Spatial and Temporal Patterns of Amazon Rainfall

Author(s): Wim Sombroek

Source: AMBIO: A Journal of the Human Environment, 30(7):388-396.
Published By: Royal Swedish Academy of Sciences
https://doi.org/10.1579/0044-7447-30.7.388
URL: http://www.bioone.org/doi/full/10.1579/0044-7447-30.7.388

BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the biological, ecological,
and environmental sciences. BioOne provides a sustainable online platform for over 170 journals and books
published by nonprofit societies, associations, museums, institutions, and presses.
Your use of this PDF, the BioOne Web site, and all posted and associated content indicates your acceptance of
BioOnes Terms of Use, available at www.bioone.org/page/terms_of_use.
Usage of BioOne content is strictly limited to personal, educational, and non-commercial use. Commercial
inquiries or rights and permissions requests should be directed to the individual publisher as copyright holder.

BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit publishers, academic institutions,
research libraries, and research funders in the common goal of maximizing access to critical research.
 Report Wim Sombroek

Spatial and Temporal Patterns of

Amazon Rainfall
Consequences for the Planning of Agricultural Occupation and
the Protection of Primary Forests

MATERIALS AND METHODS Annual rainfall (mm)

The spatial and temporal pattern of annual rainfall and the < 1200
strength of the dry season within the Amazon region are The study area was demarcated based 12001600

poorly known. Existing rainfall maps are based on the data on hydrographic, vegetational and le- 14001600

from full-scale, long-term meteorological stations, operated gal considerations. It concerns the ar- 16001800
18002000
by national organizations linked to the World Meteor- eas of the Amazon river basin down- 20002200
ological Organisation, such as INMET in Brazil. Stations stream of the Andean Cordillera, the 20002400

with 30 or more years of uninterrupted and reliable re- Araguaia-Tocantins basin, the minor 24002800

cordings are very few, considering the size of the region, catchments directly bordering the At- 28003200
32003600
and most of them are located along the major rivers. It has lantic ocean, as well as the river sys- 32004000
been suggested that rainfall conditions away from these tems of the Guyanas and southern 36004000
rivers are substantially different. An analysis has been Venezuela. The vegetational criterion 40006400
made of the records of a network of simple pluviometric is the Amazon humid tropical forest < 6400

sites in the Brazilian part of the region as maintained by biome, including its internal edaphic July, October:
the National Agency for Electric Energy (ANEEL) since savannahs. Legally, it concerns the 9
Central month of dry season

1970. The latter data sets were used to draw more detailed countries of the Amazon Cooperation
maps on annual rainfall, and on the strength of the dry Treaty, and the 9 states of Brazilian
Meteostation

season in particular; average number of consecutive Amazonia Legal. The total area is
Full meteostation, included
in 30 year normals of WMO/FAO
months with less than 100 mm, 50 mm, and 10 mm, about 6 x 10 km2, which is equiva- Full meteostation, not included
in WMO/FAO database
respectively. Also, some data were obtained on the spatial
lent to the size of Europe, excluding Pluviometric station of ANEEL
expression of El Nio events within the region. Sub-
Russia. Pluviometric station of ANEEL
regional differences are large, and it is argued that they with short and/or less reliable data

are important for the success or failure of agricultural The digitized topographic base
settlements; for the hazard of large-scale fire damage of map of Amazonia Legal, scale
Other
the still existing primary forest vegetation; for the 1:3 000 000, of the Brazilian Geo-
Outside study area
functioning of this land cover as stock and sink of CO2, and graphic and Statistics Institute (2),
Ocean and inland water
for the likelihood that secondary forests on abandoned was used as cartographic base. The Altitude above 500 m
agricultural lands will have less biomass. The effects of western and northern edges were
past El Nio rainfall anomalies on the biodiversity of the taken from a map of the World Me- River

natural savannahs within the forest region are discussed. teorological Organisation (WMO) International boundary
State boundary
(3). Areas above 500 m altitude are Lowland boundary
indicated on the maps with a shading, Road

on the basis of various cartographic

sources. The location and data of full
INTRODUCTION meteorological stations of long-term
The current efforts for ecologic-economic zoning within the functioning in the Brazilian Amazon
Amazon region, as for instance carried out for the Brazilian part were taken from various publications
under the patronage of the MMA/PPG7/SPRN project (1), re- of the National Meteorological Insti-
quire information on the spatial variation of climatological char- tute INMET (4). The non-Brazilian
acteristics as one of the criteria. stations were taken from the WMO
Until recently the only climatological data available were those database (3). The total number of sta-
of the fully equipped meteorological stations in the region that tions is about 50, or 1 per 120 000 0 100 200 300 Km

have at least 30 years of continuous recording (50 stations, of km2.

which 30 in the Brazilian part). The network of simple pluvio-
These stations are located by-and-large along the major riv- metric recording sites within the
ers and are therefore not necessarily representative for the up- Brazilian Amazonabout 800, of
per catchments. A much larger network of simple pluviometric which about 700 within the forest
Base map: IBGE 1988
stations, about 800 in total, has been operational in the Brazil- zonewas copied from draft maps of
Map digitized by International Soil
ian part of the Amazon region since the 1970s. The network is the Rede Bsica Hidrometeoro- Reference and Information Centre,
Wageningen, The Netherlands
fairly evenly distributed over all subcatchment areas. It is there- lgica Nacional, which is maintained for PPG 7/GTZ, Manaus, Brazil.

fore a valuable source of additional information, especially on by the National Agency of Electric
the absence or strength of a dry season, which is decisive for Energy (ANEEL, formerly DNAEE)
planning of agricultural settlement, avoidance of forest fires, as- on the basis of field recordings car-
sessment of biomass and biodiversity, and ecological-economic ried out by institutions such as CPRM
zoning in general. (Geological Service) and IDESP

388 Royal Swedish Academy of Sciences 2001 Ambio Vol. 30 No. 7, Nov. 2001
http://www.ambio.kva.se
 Annual rainfall (mm)
Map no. 1

