The biology of Orbignya martiana (PALMAE), a tropical dry forest dominant in Brazil


Material Information

The biology of Orbignya martiana (PALMAE), a tropical dry forest dominant in Brazil
Physical Description:
Anderson, Anthony B
Publication Date:


Subjects / Keywords:

Record Information

Source Institution:
University of Florida
Rights Management:
Permission granted to the University of Florida to digitize, archive and distributed this item for non-profit and educational purposes only. Any reuse of this item in excess of fair use requires permission of the copyright holder.
Resource Identifier:
aleph - 30349357
oclc - 11437530
System ID:

Table of Contents
    Title Page
        Page i
        Page ii
        Page iii
        Page iv
    Table of Contents
        Page v
        Page vi
    List of Tables
        Page vii
        Page viii
    List of Figures
        Page ix
        Page x
        Page xi
        Page xii
        Page xiii
    Chapter 1. Introduction
        Page 1
        Page 2
        Page 3
        Page 4
        Page 5
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
    Chapter 2. Taxonomy and phytogeography
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
    Chapter 3. Reproductive biology
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
        Page 53
        Page 54
        Page 55
        Page 56
        Page 57
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
        Page 69
        Page 70
        Page 71
        Page 72
        Page 73
        Page 74
        Page 75
        Page 76
        Page 77
        Page 78
        Page 79
        Page 80
        Page 81
        Page 82
        Page 83
        Page 84
        Page 85
        Page 86
        Page 87
    Chapter 4. Establishment
        Page 88
        Page 89
        Page 90
        Page 91
        Page 92
        Page 93
        Page 94
        Page 95
        Page 96
        Page 97
        Page 98
        Page 99
        Page 100
        Page 101
        Page 102
        Page 103
        Page 104
        Page 105
        Page 106
        Page 107
        Page 108
        Page 109
        Page 110
        Page 111
        Page 112
        Page 113
        Page 114
    Chapter 5. Growth and productivity
        Page 115
        Page 116
        Page 117
        Page 118
        Page 119
        Page 120
        Page 121
        Page 122
        Page 123
        Page 124
        Page 125
        Page 126
        Page 127
        Page 128
        Page 129
        Page 130
        Page 131
        Page 132
        Page 133
        Page 134
        Page 135
        Page 136
        Page 137
        Page 138
        Page 139
    Chapter 6. Population structure and dynamics
        Page 140
        Page 141
        Page 142
        Page 143
        Page 144
        Page 145
        Page 146
        Page 147
        Page 148
        Page 149
        Page 150
        Page 151
        Page 152
        Page 153
        Page 154
        Page 155
        Page 156
    Chapter 7. Implications for management
        Page 157
        Page 158
        Page 159
        Page 160
        Page 161
        Page 162
        Page 163
        Page 164
        Page 165
        Page 166
        Page 167
    References cited
        Page 168
        Page 169
        Page 170
        Page 171
        Page 172
        Page 173
        Page 174
        Page 175
        Page 176
        Page 177
        Page 178
    Appendix A. Species abundance, frequency, dominance, and importance in 1 HA of primary forest, Lago Verde, Maranhao
        Page 179
        Page 180
        Page 181
        Page 182
        Page 183
    Appendix B. Species abundance, frequency, dominance, and importance in 1 HA of secondary forest, Lago Verde, Maranhao
        Page 184
        Page 185
        Page 186
    Appendix C. Morphological descriptionof Orbignya martiana Barb. Rodr.
        Page 187
        Page 188
        Page 189
    Appendix D. Selective comparison of gross morphological characters in Orbignya martiana, "O. teixeirana", and O. eichleri
        Page 190
        Page 191
    Appendix E. Selective comparison of gross morphological characters in Orbignya martiana, "Markeleya dahlgreniana", and Maximiliana maripa
        Page 192
        Page 193
    Biographical sketch
        Page 194
        Page 195
        Page 196
Full Text






This study is dedicated to the late Harold E. Moore, Jr., who inspired it.


