Farming, fishing, and fire in the history of the upper Río Negro region of Venezuela

Human Ecology, Vol. 15, No. 1, 1987

Farming, Fishing, and Fire in the History of the Upper Rio Negro Region of Venezuela

Kathleen Clark 1 and Christopher Uhl 2.3

Studies o f Rfo Negro subsistence farming and fishing activities are used to estimate the human carrying capacity for the region and the likely pattern o f human land-use during prehistory. Ceramic evidence suggests human presence in the region more than 3000 years ago. Traditional farming is labor intensive and relatively unproductive. Nevertheless, farmers achieve an energy return o f 15.2:1, and produce 2600 kcal per work hour. Fish are the major protein source, but fish catch per unit o f effort and fish yield per hectare o f floodplain are very low; fishermen are probably exploiting local fish resources very close to their limit. The low human population density would suggest that the R[o Negro forest has been relatively undisturbed. Never- theless, charcoal is widespread and abundant in forest soils. This charcoal is probably f rom anthropogenic or natural wildfires. These results suggest a much more complex history for Amazonia than previously thought.

KEY WORDS: Amazon; fire; fishing; agriculture; energetics.

I N T R O D U C T I O N

The increasing destruction of lowland rain forests in Amazonia as farming, ranching, and wood-harvesting activities proliferate is a source of widespread concern (Goodland and Irwin, 1975; Pires and Prance, 1977; Na- tional Academy of Sciences, 1980; Fearnside, 1982). The capacity of forest vegetation to recover from such large-scale disturbance is not well understood, and considerable research effort has been directed at assessing the long-term ira-

~K. Clark is a free-lance biologist residing in Lima, Peru; mailing address: c/o Agency for International Development, Lima, Peru.

2Assistant Professor of Biology, The Pennsylvania State University, University Park, Penn- sylvania 16802.

3To whom all reprint requests should be addressed.

1

0300-7839/87/0300-0001505.00/0 �9 1987 Plenum Publishing Corporation

2 Clark and Uhl

pact of various land-use practices (Uhl, 1982). At the same time, it has become increasingly evident that large parts of the Amazon Basin have been disturbed in the past, and that much of what was previously considered virgin primary rain forest is, in fact, regrowth forest (Sanford, Saldarriaga, Clark, Uhl, and Herrera, 1985; Saldarriaga, 1985; Saldarriaga and West, 1986). It is of interest, then, t o consider the landscape history of Amazonia and, in particular, the nature of past human and natural disturbance events from which the forest has subsequently recovered.

We will present a picture of Amazon history that is much more complex than the usual steady-state view. It is a history composed mainly of small agricultural disturbances, widely separated in time and space, but punctuated by occasional, large-scale forest fires. Our evidence comes from studies in the Upper Rio Negro region of Venezuela and Colombia, a region with abundant evidence of disturbance by fire. This is a part of Amazonia where steady- state conditions might have been expected to prevail because of the high, relatively aseasonal precipitation (Salati, Mfirques, and Molion, 1978) and the low density of human settlement (prehistoric human populations are also thought to have been sparse; Meggers, 1971; Janzen, 1974; Gross, 1975).

We will first review the evidence of historical and contemporary fire disturbance in the Upper Rio Negro. Then we will use our studies of current subsistence farming and fishing activities to estimate regional human carrying capacity and to reconstruct the likely pattern of human land use and disturbance during prehistory.

FIRE DISTURBANCE: HISTORICAL AND MODERN

Contemporary disturbance of Amazonian rain forest vegetation generally involves fire. This is so because burning is the easiest way to clear land for farming or pasture, and because the ash formed serves as a fertilizer sup- plement for crop production on what is otherwise relatively infertile soil (Fittkau, Junk, Klinge, and Sioli, 1975; Wambeke, 1978; Herrera, Jordan, Klinge, and Medina, 1978). Cleared land, especially pasture, is often periodically reburned in an effort to eliminate invading weedy vegetation (Hecht, 1981). Even where the initial forest disturbance does not involve the use of fire as a tool, it may increase the likelihood of accidental forest fire. For example, the selective harvesting of timber trees opens the forest canopy and leaves sufficient downed slash to permit the spread of ground fires into rain forest in eastern Amazonia (Uh 1 and Buschbacher, 1985). It has become clear that fire poses a threat to rain forest vegetation which previously had been considered largely immune to burning (Malingreau, Stephens, and Fellows, 1985).

Fire disturbance occurred during prehistory as well. In the Upper Rio Negro region (Fig. 1) of southern Venezuela, there is considerable evidence of

History of the Upper Rio Negro Region of Venezuela

R , Orinoco

COLOMBIA .S. Fernando de Atabop~

1 0 ~

I l 200

N

R, Guainia

R,Ne

SAN CARLOS de RIO NEGRO

BRAZIL

Fig. 1. The study region (stippled) consisted of the blackwater drainage area of the Upper Negro, Guainia, and Casiquiare Rivers (total area = 7,000 km~). Agricultural activities are concentrated along the river margins.

prehistoric fire disturbance (Sanford et al., 1985; Saldarriaga and West, 1986). Soil charcoal is widespread and abundant in the area and is found in several soil/vegetation types (see Klinge, Medina, and Herrera, 1977 and Uhl and Mur- phy, 1981a for a discussion of Upper Rio Negro soil and vegetation types). Radi- ocarbon dating of soil charcoal samples (Sanford et al., 1985; Saldarriaga, 1985) indicates that numerous fires have occurred during the past 6000 years. However, most of the dated samples are less than 2000 years old. While few data are available, it is interesting that the radiocarbon dates reported by San- ford et al. (1985) and by Saldarriaga (1985) correspond with what are believed to have been dry episodes during recent Amazon history (Fig. 2). These hypothesized dry periods are based upon palynological evidence from the Amazon basin and peripheral areas (Wijmstra and Van der Hammen, 1966; Van der Hammen 1972; Absy, 1982; Markgraf and Bradbury, 1982). These authors indicate dry episodes from 6000-4000 BP, from 2700-2000 BP, and at about 1500 BP, 1200 BP, 700 BP, and 400 BP. Near San Carlos de Rio Negro, of nine forest sites from which charcoal has been dated (Sanford et al., t985; Saldarri-

4 Clark and Uhl

5 U- o 4 n . ' ( D ww 3 m ~

z I

0 0 0 o

t,.-

o

0. ~ o_

Q I~ a

0 0 o o

u.I 0..

I1:

L N M ~ g g g g g g g g g g g g g

& 8 8 ,!, d ,& d,,~ d, 8 d, 6 8 & ~ ~ 1 7 6 0 o 0 o o 0 8 o o o

-

/ / m O O @

i O O

YEARS BEFORE PRESENT

Fig. 2. Distribution of radiocarbon dates for soil charcoal found at nine primary forest sites in the Upper Rio Negro, Venezuela (Sanford et al., 1985; and Saldarriaga, 1985; Saldarriaga and West, 1986).

aga and West, 1986), five contain charcoal 1100-1300 years old and five have charcoal 1500-1700 years old.

The distribution of charcoal across the Upper Rio Negro landscape (Fig. 3) suggests that two kinds of fire disturbance have occurred: agricultral and wildfires. Charcoal is particularly common (86~ of sampled sites; Sanford et al., 1985) in tierra firme Oxisols and Ultisols, the soil types utilized for shifting cultivation by the region's current population. Past agricultural activity is a likely source of much of the charcoal found in tierra firme soils. Charcoal is also present, although less common, in the hydromorphic clay soils which underlie the seasonally flooded igap6 forests (36~ of sampled sites; Sanford et al., 1985) and in the white sand Spodosols (25~ of sampled sites) which support caatinga vegetation. Neither igap6 nor caatinga is ever cut and burned by the modern inhabitants of the Upper Rio Negro, and it is highly unlikely that the pre-Colombian population farmed such sites either. Thus, it appears that the charcoal in caatinga and igap6 sites originated when standing forest was swept by fire.