Ambio Vol. 30 No. 7, Nov. 2001 Royal Swedish Academy of Sciences 2001 389
http://www.ambio.kva.se
Map no. 2
Number of conseq. months with
less than 100 mm rainfall
(Socio-economic Development Institute of Par State). The Hy-
drologic Resources Department of ANEEL kindly provided all
computerized pluviometric recording data, on a monthly basis,
for all of the Brazilian Amazon. Such pluviometric information
networks also exist for the Amazon parts of neighboring coun-
tries, but they are of much lower density, except for the Guy-
anas, and their data were not available to the author.
There are 2 types of fully equipped meteorological stations.
Those qualifying for the 30-year averaged normals criterion
of WMO (full periods 19301960 and 19601990) are indicated
with large black dots within circles, while those that do not have
recordings of sufficient length, gauge, or reliability to be used
statistically are indicated with large black dots without circles.
The recordings of the simple pluviometric sites are of varying
quality. Some are quite long-term, having started around 1965
and continued uninterruptedly till 1998. Others started recently
or have been abandoned 10 to 15 years ago. Those sites with
less than 10 years of recording, and/or low reliabilitygaps in
the monthly data or obvious anomaliesare indicated on the
maps with small open dots, the others with small black dots. The
poor accessibility of higher-altitude and broken landscape situ-
ations of the Guyana and Brazilian Shields, and the occurrence
of Indian Reserveswhich often coincide with high and inac-
cessible terrainhave resulted in several subareas of poor or in-
existent rainfall recording sites. Examples are the whole fron-
tier zone of the Guyanas with northern Par, and the interfluve
zone between the Araguaia and the Middle Xingu rivers.
Interpolation lines of annual and monthly rainfall character-
istics were drawn manually, taking into account the variation in
390 Royal Swedish Academy of Sciences 2001
http://www.ambio.kva.se
Number of months
reliability as described above. To a degree, the hydrographic pat-
tern was taken into account, once it became apparent that away
from the main rivers and their broad floodplain the annual rain-
fall is higher and the dry season less pronounced than near these
rivers, in agreement with a hypothesis of Molion and Dallarosa
(5). This Manaus rule or brisa fluvial is defined as the effect
of local river breeze circulation away from the waterbodies, due
to early-daytime preferential ascending cloud formation over the
forest-covered land area. It apparently holds true for the stretches
Solimes Meio Amazonas Baixo Amazonas, but is replaced
by a contrasting rule for the very wide water-bodies at the mouth
of the river system, where the tropical Atlantic ocean causes high
annual rainfall but a pronounced dry season. This is called
Maraj rule or brisa do mar (6). In the southeastern part of
the region the parts of higher altitude (above 500 m) appeared
to have somewhat higher annual rainfall and lower dry-season
strength. This phenomenon was subsequently used in the fine-
tuning of interpolation lines.
Interpolation lines for the non-Brazilian parts of the region are
very tentative. They are based on the WMO (3) map on annual
rainfall, while for the seasonal characteristics use was made of
the Length-of-Growing-Period (LGP) approach of FAO (7) and
its database on monthly rainfalls. Also a French study (8) on the
geography of the rainfall in the Guyanas and southern Ven-
ezuela was taken into account.
Computer-driven kriging of the interpolation lines was not at-
tempted, because both the reliability categories of the recording
stations and sites, and the influence of large waterbodies or
orographic features are as yet too vague to be put in formal
Ambio Vol. 30 No. 7, Nov. 2001
Map no. 3
Number of conseq. months
with less than 50 mm rainfall
weight classes. However, gridding of the pattern that results from
the present manual interpolation is probably a relatively simple
exercise for ecosystem modelling specialists. Translation of the
map pattern into the Kppen or Thornthwaite climatic schemes
is feasible as well.
Four different digital maps were prepared (Maps 14):
Map #1: annual rainfall pattern, with isohyets of the WMO sys-
tem: 4000, 3600, 3200, 2800, 2400, 2000, 1600, 1400, 1200 and
1000 mm, and 2 extra isohyets in the lower range: 2200 and 1800
mm.
Map #2: the number of consecutive dry months, averaged over
the years of observation, with 100 mm monthly rainfall as cri-
terion.
Map #3: number of consecutive dry months, averaged over the
years of observation, with 50 mm monthly rainfall as criterion.
Map #4: number of consecutive dry months, averaged over the
years of observation, with 10 mm as criterion, also implying a
low atmospheric humidity during the dry months.
In all 4 cases the central month of the dry season is indicated,
but without line boundaries.
The set of 3 maps (Maps 24) on the characteristics of the
dry season is rather uncommon. Many studies on the Amazon
climate provide maplets with isohyets of selected dry months,
for instance July and December, but examples of the integration
of all dry months providing quantitative information on the
length, strength and regularity of the whole dry season are very
few. The Length-of-Growing-Period map of FAO (7) at scale
1:5 000 000 shows, understandably, the wet season instead of the
dry one.
Ambio Vol. 30 No. 7, Nov. 2001 Royal Swedish Academy of Sciences 2001
http://www.ambio.kva.se
Number of months
A recent NGO study on the risks of forest fires (9) recognizes
the importance of the dry season, but does not show it in map
form, though it refers to an attempt on gridded input in a NASA-
CASA ecosystem production model (10). Characterizing a dry
month as having less than 45 mm rain, it used only the normal-
ized 19611990 records of the recognized WMO meteorologi-
cal stations for the whole of Brazil. An early attempt was made
by the present author (11), applying the 100 mm month1 and
50 mm month1 criteria to the recordings of the 25 meteostations
available in 1959 (12).
Attention was given to the occurrence of years with excep-
tionally low rainfall, which would give indications of local in-
fluence of the El Nio phenomenon. However, the relatively
short length of many of the pluviometric records did not allow
for delineation of spatial drought patterns with any degree of ac-
curacy on an additional map.
RESULTS
The use of the computerized pluviometric database of ANEEL
yields a more detailed pattern of annual rainfall (Map #1) for
the Brazilian Amazon, when compared to published maps (3, 13,
14) and a number of page-size maps in review articles (1518).
The annual rainfall away from the main rivers is higher. The spa-
tial pattern is complex, not only in the water-divided areas where
orographic unevenness may have an influence, but also in ho-
mogeneously low altitude areas. An example is the configura-
tion for Maraj island in the mouth of the Amazon river. A far
more intricate pattern was found than assumed thus far: less rain-
391

