A palynological study of the upper quaternary sequence in the El Abra corridor and rock shelters (Colombia)

Palaeogeography, Palaeoclima tology, Palaeoecology, 25( 1978): 1--109 © Elsevier Scientific Publishing Company, Amsterdam -- Printed in The Netherlands

A PALYNOLOGICAL STUDY OF THE UPPER QUATERNARY SEQUENCE IN THE EL ABRA CORRIDOR AND ROCK SHELTERS (COLOMBIA)

ELISABETH J. SCHREVE-BRINKMAN

Hugo de Vries-Laboratory, Amsterdam (The Netherlands) (Accepted for publication December 29, 1977)

ABSTRACT Schreve-Brinkman, E. J., 1978. A palynological study of the Upper Quaternary sequence in the El Abra corridor and rock shelters (Colombia). Palaeogeogr., Palaeoclimatol., Palaeoecol., 25: 1--109. This palynological study is a contribution to the reconstruction of the vegetational, climatological and stratigraphical sequence in the E1 Abra corridor principally during the Last Glacial and the Holocene. It seems likely that the sequence reaches into the Last Interglacial and even further down into the last part of the penultimate Glacial. The E1 Abra corridor forms part of the Sabana de Bogot~ located in the Colombian Eastern Cordillera (altitude of site 2,570 m). The sequence starts with a succession of several warmer and colder intervals (El Abra zones I and II) during which the El Abra valley lay mainly in the Andean forest belt with a shift to the subp~ramo vegetation belt during the relatively coldest periods. Weinmannla was well represented among the trees. A series of peat, peaty clay and clay was deposited. We have named the warmer intervals, E1 Abra zones Ib, Id, IIb, IId, respectively, the Rocas de Sevilla I, II, IIIa, IIIb intervals. At the transition of the E1 Abra subzones Ild to IIe, a 14C-dating yields an age of approximately 50,000 years. Thus the El Abra subzones IIa--d probably correspond with Early Glacial interstadials and stadials, subzone IIe with the Lower Pleniglacial and subzone Id with the Last Interglacial (Europe: Eemian). However, these correlations are very tentative. During the next series of warmer and colder intervals (El Abra zone III), conditions changed in the valley. During the first two warmer intervals, named Rocas de Sevilla IVa and b (El Abra subzones IIIal and IIIas), the valley was inundated due to a rise in the level of the Sabana Lake caused by an increase of the effective precipitation. From-the end of the third warmer interval, named Rocas de Sevilla V interval (El Abra subzone IIIc), there is a 14C-dating of ca. 28,000 years B.P., which correlates this interval with the santuario interval and approximately with the European Denekamp interstadial. El Abra zone III likely corresponds with the Middle Pleniglacial and the intervals RIVa and RIVb probably with the European Moershoofd and Hengelo interstadials. Lake clay was deposited during inundation periods and is followed by a series of "soils". Alnus rather than Weinmannia becomes dominant among the trees; characteristic elements of the Andean forest disappear. The vegetation belts shift from the Andean forest/subp~ramo zones to the subp~ramo/ p~ramo zones. El Abra zone IV forms the coldest part of the Last Glacial (Upper Pleniglacial). During the Fuquene stadial (actually the coldest part), conditions were dry and the effective

precipitation was probably lower than it is today. A series of soils was deposited (or formed), with a high silt content ("loessic"), during the coldest phase. At the end of zone IV the humidity increased again, which resulted in a renewed inundation of the E1 Abra corridor during part of the following Late Glacial (El Abra zone V). This renewed rising of the Sabana Lake took place in the Guantiva interstadial, corresponding approximately with the European B011ing and AllerCd interstadials. The Andean forest again invaded the valley. During the following El Abra stadial (type section: the succession in II-B3), conditions became colder and drier again. The Late Glacial sedimentary sequence consists of "soily" material, lake clay and peaty clay. At the beginning of the Holocene (El Abra zones VI, VII, VIII), the more definite amelioration of the climate took place. The E1 Abra corridor formed part of the Andean forest belt. Several species, representative of the Andean forest, reappear after a long absence or occur for the first time in the diagram. The sediment consists of a partly homogenized "soily" material. During the upper part of the Holocene, human influence on the vegetation is apparent. Apart from the pollen analysis, opaline silica bodies were studied; these bodies (= phytoliths) are remnants of plants. The main contributors are grasses. Each genus may have its own characteristic types. It appeared that the opaline silica bodies may provide very useful information in two ways. A quantitative analysis offers methods for correlation. A qualitative analysis offers the possibility to identify various taxa within the Gramineae, which is otherwise impossible on the basis of pollen grains. RESUMEN E s t u d i o p a l i n o l b g i c o d e la secueneia del Cuaternario S u p e r i o r en el c o r r e d o r y los abrigos rocosos de E l A bra, Colombia. ...... ,

i

Este estudio palinol6gico es una contribuci6n a la reconstrucci6n de la secuencia de vegetaci6n y clima y a la estratigrafia en el corredor de E1 Abra, principalmente para el Ultimo Glacial y el Holoceno. Parece probable que la secuencia incluye tambi~n el Ultimo Interglacial y hasta la 61tima fase del Penfiltimo Glacial. La informaci6n sobre esta parte de la secuencia no es muy detailada. Despu~s del anterior estudio sobre la perforaci6n profunda en la Ciudad Universitaria (Bogota), esta publicaci6n es la primera en dar un estudio detallado de toda la secuencia del Ultimo Glacial en la Sabana de Bogota. E1 ~rea de E1 Abra es parte de esta altiplanicie en la Cordillera Oriental, a una altura de aprox. 2, 570 m. La secuencia principia con la terminaci6n de una fase de clima relativamente m~s caliente (zona Ib de E1 Abra), posiblemente correspondiendo al ~ltimo " i n t e r s t a d i a l " de la pen~ltimaglaciaci6n. La fase siguiente, m~s fria (zona Ic del Abra), corresponderia entonces a la ~ltima fase fr~a del mismo glacial. E1 siguiente intervalo de elima m~s caliente (zona Id) podria corresponder al Ultimo Interglacial y le dimos el hombre de Intervalo Rocas de Sevilla II. Durante esta parte inferior de la seeuencia (hasta la zona II) el eorredor de E1 Abra se hallaba en la zona del bosque Andino, presentando cambios hacia la zona de vegetaci6n de subp~ramo durante algunos periodos m~s frios de la secuencia. W e i n m a n n i a (enceniUo) era bien representado entre los ~rboles. Se depositb una serie de turba, turba arcillosa y arcilla. La zona siguiente (II) consiste de ratios intervalos de clima m~s caliente, alternando con periodos m~s frios y corresponde probablemente a la parte temprana del Ultimo Glacial ("Early Glacial") y al Pleniglacial Inferior. Los dos intervalos de clima relativam e n t e m~s caliente, los llamamos Intervalos Rocas de Sevilla IIIa y IIIb. Estos podr/an corresponder a los Interestadiales de Br~rup y Odderade, pero hasta ahora no hay modo de probar esta correlaci6n. Durante el Pleniglacial Medio (zona III de E1 Abra), hay dos intervalos de clima m ~

caliente, que ilamamos Rocas de Sevilla IVa y IVb; durante estos intervalos el Valle de E1 Abra fue inundado completamente, debido a una subida del nivel de la Laguna de la Sabana de Bogota. Esta subida fue causada por un incremento considerable de la precipitaci6n efectiva (hasta m~s alto que hoy d~a). E1 per~odo intermedio m~s frio fue relativamente seco. La cantidad de Botryococcus presente en el material estudiado, parece dar una buena indicaci6n sobre la cantidad o profundidad de agua abierta en el sitio. E1 tercer intervalo m~s caliente se ha llamado Rocas de Sevilla V y debe corresponder en tiempo al Intervalo del Sanctuario (de la Laguna de Fuquene) y aproximadamente al Interestadial Europeo de Denekamp. Hay un incremento de humedad hacia el final de este intervalo, posiblemente debido a una diminuci~n de la evaporaci6n causada por temperaturas m ~ bajas. Los Interv~los Rocas de Seviila IVa y IVb posiblemente corresponderian a los Interestadiales Europeos de Moershoofd y Hengelo. Despu~s de la deposici~n de arcillo de laguna durante los perfodos de inundaci6"n, la secuencia sedimentaria consiste de una serie de "suelos"; principalmente en un limo m~s arenoso. Las zonas de vegetaci6n que dominan en el corredor de El Abra, cambian de zonas de Bosque Andino/zona de subp~ramo a zonas de subp~amo/p~ramo. Weinmannia ya no domina entre los ~rboles, siendo Alnus el ~rbol mejor representado. Elementos caracter~sticos para el Bosque Andino desaparecen casi por completo de los diagramas. La siguiente zona IV del Abra corresponde a la parte m~s fria del Ultimo Glacial (Pleniglacial Superior). Durante la parte m~s fria de esta zona al Estadial de Fuquene, las condiciones clim~ticas eran secas y la precipitaci6n efectiva era probablemente m~s baja que la actual. Se deposit6 (resp. se form6) una serie de "suelos"; los que se formaron durante la parte m~s frla contienen bastante silt ("lo~sicos"). Hacia el final de la zona IV aumento nuevamente la humedad, resultando finalmente durante la zona V (Tardiglacial), en una nueva inundaci6n. Esta subida del nivel de la Laguna de la Sabana tuvo lugar en el Interestadial de Guantiva, relativamente caliente y correspondiendo en edad al los Interestadiales Europeos de Aller~d y Bcflling. Durante el siguiente Estadial de E1 Abra (localidad y secciSn tipica E1 Abra II-B3), el clima era nuevamente m~s frio y tambi~n mas seco. La presencia de Cactaceae nos da una indicaci6n m~s de la relativa aridez del clima. A1 principio de la zona V (Tardiglacial) hubo un cambio importante en las zonas de vegetaci6n dominantes en E1 Abra. E1 Bosque Andino nuevamente invade el v a l l e y se retir6 solo temporalmente durante el Estadial de El Abra. La secuencia sedimentaria consiste de material con caracter~sticas de suelos, de arcilla de laguna y de arcilla turbosa. En la transici6n hacia el Holoceno, hubo una fluctuaci6n clim~tica de corta duraciSn, representado en la secci6n II-B3. En el principio del Holoceno (zonas VI, VII y VIII de El Abra), se realis6 un mejoramiento m~s definitivo del clima. Nuevamente el ~rea del Abra lleg6 a set parte de la zona de Bosque Andino. Reaparecen despu~s de una larga ausencia varias especies tipicas para este Bosque Andino, o aparecen por primero vez en el diagrama de polen. Los sedimentos consisten en material con caracteres de suelo, en parte "homogeneizado". Durante la parte superior del Holoceno la influencia humana resulta en deforestaci~n progresiva y finalmente en cultivaciSn completa del ~rea. Fuera del analisis de polen (y de fechar, correlacionar y interpretar las zonas de polen), se estudiaron fitolitos. Estos pequefios objetos de silica opalina son restos de plantas. Se pueden formar en ciertas celulas de la planta viva. Son especialmente abundantes en la familia de las Gramineas y cada genero puede tener sus propios tipos caracteristicos. Aunque falta por estudiar mucho sobre esta materia, ya se puede decir que fitolitos "f6siles" pueden dar informaci6n muy dtil. Analisis cuantitativo ofrece un m~todo para correlaci6n, tambi~n en el caso que falta polen en los estratos, ya que se pudo establecer que las m~ximos y m~nimos de fitolitos en varios diagramas son paralelos a los de la curva de polen de Gramineas. Analisis cualitativo ofrece la posibilidad de identificar varios taxa dentro de las Gramineas, imposible de diferenciar a base de granos de polen. En esta manera se puede obtener importante informaci6n ecol6gica adicional.

4

INTRODUCTION The E1 Abra corridor forms part of the Sabana de Bogoth in the Colombian Eastern Cordillera (Fig. 1). It is a north--south running, corridor-like gap in a hill ridge of Upper Cretaceous sandstone (Fig.l). The road from Zipaquir~ to Tocanciph passes through this corridor. The rock walls, called " R o c a s de Sevilla", have a height of a b o u t 50 m above the valley surface. The actual elevation of this surface is ca. 2,570 m above sea level. The rock shelters referred to in this study are in the most westernly situated rock wall (Fig.l). This palynological study of the Upper Quaternary sequence in the E1 Abra corridor forms part of a more extensive project of investigations: a study of the Quaternary and a search for and study of possible cultural sequences of early Man in Colombia. This project was sponsored by WOTRO (the Netherlands Foundation for the Advancement of Tropical Research) and the Instituto Colombiano de Antropologfa, and later also by the National Science Foundation (U.S.A.) as far as the archeological part is concerned. The palynological studies were carried out by the author at the Department of Palynology o f the University of Amsterdam. The first test excavation in rock shelter II was carried o u t in O c t o b e r 1967 by G. Correal and T. van der Hammen; J. C. Lerman did the '4C-dating (Correal et al., 1970). In 1968 a new test excavation was made by G. Correal at the site of E1 Abra III. Finally, in 1969, excavations on a larger scale were made in three rock shelters: II, III and IV. The first results of this field work and the subsequent laboratory studies were published in a short article (Hurt et al., 1972), and an analysis of the archeological investigations together with a summary of the stratigraphical and palynological results in Hurt et al. (1976). The present publication reports the final results of pollen analysis in combination with the stratigraphy. In addition, evidence of opaline silica bodies present in the samples is provided; the quantitative and qualitative aspects of these " p h y t o l i t e s " are also discussed. Other relevant contributions on stratigraphy, environments, faunal remains, etc., will be published concurrently PRESENT CLIMATE, VEGETATION AND POLLEN RAIN The E1 Abra area, situated on the high plain of Bogota, roughly some 5 ° north of the equator and 2,570 m above sea level, has a tropical montane climate. The mean annual temperature is a b o u t 14°C. The mean m o n t h l y temperatures vary b u t little through the year; they are lowest during the dry months, when sometimes night frost occurs. The total annual rainfall is between 700 mm and 900 mm. There are t w o wet and t w o dry seasons, the wet ones from April to May and from October to November. Little is left of the original vegetation o f the area. The high plain of Bogot~ lies in the lower part of the Andean forest belt of Cuatrecasas (1958), or in the dry lower montane forest zone of Holdridge (Espinal and Montenegro, 1963).

74"

F/ %,

/

EL ABRA

BOGOTA

j

i

"l

/

5"

'

q 1 . o\

I,,

'1

/ i £__,_~2oo \.,

Z l p & q u ra

H-3.(.:~ rJda El ,\t

2575 m

263! m

E__~4q~G 107N, (

~J

f

J

'~ ,..,,,. El Abram 12E,1'

? {r

Bll

~2619m

A b r a I[

100N 10E

El Abra \~\ \\ \\

--J --4

//

.;".".'. a nt ~q uo vallecito •" : . ' old gully

\'

\\\',

p~redones de roca rock wa!ls 0

100

200

\'\ \'~\ 200 m

\~\ \%

")

Fig. 1. Map of part of the western rock wall of the El Abra corridor with sites of rock shelter sections (II, III and IV) and handborings (B).

Low scrub of secondary vegetation locally covers the hills, whereas the fiat part of the area is arable land. Only at the base of the rock walls a strip of, partly secondary, forest is left, which probably still reasonably well represents the original forest. A special vegetation type is locally developed in the cracks and fissures of the rocks of the escarpments. The forest at the foot of the escarpments is composed principally of Weinmannia (Cunoniaceae), Melastomataceae (including Miconia ), Oreopanax (Araliaceae), Eugenia (Myrtaceae), and Daphnopsis (Thymelaeaceae). Two samples were taken, each from a number of moss cushions, one more in the interior of the forest (sample 1) and the other one close to its outer border. The samples were prepared in order to obtain the recent pollen rain of the sites, and the slides were studied by Mr. G. Noldus. The results are given in Table I. Approximately 250 pollen grains of trees, shrubs (including Compositae and Hypericum} and grasses were counted in each sample (pollen total), and the percentage of each type was calculated. Other elements (herbs, ferns, fungi) were likewise calculated on the basis of this pollen total. Both pollen spectra are rather similar. Tree pollen dominates, Weinmannia being the most abundant, followed by Melastomataceae, Araliaceae, Daphnopsis, and Myrtaceae. This seems to be a fair representation of the tree composition of the forest, with the exception of Piperaceae which, although common in the forest, are not represented in the pollen spectrum. This is certainly caused by the fact that the pollen grains of Piperaceae are extremely small, difficult to recognize, and easily destroyed. In sample 2, the percentage of grass pollen is higher than in sample 1; this is easily explained by the fact that the former sample was taken close to the border of the forest, where the inflow of pollen from the cultivated land and meadows is undoubtedly higher. For the same reason, sample 2 contains some pollen of anemophilous trees and shrubs not present in the local stand of forest but forming a part of the "back-ground" of atmospheric pollen in the area (Alnus, Podocarpus, Hedyosmum, Myrica). Under present circumstances the woody vegetation of the high plain of Bogot~ and surrounding slopes would have been a mixed Weinmannia-forest similar to the remnants still present at the foot of the E1 Abra escarpments. On the slopes of the mountains forming the eastern border of the Sabana de Bogota, near a site called Torca, there are still rather extensive remnants of forest, between about 2,600 m and 3,100 malt. There is no Weinmannia in the forest found at 2,650 m, more abundant elements being Oreopanax (Araliaceae), Vallea (Elaeocarpaceae), Miconia (Melastomataceae), Eugenia (Myrtaceae), Psychotria and Palicourea (Rubiaceae), Rhamnus (Rhamnaceae), Phyllanthus and Croton (Euphorbiaceae). Between 2,700 m and 3,100 m altitude Weinmannia is the dominant tree, with Clusia and Drymis as codominants. Other elements are Ilex, Rhamnus, Palicourea, Psychotria,

Daphnopsis, Myrica, Oreopanax, Cedrela, Hydiosmum, Viburnum, Rapanea, Symplocos, Cestrum, and different genera of Ericaceae (T. van der Hammen, pers. comm.): ~

On t h e o u t e r slopes o f t h e hills b o r d e r i n g t h e high plain on t h e w e s t side, o a k f o r e s t s occur: t h e d o m i n a n t e l e m e n t in this t y p e o f f o r e s t (Quercetum) is Quercus humboldti. On t h e m a r s h y p a r t s o f t h e f o r m e r lake b o t t o m an a l d e r carr (Alnetum ]orulliensis) m a y occur. In o t h e r places in t h e E a s t e r n TABLE I Percentages of pollen grains from two recent samples from moss cushions in the forest at the foot of the escarpment of the E1 Abra Valley Pollen % (total 239) E1 Abra forest sample 1 Gramineae Compositae (Tu bulifl. ) Hypericurn type

Alnus Podocarpus Hedyosmum My rica Dodonaea Weinmannia Melastomataceae Miconia Araliaceae Daphnopsis Myrtaceae Rapanea

Hydrocotyle Rumex Chenopodiaceae type Cyperaceae Thalictrum- type Cyatheaceae type L ycopodium reticulate Jamesonia-type Monolete psilate Monolete verrucate Fungal spores

3.5 1.5 0.5 --

Pollen % (total 252) E1 Abra forest sample 2 10.5 1 2 0.5

2

--

0.5

--~

- -

0.5

65 18 0.5 5.5 2.5 2.5 0.5

47 12 1.5 16 6.5 0.5 0.5

100%

100%

--

1

-

1

"-= ~

o (O

1.5

_

-

0.5

1

0.5

--

0.5

-

5 0.5 57

r~

0.5

ii 8.5 48

.~ ..~ Z

Cordillera m i x e d Alnus-Weinmannia f o r e s t s m a y be d e v e l o p e d o n p e a t y soil (T. v a n d e r H a m m e n , pets. c o m m . ) . F o r f u r t h e r d a t a o n c l i m a t e a n d vegetat i o n a n d o n p o l l e n t y p e s t h e r e a d e r is r e f e r r e d to Van d e r H a m m e n a n d Gonzalez (1960).

8

WORKING METHODS All pollen samples were prepared in the same way. They were subjected to the following treatment: first they were heated with KOH till boiling point, secondly, gravitative separation was carried o u t in a mixture of b r o m o f o r m and alcohol with a specific density of 2. The following pollen types (genera, families) are included in the pollen sum: Gramineae llex (Aquifoliaceae) Compositae (Tubiliflorae) Styloceras (Stylocerataceae) Hypericum-type (Hypericaceae) Daphnopsis (Thymelaeaceae) Ericaceae Juglans (Juglandaceae) A cae na/Po ly lepis (Rosaceae) Bocconia (Papaveraceae) Quercus (Fagaceae) Dodonaea (Sapindaceae) Podocarpus (Podocarpaceae) Acalypha (Euphorbiaceae) Hedyosmum (Chloranthaceae) Alchornea (Euphorbiaceae) Weinmannia (Cunoniaceae) Miconia (Melastomataceae) Rapanea (Myrsinaceae) Melastomataceae Symplocos (Syrnplocaceae) Myrtaceae Drymis (Winteraceae) From each sample between 200 and 300 pollen grains of these types were counted. When the material was t o o scanty to provide this number of pollen grains, as many grains as possible were counted. If necessary, much larger pollen totals were c o u n t e d (mainly in the case of a strong dominance of Alnus pollen reducing the percentages of other types). In the general diagrams the curves of the more important forest elements (Alnus, Quercus and Podocarpus) are shown separately. The remainder of the forest elements of the Andean forest is compiled in one curve under the heading: " o t h e r forest elements". These curves are represented from left to right. Of the principal elements of the p~ramo, Gramineae, and, if sufficiently represented, Acaena/Polylepis are shown separately. The typical elements that are most abundant in the dwarf scrub of the subp~ramo or uppermost transitional forest, Compositae tubuliflorae, the Hypericum type, and Ericaceae are compiled in one curve. The curves of elements of open vegetation and of the dwarf scrub are added up, one on t o p o f the other, from right to left. To the right of the general diagram, separate curves are drawn for all identified genera and families. DESCRIPTION OF THE POLLEN

DIAGRAMS

(for general legend, see Fig.2)

General remarks The sections can be divided into three groups, viz., the rock-shelter sections, the soil sections, and the sections from the gully in front of :rock shelter II plus t w o other long sections. F r o m many of the excavation pits in the rock shelters and borings made in the E1 Abra valley, pollen samples have been taken. The exact sites are indicated i n the stratigraphical profiles (Van, der Hammen, 1978) and on the map of Fig. 1.