At present, the Upper Rio Negro region is a relatively wet and aseasonal part of the Amazon basin. It receives 3500 mm of precipitation annually, and all months average more than 200 mm of rainfall (Heuveldop, 1980).

,=,,

I'-

"1"

(..9

'l-

r,r

LU

_>

n,.

z ,,<

uJ

iJJ

:>

o m

i-

I W

I

GA

PO

F

OR

ES

T

40-

20-

TIE

RR

A

FIR

ME

F

OR

ES

T

CA

AT

ING

A

FO

RE

ST

t IS

PE

CIE

S

DO

MIN

AN

T

MIX

ED

S

PE

CIE

S

I T

IER

RA

F

IRM

E

TIE

RR

E

FIR

ME

1

HIG

H

CA

AT

ING

A

LOW

C

AA

TIN

GA

B

AN

A

i r ~

'~ .

....

.

0-4

< -

-

1000

M

ETER

S ->

Fig

. 3.

A s

chem

atic

rep

rese

ntat

ion

of

the

rela

tion

bet

wee

n hy

drol

ogy,

veg

etat

ion,

and

soi

ls i

n th

e U

pp

er R

io N

egro

reg

ion

of

Ven

ezue

la.

Are

as c

lose

to

riv

ers

and

stre

ams

cont

ain

hydr

omor

phic

cla

y so

ils

and

supp

ort

igap

6 (s

easo

nall

y fl

oode

d) f

ores

t. M

ovin

g aw

ay f

rom

riv

ers,

the

veg

etat

ion

chan

ges

to e

ithe

r ca

atin

ga o

r ti

erra

fir

me

fore

st.

Caa

ting

a fo

rest

are

as d

o no

t ex

peri

ence

ann

ual

floo

ding

and

con

tain

san

dy s

oils

(S

podo

sols

). H

igh

caat

inga

oc

curs

whe

re t

he s

and

is f

ine

grai

ned

and

the

wat

er t

able

rem

ains

clo

se t

o (w

ithi

n 10

0 cm

of)

the

soi

l su

rfac

e th

roug

hout

the

yea

r; l

ow c

aati

nga

and

bana

occ

ur o

n so

mew

hat

mor

e el

evat

ed,

bett

er-d

rain

ed s

ites

con

tain

ing

coar

se s

ands

. W

ith

furt

her

incr

ease

s in

ele

vati

on a

bove

the

mea

n ri

ver

leve

l, th

e cl

ay c

onte

nt o

f th

e su

rfac

e so

il in

crea

ses

and

oxic

con

cent

rati

ons

are

som

etim

es p

rese

nt i

n th

e lo

wer

soi

l ho

rizo

ns.

Tie

rra

firm

e fo

rest

occ

urs

in

thes

e hi

gher

sit

es e

ithe

r as

leg

ume-

dom

inat

ed f

ores

t on

Ult

isol

s or

mix

ed-d

omin

ant

stan

ds i

n O

xiso

ls.

r C

~~

6 C l a r k a n d U h l

Z 0

0 LU r r Q. >-

l- Z 0

Z

w

(/3 >-

a

123

W

Z Z

W

5 0 0

4 0 0

300

200

I00

o ~ ~ t SAN CARLOS i I

- / ~ SAN FERNANDO o---o

..o

o~ / ,

o- - -o ~ l I I I I I I I I I I I

.J 20

.J

z

~ m

Vi

5

.o .o

~ t ~ -o""

\ /o

i i i i i i i i i i i i

J F M A M d d A S 0 N D

Fig. 4. Mean monthly rainfall (top), and mean monthly dry days (days receiving ~ 1 mm of rain) (bottom), for San Carlos de Rio Negro and San Fernando de Atabapo, Venezuela. Monthly means are calculated for22 years (San Carlos) and 10 years (San Fernando).

These conditions are such that forest fires are extremely rare, and fire very seldom escapes human control. In nearby regions, forest fires have been reported to occur only on remote mountaintops with low vegetation and thin soils (Spruce, 1908; Gleason, 1931). During 10 years of study in the Upper Rio Negro we have observed only two instances of (anthropogenic) forest fire. Both of these fires occured in low, open-canopied vegetation (one in low igap6 forest and one in poorly developed tierra firme vegetation on a rocky hilltop), and both occurred following 15-30 rainless days (a prolong- ed drought by Upper Rio Negro standards). These fires were confined to the litter and duff layer on the forest floor and caused little plant tissue destruction above ground, evidently killing the trees by heat damage to the roots. Local inhabitants have reported other instances of such fires.

While ground fires are currently small ( < 20 ha) and infrequent in the Upper Rio Negro, they are more common and more extensive about 250 km to the north of San Carlos near the town of San Fernando de Atabapo (see Fig. 1) where climatological conditions are more favorable for the spread


of forest fire. Yearly rainfall in the vicinity of San Fernando is quite high (2900 mm) but it is very seasonal in distribution as compared to that in the San Carlos area (Fig. 4, top). At San Fernando, three consecutive months receive less than 100 mm of rainfall. More importantly, at San Fernando, the number of dry days (those with _<_ 1 mm of precipitation) per month exceeds 20 during six consecutive months; while near San Carlos, no month averages as many as 20 dry days (Fig. 4, bottom). These circumstances permit forest fires to kill hundreds of hectares at a time. Indeed, in the San Fer- nando region, wildfires can destroy not only low and open vegetation types, but taller, closed-canopy tierra firme forest as well. If, during Holocene dry episodes, climatological conditions in the Upper Rio Negro approximated those we see today near San Fernando, it is conceivable that considerable tracts of standing forest might have succumbed to forest fires. 3

H U M A N PRESENCE IN U P P E R RIO NEGRO

Very little is known of the pre-Colombian inhabitants of Amazonia. In- deed, the number and distribution of rain forest Indians prior to the conquest is a matter of considerable debate (Denevan, 1976; Hemming, 1978). Following the conquest, the native population of Amazonia was rapidly reduced by introduced epidemic diseases and by slave raiding (Ribeiro, 1970; Hemming, 1978), and it has subsequently proved difficult to reconstruct the way of life of pre-Colombia populations or to assess their impact on the Amazon landscape.

There is evidence that humans have been present in the Upper Rio Negro for several thousand years. Ceramic pieces, stone tools, and other artifacts are common, and are often found in association with terrapreta Anthrosols. These soils are believed to represent the cumulative effect of long-term or episodic human occupation (Smith, 1980). Terrapreta soil and ceramics are to be found at most high, rocky, tierra firme points along the river banks (that is, at suita- ble home or village sites separated by stretches of low-lying igap6 forest). On the lower Casiquiare, we found 11 terra preta sites in a 40-km stretch of river (Clark and Sanford, unpublished). A ceramic sherd from the region was dated by the thermoluminescence method at 3750 ybp _+ 10% (Sanford et al. 1985; Saldarriaga and West, 1986).

3Whether past forest fires in the Upper Rio Negro region were caused by lightning or by human activity or both is unclear. Lightning strikes are frequent in the Amazon region today (Komarek, 1971), but the accompanying heavy rainfall makes the subsequent spread of forest wildfires uncommon. We believe most forest fires would have been of anthropogenic origin.