Number of conseq. months with
less than 10 mm rainfall
fall in its western interior part without a dry season, contrasting
with high rainfall on its coasts, but partly with a pronounced dry
season. Areas with low annual rainfall (< 1800 mm) occur in
Roraima, the upper Trombetas-Jar area of northern Par, south-
west of Altamira on the Trans-Amazonica, and the Aailndia
area on the frontier Par-Maranho. Together they constitute a
sort of dry belt or corridor. However, these are also the areas
with relatively low-density field information, and many of their
recording sites have moreover become nonoperational.
The bulge of dryness in the Paragominas-Aailndia area, with
its high degree of deforestation since the early 1960s, is note-
worthy. During the combined forestry- and soil survey of FAO
in 1960/61, when the primary forest coverage was still complete,
both the high timber volumes per ha and the low base satura-
tion status of the soils in the Paragominas area suggested that
in those days the dry season was little pronounced. Only 200
km south of Paragominas town the timber volumes dropped be-
low 100 m3 ha1 and the base saturation of the soils reached levels
above 35% (19), indicating a strong seasonal drought. It is pos-
sible that the strength of the dry season and its duration have
become more evident because of the large-scale deforestation.
However, records of the Paragominas meteostationstarted only
in 1965 and out of operation since 1995 because of lack of mu-
nicipal interestdo not show such a negative trend. Any ten-
dency to decreasing rainfall in the region after large-scale de-
forestation may be verified easier in northern Mato Grosso or
central Rondnia where pluviometric records are of longer term.
An indirect indication of a negative effect of large-scale defor-
estation on the amount and regularity of subregional rainfall is
given by the 2-lobed extension of drier conditions in the
Barcelos-Rio Branco area in the south of Roraima State. The
abundance of edaphic white-sand savannahs there would imply
reduced evapotranspiration, hence less cloud formation in the dry
season.
Screening of the 800 individual pluviometric records; analy-
sis of the data of the full Brazilian meteostationsthe first ones
dating from 1910as well as scanning of the oscillations in the
discharge regime of the Rio Negro-Rio Solimes system at
Manaus, measured since 1902 (20, 21) give the following indi-
cations on El Nio occurrences since 1900:
392 Royal Swedish Academy of Sciences 2001
http://www.ambio.kva.se
Number of months
weak droughts in: 1903 (Manaus), 1911 (Manaus), 1930
(Parentins), 1933 (Manaus), 193940 (Barcelos), 194142
(Taperinha), 195152 (Porto Velho, Taperinha), 1955 (Manaus),
1963 (Porto Velho, Taperinha), 198081 (Tapau, So Felix do
Araguaia), 198788 (Aailandia), 199596 (Apui);
moderate droughts in: 1909 (Sena Madureira, Manaus),
191516 (Sena Madureira, Taperinha, Manaus), 1936 (Manicor,
Manaus), 195758 (Itacoatiara, Parentins), 196768 ( Coar,
Humait, Manaus), 197172 (Boa Vista, Porto Velho, Tarauac),
1976 (Rio Branco), 199192 (Prainha, Humait, Coar,
Barcelos);
strong droughts in: 1906 (Manaus), 192527 (Fonte Boa,
Erunep, So Gabriel de Cachoeira), 1948 (Manaus), 1983 (eve-
rywhere), 19971998 (everywhere).
The ANEEL data set provides also some information on the spa-
tial variation in the strength of the El Nio phenomenon within
the Amazon region. The central-northern zone apparently expe-
riences stronger El Nio effects than the southern zone. In the
north, the dry season has a gradual start and an irregular length
(wobbly character); in some years there are one or more com-
pletely dry months outside the long-term average period, which
has its center in January. In the south, the dry season starts quite
regularly and abruptly, with the center in JulyAugust (south-
west) or September (southeast). The dry season in the eastern
part of the Brazilian Amazon, near the ocean coast, is transitional
between the 2 systems, with its center in October.
CONSEQUENCES FOR ECOLOGIC-
ECONOMIC ZONING
The new set of rainfall maps can be used in various ways. In
their present vectorized form they have applications at regional,
state and municipal levels. In a gridded and digital form, they
can be helpful in modelling biosphere processes for the region
as a whole, in different scenarios of human occupation and cli-
mate change. Gridding with 30' x 30' latitude/longitude for glo-
bal models, or at 5' x 5' pixels for Amazon-level models (10)
may want to combine the data of Maps #2, #3 and #4 through a
composite legend.
Ambio Vol. 30 No. 7, Nov. 2001
 The map on annual rainfall (Map #1) is relevant for hydro-
logical studies as it provides information, together with the ex-
isting ANEEL set of fluviometric data, on the minimum, aver-
age and maximum water flows of subcatchment areas. These are
of use for the planning of road constructional work, for river traf-
fic, groundwater recharge, aqueous pollution control, urban-
center water supply and sanitation discharge, local hydropower
generation, etc.
For agricultural and ecological studies, the dry-season Maps
(#2, #3, and #4) are more important. Viability and sustainability
of agricultural settlements, forest biomass growth, risks of for-
est fires, chances of forest recuperation, atmospheric carbon se-
questration, supplementary soil moisture storage, and El Nio
effects in ecological tension areas will be discussed here.
Agricultural Settlements
Building of rural access roads and their maintenance is much
less problematic when there is a distinct dry season. The same
holds for infrastructural facilities at the settlement centers,
whether such centers are oriented on agricultural produce, or en-
gaged in mining, logging or nontimber forest product gathering.
Human and animal health is favored by a dry period, restricting
the spreading of endemic diseases and the multiplication of their
vectors. Even more significant for the viability and sustainability
of agricultural settlements are the dry-season effects on land
clearing practices and crop growth. In the absence of a dry sea-
son (the 0 and 1 dry-months zones under the less than 100 mm
month1 criterion; Map #2) the burning of primary or secondary
forest vegetation after complete slashing remains incomplete.
This makes the newly cleared land surface less suitable for the
planting of food or cash crops or the implantation of pastures.
Crops such as upland rice, maize and beans need a dry spell for
ripening and hardening of their grains, and the prevention of rot-
ting after harvesting. The yield of such crops is anyhow bound
to be lower because the near year-round cloud cover implies
lower daylight intensities, hence lower photosynthesis. This is
