Late Holocene history of savanna gallery forest from Carimagua area, Colombia

Review of Palaeobotany and Palynology 111 (2000) 295–308www.elsevier.nl/locate/revpalbo

Late Holocene history of savanna gallery forest fromCarimagua area, Colombia

J.C. Berrio a, H. Hooghiemstra a,*, H. Behling a,1, K. van der Borg ba University of Amsterdam, Hugo de Vries-Laboratory 2, Dept. of Palynology and Paleo/Actuo-ecology,

Kruislaan 318 SM Amsterdam, The Netherlandsb R.J. van der Graaff Laboratory, Utrecht University, Princetonplein 5, 3508 TA Utrecht, The Netherlands

Received 27 July 1999; accepted for publication 6 April 2000

Abstract

The pollen record of a 65 cm long core Laguna Carimagua–Bosque (4°04∞N, 70°13∞W ) shows the late Holoceneenvironmental history from a lake located within the gallery forest of the savannas of the Llanos Orientales ofColombia. Nine AMS radiocarbon dates of the organic deposits show that the core represents the period from ca.1300 14C yr BP to the present. The lake evolved from an active drainage system.

During the period from ca. 1300 to 875 14C yr BP (zone CMB-Ia), Mauritia-dominated swamp and gallery forestwas present, dominated by Cecropia, and later also Acalypha and Alchornea. From 875 to 700 14C yr BP (zoneCMB-Ib), the lake was completely surrounded by gallery forest. Mauritiella and Cecropia occurred around the lake.Cecropia pioneer forest reached its greatest abundance and became gradually replaced by a more species-rich galleryforest, including Acalypha, Alchornea, Euterpe/Geonoma, Moraceae/Urticaceae, Piperaceae, and Virola. From 700 to125 14C yr BP (zone CMB-II ), Cecropia lost its dominant role, and Mauritiella palms became more frequent. Themain vegetation categories were swamp forest, gallery forest, understory elements, savanna shrubs and trees, andgrass savanna. From 125 14C yr BP to recent (zone CMB-III ), the plant diversity in the gallery forest became highest,Mauritiella became very abundant, and among the savanna elements, woody Didymopanax increased.

Comparison of four pollen records from savanna sites shows that pollen of savanna vegetation is markedlyunderrepresented in lake sediments when the lake lies within the gallery forest. As most of the drainage system of asavanna is hidden by gallery forest, we also expect a significant underrepresentation of the savanna ecosystem inriver-transported pollen assemblages. © 2000 Elsevier Science B.V. All rights reserved.

Keywords: Colombia; gallery forest; Laguna Carimagua; late Holocene; Llanos Orientales; paleoecology; pollen representation;savanna; vegetation history

* Corresponding author. Tel.: +31-20-5257857; fax: +31-20-25257662.E-mail address: [email protected] (H. Hooghiemstra)1 Present address: Centre for Tropical Marine Ecology, Fahrenheitstraße 1, 28359 Bremen, Germany.2 Participating in The Netherlands Centre for Geo-ecological Research (ICG), and the Institute of Biodiversity and Ecosystem

Dynamics (IBED)

0034-6667/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved.PII: S0034-6667 ( 00 ) 00030-0

296 J.C. Berrio et al. / Review of Palaeobotany and Palynology 111 (2000) 295–308

1. Introduction available to wet the soil during the dry season(Monasterio and Sarmiento, 1971).

The present paper aims to show the environ-For a long time, the history of the savannasnorth of the Amazon basin was poorly known. mental history of the savanna ecosystem based on

a pollen record of Laguna Carimagua–Bosque andWijmstra and Van der Hammen (1966) publishedthe first pollen records of this ecosystem from to compare the vegetation development with other

pollen records from the Colombian LlanosLaguna de Agua Sucia in the Colombian LlanosOrientales and from the Rupuni savannas in Orientales. In particular, we aim to compare

records from two lakes at very close distanceGuyana (Wijmstra, 1971). A synthesis of thehistory of the neotropical savannas was published within a significantly different setting. Laguna

Carimagua–Bosque (this paper) today is locatedby Van der Hammen (1983). More recently, savan-nas in the Gran Sabana of Venezuela were studied in the gallery forest, whereas Laguna Carimagua

(Behling and Hooghiemstra, 1999) at a distanceby Rull (1992, 1999). During explorations in 1996and 1997 by H. Hooghiemstra and H. Behling in of 3 km is surrounded by open savanna vegetation,

and only a few patches of Mauritia-dominatedthe Colombian provinces of Meta and Vichada,sediment cores were collected from over 10 new palm forest reach the lake shore. Finally, we infer

conclusions of more general value for the recon-locations in the savannas of the Llanos Orientales.Results from these cores have been published struction of savanna environments from pollen

records.recently: Lagunas Angel and Sardinas (Behlingand Hooghiemstra, 1998), Lagunas El Pinal andCarimagua (Behling and Hooghiemstra, 1999).Other pollen records concern Laguna Loma Linda 2. Environmental setting(Behling and Hooghiemstra, submitted for publi-cation), Laguna Las Margaritas (Wille et al., in Laguna Carimagua–Bosque (4°04∞N, 70°13∞W,

Province of Meta; 180 m altitude) lies in the Llanospreparation), and Lagunas Mozambique andChenevo (Berrio et al., in preparation). An over- Orientales and is part of a series of studied sites

along a 500 km long west-to-east transect (Fig. 1).view and comparison of the savanna ecosystemsnorth and south of the Amazonian rain forest The distance to the lagunas Carimagua and El