8 Clark and Uhl

Despite the abundance of human remains, no archaeological investigations have been undertaken in the region. Because the Rio Negro and the Casiquiare form a navigable fluvial link between the Orinoco and Amazon Rivers, archeologists have speculated that these rivers have been an important navigation route for human populations moving between the two basins (Meggers, 1982; Migliazza, 1982). The subsistence activities of these early inhabitants are unknown, however, and can only be guessed from what is known of other lowland tropical forest Indians and of the current inhabitants of the Upper Rio Negro region. By considering contemporary subsistence behavior, we can gain some insight into the impact of the pre-Colombian inhabitants on Rio Negro forests. The modern inhabitants 4 of the area depend upon shifting cultivation and fishing to provide most of their diet. We have concentrated on these activities in our investigations of subsistence strategies.

Traditional Slash and Burn Agriculture

In the Upper Rio Negro, as elsewhere in Amazonia, shifting cultivation begins with the felling of small (0.5-1 ha) patches of mature or late second-growth forest, generally at the beginning of the dry season (September-November). After allowing the wood to dry for 1-2 months, the plots, called "conucos," are burned. Nutrients from the cut vegetation are released in the burn, and thereafter through the slow decomposition of unburned logs. Shortly following the burn, the conuco is planted. A variety of crops are cultivated, but the overwhelming dominant, generally accounting for more than 9007o of edible crop production, is cassava, Manihot esculen- ta Crantz. ~ Cassava produces edible tubers high in carbohydrates (32-3507o) but very low in protein (0.7-2.6070) (Doku, 1969). The roots mature in about 12 months at San Carlos, but can be left in the ground for longer periods. The crop is periodically harvested in small quantities. Two, or sometimes three crops of cassava are produced in the conuco, after which levels of available nutrients in the soil are sufficiently reduced to make further cultivation unrewarding. Subsistence farmers must, therefore, clear new areas of forest approximately every 2 years. Later clearings are frequently made ad-

4The modern inhabitants of the region are mostly detribalized descendants of Arawakan groups: Bar6, Baniwa, Guaraquena, and Wakuenai (Curripaco). Many are speakers of Geral, a trade language formerly common in the Brazilian Amazon. Some mestization has occurred, but the region remains isolated and uncolonized.

5No grain crops are grown, and beans are but a very minor component of a very few conucos. Soil conditions do not permit the cultivation of such crops and presumably prevented their cultivation in the past as well. Indeed, even bananas do poorly on Rio Negro soils, and are only a minor crop.

History of the Upper Rdo Negro Region of Venezuela 9

jacent to the original one, such that a farm site eventually becomes a patchwork of newly felled, actively farmed, and abandoned plots.

Energetics

The cultivation of cassava requires a great deal of arduous and time- consuming human labor. Conucos are felled with axes and machetes, weeds are controlled by hand clearing, the crop is harvested by hand, and travel to and from the conucos is usually on foot or by canoe. Harvested cassava roots are peeled, grated, and soaked. This procedure provides a favorable environment for linase, a naturally occurring enzyme, to break down the poisonous glucosides in the roots to glucose, acetate, and hydrocyanic acid. Subsequent squeezing and pressing removes the water-soluble hydrocyanic acid. Sieving and baking produce a coarse, dry, granular flour called "mafioco." Other cassava products include large flat cakes called "casabe," and "almid6n," the purified starch from cassava tubers.

We have estimated that 3230 hours of field and processing work are required to establish, plant, harvest, and process the yield from a 1-hectare area during a two-crop cycle at San Carlos (Fig. 5; Uhl and Murphy, 1981b). This time expenditure requires approximately 606,000 kcal of human energy; slightly more energy (53%) is spent in field work than in root processing (47%). Although human labor inputs are high in San Carlos conucos, the energy yield (as processed food) is estimated at 2600 kcal per hour of work. Thus, the Rio Negro farmer can satisfy his daily caloric needs (2500 kcal) with about 1 hour of work a day; two adults each working 2 hours a day can provide the needed calories for themselves and four children. 6

We have also estimated the amount of active cropland required per person. Assuming an average daily intake of 2500 kcal per capita, and that 80% (2000) of these calories are provided by cassava products, then 0.13 ha of cassava crop will be needed by each adult inhabitant each year.

Regional Carrying Capacity Based on Caloric Considerations

The potential agricultural production of the Upper Rio Negro region under shifting cultivation can be determined through estimates of crop pro-

6Nevertheless, calorie production at San Carlos is generally less efficient than that reported elsewhere in lowland South America. Werner, Flowers, Ritter, and Gross, (1979) surveyed subsistence productivity in four Brazilian Indian groups. Only one group (the Bororo, living in a degraded cerrado region) was less efficient than San Carlos in agricultural production (1550 kcal hr-1). The other villages ranged from 5,000-17,500 kcal hr -1.

I0 Clark and Uhl

Cutting forest 48,496 kcal 8.0 %

Burning 2,509 kcal 0.4 %

Gathering stem stocks ]0,197 keal 1.7%

Planting 42,320 kcal 7.0%

Cleaning and weeding 101,589 kcal ]6.7% Harvesting 81,578 kcal 13.4%

Peeling 94,439 kcal 15.6%

Grating and soaking 54,585 keel 9.0%

Squeezing and sifting 15,795 keal 2.6%

Fire making end baking 45,728 kcal 7.2% Traveling 111,630 kcal 18.4%

Fig. 5. A summary of the human energy inputs to cassava crop production for a 1-ha farm over a two- crop cycle in the Upper Rio Negro, Venezuela.

duction, land available for farming, and required fallow time. In this sec- tion, we discuss each o f these factors and conclude by estimating the potential cassava yield for the region, and the theoretical human population that it could support.

Conuco Crop Production. Estimates of conuco cassava crop produc- t ion in the Upper Rio Negro area are based on sampling in seven conucos all near the town of San Carlos de Rio Negro (Clark and Uhl, 1984). All conucos studied were planted and maintained by local farmers using traditional practices. Cassava yield estimates were made by first regressing mature cassava stem diameter on tuber weight. Then the number and diameter o f all cassava plants were determined in several 1 • 50 m transects in each of the seven conucos just prior to plant harvest. Average cassava root yield,

History of the Upper Rio Negro Region of Venezuela 11

based on this survey, was 4.7 t wet weight ha -1, and is taken to be a represen- tative value for the region. 7 Petriceks (1968) estimated average cassava production to be 7 t ha -1 for northern Venezuela. The world average cassava production is 8.4 t ha -~ crop -1 (Kay, 1973). Kay indicates, however, that production can drop to 3 t ha -~ on poor soils. The extreme infertility of the San Carlos soils (Herrera et al., 1978) undoubtedly explains the relatively low cassava production for this region.

Available Land for Traditional Agriculture. We have traveled exten- sively through the Upper Negro-Guairda-Casiquiare region, visiting numerous conucos, and our overall impression has been that farming practices and yields are fairly uniform. The results of cassava production studies at San Carlos de Rio Negro apply to the entire Negro-Guainla-Casiquiare region of Venezuela and Colombia (see Fig. 1). This region includes that land whose streams drain into the Guainia and Negro Rivers (from the Colombia border on the north to the Brazilian border on the south) and into the Casiquiare River f rom its mouth up to the mouth of the Pacimoni River. This is a total river length of about 250 km and an area of about 7000 km 2. However, within this region, only the area adjacent to major rivers represents a potential settlement zone because fish and game are most abundant in the floodplain along these rivers. The rivers provide access to the proteins lacking in conuco crops. Hence, we consider the practically usable land to lie within a 5-km wide strip along each bank of the major rivers and calculate the theoretically usable land area to be about 2500 km 2. Conucos can only be cut f rom parcels of tierra firme forest growing on non-flooded Oxisols or Ultisols (Fig. 3). The caatinga forest complex, so common in this region, occurs on bleached white sands (Spodosols) which are unsuitable for agriculture. Igap6 forest sites, likewise, are unsuitable for agriculture because of seasonal flooding. We estimate that tierra firme vegetation accounts for roughly 30% of the land area in the region. Given these restrictions, about 750 km 2 are available within the region for agricultural use.