My field research and data analysis were financed by Grant No. 5107-09-07 to John Ewel from the USDA Forest Service, Consortium for the

Study of Man's Relationship with the Global Environment. Additional support was provided by a fellowship from the Superintendencia de Desenvolvimento Cientifico e Tecnol6gico (CNPq). I am grateful to the Museu

Paraense Emilio Goeldi and its Director, Dr. Jose Seixas Lourenco, for invaluable logistic as well as moral support during the course of my research.

Special thanks are due to Bernardo da Silva and Mario Costa, who gave unfailing assistance in the field, and to Michael Balick and William Overal, who collaborated with me on various aspects of the research. I am also grateful to the following people and their associated institutions: Jos6 Mdrio Frazao and Claddio Pinheiro of the Instituto Estadual do Babapu (INEB) for their collaboration in studies on germination, phenology, and taxonomy of babassu; Heraclito Aquino and

Walbert Carvalho of the Empresa Maranhense de Pesquisa Agropecuiria (EMAPA) for providing field workers; S6rgio Alves, Mirio Dantas, Elizabeth de Oliveira, Raimundo Rego, and Waldemar Ferreira of the Centro de Pesquisa Agropecuaria do Tropico Umido (CPATU) for invaluable logistic support; and Raimundo Bittencourt, Waldemar Franca, Nelson Rosa, and Sr. Milton of the Museu Goeldi for their assistance in the field.


The excellent drawings were by George Fuller, and my battery of dedicated typists included Bob Epting, Donna Epting, and Leslie Rigg. Less tangible but.nonetheless meaningful assistance was provided by Helda Lenz Cesar, Warwick Kerr, Mario Leal, Peter May, and Ghillean Prance.

I would like to express my sincere appreciation to the people who made my dedication to the research possible: Clemente Rodrigues and his extended family; Isa and Nelson Carvalho; my parents, Martha and Seneca Anderson; and especially my wife, Suely Anderson.

Finally, very special thanks go to my supervisory committee, Drs. Peter Feinsinger, Walter Judd, Hugh Popenoe, Francis Putz, and especially my chairman, John Ewel, who visited the field sites and provided patient guidance in all phases of the research.



ACKNOWLEDGEMENTS . . . . . . . . . . . .

LIST OF TABLES . . . . . . . . . a a vii

LIST OF FIGURES . . . . . . . . . . . . ix

ABSTRACT . . . . . * * * * * * *xii


1 INTRODUCTION . . . . . . . . . . 1

Methods * * . * . . e 29 Results and Discussion . . . . . 32

3 REPRODUCTI EBIOOLO... .. .. .. . ... 38
Methods . . . . . . . . . . 39
Results . . . . . . . . . . 46
Discussion . . . . . . . . . 74

4 ESTABLISHMENT . . . . . . . . . . 88
Me thods . . . . . . . . . . 90
Discussion . . . . . . . . . 112

5 GROWTH AND PRODUCTIVITY ... .. .. . .. 115
Results . . . . . . ... 1216
Discussion. . * * e * 9 . . 129


Discussion., 154

.People and the Palm Forest . . 157 Limiting Factors. .....* *159
Possible Solutions ............. . 163


REFERENCES CITED . . . . . . . . . . . 168


MARANHA0 . . . . . . . . 179

MARANHAO . . . . . . . . . . .. 184

Barb. Rodr. .. . . .. .. . . . 187

CHARACTERS IN Orbignya martiana, "0.
teixeirana", and 0. eichleri . . . . . . 190

CHARACTERS IN Orbignya martiana, "Markeleya
dahlgreniana", and Maximiliana maripa . . 192

