1500 1200^1400 10,
11
Numbers denote the location of the £oras on Fig. 1. Method,
method used to estimate paleoclimate and paleoelevation; A, mod-ern
analog; M, leaf morphology. MAT, mean annual temperature; MAP, mean
annual precipitation; PaleoZ, paleoelevation.Refs, references: 1,
Hinojosa and Villagran, 1997; 2, Charrier et al., 1994; 3, Mun‹oz
and Charrier, 1996; 4, Gregory-Wodzickiet al., 1998; 5, Berry,
1939; 6, Troncoso, 1991; 7, Hinojosa, 2000, in preparation; 8,
Berry, 1922b; 9, this study; 10, Berry,1922a; 11, Graham et al. (in
press).
PALAEO 2793 21-5-02
K.M. Gregory-Wodzicki / Palaeogeography, Palaeoclimatology,
Palaeoecology 180 (2002) 331^348 345
-
The interpretation of an arid climate in thenorthern Altiplano
in the late Miocene is consis-tent with the accumulation of
evaporite depositsin the southern Altiplano Puna of
Argentinastarting at 15 Ma, which is thought to mark theonset of
the modern dry climate regime (Vander-voort et al., 1995). The
higher amounts of rainfallestimated for the Pislepampa £ora from
the east-ernmost slopes of the Eastern Cordillera as com-pared to
the other £oras from further west sug-gests the presence of
gradient in rainfall from eastto west, and thus a rain shadow in
the Altiplanoarea by at least the late Miocene.
5.2. Implications of elevation analysis
The warmth of the late Miocene Jakokkota siteas compared with
the modern cool climate regimecan not be explained by global
climate change orcontinental drift, and thus is most likely due to
asigni¢cantly lower elevation than at present. Thelapse rate-based
calculation used in this studyestimates a paleoelevation of 11600
600 m, whichis almost 2800 m lower than the modern eleva-tion.This
paleoelevation is consistent with paleoele-
vation estimates of other paleo£oras from the Al-tiplano and
Eastern Cordillera (Table 8). The 21Ma Chucal £ora from the Chilean
Altiplano hasan estimated paleoelevation of 10000 1500 m,while the
early^middle Miocene Potos|¤ and 6^7Ma Pislepampa £oras from the
Eastern Cordillerahave paleoelevation estimates of 0^13200 1200and
1200^14000 1000 m, respectively (Fig. 1, Ta-ble 8) (Mun‹oz and
Charrier, 1996; Gregory-Wod-zicki, 2000a; Graham et al., in press).
The paleo-elevation of the Jakokkota £ora is also consistentwith
paleoelevations estimated using other paleo-altimeters; Kennan et
al. (1997) estimated thatremnants of a 10 Ma erosion surface that
capsthe Eastern Cordillera formed at an elevation of1000^1500 m.If
these data are correct, they imply that the
Altiplano and Eastern Cordillera underwent sig-ni¢cant amounts
of uplift since the late Miocene.The Jakokkota £ora suggests that
on the order of2/3 of the modern elevation of the Altiplano
wascreated since the early late Miocene, and the ero-
sion surfaces and the Pislepampa £ora suggestthat on the order
of 1/3^1/2 of the modernelevation of the Eastern Cordillera was
createdsince the late late Miocene (Graham et al., inpress).Such
signi¢cant amounts of recent uplift would
make the Central Andes the youngest of the ma-jor world orogens;
the Western Cordillera ofNorth America is thought to be
Eocene^Creta-ceous in age (Chase et al., 1998; Wolfe et al.,1998),
while the Himalayas and southern TibetanPlateau appear to have
attained their modern el-evations by at least 10 Ma (Garzione et
al., 2000;Rowley et al., 2001).