9 From each of the three rock shelters II, III and IV (1969-excavations) one or two sections provided minor pollen diagrams (II-10E, II-100N, III-10E, III-12E, IV-107N). Borings in the valley bottom itself have provided some very fine pollen diagrams (II-B1, II-B3, B9, B l l , B16, B19). The borings of LEGEND

-

L E Y E N DA

GREY TO BROWNGREY SANDY TO LOAMY "SOIL" SUELO A R E N O S O - L I M O S O , GRIS HAETA PARDO

DARK HUMIC "SOIL" SUELO HUMOSO OSCURO

SAND ARENA

SANDY CLAY ARCILLA ARENOSA

LAKE CLAY A R C I L L A DE L A G U N A

YELLOW M O T T L E D CLAY A R C I L L A CON MANCHAS

-~

~-~

AMARILLAS

Y E L L O W TO G R E E N I S H "LOESSIC" LOAM LIMO LOESICO~ A M A R I L L O HASTA V E R D O S O

PEAT

TURBA

HUMIC TO PEATY CLAY OR LOAM A R C I L L A ( O LIMO)I HUMOSA HASTA TURBOSA

Y E L L O W LOAMY F R A G M E N T S TO M O T T L E S FRAGMENTOS ( M A N C H A S ) L i M O S O S A M A R I L L O S

D

SMALL STONE F R A G M E N T S PEQUE~OS FRAGMENTOS DE ROCA

V O L C A N I C ASH ( M I C A ) CENIZA VOLCANICA (MICA)

A

CHARCOAL CAR ~ O N V E G E T A L

Fig. 2. Legend for the stratigraphical column of all figures.

the sections II-B1, II-B3 and B16 are situated in front of rock shelter II; the former two in a slight depression, the so-called secondary gully, and the latter (B16) exactly at the rim of this gully. Any mutual correlation between the different groups of diagrams (rock shelter, soil, gully) are not always immediately evident. Therefore, we have

10 distinguished separate zone systems for two of the three groups of diagrams, namely for the soil diagrams and for the gully diagrams in front of rock shelter II (including the related diagrams), which are indicated with a capital S and G, respectively, in combination with an Arabic cypher. For the description of the more fragmentary pollen diagrams of the rock shelters, introduction of a separate zone system seemed unnecessary. Finally, local E1 Abra pollen zones are introduced here. They are proposed in the description of the composite pollen diagram described in the section "Reconstruction and description of the composite pollen diagram". Correlation of all the different zone systems is given in Fig.14, together with an abstract of the various diagrams of which only the most striking features are mentioned. The pollen diagrams of the various sections will be described in the following order: 1. the rock-shelter sections (II-10E, II-100N, III-10E, III-12E, IV-107N); 2. the soil section (B9, B l l ) ; 3. the sections in the gully in front of rock shelter II (II-B1, II-B3) and other long sections near and in the slope of the valley (B16, B19).

The rock-shelter sections II-10E, II-100N, III-10E, III-12E, IV-107N (Figs.3--7) Pollen diagrams could only be made from parts of the rock shelter sections. The pollen is mostly poorly preserved. From the stratigraphical data we know that there are hiatuses in the different sections. Although for these reasons the final interpretation is sometimes complicated, a correlation with the other sections can be established.

Section II-IOE (Fig.3). In none of the spectra could a sufficient amount of pollen be counted. Moreover, the pollen grains are poorly preserved. The base of the section consists only of a "yellow mottled" layer. The material of these "yellow mottles" is most likely of volcanic origin (see section "Analysis of material of volcanic origin"). In the diagram just above this layer we find high percentages of Compositae tubuliflorae. In the stratigraphical column we see above the "yellow mottles" a dark humic layer succeeded by a relatively less humic and light-coloured layer on top of which a loess-like sediment (silty loam with a well-represented loess fraction) starts at the level of sample no. 3-23. This layer will be referred to as the "loessic" layer (Van der Hammen, 1978). This name only indicates a certain similarity to loess, and does not imply the same genesis. A little below this "loessic" layer we find clear indications of the presence of Botryococcus and of high percentages of fungal spores. Section II-IOON (Fig.4). In this section the pollen grains are also poorly preserved. The "yellow mottled" layer from the base of section II-10E is found approximately in the middle of this section. We find also an indication of high percentages of Compositae tubuliflorae associated with this "yellow

ca.

2570m

---

10

.

20

30

40

Brinkman

50

60

70

DEL

80

-

_21

" "~2_15~

3

""

9

5

1_1

1

.

.

"

.

.

.

.

.

.

.

.

COMPOS}TAE (- HYPERrCUM

...

i m

50

-

-

40

30 ~

"''I

TUBUUFLORAE:GRA TYPE )

100 9 0 8 0 7 0 6 0 ~, . . . . . . . . L

1_ 7

> 1

,--"/~ • 2_1

~ 2 - 1

•

~

"

10

MI N EA E

20

--

"1,

°/o

>_~

NUMERO! 6 mn~'DE 100 300 500

LA P L A C A 700 900

,D 10

20

30%

7

1

42

53

80

78

IT I]

IEi

Ir

Jill

I

iNCLUDED INCLUI°OS

~

~

~

~

i

:

10

~

~

I

~

20

-

,

-

'

I

i

-

~

- + ~ -

50%

I

i

5

1_ 1

1--

1_ 7

1_ 9

2_11

2_13

2_15

3_21

3_31 3_29 3_27 3_25 p ~ ~ 3_23

30 40

)NT i l e P O L L E N S U M EN L A SUMA DE P O L E N

,-n

0

~,. v) ~

OD-

o U

o

~:

za

~wO

~EO~O___

z~

~u 27

"3 ~O

~'ig.3. Pollen diagram of section [I-10E. For legend, see Fig. 2.

150`

,oo_

_

•

90

MAR

_)50-Cm--/./ 3_29~ 3_31 ~ , --~ ~ 3_27 3-25-. . . . . . . [ 3-23 --- --

0

-

SEALEVEL ELNIVEL

-ABOVE

-SOBRE

COLOMBIA, SOUTH AMERICA)

- 10 E

DE BOGOTA,

II

kNAL E.J. Schreve

~LT.

SABANA

-L ABRA

,,.a

,

-

lOO

N

co.

m

>'

/"

e

./..

77

5. v a n

4~

MAR

310 Z'O 10

OEL

%

I,OPAMI NEAE "

~'.-.-...-.-.-..>~-~,

TREES ',COMPOSITAE ARBOLES'~ TUBUUFLORAE

t

100 9 0 ~ 0 7,0 r~O 5 0

Geel

SEALEVEL EL NIVEL

- E~rinkmon

- ABOVE -SOP, RE

E.J. Schreve

2.570

ALNUS OTHER FOREST E L E M E N T S - O T R O S ELEMENTOS DE BOSQUE

IN(.;LUI3LD IN IHp- I"ULLI~NbUM EN LA SUMA DE POLLEN I NNCLUIDO$ ( 10 2 0 3 0 4 0 5 0 6 0 70 8 0 9 0 100 % I I t I I I I I I / ~-..ALE FOR A L L { POLLIEN) CURVES UNLESS OTHERWISE INOICATED ESCALA PAPA LAS CURVAS (DE P O L E N } SALVO A L G U N O S CASOS I N D I C A D O S

o)

zl:l ~ w ~ Z J,
uz j~~W -Jwm

~,

(~_~

J ~0 z---

~ig.4. Pollen diagram of section II-100N. For legend, see Fig.2.

100 -

!50

:m

~,NAL.

~LT

SABANA DE BOGOT,~. COLOMBIA. SOUTH AMERICA)

-L ABRA

u.

o o

lOT (NCLUDCD IN THE POLLENSUM I0 INCLUIDOS EN LA SUMA DE POLLEN

Ez:E

o,~o-~

>i--w (juJ

2,

,3

13

( I~1rl

o

w

o z w

~2

~o~ zaz

oex

( SOJ.I]Olr: N310d 30 Y~lns ~1~ s N 3 ] " l O d

U 0 i

~E

i

j{3

-I

I-

>

0"I"

o , =

.~-~

= 0

wj

~

~

o

o

~

~

o

0

,~v)

t-

o~ ,,l,,l,a,rl,11,1111,1,,,,i,,l, m m m m m m m m m m ~ ~

..........

a=

E

o z

~

-~

69~9-NJ~

14

EL

ABRA

(SABANA

-

12 E

DE BOGOTA, COLOMBIA, SOUTH AMERICA)

ALT ca, 2 5 7 0

ANAL,

III m

-ABOVE -SOBRE

E.J,Schreve-

SEALEVEL EL N I V E L DEL M A R

Brinkman

, _.~ XQI.-

O>.1~ dl~W L)~ u 3 NUM~ERI6mrr~OF

crrl

NUMERO!

•;

~

100

SLIDE eirnrr~DE LA PLACA 300 500 700

" INSERT.

SED. 4-FUNGI

INSERT.

SED. 4

;I-H~

4

BOTRYOCOCCUS

B,OT R Y O C O C C U S lllllll~

"f'r'r~r ~~

2-t+1111 II

FUNGi

II

BOTRYOCOCCUS

~ t-/+L/+h ~:IIIIII ~l],llll ~lJJllll ~ ~ llE±L±,llrl ll ]

~', -'1', ', l / ~,~

BOTRYOCOCCUS

~1]]1111

;H-~.

RAMINEAE

f~

RAMINEAE

. i II ', ','[ ', L ~ll,,,,,

40%-COMPOSITAE

TUD, ULIFL. 6 0 % (POLLEN }': 15) 76 % - C O M P O S I T A E TUB, ULIFL. 2 4 % ( P O L L E N ~" : 9 3 )

h~

Fig.6. Diagram of section III-12E. For legend, see Fig. 2.

15 m o t t l e d " layer, and also an increase of the Botryococcus curve between this layer and the base of the "loessic" layer starting at the top of the section. At the base o f the diagram the end of a possibly warmer fluctuation is visible. The curve of the Gramineae shows a high peak between this warmer fluctuation and the high percentages o f Compositae tubuliflorae mentioned above.

Section III-10E (Fig. 5). The pollen grains of this section are better preserved than those of the previously described sections. Just above the base of this section we find again the " y e l l o w m o t t l e d " layer, but unfortunately without pollen. Below this layer the curve of Gramineae shows high values (around 70%). In sample no. 1-7, where the pollen record starts again, there is a possible indication o f the end o f a slightly warmer fluctuation. In the further course of the diagram the Compositae tubuliflorae percentages increase. The percentages of Gramineae remain fairly high throughout the whole diagram. In the upper part, from sample no. 1-8 onward, Botryococcus appears in low percentages. This part corresponds with a dark loamy layer which is situated between the layer with " y e l l o w mottles" and the "loessic" layer, which starts from sample no. 2--14 onward. From the combined stratigraphical evidence it appears that there is a hiatus between this dark layer and the "loessic" layer. During the main period which is represented in this diagram, the climate has been fairly cold, with the exception of the relatively warmer fluctuation. Section III-12E (Fig.6). We have no pollen diagrams from this section, because pollen grains are lacking. The stratigraphy corresponds with that of section III-10E. Section I V - I O 7 N (Fig. 7). Also from this section we have no continuous pollen record. In the upper part of the section pollen is completely lacking. The discontinuity of the pollen record in the lower part of the section is caused by the fact that no material could be collected from this interval. The lower part o f the diagram, consisting of four samples of which only three contained pollen, forms part of a strongly humic to peaty clay deposit. The total percentages o f the elements forming open vegetation reach about 60%. The curve of Alnus is low. The high percentages of Weinmannia (sometimes exceeding 30%) are remarkable. Miconia pollen is also relatively well represented. The percentages of Botryococcus are low. In contrast, the curve of fungal spores shows very high values in the lower three spectra. However, in the uppermost of the four spectra this curve decreases to about 30%. The climate must have been relatively cool, certainly not really cold, because Weinmannia, a tree from the Andean forest, must have grown in situ according to the high percentages of its pollen. The climate may n o t have been very humid, b u t soil humidity was high enough for the accumulation of peaty clay. The next part of the diagram, after the gap in the pollen record, could be

16 divided into t w o parts. The lower part ranging from sample no. 1-1 through sample no. 2-12 shows relatively high tree-pollen percentages, whereas the upper part, ranging from sample no.2-13 through sample n0.3-25, shows high percentages of non-arboreal pollen, especially of grasses. The aspect of the lower part of this section is dominated by negatively correlated fluctuations in the curves of Alnus and Myrica. During the minimum in the Alnus curve the curve of Gramineae is relatively high. In the stratigraphical column at the levels of samples no.l-1 and no.l-2, a relatively coarse, dark-coloured and loose sand is found, followed by an alternation of very dark loamy layers with intercalations of light-coloured and relatively coarse sand. As we move upwards in this part of the section, the role of the sandy c o m p o n e n t increases. During this interval, the climate has been moderately warm with possibly cooler intermittent periods (corresponding with a greater abundance of Gramineae). The upper section of the diagram is characterized by high percentages of Gramineae. However, before the curve of the Gramineae reaches its maximum between 65% and 80%, the curve of the Compositae tubuliflorae shows a small peak o f a b o u t 20%. There are also two relatively warmer fluctuations in this part of the diagram, which are mainly manifested in the Alnus curve. At the same time the curves of Gramineae and Compositae tubuliflorae show a slight decline. In the last two spectra we see the beginning of an increase of the Compositae tubuliflorae percentages. In the same spectrum Botryococcus is represented by very low percentages. The upper surface of the grey layer is in this part of the profile clearly uneven and locally somewhat eroded. Immediately on top of this level rests the "loessic" layer.

The soil sections (B9 and B l l ) (Figs. 8, 9) Section B l l corresponds with the soil of the E1 Abra valley just outside the influence of the secondary gully, which is clearly visible in front of rock shelter II. Section B9 corresponds with the soil in the middle of the E1 Abra valley. Thus we can expect that in both sections the stratigraphy and the succession of the pollen zones will be comparable. The t o p of section B9 is situated 60 cm higher than the t o p of section B l l . When studying these sections, we have to bear in mind that they represent a series of (paleo) soils, alternating with and including true sedimentary layers. For the differentiation of zones, certain lithological data have been used besides the pollen data.: Given the nature of these sections, hiatuses may be expected, and thin layers may correspond with large time intervals. The sample distance is 5 cm. Regarding the possibly slow accumulation or sedimentation rate, this sample distance is considerable. It is, therefore, to be expected that pronounced maxima corresponding with thin deposits in one diagram, may be almost or completely lacking in the other one. It will be tried to detect the really significant changes in the pollen curves.

EL

ABRA

(SABANA

IV - 1 0 7

N

o

DE BOGOT,~, COLOMBIA, SOUTH-AMERICA)

pp. 1 7 - - 2 2 .

o

"u4 ~m 5w

ALT

ca.2570

m

- A~OVE

SEALEVEL

SOE~RE E L N I V E L

ANAL

U

z~: DEL

MAR

E2~ 8

tJ~

v) LUO NUMBER / z ~ -6ram ~ z~ w~ OF SLIDE j _ NUMERO / 0 ~ -~ 6 m m I OE * "^ ~ LA PLACA (9 1(30 200 3(

ELI. S c h r e v e - B r i n k m a n cm

o

~

>- < > - <

....

<[

o~

<

~ ~ ~ ~oo<~

~

~-~-

DWm_

o ~ o 8o o~ o~

~ m e:rn J OO'~<©

j

a_

~:E

U

u1-4

o
~

~:~°~

o

o

eJ

~"

O~:E J

~

S

D LL

80

%

~

-14_59 13_58 -~2_57 d 11-56 10_55 9 - 54 8_53 -7_52 6_51 5_48 5_47 5_45 5_45 5_44 --

i

5_43

5_42 5_41 4-40 4_39 4_38 4_37 4_36 4_35 4_34 4_33 4_32 4_31 3_30 3-29 3_28 3_27 3 _ 26 3_25 3 - 24 3 - 23 3_22 3_21 2_20 2_19 2_18 2_17 2_16 2_15 2_14 2_13 2_12 2_11

,

.... k:)

--

--

---

--"

_~

I

f

I

/

!

-

. . . . . . . .

....

J,

'

-

-

-

1_ "io

-

"

"

1_ 1_ 1_ 1 1_ 1_

9 8 7 6 5 4

1_ 1_

3 2

1_ 1

k

.

LTREES

t.:.:.:.:.~::,:.l

-~

[COM POSITAE IGRAMINEAE ARBOLES ~TUBULIFL * :HYPER CUM TYPE! ~+ ERICACEAE

10

20

30

40

50

60

70

80

90

100°/o

SCALE FOR ALL [POLLEN) CURVES UNLESS OTHERWISE INDICATED ESCALA PARA LAS CURVAS (DE P O L E N ) S A L V O A L G U N O S CASOS INDICADOS o • x

ALNUS PoDOCARPUS OTHER FOREST ELEMENTS -OTROS E L E M E N T O S

Fig. 7. P o l l e n diagram o f s e c t i o n I V - 1 0 7 N . F o r legend, see Fig.2.

.

.

.

....

N O T INCLUDED IN THE POLLENSUM NO INCLUIDOS EN LA S U M A DE POLEN

INCLUDED IN THE POLLENSUM INCLUIDOS EN LA SUMA DE POLEN 0

.

DE B O S Q U E

--

I

-J

q

L

EL

ABRA

B 11

(SAIAANA DE B O G O T A , C O L O M B I A , SOUTH

A_u o uLn i--

AMERICA)

<

Pp. 23--28.

£ 0

:E

,T

-i

~o

~ 3

,Jp_

50

ALT. c o ,

elm Z>-

2570

m

- ABOVE

SEALEVEL

SOBRE

gu Z<

£>

ANAL.

_jW <:l: UI-

cm

9~

E.J Schreve-

EL NIVEL

DEL

50

MAR

L~

~

u~O hi Z ~ tn N U M B E R / 6 m m ~ OF eJ .J< < SLIDE j :3E ( 2~ N U M E R O / <~ 0 ~-~ ~ 6 m m ~ D E ~9 LA PLACA

Brinkmon

~)

o

:E

z

~

5

:E D

i,~1<
u o

-

_~

z

E

J_w

~

w~

w

o~

mOOJz< Oz~m
;#F ow LL~

j

-~ ~q

:Ec~

oo~

#

OOI~,

~w

~,o

o o. o

>-<

~ o u u ~ u

w

O

Z

s

o

Ili,

I

I

____+

- - ~

2

-

--

3

4 5 6

1_ 7 11-

8 9

1_10 1_1 1_12 1_13 1_14 1_15 1_16 1_17

1DO

2.18 - 2_19

2-20 -2.21 • 2_22 2-23 2-24 -2-25

15

2-26

• 2-27

i

F: . . . . . . .

~

TREES :COM POSITAE IGRAM[NEAE ARBOLES ,TUBULIFL. * : HYPERICUM TYPE: ,~ ERICACEAE

i

0

NOT INCLUDED IN THE POLLENSUM NO INCLUIDOS EN LA SUMA DE POLEN

INCLUIDOS EN LA SUMA DE POLEN 10 20 30 40 50 60 70 80 90 100 %

SCALE FOR ALL ( P O L L E N ) C U R V E S UNLESS OTHERWISE INDICATED ESCALA PARA LAS CURVAS (DE POLEN) SALVO A L G U N O S CASOS INDICADOS c] • • x

ALNUS QUERCUS PODOCARPUS OTHER FOREST E L E M E N T S - O T R O S ELEMENTOS DE B O S Q U E

Fig.8. Pollen diagram o f section B l l . For legend, see Fig.2.

EL

ABRA

B - 9

£ 7

(SABANA

DE BOGOTA, C O L O M B I A , SOUTH-AMERICA)

,?

=~ '-<

o

05

o

:E

~,£

wJ

w ALZ

uz j~t n ~ Jwo SOBRE EL NIVEL DEL MAR o_~ 5" 0 n ~J E.J. SchreveBrinkmon z~omu zw w<[< -J O NUMBER / 6 r a m * <~ J:E-O F SLIDE Om J NUMERO/ 9 r a m ~ ~9 O 10 20 30 40 50 60 70 80 90 % a_ u~ Ln m r I ~: Pa z~P~

co.

Z - > -~

.J ANAL,

N

cm

, - -

2570m

-

ABOVE

SEALEVEL

<

~,== ~_~

-

8

~ ~J

<

z

~

)

0 d]

(9~DO

~>-~ J_IE 00< o_a.o

oJ

>-

w o: < w w:E< w

~

< -

w c~ v~ o_

a

:E

D

o_

z~ LLI:E :zO Ou ,

~2

~ wo

O0

~-z ww

_-2:E

O

Oj

:E

O~ :E'~'

zo

d ~J J

oB

z Z OD :ELL

<

_z

I

1 2 3

./

A

5(

S6

(

i

S5

c 1( _1

IO(

t

_1: _1, _1! _11 _1

S4

.ll

.1! _2( _2 2 _2: 2 _2:

E $3

15£

• 2_2 ~ 2_2~

S2

i

j

L TREES ',COM POS TAE nGRAMINEAE ARBOLES I TUBULIFLORAE • '] IHYPERICU M TYPE ,* ERICACEAE I i

INCLUIDOS EN LA S U M A DE POLEN O 10 20 30 40 50 60 70 8 0 90 100 % SCALE FOR A L L ( P O L L E N ) CURVES UNLESS OTHERWISE INDICATED ESCALA PARA LAS CURVAS (DE P O L E N ) SALVO ALGUNOS CASOS INDICADOS

[] • • x

Fig.9. Pollen diagram of section B9. For legend, see Fig.2.