BIOGRAPHICAL SKETCH. . . . . . . . . . .. 194




1 Uses of occupied lands in Maranhio during 1950 and 1975 . 22

2 Nomenclatural history of babassu and its suspected hybrids. 30

3 Fruit yields of babassu per ha and per palm during 1980-81
and 1981-82 at three sites. .0. . . . . . . .. 60

4 Sources of increased fruit yields per ha between 1980-81
and 1981-82 at three sites ................ 61

5 Attack of babassu fruits and seeds by the seed-predaceous
beetle, Pachymerus nucleorum .. . . . . .. 68

6 Germination of babassu in four experiments. .*. . . 73

7 Preliminary checklist of insects that feed on babassu .... 80

8 Hypothesized success of babassu in four experiments .... 89

9 Palms per m2 (a), leaf area per palm (b), and leaf area
index (c) of babassu at end of irrigation experiment.. . .102

10 Palms per m2 (a), leaf area per palm (b), and leaf area
index (c) of babassu at end of fertilizer experiment.. . .103

11 Palms per m2 (a), leaf area per palm (b), and leaf area
index (c) of babassu at end of insecticide experiment . .104

12 Palms per m2 (a), leaf area per palm (b), and leaf area
index (c) of babassu at end of climate experiment, with
climate and weeding as main effects 0 . . 0 . . .105 13 Palms per m2 (a), leaf area per palm (b), and leaf area index (c) of babassu at end of climate experiment, with
climate and ecotype as main effects . . . 0 0 0 $106

14 Leaf area index (LAI) of babassu and competitors in four
experiments 0 . . . . .. .. .. . .107

15 Dry mass of babassu and competitors in four experiments . .108


16 Soil chemistry of four treatments at end of nutrient
experiment . . . . . . . . . . . . .109

17 Soil chemistry of experimental sites at Lago Verde and
Belim . . . . . . . . . . . . . *110

18 Definition of 12 life stages of babassu used in study . .117 19 Estimated growth rates and ages of babassu palms on the
primary forest site at Lago Verde . . . . . . 122

20 Estimated growth rates and ages of babassu palms on the
secondary forest site at Lago Verde . . . . . .123

21 -Partitioning of above-ground dry matter production of
babassu over an estimated life span of 184 yr . . . .126 22 Annual above-ground dry matter production of mature babassu
palms on three study sites at Lago Verde. . ... . . 127 23 Annual above-ground dry matter production of all babassu
palms on the primary and secondary forest sites at Lago
Verde . . . . . . . . . . . .. .128

24 Life table of the babassu population on the 1-ha primary
forest site at Lago Verde . . . . . . . . .144

25 Annual changes in number of palms per life stage on the 1ha primary forest site at Lago Verde. . . . . . .145

26 Annual changes in number of palms per life stage on the 1ha secondary forest site at Lago Verde. . . . . .. .146

27 Annual changes in number of palms per life stage on the 1ha pasture site at Lago Verde . . . ... . . .147




1 Approximate distribution of high-density stands of babassu
(Orbignya martiana) in Brazil . . . . . . . . 2

2 Aerial view of babassu stands in the county of Bacabal,
Maranhao. . . . . . . . . . . . . .. 3

3 Specimen of babassu (0. martiana) in Pirapora, Minas
Gerais. .0. . . . . . . . . . . . . 3

4 Deforestation in the Tropical Moist Forest zone of
Maranh o. . . . . . . . . . . . . 4

5 Regrowth of babassu on a deforested site in the Tropical
Moist Forest zone of Maranhao . . . . . . . . 4

6 Production of babassu kernels in Maranhio and Brazil, 19201979. . . . . . . . . . . . . . . 5

7 Flowchart of products derived from babassu fruits by industries in Bacabal, Maranhao, during 1980 . . . . . 6

8 Flowchart of products derivable from babassu fruits via
current (1983) technology . . . . . . . . 7

9 Maranhao and neighboring states, with locations of research
sites . . . . . . . . . . . . 10
10 Major soil groups of Maranhao, according to the Brazilian
Classification System . . . . . . . . . 11

11 Principal ecological zones of Maranhao. . . . . .. 12

12 Coverage of babassu stands in Maranhao. . . . . . 13

13 Monthly rainfall at four research sites . . . . . 18

14 Distribution of land and households according to size of
landholding in rural areas of Maranhao during 1950 and
1975. . . . . . . . . . . . . . . 20

15 Distribution of specimens of Orbignya martiana and its
suspected hybrid complexes. . . . . . ....... 31


16 Mean monthly gain and loss of leaves per palm on three
sites in Maranhao . . . . .. . . . .. . 47

17 Mean monthly gain and loss of leaves per palm at three
study sites at Lago Verde . . . . . . . . .. 48

18 Mean monthly production of male and functionally female
inflorescences per palm at three sites in Maranhao.... . 50 19 Mean monthly production of male and functionally female
inflorescences per palm at three sites at Lago Verde . . 51 20 Distributions of female inflorescences per palm in six
populations of babassu. . . . . . . . . .. 54