6. Conclusions
The analysis of the upper Jakokkota £ora sug-gests the following
conclusions:
1. The upper Jakokkota £ora is probably notmore than 0.2^0.5 Ma
older than the 10.66 0 0.06Ma lower Jakokkota £ora.2. There was not
a signi¢cant climate change
between the two levels of the Jakokkota £ora;leaf morphologic
analysis suggests a mean annualtemperature of 20.1 0 2.5‡C and a
mean annualprecipitation of 5500 180 mm for the lower Ja-kokkota
£ora, and 23.0 0 2.7‡C and 5500 180mm for the upper Jakokkota
£ora.3. The paleoclimate estimate for the combined
Jakokkota £ora is MAT=21.5 0 2.0‡C andMAP=5500 180 mm. Given the
apparent lackof climate change and taphonomic bias betweenthe two
levels, this estimate is likely a good rep-resentation of the late
Miocene climate of thenorthern Altiplano.4. Comparison with other
Miocene fossil £oras
suggests that subtropical-dry forest covered alarge portion of
the Altiplano. The area is nowcovered by alpine tundra.5. Using a
lapse rate-based calculation, the
paleoelevation of the Jakokkota £ora was11600 600 m. This
estimate implies that on theorder of 2/3 of the modern elevation of
the Al-tiplano was created since the early late Mio-cene.
PALAEO 2793 21-5-02
K.M. Gregory-Wodzicki / Palaeogeography, Palaeoclimatology,
Palaeoecology 180 (2002) 331^348346
-
Acknowledgements
The author was supported by U.S. NationalScience Foundation
grant EAR-99-09114. Manythanks to J. Argollo for assistance with
relocatingthe Jakokkota site and for use of laboratoryfacilities at
the Universidad Mayor de San Andres,to A. Auza for assistance in
the field, to M. Neefor assistance in identifying the fossil forms,
and toD.R. Greenwood and J.A. Wolfe for reviewswhich significantly
improved the manuscript. Thisis Lamont-Doherty Earth Observatory
contribu-tion number 6319.
References
Axelrod, D.I., Bailey, H.P., 1976. Tertiary vegetation,
climateand altitude of the Rio Grande depression, New
Mexico-Colorado. Paleobiology 2, 235^254.
Berry, E.W., 1922a. Pliocene fossil plants from eastern
Bolivia.John Hopkins Univ. Stud. Geol. 4, 145^202.
Berry, E.W., 1922b. Late Tertiary plants from Jancacota,
Bo-livia. Johns Hopkins Univ. Stud. Geol. 4, 205^221.
Berry, E.W., 1939. The fossil £ora of Potos|¤, Bolivia.
JohnHopkins Univ. Stud. Geol. 13, 1^67.
Burnham, R.J., Pitman, N.C.A., Johnson, K.R., Wilf, P.,2001.
Habitat-related error in estimating temperaturesfrom leaf margins
in a humid tropical forest. Am. J. Bot.88, 1096^1103.
Charrier, R., Mun‹oz, N., Palma-Heldt, S., 1994. Edad y
con-tenido paleo£oristico de la Formacio¤n Chucal y
condicionespaleoclimaticas para el Oligoceno Tardio - Mioceno
Inferioren el Altiplano de Arica, Chile. Proceedings, Congreso
Geo-lo¤gico Chileno, 7th, Concepcio¤n, Chile, vol. 1, pp.
434^437.
Chase, C.G., Gregory-Wodzicki, K.M., Parrish-Jones,
J.T.,DeCelles, P., 1998. Topographic history of the western
Cor-dillera of North America and controls on climate. In: Crow-ley,
T.J., Burke, K. (Eds.), Tectonic Boundary Conditionsfor Climate
Model Simulations. Oxford Monographs onGeology and Geophysics vol.
39. Oxford University Press,Oxford, pp. 73^99.
Forest, C.E., Molnar, P., Emanuel, K.A., 1995. Palaeoaltim-etry
from energy conservation principles. Nature 374, 347^350.
Forest, C.E., Wolfe, J.A., Molnar, P., Emanuel, K.A.,
1999.Paleoaltimetry incorporating atmospheric physics and
bota-nical estimates of paleoclimate. Geol. Soc. Am. Bull.
111,497^511.
Garzione, C.N., Quade, J., DeCelles, P.G., English, N.B.,2000.