ALNUS QUERCUS PODOCARPUS OTHER FOREST E L E M E N T S - O T R O S ELEMENTOS DE ~ O S Q U E

NOT INCLUDED IN THE POLLENSUM NO INCLUIDOS EN LA SUMA DE POLEN

2 -2( 2_2; • 2_2~ 2_2¢.

- 2-3(

2-31

29

Section B11 (Fig. 8) Zone $1. The base of section B l l forms part of a lake-clay layer with yellow mottles o f pedogenic origin. It is important to separate this part of the section (as zone $1 } from the rest of the section, because this is the only part of the section in pure lake-clay. Zone $1 is characterized by high percentages of Alnus and Myrica pollen (the latter up to 60%), and by low values for the curves of Gramineae and Compositae tubuliflorae. Only in the lowest spectrum was a somewhat higher percentage of the latter (30%) found. The percentages of Ericaceae are distinctly higher than in the other parts of the diagram. The curve of Alnus, and those of Myrica and of the Symplocos show alternating maxima and minima and are negatively correlated. The curves of the spores show high values, the Hymenophyllum curve excepted. The curve of Botryococcus also attains very high percentages, but decreases considerably towards the end of zone S1. The sedimentation of lake-clay and the high percentages of Botryococcus indicate open water. The correlation of high percentages of Botryococcus and of Monolete verrucate fern spores is rather striking. Judging from the high percentages of tree pollen in zone $1, a w o o d y vegetation must have existed. Alnus must have been succeeded by Myrica and Symplocos as most important vegetation elements, and vice versa. Possibly, this succession is related with the progressive desiccation of the E1 Abra corridor. At the site of the section itself there must still have been open water, judging by the high percentages of algae. Towards the end of zone-S1 conditions become drier, however. This might also be represented in the stratigraphical column by the yellow mottles caused by pedogenesis. The position of this section in the E1 Abra corridor indicates that the valley itself must have been inundated during zone S1. The climate must have been relatively warm and moist. Only towards the end of zone S1 did the conditions become drier. The transition from zone S1 to zone $2 is characterized by a strong increase of the Gramineae percentages, a slight rise of the curve o f Compositae tubuliflorae and a decrease of the curves of Alnus and Myrica. In the curves of the Jamesonia type and the Monolete verrucate fern spores we see also a sharp fall. The curve of Botryococcus is almost reduced to nil. The local vegetation became more open. Zone $2 is characterized by high percentages of Gramineae. The values of the curves o f Compositae tubuliflorae and Alnus are low. The curve of Myrica shows relatively low values as compared to the preceding zone S1. The curve of Hymenophyllum spores shows a distinct peak. The curve of Botryococcus is almost zero. This indicates the completion of the desiccation of the E1 Abra corridor, a process which already started in the uppermost part of zone S1. At the same time the Monolete verrucate fern spores curve attains a minimum. In the stratigraphical column we see a dark, almost black, and very loamy layer. An open t y p e of vegetation must have prevailed. The climate was relatively cool and dry. The transition from zone $2 to $3 is determined by

30 a decrease of the Gramineae curve and a rise of the Alnus curve. In the stratigraphical column the same kind of " y e l l o w mottles" of volcanic origin (see the section "Analysis of material of volcanic origin") occur in the t o p m o s t part of the very dark loamy layer, as reported for various rock-shelter sections (although the stratigraphic position might n o t be the same as will be discussed later).

Zone S3.]Although the Gramineae curve decreases at the beginning of zone $3, the percentages are still higher than in zone S1 (ca. 45--50%). The curve of Compositae tubuliflorae shows a distinct rise towards the t o p of zone $3. The Alnus curve starts with values of ca. 45% but half-way through zone $3 the curve falls to values of a b o u t 20%. During zone $3 we see the sudden appearance of some herbs: Borreria, Polygonum, Caryophyllaceae, Chenopodiaceae-Amaranthaceae type, Compositae liguliflorae, and UmbeUiferae. The experience obtained from similar p h e n o m e n a in other dated diagrams indicates a possible connection with a warmer climate (especially high percentages of Borreria seem to be good pointers). The curve of Cyperaceae has a distinct maximum, which may indicate wetter conditions. The slight rise of the Botryococcus curve confirms this. The vegetation is more w o o d y as compared with zone $2, although the sum total of Gramineae and Compositae tubuliflorae clearly indicates that a veritably dense forest vegetation cannot have existed. At the beginning of zone $3 the climate was relatively warmer and more humid than in zone $2, to become colder again towards the end of zone $3. The diagram gives the impression of the existence of a warmer fluctuation, which can be subdivided into an earlier period with a relatively more closed, w o o d y vegetation and a later period with a relatively more open and more herbaceous vegetation. At the transition of these t w o periods a minor peak in the Gramineae curve is found, which seems here of no great importance. It must be borne in mind, however, that the sample distance is relatively great. The transition from zone $3 to zone $4 is characterized b y an increase of the curve of Gramineae and a strong decrease of the Alnus curve in addition to an important change in the sediment deposition from a humic sandy loam to a "loessic" loam. Zone $4. In zone $4 t h e tree pollen curves reach e x ~ e m e l y low values (under 10%). The curves of Gramineae, Compositae tubuliflorae and the Hypericum t y p e have very high values (in total over 90%). The vegetation must have been completely open. The curve of Cyperaceae shows a distinct minimum and is in parts reduced to almost nil. The Botryococcus curve also shows a minimum. This indicates drier conditions than in the preceding zone $3. The climate must have been relatively cold and dry during the main part of zone $4. The stratigraphic column shows that during zone $4 a loess-like sediment was deposited (silty loam with a well-represented loess fraction). In the lower part of zone $4, Weinmannia appears as a low and small peak. We cannot give a satisfactory explanation, because the local conditions cannot

31 have permitted the establishment of Weinmannia at the site. Towards the end of zone $4 the curve of Botryococcus starts rising again synchronously with a slight increase of the Alnus curve. Apparently the conditions became wetter. Again the rise o f the Monolete verrucate fern spores curve coincides with an increase of the Botryococcus curve. The transition from zone $4 to zone $5 is characterized by a further increase of the percentages of Alnus and Botryococcus. Conditions became relatively warmer and wetter and the influence of pedogenesis became more important. A further subdivision of section B l l becomes rather problematic.

Zones $5, $6, $7. An a t t e m p t will be made to make a further differentiation into zones with the help of the diagram of section B9. As stated above, a further differentiation based on the data of section B l l alone is very unreliable. It is perfectly clear that there is a top in the Botryococcus curve in zone $5 immediately overlying zone $4, again coinciding with a top in the curve of Monolete verrucate fern spores. After the final decrease of the Monolete verrucate fern spores, some herbs appear on the scene, which also appeared in zone S3. This point will be discussed below when the diagram of section B9 is described. Section B9 (Fig. 9) The diagram of section B9 is, generally speaking, a repetition of the diagram o f section B l l . Zone $1. The diagram of section B9 starts with a high peak in the Gramineae curve. This very narrow distinct peak does not appear in the diagram of section B l l . The small Gramineae peak of sample no.2-25 of section B l l could possibly be related to this high peak. In the stratigraphy there is no difference as far as section B l l is concerned. For the same reason as indicated in the description of section B l l , we differentiate this part situated in the lake-clay from the rest of the section. Again, the Myrica curve shows high values. Also the Botryococcus curve reaches high percentages. The vegetation must have been alternatingly a forest and an open grassland. The climate was relatively humid judging by the lake facies with high percentages of Botryococcus, and, temporarily, relatively warm. The E1 Abra valley must have been inundated and must have formed a part of the great lake of the Sabana de Bogota. The transition from zone S1 to zone $2 is determined by the rise of the Gramineae curve and a marked decrease of the Myrica curve. Zone $2 shows the same characteristics as described in section B l l . Zone $3 starts with a relatively warmer fluctuation judging by high percentages of the Alnus curve and low values of the Gramineae curve. The situation as described in section B l l is repeated. However, the Gramineae peak appears to be far more distinct in this section than the corresponding minor peak in

32 B l l . This high Gramineae peak must represent a completely open vegetation. The total percentages of tree pollen remain below 10%. After this peak in the Gramineae curve a slight rise of the curve of the other forest elements t o o k place again, the curve of Alnus retaining the same percentages. The curve of Compositae tubuliflorae shows a very distinct rise. In the diagram of section B l l this rise already started at a lower level. In the t o p of zone $3 the curve of Botryococcus shows a more distinct rise (to a b o u t 40%) than it does in section B l l . The concurrent change in the curve of Monolete verrucate fern spores is rather striking. Unfortunately the group of herbs which shows such remarkable peaks in zone $3 of section B l l is completely lacking in this section. There is no difference in the stratigraphy of this part between section B9 and B I I . At first the climate was relatively warmer in zone $3 as compared to zone $2, but during the Gramineae maximum the climate must have been colder in zone $3 than it was in zone $2. Only towards the end of zone $3 did the climate improve again somewhat. Throughout zone $3 the humidity increased. The percentages of Botryococcus are n o t so high that the occurrence of open water in the whole El Abra corridor may be accepted. However, periodically, lower parts of the valley may have become inundated. The transition from zone $3 to zone $4 is characterized by an increase of the Gramineae percentages and a further falling off of the Alnus curve.

Zone $4. As in the corresponding section B l l , zone $4 is characterized by high percentages of elements of open vegetation types. In the stratigraphical column only in the lower part of zone $4 the "loessic" material is well developed. The upper half of the "loessic" layer is influenced by pedogenesis. Towards the end of zone $4 the "loessic" character disappears completely; instead the material more resembles a very sandy loam. The Botryococcus curve has a minimum of a b o u t 30% in the lower part of zone $4, which in conjunction with the occurrence of an open vegetation type, is indicative of the prevalence of relatively dry conditions. Half-way through zone $4 the Botryococcus curve starts rising again towards values of a b o u t 40% in the top part. In the stratigraphical column this rise coincides with the level where the material becomes more humic, and subsequently the "loessic" material is replaced by the accumulation of a humic sandy to l o a m y "soil". In the beginning the climate must have been relatively cold and dry. In the upper part of zone $4 the conditions became wetter, the climate remaining cold. Towards this point the various changes in the diagrams of both sections run more or less parallel. Through the remainder of section B9 the various changes seem to be better reflected than in section B l l . Section B9 will first be described in more detail, subsequently to be compared in its course with that o f section B l l . The transition from zone $4 to zone $5 is characterized by a marked

33 decrease of the Gramineae curve (from 70% to about 40%). The appearance of pollen of some specific herbs is also indicative o f the beginning of zone $5.

Zone $5. In zone $5 the Alnus curve attains a distinct maximum, the curve of Gramineae showing a distinct minimum. The percentages of Compositae tubuliflorae are also low. An important feature is the occurrence of some herbs, apparently at least partly from the undergrowth of the Andean forest, such as Borreria, Caryophyllaceae, Chenopodiaceae-Amaranthaceae type, Compositae liguliflorae, Jussiaea, Hydrocotyle type, and Monolete verrucate fern spores, which are not found in the zones $2, S3 and $4. The rise of Botryococcus, which started already in the upper part of zone $4, continues. Again, the Monolete verrucate fern spores curve follows the curve of Botryococcus towards a m a x i m u m of over 80% in the top part of zone $5. Througho u t zone $5 the percentages of Cyperaceae are relatively high (15%--28%). The combination of moderately high values of Botryococcus percentages and the permanently well-represented Cyperaceae points to relatively h u m i d conditions. It is not very likely that the E1 Abra corridor itself was inundated; however, the valley b o t t o m must have been marshy. In lower parts of the valley there may have been open water (e.g., in the secondary gully in front of rock shelter II). The climate must have been relatively warm and humid. Zones $6 and $7. In the diagram there is a distinct limit between zone $5 and $6. The Alnus percentages fall to values below 10%. The percentages of Gramineae increase and the curve of Compositae tubuliflorae is also rising (from about 20% to about 45%). The curve of Botryococcus decreases rapidly. At the same time the curve of fungal spores rises to remain at a high level in the rest of the diagram. The transition from zone $6 to zone $7 is n o t very distinct. However, we could place this transition at the point where the Alnus curve begins to increase again and the Gramineae curve starts to decrease. In this way the peak of about 45% of the Compositae tubuliflorae belongs to zone $6, whereas zone $7 shows somewhat decreasing values. In this part of the diagram the stratification of the pollen material seems highly disturbed by homogenization associated with pedogenesis. Section B11 Zones $5, $6 and $7. In Section B9, zone S5 is characterized by high Alnus percentages and a m i n i m u m in the Gramineae curve. These characteristics are n o t reflected in the diagram of this section. Fortunately, however, the peak in the Botryococcus curve is very distinct. Most probably this peak may be correlated with the m a x i m u m in section B9. This zone S5 is only manifest in one single spectrum (1-9). Zone S6 is characterized by a m a x i m u m in the Gramineae curve, by high values of the curve of Compositae tubuliflorae, and by low percentages of Alnus. In section B l l only a rise of the curve of Gramineae can be noted. Z o n e $7 might start with the appearance of some herbs, e.g., Borreria,

34

Polygonum, CaryophyUaceae, Chenopodiaceae-Amaranthaceae t y p e and Compositae liguliflorae. The occurrence of some of these is mainly limited to the Andean forest belt. At the same time the percentages of Myrica are increasing and the curve of Hedyosmum starts again. The presence of these species clearly indicates a relatively warmer climate. Also in this part of the section a homogenization of the pollen material, related to pedogenesis, seems to have taken place. A striking feature in both diagrams is the appearance of maxima in the curves of Insert(ae)sed(es) 2, together with maxima in the curve of Botryococcus, after zone S1. During the interval S1 the presence of open-water facies in the E1 Abra valley is very likely. In the rest of the zones a marshy facies will have existed only periodically. It is only during these periods that maxima occur in the curves of Insert.sed.2. From this evidence we may tentatively conclude that the occurrence of Insert.sed.2 is connected with relatively humid conditions. Recently, these micro-fossils were determined as zygospores of the alga Debarya, Zygnemataceae (Van Geel and Van der Hammen, 1978) and they are indicative of a relatively cooler climate. The sections in the gully in front o f rock shelter H (II-B1 and II-B3), and two additional long sections (B16 and B19) (Figs. 10--13) In front o f rock shelter II there is a slight but manifest depression. As we shall learn from the diagrams of the sections taken in this depression (II-B1, II-B3), it has at times played an important role as a gully or lower-lying place with stagnant water in the past vegetational history of the valley. In earlier times, when the gully apparently did not y e t exist, the succession in the stratigraphy and vegetation, in particular of the diagram of section B16, is very well comparable with that of section II-B1 (II-B3 is not important in this connection because it only represents the uppermost strata).

Section II-B1 (Fig. 1 O) Section II-B1 is almost situated in the deepest part of the gully in front of rock shelter II. The boring of this section, like those of the sections B16 and B19, reaches a far lower level than the other sections. Zone G1. We place the basis of this section in the provisional zone G1. This zone is characterized by low values of the percentage of Gramineae and Compositae tubuliflorae. The percentages of Alnus pollen are between 20% and 30% with one t o p of a b o u t 50%. The percentages of Weinmannia are remarkably high (about 50%). This implies that Weinmannia was growing in situ at that time. Miconia pollen is also well represented by a b o u t 20%. Extreme values of up to 900% can be seen in the curve of the fungal remains. This maximum of 900% originates from a very peaty layer (the whole lower half of the section consists of peaty clay). The local stand of vegetation must have been a dense forest, and hence, the climate must have been relatively warm.

35 The transition from zone G1 to zone G2 is characterized by a marked increase o f the Gramineae percentages.

Zone G2. Throughout zone G2 the Gramineae are well represented (about 25%--50%). The Alnus curve continues with values lying between about 20% and 45%. The curve of Weinmannia still shows relatively high values (of about 20%), with one m a x i m u m in the upper part of zone G2 of about 50%. The Hypericum-type curve shows a peak of over 20% in the lower part of zone G2. Fungal spores remain very well represented with the exception of two minima of about 35% and about 3%, synchronous with the maxima in the curves of the pollen of the Hypericum type and of Weinmannia, respectively. In the stratigraphy we find again the peaty clay of zone G1. The vegetation must have been of a more open type than in zone G1; however, trees still played an important role in the whole stand of vegetation. The climate must have been somewhat cooler than it was in zone G1. The transition from zone G2 to zone G3 is determined by a strong decrease in the curve of Gramineae and an important increase of the Alnus pollen percentages. Zone G3. The diagram has rather constant characteristics t h r o u g h o u t zone G3. The percentages of Gramineae are extremely low, as are those of the Compositae tubuliflorae. The Alnus curve attains high values, mostly above 50%. Weinmannia pollen is still well-represented (mostly by more than 10%). The fungal spores have decreased in importance. Between the two maxima in the curve at the base and at the top of zone G3, the mean values of the curve remain low. In a few spectra some specimens of Botryococcus (not more than 1%) have been observed. In the stratigraphy the picture is still the same: peaty clay, locally becoming more peaty or more clayey. The more peaty parts coincide with the maxima in the curve of the fungal spores. During this interval the climate must have been relatively warm. The transition from zone G3 to zone G4 is characterized by the decrease of arboreal pollen, manifestly so in the Alnus curve, and the increase of elements typical of open vegetation. The extreme increase of the Hypericum type curve is the most remarkable feature. Zone G4. In this section, the interval G4 is subdivided into three subzones, viz., G4a, G4b, G4c. In the lower part of zone G4 (subzone G4a) the Hypericum-type curve has a clear m a x i m u m of about 40%, synchronous with a less pronounced m a x i m u m in the Gramineae curve (of about 25%). These maxima of elements of open vegetation types coincide with a marked depression in the Alnus curve. In the adjacent subzone G4b the Alnus curve reaches a m a x i m u m (of about 70%), and at the same time the curves of Gramineae and o f the Hypericum type fall to extremely low values. In subzone G4c the conditions of subzone G4a seem to have been restored, but the peak in the Hypericum-type curve is followed by a peak in the curve of Compositae tubuliflorae. In between these two maxima, the Alnus curve reaches a value

36 of a b o u t 40%. Remarkable is the curve of Jamesonia-type spores, which reaches values lying between 10% and 20% in the subzones G4a, G4b, and the first part of G4c. In zone G4 an open vegetation prevailed in subzones G4a and G4c. This open vegetation was replaced by an alder forest in subzone G4b. The climate must have been alternately relatively cold and warm judging by the occurrence of more open and more silvan vegetation types, respectively. The transition from zone G4 to zone G5 is characterized by a marked decrease of the curves of Gramineae and Compositae tubuliflorae, while the Alnus curve is increasing rapidly (up to a b o u t 85%).

Zone G5. The sudden appe .ar_ance of Monolete verrucate fern spores and of Botryococcus, and the almost complete disappearance of Weinmannia pollen in the beginning of zone G5 is noteworthy. As will be discussed in the section " R e c o n s t r u c t i o n and description of the composite pollen diagram", from n o w on there was a gully in front of rock shelter II. Due to erosion activities, during periods with a relatively low water table when the gully functioned as a little stream bed, there are several hiatuses in the sedimental succession from n o w on. For this reason zonations are n o t always very clear. In zone G5 the curve of Gramineae rises gradually from a b o u t 5% to a b o u t 45%, whereas the Alnus curve, starting with a value of ca. 85%, gradually decreases. The percentages of Compositae tubuliflorae remain at a b o u t 10%. Myriophyllum reappears (after a previous occurrence of short duration in the uppermost part of interval G2), and pollen of the Hydrocotyle t y p e is better represented. The curves o f various spores exhibit a small rise. The increase of the Hymenophyllum and Monolete verrucate fern spores percentages is very substantial. The Botryococcus curve shows t w o peaks of a b o u t 60%, somewhat retarded as compared with the curve of Monolete verrucate fern spores. For the first time in this section the Insert(ae) sed(es) 2 t y p e appears on the scene (zygospores of the alga Debarya; Van Geel and Van der Hammen, 1978). The presence of Botryococcus indicates wetter conditions. The presence of lake clay in the stratigraphical sequence of the gully points in the same direction. Temporarily, at least in the lowest parts of the valley, there may have been open water. In the beginning of zone G5 there must have been an alder forest, but elements of more open vegetation types increasingly infiltrate throughout this zone. In the very beginning of zone G5 the climate must have been relatively warm and humid, gradually to become colder and also more humid as the pollen types representing a more open vegetation type increase in number. The presence of Myriophyllum as well as Insert.sed.2 is indicative of a relatively cooler climate. Zone G6. During zone G6 the curves of Gramineae and Alnus pollen maintain average values lying between 40% and 35%, respectively. The curve of Botryo-

EL

ABRA

i.I -

(SABANA u~

ALT.

ca

DE BOGOT~k, COLOMBIA, SOUTH- AMERICA)

2570

m

o~

~

ANAL.

NO~

B1

ABOVE

5EALEVEL

SOB'RE

E.L N I V E L

E. J, S c h r e v e B. von

7,

~ ~

ow2

mE

E~r~¢INUIMBER/ w ¢h w O w ~ 6 m m , "{ I~

B.rtnkmon

Geei

d

u~ u

za.c=

MAR

w~

w

zw°~w DEL

pp. 3 7 - - 4 4 .