21 Mean monthly number of immature and mature infructescences
per palm at three sites in Maranhio . .. . ... . 55

22 Mean monthly number of immature and mature infructescences
per palm at three sites at Lago Verde . . .. . .. 56

23 Monthly fruit production per ha on three study sites in
Lago Verde. . . . . . . . . . . . 57

24 Monthly fruit production per palm on three study sites at
Lago Verde. . .. . . . . . . . . 58

25 Removal of babassu fruits on three study sites at Lago
Verde . . 0. . . . . . . . . . 66

26 Attack of babassu fruits and seeds by the Bruchid beetle,
Pachymerus nucleorum, over time . . . . . . . 67

27 Seed fate over time in three study sites at Lago Verde . 69 28 Cumulative germination of babassu fruits over time. . . 71 29 Germination in fruits obtained from 10 babassu palms . . 72 30 Force required to break palm fruits . . . . . .. 84

31 Cryptogeal germination in babassu . . . . . . .. 86

32 Leaf area as a function of leaf length in seedlings of
babassu . . . . . . . . . . . . 93

33 Palms per m2 (a), leaf area per palm (b), and leaf area
index (c) of babassu over time in irrigation experiment . 96


34 Palms per m2 (a), leaf area per palm (b), and leaf area
index (c) of babassu over time in fertilizer experiment . 97 35 Palms per m2 (a), leaf area per palm (b), and leaf area
index (c) of babassu over time in insecticide experiment. . 98 36 Palms per m2 (a), leaf area per palm (b), and leaf area
index (c) of babassu over time in climate experiment ..... 99 37 Absolute allocation of annual above-ground dry matter
production per babassu palm as a function of age.. ... .130 38 Relative allocation of annual above-ground dry matter
production per babassu palm as a function of age. ... 0131 39 Mean annual number of inflorescences produced per palm as a
function of mean age per life stage . . . . . . .134

40 Mean annual number of seeds produced per palm as a function
of mean age per life stage. . . . . . * e .135

41 Stage distributions of babassu on the primary forest,
secondary forest, and pasture sites at Lago Verde .. 148 42 Age distribution of babassu on the primary forest site at
Lago Verde . . . . . 0 0 0 0 .149

43 Age distribution of babassu on the secondary forest site
at Lago Verde . .. . . . .. . .. 0 .150
44 Population dynamics of babassu on the 1-ha primary forest,

secondary forest, and pasture sites at Lago Verde .. . .151 45 Shifting cultivation site just prior to burn at Lago Verde. .160 46 Clearcut of palm forest prior to pasture establishment,
near Bacabal, Maranhdo . . ............ 160


Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Anthony B. Anderson

December, 1983

Chairman: John J. Ewel
Major Department: Botany

Monospecific stands of the babassu palm (Orbignya martiana) occupy widespread areas of Brazil, attaining their greatest coverage (102,970 km2) in the state of Maranhio. Near Lago-Verde, Maranhio, I conducted comparative studies of babassu's life history to determine how the palm is maintained in mature forests as well as how it dominates disturbed sites.

Flowering in babassu peaked during the rainy season (January-March) and coincided with a flush of new leaves (x = 4.1 per yr); fruiting peaked during the following dry season (October-December). Outcrossing was predominant and was effected by wind and nitidulid beetles (MystroDs

sp.). Dispersal of the massive ( >100 g) fruits was mediated primarily by gravity and by rodents. Due to the hard fruit endocarp, seed predation was virtually limited to bruchid larvae (Pachymerus nucleorum), which entered through the germination pores; ca. 40% of seeds suffered


predation. Germination required 3 mo and was enhanced in shade. Stemless palms persisted in the understory of a primary forest for about 50 yr, permitting build-up of high densities (6,000 per ha). Escape from seed predators, shade tolerance, and an estimated lifespan of 184 yr account for babassu's importance in such forests. Due to burial of

their apical meristems via cryptogeal germination, stemless palms survive when sites are burned for shifting cultivation. During the subsequent fallow, these palms are released and form virtually monospecific stands. One such stand was still increasing in numbers after a fallow of 33 yr. Annual above-ground dry matter production of this stand was ca. 25 t'ha-1"yr-1, 70% of which was allocated to leaves. With fallows of 4-5 yr, enough fuel accumulates in the form of leaves to permit the hot fire necessary for weed control and nutrient release. Thus shifting cultivation can be practiced with minimal thinning of the stands.