Pinal is ca. 3 and 22 km, respectively (Behling andwere published by Behling and Hooghiemstra (inpress). Hooghiemstra, 1998). Other lakes studied palyno-

logically were lagunas Sardinas and Angel (BehlingAccording to Sarmiento and Monasterio(1975), savannas are stable ecosystems, floristically and Hooghiemstra, 1998), both more than 100 km

from the site Carimagua–Bosque. The distance todominated by grasses but also including herbs,shrubs, and several characteristic trees (such as the foot of the Eastern Cordillera at Villavicencio

is ca. 270 km. Laguna Carimagua–Bosque is aCuratella and Byrsonima). Laguna El Pinal showsthe longest record and reaches as far back in time shallow lake, maximally 1–2 m deep with a diame-

ter of ca. 150 m (Fig. 2). It is located in anas the last glacial maximum (LGM). It showsshifts in the contribution of open grass savanna, extended slightly undulating savanna landscape.

The maximum difference in elevation is ca. 50–wooded savanna, and gallery forest, but the floralcomposition of these types seems to be relatively 80 m, but in the surroundings of the Carimagua–

Bosque site, it is 10–20 m. The origin of thestable (Behling and Hooghiemstra, 1999).However, Beard (1953), Blydestein (1967), and depression of Laguna Carimagua–Bosque is a

drainage system that is still present in the area butHills and Randall (1968) have supported the ideathat the extension of savanna and savanna-forest only active today at some distance. At the shores

of Laguna Carimagua, an agricultural researchis edaphically determined as a consequence of poordrainage in combination with strong seasonal pre- station is located, called ‘Centro de Investigaciones

para la Altillanura, Carimagua’.cipitation. Moreover, gallery forest can occur inthe lower parts of the drainage system if water is The climate of the Llanos Orientales is seasonal

297J.C. Berrio et al. / Review of Palaeobotany and Palynology 111 (2000) 295–308

Fig. 1. Map showing the geographical location of sites Laguna Carimagua and Laguna Carimagua–Bosque in the Llanos Orientalesof Colombia. Other pollen sites mentioned in the text are also indicated on the map.

with a rainy period between April and October Hedyosmum bonplandianum, Trichantera gigantea,Miconia scorpioides, Caraipa llanorum, Alchornea(Lowy and Rangel, 1993). In the Llanos

Orientales, there is a precipitation gradient from triplinervia, and species belonging to the generaProtium, Vochysia, and Cecropia. Hedyosmum bon-1200–2000 mm/yr in the northern region near the

Venezuelan border to about 2000–2500 mm/yr in plandianum is of special interest as, in palynologicalstudies, this genus is often regarded as a montanethe southern and southwestern parts of the

savanna area. Based on the records of the tree. However, these shrubs or small trees occurin Colombia at elevations of 100–2350 m in openCarimagua Reseach Station, the local mean annual

precipitation is 1860 mm. The mean annual tem- habitats along stream edges, ravines, and clearedslopes, as well as in wet montane forest (Todzia,perature in the study area is 26–27°C, with less

than 3°C variation between monthly means. 1988).Soils in the surrounding area are called ‘soils ofThe floral composition of the Colombian savan-

nas was studied by Blydestein (1967), Huber the esteros’, characterized by a fine to mediumtexture, with moderately high organic matter and(1987), Cuatrecasas (1989), and Pinto-Escobar

(1993). According to Rangel et al. (1995), the a very poor drainage capacity (Goosen, 1971).The same author pointed out that soils fromstudy area is characterized by ‘bosque tipo mor-

ichal’, and the main taxa are Mauritia flexuosa, Carimagua area are structurally weak and that


Fig. 2. Photograph of Laguna Carimagua–Bosque (photograph: H. Hooghiemstra).

water infiltrating the soils causes translocation of samples were mounted in a glycerin–gelatinmedium. A minimum of 300 pollen grains frommineral particles.terrestrial taxa were counted. Carbonized particleswere not counted because during the heavy liquidseparation, many charcoal particles were elimi-3. Methodsnated. Pollen identification was based on morpho-logical descriptions by Absy (1979), HooghiemstraThe 65 cm long core Carimagua–Bosque was

collected by a modified Livinstone piston corer (1984), Roubick and Moreno (1991), Behling(1993), Herrera and Urrego (1996), and thefrom a wooden platform, which was fixed on two

inflatable rubber boats. The core was transported modern reference collection of pollen and sporesat the Hugo de Vries Laboratory. Six bulk samplesto the Hugo de Vries Laboratory and stored in a

cold room (4°C) before sediment samples were followed by three additional samples of selectedplant material were dated by accelerator masscollected. For pollen analysis, samples of 1 cm3

were taken at 2.5 cm intervals along the core. Prior spectrometry (AMS 14C) at the Van der GraaffLaboratory, University of Utrecht (Van der Borgto processing, one tablet of exotic Lycopodium

spores was added to each sample for calculation et al., 1987).Pollen grains were classified into ecologicalof the pollen concentration and pollen influx

values. All samples were prepared using the stan- groups following previous palynological studies ofthis area (Behling and Hooghiemstra, 1998, 1999):dard pre-treatment techniques including sodium

pyrophosphate, acetolysis, and heavy liquid sepa- (1) forest, gallery forest, shrubs and trees; (2)savanna shrubs and trees; (3) savanna herbs; (4)ration by bromoform (Faegri et al., 1989). Pollen