also reflected in the 365+ isoline of length-of-growing-period
maps of FAO (7, 22). The above restrictions hold even more for
large-scale soybean growing. This crop is sensitive to attacks by
a number of diseases and pests in its vegetative stage if near-
surface air humidity remains high during a good part of the day,
such as in the edaphic savannah area of Humait. The use of
heavy machinery at the required high degree of mechanization
for the crop to become commercially viable, is only feasible
when the land surface is reliably dry at both planting and har-
vesting times. Only zones 1, 2 and 3 of Map #4 (which uses the
10 mm month1 criterion) are therefore suitable for mechanized
soybean cultivation. FAO recognizes a growing period of only
120 to 300 days as suitable to very suitable for the crop (7).
Suitable food crops in the continuously moist regions are cas-
sava, bananas, and certain tree-fruit crops (23, 24). The health
of cattle, for household use or meat commercialization, is prob-
lematic in near-constant rainy conditions because of abundance
of ecto-parasites and blood-sucking insects (25), as well as nu-
trient imbalances in the soils (26) and long distance to markets
and supply centers. It is noteworthy that the zones 0 and 12 of
Map #2 coincide with the near absence of sustained agricultural
settlements of substantial size. Whether government-organized
or spontaneous, many recent settlements in these zones have
been abandoned. This is not clearly indicated on Remote Sens-
ing based maps of deforestation areas, because they do not nor-
mally distinguish between active settlement and frequency of old
regrowth.
There are indications, from the denseness of patches of terra-
preta-do-indio soil (11) and early historical records, that the pre-
Colombian Amerindian populations also preferred dry-season
areasunless soil and river-strategic conditions were particu-
larly favorable.
Ambio Vol. 30 No. 7, Nov. 2001 Royal Swedish Academy of Sciences 2001
http://www.ambio.kva.se
The largest areas with active deforestation for agricultural pur-
poses and permanence of rural settlements coincides with Zone
5 of the 100 mm month1 Map (#2), as confirmed by Landsat-
TM imagery, by the reas antrpicas of the 1995 IBGE map
on Amaznia Legal (2) and the network of feeder roads on the
present maps. These are apparently the areas where one or two
crops per year, and several perennials, have secure yields.
Among the commercial perennials, only oil palm thrives in
the permanently humid zones, as proven by several successful
plantations near Belm. The lack or neglect of processing facili-
ties has thus far been an impediment for sustainable oil palm
enterprises in government-led efforts in Amazonas state (Tef;
Distrito Agropecuario north of Manaus). Rubber plantations
have been a failure nearly everywhere in the region, because of
their sensitivity to the Dothidella ulei fungus. A FAO/UNDP
committed study by the Agronomic Institute of Campinas (27)
concluded that only areas with a distinct dry season and a low-
temperature spell would be safe niches for rubber plantations.
The southern higher-altitude areas, indicated with a shading on
the present maps, would therefore be of potential interest for rub-
ber plantations.
Timber Volume, Biomass Productivity,
and Carbon Storage
Early FAO forest and soil inventories (11, 28), the more recent
RADAMBRASIL multidisciplinary inventories (29) and their
spatial generalization (30), all indicate that the highest gross tim-
ber volumes and thus aboveground biomass values, > 200 m3
ha1 implying > 100 tonnes C ha1, occur in areas where the to-
tal annual rainfall is about 2000 mm and there is a dry season
of 2 to 3 months (50 mm criterion) and 4 to 5 months (100 mm
criterion), respectively. These are also the areas preferred for
agricultural settlement, especially where the soil conditions are
favorable, as is the case for the diabase- or limestone-derived
dusky red soils near Altamira and the high base status soils in
Rondnia. Safeguarding the remaining forest patches in such ar-
eas, as permanent ecological reference sites, should be given high
priority in the hot spot approach to tropical forest conserva-
tion.
Forest growth apparently benefits from a short dry spell, pos-
sibly because of smaller damage from diseases and pests. Prob-
ably more important is the lower interception of sunlight, an es-
sential plant growth factor, in comparison with the near-continu-
ously cloudy areas. Modelling of the potential Net Primary Pro-
ductivity of biomass (NPP), based on remotely sensed average
cloud cover as derived from NDVI-AVHRR data sets of NASA,
gave NPP values of 10001200 g C m2 yr1 for the continuously
moist zone vs. 12001400 g C m2 yr1 for other parts of the re-
gion (10).
Timber volumes are lower and the chances for high-forest re-
cuperation are poorer when soil conditions in the short-dry-sea-
son zones are marginal (shallowness; compactness of the sub-
soil, swell-shrink behavior, or extreme sandiness). Under such
conditions anthropical disturbances, whether present-day or from
pre-Columbian Indian population groups, will result in bamboo
forest (bambuzal of Acre and adjoining parts of Amazonas (31);
creeper-and-vine forest areas (cipoal areas of the lower Xingu
planalto stretches (11, 32); Orbignia palm forest (babaual ar-
eas in the middle Xingu area (32, 33), or even shrub savannah
(campina areas near Santarm and Manaus). All are divergent
vegetation types, with lower stocks of carbon and often also
lower biodiversity than the closed-canopy high forest.
Forest Fires and Forest Recuperation
As demonstrated by the 1997/98 extensive forest fires in Roraima
state, there is a large spatial variation in the risk of forest fires,
which may spread from slash-and-burn sites or are caused by
lightning. Fire risks are low if logging takes place very selec-
393
tively for rare precious timber or with careful planning for long-
term sustainable yields of more tree species, say only 30 m3
ha1, once in 20 years or so (34). If higher volumes are cut and
towed away then the remaining off-fall can become a source of
easily flammable dry matter, augmented by litter from semi-
deciduous species where the dry season is pronounced. The re-
sulting below-canopy fires often remain undetected by radar- or
visual-light-based Remote Sensing, but may cause considerable
damage to floral and faunal diversity (9).
The chances for accidental extensive forest fires are largest
in the zones 5 and 6 of the 100 mm Map (#2), and the zones 3
and 4 of the 50 mm Map (#3). Variation in soil conditions, es-
pecially the moisture storage capacity, plays a modifying role.
It is generally assumed that it will take 300 to 1000 years be-
fore a completely burned-down humid tropical forest regains its