Fallow Time Required to Restore Site Productive Capacity. The third factor that must be considered in estimating regional cassava yield under shifting cultivation is the fallow period required between farming episodes. When conucos are abandoned, they are gradually colonized by fast-growing herbaceous and woody species, and later by pr imary forest species. As regeneration proceeds, nutrients lost during the farming episode by burn-

7Cassava yield was also measured directly in a 0.2-ha farm site over a 2-year period by weighing all roots. Root yields were 4.3 and 2.8 t ha -1 for years 1 and 2, respectively (Uhl and Murphy, 1981b). On numerous other occasions we have weighed cassava from small parcels as it was being harvested by native farmers. Yields, when extrapolated to 1 hectare, invariably ranged between 3-8 t. Hence, we feel that 4.7 t ha -1 is a reasonable average yield for this region.

12 Clark and Uhl

ing, leaching, and crop harvest, reaccumulate on the site. These nutrients are chiefly located in the regrowing plant biomass. Hence, the regrowth vegetation represents the fertilizer supply for the next farming effort.

An appropriate fallow period is difficult to determine. Farmers in tropical Africa and Asia usually employ fallows ranging from 5-15 years (Vasey, 1979). In the Upper Rio Negro, farmers occasionally farm land that has been fallow for only 10 years; however, such plots produce only one cassava crop and are then abandoned. Experienced local farmers suggest that two crops of cassava can be produced on land fallowed for about 25 years, but it is unlikely that such a cycle could be indefinitely repeated without eventual soil degrada- tion. Vasey (1979) points out that Amazon fallows often exceed 25 years.

Studies of the regrowth vegetation on 10- to 80-year-old abandoned agricultural sites in theUpper Rio Negro (Saldarriaga, 1985) indicate that plant biomass accumulates quite rapidly, in a nearly linear fashion, for the first 50 years of succession, but, thereafter, biomass accumulates very slowly. By 50 years, fallows have regained about 55% of the total biomass typical of the mature forest, but the most nutrient-rich biomass components, i.e., leaves, branches, and fine roots, have attained mature forest levels by this time. Given this pattern of biomass accumulation, 50-year fallows would ensure high nutrient stores for farming, but the benefit of extending fallows beyond 50 years would dimin- ish. Based on the infertility of the region's soils and our observations, we believe 50 years to be a minimum fallow period if the shifting agricultural cycle is to be repeated indefinitely.

Potential Cassava Yield for Region. Combining the estimates of cassava production, farmable land area, and fallow time allows an estimate of the potential regional cassava yield using traditional farming practices (Table I). An estimated 13,600 t wet weight of cassava could be produced per year in this 7000-km 2 area. It would supply sufficient calories for 21,600 people (at 2000 kcal day-l), or 3.1 persons per square kilometer (8.0 persons per square mile). This is an extraordinarily low carrying capacity compared to other tropical regions. Ooi Jin Bee (1958) provides a list of carrying capacities for nine tropical areas under shifting cultivation. Values range from 25-160 persons per square mile, with most of the figures above 80. Fearnside (1978) analyzed carrying capacity potential for the region along the Transamazon Highway in Brazil and concluded that as many as 60 people might be sup- ported per square mile.

Annual Level of Current Agricultural Disturbance

While the estimated carrying capacity (based upon caloric considerations only) of the Upper Rio Negro region is remarkably low, it is, in fact,


Table I. Potential Cassava Yield under Traditional Farming Practices in the Negro- Guainia-Casiquiare Region of Venezuela and Colombia Based on Considerations

of Crop Production, Land Area, and Fallow Time

13

Crop production Cassava root yield (wet weight) Caloric content of yield

Land area Total Negro-Guiania-Casiquiare region of

Venezuela and Colombia Total area within 5 km of major r.ivers Tierra firme, i.e., farmabte land,

within 5 km of rivers Farm-fallow cycle

Period of farm plot use Minimum fallow period

Regional yield Land area available each year for

farming Cassava yield for region (cassava wet

weight) Caloric content of yield

Carrying capacity Supportable population for region

given restrictions discussed in text Supportable population per km ~

4.7 t ha-' yr-' 5A8 x 106 kcal ha -I yr -t~

7000 km ~ 2500 kin:

750 km 2

2 years 50 years

29 km 2b

13,600 f 1585 • 101~ kcal

21,600 d 3.1 *

~In the processing of cassava, skins, fibers, and water are removed from the roots, reducing the raw crop average of 4.7 t ha -I yr-' to 1.5 t ha -* yr -~ of edible flour. The conversion factor of wet root weight to mafioco is 0.31 (Uhl and Murphy, 1981 b). One gram dry weight of cassava contains an estimated 3.76 kcal (Mon- taldo, 1977). Therefore, the cassava yield ha-' in kcat is 5.478 • 106 [(4.7) (1,000,000) (0.3 t) (3.76)].

bThere are 750 km 2 of available land for farming. Each hectare can be farmed for 2 years during a 52-year period. Therefore, there are 29 km ~ (2900 ha) of farmland available each year [(750) (2)/52 = 28.8 kin2].

~Cassava yield is estimated as 4.7 t wet weight ha -~. Hence, t3,600 t wet weight of cassava roots could be produced each year (2900 ha • 4.7 = 13,630).

aThis estimate assumes that daily per capita caloric intake is 2500 kcal and that 80% of this (2000 kcal) is supplied by cassava. Hence, 2000 kcal • 365 days = 730,000 kcal per person per year, and 1585 • 101~ kcal/7.3 • 10 ~ kcal = 21,600.

*Given the estimate of 7000 km 2 for the upper Rio Negro, Guainia, and Casi- quiare region, a population density of 3.1 persons per km 2 is possible (21,600 people/7000 km~) if only caloric considerations are taken into account.

a b o u t s e v e n t i m e s g r e a t e r t h a n t h e a c t u a l c u r r e n t p o p u l a t i o n ( T a b l e I I ) . N o r

is t h e r e r e a s o n t o b e l i e v e t h a t t h e p o p u l a t i o n h a s b e e n s u b s t a n t i a l l y g r e a t e r

a t a n y p o i n t d u r i n g t h e r e c e n t h i s t o r i c a l p e r i o d f o r w h i c h s o m e s c a n t y e v i d e n c e

is a v a i l a b l e ( V e n e z u e l a , C O D E S U R , 1973). W h i l e a b o u t 29 k m 2 o f f a r m l a n d

a r e a v a i l a b l e f o r c l e a r i n g e a c h y e a r , t h e r e g i o n ' s p o p u l a t i o n o f 3000 is c l e a r -

14 Clark and Uhl

Table II. Human Population of the Negro-Guainia- Casiquiare Region of Venezuela and Colombia"

Venezuela b Large towns (San Carlos and Maroa) 1216 Villages 1101

Colombia c Villages +_ 700

Total + 3000

aSee Fig. 1. bBased on an unpublished 1981 medical census by the Venezuelan Ministerio de Sanidad y Asistencia Social, Nticleo Rio Negro.

CNo census data available. Populations estimated from data on village size and typical regional household size.

ing only about 150 ha (at 0.13 ha adult year-l). To understand why the region remains sparsely populated and lightly farmed despite the availability of usable tierra firme land, it is necessary to consider another aspect of the subsistence economy: the acquisition of proteins.

Subsistence Fishing

While cassava is a good and reliable source of calories, it is a poor source of protein. The major protein source for the inhabitants of the Upper Rio Negro region is fish. Holmes (1981) found that the residents of small set- tlements along the river consumed about twice as much fish per person per day as game (24 vs. 13 g of protein). The inhabitants of the region's largest town, San Carlos de Rio Negro, consume somewhat less fresh fish because they are able to purchase imported protein foods (chieflycanned fish and meat) and because there is a small local production of chickens, pigs, and cows (Clark and Uhl, 1984). Even in San Carlos, however, locally caught fish is the single most important protein source, accounting for about 30-40% of the animal protein consumed.