2

o

~,;,< <

~U

o-°W

O~ z

w <~=~Oz ~ On

w <

w~

~u<

3z~oE ~ o~ ~oozw ~o~=~

w ~w P~U

OW

E<

~D

~0<

2 ~n

~

~oo o ~ 0w ~

u~I

~

~

w

::"

c~

w

w

G J

~o

oo ~ =

w

w

G o ~

o

oo e~=

(

C

~TA

5_1

5_2 :v: 6_1 6-2

II

7_2

I I I

8-1

I i

8_2 9_1 9-2

i

10_1 ro_2 t l _1

y

--

v v

v

v

m

v I v vt v v v v cvvw

__

vv v,¢, v

g=

v

L

450

Z

V

~v v ~vv v v

500

~

I

u v

v~ 600

-v--

650 t

70(?

'.

1 v

v

v~

I

5T.:.:.:.:.:.:,:.:.:_~

TREES

~,COMPOSITAE

GRAMINEAE

ARBOLE$ ITUBULIFLORAE ~ i ',HYPERICUM TYPE ' i+ E R I C A C E A E

/

INCLUIDOS EN LA S U M A 0

10

20 3 0

40

50

60

N(JI INL.LUIJb. IJ 11"4 [~1~ M U L L E N b U ~ NO INCLUIDOS EN LA SUHA DE P O L E N

DE POLEN 7Q 8 0

90

1Q0%

SCALE FOR ALL (POLLEN) CURVES UNLESS OIHERWrSE INDICATED ESCALA PARA L A S CURVAS ( D E P O L E N ) SALVO A L G U N O S CASOS INDICADOS

Fig.10. Pollen diagram of secion II-B1. For legend, see Fig.2.

13 A L N U S

• QUERCUS • PODOCAR PUS x OTHER FOREST ELEMENTS - OTROS E L E M E N T Q S

DE ~,OSCLUE

I I

EL ABRA II-B3

'

u~ o F-

(SABANA DE BOGOTA,COLOMBIA,

i

~J <

=~
SOUTHAM£RICA) "'" i,i '1t-)-

PP. 45--52.

n >U P

ul J

J

0 L

z u

J

ALT co. 2570 m - ABOVE SEALEVEL

~Zvk~ ~0 .J. (L

- SOBREELNIVELDELMAR

O~ m~O ANAL

EJSchreve-BrJnkrnan

.JZ ,<,~ UU

q~

Cm A

.0-~ U m~rnU z a . NUMBERt6mm:~ z "~ OF SLIDE Jz0~,a NUWERO!6ram= ~.m~n DE LA PLACA

D

L~

F ~J < _~

F

~L~D 0

^i

U

u.,I

I u -~ <

(j

[d I~

U

U404~

D: 0

~'z

<~J~<

,¢ D N t-

p- I-" ii

-, b~O~'n D_(1.~ J~U v ~.

.

.

.

.

.

.

<~:,

,(OFo

plm m ~).

gz ~.U

zO

>F O~ U.E

~

J

,t~j

h~

~g

~g

>a_

I: ~D O0 O0

}hi 0 .

I,.-

O0 UU

.-I

Z

0 Z 0

0 0--4-

.

) 5{

xt

j,I. EOX

>

U O L~ 0 rr

¢,..

r~ bJ

]

I--

~[

lk

hi U3 7

_J 0

N <

I

-|-

-k :

m /

I

40 39 38 37 36 35 34 33 32 3l 3O 2g 28 27 25 25 24 Z3 22 21 20 1B 18 17 16 15 14 33 12 11 10 g 8

VO

7

I

t 0

I/

J

INCLUBED HHTHE POLLENSUM INCLUIDOS Eq LA SUMA DE POLEN

TREES ARBOLES

,COMROSITAEIGRAM!NEAE 'TUBULIFL.+~ BYPERICIJM i ;TYPE ~. i IERICACEAE {

0 10 20 30 40 50 60 70 80 90 100%

SCALEFO'RALL {POLLEN)CURVES UNLESSOT'HER~ISF.INDICATED ESCALA PARALAS CURVAS{OE POLEN] SALVOAL~UNOS CASOSINDICAOOS

a ALNUS • &~ERCUS + PODOCARPUS

Fi~tL PoI[en.dia~ar~ of section I[-B3. For legend, see Fi~.2,

x OTHERFORESTELEMENTS~ OTROSELENENTOS DE BOSQUE

~UI I~<~LUL~L~ IN C~ VULLIr.~bU~ NO INCLUIDO5 EN IA SUMA ~£ POLEN

6

5 4

3 2 -I

53 coccus has a distinct minimum at the start of zone G6 and increases towards the end o f zone G6 to a small maximum. As already indicated before, there must be several hiatuses in the sedimentary succession o f the gully owing to erosion. Conditions must have been somewhat colder and less humid than in the preceding zone G5. The transition to zone G7 is marked by a distinct rise of the Botryococcus curve and a slight rise o f the Alnus curve. Zone G7 is marked by high values of the Botryococcus curve. The Alnus pollen percentages, initially a b o u t 45%, decrease gradually. At first, conditions were relatively warm and humid to become colder and more humid towards the end of this zone. In the part of the diagram corresponding to G5, G6 and G7, the curves of trees and herbs belonging to the Andean forest zone are only represented b y very low percentages (with the exception of the Alnus curve); some of these curves show a slight depression in G6. In the stratigraphical column there is no change in the part corresponding to G5, G6 and GT. The sediment consists of lake clay, with only at the base (zone G5) some admixture o f sand. The transition from zone G7 to zone G8 is determined b y the beginning of a minimum in the Alnus curve and the initiation of a high representation of Compositae tubuliflorae. Apparently the climate between relatively cold but remained wet. Z o n e G8 is characterized by high percentages of elements of open vegetation and low Alnus percentages. In order to describe the events that occurred in the interval G8 better, we divide this zone into three subzones: G8a, G8b and G8c. S u b z o n e G8a shows a maximum in the curve of the Gramineae ( o f about 65%). The percentages of Compositae tubuliflorae are higher than they are in zone G7. The Alnus percentages hardly exceed 10%. The curve of Jamesoniatype spores falls remarkably quickly from values of a b o u t 30% in zone G7 to values under 5%. At first, the percentages of Monolete verrucate fern spores and Botryococcus remain high, but they fall rapidly to values of a b o u t 10%. Apparently, conditions were rapidly becoming drier. In the stratigraphic column this change towards drier conditions is reflected in a transition of the previously deposited pure lake clay into dark peaty clay. The vegetation had a markedly open character. The climate must have been cold and changing from a wet to a dry condition. Subzone G 8 b is characterized by a maximum in the curve of Compositae tubuliflorae and a declining curve of Gramineae. The vegetation maintained its open character. The percentages of Alnus remain low as in subzone GSa, only in the t o p of subzone G8b showing a slight increase to a peak of a b o u t 30%, which is reached at the beginning of subzone G8c. Presumably the climatic conditions began to ameliorate towards the very end of subzone G8b.

54 The percentages of Botryococcus remain low. The curve of Cyperaceae shows a maximum of more than 60%. This is an indication of the desiccation of the gully. After this maximum the curve of Cyperaceae falls again towards the beginning of subzone G8c. In the stratigraphic column the dark peaty clay facies of subzone G8a is still present. The climate remained relatively cold. Only in the t o p of subzone G8b did a minor amelioration commence. The drier conditions that started in the u p p e r m o s t layer of subzone G8a, apparently persisted. Subzone G8c starts with a relatively short peak of 30% in the Alnus curve already mentioned above. This maximum follows the peak in the Cyperaceae curve from the t o p of subzone G8b onward. Possibly, this succession reflects the drying up of the gully. Immediately thereafter the Gramineae curve increases again to be followed by a peak in the curve of the Compositae tubuliflorae. Towards the top of subzone G8c the Alnus curve starts rising again. Also the curve of Cyperaceae shows another maximum (of almost 80%). Botryococcus disappears completely in the uppermost part o f subzone G8c. At the same time the curve of fungal spores increases. After the minor maxim u m in the Alnus curve at the beginning of subzone G8c (which might indicate a slight amelioration of the average annual temperature), the elements of an open vegetation dominated again and the climate was relatively cold once more. At the base of subzone G8c a sandy clay occurs in which a certain kind o f yellow " m o t t l e s " , consisting of material of volcanic origin, is discernible. Signs of pedogenetic activity can be observed. Summarizing, the following conclusion can be drawn. At first the climate rapidly became colder and drier in the course of zone G8. At the base of subzone G8c there is a short fluctuation, during which " w a r m e r " elements played a role, so that the climate must temporarily have improved somewhat. Immediately thereafter the climate turned cold again, although in all likelih o o d it was n o t so cold as it was during the initial part of zone G8 or during the main part of the subzones G8a and G8b. Finally a definite improvement of the climate started towards the end of zone G8. The transition from zone G8 to zone G9 is characterized by the as y e t final and definitive rise of the Alnus curve, a steep fall of the curve of Compositae tubuliflorae, and the beginning of a low representation of Gramineae.

Zone G9. During zone G9, forest elements gradually started to dominate: the percentages o f Alnus increase to over 50%. Moreover, other forest elements become better represented, e.g., Miconia, Myrtaceae and Borreria (a herb whose presence is often associated with an increase of the Andean forest). The curve of the Cyperaceae decreases. Fern spores are no longer well represented. Only fungal spores are still numerous and represented by a maxim u m of a b o u t 60%. The climate definitively improved, which situation persisted t h r o u g h o u t this zone. In the stratigraphical column the " y e l l o w mottles" disappear. Pedogenetic processes seem to have been active.

55

Zone GIO. It is n o t immediately clear from the diagram why a separate zone, viz., the zone G10, can be distinguished, but the evidence of the diagram of section II-B3 from a somewhat more elevated site on the slope of the same gully renders this further differentiation possible. In section II-B1 the only criterion for a further subdivision is the point where the curve of Gramineae begins to rise again. There is no appreciable change in any other curve possible or in the stratigraphy either. The only possible conclusion regarding the climate is that conditions remained relatively warm. Section II-B3 Section II-B3 is situated somewhat higher on the slope of the gully towards rock shelter II. Zone G 7. The base of section II-B3 is situated in the uppermost part of zone G7 with, still, a high representation of forest elements. On the other hand the percentages o f Gramineae and Compositae tubuliflorae are increasing. The landscape assumed a more open character, and the climate became relatively colder. The percentages of Botryococcus remain high although they show a slight decrease, Pediastrum is also represented; both taxa indicated the presence of water in the gully at that time. Spores of Monolete verrucate ferns are numerous but, in this case also, the percentages gradually decrease. Finally, the percentages of the Jamesonia-type spores decrease as in section II-B1. The curve of Geranium shows the beginning of a rise, which points to relatively cooler conditions. The transition from zone G7 to zone G8 is characterized by low Alnus percentages and by the increase of the curves of Gramineae and Compositae tubuliflorae. This transition still t o o k place in an environment in which lake clay was deposited. Apparently wet conditions prevailed during the cooling of the climate. Zone G8. On the whole, zone G8 is characterized by extremely low Alnus percentages (5%--10%). As in section II-B1 this zone is divided into three subzones: G8a, G8b and G8c. In the subzone G8a the conditions at first were still wet. The Botryococcus percentages, however, decrease rapidly, as do those of the Monolete verrucate fern spores. Conditions were becoming drier. This trend is reflected in the stratigraphical column by the transition of the lake clay into dark peaty clay. The vegetation must have had an open character according to the high Gramineae percentages (over 60%). The climate must have been relatively cold, while conditions became drier. In subzone G8b we see a m a x i m u m in the curve of Compositae tubuliflorae (around 50%) and a decrease of the Gramineae curve {to 30%). In the upper half of this interval there is a slight rise of the Alnus curve culminating in a minor peak of 10% at the base of subzone G8c. The curve of other forest elements shows the same, slight

56 increase. The presence of Cactaceae pollen points to very dry conditions. The highest elevation where nowadays Cactaceae are found growing in the Eastern Cordillera is a b o u t 3,000 m, at the transition of xerophytic vegetation to dry p~ramo (locally, on the upper slopes of Chicamocha Valley; Cleef and Van der Hammen, pets. comm.). In Pen], Cactaceae are widespread in the puna, the high-lying and arid Andean equivalent of the p~ramo, however. A peak in the curve of Cyperaceae indicates the continued desiccation of the gully, a development fully in accordance with the indications of surrounding dry conditions. The curve o f Botryococcus maintains its decline. In the diagram, subzone G8c starts with a minor maximum in the Alnus curve mentioned above, combined with a slight increase o f the other forest elements. Also some curves of herbaceous plants, which occur mainly in the upper part of the diagram (in zones G9 and G10), show some increase. They either appear for the first time or occur in higher percentages and represent mainly elements o f the Andean forest. Notwithstanding the open character of the vegetation (and the still relatively cold climate), this fluctuation points clearly to an amelioration of the climatic conditions in the lowest part of the subzone G8c. This warmer fluctuation is followed by a phase with high percentages of Gramineae (up to 50%) and a slight falling o f f of the curve of Compositae tubuliflorae. At the same time the percentages of Alnus decrease a little, b u t the values remain higher than in the t o p part of subzone G8a and also in subzone G8b before the minor maximum at the base of subzone G8c. Botryococcus has disappeared completely which indicates that there was no longer open water in the gully. The curve of Cyperaceae shows a second maximum. The maxima of the curves of Lycopodiumreticulate and Hymenophyllum, which are synchronous with the decline in the two-peaked maximum of the curve of Cyperaceae, are remarkable. Conceivably this sequence has something to do with a local succession in the marshy gully. The maxima in the curves of the Relbunium type, Papilionaceae (here mainly Vicia), and Monocotyledoneae are also noteworthy. For the first time, fungai spores appear in the diagram. The absence of fungi in an open-water facies is well known. In the stratigraphic column the peaty character o f the sediment has changed; the sediment consists of sandy clay with yellow " m o t t l e s " of volcanic origin. Pedogenetic processes probably became more important. The climate is somewhat colder after the small warmer fluctuation at the beginning of subzone G8c, b u t it remained distinctly warmer than at the beginning of subzone G8b. Throughout zone G8a a group of palynomorphs could be distinguished (types 1 through 4) partly with an Artemisia-like appearance. Unfortunately up to n o w this group could not be identified to the generic level. They seem to be s o m e h o w characteristic of this part o f the diagram. The transition from zone G8 to zone G9 is determined by a marked increase of the Alnus percentages while the curves of Gramineae and Compositae tubuliflorae show an appreciable decline.

Zone G9. In zone G9 there is a rapid rise of the percentages of elements

57 which play an important role in the Andean forest, such as Hedyosmum, Weinmannia, Symplocos type, Ilex, Dodonaea, Acalypha, Melastomataceae (here mainly Miconia), and Myrtaceae. The rapid rise of the curves of Viburnum and Borreria, belonging, respectively, to the shrubby and the herbaceous undergrowth of the Andean forest, is the most striking. Of the remainder o f the (partly herbaceous) elements which are better represented from n o w onward the following are n o t e w o r t h y : Polygalaceae, Polygonum, Compositae liguliflorae, and Ranunculaceae. It will be clear that all this indicates a notable increase of elements preferring warmer climatic conditions. The vegetation becomes denser. The peak in the curve of Hydrocotyle type seems to be associated with the local succession of stands of Cyperaceae and some ferns on the lower and more humid parts of the gully. The Jamesonia t y p e (possibly representing Eriosorus in this case) appears again. The fact that the previous higher representation fell in the last, truly warmer zone G7, is noteworthy. The percentages of Monolete verrucate fern spores also start to increase again. It seems probable that the species represented here is different from the one of zone G7 during which the curve of Monolete verrucate fern spores runs synchronously with that of Botryococcus. It is clear that the climate had become considerably warmer. This situation prevails during the rest of the diagram. In the sediments of this interval pedogenetic processes seem to have played an important role. The transition from zone G9 to zone G10 is characterized b y a slight falling off of the Alnus curve, which coincides with a distinct rise of the percentages of Gramineae and also with a small increase of the percentages of the Compositae tubuliflorae.

Zone GIO. Throughout zone GIO the ample representation of elements of a more open vegetation type is maintained, which is most probably attributable to human influence. It is there that Gramineae pollen of a size comparable to maize appears. This agrees satisfactorily with a slight increase of the curves of Plantago, of Chenopodiaceae and possibly also of Caryophyllaceae. The strong decrease of the curve of Viburnum, a typical forest element, also fits in nicely, the percentages of Borreria, a herbaceous taxon also occurring in more open vegetation types, remaining high. The sudden strong increase of fungal spores is n o t e w o r t h y . As in zone G9, pedogenetic processes seem to have been important during accumulation of the sediments. The end of the diagram may approximately correspond with the present_ situation.

Section B16 (Fig. 12) Section B16 is situated directly on the edge of the gully in front o f rock shelter II, apparently almost outside its direct influence.

Zone G3. In all probability the lowest part of section B16 represents zone G3. Forest vegetation must have played an important role, although elements of

58 open vegetation are relatively well represented. The Weinmannia curve shows two distinct peaks in the two intervals during which the total of forest elements increases. The sediment consists of peaty clay with the exception of the lowest 20 cm which consists of peat. In section II-B1 probably the same peat layer lies almost at the basis of zone G3 and at approximately the same level. The vegetation most likely represents an alder brook with Weinmannia growing on the drier sites. The presence of Azolla and Botryococcus points to spots with open water in the marshy valley bottom. The climate must have been relatively warm and humid. However, the presence of Iso~tes seems n o t to be in accordance with a relatively higher annual temperature corresponding with the climatic conditions as described above. In this respect the presence of Acaena/Polylepis pollen is also somewhat problematic. The possible explanation is that during this relatively warm period there have been local small streams or run-offs intermittently running down the slope of the E1 Abra corridor causing re-sedimentation of " s e c o n d a r y " sporomorphs. In section B19, as we shall see later on, and in other nearby sections, there are cogent, relevant indications (Van der Hammen, 1978). The relatively ample representation of fungal spores in this part of the section is remarkable. The transition from zone G3 to zone G4 is determined, as in the diagram o f sections II-B1, by a sudden rise of the pollen percentages of elements belonging to the open vegetation.

Zone G4 consists of a series of fluctuations representing alternating, relatively colder and warmer periods, corresponding with subzones G4a, b, c and d, respectively. It is in the last phase during which Weinmannia, although only temporarily, forms an important element in the local stand of vegetation. The same situation is obtained in the diagram of section II-B1, where zone G4 terminated with subzone G4c after which we observed a probable hiatus. If the stratigraphy and the diagrams of both sections are compared, the conclusion can be drawn that this is the same hiatus as the one observed at the end o f subzone G4d (see also Fig.14). Subzones G4a and G4c are characterized by a maximum in the curve of Hypericum-type pollen, whereas the curve of Ericaceae only has a maximum in subzone G4c. It is n o t e w o r t h y that the curves of spores of the Jamesonia t y p e and Monolete verrucate fern spores follow the same course as the curves o f the Hypericum type and the Ericaceae, respectively. The minimum between the two maxima in the Hypericum type curve is filled in by a maxim u m in the Alnus curve falling in subzone G4b. Subzone G4d following after subzone G4c is characterized by a relatively high m a x i m u m in the Weinmannia pollen curve. At the beginning of zone G4 there is in the stratigraphical column a transition of peaty clay to a somewhat humic-lake c l a y . A t the same time fungal spores disappear, and Botryococcus reappears. This indicates that conditions were becoming wetter. There must have been some shallow spots with open water at the location of section B16, Possibly an alder carr was growing in situ during the relatively warmer subzone G4b. In subzone G4c

59 this carr was replaced by open vegetation of the s u b - p ~ a m o type as in subzone G4a. In the stratigraphical column a change occurs at the beginning of subzone G4d. The clay becomes more sandy and pedogenetic processes begin to play a more important role. There must have been a Weinmannia wood at that time. Zone G4 is terminated by a hiatus as stated before. Zone G5 is characterized by relatively high percentages of Alnus pollen. Weinmannia pollen has almost completely disappeared. The so-called layer with " y e l l o w m o t t l e s " of volcanic origin is present in the stratigraphical column at the base of zone G5; the same layer has also been noted in the soil sections B9 and B l l and in some of the sections from the rock shelters. The relatively ample representation of elements of open vegetation just below this characteristic layer is noteworthy. The same happened in the two soil sections. Zone G5 starts with relatively cool conditions during which elements of open vegetation types are well represented. Thereafter the conditions became somewhat warmer, which resulted in an increase of humic components in the sediment. At that time an alder forest must have been growing in situ. As in the diagram of section II-B1, a manifest change is reflected in this part of the diagram. This must have happened at some time between the zone G4d (not represented in II-B1) and the zone G5, presumably during a very cold phase. Some fern spores become very abundant. Fungal spores disappear for the time being. The decreasing variety of the tree pollen and at the same time its decrease in numbers is very interesting. Only in the upper part of the diagram tree pollen are again relatively better represented. At the transition of zone G5 to zone G6 the climate clearly became very cold, for which the increase in representation of elements of open vegetation to over 90% provides adequate evidence. Zone G6 is characterized by high percentages of Gramineae and a good representation of Compositae tubuliflorae pollen on the one hand, and by very low values of the Alnus percentages on the other. All other curves of forest elements including herbs show minima or are even completely absent. In the upper half of zone G6 there is an increase of the Botryococcus percentages, the curve of Monolete verrucate fern spores attaining a m a x i m u m at the same time. The stratigraphical column shows relatively loose, loamy sand. The climate was comparatively cold, and in the first half of the interval G6 also dry, whereas in the second half the h u m i d i t y increased. The transition from zone G6 to zone G7 is n o t very distinct. Zone G7. In zone G7 the percentages of Alnus and those of some herbs are beginning to increase again. The curve of Cyperaceae pollen also shows a rise. The rise o f the curve of Botryococcus already started in G6, continues. Conditions continued to become relatively wetter in comparison with zone G6. The climate was slightly warmer and more humid. In the stratigraphy pedogenetic processes seem to have become quite apparent. The transition from

6O zone G7 to zone G8 is characterized by a decrease of the Alnus percentages and a strong increase of the Gramineae percentages.