original structure and species composition (35). However, recu-
peration to a high, closed or open-canopy forest is far from sure,
as demonstrated by several field studies (36). As soon as there
is a long dry season and soil physical conditions are poor, recu-
peration becomes unlikely. Zones 1, 2 and 3 of Map #4 (the 10
mm month1 criterion) with their very pronounced dry season
are areas where the primary forest, already semideciduous, is a
relict from former, more humid climatic conditions. It will never
regrow to its present status if subjected to large-scale slash-and-
burn practices for ranching or soybean cultivation. There are,
however, several large Indian Reserves in these zones that may
provide protection from massive deforestation.
Carbon Sequestration Potential
The Amazon forest ecosystem is a large store of carbon. Until
recently it was assumed that in mature forests the uptake of at-
mospheric CO2 for daytime assimilation is balanced by an equal-
strength CO2 emission through nighttime canopy respiration and
heterotrophic soil respiration. Only young secondary forest
would act as a CO2 sink. However, studies in Rondnia
(ABRACOS project in the Ji-Paran area (37)) and north of
Manaus (BIONTE project (34)), using eddy-correlation flux
measuring techniques from towers in old-growth forest and
partly also tree-girth measurements, suggest otherwise for recent
years. Apparently the increased CO2 concentration in the atmos-
phere stimulates tree growth, resulting in a net uptake, called se-
questering or sink condition for CO2, by the forest ecosystem in
most years. The amount sequestered in the Rondnia case was
13 tonnes C ha1 yr1, and in the Manaus case even 57 tonnes
C ha1 yr1 (38). A net uptake is consistent with theoretical con-
siderations and actual measurements in temperate regions on C3
plantsto which nearly all tropical tree species belongand is
known as the CO2 fertilization effect. Apparently even in the
humid tropical forest environment, with its low chemical soil fer-
tility, the atmospheric CO2 concentration has been in the mini-
mum as a growth factor until some decades ago. Measured cu-
mulative change in aboveground dry biomass in Amazonia since
1977 points in the same direction (39). The spatial and tempo-
ral consistency of this fertilization phenomenon will be further
researched by the new interdisciplinary research program LBA
(40). It is a cooperative program of Brazilian institutions led by
INPE (So Paulo) and INPA (Manaus), with participation of
American and European institutional consortia funded by NASA
and the European Community, respectively. Additional CO2 flux
measurement sites will be Caxiuan and Santarm, both in mid-
Par state, and possibly Tef and So Gabriel do Rio Negro in
Amazonas state.
It is debatable whether the extra growth will continue as higher
aboveground biomass in the long run. It is more likely that a
good part of the generated higher biomass will end up in the for-
est soil, in the form of higher soil organic matter content of a
stable nature. Soil organic matter is already nowadays a very sig-
nificant stock of carbon (4144). Increased above- and below-
394 Royal Swedish Academy of Sciences 2001
http://www.ambio.kva.se
ground tree growth implies earlier maturity and thus earlier se-
nescence, hence higher tree fall intensity and associated higher
input of organic material over the rootable depth of the soil.
There are indications that the function of the Amazon primary
forest as a CO2 sequestering mechanism is interrupted during El
Nio years. It has been suggested that there would be insuffi-
cient soil moisture reserve in such years (45), thus moisture
rather than CO2 concentration becoming the limiting growth fac-
tor. In contrast, the Net Primary Productivity in those areas that
are normally without any dry season, such as west of Manaus,
may benefit from an El Nio year because of the temporary
availability of more sunlight for photosynthesis.
Soil Moisture Storage
At the beginning of the dry season, plant growth can continue
unabatedly for some time because of a reserve of moisture in
the rootable soil. However, the plant-available moisture storage
capacity (PAM, moisture in the suction range of pF 2.04.2) of
the various Amazon soils is poorly quantified. There is an acute
scarcity of direct measurement of soil moisture suction charac-
teristics on undisturbed samples and of their bulk densities. The
few hard data carried for Brazilian soils are summarized by
Tomasella and Hodnett (46). About 30 of the profiles with pF
data are from the Amazon region itself, mainly from Par State
and the neighborhood of Manaus, where Xanthic Ferralsols
(Latosssolos amarelos) predominate. There are no soil profiles
with pF data for the Brazilian or Guyana Shield areas, nor for
the vast area (700 000 km2) southeast and east of Manaus that
has fine sandy to silty sediments of the Late Pleistocene I for-
mation and Plinthic Acrisol as main soil (47). Xanthic Ferralsols
and related soils are deep but are known to have low values of
PAM, viz. 510 vol.% in the central and lower parts of the pro-
files, even if they are very clayey. Hydrologically, they act like
sandy soils of temperate regions. Less weathered soils of the
Amazon and Brazil-at-large, such as Orthic Acrisols, Luvisols
and Cambisols have substantially higher values, viz. 1015 vol.%
and even more in topsoils when rich in humus, but these soils
are less deep.
Comparison of analytical data on disturbed samples with tex-
turally equal, but mineralogically different temperate-zone soils
will give too high PAM values. Special equation calibration is
required (48) to arrive at estimates that are in line with the few
existing pF data. The overestimation problem is also pointed out
in the recent NGO study (9) on fire and deforestation risks. The
study could not use the RADAMBRASIL data set of 1147
analyzed soil profiles of the Brazilian Amazon (29) for compari-
son with soil textural situations of temperate zones because none
of these profiles have pF or bulk density data. An additional
problem is the relevance of the maximum soil moisture-suction
value for the humid tropical forest environment. The internation-
ally used value of pF 4.2 (15 atm. suction) was established for
the wilting point of temperate-zone annual crops such as sun-
flower and wheat. It is possible that tropical forest species have
smaller suction force, but this was never investigated, either in
the field or in laboratory growth chambers. With respect to maxi-
mum photosynthesis and CO2 uptake, the leaf stomata should
be fully opened during daytime. Therefore, the more readily
available moisture (pF 3.5 minus pF 2.0 values), which may be