Andean forest taxa; (5) aquatics; (6) unknown; Obviously, the ages obtained for the bulk(7) fern spores; (8) fungal spores; (9) samples do not show a distinct time marker butBotryococcus, and (10) other algal remains. The rather the age of a carbon fraction, which seemspollen and spore data are presented as a percentage to have penetrated the whole core. The ages of theof the pollen sum, consisting of the groups 1–4 plant remains show a rapid increase in age betweenthat represent the regional vegetation elements and 35 and 65 cm which corresponds to a low sedimentare in competition with each other. For calculation accumulation rate during this period. The lowof the pollen and spore frequencies, graphing of pollen concentration values in this interval (Fig. 4)the pollen diagrams, and cluster analysis, the soft- support this interpretation. There is no indicationware packages TILIA, TILIAGRAPH, and from the pollen diagram or from the observedCONISS were used (Grimm, 1987). The pollen changes in the sedimentary column that a hiatusdiagrams include individual records of the most is present.important pollen and spore taxa (Fig. 3), as well We discuss here some aspects relevant to theas records of the sums of the ecological groups, problems in the chronology. Goosen (1971)downcore changes of the pollen concentration and reports soil instability as an important characteris-pollen influx values, and a cluster analysis dendro- tic of this area, allowing the transport of mineral-gram used for zonation (Fig. 4). ogical soil particles: organic material of a narrow

age range could have vertically migrated throughthe soil column. However, in such a case, also,

4. Results pollen grains might have migrated vertically.However, the vegetation changes illustrated in the

4.1. Stratigraphy pollen diagram show a plausible development, alsowhen compared to other pollen records from the

The sediment core consists of gray sand from Llanos Orientales. Therefore, vertical migration of65 to 60 cm core depth. From 60 to 52.5 cm depth, pollen grains is very unlikely. It remains doubtfula mixture of sand and muddy dark gray clay is as to whether migration of only mineralogical soilpresent. The interval from 52.5 to 35 cm contains

particles is possible. A second explanation is thathomogeneous muddy dark gray clay, rich inthe total sediment sequence has been deposited inorganic material. Macro remains such as leavesa very short time around 800–850 14C yr BP andand other plant debris is observed. The uppermostthat the ages on bulk material at 24 and 50 cm35 cm consist of homogeneous and compact darkcore depth are, for some reason, too old. However,gray clay.for the same reason as that given for the firsthypothesis, this is very unlikely.4.2. Chronology and zonation

Notwithstanding the uncertainty in the timecontrol, we conclude that this core representsTime control of the 65 cm long core wasvegetation conditions of the latest Holocene. Forattempted with radiocarbon analysis of six bulkthis reason, the comparison of the vegetation devel-samples; the organic residue was taken for analysisopment shown by core Carimagua–Bosque may(Table 1). However, the dates obtained for the sixbe compared to the uppermost parts of the coresbulk samples did not show the expected increaseCarimagua, El Pinal, and Sardinas, which reflectof age with depth but rather showed a constantthe same time interval.age with an average of about 940 14C yr BP from

In conclusion, we accepted the ages at 5–9, 35–the top to 45 cm depth, and an average of39, 44, 55, and 60–64 cm to calculate the age of1030 14C yr BP in case we consider the top 55 cmthe periods represented by the pollen zones. Weof the core. For that reason, three samples ofrejected the samples at 2.5, 9, 24, and 50 cm (theseidentified plant remains were used for radiocarbonages are shown in brackets in Fig. 4). The pollenanalysis and showed ages from modern for the topzones, based on significant changes in representa-40 cm of sediments to 1090±60 14C yr BP (=

cal yr BP 1060–930) for the 60 cm core depth. tion and/or changes in the composition of the


Fig

.3.

Pol

len

perc

enta

gedi

agra

mof

65cm

core

Lag

una

Car

imag

ua–B

osqu

e,lo

cate

dat

180

mal

titu

dein

the

Lla

nos

Ori

enta

les

ofC

olom

bia.

Ase

lect

ion

ofth

em

ost

freq

uent

indi

vidu

alta

xa,

arra

nged

into

ecol

ogic

algr

oups

,is

show

n.


Fig

.4.P

olle

npe

rcen

tage

diag

ram

(sum

sof

ecol

ogic

algr

oups

)of

65cm

core

Lag

una

Car

imag

ua–B

osqu

e,lo

cate

dat

180

mal

titu

dein

the

Lla

nos

Ori

enta

les

ofC

olom

bia.