Predicting paleoelevation of Tibet and the Himalayafrom delta O-18
vs. altitude gradients in meteoric wateracross the Nepal Himalaya.
Earth Planet. Sci. Lett. 183,215^229.
Graham, A., Gregory-Wodzicki, K.M., Wright, K.L.,
2001.Mio-Pliocene palyno£ora from the Eastern Cordillera, Bo-livia:
implications for the uplift history of the Central An-dean Plateau.
Am. J. Bot. 88, 1545^1557.
Greenwood, D.R., 1992. Taphonomic constraints on
foliarphysiognomic interpretations of late Cretaceous and Terti-ary
palaeoclimates. Rev. Palynol. Paleobot. 71, 149^190.
Gregory, K.M., McIntosh, W.C., 1996. Paleoclimate and
pa-leoelevation of the Oligocene Pitch-Pinnacle £ora, SawatchRange,
Colorado. Geol. Soc. Am. Bull. 108, 545^561.
Gregory-Wodzicki, K.M., 2000a. Uplift history of the Centraland
Northern Andes: A review. Geol. Soc. Am. Bull. 112,1091^1105.
Gregory-Wodzicki, K.M., 2000b. Relationships between
leafmorphology and climate, Bolivia: Implications for estimat-ing
paleoclimate from fossil £oras. Paleobiology 26, 668^688.
Gregory-Wodzicki, K.M., McIntosh, W.C., Velasquez, K.,1998.
Climatic and tectonic implications of the late MioceneJakokkota
£ora, Bolivian Altiplano. J. South Am. Earth Sci.11, 533^560.
Haq, B.U., Hardenbol, J., Vail, P.R., 1987. Chronology
of£uctuating sea levels since the Triassic. Science 235,
1156^1167.
Hinojosa, L.F., 2000. Ana¤lisis ¢siogno¤mico-foliar de las
pale-o£oras del Mioceno de Chile Central: magnitud del
cambioclima¤tico Neo¤geno-Paleo¤geno. XI Simposio Argentino
dePaleobota¤nica y Palinolog|¤a, Tucuman, Argentina, p. 48.
Hinojosa, L.F., Villagran, C., 1997. Historia de los bosques
delsur de Sudame¤rica, I: antecedentes paleobota¤nicos,
geolo¤gi-cos y clima¤ticos del Terciario del cono sur de Ame¤rica.
Rev.Chil. Hist. Nat. 70, 225^239.
Jacobs, B.F., 1999. Estimation of rainfall variables from
leafcharacters in tropical Africa. Palaeogeogr.
Palaeoclimatol.Palaeoecol. 145, 231^250.
Jordan, G.J., 1997. Uncertainty in palaeoclimatic
reconstruc-tions based on leaf physiognomy. Aust. J. Bot. 45,
527^547.
Jordan, T.E., Isacks, B.L., Allmendinger, R.W., Brewer,
J.A.,Ramos, V.A., Ando, C.J., 1983. Andean tectonics related
togeometry of subducted Nazca plate. Geol. Surv. Am. Bull.94,
341^361.
Kennan, L., Lamb, S.H., Hoke, L., 1997. High-altitude
pa-laeosurfaces in the Bolivian Andes: evidence for late Ceno-zoic
surface uplift. In: Widdowson, M. (Ed.), Palaeosurfa-ces:
Recognition, Reconstruction and PalaeoenvironmentalInterpretation.
Geological Society Special Publ., No. 120,pp. 307^323.
Kennedy, E.M., 1998. Cretaceous and Tertiary mega£orasfrom New
Zealand and their climate signals. UnpublishedPh.D. thesis, Open
University, Milton Keynes, 262 pp.
Killeen, T.J., Garc|¤a, E., Beck, S.G., 1993. Gu|¤a de Arboles
deBolivia. Herbario Nacional de Bolivia-Missouri BotanicalGarden,
958 pp.
King, T., 1996. Equatorial Paci¢c sea surface
temperatures,faunal patterns, and carbonate burial during the
Pliocene.Mar. Micropaleontol. 27, 63^84.