Like slash and burn agriculture, fishing in the Venezuelan Rio Negro region is a small-scale, subsistence-level activity. The region's fishermen do have access to a variety of modern equipment, most importantly outboard motors, gasoline, 12-volt batteries, and headlamps. Despite the availability of such labor-saving equipment, fishing remains difficult and unrewarding by Amazon basin standards. The difficulty of obtaining protein by fishing is a good indicator of the general difficulty of protein acquisition in this oligotrophic black-water region.


Catch Composition

15

The composition of the fish catch in the Upper Rio Negro is very distinct from that reported from the central Amazon region (Smith, 1979; Goulding, 1981) or from the Orinoco drainage (Mago-Leccia, 1970, 1978). This is because several large-growing species of considerable economic importance elsewhere in lowland South America are notably absent from the Upper Rio Negro. Among these are the pirarucfi and aruanfi (Osteoglossidae), the bocachicos of the family Prochilodontidae, the large serrasalmids of the genus Colossoma, and a variety of large pimelodid catfishes. The absence of these fishes means that Upper Rio Negro fishermen must depend upon many smaller species.

The number of fish species regularly exploited and the size of the fish caught are good indications of the poverty of Upper Rio Negro fish resources. Based on an examination of 8900 fishes caught during sampling from 1979-1981 by the seven most important local fishing methods (spear, speargun, trot line, drift line, poison, set line, and trap) and representing some 500 man days of effort, it is possible to characterize in general terms the composition of the San Carlos fish catch (Table III). The hallmark of the San Carlos fishery is the large number of species exploited: by these seven methods more than 100 species in 24 families. The catch diversity of all methods except those which focus on large catfishes (trot line and drift line) is high. Smith (1979) described the catch of 12,850 fishes of 68 species caught by ten methods at Itacoatiara, Brazil, in the central Amazon. The catch diversity computed from his data is 1.16, while the overall diversity of the San Carlos catch was 3.72 (Table III). The Itacoatiara fishery is commercial, and the fishermen can depend to a large extent on just a few species. This is not possible at San Carlos where no single species or small group is sufficiently abundant to bear the major burden of fishing pressure. Similarly, Goulding (1981) found that five species made up 70~ (by weight) of the Rio Madeira (Brazil) commercial catch, and Bayley (1981) reported that three species accounted for 71% of the commercial landings at Manaus. At San Carlos, 15 species accounted for 70% of the catch weight. Moreover, the individual fishes caught are small, averaging only 198 g in the rainy season and 376 g during the dry season.

Catch Seasonality

Catch rates at San Carlos depend to a large extent on river height. The level of the river depends upon rainfall regimes to the north (in the upper Guainia of Colombia and the upper Orinoco of Venezuela) and therefore is markedly seasonal, whereas, precipitation in the immediate vicinity of San

7,

Tab

le 1

11. F

ish

Cat

ch a

t S

an C

arlo

s de

Rio

Neg

ro,

1979

-198

1 a

Nu

mb

er o

f T

ota

l M

ean

fis

h

Nu

mb

er o

f N

um

ber

of

fish

w

eigh

t (k

g)

wei

ght

(g)

fish

fam

ilie

s fi

sh s

peci

es

H ~

div

ersi

ty b

Rai

ny s

easo

n

5468

10

85.6

19

8 23

+_

100

3.

31

Dry

sea

son

35

00

1315

.4

376

17 c

71 e

3.13

T

ota

l 89

68

2401

.0

268

24

+

105

3.72

aBas

ed o

n s

amp

lin

g o

f th

ree

maj

or

rain

y s

easo

n a

nd

fo

ur

maj

or

dry

seas

on

fis

hing

met

ho

ds.

bH

~ =

S

han

no

n W

ien

er I

nd

ex =

E

(P

3 (In

Pi)

, w

here

Pi

=

per

cen

t ab

un

dan

ce o

f ea

ch s

peci

es.

"Not

inc

lude

d ar

e fi

shes

wei

ghin

g le

ss th

an 1

0 g

cap

ture

d b

y f

ish

po

iso

nin

g ("

barb

asco

");

thu

s, m

any

spe

cies

an

d s

ever

al f

amil

ies

(esp

ecia

lly

char

aco

ids)

are

no

t re

pres

ente

d.

ex~


Carlos (1 ~ lat.) is less clearly so. There is an average annual change of about 7 m in the river's height between the dry season low (February) and the rainy season peak (July). Despite the seasonal regularity of rise and fall, there is considerable variability from year to year in the timing of onset and the extent (temporal and spatial) of flooding.

During the rainy season (April-September), extensive areas of low-lying forest are inundated. As the water rises, many fish species abandon the river channel to exploit the food resources available within the flooded forest; moreover, many fishes also utilize the flooded forest as a spawning ground. Fishing during this period is poor because the fish are dispersed over a wide area and because travel and the use of fishing gear are difficult within the forest. During the dry season the fish are concentrated within a greatly reduced water volume and are therefore more accessible to fishermen.

Catch Per Unit of Effort

Monthly catch per unit of effort (kilos of raw fish per man hour of labor) by all fishing methods combined was calculated for the period April 1979-April 1981 (Fig. 6). Work time includes the hours spent traveling, fishing, building traps, and collecting bait, but does not include the value of boats, motors, gasoline, or fishing gear. The monthly catch rates are between 0.15-1.10 kg per man hour, with an overall mean of 0.62 kg hr- 1. Fishing is most productive during the driest months of the year when fish are concentrated in shallow, clear creeks, and during the early part of the rainy season, when fish are active and the cacure (trap) catch rate is high.

RIVER HEIGHT 1.2 . . . . . CATCH 16

g Lo 14 "~ Frl

# P: o.8 p'--Z\ ~ ^ /1'~ % . s ~ I I , " ,2 E -~ "l ,' ~ # l l \ l U I / ~m

0.6 I0 -7- i % - I o ,, 5o.4 , 8 7

'i_..,-" '%.,.-,J I I

A J A 0 O J M M J S N J M M

1979 1980 1981

Fig. 6. The relationship between fish catch rates and river level in the Upper Rio Negro, Venezuela.

18 Clark and Uhl

Compared to fish catch rates elsewhere in lowland South America, the San Carlos fishery is very poor. The catch per unit of effort is similar to that achieved by relatively unacculturated Indian groups which depend upon fish as a major protein source (Werner et al., 1979; Beckerman, 1980). However, San Carlos fishermen utilize technology unavailable to less ac- culturated Indians. The Bari, for instance, caught 0.42 kilos of fish per man hour without steel hooks or other introduced technology (Beckerman, 1980). The San Carlos catch per unit of effort is only one-third that of fishermen at Itacoatiara, Brazil, using similar small-scale fishing gears, e.g., trot lines, hand lines, and spears (Smith, 1979).

Protein

Since fish make their most important contribution to the diet as protein, it is convenient to convert the raw catch rate to grams of protein caught per man hour of labor. The edible portion of a typical fish from the Upper Rio Negro, taking into account the species composition of the catch, is 82% (Clark, unpublished data). Based on literature values (Jaff6, Nolberga, Embden, Garcia, Olivares, and Gross, 1956; Smith, 1979), the edible part is about 18% usable protein. There is no universally accepted minimum daily protein requirement. Fifty grams was chosen as a daily per capita protein requirement because the population includes a large number of growing children and pregnant or nursing women, because the population is heavily parasitized (Holmes, 1981), and because the local diet depends largely on root crops (principally cassava). At San Carlos, considering typical dry and rainy season catch rates, it would take from 20 minutes to 2�88 hours per capita per day to furnish the required daily protein (50 g) by fishing. It is unlikely that better rates of protein capture can be achieved by hunting in the Upper Rio Negro. Thus, fish capture rates can be taken as an estimate of the time investment required to provide adequate animal protein by current subsistence methods at San Carlos.