Zone G8. During zone G8 the Alnus curve has a distinct minimum, while the curve of Gramineae shows a maximum. The curve o f Cyperaceae rises towards a m a x i m u m of over 150%. The Botryococcus curve also shows a maximum. Apparently conditions were relatively wet. Judging by the high percentages o f Cyperaceae pollen, it is n o t very likely that there were extensive stretches of open water at the site of section B16. Also in the stratigraphy there is nothing indicating open water. However, the m a x i m u m in the curve of Botryococcus at least seems to indicate some open water, possibly in the centre of the gully in front of rock shelter II. When the values of the Botryococcus percentages are compared with those of other diagrams, it becomes clear that much higher percentages would have been recorded if there would have been open water at the boring site. In the section "Movement of vegetation belts" this point will be discussed more extensively. The climate was relatively cold and wet. Zone G8 is subdivided into the subzones a, b and c, as in the diagrams of the sections II-B1 and II-B3. This subdivision does not seem relevant in the diagram of this section. In the stratigraphy there is more reason for this subdivision, however. From the beginning of subzone G8b onward the same kind of yellowish mottles as those f o u n d in the sections II-B1 and II-B3 occurs. Zone G9 begins at the point where the Alnus percentages begin to rise to a m a x i m u m o f about 70%, the Gramineae curve showing a minimum at the same time. Various herbaceous pollen curves are on the increase again. The most striking rise is that of Borreria pollen, a herb that in the pollen diagram is associated with the Andean forest; the percentages increase suddenly from almost nil to about 8%. In the stratigraphy there is no change in comparison to subzones G8b and G8c. The climate was apparently warmer and drier than during the preceding last warmer zone G7. At the transition from zone G9 to zone G10 the curves of Gramineae and Compositae tubuliflorae rise again. At the same time the Alnus curve falls to about 20%. Zone GIO. After a slight depression the Alnus curve maintains values of 20--25%. Curves of tree pollen and of herbs from the Andean forest are as well represented as they are in zone G9. After a slight increase in the beginning of zone G10, the elements of open vegetation maintain percentages o f about 75. Possibly this is mainly due to h u m a n influence. The rise of the curve of fungal spores is noteworthy. In the stratigraphy the " y e l l o w m o t t l e s " have disappeared and the influence of pedogenetic processes in sandy humic material is noticeable.

EL

ABRA

(SABANA

pp. 61--66.

II - B 1 6 DE B O G O T a , , C O L O M B I A , SOUTH AMERICA )

J >-Lu

ALT

ca,

2570

m

~_j C3 zw ~~O< Na

ANAL,

E.J. Schreve

ABOVE - SOBRE

-

-

SEALEVEL EL NIVEL

Brin~,m~n

U~(9 w

DEL


MAR

o

~ < ~w~d z~z

,,"I, o.

3

<-uE 0=¢

-~ ~

~:

:E

,.

,.,.< _~ =--~

~

w,~>-<

-~o.

~

,,. >~-

¢nLuO

--

o

U

w

~o:

Jw

o

OO

0~-.o Oz

==

o= c,:E

-

~,=.,

OW~-

1_1

r

1_2 2-1 2-2 3-1

m

3_2

r

4_1 4.2 5_1 5-2 p~

i r

T

-6_1 0_2 7_1 7_2 8_1 8_2 -9_1

i i

9_2 10_1 10_2

II

~.mm

m

11

_I

11 _ 2

d i m

r

12_1 t2_2

r

13 _ 1 13_2 1.4_ 1 14_2 15_1

.;!

m _

15_2 1 6 _

1

16_2 m _

17_1

m

17-2

-18_I

18_2

m

-19-1 -19_2 -20_ 1 m

INCLUDED IN T H E P O L L E N S U M INCLUIDOS EN LA 5UMA DE POLEN TREES ICOMPOStTAE * 'GRAMI N E A E ARBOLES, ,TUBULIF'LORAE * i ] HYPERICUM TYPE + ERICACEAE

0

10 2 0

30

40

50 60

70

80

9 0 100°/o

SCALE FOR A L L ( P O L L E N ) CURVES U N L E S S OTHERWISE IN D ICA~'E D ESCALA PARA LAS CURVAS ( DE P O L E N ) SALVO A L G U N O S CASOS I N D f C A D O S 0 • x

ALNUS p o DOCAR PUS OTHER FOREST E L E M E N T S - O T R O S ELEMENTOS DE B O S Q U E

Fig. 12. Pollen diagram of section B16. 'For legend, see Fig. 2.

-20_2 NOT I N C L U D E D ]N T H E P O L L E N S U M N O INCLUIDOS EN LA SUMA DE POLEN

EL

ABRA

-

B19

(SABANA DE BOGOTA, COLOMBIA, SOUTH AMERICA)

pp. 67--7 2. o

u~

u} ALT

OD

ca.

2570m

_

ABOVE

SEALEVEL

SOE~RE

EL

NIVEL

DEL

MAR

~o~

z ~-

ANAL.

E.J.

Schreve

w

in

~0

30

8

<~ zO

- Brinkmon

OU O~

n

_~L~ dW<

4~

5o

Bn 7n

80

~

-J:E~ 0 ~ o~. 0 . ~ 0

w

U < (OJ

<0 in< ~£~_ Ob~>-

×o

~F 0"< ~E:E

:E < )- m bJ :E ~ ~-:s:-u ~ Z

D E > - ~ < <1~

--

~:EO

E<

o=o

ww

.0 w

Oz~ ~

JJ O0

8

w

EoO , ~ZE ~{: w ~<--

~=

o=~

W<

zz oo

o~ z z OD

0 ,n

> d~U 3 v) LO>

cm

mo~O~ oo) •, col. ,

o

198

~d 300-~

i 350~-V

400-I--

I 450-1--

500 I

,

I

590

o

o °

4-60

OJ~

P

0 °

-

o o o

G2

65--01 -_

700

--"

I ooI

oolI

ITREES

~.-.-.-.-.-..-, u~F-;COM POSITAE '~GRAMI N E A E ARBOLES ~II"U~ULIFLORAE ° ~ ~HYPERICUM TYPE +~ ~,ERiCACEAE ',

~NCLUIDOS EN L A S U M A DE P O L E N 0 10 2 0 3 0 4 0 5 0 6 0 7 0 8 0 9 0 lO0=&, S C A L E F O R A L L ( P O L L E N ) CURVES UNLESS O T H E R W I S E I N D I C A T E b E S C A L A PARA L A 5 CURVAS ( D E P O L E N ) SALVO A L G U N O 5 C A S O S I N D I C A D O S 0 ALNUS • PODOCARPUS x OTHER FOREST E L E M E N T S - O T R O S

Fig.13. Pollen diagram of section B19. 'For legend, see Fig.2.

E L E M E N T O S OE B O S Q U E

...........

£D~NITHE POLLENSUM ' ' h S EN L A S U M A DE POLEM

' ' '

' '

0 ~n

73 Section B19 (Fig. 13) Section B19 is situated near rock shelter IV, at the rim of the valley b o t t o m where a sort o f talus comes down from the slope of the rocks bordering the E1 Abra corridor. To understand the zonation of section B19, the stratigraphy has to be taken into account. The base of section B19 consists of a coarsegrained sand becoming finer at the top and gradually passing into a fine sandy clay and ultimately into fine clay. This unit is overlying a peaty clay deposit still present in the lower-most part of the section. From a study of the total stratigraphy (Van der Hammen, 1978) it appears that this coarse sediment was deposited from a dry valley cut in the western rock wall of the corridor downwards during a cold phase, which in section II-B1 is called zone G2, and of which the base of this section B19 also forms a part. Zone G2. In the above-mentioned way the two isolated pollen spectra at the base o f the diagram in zone G2, as well as the lower part of the continuous diagram corresponding with the coarse sediment, can be assigned a place in the stratigraphic sequence. The origin of the coarse-grained layer renders the deposition of secondary pollen rather likely, so that the pollen diagram need not reflect the pollen rain accurately. Elements of an open vegetation type may be expected. Acaena/Polylepis is represented. According to Van Geel and Van der Hammen (1973), Polylepis was growing on the mountain slopes bordering the Sabana de Bogot~ during periods with a relatively colder climate. It is rather hazardous to give a characteristic of zone G2, b u t the climate must at any rate have been relatively cold. Zone G3. The transition from zone G2 to zone G3 is mainly marked by the sharp limit between the coarse-grained deposit and the overlying clay deposit. However, at the same time the Weinmannia pollen curve starts rising to attain relatively high values at the u p p e r m o s t part of the clay unit o f zone G3. Since the gritty sediment fans o u t in front of a dry gully cut in the rock wall of the E1 Abra corridor (a little to the south of rock shelter IV) and rises above the flat b o t t o m of the E1 Abra valley, during the relatively warm phase o f zone G3 (see section II-B1) the sedimentation remained mainly restricted to the lower-lying valley bottom. Only in gullies cut in the fan could any deposition of importance take place, which seems to be the case at the location of section B19. The relatively high percentages of elements representing an open vegetation t y p e are most likely of secondary origin. The presence of Weinmannia pollen seems to be indicative of relatively warm conditions. Zone G4. The next transition, from zone G3 to zone G4, is also marked by a sharp limit in the sedimentary sequence. In all probability there is a hiatus between the humic clay of zone G3 and the overlying peat deposit of zone G4 (see also Van der Hammen, 1978). A comparison of the high percentages o f Compositae tubuliflorae in the lower part of zone G4 with the zone G4 of section II-B1 suggests that in section B19, zone G4 starts with subzone

74 G4c after a hiatus. Although a vegetation type belonging to the s u b p ~ a m o zone is very well represented, Weinmannia pollen is also relatively abundant (about 30%). From the stratigraphical column we note the presence of a peat deposit at the location of section B19. There must have been an open vegetation with Weinmannia still growing on higher and drier places, so that, accordingly, the climate cannot have been very cold, but most probably corresponded with that favouring a vegetation found at the transition of the subp~ramo to the Andean forest. The conditions must have been relatively humid. The transition from subzone G4c to subzone G4d is marked by a sharp fall of the respresentatives of an open vegetation and a distinct rise of the Alnus curve. We see the same in the diagram of section B16. In section II-B1 subzone G4d is already cut off by the erosion activities of running water in the gully in front o f rock shelter II. In section B16 this happened after the deposition of subzone G4d. Subzone G4d is characterized by extremely low values of the percentages of taxa representing an open vegetation type, and relatively high values of the percentages of Alnus and Weinmannia. The situation is the same as in B16, although in B16 Weinmannia is better represented. In the stratigraphic column the peat deposition of subzone G4c continues in subzone G4d. Locally, the conditions must have been relatively humid and alder must have grown on the somewhat higher and drier spots in the marsh. At the location of section B16, the conditions must have been less humid, so that Weinmannia could grow there. This may explain the minor representation of Weinmannia in the diagram of section B19. The climate must have been relatively humid and warm. The transition from subzone G4d to subzone G4e is marked by a steep fall of the tree pollen and a rapid rise of the curves of pollen types belonging to more open vegetation types. In section B16 the succession was interrupted in subzone G4d by erosion activities in the glflly in front of rock shelter II, so that among the diagrams of the present study only B19 contains a subzone G4e. During the subzone G4e the curve of Gramineae rises to a value of about 50%, and the percentages o f Alnus pollen are relatively low. In the stratigraphy there is a transition from peat to humic clay. The conditions must have been relatively cold, and the formation of humic clay suggests that locally the soil conditions must have been humid. A comparison of the aliagrams of the sections II-B1, B16 and B19 with those of subzone G4e of B19 indicates that the latter was presumably as cold as zone G2 represented in all three sections. The ~*C-dating of the layer corresponding to subzone G4e of section B19 is of considerable interest because it is the oldest limited dating in the valley. It appears that section B19 plays an important role in the correlation of the soft and rock-shelter sections with the oldest part of the deeper sections. An as yet still unpublished diagram of section B22 situated somewhere between section II'B1 and B19 (based on counts b y G. Noldus) has been of some help in establishing the chronostratigraphy, as we will describe in detail in the section " R e c o n s t r u c t i o n and description of the composite pollen diagram".

75 RADIOCARBON DATINGS OF THE EL ABRA AREA (by E. J. Schreve-Brinkman and

T. van der Hammen) Below a list is given of all the known 14C-datings from the E1 Abra valley. The samples were taken in the archaeological excavation pits in the rock shelters in 1967 and 1969, and in boreholes during the survey of the valley in 1969. Fourteen of the samples were dated in Groningen, five in Bern and two by Isotopes Inc. We want to thank especially Dr. W. G. Mook (Groningen) and Dr. J. C. Lerrlan for their interest and assistance with the analyses. The datings are listed in order of increasing age. Only two of the nineteen datings are from material that was possibly slightly contaminated with younger material (penetrated rootlets, or unobserved intrusions caused by displacements by small rodents). A careful study of the relative position in respect of other datings in the same series that could clearly be correlated, revealed that the dating of Col.86 has to be considered as a minimum date. The same probably holds true for Col.130. All other datings are fully consistent, w i t h o u t contradictions in their local sequence or stratigraphy. The 14C-datings marked with an asterisk have been used for correlations in this paper. For an interpretation of these datings the reader is referred to the section " R e c o n s t r u c t i o n and description of the composite pollen diagram". All datings are years before present. B-2136

Co1.133

340 + 260

E1 Abra III, 1969 excavation. Lumps of charcoal; depth 55 cm; dates pottery. 1-6362

Co1.129

495 + 104

E1 Abra III, 1969 excavation. Charcoal, from the fill of a child burial (N2), f o u n d about 45 cm below the surface. B-2135

Co1.140

7,250 + 100

E1 Abra IV, 1969 excavation. Charcoal, in soil sealed off by a very large flat stone fallen from the ceiling of the rock shelter at a later date. Position 4 cm above top of "loessic" layer. Dates fire place. B-2137

Co1.135

8,760 + 350

E1 Abra III, 1969 excavation. Charcoal in soil; depth 100--118.5 cm, layer sloping from north to south; dates a bone layer. B-2133

Col.130

8,810 -+ 430

E1 Abra III, 1969 excavation. Charcoal in soil; depth 190--191 cm; located near top level of "loessic" layer; dates level with heavy flake and a scraper of quartsitic sandstone. Date probably too young. GrN-5710

Col.82

9,025 + 90

E1 Abra II, test excavation 1967. Charcoal in soil; depth 75--100 cm (see

76 Correal, et al., 1970). Related to GrN-5746, Col.84; GrN-5561, Col.83; GrN-5557, Col.86; GrN-5556, Col.85; see below. 1-6363

Co1.142

9,050 + 470

E1 Abra IV, 1969 excavation; field no.239. Charcoal in soil, below large rock, from feature 9, a fired area (hearth). Very slightly above top of "loessic" loam level. Locality in grit: 106 N-13E; depth 93.75 cm. GrN-5746

Col.84

9,325 + 100

E1 Abra II, test excavation 1967. Charcoal in soil; depth 125--155 cm. GrN-5561

Col.83

9,340 + 90

E1 Abra II, test excavation 1967. Charcoal in dark soil, in which thirty stone artifacts were found; depth 100--125 cm (see Vogel and Lerman, 1969; Correal et al., 1970). GrN-5557

Col.86

9,420 + 110

El Abra II, test excavation 1967. Charcoal in dark soil in which five stone artifacts were found; depth 150--175 cm; as the fine charcoal was mixed with very fine rootlets, the older date of Col.85 from a somewhat higher level seems more reliable, see below. (See Vogel and Lerman, 1969; Correal et al., 1970). B-2134

Co1.120

10,720 + 400

E1 Abra II, 1969 excavation. Charcoal in soil; depth 148 cm; dates rock fall level with artifacts. *GrN-5941

Co1.123

11,210 + 90

El Abra II-B3, 1969 excavation. Black humic clay; depth 185--190 cm; dates top of wet phase during which sedimentation of lake clay t o o k place, representing a warm fluctuation in the pollen diagram (see pollen diagram E1 Abra II-B3, F i g s . l l and 14). GrN-5556

Col.85

12,400 -+ 160

E1 Abra II, test excavation 1967. Charcoal in dark soil in which nine stone artifacts were found; depth 1 5 0 - 1 7 5 cm. (See Vogel and Lerman, 1969; Correal et al., 1970). *GrN-6281

Co1.146

21,050 + 210

El Abra IV-107N profile, 1969 excavation. Black sandy clay; depth 208--213 cm; dates top of erosion level on which the "loessic" layer was deposited (see pollen diagram E1 Abra IV-107N, Figs.7 and 14). *GrN-6267

Co1.127

22,200 + 335

E1 Abra II-B1, 1969 excavation. Lake-clay from borehole in secondary gully

77

in front of rock shelter II. Depth 205--222 cm (see pollen diagram E1 Abra II-B1, Figs.10 and 14). *GrN-6279

Co1.144

23,850 + 280

E1 Abra IV-107N profile, 1969 excavation. Black sandy clay; depth 179--185 cm, dates top of erosion level on which the "loessic" layer was deposited (see pollen diagram E1 Abra IV-107N, Figs.7 and 14). *GrN-6280

Co1.145

23,870 -+ 185

E1 Abra IV-107N profile, 1969 excavation. Black sandy clay; depth 223--232 cm; dates top of erosion level on which the "loessic" layer was deposited (see pollen diagram E1 Abra IV-107N, Figs.7 and 14). *GrN-6268

Co1.128

25,195 -+ 440

E1 Abra II-B1, 1969 excavation. Lake-clay from borehole in secondary gully in front of rock shelter II; depth 2 2 9 - 2 4 2 cm. Probably mixture of older and younger material caused by erosion in gully (see pollen diagram E1 Abra II-B1, Figs.10 and 14). *GrN-6269

Co1.137

28,140 + 440

E1 Abra III-10E profile, 1969 excavation. Dark sandy loam; depth 325--330 cm; dates top layer of this deposit (see pollen diagram E1 Abra III-10E, Figs.5 and 14). *GrN-5942

Col. 124

> 50,000

E1 Abra II-B1, 1969 excavation. Peaty clay with some fine charcoal from borehole in secondary gully in front of rock shelter II; depth 325--350 cm (see pollen diagram E1 Abra II-B1, Figs.10 and 14). *GrN-6548

Co1.198

50,720 + 4,100 - 2,700

E1 Abra B19, 1969 excavation. Dark humic clay together with some peat from borehole in valley slope betweeen the rock shelters E1 Abra III and E1 Abra IV. Depth 260--283 cm (see pollen diagram E1 Abra B19, Figs.13 and 14). RECONSTRUCTION AND DESCRIPTION OF THE COMPOSITE POLLEN DIAGRAM (Figs.14 and 15)

Correlation and general zonation With the aid of the descriptions of the pollen diagrams and the stratigraphy given above, an attempt will be made to reconstruct the vegetational history. In Fig.15 a composite diagram is shown which is based on the diagrams of the various sections. From the ~4C-datings it will be clear that the sections form a part of the Last Glacial (i.e., they are at least partly synchronous with the European Weichselian) and the following Holocene.

78 The zones indicated in R o m a n figures used in the following description are the local E1 Abra zones. These are local pollen zones, based on the changes in vegetation (mostly induced by changes of climate). They are indicated on Figs. 14 and 15. Fig. 14 shows the correlation of the local E1 Abra zones]with the provisional zone systems used in the previous section. The lower part of the composite diagram consisting of zone I and II is reconstructed from the diagrams of the sections reaching that deeper level: II-B1, B16 and B19. In general they are mutually consistent, although section B19 takes a somewhat special place. Two 14C-datings of this lower part are available: one of section II-B1 and another of section B19, of which the latter is finite. The middle part of the diagram, consisting of the zones III and IV, is mainly constructed from the diagrams of the two soil sections B9 and B l l . They provide the most detailed information. 14C-datings are available from the diagrams of the sections IV-107N, III-10E and II-B1. The dating of the latter section should be interpreted with caution, because of the presence of a hiatus. The upper part of the composite diagram consisting of zones V, VI, VII and VIII is reconstructed mainly with the aid of the diagrams of the sections II-B1 and II-B3, of which the latter gives the most detailed information. Since the diagrams of the sections II-10E and II-100N have added nothing to the reconstruction of the composite and continuous diagram, they are left o u t of consideration here. Only their position in the sequence is mentioned; their position in the stratigraphy can be found on Fig.14. The diagrams of the t w o sections in the gully in front of rock shelter II and other diagrams in which there are indications of the presence of open water, provided data concerning probable water level fluctuations during wet periods in which the gully actually functioned as a stream or f o r m e d a small lake. The results are given in a graph on Fig. 15. With the available 14C-datings an a t t e m p t was made to match the composite diagram with a linear time scale, Fig.15, in order to compare the reconstructed zonation with that of Laguna de Fuquene (Van Geel and Van der Hammen, 1973), the Andean pollen zones (Van der Hammen and Gonzalez, 1960) and, finally, the European time scale (Van der H a m m e n et al., 1971), also given on Fig.15. For the mutual correlation of the various diagrams a digest was made (Fig.14) of all the principal pollen diagrams, together with their stratigraphy and the occurrence or absence of Botryococcus, the most important indicator of open water (see the sub-section "The construction o f a diagram indicating the maximal fluctuations of the water level in the valley"). For the digests of the pollen diagrams only the curves with the most significant changes were used: Weinmannia, Myrica, Alnus, elements abundant in the subp~ramo (Compositae tubuliflorae + Hypericum t y p e + Ericaceae) and Gramineae. The abstracted diagrams are cumulative diagrams. In this way a reasonable survey of the most significant pollen data was achieved.