Table 1List of AMS 14C dates of core Carimagua–Bosque

Laboratory number Depth (cm) 14C yr BP 13C/12C r. Calendar age (cal yr BP) Material dated

UtC-8008 (Col-1162) 2.5 854±41 −27.6 784–703 Bulk materialUtC-9263 (Col-1197) 5–9 −1179±33 −30.9 Modern Plant remainsUtC-8009 (Col-1163) 9 780±46 −27.8 723–664 Bulk materialUtC-8034 (Col-1164) 24 1260±60 −28.2 1270–1077 Bulk materialUtC-9264 (Col-1198) 35–39 161±44 −31.7 Modern Plant remainsUtC-8010 (Col-1165) 44 875±49 −29.9 891–870 Bulk material

826–814792–723

UtC-8011 (Col-1166) 50 1603±46 −28.5 1535–1411 Bulk materialUtC-8012 (Col-1167) 55 837±39 −30.2 770–763 Bulk materialUtC-9265 (Col-1199) 60–64 1090±60 −30.1 1058–946 Plant remains

943–933

pollen spectra, represent the following periods: In zone CMB-II (42–21 cm), the contributionof the ecological groups of forest, gallery forest,zone CMB-Ia: 1300–875 14C yr BP, CMB-Ib:

875–700 14C yr BP, zone CMB-II 700–125 14 and shrubs and trees decreases slowly to 78–87%,whereas savanna herbs increase gradually. TheC yr BP, and zone CMB-III 125 14C yr BP to

recent. record of Mauritia-type pollen continues withvalues similar to the previous zone, but in theuppermost part, values vary from 15 to 33%.4.3. Concise description of the pollen diagramMauritiella-type pollen starts to be present with4% in this zone. Acalypha is present with 14–28%Zone CMB-Ia (65–53 cm) is characterised by a

high representation of pollen of Mauritia-type in this zone. Alchornea shows a stable contributionof 18%. Compared to the previous zone Cecropia(51%) at the base of the zone. Other important

taxa are Acalypha and Alchornea (5–15%), and decreases markedly to 2–11%. Melastomataceae/Combretaceae, and Moraceae/Urticaceae con-Cecropia reaches its maximum values (45%) at the

end of the zone. Melastomataceae/Combretaceae, stantly show 2%. Didymopanax is poorly repre-sented but rises at the uppermost part to 4%. TheMoraceae/Urticaceae, and Fabales-I reach only

2%. Savanna shrubs and trees are poorly repre- savanna herbs Poaceae (15%), Asteraceae (4%),and Cyperaceae (5%) show similar values as insented, but Didymopanax is present with 2%.

Savanna herbs are common and maximally repre- zone CMB-Ia. Aquatics increase from 3 to 6%.In zone CMB-III (21–0 cm), gallery forest con-sented with 18% Poaceae, 10% Asteraceae, and

5% Cyperaceae. Aquatics reach 5–10%. tinues to decrease to reach 77–65%, the Mauritiatype decreases to 10%, and the Euterpe/GeonomaAt the beginning of zone CMB-Ib (53–42 cm),

pollen grains of Euterpe/Geonoma-type become type decreases to 1%. Compared to the previouszone, Cecropia is poorly represented (3%). Melas-frequent (2%). Mauritia-type (21%), Acalypha

(22%), Alchornea (18%), and Cecropia (25–35%) tomataceae/Combretaceae, Moraceae/Urticaceae,and the Macrolobium type show a stable represen-are the most important taxa of the gallery forest.

Byrsonima and Didymopanax are poorly repre- tation. Less frequent taxa are Vismia (3%), andCroton, Fabales II, Psychotria, and Protium (allsented in this zone. Savanna herbs decreased and

are mainly represented by Poaceae (5–8%), 1–2%). Savanna shrubs and trees (mainlyDidymopanax) show the highest representation inAsteraceae (1%), and Cyperaceae (1%). Aquatics,

such as Ludwigia, Polygonum and Sagittaria show this zone (7%). Savanna herbs rise again to higherpercentages (30%) in the uppermost part of thevalues <1%.


zone: Poaceae (20%), Asteraceae (5%), and late dark clay rich in organic material. A significantcontributor to the local production of organicCyperaceae (4%). Other savanna herbs, such as

Borreria, Lamiaceae, and Spermacoce, have 1% as material might have been the swamp forest withCecropia and Mauritiella growing in the shallowa maximum. Aquatics are continuously repre-

sented by Ludwigia and Sagittaria (2–3%), and by parts of the lake. The pioneer forest of Cecropiaexperienced its maximum extension and graduallyXyris and Polygonum (maximally 1.5%).became replaced by ‘mature’ forest, mainly con-sisting of Acalypha, Alchornea, Euterpe/Geonoma,Moraceae/Urticaceae, Piperaceae, and Virola.5. Environmental development based on

Carimagua–Bosque record Pollen influx values reach their highest values inthe total record, indicating that the transition fromopen pioneer forest to more closed gallery forestLake Carimagua–Bosque came into existence

during the late Holocene, and the lake sediments is a favorable condition for the lake to collect amaximum of pollen.document the development of the dynamic balance

between open savanna and gallery forest during During the period from 700 to 125 14C yr BP(zone CMB-II ), forest development reached a thirdthe last 1300 14C yr BP. The interval with sand,

gradually changing to sandy clay (65–52.5 cm) phase in which the pioneer tree taxon Cecropialost its dominant role, and Mauritiella palmsdated from 1300–850 14C yr BP (extrapolated

ages), reflects the initial phase of the lake that became a new arboreal taxon competing withforest trees. The slightly undulating landscape sur-possibly developed from an abandoned meander.