While catch per unit of effort is notably poor in the Venezuelan Rio Negro region, it is apparent that, except during the worst of the high water season, adequate protein can be acquired with relatively little investment of time. In this respect, the procurement of proteins is similar to the procurement of calories from the cultivation of cassava. However, the protein situation differs from that of calorie production in that, while a large reserve supply of usable farmland is available, there is not a large reserve of unex- ploited fish. Indeed, Upper Rio Negro fishermen are probably exploiting local fish resources very close to their limit. This is apparent from a consideration of the regional fish yield.


Regional Fish Yield

Fish consumption patterns in the Upper Rio Negro region depend upon the degree of involvement in the larger national culture and economy. In the more isolated, smaller villages, fish make up about two-thirds of the animal protein consumed (135 g of uncleaned fish per person per day; Holmes, 1981). The residents of the region's largest town, San Carlos de Rio Negro, on the other hand, were found to consume only about 83 g of fish per person per day (based on 295 random household interviews over a 14-week period). Fish consumption in the Venezuelan Rio Negro, and particularly in the town of San Carlos, is thus lower than that estimated for the Manaus population (155 g of uncleaned fish per person per day) by Shrimpton, Guigliano, and Rodrlgues, 1979; Bayley, 1981). This reflects the continued importance of game animals, and, in San Carlos, the availability of inex- pensive (government subsidized) imported foods and small local livestock production. Combining the estimates of daily per capita fish consumption with the regional population data (see Table II), we estimate the yearly regional fish catch to be about 125 t. s

It is instructive to view this regional catch in terms of fish yield per unit of area. For river fisheries, the areal unit of interest is hectares of floodplain. The estimated annual regional fish yield per hectare of floodplain for the Negro-Guainia-Casiquiare region of Venezuela and Colombia (see Fig. 1) is 6.6-13.2 kg ha -1 yr -1 (Table IV). This is an extraordinarily poor yield in comparison with other tropical floodplain river fisheries. Welcomme (1979) reviewed the available data on fish yield of tropical (chiefly African and Asian) floodplain rivers and found typical yields to be from 40-60 kg per hectare of floodplain per year. The Rio Madeira (Brazil) fishery described and evaluated by Goulding (1981) yielded 52 kg ha -1 yr -', and thus is well within the typical range of tropical rivers.

Bayley (1981) compared actual yields of several Amazonian rivers to theoretical yields based on Welcomme's (1976) model of African riverine fisheries production. He found that fish production in the lower Rio Negro was only one-half of the predicted theoretical yield. Using the same model, the theoretical catch for the Negro-Guainia-Casiquiare region would be about 530 t yr-', that is, more than four times the actual catch.

The Rio Negro is characterized by its clear but darkly stained water of low pH and extremely low mineral content (Sioli, 1956; Marlier, 1973). The dark coloration of the water derives from partially decomposed organic matter (humic acids) leached from the region's Spodosols (Klinge, 1967; Sioli,

8[(1801)(135) + (1216)(83)] (365) / (1000) 2 = 125.6.

20 Clark and Uhl

Table IV. Annual Fish Yield Per Hectare of Floodplain in the Negro- Guainia-Casiquiare Region of Venezuela and Colombia a

Negro-Guainia-Casiquiare drainage basin area b

Floodplain area c (0.5-1.0~ of basin area) Regional catch Annual catch per hectare of floodplain

19,000 km 2 95-190 km 2

125 t 6.6-13.2 kg ha -1 yr -t

aSee Fig. 1. bEstimated from regional map (1:1,000,000). CAssuming "normal" floodplain development (Welcomme, 1979); that is, a floodplain less than l~ of the basin area.

1975). Because of their low productivity, black-water rivers have been called "rivers of hunger" (Janzen, 1974). The low production has been attributed to several factors, including low nutrient content and light penetration (which reduce aquatic primary productivity), and the toxic effects of low pH and high concentrations of plant secondary (phenolic) compounds (Janzen, 1974). Where aquatic primary production is low, aquatic food chains are based primarily on inputs, e.g., insects and fruits, f rom the terrestrial ecosystem. In the Venezuelan Rio Negro, the opportunity for such terrestrial inputs is reduced due to the relatively small floodplain (Bayley, 1981) and the simple, unbraided nature of the river course. Further, the Venezuelan Rio Negro is free of large rapids which, elsewhere in the Amazon system, concentrate fishes making them more easily exploited by fishermen (Goulding, 1981).

Welcome (1976) suggested that low pH and low conductivity (as an indicator of inorganic nutrient content) tend to depress fish yields below the theoretical levels calculated from floodplain development. Both the pH (4.5-5.8) and the conductivity (5-14/~mhos) of the Rio Negro at San Carlos (from unpublished data of the Estaci6n de Hidrologia, Ministerio de Am- biente y de los Recursos Naturales Renovables, San Carlos de Rio Negro) are at the very lowest extreme of those parameters in the African rivers considered by Welcomme. The very poor fish yield in the Upper Rio Negro is thus likely to result from a combination of morphological (floodplain) and edaphic (nutrient) factors and is not (considering the very low catch per unit of effort) simply an indication of underexploitation. There is little, if any, room for expansion of the regional fish catch.

DISCUSSION

The agricultural carrying capacity (by traditional shifting cultivation) of the Negro-Guainla-Casiquiare region of Venezuela and Colombia is estimated at about 21,600 persons, but local animal resources could never provide sufficient protein for such a population. The lack of abundant

History of the Upper Rio Negro Region of Venezuela 2I

animal protein resources constitutes a severe constraint on the human population of the Upper Rio Negro region. The reputation of the Rio Negro as a "hunger river" is borne out by the low catch per unit of effort at San Carlos, the low yield per hectare of floodplain, and the dependence of fishermen on a large number of very small species. The annual fish yield (125 t) is sufficient to provide 100% of the dietary protein for a combined Colom- bian-Venezuelan population of about 1000 persons although, in fact, fish currently provide about one-third of the animal protein consumed in San Carlos and two-thirds of that consumed in the small villages (for a total population of about 3000 people).

Given the scarcity of game in the Upper Rio Negro region, the low protein content of conuco crops, and the irregular availability of~ild plant foods, it is difficult to imagine any combination of protein sources that would have permitted a regional population substantially greater than the 3000 present today. Indeed, since the current inhabitants are dependent to an increasing extent on imported foods (Clark and Uhl, 1984), it is even possible that today's population has exceeded the region's carrying capacity. The dependence on imported foods is partly attributable to development policies which have tended to draw the population into the region's two largest towns (Clark and Uhl, 1984) thereby concentrating subsistence activities in a reduced area.

To have made best use of the region's poor faunal resources, the precon- quest population of the Upper Rio Negro was almost certainly organized in small, perhaps mobile groups. As a result, disturbance of the forest by shifting cultivation would generally have been small in scale and well-dispersed in time and space. However, soil charcoal is widespread and abundant in the region.

One possible reason is that a small population produced a large cumulative effect over many thousands of years. However, a population of about 3000 persons would not require as much as the 5-kin wide strip of river margin land we have calculated as "practically usable." In fact, such a population would find sufficient agricukural land within 1 km of the river. Even if the hypothetical population were doubled to 6000, sufficient land would be found within 2 km of the bank. It is thus difficult to explain extensive burning many kilometers from the river. It is even more difficult to explain the charcoal found in non-agricultural igap6 and caatinga soils.