Babassu's dominance of the landscape results from its persistence over a wide gradient of disturbances. Because of its capacity to form monospecific stands, its low maintenance requirements, its integration into major land uses, and its current importance in market and subsistence economies, babassu seems most amenable for use as an extensive, moderate-yielding crop.



Numerous species of palms dominate the landscape over extensive areas of the lowland tropics (Beard 1944, Richards 1952, Moore 1973a). Palm-dominated stands often occur on edaphically limited sites where competition is apparently reduced (Myers 1981). Several species, however, form nearly monospecific stands on sites originally covered by forests of high biological diversity. Such secondary palm forests

originate from and are apparently maintained by human activities. People, in turn, depend on the palms as a source of food, fuel, shelter and income.

This study concerns one of the most economically important and certainly the most ubiquitous of such secondary forest palms: Orbignya martiana, known in Brazil as babassu. The tree-sized babassu palm forms high-density stands over extensive areas of Brazil (Figs. 1-3). The stands attain their greatest extent in the state of Maranhdo, where they

cover an area estimated at 102,970 km2 (Anonymous 1981), or roughly equivalent to the U.S. state of Virginia. Babassu flourishes on sites subject to intensive disturbances such as cutting and burning (Figs. 45). The palm is a crucial resource for rural poor, who utilize the fruits, leaves, and stem for a wide range of subsistence products. Manual extraction of kernels from the extremely hard fruits represents the state's largest cottage industry (involving over 400,000 families) and its most important income source for women. The kernels are util1


75 70 65 60 C.5 50 45 40 3
h I ti I




-. ;. r ...- ',, --

MINAS GERAIS"' a oo 000k / 20



Figure 1. Approximate distribution of high-density stands of
babassu (Orbignya martiana) in Brazil. States containing stands are labelled. Adapted from Mendes
and Carioca (1981).


.Q. Ca cc U



ca 0 ~ca
r p

cu a

Figure 4. Deforestation in the Tropical Mo11ist Forest zone of Iflaranhao.


.. ~. ..

Figure 5. Regrowth of babassu on a deforested site in the Tropical
Moist Forest zone of Maranhdo.






w 40

1920 1930 1940 1950 1960 1970 1980

Figure 6. Production of babassu kernels in Maranhio and Brazil,
1920-79. Sources: IBGE (1981a and previous volumes to


1 3 8 S T A- R C H

MESOCARP 9.2 -023.0 23.0 -- FERTILIZER

< 8.8

I 1. 1.0 PRIMARY


1 4 .7 -. C O K E

4.20 OIL


Figure 7. Flowchart of products derived from babassu fruits by industries
in Bacabal, Maranhao, during 1980. Diamonds represent "eitheror" options. Numbers are percentages of fruit weight. Adapted
from Mendes and Carioca (1981) and Empresa Industrial de Bacabal (personal communication).



3.0 9.2 ANIMAL









Figure 8. Flowchart of products derivable from babassu fruits via
current (1983) technology. Diamonds represent "either-or"
options. Numbers are percentages of fruit weight. Adapted from Abreu (1940), Viveiros (1943), Escola Tecnica Federal
do Maranhao (1976), IPT (1979), and Mendes and Carioca


ized by industries for production of vegetable oil, soap, and feedcake. These industries represent the largest single source of tax revenues in the Maranhao economy. Over the past 60 yr, kernel production has steadily increased (Fig. 6), and as of 1979 babassu provided 89% of all vegetable oil obtained from non-domesticated sources in Brazil (IBGE 1981 a). Recently introduced machinery capable of mechanically breaking the hard fruits has enabled additional babassu-derived products (e.g., charcoal, coke) to enter national markets (Fig. 7), and technology presently exists to derive a host of additional products from the fruits (F ig 8).