79 The most important factors used for correlation are listed below: zones I and I I - - the curve of the fungal spores; the relatively high percentages of Weinmannia (in comparison with the remainder of the diagram); -- the humid to relatively wet local "soil" conditions, reflected by sediments consisting mainly of peaty clay or peat. zone III -- the high peaks in the Myrica and Alnus curve; -- high Botryococcus percentages; -- a discontinuity in the curve of fungal spores; -- the extremely dark humic layer with "yellow m o t t l e s " in the uppermost part, and the presence of these "yellow mottles" in the various sections as such. zone IV -- the presence of the "loessic" layer in the rock shelter and soil sections and in section B16, and its absence in section II-B1 due to a hiatus; -- extremely high percentages of grasses and Compositae tubuliflorae. zones V, VI, -- the high Botryococcus peak in the diagram of the soil VII and VIII sections above the "loessic" layer, the last high Botryococcus peak in the diagram of II-B1, and the one in that of II-B3; -- the congruency of the diagrams of section II-B3 and the top of section II-B1 from the last maximum in the Botryococcus curve onward; -- the correspondence of the uppermost parts of the diagrams. In addition to the composite diagram on Fig.15 curves of water indicators are indicated and next to it the estimated relative water depth fluctuations, the presence or absence of material of volcanic origin, the vertical movements of vegetation belts as abstracted from the pollen diagrams, and, finally, partly very tentatively, the correlation with various other pollen zones and chronostratigraphies.

Zone 1 (sections II-B1, B16, B19). Zone I is subdivided into subzones Ib, Ic, Id (subzone Ia was " d r o p p e d " in the course of the search for correlations). Subzone Ib is only represented in section II-B1 in the lowermost 4 to 5 spectra. Pollen of Weinmannia and Alnus is very well represented. Climatic conditions must have been relatively warm and comparable with the presentday situation. "Soil" conditions must have been wet because peat and peaty clay was formed. There was no open water in the valley. We propose to call this warm interval the Rocas de Sevilla I Interval (Rocas I or RI, after the name o f the rock walls along the E1 Abra corridor}. Subzone Ib m a y correspond with the last warm phase of the penultimate glacial (the Last Interglacial) or with an early warm phase of the Last Glacial. In subzone Ic (II-B1, B19), elements representing an open vegetation are relatively well represented, especially the grasses. In the diagram of section

8O II-B1 there is still a good representation of Weinmannia and Alnus pollen, the latter also being well represented in the diagram of section B19. This means that the E1 Abra corridor still belonged to the Andean forest zone, presumably near its upper limit judging by the ample representation of grasses. The climatic conditions must have been relatively cooler than in the preceding subzone Ib, but n o t very cold. "Soil" conditions must have been relatively wet, judging by the formation of a peaty clay deposit. There was no open water in the valley. The grit sequence of section B19 was probably deposited under relatively cold conditions. Its presence in other boreholes suggests that it formed a sort of fan-like deposit in front of a dry valley cut o u t in the rock wall of the E1 Abra corridor near rock shelter IV (all other stratigraphical information than that deduced from the pollen sections is obtained from Van der Hammen, 1978). Higher percentages of tree pollen are probably attributable to redeposition. Subzone Ic may correspond with the last cold phase of the penultimate glacial or to an early cold phase of the Last Glacial. In subzone Id (II-B1, B16, B19), pollen of Weinmannia and of Alnus is again well represented. In the diagram of section II-B1 grasses and subp~ramo elements only reach values lying between 10% and 30%. In the diagram of section B16 they are better represented (20--50%) and in the diagram of section B19 they are still more numerous. In the valley, peaty clay and peat were deposited. On the fan-like deposition of grit (B19) some minor depressions were present in which a fine, more softy material was deposited, It is, therefore, very likely that at the site of B19 pollen was secondarily deposited, which resulted in the high representation of pollen o f elements belonging to an open vegetation during a relatively warm phase. At the location of section B16 some redeposition of pollen material possibly also t o o k place (viz. of Acaena/Polylepis and of elements belonging to an open vegetation). The climatic conditions during the interval Id are comparable with the present-day situation. There was no open water in the valley, only small or occasional streams causing some resedimentation. The relatively warm phase of subzone Id m a y correspond with the Last Interglacial or with an early warm interstadial of the Last Glacial. We propose to call this warm interval Rocas de Sevilla II Interval (RII). The transition from zone I to zone II is marked by a distinct percentual rise of pollen representing the subp~ramo vegetation belt (here mainly of the Hypericum type). ......

Zone H (II-B1, B16, B19, IV-I O7N). Zone II consists of some relatively warmer and colder fluctuations, probably corresponding with the Early Glacial. During the colder phases, elements of the subp~ramo vegetation belt dominated, and during the warmer phases Weinmannia together with Alnus were the principal contributors to the tree pollen present in the sections. In the valley a sequence is deposited of peat, peaty clay and clay or sandy clay showing pedogenetic influences. The entire valley b o t t o m was covered by it

]]_10 E

]]]. IOE

4#~ r. 107 N 12

EL ABRA (Sabana de Bqgota) alt. ?_570rn above sea level

]~ _ 1 0 0 N

sobre el nivel del mar pp. 8 1 - - 8 3

~-

o

B 16

E_B1

~zm ~ i ~ ~

G~ramimeae>46%

~

$6 55

#~iiiiiiiiiiii: i

s6

i:iiiiiiiii~iiii!~ :i:;ii?i;i~

fig 9

hg 4

fig 3

~ •

IOO%

0

100%

,

.

IG7":i

:iiiiii!iiiiiiiiiii fig. 5

o

~N

1~o%

--r9o~

Weinmannia

,,;

Myrica ~ ] Alnus ~ ] other elementsof ~:AP Sub pAramo (Comp.tuD., Nypericum t., E~caceae ) I ~ Gramineae

s4

0

100% hg

. . . .

ti9. 7

B 19

B9

Bll

II i

,,

"., .,"

i

i i

% f~g13

~oo

b

Botryo¢occus

600

hq ~2

Fig. 14. S u m m a r y o f several pollen d i a g r a m s a n d their s t r a t i g r a p h y : c o r r e l a t i o n of the " s e p a r a t e z o n e s y s t e m s " with t h e local Et A b m (pollen) zones. F o r l e g e n d for s t r a t i g r a p h i c a l c o l u m n s , see Fig. 2. F o r l o c a t i o n o f sections, see Fig. 1. !

G3

f,g.10

o

, ,,&~

g

~,

~,

147data~-

~gn

ITEI:ITI~: .::" . ' - " . .

C

~LI<[

,&
.

.

.

.

-

7 1}1117;T~

~ •~

"~ ~

~ ~

~ ~

~ ~

~3

~ ~

~

~

, ~

~

, ~

~

m

'

m

'

~-

'

~

'

~

~

x 1000

yearsBR

< < < - ~

< < < ~ kltlill14iH141,4<~l~ h ' t ¢ ~ ] ~ L

~

I~I
"

P,:

~. = .~ ~..

"

. . . . . . . .

:

"

'

'

'

~"~

~

~o

>

"j/L':

N C~

4---*

0

+

"4

+

+

0-.

Y ~:g? ~'~

~

P ~

Iqca,¸ m a r ~ 5;-

,'

0

--

~ ~ a ~ ~ :q~, ~ X o>en w~ter in m L- ~*meguilyand

Material oL v¢ican~ o-.~ln Ma~ de origen

m a t e r i a l of v o T ~ a n i c o r i g i n • .

_

_

open ~Fass p~rarno

bC

4000

m.

s~b p~ram~ ]~4

; E'o=?

%~~

e+

~# .

.

.

.

.

.

.

~

.....

.......

sub~ndean "orest

5e oo J

n

G~--

~R~ m

J

E_Z :L_ (3_

[ Penultimate

Glacial

to

Last

interglacial

to Early

AS

G

T

L

A

C

---i A

- o i

]~mT

-

Very

-

~0.

I

I

ill"

--0.,

l--

I

~, I

I

-<

i -<

N

~N

N

L0cal zones Fuquene ~ o o

[HOLOCENE

P[enigtacl I ~ q ~ ~__ [(~ ~

,~ I

--

~.~

L

g l l [ Ve l l d dr le L o TP l e•n l w

Glacial

~acl mz31

-

:~ ~i

-

~

'1 ~

,

~

~

~ ~

I~

'~

-

~

~

J.

tentative

87 and later also the higher parts, namely the "fan". The peat of section B19 is deposited in a gully in the fan somewhat later but still in zone II. The diagram of section B22, prepared by G. Noldus and already mentioned earlier in this study, helped to establish the stratigraphy in this part of the diagram. Zone II is subdivided into five subzones: a, b, c, d, and e. Section II-B1 seems to be continuous at the transition of zone I to zone II. Subzone IIa (II-B1, B16) is a relatively cold phase during which the E1 Abra corridor belonged to the s u b p ~ a m o vegetation belt. Hypericum-type pollen is remarkably well represented. The Weinmannia and Alnus pollen percentages remain below 10. Jamesonia, which is associated with p~ramo and higher subp~ramo vegetation types is amply represented. In the stratigraphy of both the sections II-B1 and B16 in which subzone IIa is represented, the deposition of peaty clay continued, but apparently it became subjected to pedogenetic influences. Charcoal is found. Climatic conditions must have been relatively cold and n o t t o o wet. The transition to subzone IIb is marked by an increase of the Alnus pollen percentages and a falling off of the Hypericum-type curve. Subzone Hb {II-B1, B16) reflects a relatively warmer phase with a dominance of Alnus pollen. The Weinmannia percentages remain relatively low. In section B16, the stratigraphical column shows a sedimentation of clay subjected to pedogenetic influences, and at II-B1 a continuation of the peatyclay deposition of subzone IIa. The climatic conditions must have been relatively warm, although considerably colder than they were in the last preceding warm phase of subzone Id, and the soil relatively humid. We propose to call this warmer phase the Rocas de Sevilla IIIa Interval {RIIIa). The transition to subzone IIc is marked by a strong decline of the Alnus curve and another sharp increase of the Hypericum-type pollen percentages. Subzone IIc (II-B1, B16, B19) represents a relatively colder phase indicated by the relatively ample representation of Hypericum-type pollen and of Jamesonia spores. In section II-B1 the maximum in the Hypericum-type curve is followed by a maximum in the curve of Compositae tubuliflorae separated from each other by a minor peak in the Alnus curve. In section B16 this Compositae tubuliflorae maximum is n o t represented, whereas in section B19 the sequence starts again at this point after a considerable hiatus ranging from subzone Id to halfway through subzone IIc. In section B16 a sedimentation of sandy clay began in the course of subzone IIc. Again there is a concentration o f charcoal at the level corresponding with the maximum cold phase in section B16. In section II-B1 peaty clay occurs, and in section B19 peat apparently formed in a minor gully in the "fan". Climatic conditions were relatively cold and somewhat humid, although n o t so cold as they were during the interval IIa. The transition to subzone IId is marked by a rise in the Alnus curve and a fall o f the curves o f Gramineae and subp£ramo elements (here mainly Compositae tubuliflorae) to only a few percents. In section II-B1 the sequence is interrupted by a hiatus.

88

Subzone IId (B16, B19) represents a relatively warmer phase during which Alnus and Weinmannia were again dominant among the trees. In section B16, the deposition of sandy clay continued and in section B19 the peat forming in the gully o f the "fan". In the diagram of section B19 Alnus dominates apparently owing to the local wet conditions, whereas in the diagram of section B16 directly after the rise of the Alnus curve, Weinmannia takes over and becomes dominant. During subzone IId climatic conditions must have been relatively warm; presumably warmer than they were in subzone IIb. Locally, soil conditions may have been wet, but in general the conditions must have been only slightly humid. We propose to call this warmer phase Rocas de Sevilla IIIb Interval (RIIIb). The transition of subzone IId to subzone IIe is marked by a fall of the tree pollen curves and a steep rise of the Gramineae curve. Subzone IIe (B19) is only represented in the diagram of section B19 by t w o spectra and also in that o f B22 (not published). As mentioned already, section B22 was very useful for the correlation owing to the inclusion of subzone IIe and of the end of the preceding subzone IId. Subzone IIe represents a relatively cold phase with high percentages of Gramineae. In the diagram of section B22 this maximum is followed by a maximum of the curves representing the subp£ramo vegetation belt. According to the stratigraphy of section B19, clay was deposited as also occurred at the location of section B22. In the latter also some Botryococcus is found. Climatic conditions must have been relatively cold and wet as indicated by the deposition of clay. Apparently, the water level of the great Sabana lake began to rise and the water penetrated into the E1 Abra corridor. At the transition from subzone IId to IIe a 14C-dating yields an age of somewhere between approximately 55,000 and 48,000 B.P. (Col. 198; GrN-6548). The sequence discussed above, consisting of warmer and colder phases, probably corresponds, according to the '4C-dating, with Early Glacial interstadials and stadials (IIa--d) and with the Lower Pleniglacial (IIe). In that case the Rocas de Sevilla IIIa and IIIb Intervals could correspond in time w i t h t h e European Br~Srup and Odderade Interstadials (and the Rocas de Sevilla II Interval with the Eemian). It must be emphasized here again that these correlations are very tentative; only additional data and especially '4C-datings can render these correlations more definite. Unfortunately in none of the sections is the transition from zone II to zone III present. Zone III (B9, B l l , B16, IV-IO7N, III-IOE, II-IOE, II-IOON, II-B1). Zone III consists of a series of warmer and colder fluctuations. During the warmer phases the E1 Abra corridor became inundated, completely or only partly. This was apparently due to the rising of the lake level of the great Sabana lake in the Sabana de Bogota. During interval III the role of the elements forming an open vegetation increased in importance (total values up to a b o u t 75%). However, in the lower part forest vegetation still dominated. Zone III is divided into five subzones: a l , a2, a3, b and c.

89 The only indication of the presence of a subzone IIIal is found in the diagrams o f section B9 and IV-107N. In section B9 it is only represented by one spectrum (the lowermost one). It is a relatively warmer phase with extremely high percentages of Botryococcus in the relatively lower soil section B9. Among the trees, Alnus and Myrica are best represented. From n o w on Weinmannia pollen is only represented in very low percentages. The climatic conditions must have been relatively warm and extremely wet. The E1 Abra corridor was completely inundated, as indicated by the sedimentation of lake clay with high Botryococcus percentages on the valley b o t t o m . R o c k shelter IV-107N probably formed a part of the lake shore. The sediment o f section IV-107N consists of an almost black, humic sandy loam alternating with relatively coarse sand. The rise of the lake level was probably caused by a considerable increase of precipitation. Accordingly, the climate must have changed considerably at the transition to zone III. The following subzone IIIa2 (B9, IV-107N) is a relatively cold phase with high Gramineae percentages (ca. 50--75%). The principal arboreal pollen represented are of Alnus and Myrica. The Botryococcus curve decreases to a minimum of a b o u t 30%. The lake clay shows signs of pedogenetic influences. In section IV-107N the coarse sand c o m p o n e n t increases in importance. Apparently the E1 Abra corridor was no longer inundated; only in low-lying places there may have been some standing water. This means that the lake level of the great Sabana lake must have dropped below the 2,570 m level of the E1 Abra valley. Climatic conditions were relatively colder and less humid as compared with the preceding subzone IIIal and also with the subsequent subzone IIIa3. Subzone IIIa3 (B9, B l l , IV-107N) represents a relatively warmer phase, again with extremely high percentages of Botryococcus. The E1 Abra corridor was completely inundated once more as the result of a renewed rise of the level o f the Sabana lake. Alnus and Myrica dominated alternatingly. On the b o t t o m o f the valley sedimentation of lake clay continued. In rock shelter section IV-107N the sedimentation of a coarse sand alternating with layers of a dark humic sandy loam continued. The conditions must have been relatively warm and wet. We propose to call the two warmer phases of subzones IIIal and IIIa3 Rocas de Sevilla IVa Interval (RIVa) and l~ocas de Sevilla IVb Interval (RIVb), respectively. The transition to subzone IIIb is marked by a distinct rise of the curve of the grass pollen and a fall of the Botryococcus curve. Subzone IIIb (B9, B l l , B16, III-10E) is characterized by a p~ramo-like vegetation representing a relatively cold phase. The tree pollen curves remain relatively low. The percentages of Botryococcus decrease to almost zero. The sediment changes from lake clay, already subjected to some pedogenesis, to a dark humic sandy loam. Climatic conditions must have changed from very wet in subzone IIIa3 to relatively dry, and it must have been relatively cold. Subzone IIIb is most completely represented in the sections B9 and B l l , and part of it in B16 and III-10E. In section III-10E this subzone is repre-

90 sented at the base. In section IV-107N the succession seems to be interrupted at this point. On t o p of this dark humic sandy loam a so-called " y e l l o w - m o t t l e d " very sticky loam layer is found (apparently of volcanic origin; see section "Analysis of material of volcanic origin"). This layer is also found in the following sections: B9, B l l , B16, III-10E, III-12E, II-10E, and II-100N. There is a second "yellow m o t t l e d " layer somewhat later in the succession in subzone IVa, just below the "loessic" layer in the same profile as the one in which section IV-107N is situated. There is no p r o o f whether the " y e l l o w - m o t t l e d " layer o f sections II-10E and II-100N is related to the earlier or to the later deposits o f volcanic origin, although we favour the oldest dating. The uncertainty is due to the poor preservation of the material and to an expected discontinuous accumulation of material in these rock shelter sections. The transition to subzone IIIc is marked by an increase of the Alnus percentages and another rise of the Botryococcus curve. Subzone IIIc (B9, B l l , B16, III-10E (II-B1)) is characterized by an apparently warmer fluctuation. The humidity seems to have increased again; however, conditions did n o t become so wet as in the preceding warmer subzones IIIal and IIIa3. Instead of lake clay, a somewhat humic clay was deposited and the Botryococcus curve reaches a relatively l o w e r maximum (of only 10--20%) towards the end of the subzone. Climatic conditions were only slightly warmer and somewhat wetter than during the preceding subzone IIIb and certainly colder than in the last warmer subzone IIIa3. The E1 Abra corridor was n o t inundated and did n o t form a part of the Sabana lake as it did in subzones IIIal and IIIa3. Possibly there were only some local, l o w - .... : : lying places with standing water (e.g., the valley in front of rock shelter II). A i4C-dating was obtained from the end of subzone IIIc in section III-10E of: ca. 28,000 B.P. (Co1.137; GrN-6269). Some datings and correlations will n o w be discussed. In the soil sections, subzone IIIc forms the lower pa~t of a sequence of dark humic sandy loam which is followed by the sequence of "loessic" material. From section IV-107N some datings were obtained of about 24,000 B.P. for the last warmer fluctuation also represented in the uppermost part of the dark humic sandy loam layer of the soil sections (Co1.144 and 145; GrN-6279 and 6280). According to these 14C-datings of the dark humic layer, one at the t o p of the dark humic layer of a b o u t 24,000 B.P. and one in the middle of a b o u t 28,000 B.P., the base of this layer in the soil sections must have an estimated age of a b o u t 32,000 B.P. The base of this layer coincides with the beginning of subzone IIIc, so that the warmer subzone IIIc can be correlated with the Santuario Interval from the Fuquene sequence (Van Geel and Van der Hammen, 1973) and a correlation with the European Denekamp Interstadial also seems acceptable. It is interesting to compare this result with a section from Zipaquira (Zip.I) palynologically studied at the Hugo de Vries-Laboratory (publication in prep.). It concerns a warmer fluctuation which can tentatively be placed in the Middle Pleniglacial of the Last Glacial according to

91 the stratigraphical evidence. From this warmer fluctuation, a 14C-dating of 32,890 + 660 B.P. (GrN-5838) was obtained. Immediately above this 14C-sample, Aragoa pollen begins to appear and continues to be present during a few spectra in the very last part of this warmer fluctuation. If this diagram is compared with our diagram of section III-10E, the situation is more or less the same, but there is a ~4C-dating of 28,140 + 440 B.P. slightly above the relative maximum of the Aragoa curve. This " m a x i m u m " of Aragoa also marks the end of the warmer fluctuation. The correlation of the RV Interval with the Santuario Interval seems to be relatively certain. The Rocas de Sevilla IVa and IVb Intervals might, then, correspond with the European Moershoofd and Hengelo interstadials, respectively. Anyhow, the apparent differences in climatic conditions between subzones IIa, b, c, d, and zone III with subzone IIe as a colder transition between them corroborate the supposition that the border between zones IId and IIe/III can be correlated with the border between the Early Glacial and the Pleniglacial, the colder subzone IIe corresponding to the Lower Pleniglacial. Zone III might then correspond to the Middle Pleniglacial. The existing 14C-dates favour such a general correlation. However, it will be clear that any correlation before 30,000 B.P. remains highly tentative until more ~4C dates become available.