The drainage system possibly included swampy rounding the lake was now covered by differentvegetation types: (a) swamp forest dominated byareas, where Mauritia-dominated swamp forest

(‘morichal’) was common, as reflected by the high Mauritia and Mauritiella on poorly drained soils;(b) gallery forest dominated by Acalypha,representation of this palm pollen. The forest

development started with Cecropia, a common Alchornea and, to a lesser degree, Anacardiaceae,and Macrolobium (on better drained soils), withcolonizer, whereas the trees Acalypha and

Alchornea became more frequent in the later part the light-demanding tree taxon Cecropia as anindicator for internal forest dynamics and possiblyof this initial 450 year long period. Light-demand-

ing ferns represented by the records of ‘monolete also reflecting Cecropia stands at larger distancewhere the development of the forest may havepsilate spores’ are common in the understory of

pioneer forests. While local gallery forest taxa been at a different stage; (c) elements of theunderstory partly reflected in the records of theincreased, the Poaceae and Asteraceae taxa

decreased, documenting a reduction in open Melastomataceae/Combretaceae, and Moraceae/Urticaceae; (d) savanna shrubs mainlysavanna. Aquatic taxa, such as Xyris, Ludwigia,

and Sagittaria were common from the start of the Didymopanax and less frequent Byrsonima; and (e)savanna herbs dominated by Poaceae.record. Also, the presence of the alga Botryococcus

indicates that stagnant and shallow water was The youngest part of the reconstruction dealswith the period of the last 125 14C yr (zone CMB-present from the start of the record.

During the period from 875 to 700 14C yr BP III ). Although the last decades are probably notrepresented as the uppermost few centimeters of(zone CMB-Ib), the lake was apparently com-

pletely surrounded by gallery forest, and it is unconsolidated sediments were lost during thecoring. The pollen record reflects gallery forestpossible that there was no direct contact any more

with the open savanna in the surroundings, as dominated by Acalypha, and Alchornea, with(partly in the understory) Melastomataceae/evidenced by the low contribution of the two

groups of savanna taxa. Also, the lithological Combretaceae, Moraceae/Urticaceae, Macrolobium,Croton, Fabales, Psychotria-type, Protium, Vismia,column shows that the (aeolian) supply of sandy

material to the lake had ceased, probably pre- Anacardiaceae, and Alibertia. Cecropia is part ofthe gallery forest. In the swamp forest, Mauritiellavented by dense forest. The lake started to accumu-


became common and may have partly replaced vegetation, and in fact 100% in contact with thesurrounding grass savanna (Behling andMauritia. In the savanna, woody taxa, primarily

Didymopanax, became more frequent. Human Hooghiemstra, 1999).The location of these sites is shown in Fig. 1.impact may have contributed to this change,

including the increase in Mauritiella, which is The records of the lakes Carimagua andCarimagua–Bosque play a special part in thisfrequently used by man. Conditions were wetter

during this period on average. Aquatic vegetation comparison as they are located only some 3 kmapart and, considered on a landscape scale, arenever became important, which must be related to

the topography of the lake; because the lake shores located in the same drainage system. As shown bythe radiocarbon-dated pollen records, Lagunaare narrow once they are surrounded by forest,

there is not much shallow area for extensive Carimagua was abandoned in the early Holocene,and Laguna Carimagua–Bosque only in the lateaquatic vegetation.Holocene, stressing that the drainage system wasdynamic. The Laguna Carimagua record starts at8270 14C yr BP and shows a landscape dominated6. Comparison of savanna pollen records:

implications for savanna reconstructions by grass savanna with only a few woody savannataxa (Curatella and Byrsonima) and a low occur-rence of forest and/or gallery forest along theIncreased percentages of Cecropia recorded at

Carimagua–Bosque can be interpreted as evidence rivers. Up to 3850 14C yr BP, there are only minorchanges in the vegetation of the savanna eco-of increased human impact, as suggested also for

the Laguna Carimagua record (Fig. 5). This sug- system. A marked change occurred around theinterpolated age of 3850 14C yr BP, primarily bygests that either parts of the drainage system

nearby are still changing their routes and continue the sudden high abundance of the palmMauritiella, but also by Mauritia, Alchornea andto provide virgin soil to be colonized by Cecropia

forest, or there is disturbance of the vegetation Cecropia that became more frequent. This changeis related to wetter conditions, i.e. an increase inby fire.

In the following section, we compare arboreal precipitation and/or a shorter dry season. Behlingand Hooghiemstra (1999) tentatively suggestedvs. savanna pollen percentages in four pollen

records of the Llanos Orientales that include the that the incease in the pioneer and disturbanceindicator, Cecropia, in the record from Carimagualast 1500 years. During this period, the arboreal

vs. savanna pollen ratio within the records shows could be indicative of an increase in human impacton the vegetation after 3850 14C yr BP. Mauritiellaminimal fluctuations in comparison with the

difference of this ratio between the cores. Based today forms dense stands around lake Carimagua,which are reflected by the curve of Mauritiella inon these stable ratios in the youngest parts of the

pollen records, we consider the present-day setting the uppermost part of its record. As suggested bythe radiocarbon dates, the last thousand years areas representative:

(1) pollen record of Laguna Carimagua–Bosque missing in this record. In the pollen record fromLaguna Carimagua–Bosque (zone CMB-III ), thetoday 100% surrounded by dense gallery forest

(this paper), representation of Mauritia (10–20%) is higher thanthat of Mauritiella (2–10%) and reflects present-(2) pollen record of Laguna Carimagua >50%

surrounded by grass savanna (Behling and day conditions. We conclude that both pollenrecords do overlap for the period of 1300–Hooghiemstra, 1999),

(3) pollen record of Laguna Sardinas only 1000 14C yr BP, which means that zone CMB-Ia(this paper) is coeval with the uppermost part ofsurrounded by a narrow zone with relatively open

morichal vegetation (Behling and Hooghiemstra, zone CAR-II (Behling and Hooghiemstra, 1999).Comparing the ratio arboreal pollen vs. savanna1998), and

(4) pollen record of Laguna El Pinal surrounded pollen in these four records for the youngest periodnot older than the last 1300 years and comparingby a narrow zone of very open Mauritia palm


Fig

.5.