Despite its economic importance, babassu is not currently domesticated, and formation of the stands remains a spontaneous phenomenon. In the past, stands of babassu were generally preserved and formed an integral part of shifting cultivation systems. In recent years, however, population pressures and changing land use pa'Vterns are inducing both shifting cultivators and ranchers to clearcut the stands (Anderson and Anderson 1983). Sound management of the palm forest requires basic information concerning the biology of babassu. The purpose of this study is to provide such information.

The study covers babassu's taxonomy (Chapter 2), life history (Chapters 3-.5), and population biology (Chapter 6); levels of inquiry change respectively from the species to individual palms to stands of


babassu. The study concludes with a discussion of the implications for management (Chapter 7).

The Setting


The Brazilian state of Maranh&o (Fig. 9) encompasses a wide range of ecological conditions. To the south lies a geologically old (Mesozoic), elevated peneplain, broken by abrupt plateaus and canyons, and by low, eroded mountain ranges. The predominant soils in this region are deeply weathered and nutrient poor, and annual rainfall rarely exceeds 1,000 mm. Human population densities are consequently low (<5 inhabitants per kin2), and the principal land use is grazing on unimproved range.

In central and northern Maranh~o, elevations decline and the terrain gently undulates as one enters a broad plain comprised of geologically young (Tertiary) sediments, dissected by many of the state's major rivers. Soils are'extremely variable (Fig. 10), with an abundance of nutrient-rich alluvium, and annual rainfall ranges from 1,000 to over 2,000 mm. Agriculture (most commonly shifting cultivation) and grazing constitute the principal land uses. Outside the capital (Siio Luis), population densities in this region average >10 inhabitants per km

The babassu forests of Maranha-o are largely confined to the central and northern portions of the state, where they are associated with three broadly defined ecological zones (Fig. 11): seasonally wet savanna

(baixada); woodland, scrub, or dry savanna (cerrado); and moist or dry forest (ma ta).


48 46 44 42
0 I I1 1






6 -.



50 0 50 100 150 200 250 KILOMETERS


Figure 9. Maranhio and neighboring states, with locations of
research sites.







Figure 10. iAajor soil groups of Maranhao, according to the Brazilian
classification system (tEIBRAPA 1976). H = Hydromorphic L Pe

Laterite, Pd = Dystrophic Red-Yellow Podzolic soils, Pe =

Eutrophic Red-Yellow Podzolic soils, Ppd = Plinthic PodLr Lr

zolic soils, L = Yellow Latosol, Lr Pe= Red-Yellow Latosol,

and Q = Quartz sands. Unlabelled areas = other soil types.
Source: eBRAPA (1981).


Lr .Lr

Lr 0 100 200 km


Figure 10. Major soil groups of Maranhio, according to the Brazilian
classification system (EMBRAPA 1976). H =Hydromorphic
Laterite, Pd = Dystrophic Red-Yellow Podzolic soils, Pe= Eutrophic Red-Yellow Podzolic soils, Ppd = Printhic Podzolic soils, L = Yellow Latosol, Lr = Red-Yellow Latosol, and Q = Quartz sands. Unlabelled areas = other soil types.
Source: EMBRAPA (1981) .



Figure 11. Principal ecological zones of i1aranhao.
Source: Kuhlmann (1977).


5- 33
34-67 '



............... x .

0 00 2CCkm

Figure 12. Coverage of babassu stands in Maranhao. Source:
Anonymous (1981).


Seasonally wet savanna. An extensive area traversed by the lower reaches of Maranhdo's major rivers is flooded during the December-May rainy season and burned during the dry season. This low-lying area is called baixada. The predominant soils (ground-water laterites and gleys) are hydromorphic, the major vegetation type is a low savanna dominated by sedges and grasses, and the principal land use is grazing

of cattle and buffalo on unimproved range. Throughout this region, forests of babassu are generally limited to topographically elevated islands, which are also the principal sites for shifting cultivation.