only two-thirds of the PAM value, should be the really impor-
tant amount at modelling of the carbon sink function of the for-
est.
The total moisture availability for uninterrupted plant growth
is a combination of the PAM value per soil horizon and the to-
tal depth of the soil that is rootable. Total rootable depth varies
highly over the Amazon region, from extremely deep (510 m)
on the freely draining Xanthic Ferralsols (49, 50), to quite shal-
low (50100 cm) for many of the Plinthic Acrisols southwest
of Manaus with their relatively low timber volume (43). The
Ambio Vol. 30 No. 7, Nov. 2001
ABRACOS project monitored at weekly or half-weekly inter-
vals for 2 to 4 years the moisture content till a depth of 3.6 m
under forest and planted pasture, at 3 sites (north of Manaus;
Ji-Paran in Rondnia, and Marab in southern Par State) (51).
Soil moisture abstractions at all depths were much greater un-
der forest than under pasture, and also deeper, well into the
weathering substratum. Per soil horizon there was a much
smaller abstraction in the Ferralsols than in the other 2 soils,
which is consistent with its assumed lower PAM values (no pF
curves established).
It can be concluded that forests on Ferralsols, when a distinct
dry season is present, need to root deeper to secure sufficient
soil moisture for uninterrupted growth than those on less weath-
ered soils. In general, representativeness of analyzed soil pro-
files for broad landscape-related soil units, rather than indiscrimi-
nate statistical analysis of the total number of samples, should
be used to arrive at spatially improved estimates on effective
depths and moisture storage capacities for forest growth, includ-
ing validation by measurements on undisturbed samples (51).
El Nio and Ecotonal Biodiversity
Among the El Nio occurrences in the Amazon as listed above,
the period 19251927 stands out. More than a full-year drought
occurred in the Rio Negro catchment area, with many forest fires
and extremely low river-water levels that impeded river travel.
This event is well documented in a report by the Salesian bishop
of the area to the Vatican (52).
Very strong El Nio events such as the 19251927 one will
have occurred earlier as well, perhaps at intervals of 70 years.
There are reports (24) of very strong droughts in 1860 (Purus
area) and in 1774 (Rio Negro area). They must have had pro-
nounced effects in ecotonal field situations at the edge of the
forest biome, and in areas of ecological tension within the sys-
tem. The latter would apply, for example, to the edaphically de-
termined patches and stretches of savannahs and savannah-for-
ests, which are particularly frequent in Amazonas state (campos,
campinas, campinaranas, caatinga amaznica). Their present-
day specific biodiversity, including endemism, and their
vegetational structureboth not well studied thus farwould
be reimpacted in such exceptionally dry years (53). Physiologi-
cal drought effects would be enhanced by accidental fires, or
purposely started ones by Amerindians or early settlers, result-
ing in sharp boundaries with the surrounding forest.
From studies on charcoal occurrences in the So Carlos do
Rio Negro area (54) it was concluded that extensive fires must
have occurred 250 to 400 years ago. Still larger effects will have
had the very broad-periodicity prehistoric Mega El Nios
recognized by Meggers (55, 56) on the basis of pollen sequences
and river-level fluctuations in the northern part of the region
(stone-grinding sites of Amerindians are occurring below
present-day low-water levels of rivers). She identified dry peri-
ods around the years 400, 700, 1200 and 1500 BP. The occur-
rence of 3 levels of broad low terraces of the Late Pleistocene
Ia-formation in western Amazonas (47) suggests that also be-
fore 1500 BP there were times with lower rainfall and/or stronger
dry seasons, succeeding an overall wetter period between 12 000
and 5000 BP (17).
An example of sharp vegetation boundaries induced by fire
is the large forest-encircled open savannah area in southeastern
Amazonas state, below the southern dip of the Trans-Amaznica
highway. It is a surmised source area of the Tup groups of Am-
erindian languages (57). These campos de Tenharim may have
undergone extensive anthropogenical vegetation degradation ac-
companied by water and wind erosion during these early drought
periods. Low dunes are still present, while water-erosion prod-
ucts apparently have accumulated in the original downstream riv-
erbeds, blocking their discharge function. Nowadays, these
downstream stretches feature as winding and narrow, elongated
Ambio Vol. 30 No. 7, Nov. 2001 Royal Swedish Academy of Sciences 2001
http://www.ambio.kva.se
strips of shrub savannah vegetation on a white-sand soil surface
(campinas de Manicor) (11, 43).
A large area of uninterrupted shrubby savannah or savannah-
forest vegetation (campina and campinarana) occurs northwest
of Barcelos in the Rio Negro region. On Landsat-TM images
there are clear traces of former longitudinal and barchane dunes,
which must have developed on the coarse sandy soils concerned,
during a dry period of the Late Pleistocene or Early Holocene.
CONCLUSIONS AND RECOMMENDATIONS
Interpolation of the data of simple pluviometric recording sites
away from the main rivers is helpful in assessing and mapping
the spatial variation in annual rainfall, as well as the length,
strength and regularity of the dry season in the Amazon region.
The dry season characteristics are important for the success or
failure of agricultural settlements, the variation in aboveground
biomass, the risk of forest fires, and the chances for forest re-
generation. They also provide information on the extent of the
CO2 sink function of the primary forest, as well as the spatial
variation and strength of El Nio drought anomalies. All of these
factors need to be taken in consideration with the current efforts
to establish national and state-level policies for forest protection
in harmony with sustainable rural settlement.
Many of the pluviometric recording sites in the Brazilian part
of the region have recently become nonoperational. For the ben-
efit of agro- or forest-ecological and socioeconomic zoning pur-
poses at subregional or municipal levels, it is recommended that
this situation be remedied soonest, and that extra sites be estab-
lished in the so-called dry corridor of the eastern part of the re-
gion. At the same time, several fully automatic and satellite-
linked complete meteorological stations should be established
and maintained in remote water divide areas, to substantiate the
information from the pluviometric network.
There is a great need to correlate rainfall patterns with data
on key soil physical parameters, with specific attention to meas-