Kowalski, E.A., 2001. Middle to Late Miocene Environments
PALAEO 2793 21-5-02
K.M. Gregory-Wodzicki / Palaeogeography, Palaeoclimatology,
Palaeoecology 180 (2002) 331^348 347
-
of Southern Ecuador: Temperature, Elevation, and FossilPlants of
the Nabon Basin. Ph.D. Thesis, University ofMichigan, Ann Arbor,
MI.
Lenters, J.D., Cook, K.H., 1995. Simulation and diagnosis ofthe
regional summertime precipitation climatology of SouthAmerica. J.
Clim. 8, 2988^3005.
Lenters, J.D., Cook, K.H., 1997. On the origin of the
BolivianHigh and related circulation features of the South
Americanclimate. J. Atmos. Sci. 54, 656^677.
Loutit, T.S., Kennett, J.P., Savin, S.M., 1983. Miocene
equa-torial and southwest Paci¢c paleoceanography from
stableisotope evidence. Mar. Micropaleontol. 8, 215^233.
Meehl, G.A., 1992. E¡ect of tropical topography on
globalclimate. Annu. Rev. Earth Planet. Sci. 20, 85^112.
Meyer, H.W., 1992. Lapse rates and other variables applied
toestimating paleoaltitudes from fossil £oras. Palaeogeogr.
Pa-laeoclimatol. Palaeoecol. 99, 71^99.
Mun‹oz, N., Charrier, R., 1996. Uplift of the western border
ofthe Altiplano on a west-vergent thrust system, NorthernChile. J.
South Am. Earth Sci. 9, 171^181.
Povey, D.A.R., Spicer, R.A., England, P.C., 1994.
Palaeobo-tanical investigation of early Tertiary palaeoelevations
innortheastern Nevada: initial results. Rev. Paleobot. Palynol.81,
1^10.
Roth, J.L., Dilcher, D.L., 1978. Some considerations in leafsize
and leaf margin analysis of fossil leaves. Cour. Forsch.inst.
Senckenb. 30, 165^171.
Rowley, D.B., Pierrehumbert, R.T., Currie, B.S., 2001. A
newapproach to stable isotope-based paleoaltimetry: implica-tions
for paleoaltimetry and paleohypsometry of the HighHimalaya since
the Late Miocene. Earth Planet. Sci. Lett.188, 253^268.
Savin, S.M., 1977. The history of the earth’s surface
temper-ature during the past 100 million years. Annu. Rev.
EarthPlanet. Sci. 5, 319^355.
Savin, S.M., Abel, L., Barrera, E., Hodell, D., Kennett,
J.P.,Murphy, M., Keller, G., Killingley, J., Vincent, E., 1985.The
evolution of Miocene surface and near-surface marinetemperatures:
Oxygen isotopic evidence. Geol. Soc. Am.Mem. 163, 49^82.
Savin, S.M., Woodru¡, F., 1990. Isotopic evidence for
temper-ature and productivity in the Tertiary oceans. In:
Burnett,W.C., Riggs, S.R. (Eds.), Phosphate Deposits of the
World.Cambridge University Press, Cambridge, pp. 241^259.
Sirvas, F., Torres, E., 1966. Consideraciones geolo¤gicas de
lazona noroeste de la provincia Pacajes del Departamento deLa Paz.
Bol. Inst. Boliv. Pet. 6, 54^64.
Smith, A.G., Hurley, A.M., Briden, J.C., 1981.
PhanerozoicPaleocontinental World Maps. Cambridge University
Press,Cambridge, 102 pp.
Stranks, L., England, P., 1997. The use of a resemblance
func-tion in the measurement of climatic parameters from
thephysiognomy of woody dicotyledons. Palaeogeogr. Palaeo-climatol.
Palaeoecol. 131, 15^28.
Suarez Soruco, R., Diaz Martinez, E., 1996. Lexico
estratigra-¢co de Bolivia. Rev. Tec. Yacim. Pet. Fisc. Boliv. 17,
100^102.