Woodland, scrub, or dry savanna. The term cerrado refers to both an extensive phytogeographic region of Brazil and the region's predominant vegetation. Within Maranh~o, cerrado is characterized by annual rainfall >1500 mm, a dry season of 6 mo or longer, and well- to excessively

drained soils. The vegetation consists of deciduous woodland, scrub, and occasional savanna. The predominant land use is grazing of cattle, goats, and sheep on unimproved range. Shifting cultivation and permanent plantations of sugarcane and cotton are generally confined to the moist bottomlands, which are also the habitats occupied by babassu forests.

Moist or dry forest. This zone is characterized by moderate to high rainfall and a dry season of 6 mo. It extends across both Tropical Moist Forest and Tropical Dry Forest Life Zones (sensu Holdridge 1967). The vegetation was originally tall, diverse forest ranging from evergreen to deciduous. Due to widespread deforestation during the past 60 yr, the original forest has been virtually eliminated, and today stands


of babassu dominate the landscape. During the past 25 yr, rapid influx of settlers from the arid Brazilian northeast--combined with an ambitious road-building program throughout the state--have pushed human settlement deep into Tropical Moist Forest, and extensive areas are presently occupied by young stands of babassu. The principal land uses throughout the moist or dry forest zone are shifting cultivation and cattle ranching. In contrast to other ecological zones in Maranhio, ranching here is carried out primarily on improved pastures.

The palm forest. In the major ecological zones of Maranhao described above, babassu stands are generally found on the sites most intensively utilized by people. On such sites, crops are established and pastures maintained by periodic cutting and burning. These practices account for the origin of the babassu stands in Maranhao.

Prior to the arrival of Europeans, Maranhio had long been occupied

by numerous indigenous groups (described in Steward 1963). Among the dominant sedentary tribes (e.g., Tupinamba, Timbira), agriculture was generally confined to the moist bottomlands, and villages were established nearby. During the dry months, bottomland stands of babassu and

buriti (Mauritia flexuosa) palms provided an important dietary supplement. As with the more nomadic tribes, territorial rights were identified with access to these stands, which served as a source of fiber and fuel, as well as a hedge against famine. High densities of babassu were probably limited to these bottomland sites. The original nature of the far more extensive upland forests can be discerned in the remnants that


still exist; babassu was commonly present but invariably associated with a high diversity of hardwoods.

With the arrival of Europeans during the 1600s, the landscape of Maranhi.o changed. In response to the needs of the colonial economy, plantations of rice, cotton, and sugarcane were established on bottomlands over an extensive area of the state. Yet more than two centuries elapsed before the first historical reference was made to the babassu forests of Maranhi.o (Prazeres 1820, cited in Abreu 1940: 11), which suggests that these forests were not as extensive as they are today. With the abolition of slavery in 1888, the plantation economy collapsed afid vast areas of cultivated land were abandoned. These sites were recolonized by high-density stands of babassu. Like the Indians before them, the freed slaves and their descendants were dependent on the palm for food, fuel, and fiber. As their populations increased, agriculture

in the form of shifting cultivation moved out of the bottomlands into the mixed upland forests, and stands of babassu gradually spread across

the landscape. The size of these stands today (Fig. 12) indicates the extent and nature of human impact on the landscape of Maranh~o. The Study Region

Climate and soils. The ecological research was located primarily in the central Mearim Valley of north-central Maranhao (Fig. 9), which I shall henceforth refer to as the study region. Climate and soils in this region are extremely favorable for agriculture. Meteorological data from the region's principal town, Bacabal, show a mean annual precipitation of 157.5 cm; the 6-mo wet season (December-May) is suffi-


cient for rain-fed agriculture on upland sites (Fig. 13). Mean monthly temperature and relative humidity are subject to moderate variations in

response to rainfall; the former ranges from 25.7C (February) to 28.4C (November), and the latter from 67.8% (October) to 85.2% (JanuaryFebruary). Throughout the region, the terrain is flat to gently rolling

and is dissected by numerous streams that feed into the Mearim River. The predominant soil (Fig. 10) is the Eutrophic Red-Yellow Podzolic type (Brazilian classification system), roughly equivalent to a Ferric Luvisol (FAO legend), or an Ustalf or Udalf (U.S.D.A. Soil Taxonomy). This soil is characterized by high base saturation ( 50%) and relatively high fertility compared to other regional soil types; soil structure is generally excellent (EMBRAPA 1975).