urement of total rootable depth, bulk density and net soil mois-
ture storage capacity, in actual field situations and on undisturbed
samples at laboratory analysis. These properties need to be de-
termined also for the deeper subsoil and the weathering substra-
tum, particularly in areas where there is a long dry season. The
resulting Amazon soil moisture storage map should be linked
with the 3 dry-season maps (Maps 24), to arrive at quantita-
tive estimates of the occurrence of a CO2 sink function of the
Amazon old-growth forests, spatially from per-humid to semi-
deciduous and temporally in and between El Nio years. One
of the outcomes may be that the sink function is more substan-
tial and continuous in the zone with little or no dry season, like
most of the state of Amazonas and the adjoining parts of Peru
and Colombia, even though this zone does not have the highest
timber volume at present. They would then qualify even more
for protection through the Ecological Corridors idea of the PPG7
program.
Quantifying the spatial and temporal carbon sink function of
the forest also requires that the network of CO2 flux monitoring
of the ongoing LBA program be extended to areas beyond the
axial zone Manaus-Santarm-Belm, to incorporate different
rainfall and soil conditions in western Amazonia and in the Bra-
zilian Shield area.
The origin and the specific biodiversity values of the many
intraforest natural savannahs merit additional study, also in re-
lation to past and present El Nio droughts and early Amer-
indian land use-activities.
395
References and Notes
1. Ministrio do Meio Ambiente Pilot Programme for the protection of the Brazilian tropi-
cal forests of the G7 countries / Subprogramma de Poltica de Recurson Naturais. (In
Portugese).
2. IBGE 1995. Map of Amazonia Legal 1:5 000 000. Inst. Brasileira de Geografia e
Estadistica, Rio de Janeiro.
3. WMO 1995. Atlas Climtico de America del Sur. Vol. I. Scale 1:5 000 000. WMO,
Geneva.
4. Marques, J.C.J. (ed.). 1992. Normais Climatolgicas 19611990. Ministrio de
Agricultura e Reforma Agrria, Instituto Nacional de Meteorologia INMET, Brasilia.
(In Portugese).
5. Molion, L.C.B. and Dallarosa, L.R.G. 1990. Pluviometria da Amaznia: so os dados
confiaveis? Climanlise, Bol. Monitor. Anl. Clim. 5, 4042. (In Portugese).
6. Molion, L.C.B. 1987. Climatologia dinmica da regio Amaznica: mecanismos de
precipitao. Rev. Brasil. Meteo. 2, 107117. (In Portugese).
7. FAO 1978. Report on the Agro-Ecological Zones Project. Vol. I. Methodology and
Results for Africa. [with 1980 map at 1:5.000 000 for South America]. World Soil Re-
sources Report 48 FAO, Rome.
8. Petit, M. 1978. Geographie des pluies entre Amazone et Orinoque (les Guyanaes entre
lEquateur et le 9 (degree de latitude nord). In: Travaux et Documents de Geographie
Tropicale 34. CEGET-CNRS, Domaine Universitaire, Bordeaux, France. pp. 4169.
9. Nepstad, D.C., Verissimo, A., Alencar, A., Nobre, C.A. et al. 1999. Large-scale im-
poverishment of Amazonian forests by logging and fire. Nature 398, 505508.
10. Potter, C.S., Davidson, E., Klooster, S., Nepstad, D.C., Negreiros, G.H. and Brooks,
V. 1998. Regional application of an ecosystem production model for studies of bio-
geochemistry in the Brazilian Amazon. Global Change Biol. 4, 315333.
11. Sombroek, W.G. 1966. Amazon Soils. Pudoc. Wageningen. 303 pp.
12. Guerra, A.T. 1959. Geografia do Brazil Grande Regio Norte. IBGE-CNG, Biblioteca
Geogrfica Brasileira. Rio de Janeiro Vol. 1, Ser. A. Publ. 15, 421 pp. (In Portugese).
13. DNAEE 1995. Isoietas mdias annuais do Brasil [map at scale1:13 000 000], Ministrio
de Minas e Energia, Brasilia. (In Portugese).
14. SUDAM 1984. Projeto de Hidrologia e Climatologia da Amaznia, Atlas Climatolgico
da Amaznia Brasileira. SUDAM, Belm, Brazil. (In Portugese).
15. Salati, E. and Marques, J. 1983. Climatology of the Amazon region. In: The Amazon;
Climatology and Landscape Ecology of a Mighty Tropical River and its Basin. Sioli,
H. (ed.). Dr. W. Junk Publishers, Dordrecht/Boston. pp.85126. [See also: Salati, E.,
Marques, J. and Molion, L.C.B. 1978. Orgem e distribuio das chuvas na Amaznia.
Interciencia 3, 200206]. (In Portugese).
16. Bastos, T.X., Rocha, E.J.P., Rolin, P.A.M., Diniz, T.D.S., Santos, E.C.R., Nobre, R.A.A.
and Cutrim, R.C. 1986. O estado atual dos conhecimentos do clima da Amaznia
Brasileira com finalidade agricola. Proc. Primeiro Symposio do Trpico Hmido. Vol.
I, EMBRAPA-CPATU, Belm. (In Portugese).
17. Dickinson, R.E. and Virji, H. 1987. Climate change in the humid tropics, especially
Amazonia, over the last twenty thousand years. In: The Geophysiology of Amazonia;
Vegetation and Climate Interactions. Dickinson, R.E. (ed.). John Wiley and Sons,
Chichester, UK pp. 91101.
18. Fisch, G., Marengo, J.A. and Nobre, C.A. 1998. Uma reviso geral sobre o clima da
Amaznia. Acta Amaznica 28, 101126. (In Portugese).
19. Sombroek, W.G. 1962. Reconnaissance soil survey of the Guam-Imperatriz Area (area
along the upper part of the Belm-Brasilia highway). Mimeo report of SPVEA/SUDAM,
Belm and FAO, Rome.
20. Richey, J.E., Nobre, C. and Deser, C. 1989. Amazon river discharge and climate vari-
ability: 1903 to 1985. Science 246, 101103.
21. Courtesy Dr. Ren Levi Aguiar, Secretaria Municipal do Meio Ambiente. Manaus, Bra-
zil.
22. FAO 1985. Agroclimatological Data for Latin America and the Caribbean. Plant Pro-
duction and Protection Series 24. FAO, Rome.
23. FAO 1998. ECOCROP, the Crop Environmental Requirements Database, and the Crop
Environment Response Database. Land & Water Digital Media Series 4, FAO, Rome.
24. Smith, N.J.H., Serro, E.A.S.S., Alvim, P.T. and Falesi, I.C. 1995. Amazonia: Resil-
iency and Dynamism of the Land and its People. United Nations University Press, To-
kyo. 253 pp.
25. OIE 2000. Handistatus II Database on Animal Health World-wide. Office International
de Epizooties, Paris.
26. Sutmller, P., Abreu, A.V., Van der Grift, J. and Sombroek, W.G. 1963. Mineral Im-
balances in Cattle in the Amazon Valley. Communication 53. Royal Tropical Institute,
Amsterdam. 135 pp.
27. IAC 1991. Zoneamento Agroclimtico da Heveicultura no Brazil, relatorio final, projeto
FAO/UNDP/BRA/89/006. FAO, Rome and MINTER/SAE, Brasilia. (In Portugese).
28. Heinsdijk, D. and Miranda Bastos, A. 1963. Inventarios Florestais na Amaznia.
Ministrio de Agricultura, Servio Florestal, Setor de Inventrios Florestais. Bol. 6.
Rio de Janeiro. (In Portugese).
29. RADAMBRASIL 19721978. Levantamento de Recursos Naturais. Vol. 118.
Ministerio das Minas e Energia, Dep. Nac. da Produo Mineral. Rio de Janeiro. (In
Portugese).
30. Brown, S. and Lugo, A. 1992. Above ground biomass estimates for tropical moist for-
ests of the Brazilian Amazon. Interciencia 17, 818, 6061.