Troncoso, A., 1991. Paleomega£ora de la Formacion Navidad,
Miembro Navidad (Mioceno), en el a¤rea de Matanzas, ChileCentral
Occidental. Bol. Mus. Nac. Hist. Nat. Chile 43, 155^178.
Vandervoort, D.S., Jordan, T.E., Zeitler, P.K., Alonso,
R.N.,1995. Chronology of internal drainage development and up-lift,
southern Puna plateau, Argentine central Andes. Geol-ogy 23,
145^148.
Vose, R.S., Schmoyer, R.L., Steurer, P.M., Peterson, T.C.,Heim,
R., Karl, T.R., Eischeid, J.K., 1992. The Global His-torical
Climatology Network: Long-Term Monthly Temper-ature, Precipitation,
Sea Level Pressure, and Station Pres-sure Data. CDIAC’s Numeric
Data Package Collection,Oak Ridge National Laboratory, Tennessee,
version 1.02.
Wiemann, M.C., Manchester, S.R., Dilcher, D.L., Hinojosa,L.F.,
Wheeler, E.A., 1998. Estimation of temperature andprecipitation
from morphological characters of dicotyledo-nous leaves. Am. J.
Bot. 85, 1796^1802.
Wilf, P., 1997. When are leaves good thermometers? Paleobi-ology
23, 373^390.
Wilf, P., Wing, S.L., Greenwood, D.R., Greenwood, C.L.,1998.
Using fossil leaves as paleoprecipitation indicators:an Eocene
example. Geology 26, 203^206.
Wilf, P., Wing, S.L., Greenwood, D.R., Greenwood, C.L.,1999.
Using fossil leaves as paleoprecipitation indicators:an Eocene
example - REPLY. Geology 27, 92.
Wing, S.L., Greenwood, D.R., 1993. Fossils and fossil
climate:the case for equable continental interiors in the
Eocene.Philos. Trans. R. Soc. Lond. B Biol. Sci. 341, 243^252.
Wolfe, J.A., 1971. Tertiary climatic £uctuations and methodsof
analysis of Tertiary £oras. Palaeogeogr. Palaeoclimatol.Palaeoecol.
9, 27^57.
Wolfe, J.A., 1979. Temperature parameters of humid to
mesicforests of eastern Asia and relation to forests of other
re-gions of the Northern Hemisphere and Australasia.
USGSProfessional Paper 1106, 37 pp.
Wolfe, J.A., 1992. An analysis of present-day terrestrial
lapserates in the western conterminous United States and
theirsigni¢cance to paleoaltitudinal estimates. USGS
ProfessionalPaper 1964, 35 pp.
Wolfe, J.A., 1993. A method of obtaining climatic parametersfrom
leaf assemblages. U.S. Geological Survey Bulletin vol.2040, 71
pp.
Wolfe, J.A., 1994. Tertiary climatic changes at middle
latitudesof western North America. Palaeogeogr. Palaeoclimatol.
Pa-laeoecol. 108, 195^205.
Wolfe, J.A., 1995. Paleoclimatic estimates from Tertiary
leafassemblages. Annu. Rev. Earth Planet. Sci. 23, 119^142.
Wolfe, J.A., Forest, C.E., Molnar, P., 1998.
Paleobotanicalevidence of Eocene and Oligocene paleoaltitudes in
midlati-tude western North America. Geol. Soc. Am. Bull. 110,
664^678.
Wolfe, J.A., Schorn, H.E., Forest, C.E., Molnar, P.,
1997.Paleobotanical evidence for high altitudes in Nevada duringthe
Miocene. Science 276, 1672^1675.
Wolfe, J.A., Uemura, K., 1999. Using fossil leaves as
paleo-precipitation indicators: an Eocene example - COMMENT.Geology
27, 91^92.
PALAEO 2793 21-5-02
K.M. Gregory-Wodzicki / Palaeogeography, Palaeoclimatology,
Palaeoecology 180 (2002) 331^348348