Com

pari

son

ofm

ain

polle

npe

rcen

tage

diag

ram

sof

Lag

una

Car

imag

ua(p

erio

dof

8000

–135

014

Cyr

BP

;B

ehlin

gan

dH

oogh

iem

stra

,19

99)

and

Lag

una

Car

imag

ua–B

osqu

e(1

30014

Cyr

BP

tore

cent

;th

ispa

per)

.Si

tes

are

loca

ted

ca.

3km

apar

tat

180

mal

titu

dein

the

Lla

nos

Ori

enta

les

ofC

olom

bia.


this ratio in the pollen assemblages with the ratio Carimagua–Bosque prevents pollen grains fromsurrounding savanna vegetation to be included inin the present-day vegetation surrounding the

lakes, there is a significant difference between the lake sediments. This is an important observa-tion as it makes clear that the signal of grass pollenthese sites:

(1) The record of Laguna Carimagua–Bosque, in a record from a savanna ecosystem is maskedto a significant degree when gallery forest sur-today completely surrounded by dense gallery

forest up to ca. 2 km distance from the lake, shows rounds the lake. This has important implicationsas most of the drainage system in a savannathe highest percentages of arboreal pollen: 70–

90%. This lake is a closed system about 150 m in ecosystem is hidden by gallery forest, a fact thatcan be easily seen from an aircraft and from aerialdiameter, 1–2 m deep, and located at 180 m eleva-

tion at 4°04∞N, 70°13∞W. photographs. As shown in the next paragraph, thisaspect, which is observed thanks to the fact that(2) The record of Laguna Carimagua, today

>50% surrounded by grass savanna and with we can compare pollen assemblages from twoneighboring lakes in different environmental set-some patches of forest of several hectares reaching

the border of the lake, shows 30–50% arboreal tings, potentially has significant implications forthe interpretation of pollen records from sitespollen. This lake is a closed system about 3 km

long and 1.5 km wide, 0.8 m deep, and located at where pollen assemblages are mainly supplied bywater currents.180 m elevation at 4°04∞N, 70°14∞W.

(3) The record of Laguna Sardinas, today sur-rounded by a narrow zone of open palm forest(morichal ) and shrub up to some 50–100 m dis- 7. Potential implications for ice-age history of

Amazoniatance from the border of the lake, shows 60–65%arboreal pollen. This lake is a closed system about800 m in diameter, 1.1 m deep, and located at 80 m In the discussion of the ice-age Amazon, the

central question is whether present-day rain forestelevation at 4°58∞N, 69°28∞W.(4) The record of Laguna El Pinal, surrounded had been replaced significantly by dry forest, decid-

uous forest, and wooded and open savanna-likeby a ca. 50 m wide zone with sparsely distributedMauritia palms, and in fact almost completely vegetation during the last glacial maximum (=

LGM ) (e.g. Colinvaux, 1996, 1997; Colinvauxsurrounded by open grass savanna, shows 7–15%arboreal pollen. This lake is a closed system about et al., 1996, 2000; Hooghiemstra, 1997;

Hooghiemstra and Van der Hammen, 1998; Van1200 m long and 300 m wide, 0.7 m deep, andlocated at 180 m elevation at 4°08∞N, 70°23∞W. der Hammen and Hooghiemstra, 2000). As ter-

restrial records from the Amazonia are scarce,We observe that the contribution of arborealpollen increases with the density of the forest attention was focused on pollen records from

Amazon Fan cores in order to reconstruct thearound the lake. Apparently, the more lakes aresurrounded by gallery forest, the more pollen large-scale vegetation history of the Amazonian

rain forest. Haberle (1997), Hoorn (1997), andspectra from these lake sediments are unable todocument savannas in the large surroundings. In Haberle and Maslin (1999), showed that the

Amazon Fan pollen spectra did not evidence athis respect, it is important to observe that theLaguna Carimagua record shows 45–55% of significantly increased contribution of savanna

taxa during the LGM. Colinvaux and co-authorssavanna herb taxa, whereas Laguna Carimagua–Bosque shows only 8–20(30)%. It is very plausible interpreted this pollen signal as evidence that

Amazonian forests were not extensively replacedthat this difference is directly related to the differ-ence in environmental setting of both lakes: by the above mentiond drier vegetation types, and

thereby contradicted the ‘forest refugia’ hypothesisLaguna Carimagua is surrounded by open savannawith over 50%, and grass pollen can easily be (see references mentioned above). According to

Hooghiemstra (1997), Hoorn (1997), andsupplied to the lake sediments. On the contrary,the presence of gallery forest around Laguna Hooghiemstra and Van der Hammen (1998), such


pollen curves without any major oscillation reflect Referencesredeposited and mixed sediments, originating frommany places along the Amazon River and finally Absy, M.L., 1979. A palynological study of Holocene sediments

in the Amazon basin. Ph.D. thesis, University of Amster-deposited offshore.dam, Amsterdam, 1–105.For the first time, our data show evidence that