Zone I V (sections B9, B l l , B16, IV-IO7N (II-B1)). Zone IV is characterized by extremely high percentages of Gramineae and/or Compositae tubuliflorae. The apparently slightly warmer fluctuations which interrupt this general picture are of far less importance than those found in the preceding zone III. Zone IV is subdivided into five subzones: a, b, c, d, and e. Subzone IVa (B9, B l l , B16 (II-B1, IV-107N)), a relatively colder subzone, starts with high percentages of Compositae tubuliflorae already present in the u p p e r m o s t part of zone III. The humidity, which had increased somewhat in subzone IIIc, decreases again during subzone IVa. In the diagram of section IV-107N the Compositae tubuliflorae maximum is not manifestly developed, which is possibly attributable to the hiatus mentioned before (see under subzone IIIb). This maximum is followed by a maximum in the Gramineae curve, which is very distinct in the diagram of section B9, but in which on the other hand the Compositae tubuliflorae maximum is n o t apparent. It m u s t be the same Gramineae maximum as the one in the t o p part of the diagram of section IV-107N. Apparently, this signifies the end of the hiatus in section IV-107N. In section III-10E, this Gramineae peak is missing. Here the "loessic" layer follows immediately after the phase with high Compositae tubuliflorae percentages. In the stratigraphy this hiatus is very distinct. There is a succession from a warmer interval (subzone IIIc), represented mainly by Alnus, to the colder subzone IVa, represented by Gramineae, via a marked increase in the Compositae tubuliflorae curve. Apparently, the vegetation belt in the E1 Abra valley moved from the upper limit of the Andean forest zone via the subpdramo with abundant Compositae tubuliflorae scrub

92 towards the grass paramo. Climatic conditions must have been relatively cold and slightly humid. Subzone IVb (B9, B l l , IV-107N (B16)) represents an apparently milder fluctuation. Only the Alnus pollen percentages increase somewhat. According to the Botryococcus curve of the diagram of section B9, the humidity may have increased somewhat. This interval could be correlated with the Suta Interval (Van Geel and Van der Hammen, 1973). Subzone IVc (IV-107N) consists of a colder fluctuation of short duration during which the curve of Compositae tubuliflorae increases somewhat; this is only visible in the diagram of section IV-107N. Subzone IVd (IV-107N) represents a last warmer fluctuation of zone IV, which precedes the relatively extended cold subzone IVe; this fluctuation is only visible in the diagram of section IV-107N. Subzones IVb, IVc, and IVd cannot be separated in the diagrams of sections B9 and B l l . The relatively milder fluctuations of subzones IVb and IVd appear as only one single fluctuation. During this fluctuation a very modest appearance of Weinmannia pollen is visible only in section B9. The Botryococcus curve rises in these t w o sections. This may point to increasingly humid conditions. In section IV-107N, Botryococcus is even f o u n d in subzone IVd, whereas this alga was never f o u n d in the rest of the section. The valley must have been very wet, and there may have been standing water in low-lying places in the valley. In the u p p e r m o s t part of subzone IVd a 14C-dating from section IV-107N is available o f 23,850 + 280 B.P. (Co1.144; GrN-6279). This interval could be correlated with the Saravita Interval (Van Geel and Van der Hammen, 1973). From the same profile there are t w o additional 14C-datings o f 23,870 + 185 B.P. (Co1.185; GrN-6280) and 21,050 + 210 B.P. (Co1.146; GrN-6281). All three samples were taken just below the "loessic" layer. In the profile, a local erosion level is present on which the "loessic" layer was deposited. This explains the slightly diverging 14C-datings. At the transition from subzone IVd to subzone IVe a further increase of the Compositae tubuliflorae t o o k place (see the diagram of sections B9, B l l , B16 and IV-107N) coinciding with a rapid lowering of the Botryococcus curve. In the E1 Abra valley the wet conditions apparently ceased once more. During subzone IVe (B9, B l l , B16) the total percentages of elements common in the p~ramo exceed 90%. This must result from a substantial drop of the mean annual temperature. The transition from subzone IVd to subzone IVe coincides with the base of the "loessic" layer. The only pollen data available are provided by the soil sections B9 and B l l , and by section B16. In the gully in front of rock shelter II the sequence is discontinuous, due to erosion activities during periods with a greater run-off of water. The rockShelter sections do not provide any pollen through the rest of the sequence from the beginning of subzone IVe onward. The diagrams of the soil sections B9 and B l l provide the most differentiated picture. At the base of subzone IVe the curve of Compositae tubuliflorae attains a maximum. Through subzone IVe the Gramineae percentages increase towards a distinct maximum

93 in the top layer of this subzone, apparently representing the coldest part. This Gramineae peak is only well developed in the diagram of section B9; at that point the "loessic" layer has changed into a more sandy deposit with distinct indications of pedogenesis. This change coincides with a distinct increase in humidity. During the first part of subzone IVe, climatic conditions were dry and cold. In the top part of subzone IVe, the climate apparently became colder. At the same time the humidity must have increased and there may have been standing water in low-lying places in the valley. The approximate age of the basal part of subzone IVe is 21,000 years B.P., according to 14C-data from the top layer of subzone IVd. Subzone IVe apparently represents the principal cold interval of the Upper Pleniglacial, and can be correlated with the Fuquene Stadial (Van Geel and Van der Hammen, 1973).

Zone V (II-B1, II-B3, B9, B l l , B16). This zone represents the period during which warmer and colder phases alternate before a definite amelioration of the climate in the zones VI through VIII t o o k place. Zone V is subdivided into two subzones: a and b. Subzone Va (II-B1, II-B3, B9, B l l , B16) represents a warmer phase during which the Alnus curve in particular shows a maximum. At the same time the Botryococcus percentages increase rapidly. Apparently, the increase of humidity which started at the end of subzone IVe continues at a fast rate. The b o t t o m of the E1 Abra corridor must have become further inundated. In section B9 this climatic amelioration is clearly reflected; in section B l l only a peak in the Botryococcus curve could be recorded. The diagrams.of the gully in front of rock shelter II show the change more distinctly. There must have been extensive patches of open water in the valley. Probably the valley b o t t o m was inundated. It seems as if the highest water level was reached towards the end of this subzone Va. From the t o p layer of subzone Va of section II-B3 a 14C-dating suggests an age of 11,210 + 90 B.P. (Co1.!23; GrN-5941), so that this fluctuation must be synchronous with the Guantiva Interstadial (Van Geel and Van der Hammen, 1973). It seems very likely that this interstadial may be correlated with b o t h the Belling and the Aller0d i nterstadials of Europe. Subzone Vb (II-B1, II-B3, B16, B9, B l l ) . Neither the diagrams o f the soil sections B9 and B l l nor that of section B16 provide us with more detailed information. Only the sections from the gully in front of rock shelter II, II-B1 and II-B3, yield useful data, especially the latter. They very distinctly reflect a Considerable drop in temperature and a decrease of the humidity at the beginning of subzone Vb. At first the Gramineae percentages are high. This peak is followed by a steep maximum in the curve of Compositae tubuliflorae. The occurrence of Cactaceae seems to point to rather extreme arid conditions. The occurrence of pollen grains of still unidentified plant species in section II-B3 must be noted. During the last part of subzone Vb the climate

94 improved somewhat, a short, warmer phase again being followed by a peak in the Gramineae curve. The climate, however, did not become so cold again as it was at the beginning of subzone Vb. The transition to zone VI is determined by the start of a renewed manifest dominance of forest elements. The succession in subzone Vb is so well marked in the E1 Abra sequence that it was decided to select the El Abra valley as the type site, and to call this colder phase the E1 Abra Stadial. The name was used for the first time for deposits at the Fuquene site (Van Geel and Van der Hammen, 1973). The larger part of this stadial undoubtedly correlates with the European Younger Dryas Stadial. Apparently this equivalent of the European Younger Dryas Stadial has been notably cold, and later also became considerably dry. The exact placing of the lower limit of the Holocene (casu quo: upper limit of the Late Glacial) is still uncertain. The beginning of the Holocene, according to international agreement should be placed at the beginning of the amelioration of climate of ca. 10,000 B.P. This might be the beginning of the short, warmer interval preceding the second Gramineae peak and following after the Compositae tubuliflorae maximum. In that case this short warmer interval might be approximately contemporaneous with the Friesland phase, and the second Gramineae peak with the Rammelbeek phase of the European preBoreal (Wijmstra and De Vin, 1971). In that case the E1 Abra Stadia] would include the lowest part of the European pre-Boreal (see also Van Geel and Van der Hammen, 1973). Since absolute datings are missing in this part of the sections, as yet no definite conclusions can be drawn.

Zone VI (II-B1, II-B3, B16 (B9, Bll)). The curves of Gramineae and Compositae tubuliflorae fall off rapidly, while the percentages of tree and herbaceous pollen increase. This zone can approximately be correlated with the Andean pollen zones IV and V (approximately corresponding to the European pre-Boreal and Boreal).

Zone VII (II-B1, II-B3, B16, (B9, Bl l ) ).During this zone the tree pollen percentages reach their maximum values, altogether to about 75%. This zone must approximately coincide with the Andean pollen zones VI and VII (that are approximately equivalent to the European Atlantic and subBoreal). The limit between these two zones possibly lies there where the percentages of the so-called "other forest elements" start decreasing.

Zone VIII (II-B1, II-B3, B16, B9, Bll). Throughout this zone human influence is reflected on the vegetation. The decline of the arboreal pollen curves and the increase of curves of elements of more open vegetation types is certainly attributable to human activities; Gramineae ponen ,comparable to maize were found in section II-B3.

95

MOVEMENTS OF VEGETATION BELTS, CHANGES OF TEMPERATURE AND F L U C T U A T I O N S OF THE WATER LEVEL IN THE VALLEY

The construction o f the curve indicating movements o f vegetation belts and temperature changes In this chapter the construction of the curve of vertical shifts of the vegetation belts and of approximate changes in the average annual temperature (see Fig. 15) will be discussed. Firstly, the altitudinal ranges of various vegetation types must be considered and also the vertical range of certain characteristic species which played an important role in the local E1 Abra vegetational succession. Weinmannia forest, an important part of the stands of vegetation in the lower part of the composite diagram, presently does not occur above 3,500 m altitude. Alnus, consistently present t h r o u g h o u t the diagram, likewise does n o t grow above an altitude of 3,500 m. However, Alnus, is a much more profuse pollen producer than Weinmannia, so that the presence of relatively low percentages of Alnus pollen does n o t imply that Alnus formed part of the local stand of vegetation. In the Eastern Cordillera the Andean forest proper does not exceed a m a x i m u m altitude of 3,500 m, its upper limit lying in any case between 3,200 and 3,500 m. Above that limit dwarf forest (scrub) and subp~ramo occur with an abundance of characteristic taxa of Compositae tubuliflorae, Hypericum and Ericaceae. At higher altitudes from about 3,500 m upwards, a proper grass p~ramo becomes increasingly more c o m m o n to dominate entirely at altitudes above 3,700--4,000 m. The relation between the altitudinal range of certain vegetation types and the average annual temperatures will be used in an a t t e m p t to reconstruct a curve showing shifts of the vegetation belts and the corresponding temperature changes. From previous studies of the recent pollen rain in the Eastern Cordillera data were obtained for the correlation of the relative fluctuations of the Gramineae curve with absolute values (in metres) of the displacements of the tree line {Van der Hammen and Gonzalez, 1960). A difference of ca. 5% in Gramineae pollen, corresponding with a vertical shift of 100 m of the tree line, seems a reasonable measure. In this way the curve of Gramineae can be converted into a curve which indicates the probable vertical displacement (in metres} of the tree line. Actually, the curve of the real fluctuations of the tree line is the result of two factors, viz. of changes in temperature and of changes in humidity, the first being far more important. On the other hand, it is k n o w n that the temperature decreases or rises more or less accurately, rather constantly by 2/3°C per every 100 m of altitude. The curve showing tree-line displacements calculated on the base of the relative pollen curve of Gramineae has been adapted to some extent, taking other factors into consideration, especially the occurrence or dominance of (groups of) taxa typical or a b u n d a n t in the different vegetation belts.

96

The construction o f a diagram indicating the maximal fluctuations of the water level in the valley The construction of this diagram is mainly based on the occurrence of Botryococcus in the pollen diagrams of the various sections. The use of the presence of Botryococcus as an indicator of fluctuations of the level of open waters is supported by incidence of corresponding sedimentation types in the E1 Abra valley, viz., of layers of a pure lake clay and transitions of such a clay to a more peaty facies subjected to pedogenesis. Generally speaking, each former situation corresponds with a p r o n o u n c e d maximum in the Botryococcus curves, and each latter one with a less pronounced rise or fall of the Botryococcus percentages. It must be emphasized that these observations are made on the basis of all diagrams and n o t merely on a single one. Therefore, the assumption seems warranted that extremely high percentages of Botryococcus correspond with an open-water facies all over the valley and lower percentages are probably indicative of smaller and shallow (and scattered) bodies of water. In this way the highest values of the open-water depth, at times when the whole valley was inundated, can be estimated. At such times the water depth indicated by the curve corresponds with the height difference in metres between the estimated lowest point in the gully in front o f rock shelter II and the highest level in the valley itself where most probably open water was present. If there is only an indication of some open water in the gully, the water depth indicated b y the curve corresponds with the height difference b e t w e e n the surface level of the gully and the possibly highest level of open water in that gully. However, in that case, there were probably smaller and shallow bodies of water in the b o t t o m valley itself. ANALYSIS OF MATERIAL OF VOLCANIC ORIGIN

In this section the results are given of analyses of heavy minerals of soil material from the so-called "yellow-mottled and very sticky layer", and from the so-called " y e l l o w mottles to fragments" which occur, respectively in the zones IIIb (and IVa/b), and Vb to V1, and of the "loessic" material of zone IVe when mica is abundant (sections B9, B16, III-10E). These analyses were carried o u t at the Physical-Geographical Laboratory of the University of Amsterdam by Mr. H. Stoltenberg. The analysed heavy minerals all have a grain-size exceeding 32 pm. The list of all analyses showing the heavy-mineral contents follows below; minerals provided with an asterisk may be indicative o f a volcanic origin of the material studied. Section II-10E, depth 350 cm (zone IIIb; possibly zone IVa/b): *green Amphibole *green to brown Amphibole anatase epidote *hypersthene

*opaque futile tourmaline zircon

97 Section II-100N, depth 268 cm (zone IIIb; possibly zone IVa/b): *aegirite * brown amphibole * green amphibole

epidote *hypersthene * opaque

Section III-10E, depth 375 cm (zone IIIb): *green amphibole *hypersthene

*opaque

Section B l l , depth 125 cm (zone IIIb): *aegirite *brown amphibole *augite

*hypersthene *opaque

Section B9, depth 135 cm (zone IVe): *aegirite *brown amphibole *green amphibole

epidote *hypersthene *opaque

Section II-B3, depth 115 cm (zone Vb): *aegirite *brown amphibole * green amphibole

* biotite epidote *opaque

All the samples, except the one from section B9, were taken from the so-called " y e l l o w m o t t l e d and very sticky" layer (IIIb, IVa/b) and from the layer with so-called " y e l l o w mottles to fragments" (Vb). The sample of section B9 was taken from the layer with "loessic" material. Only with sections II-10E and III-10E is a volcanic origin somewhat doubtful. The a m o u n t and the variety of heavy minerals of volcanic origin is too small in both samples to be conclusive. From stratigraphically corresponding levels of nearby sections there is evidence of the volcanic origin, of both samples, however. It follows that several horizons containing material of volcanic origin occur in the stratigraphic sequence identifiable by the occurrence of the socalled " y e l l o w mottles to fragments". These horizons provide an important contribution towards the stratigraphical correlation. They are indicated on Fig. 15 (in a special column). Apart from these levels with weathered volcanic material, concentrations of mica were found in several sections. The material rich in mica in section B9 was also found to be of volcanic origin, and the presence or abundance of mica in the sections is therefore indicated in the separate column (Fig.15) likewise. In the excavation pit of rock shelter IV another layer with "yellow m o t t l e s " was noticed. Since no section was taken at that site, no samples were available

98

for an analysis o f the heavy-mineral content. However, we consider this layer also to be of volcanic origin. This layer is situated at the transition from subzone IVa to subzone IVb and is also indicated in the column of Fig.15 mentioned before. Van Geel and Van der H a m m e n (1973) found thin layers of volcanic ash rich in mica in thei sediments of Lake Fuquene; it is interesting to note that t h e y are likewise abundant in the interval b e t w e e n ca. 30,000 and 20,000 years B.P. DESCRIPTION OF SOME UNIDENTIFIED PALYNOMORPHS (Plate I)

In the samples six u n k n o w n pollen types were found which seem to be of some special interest. Types 1--4 were found in section II-B3, mainly in subzone Vb and with some finds at the end of subzone Va and the beginning of zone VI. Types 1--3 seem to be related and we suppose that they belong to Compositae. T y p e 5 appeared also only in section II-B3 b u t was found only in zones VI, VII and VIII. Type 6 appeared only in section IV-107N, z o n e IIIa. A pollen t y p e probably referable to Dalea could be the first fossil record of the genus. It was f o u n d in several sections and n o t restricted in its occurrence to special parts of the sections. At least two pollen representing Cactaceae were f o u n d in section II-B3, subzone Vb. The importance of the presence of Cactaceae was pointed o u t in the preceding chapters. Only short descriptions of this t y p e are shown below: type 1 (Plate I, 1):

tricolporate, microechinate, with distinct columellae, a very thick exine and distinct pores; tricolp(or)ate, without distinct pores, otherwise as type 1; type 2 (Plate I, 2): tricolporate, psilate, otherwise as type 1; type 3 (Plate I, 3): stephanocolp(or)ate, (4 colp(or)ate), perforate, pores not very type 4 (Plate I, 4): distinct, columellae very coarse, exine very thick; tricolpate, striate; type 5 (Plate I, 5): tricolpate, perreticulate, with very distinct columellae; type 6 (Plate I, 6): tricolpate, psilate, colpi slightly constricted, and with a very cf. Dalea (Plate I, 7): small polar area. This grain is morphologically very similar to Dalea. pericolpate, eehinate, perforate, the genus could not be deterCactaceae (Plate I, 8): mined. Sporae insertae sedis 2: description in: Van der Hammen et al. (1973), according to Van Geel and Van der Hammen (1978) recently recognized as zygospores of Debarya. Sporae insertae sedis 4: description in: Van der Hammen et al. (1973). i

THE OPALINE SILICA BODIES (= PHYTOLITHS) ANALYSIS (see Fig.16)

General remarks

During the microscopic analysis of the pollen material, bodies were observed which do n o t seem to be of organic origin. Initially they were supposed to be soil particles o f volcanic origin, but after a consultation with the physicalgeographic department, it became apparent that they represent opaline silica

99

PLATE I (x 1000, except Figs. 11--12: × 500) 1--8 Unidentified palynomorphs; 9--10 fossil opaline silica bodies: Cortaderia (9), Chusquea (•0); 11--13 Recent opaline silica bodies: Cortaderia (11), Chusquea (12), Calamagrostis (13).

Ia

Ib

J

i

i

i

i

5t~ J

t-~ . . . .

~zb

,",f ,~nnllnP_

,NN

~

,

\, \

r-lOE

,

yellow mottles ~aterfa(s of volcanic origin) manchas amarillas (material de ¢¢~en volcardco)

r

I

/

/

I

silica b o d i e s ( D h v t o l i t h s ) o f t h e v a r i o u s s e c t i o n s .

I . . . . .

F

~rnber per 6mmz of slide ~umero pet- 5rf~cn2 de piaca

)_

h~us . . . . .

1~'-I07N

,";,., 1 ~

Ed

I

T

~g

I

ocal zones !L ABRA

]I-12E

;ILICA BODIES :ITOLII"OS

I-B3 E-B1

Ill

~9

:)

101

bodies, which are known to be originated in plants. It was decided to start an investigation o f the possible significance of the various degrees of representation of the opaline silica bodies in the pollen samples. Opaline silica bodies range in size from (10 --) 25--60 (-- 80) ttm; t h e y are isotropic. They originate inside certain cells of plant tissues. (Netolitsky, 1929; Parry and Smithson, 1957; Metcalfe, 1960; Kaufman et al., 1970). Gramineae and m a n y Cyperaceae form the principal source of the opaline silica bodies (Netolitsky, 1929). The highest concentrations are found in the leaf sheath (Parry and Smithson, 1964). Each genus often forms silica bodies of a distinct and characteristic shape; therefore this can be of importance to distinguish various genera of grasses (Parry and Smithson, 1964). Studies of the silica bodies of recent grasses carried o u t in England yielded interesting results. Their occurrence in soils was used for the reconstruction of the " r e c e n t " history of the soils in connection with land reclamation and agriculture (Smithson, 1956a, 1958). Since opaline silica bodies are found in plants and the plant material has completely disintegrated, they remain in the soil as isolated particles. Their integration in the soil renders it likely that, unless retransportation of the sediment occurred, t h e y accumulated in situ and that their presence signifies the erstwhile occurrence o f sedimental areas with a vegetation cover. The opaline silica bodies were f o u n d in nearly all samples of the various sections studied (with the exception of the oldest ones), but t h e y occur in different concentrations. It, therefore, became necessary to elucidate the meaning of the different concentrations of the silica bodies in the slides, and to determine their different forms in connection with their indicative taxonomic significance. The preparation m e t h o d of the slides and the principal differences between the counting of pollen grains and of opaline silica bodies in the same slides in order to try to establish a possible relationship must be explained first. It was not deemed necessary to work quantitatively with samples of the same weight, because of the lack of information concerning the prospective results: the a m o u n t of work involved would not warrant such a refinement. The differences between the amounts of silica bodies in various samples is, moreover, so large (100--1000%) that the relative error in the preparation process could be, admittedly, relatively substantial. Therefore, from each sample approximately the same a m o u n t by volume was taken, and these quantities were processed exactly the same way. Taking the appreciable differences in concentration of silica bodies into account, this m e t h o d is thought to be adequate to establish the principal variations. For the preparation o f pollen samples a separation liquid is used with a specific gravity of 2. Since the gravity of opaline silica bodies is slightly higher and averages 2.14, the shortest possible centrifugation time was used. An experiment with a separation liquid with a specific s.g. of 2.2 produced a centrifugate o f opaline silica bodies contaminated with so much debris that the material could n o t be properly counted. Nevertheless, concentration differences between relatively poorer and richer samples are manifest, as stated before. To obtain an overall result the m e t h o d applied appears to be sufficiently reliable, if the sample is centrifuged in the shortest possible

102 time and, almost needless to say, n o t subjected to a H F treatment. First, something must be mentioned a b o u t the quantitative approach, i.e., the measuring of the differences in concentration between the various samples. The counting of the silica bodies is carried o u t as follows: per slide area (circular, with a diameter o f ca. 12 mm) three surfaces of 2 x 2 mm 2 were searched for silica bodies; the three equidistant surfaces of 4 mm 2 each were selected along the diameter of the sample with the aid of the graduation of the objective table of the microscope. Now and then surfaces of 4 mm 2 were selected along diameters running in a different direction to check any appreciable difference in the number of silica bodies, b u t substantial differences were never f o u n d within one slide. The number of silica bodies counted per slide was n o t expressed in percentages of the pollen total. In some parts of the sections the pollen total is t o o small, or pollen m a y be lacking altogether. For this reason the absolute numbers of opaline silica bodies per tested unit o f slide surface (3 x 4 mm 2 = 12 mm:) were recorded.