We believe that forest fires have disturbed far greater areas of the Upper Rio Negro than human agricultural activity. Moreover, the presence of abundant charcoal in RIo Negro soils is not an anomaly. We have done informal sampling in the Peruvian, Brazilian, and Ecuadorian Amazon and encounter charcoal everywhere we look. Amazon researchers from Empresa Brasileira de Pesquisa Agro Pecu&ia and Instituto Nacional de Pesquisas de Amazonia in Brazil concur that charcoal is common in the soils

22 Clark and Uhl

of the central and eastern Amazon. Indeed, it appears to be much more difficult to find unburned sites than to find sites with charcoal.

We speculate that 10% or less of the Upper Rio Negro forest land area has been used for slash and burn agriculture over the last 10,000 years, but that the vast majority of the land has been subjected to wildfire disturbances, perhaps on the order of every 1000 years.

There may, of course, be a link between wildfire disturbances and human activities. During wet periods of Amazon history, human agricultural disturbances would have been small and widely dispersed, as they are today. However, this pattern might have been altered during periods of unusual drought. At such times, fires in Amazonia would not have been confined to agricultural clearing but would have spread into standing forest as they do today near San Fernando de Atabapo. The effects of such wildfires would have been far reaching, destroying not only potential cropland but game and fish habitat as well. Under such circumstances of increased human pressure on forest resources, it is easy to imagine human migration from the drier and more fire-prone core areas of the Amazon basin into wetter and less hospitable peripheral areas such as the Upper Rio Negro. This human disper- sal, in turn, would have increased the incidence of forest fires even in these marginal parts of the basin. Indeed, some linguistic and ethnographic evidence points to large-scale migration of human population at times roughly cor- responding with the hypothesized Holocene drought periods in Amazonia (Meggers, 1982; Migliazza, 1982). Moreover, Rio Negro mythology records the occurrence of a devastating fire(s) which destroyed crops and game and led to widespread wandering and starvation. The soil charcoal evidence from the San Carlos region suggests that this has indeed occurred, perhaps twice, during the past 2000 years.

Under the present climatic regime, tropical rain forests have generally been regarded as immune to fire. Yet, we report that certain forest types at San Carlos can burn during short droughts; and Uhl and Buschbacher (1985) report that forest fires are common in selectively-logged forest in the eastern Amazon. The most spectacular example of contemporary fire occurred in Bornean rain forest in 1983 when thousands of square kilometers of forest (an area equal to Taiwan) burned following selective logging. Selec- tive logging promotes fire in both Amazonian and Malaysian rain forest; the extensive canopy openings and the addition of slash on the forest floor associated with forest exploitation turn the normally fire-resistant rain forest ecosystem into a fire-prone ecosystem. When fires do occur, they are usually restricted to the ground, but in vine-laden forests, canopy fires can also occur (Uhl and Buschbacher, 1985).

Given the prevalence of fire in Amazon forest history, how does Amazon vegetation respond to fire disturbance? Studies of forest succession following forest cutting and burning disturbances at San Carlos, (Uhl and


Jordan, 1984) provide an indication of regrowth potential following fire. Uhl and Jordan found that forests regrow vigorously on burned sites. This is because many Amazonian tree species have the ability to sprout after damage; and, al though fires do kill many stems outright, a pool of individuals sur- vives burning and quickly sprouts. In addition, Amazon forests have a rich seed bank (500-1000 seeds of successional woody species m -2) and a fair proport ion of these seeds survive burning and then establish (Uhl and Clark, 1983). The ability of some rain forest species to withstand fire disturbances, either through sprouting or the buried seed strategy, may be the result of natural selection, i.e., fire may have been a selecting agent for these character- istics.

Our report should underscore that the disturbance history of Amazo- nia and the role that humans have played in this history is still very poorly understood. We do know that, in the case of the Rio Negro, environmental constraints related to protein capture must have severely restricted population size. Humans engaged in slash and burn agriculture to meet their car- bohydrate needs, but probably only disturbed land near streams and rivers. In these settings, disturbances were probably widely dispersed in time and space.

The vast expanse of land removed from rivers was probably never farmed in the Upper Rio Negro, but fire may have been much more common than ever thought in these areas. It has been customary to regard Amazonia as having had a stable vegetation history, but the fire reports f rom the Rio Negro suggest a much more complex history. The role that humans have had in this history, as orchestrators or victims, remains to be unraveled.

A C K N O W L E D G M E N T S

In developing the ideas presented in this paper, we benefitted f rom discussions with Juan Saldarriaga, Robert Sanford, and Rafael Herrera. We thank Gladys Marshall for helping us prepare the manuscript for publication.

R E F E R E N C E S

Absy, M. L. (1982). Quaternary palynological studies in the Amazon Basin. Prance, G. T. (ed.), BiologicalDiversification in the Tropics. Columbia University Press, New York, pp. 67-73.

Bayley, P. B. (1981). Fish yield from the Amazon in Brazil: Comparison with African river yields and management possibilities. Transactions American Fisheries Society i 10:351-359.

Beckerman, S. (1980). Fishing and hunting by the Barf of Colombia. Haines, R. B. and Kens- inger, K. M. (eds.), Studies in Hunting and Fishing in the Neotropics. Working Papers on South American Indians No. 2, Bennington College, Bennington, Vermont, pp. 68-109.

Clark, K. E., and Uhl, C. (1984). Deterioro de la vida de subsistencia tradicional en San Carlos de Rio Negro. Interciencia 9: 358-365.

24 Clark and Uhl

Denevan, W. M. (1976). The aboriginal population of Amazonia. In Denevan, W. M. (ed.), The Native Population o f the Americas in 1492. The University of Wisconsin Press, Madison, pp. 205-234.

Doku, E. V. (1969). Cassava in Ghana. Ghana University Press. Fearnside, P. M. (1978). Estimation of carrying capacity for human populations in a part of

the Transamazon Highway colonization area of Brazil. Ph.D. Dissertation, University of Michigan, Ann Arbor.

Fearnside, P. M. (1982). Deforestation in the Brazilian Amazon: How fast is it occurring? In- terciencia 7: 82-88.

Fittkau, E. J., Junk, W., Klinge, H., and Sioli, H. (1975). Substrate and vegetation in the Amazon region. In Dierschke, H. (ed.), Vegetation and substrate. J. Cramer, Vaduz, Liechtens- tein, pp. 70-93.

Gleason, H. A. (1931). Botanical results of the Tyler-Duida Expedition. Bulletin Torrey Botanical Club 58: 277-506.

Goodland, R. J. A., and Irwin, H. S. (1975). Amazon Jungle." Green Hell to RedDesert, Elsevier, New York.

Goulding, M. (1981 ). Man and Fisheries on an Amazon Frontier. Developments in Hydrobiology 4. W. Junk, The Hague.

Gross, D. (1975). Protein capture and cultural development in the Amazon basin. American Anthropologist 77." 526-549.

Hecht, S. B. (1981). Cattle ranching in the Amazon: Analysis of a development strategy. Ph.D. dissertation, University of California, Berkeley.

Hemming, J. (1978). Red Gold: The Conquest o f the Brazilian Indians. Harvard University Press, Cambridge.

Hen'era, R., Jordan, C. F., Klinge, H., and Medina, E. (1978). Amazon ecosystems: their structure and functioning with particular emphasis on nutrients. Intercienc!a 3:223-231.

Heuveldop, J. (1980). Bioklima von San Carlos de Rio Negro, Venezuela. Amazoniana 7: 7-17. Holmes, R. (1981). Estado nutricional en cuatro aldeas de la selva Amaz6nica-Venezuela: Un

estudio de adaptaci6n y aculturaci6n. M.S. thesis, Centro de Estudios Avanzados, In- stituto Venezolano de Investigaciones Cientificas (I.V.I.C.), Caracas.