Beard, J.S., 1953. The savanna vegetation of northern tropicalthe presence of a savanna ecosystem is masked in America. Ecological Monographs 23, 149–215.case gallery forest is present, preventing pollen Behling, H., 1993. Untersuchungen zur spatpleistozanen und

holozanen Vegetations- und Klimageschichte der tropischenfrom entering the water body. Today, most of theKustenwalder und der Araukarienwalder in Santa Catarinadrainage system of a savanna is hidden by gallery(Sudbrasilien). Dissertationes Botanicae J. Cramer, Vaduzforest. As a consequence, we expect significant206, 1–149.

underrepresentation of the savanna ecosystem in Behling, H., Hooghiemstra, H., 1998. Late Quaternary palaeo-river-transported pollen assemblages, and we add ecology and palaeoclimatology from pollen records of the

savannas of the Llanos Orientales in Colombia. Palaeogeog-here new evidence to the earlier argumentionraphy, Palaeoclimatology, Palaeoecology 139, 251–267.(Hooghiemstra, 1997) as to why an absence of

Behling, H., Hooghiemstra, H., 1999. Environmental history ofincreased percentages of savanna elements in off-the Colombian savannas of the Llanos Orientales since the

shore Amazon Fan cores during the LGM is Last Glacial Maximum from lake records El Pinal and Cari-unsufficient evidence to conclude that at LGM magua. Journal of Paleolimnology 21, 461–476.

Behling, H., Hooghiemstra, H., 2000. Neotropical savannatime Amazonian rain forest was not significantlyenvironments in space and time: Late Quaternary interhemi-replaced by savanna-like vegetation. Additionalspheric comparisons. In: Markgraf, V. (Ed.), Inter-hemi-pollen records and more systematic studies on thespheric Climatic Linkages (Present and Past Inter-

relationship between pollen representation and hemispheric Climate Linkages in the Americas and theirvegetation cover in Amazonian lowlands are Social Effects). Academic Press, New York, in press.

Behling, H., Hooghiemstra, H., 2000. Holocene Amazon rainneeded to improve the argumentation of bothforest-savanna dynamics and climatic implications: high‘schools’ of the ice-age Amazon.resolution pollen record Laguna Loma Linda in easternColombia. Palaeogeography, Palaeoclimatology, Palaeo-ecology. submitted for publication.

Acknowledgements Berrio, J.C., Hooghiemstra, A.A., Behling, H., Van der Borg,K., submitted. Holocene environmental history of the west-ern Colombian savannas of the Llanos Orientales fromWe thank Pedro Botero (Instituto GeograficoLagunas Mozambique and Chenevo. The Holocene Journal‘Augustin Codazzi’, Bogota; at present,of Quaternary Science.

‘Fundacion Tierra Nueva’), and Carlos Botero Blydestein, J., 1967. Tropical savanna vegetation of the Llanos(Villavicencio) for logistic support, valuable assis- of Colombia. Ecology 48, 1–15.

Colinvaux, P.A., 1996. Quaternary environmental history andtance, and good companionship during the fieldforest diversity in the neotropics. In: Jackson, J.B.C.et al.,work. Guido van Reenen (Amsterdam) assisted(Eds.), Evolution and Environment in Tropical America.during part of the field work. Elly Beglinger andThe University of Chicago Press, Chicago, IL, pp. 359–405.

Annemarie Phillip (Amsterdam) are thanked for Colinvaux, P.A., de Oliveira, P.E., Moreno, J.E., Miller, M.C.,preparing the pollen samples. Juan Saldarriaga Bush, M.B., 1996. A long pollen record from lowland Ama-

zonia: forest and cooling in glacial times. Science 274, 85–88.(former director of Tropenbos-Colombia), CarlosColinvaux, P.A., 1997. In: The Ice-age Amazon and the Prob-Rodriguez (present director of Tropenbos-

lem of Diversity. NWO/Huygenslezing, Netherlands Organ-Colombia), and Thomas Van der Hammen (Chia,isation for Scientific Research, The Hague, pp. 7–30.

Colombia) are thanked for continuous support. Colinvaux, P.A., de Oliveira, P.E., Bush, M.B., 2000. Ama-We thank Vera Markgraf and an anonymous zonian and neotropical plant communities on glacial time-

scales: the failure of the aridity and refuge hypotheses. Qua-reviewer for constructive comments on an earlierternary Science Reviews 19, 141–169.draft of this manuscript. This research was sup-

Cuatrecasas, J., 1989. Aspectos de la vegetacion natural deported by the Netherlands Organisation forColombia. Perez-Arbelaezia Jardın Botanico de Bogota

Scientific Research (NWO-GOA project number ‘Jose Celestino Mutis’ 2, 155–283.750.195.10 to H.H./H.B., and WOTRO-DGIS Faegri, K., Kaland, P.E., Krzywinski, K., 1989. Textbook of

Pollen Analysis. fourth ed., Wiley, Chichester, UK.project number WB 84-461 to H.H./J.C.B.).