The quantitative and qualitative analyses; discussion and conclusions The curves showing the variation in the numerical amounts of silica bodies per surface unit are represented in the pollen diagrams in the first column to the right of the main diagram. This was purposely done to facilitate the comparison of this curve with the grass-pollen curve, grasses apparently being numerically the most importantsuppliers of silica bodies. A comparative interpretation of the curves of the opaline silica bodies and a comparison with the pollen diagram can only be made if one takes the following restrictions into account: (1) the opaline-silica-body curves do n o t show relative values as do the pollen curves, b u t the absolute amounts per sample volume, i.e., per surface unit of the slides; this means that when maxima and minima coincide, the recorded data may differ considerably in their absolute values; (2) the curves suggest that they provide exact data; one should, however, n o t try to attach any considerable importance to minor fluctuations because the m e t h o d is not sufficiently exact. As indicated above, Gramineae and Cyperaceae are the most important suppliers; however, considering that, in general, the grasses dominate over the Cyperaceae in the pollen diagrams, the former have probably been the most important producers of silica bodies in the deposits under discussion. The silica bodies were, as said above, probably produced by in situ stands of vegetation; the possible incidence of allochtonous silica bodies (resulting from resedimentation) has, n o t been taken into account. A mutual comparison of the curves of opaline silica bodies (Fig.16), reveals that the curves run largely parallel. In nearly all curves the highest numbers coincide with the maximum cold in E1 Abra zone IV and with the E1 Abra Stadial, the coldest part of the Late Glacial. This is completely according to

103

expectation since it is generally assumed that during these intervals open vegetation types with a domination of grasses prevailed. The following Holocene deposit contains extremely small numbers of opaline silica bodies, and the period preceding the Santuario Interval, as far as represented in the sections and as far as silica bodies counts were carried out, contains even lower numbers or none at all. Generalizing, one m a y accept that the relatively coldest periods, when the E1 Abra valley mostly lay within the subp~ramo/p~ramo zone, are characterized by an abundance of opaline bodies in the sediment. The curve of the silica bodies of section III-12E, from which no pollen could be extracted, is of special interest, because the silica bodies open up new perspectives for correlation. A comparison of this curve of section III-12E with that of the nearby section IV-107N shows that this new possibility of correlation is highly promising. We wish to stipulate that the curves of opaline silica bodies were n o t used to look for correlations between the pollen diagrams and the stratigraphic sequences of the sections in the E1 Abra valley. This renders the similarity of the evidence from the silica b o d y curves and that from other data even more striking. To pursue the second objective, viz., the qualitative approach to extract useful information from the silica bodies, ten leaf samples of recent grasses were compared with the fossil samples. These recent grasses are of c o m m o n occurrence in the present vegetation cover (Plate I): Agrostis thrichoides Roem. et Schult Bromus lanatus H.B.K. Calamagrostis nuda (Pilger) Pilger Chusquea tessellata (Munso) McClure Cortaderia colum biana Pilger

Festuca dolichophylla Presl Paspalum hirtum H.B.K. Poa annua L. Stipa ichu (R. et P.) Kunth Zea mays L.

In order to make their opaline silica bodies visible for a comparison with the "fossil" ones recovered from the sections, the recent leaves were treated in the following, very simple, way which is n o t very time-consuming: leaf fragments are heated in a crucible and carbonized. The carbonized remains are placed on a glass slide in a 1': 3 mixture of phenol and xylol. This mixture is volatile, so that it is necessary to seal o f f the sample immediately after mounting. If the carbonized pieces have retained their structure, the silica bodies can be located within the original cell pattern. Three of the ten species of grasses could thus be recognized in the fossil samples recovered from sections, viz., Calamagrostis, Chusquea, and Cortaderia (Plate I). Using the silica bodies o f recent grasses as reference samples, it appears to be possible to identify them in fossil samples to the generic level. Therefore, the use of opaline silica bodies to identify grass genera in fossil material is to be strongly recommended, because, unfortunately, it is n o t possible to distinguish them by their pollen morphology. Although the investigation o f the opaline silica bodies d i d n o t contribute

104 directly towards the formulation of the conclusions regarding the stratigraphic and vegetational sequences in the E1 Abra corridor, the quantitative approach supports the findings of the pallynological analysis, which constitutes an important asset in itself. It is even complementary to the pollen analysis when the pollen record is incomplete but opaline silica bodies are amply represented in the samples. The importance of the application of the qualitative approach for identification purposes can hardly be overrated. This m e t h o d might eventually provide cogent evidence for the reconstruction of past climates in, for example, intervals with p£ramo vegetation, because certain genera of grasses are restricted in their occurrence to special environments and may serve as indicators of the micro- and macroclimatological conditions in the past. GENERALCONCLUSIONS Summarizing, one can distinguish three main phases in the vegetational sequence that took place in the E1 Abra corridor. During the first phase, consisting o f the E1 Abra zones I and II, the El Abra valley was most of the time covered with a dense stand of vegetation belonging to the Andean forest belt with Weinmannia as one of the most prominent trees. During the second phase, consisting of E1 Abra zones III, IV and V, mostly an open-vegetation type o f the subp~ramo belt prevailed in that valley, zones IIe and I!Ia, and V forming transitional phases of an increasing humidity, ultimately resulting in the inundation of the E1 Abra corridor. The third phase consists of the E1 Abra zones VI, VII and VIII. The valley again lay within the Andean forest belt, Alnus now being one of the most prominent trees. Unfortunately subzone IIe, presumably representing the Lower Pleniglacial, is only represented by two pollen spectra in the E1 Abra sections available for the present study. A very recently studied section of B22, in which subzone IIe is represented, provided some useful information, however. So far this section yielded the only diagram which fully includes subzone IIe. For an overall survey, showing a "digest" of the various pollen diagrams, Fig. 14, was drawn up. These diagrams are cumulative. To the left of each diagram the various zone systems used in this study are indicated in addition to the stratigraphy. To the right, the presence o f Bot~yococcus and Iso~tes (the socalled "water indicatgrs" ) is indicated. The three phases will be described below in some greater detail so as to formulate the general conclusions arrived at in this study. Zones I and H (Phase 1)

During the intervals I and II, the E1 Abra corridor was most of the time lying within the Andean forest belt, with the exception of a few relatively short periods during which the subp~ramo vegetation belt descended into the valley (subzone Ic partly and subzones IIa, IIc partly and IIe). When the E1

105 Abra corridor lay inside the Andean forest belt, Weinmannia dominated among the trees. At wetter sites (i.e., in marshy places), A l n u s must certainly have dominated. The sedimentary sequence consists mainly of layers of peat, peaty clay or clay. There were no periods during which there were extensive bodies of standing water in the valley, although the soil conditions were apparently relatively humid. Fungal spores are well represented. The unit of grit and finer material washed down from the western rock wall and deposited near rock shelter IV to form a sort of fan, is of fluviatile origin. The level of the Sabana lake in the Sabana de Bogot~ must have remained below the 2, 570~m level. The sequence starts with a relatively warmer phase -- the Rocas de Sevilla I Interval -- presumably corresponding with the last warmer phase of the penultimate Glacial. It will be assumed here that during the next colder phase (the subzone Ic) the climate must have been somewhat more severe than it is at present and only slightly humid. During the following, warmer, Rocas de Sevilla II Interval (subzone Id), in time presumably corresponding with the European Eemian Interglacial, the climatic conditions were comparable with the present ones. During subzone IIa, presumably corresponding with the initial phase of the Early Glacial, the subp~ramo vegetation belt descended into the valley. Apparently, climatic conditions had become relatively colder. During the t w o intervals Rocas de Sevilla IIIa and IIIb (the subzones IIb and IId) the climate must have been relatively warm again. In the RIIIb Interval the last occurrence of relatively high percentages of Weinmannia was recorded. The intermediate subzone IIc was relatively colder, the Andean forest still dominating in the valley although the tree line must have descended to quite near the valley. During the interval IIe (only represented by two pollen spectra in our sections) the climate was also somewhat colder as compared to subzone IIc. The humidity seems to have increased and this initiated the situation of the beginning of zone III. Zones III, I V and V (Phase 2) During most of the intervals III, IV and V, corresponding with the Middle Pleniglacial, the Upper Pleniglacial and the Late Glacial, respectively, the E1 Abra corridor lay within the (sub)p£ramo vegetation belt. Only in the beginning of zone III, the Andean forest belt twice ascended into the valley again, viz., during the Rocas de Sevilla IVa Interval (subzone IIIal) and during the Rocas de Sevilla IVb Interval (subzone IIIa3). During the RIVa Interval Alnus was dominant among the trees and during the R I V b Myrica was also prominent, alternatingly dominating with Alnus among the stands of arboreal vegetation. Apparently Weinmannia had almost disappeared from the local forests. During the two above-mentioned intervals the E1 Abra corridor was inundated: lake clay was deposited and Botryococcus is amply represented. Apparently the Sabana lake rose above the 2,570-m limit corresponding with the actual height o f the floor of the E1 Abra valley. This rising of the lake level must be

106 attributed to a considerable increase of effective precipitation to a rate higher than today. This climatic change had already been initiated in the Lower Pleniglacial. Climatic conditions must have been relatively warm (comparable with the preceding warmer intervals of the Early Glacial) and also extremely humid. Presumably these two warmer intervals RIVa and RIVb correspond with the European Moershoofd and Hengelo Interstadials, respectively. During the intermediate colder interval IIIa2 the lower limit of the p~ramo vegetation belt almost reached the valley. Climatic conditions were relatively cold and dry. From the beginning of subzone IIIb onward till the end of zone IV the E1 Abra corridor lay in the (sub)p~ramo belt. Even during the warmer fluctuations of subzones IIIc and during those of the following zone IV the Andean forest did not invade the valley. A series of sandy to loamy "soily" material was deposited. Apparently conditions became less humid. Subzone IIIb represents a relatively cold and dry period. The upper limit corresponds with the previously mentioned "yellow mottled", very sticky layer of volcanic origin. It forms a guide horizon not only through the various E1 Abra sections, but also in other parts of the Sabana de Bogot~ (Van der Hammen, 1978). During the following interval IIIc, named The Rocas de Sevilla V Interval and corresponding with the Santuario Interval and in time with the European Denekamp Interstadial, the upper limit of the Andean forest almost reached the El Abra corridor. Apparently, climatic conditions became somewhat warmer. Towards the end also the humidity increased. However, there were no extensive bodies of standing water. During zone IV, corresponding with the Upper Pleniglacial, climatic conditions gradually became colder still and also drier. Presumably the effective precipitation was lower than it is at present (Van der Hammen, 1974). In zone IV there are a few warmer fluctuations corresponding with the Suta and Saravita Intervals (Van Geel and Van der Hammen, 1973) during which climatic conditions became only slightly warmer. During the final phase of zone IV, the Fuquene Stadial (subzone IVe), a pure open-grassland vegetation descended into the valley. Apparently the Fuquene Stadial formed the coldest and driest part of the Last Glacial. In almost all sections material with a "loessic" appearance was deposited. Towards the end the humidity seemed to increase somewhat, ultimately to cause a renewed inundation of the valley in the following Guantiva Interstadial (subzone Va). Zone V represents the Guantiva Interstadial and the E1 Abra Stadial, corresponding in time approximately with the European BOlling and Aller@d Interstadials, and the Late Dryas, respectively. During the Guantiva Interstadial the Andean forest again invaded the valley, Alnus being the most prominent tree. At the same time the valley was inundated once more due to an increase of the effective precipitation. Climatic conditions were relatively warm and humid. During the following E1 Abra Stadial (named after the El Abra corridor), the subp~ramo vegetation belt again descended into the valley,

107

and climatic conditions became relatively cold and dry again. It is n o t e w o r t h y that in the same period in tropical East Africa an increase of the effective precipitation also occurred in the Lake Victoria basin, followed by a relatively dry period (Kendall, 1969). The sedimental sequence in the "higher" places consists principally of "soily material". At the inundated places lake clay was deposited. When the water level dropped, peaty or humic clay was deposited. Summarizing, the following conclusions can be drawn: the Early Glacial and the Lower Pleniglacial differ from the Middle and Upper Pleniglacial by a shift of the vegetation belts from the Andean forest-zone/subp~ramo-zone to the subp~ramo-zone/p~ramo-zone, respectively. There must also have been a considerable increase of effective precipitation to a rate higher than it is today, by the beginning of the Middle Pleniglacial ultimately resulting in the inundation of the E1 Abra corridor. This was already initiated in the Lower Pleniglacial. There was a gradual decrease of the effective precipitation, probably to a rate lower than the present one, during the further course of the Middle and Upper Pleniglacial culminating in the Fuquene Stadial. Similar observations have been recorded elsewhere in tropical South America (Van der Hammen, 1974). The main cold period became terminated in the Late Glacial by an increase of the average temperature resulting in a shift of the Andean forest-zone/subpgtramo-zone into the valley. This prominent shift, uphill this time, of the vegetation belts is again accompanied by an increase of the effective precipitation resulting in the inundation of the E1 Abra corridor. The same happened, as we described above, when at the transition from the Early Glacial/Lower Pleniglacial to the Middle Pleniglacial the vegetation belts also shifted, that time, however, downwards. Zones VI, VII and VIII (Phase 3)

Zones VI, VII and VIII represent the Holocene. The E1 Abra valley again became covered with dense vegetation of the Andean forest-belt type. The amelioration of the climate after the Last Glacial reached its o p t i m u m during zone VII, corresponding with the Andean pollen zone VI (and approximately with the European Atlantic). Gradually a situation became established comparable with the present. Human influences (agricultural practice) resulted in the development of a somewhat more open vegetation. The sediment consists of homogenized "soily" material. This study o f the pollen and of the stratigraphical material of the E1 Abra corridor represents the first detailed study of the entire Last Glacial sequence in the Sabana de Bogotfi. For the definite establishment of the chronostratigraphy as described above, more detailed studies are required. Also for the long-distance correlation more detailed information is needed. It will, however, be apparent that the Glacial/Interglacial cycles (and minor climatic changes) of the polar and temperate zones also played an important role in

108

the vegetational history of the tropics (Van der Hammen, 1974). The analysis of opaline silica bodies was not used for correlations or for chronostratigraphic purposes. Nevertheless, the following conclusions could be drawn: all major maxima in the curves of opaline silica bodies correspond with the Upper Pleniglacial, principally with the Fuquene Stadial and with the Late Glacial. ACKNOWLEDGEMENTS

First, I wish to thank the Netherlands Foundation for Advancement of Tropical Research (WOTRO) who sponsored the laboratory work at the Department of Palynology and Palaeoecology of the Hugo de Vries-Laboratory of the University of Amsterdam. Without their financial support this study could not have been accomplished. However, without the interest and enthusiasm of Prof. Thomas van der Hammen I never would have finished this study. I wish to thank him specially for his cordial support. Furthermore, I wish to thank Dr. W. G. Mook for his interest in the special problems of correlation in the E1 Abra sequence and the work on the 14Cdatings; Prof. A. D. J. Meeuse for the revision of the English text; Bas van Geel for the analysis of some samples in the sections II-B1, II-10E and II-100N (when at short notice some results were necessary for a first report on the 1969 excavations of the E1 Abra rock shelters) and for his help in reading the first draft of the manuscript, for which I wish to thank also Dr. T. A. Wijmstra; Mr. G. Moldus (analysis of the recent samples and section B22); Mr. J. Oosting (drawings); Mr. H. J. Koerts Meijer (drawings and plates); Mrs. T. van Leeuwen nde Voskuilen (preparation of samples); Miss E. Romijn and Mrs. M. J. Content nde Besselink (typing of the manuscript); and last but not least my family for their patience and the friends who took care of my children and so made it possible for me to finish the publication of my laboratory work from 1969 to 1972.

REFERENCES Cuatrecasas, J., 1958. Aspectos de la vegetaci6n natural de Colombia. Rev. Acad. Col. Cienc., Ex. Fis. Nat., 10(40): 221--263. Correal, G., Vander H a m m e n , T. and Lerman, J. C., 1970. Artefactos liticosde abrigos rocosos en El Abra, Colombia, South America. Rev. Colomb. Antropol, 14: 11--46. Espinal, L. S. and Montenegro, E., 1963. Formaciones Vegetales de Colombia. Inst. Geogr. "Augustin Codazzi", Bogota, 201 pp. Hurt, 'W. R., Van der H a m m e n , T. and Correal, G., 1972. Preceramic sequences in the El Abra rock shelters, Colombia. Science, 175: 1106--1108. Hurt, W. R., Van der H a m m e n , T. and Correal, G., 1976. The El Abra rock shelters, Sabana de Bogot~ Colombia, South America. Indiana Univ. Mus. Occas. Pap. Monogr., 2. (The Quaternary of Colombia). Kaufman, P. B., Petering, L. B. and Smith, J. G., 1970. Ultrastructural development of

109

Cork -- ~ilica Cell Pairs in Avena Internodal Epidermis, Botanical Gazette, 131, 3: 173--185. Kendall, R. L., 1969. An ecological history of Lake Victoria Basin. Ecol. Monogr., 39: 121--176. Metcalfe, C. R., 1960. A n a t o m y of the Monocotyledons, 1. Clarendon Press, Oxford. Netolitsky, F., 1929. Die Kieselk~rper: Kieselk~rper der einzelnen Pflanzenfamilien. Die Inhaltsstoffe der Epidermiszellen II Kieselsa~re. Linsbauer's Handbuch der Pflanzenanatomie, pp. 16--19 and 157--161. Parry, D. W. and Smithson, F., 1957. Detection of opaline silica in grass leaves. Nature, 179: 975. Parry, D. W. and Smithson, F., 1964. Types'of opaline silica depositions in the leaves of British grasses. Ann. Bot., 28: 169--185. Smithson, F., 1956a. Silica particles in some British soils. J. Soil Sci., 7(1): 122. Smithson, F., 1956b. Plant opal in soil. Nature, 178: 107. Smithson, F., 1968. Grass opal in British soils. J. Soil, Sci., 9(1): 148. Van der Hammen, T., 1974. The Pleistocene changes of vegetation and climate in tropical South America. J. Biogeogr., 1: 3--26. Van der Hammen, T., 1978. Stratigraphy and environments of the Upper Quaternary of the E1 Abra corridor and rock shelters (Colombia). Palaeogeogr., Palaeoclimatol., Palaeoecol., 25: 111--162. Van der Hammen, T. and Gonzalez, E., 1960. Upper Pleistocene and Holocene climate and vegetation of the "Sabana de Bogot~ (Colombia, South America). Leidse Geol. Meded., 25: 26--315. Van der Hammen, T., Wijmstra, T. A. and Zagwijn, W. H., 1971. The floral record of the Late Cenozoic of Europe. In: K. K. Turekian (Editor), The Late Cenozoic Glacial Ages. Yale University Press, New Haven, Conn., pp. 391--424. Van der Hammen, T., Werner, J. H. and Van Dommelen, H., 1973. Palynological record of the upheaval of the Northern Andes: a study of the Pliocene and Lower Quaternary of the Colombian Eastern Cordillera and the early evolution of its high-Andean biota. Rev. Palaeo-bot. Palynol., 1 6 : 1 - - 2 2 (The Quaternary of Colombia, 2). Van Geel, B. and Van der Hammen, T., 1973. Upper Quaternary vegetational and climatic sequence of the Fuquene area (Eastern Cordillera, Colombia). Palaeogeogr., Palaeoclimatol., Palaeoecol., 14: 9--92. (The Quaternary of Colombia, 1.) Van Geel, B. and Van der Hammen, T., 1978. Zygnemataceae in Quaternary Colombian sediments. Rev. Palaeobot. Palynol., 25(5): 377--392. Vogel, J. C. and Lerman, J. C., 1969. Radiocarbon dates VII. Radiocarbon, 2(2). Wijmstra, T. A. and De Vin, E., 1971. The new Dinkel canal section. In: T. van der Hammen and T. A. Wijmstra (Editors), The Upper Quaternary of the Dinkel Valley. Meded. Rijks Geol. Dienst, New Ser., 22: 55--214. Ijzereef, G. F., 1978. The faunal remains from the E1 Abra rock shelters (Colombia). Palaeogeogr., Palaeoclimatol., Palaeoecol., 25: 163--177.