Jaff6, E. H., Nolberga, B., Embden, C., Garcia, S., Olivares, H., and Gross, M. (1956). Com- posici6n de pescados Venezolanos. Archivos Venezolanos de Nutrici6n 7: 163-166.

Janzen, D. (1974). Tropical blackwater rivers, animals, and mast fruiting by the Dipterocar- paceae. Biotropica 6: 69-103.

Kay, D. E. (1973). Root crops. Crop and Production Digest, Tropical Production Institute. Klinge, H. (1967). Podzol soils: A source of blackwater rivers in Amazonia. Atlas do Simp6sio

s6bre a Biota Amaz6nica 3: 117-125. Klinge, H., Medina, E., and Herrera, R. (1977). Studies on the ecology of Amazon caatinga

forest in southern Venezuela. I. General features. Acta Cient~ca Venezolana 28" 270-276. Komarek, E. V. (1971). Lightning and fire ecology in Africa. Proceedings of the Tall Timbers

Fire Ecology Conference Tallahassee No. 10, pp. 473-511. Mago-Leccia, F. (1970). Estudios preliminares sobre la ecologia de los peces de los llanos de

Venezuela. Acta Bioldgica Venezolana 7: 71-102. Mago-Leccia, F. (1978). Los Peces de Agua Dulce de Venezuela. Cuadernos Lagoven. Malingreau, J. P., Stephens, G., and Fellows, L. (1985). Remote sensing of forest fires: Kaliman-

tan and North Borneo in 1982-1983. Ambio 14: 314-321. Markgraf, V., and Bradbury, J. P. (1982). Holocene climatic history of South America. Striae

16: 40-45. Marlier, G. (1973). Limnology of the Congo and Amazon Rivers. In Meggers, B. J., Ayensu,

E. S., and Duckworth, W. D. (eds.), Tropical Forest Ecosystems in Africa and South America: A Comparative Review. Smithsonian Institution Press, Washington, D.C., pp. 223-238.

Meggers, B. J. (1971). Man and Culture in a Counterfeit Paradise. University of Chicago Press, Chicago.

Meggers, B. J. (1982). Archaeological and ethnographic evidence compatible with the model of forest fragmentation. Prance, G. T. (ed.), Biological Diversification in the Tropics. Columbia University Press, New York, pp. 483-496.


Migliazza, E. C. (1982). Linguistic prehistory and the refuge model. In Prance, G. T. (ed.), Biological Diversification in the Tropics. Columbia University Press, New York, pp. 497-519.

Montaldo, A. (1977). Cultivo de Raices y Tubdrculos Tropicales. Instituto Interamericano de Ciencias Agricolas (OEA), San Jos6, Costa Rica.

National Academy of Sciences (1980). Conversion o f Moist TropicalForest. NAS, Washington, D.C.

Ooi, Jin Bee (1958). The distribution of present-day man in the tropics: Historical and ecological perspective. Proceedings of the Ninth Pacific Science Congress of the Pacific Science Association (Vol. 20), Published by the Secretariat, Ninth Pacific Science Congress, Bangkok, Thailand.

Petriceks, J. (1968). Shifting cultivation in Venezuela. Ph.D. dissertation, SUNY College of Forestry, Syracuse.

Pires, J. M., and Prance, G. (1977). The Amazon forest: A natural heritage to be preserved. In Extinction isForever Symposium. New York Botanical Garden, New York, pp. 158-194.

Ribeiro, D. (1970). Fronteras Ind[~enas de la Civilizaci6n (3rd. Ed.). Columbia University Press, New York.

Salati, E., M~ques, J., ancl Molion, L. C. B. (1978). Origem e distribuc~o das chuvas na Amaz6nia. lnterciencia 3: 200-205.

Saldarriaga, J. (1985). Forest succession in the Upper Rio Negro of Colombia and Venezuela. Ph.D. dissertation, University of Tennessee, Knoxville.

Saldarriaga, J. G., and West, D. C. (1986). Holocene fires in the northern Amazon Basin. Quater- nary Research 26: 358-366.

Sanford, R. L., Jr., Saldarriaga, J., Clark, K. E., Uhl, C., and Herrera, R. (1985). Amazon rain forest fires. Science 227: 53-55.

Shrimpton, R., Giugliano, R., and Rodrigues, N. M. (1979). Consumo de alimentos e algums nutrientes em Manaus, 1973-1974. Acta Amaz6nica 9:117-141.

Sioli, H. (1956). As ~iguas da regi~o do alto Rio Negro. Instituto Agron6mico do Norte No. 32, pp. 117-155.

Sioli, H. (1975). Tropical rivers as expressions of their terrestrial environments. In Golley, F. G., and Medina, E. (eds.), Tropical Ecological Systems: Trends in Terrestrial and Aquatic Research. Springer-Verlag, New York, pp. 275-288.

Smith, N. J. H. (1979). A Pesca no Rio Amazonas. Conselho Nacional de Desenvolvi- miento Cientifico e Tecnol6gico, Instituto Nacional de Pesquisa da Amaz6nia, Manaus, Brasil; Falangola, Bel6m, Brasil.

Smith, N. J. H. (1980). Anthrosols and human carrying capacity in Amazonia. Annals o f the Association o f American Geographers. 70: 553-566.

Spruce, R. (1908). Notes o f a Botanist on the Amazon and Andes (Vol. 1). Johnson Reprints Corp., New York (reprinted 1970).

Uhl, C. (1982). Recovery following disturbances of different intensities in the Amazon rain forest of Venezuela. lnterciencia 7: 19-24.

Uhl, C., and Buschbacher, R. (1985). A disturbing synergism between cattle-ranch burning practices and selective tree harvesting in the eastern Amazon. Biotropica 17." 265-268.

Uhl, C., and Clark, K. (1983). Seed ecology of selected Amazon Basin successional species. Botanical Gazette 144: 419-425.

Uhl, C., and Jordan, C. F. (1984). Succession and nutrient dynamics following forest cutting and burning in Amazonia. Ecology 65: 1476-1490.

Uhl, C., and Murphy, P. (1981a). Composition, structure, and regeneration of a tierra firme forest in the Amazon Basin of Venezuela. Tropical Ecology 22: 219-237.

Uhl, C., and Murphy, P. (1981b). A comparison of productivities and energy values between slash and burn agriculture and secondary succession in the Upper Rio Negro region of the Amazon Basin. Agro-Ecosystems 7" 63-83.

Van der Hammen, T. (1972). Changes in vegetation and climate in the Amazon Basin and sur- rounding areas during the Pleistocene. GeoL en Mijnbouw 51: 641-643.

Vasey, D. E. (1979). Population and agricultral intensity in the humid tropics. Human Ecology 7: 269-283.

26 Clark and Uhl

Venezuelan, Comisi6n para el Desarrollo del Sur de Venezuela (CODESUR) (1973). Atlas del Territorio Federal Amazonas y Distrito Cede~o del Estado Bolivar. Minesterio de Obras Pfiblicas.

Wambeke, A., van (1978). Properties and potentials of soils in the Amazon Basin. Interciencia 3: 233-242.

Welcomme, R. L. (1976). Some general and theoretical considerations on the fish yield of African rivers. Journal Fisheries Biology 8:351-364.

Welcomme, R. L. (1979). Fisheries Ecology of Floodplain Rivers. Longman, London. Werner, D., Flowers, N. M., Ritter, M. L., and Gross, D. R. (1979). Subsistence productivity

and hunting effort in native South America. Human Ecology 7: 303-315. Wijmstra, T. A., and Van der Hammen, T. (1966). Palynological data on the history of tropical

savannas in northern South America. Leides Geol. Meded. 38: 71-90.

Documents

Farming, fishing, and fire in the history of the upper Río Negro region of Venezuela