31. Nelson, B.W. and Irmo, M.N. 1998. Fire penetration in standing Amazon forests. Proc.
IXth Brazilian Remote Sensing Congress, Santos, SP, 1318 Sept. 1998.
32. Bale, W. 1989. The culture of the Amazon forests. Adv. Econ. Bot. 7, 121.
33. Bale, W. 1988. Indigenous adaptation to Amazonian palm forests. Principes 32, 47
54.
34. Higuchi, N., Santos, J., Ribeiro, R. and Biot, Y. 1997. Base Cientifico para Manejo
Florestal Sustentavel. Draft final report BIONTE project, INPA, Manaus (final ver-
sion in press at DFID, UK). (In Portugese).
35. Terborgh, J. 1992. Diversity and the Tropical Rain Forest. Scientific American Library
Series 38. W.H. Freeman & Co., New York.
36. Nepstad, D.C., Jipp, P., Moutinho, P., Negreiros, G. and Vieira, G. 1995. Forest re-
covery following pasture abandonment in Amazonia: Canopy seasonality, fire resist-
ance and ants. In: Evaluating and Monitoring the Health of Large-scale Ecosystems.
Rapport, D. (ed.). Springer Verlag, New York. pp 333349. See also: Uhl, C. and
Kauffman, J.B. 1990. Deforestation, fire susceptibility and potential tree responses to
fire in the eastern Amazon. Ecology 71, 437449.
37. Grace, J., Lloyd, L., McIntyre, L., Miranda, A.C., Meir, P. and Miranda, H.C. 1996.
Carbon dioxide flux over Amazonian rainforest in Rondnia. In: Amazonian Defor-
estation and Climate. Gash, J.H.C. et al. (eds). John Wiley & Sons, Chichester, pp.
307318.
38. Grace, J. et al. 1999. Carbon fluxes and productivity of tropical rainforests. In: Terres-
trial Global Productivity: Past, Present and Future. Mooney, H. et al. (eds). Academic
Press, San Diego.
39. Phillips, O.L., Malhi, Y., Higuchi, N., Laurance, W.F., Niez, P.V., Vsquez, R.M.,
Laurance, S., Ferreira, L.G., Stern, M., Brown, S. and Grace, J. 1998. Changes in the
carbon balance of tropical forests: Evidence from long-term plots. Science 282, 439
442.
40. LBA 1996. The Large Scale Biosphere-Atmosphere Experiment in Amazonia (LBA):
Concise Experimental Plan. INPE, Cachoeira Paulista, SP, Brazil. 44 pp.
396 Royal Swedish Academy of Sciences 2001
http://www.ambio.kva.se
41. Morais, J.L., Cerri, C.C., Melillo, J.M., Kicklighter, D., Neil, C., Skole, D.L. and
Steudler, P.A. 1995. Soil carbon stocks of the Brazilian Amazon Basin. Soil Sci. Soc.
Am. J. 59, 244246.
42. Sombroek, W.G. 1992. Biomass and carbon storage in the Amazon ecosystems.
Interciencia 17, 269271.
43. Sombroek, W.G., Fearnside, P.M. and Cravo, M. 1999. Geographic assessment of car-
bon stored in Amazonian terrestrial ecosystems and their soils in particular. In: Global
Climate Change and Tropical Ecosystems. Lal, R. et al. (eds). CRC-Lewis Publishers,
Boca Raton, USA pp 375389 + insert page.
44. Batjes, N.H. and Dijkshoorn, J.H. 1999. Carbon and nitrogen stocks in the soils of the
Amazon region. Geoderma 89, 273286.
45. Tian, H., Melillo, J.M., Kicklighter, D.W., McGuire, A.D., Helfrich, J.V.K., Moore,
B. and J. Vrsmarty, Ch.J. 1998. Effect of inter-annual variability on carbon storage
in Amazonian ecosystems. Nature 396, 664667.
46. Tomasella, J. and Hodnett, M.G. 1998. Estimating soil water retention characteristics
from limited data in Brazilian Amazonia. Soil Sci. 163, 90202.
47. Sombroek, W.G. 2000. Amazon land forms and soils in relation to biological diver-
sity. Acta Amazonica 30, 81100.
48. Van den Berg, M., Klamt, E., Reeuwijk, L.P. and Sombroek, W.G. 1997. Pedotransfer
functions for the estimation of moisture retention characteristics of Ferralsols and re-
lated soils. Geoderma 78, 161180.
49. Nepstad, D.C. et al.1994. The role of deep roots in the hydrological and carbon cycles
of Amazonian forests and pastures. Nature 372, 666669.
50. Jipp, P., Nepstad, D., Cassle, K. and Carvalho, C.R. 1998. Deep soil moisture storage
and transpiration in forests and pastures of seasonally-dry Amazonia. Climate Change
39, 395412.
51. Hodnett, M.G., Oyama, M.D., Tomasella, J. and Marquez, A.O. 1996. Comparison of
long-term soil water storage behaviour under pasture and forest in three areas of
Amazonia. In: Amazonian Deforestation and Climate. Gash, J.H.C. et al. (eds). John
Wiley & Sons. Chichester, UK. pp 5777.
52. Marchesi, G. 1975. Tra Fiumi e Foreste. Instituto Storico Salesiano, Roma. 2631. (In
Italian).
53. Anderson, A.B. 1981. White-sand vegetation of Brazilian Amazonia. Biotropica 13,
199210.
54. Sanford, R.L., Saldarriaga, J., Clark, K.E., Uhl, C. and Herrera, R. 1985. Amazon rain-
forest fires. Science 227, 5355.
55. Meggers, B.J. 1994. Mega El Nio events in Amazonia during the past two millenia.
Climate Change 28, 321338.
56. Meggers, B.J. 1984. The indigenous people of Amazonia, their cultures, land use pat-
terns and effects on the landscape and biota. In: The Amazon, Limnology and Land-
scape Ecology. Sioli,H (ed.). Dr W. Junk Publishers, Dordrecht/Boston, pp 627648.
57. Heckenberger, M.J., Neves, E.G. and Peterson, J.B. 1998. De onde surgem os modelos?
As origens e expanses Tupi na Amazonia Central. Rev. Antropol. 41, 6996.
Dep.Antropologia, USP, So Paulo. (In Portugese).
58. The author wishes to thank the ANEEL organisation in Brasilia (Eng. Vinicius de S
e Benevides; Eng. Valdemar Guimares) for providing all digitized monthly data of
its pluviometric stations in the Brazilian Amazon region. He also wishes to acknowl-
edge the critical reading of an earlier draft text, with valuable suggestions, by Dr.
Therezinha Xavier Bastos of EMBRAPA in Belm, Prof. Jan Sevink of the University
of Amsterdam / Tropenbos-Araracuara, and Ir Johannes van Leeuwen of INPA, Manaus.
59. A CD-Rom with the digital version of the four maps may be obtained from the GTZ
office at IPAAM, Manaus <gtz@ipaam.br>
60. First submitted 14 March 2000. Accepted for publication after revision 7 Feb. 2001.
Wim Sombroek holds a PhD in agricultural and environ-
mental sciences from Wageningen University. From 1959 to
1979 he participated in tropical soil survey and land
evaluation projects in Latin America (Brazil, Uruguay) and
Africa (Nigeria, Kenya). From 1979 till 1991 he was director
of the International Soil Reference and Information Centre
(ISRIC) in Wageningen and during those years he also
served as Secretary-General of the International Society of
Soil Science (ISSS), a scientific associate of ICSU. In 1991,
he became Director of the Land and Water Division of the
Food and Agriculture of the United Nations (FAO) and was
its focal point on Climate Change matters. He retired from
the FAO in 1996 and then accepted an invitation from GTZ
to become a team member/consultant in a Manaus-based
project on natural resources policies, within the Pilot
Programme for the Protection of the Brazilian Tropical
Forests of the G7 countries and Brazil (PP/G7). His address:
ISRIC, P.O. Box 353, 6700 AJ Wageningen, The Netherlands.
E-mail: sombroek@isric.nl
Ambio Vol. 30 No. 7, Nov. 2001