Goosen, D., 1971. Physiography and soils of the Llanos Orien- de la Diversidad Biotica de Colombia. Convenio Inderena-Universidad Nacional de Colombia (Documento Interno).tales, Colombia. Publicaties van het Fysisch-Geografisch en

Monasterio, M., Sarmiento, G., 1971. Ecologıa de sabanas deBodemkunding Laboratorium van de Universiteit vanAmerica Tropical. I. Analisis macroecologico de los llanosAmsterdam 20, 1–200.de Calabozo, Venezuela. Cuadernos Geograficos 4, 1–126.Grimm, E.C., 1987. CONISS: A Fortran 77 program for strati-

Pinto-Escobar, P., 1993. Vegetacion y flora de Colombia. Edito-graphically constrained cluster analysis by the method ofrial Guadalupe, Bogota. 72 pp.the incremental sum squares. Computer Geoscience 13,

Rangel, J.O., Sanchez, H., Lowy, P., Aguilar, M., Castillo, A.,13–35.1995. Region de la Orinoquia. In: Rangel, J.O. (Ed.),

Haberle, S., 1997. Upper Quaternary vegetation and climateColombia diversidad biotica I. Instituto de Ciencias Natura-

history of the Amazon basin: correlating marine and ter- les. Universidad Nacional de Colombia, Colombia,restrial pollen records. Proceedings Ocean Drilling Program, pp. 239–254.Scientific Results 155, 381–396. Roubick, D.W., Moreno, J.E., 1991. Pollen and spores of Barro

Haberle, S., Maslin, M.A., 1999. Late Quaternary vegetation Colorado Island. In: Missouri Botanical Garden 36, 270 pp.and climate change in the Amazon basin based on a 50,000 Rull, V., 1992. Successional patterns of the Gran Sabana (south-year pollen record from the Amazon Fan, ODP Site 932. eastern Venezuela) vegetation during the last 5000 years andQuaternary Research 51, 27–38. its responses to climatic fluctuations and fire. Journal of

Herrera, L.F., Urrego, L.E., 1996. Atlas de polen de plantas Biogeography 19, 329–338.Rull, V., 1999. A palynological record of a secondary successionutiles y cultivadas de la Amazonia Colombiana (Pollen atlas

after fire in the Gran Sabana, Venezuela. Journal of Quater-of useful and cultivated plants in the Colombian Amazonnary Science 14, 137–152.region) Estudios en la Amazonia Colombiana XI. Tro-

Sarmiento, G., Monasterio, M., 1975. A critical considerationpenbos-Colombia, Bogota. also published in: The Quater-of the environmental conditions associated with the occur-nary of Colombia 24 (ed. T. van der Hammen), Amsterdam.rence of savanna ecosystems in tropical America. In: Golley,Hills, T.L., Randall, E. (Eds.), 1968. The Ecology of the Forest-F.B., Medina, E. (Eds.), Tropical Ecological Systems Trends–Savanna Boundary, McGill University, Savanna Researchin Terrestrial and Aquatic Research. Springer, Berlin,

Series Vol. 13. 1–128.pp. 223–250.

Hooghiemstra, H., 1984. Vegetational and climatic history of Todzia, C.A., 1988. Flora Neotropica, 48. Chloranthaceae:the high plain of Bogota, Colombia: a continuous record of Hedyosmum. The New York Botanical Garden, New York.the last 3.5 million years. Dissertationes Botanicae 139 pp.J. Cramer, Vaduz 79, 1–368. (Also published in: The Van der Borg, K., Alderliesten, A., Harnton, C.M, De Jong,Quaternary of Colombia 10 (ed. T. van der Hammen), A.F., Van Zwol, N.A., 1987. Accelerator mass spectrometryAmsterdam). with 14C and 10Be in Utrecht. Nuclear Instruments Methods

Hooghiemstra, H., 1997. Tropical Rain Forest Versus Savanna: B 29, 143–145.Van der Hammen, T., 1983. The palaeoecology and palaeogeog-Two Sides of a Precious Medal? A Comment. NWO/Huy-

raphy of savannas. In: Bourliere, F. (Ed.), Tropical Savan-genslezing, Netherlands Organisation for Scientificnas. Ecosystems of the World 13. Elsevier, Amsterdam,Research, The Hague.pp. 19–35.Hooghiemstra, H., Van der Hammen, T., 1998. Neogene and

Van der Hammen, T., Hooghiemstra, H., 2000. Neogene andQuaternary development of the neotropical rain forest: theQuaternary history of vegetation, climate, and plant diver-forest refugia hypothesis, and a literature overview. Earth-sity in Amazonia. Quaternary Science Reviews 15, 725–742.Science Reviews 44, 147–183.

Wijmstra, T.A., Van der Hammen, T., 1966. Palynological dataHoorn, C., 1997. Palynology of the Pleistocene glacial/intergla-

on the history of tropical savannas in the northern Southcial cycles of the Amazon Fan (Holes 940A, 944A and America. Leidse Geologische Mededelingen 38, 71–90.946A). Proceedings Ocean Drilling Program, Scientific Wijmstra, T.A., 1971. Palynology of Guiana coastal basin.Results 155, 397–409. Ph.D. thesis, University of Amsterdam.

Huber, O., 1987. Neotropical savannas: their flora and vegeta- Wille, M., Behling, H., Hooghiemstra, H., preparation. Vegeta-tion. Tree 2, 67–71. tion history in the western savannas of the Llanos Orientales

Lowy, P., Rangel, O., 1993. Caracterizacion climatica de la of Colombia based on a 9800-yr pollen record from LagunaLas Margaritas. in preparation.Orinoquıa Colombiana. Contribucion del Proyecto Estudio

Late Holocene history of savanna gallery forest from Carimagua area, Colombia

Documents

Transcript of Late Holocene history of savanna gallery forest from Carimagua area, Colombia