Settlement. Even after the first wave of non-indigenous settlers arrived during the 1920s, the region remained thinly populated. At that

time, Babacal was a small village, named after an adjacent stand of bacaba palms (0enocarpus bacaba) that has long since disappeared. Hunting, fishing, gathering, and shifting cultivation continued to be the principal economic activities. During the 1950s and 1960s, droughts in

the Brazilian northeast and roadbuilding throughout the state brought waves of migrants into the region. Populations soared, and by 1980 the

town of Bacabal contained 43,229 inhabitants (IBGE 1981b) and was the fifth largest urban center in Maranhao. A roadbuilding program begun

during the 1960s linked the major towns of the Mearim Valley with a well-developed system of roads. Influx of settlers, enhanced access, ample rainfall, and excellent soils transformed The region into one of



50 o (273.9cm) (206.7cm)
50- o

40. oo

20- *

10- .
I0 e


(157.5cm) (129.Ocm)

40- a

2 0



Figure 13. Monthly rainfall at four research sites. Numbers are longterm annual means, bars are long-term monthly means, shaded
circles are monthly data for 1981, and open circles are monthly data for 1982. Period over which long-term means were calculated and source of unpublished data for each
site are: Belem, 1931-60, Centro de Pesquisa Agropecuaria do Tropico Ifmido (CPATU); Pindare Mirim, 1966-80, Superintendincia de Desenvolvimento do Nordeste (SUDENE); Bacabal,
1972-80, Instituto Nacional de Heteorologia (INMET); and
Caxias, 1966-80, INMET.


the most agriculturally productive areas in Maranh-ao. Yet agricultural productivity has not brought prosperity to most of the people living there.

Land tenure. The main front of agricultural expansion in Maranh&o swept westward through Bacabal during the 1950s. But the pattern by which the study region was settled can still be seen today in other areas of frontier expansion throughout Brazil (Velho 1972). As settlement pushes forward, once isolated areas become increasingly accessible, and the process of land concentration soon commences. The aftermath of this process can be seen from the data on land distribution in MaranhAo (Fig. 14). In 19,50, 61.8% of the state's occupied lands were owned by 2.0% of its households. Despite a rapid expansion of area under occupation, there had been no discernible improvement by 1975, when 41.4% of the land was controlled by 0.5% of the households. In absolute terms, the amount of land in smallholdings (<100 ha) increased by a factor of 4.0; however, the number of small-holdings increased by a factor of 5.6. Smallholdings became, on the average, smaller, and the number of landless increased.

As throughout most of Maranh&o, land tenure in the study region is now well defined. The vast majority of rural inhabitants live in latched huts strung like beads along the roads, footpaths, and watercourses that form a network throughout the region. Behind these hamlets and villages, fences are invariably present, and beyond these fences lie the resources upon which people depend.


2.0 LAND (%)

5 .9

~- .. .. .

(95,112k9) 17
4 6.1( ............

( 95,2k9)1-7



D < IOha
E] 0-100 ha

100-1000 ha 1975
(n= 494,596) 1000-10,000 ha

S> 10,000 ha

Figure 14. Distribution of land and households according to size of
landholding in rural areas of Maranhao during 1950 and
1975. Size of holding defined by effective occupation of the land. Amount of land increased from 1950 to 1975 due
to frontier expansion. Sources: IBGE 1956, 1979.


Shifting cultivation. Throughout Maranhio, sharecropping is the means by which the vast majority of cultivators gain access to land. Crops are raised principally by shifting cultivation (Table 1). The forest is felled by axes and burned toward the end of the dry season. Planting usually takes place in December with the first wet season rains. In the study region, the principal crops are upland rice, cassava, maize, and beans. As rice is sensitive to drought, and cassava is intolerant of prolonged waterlogging, farmers often plant both, and intercropping is the rule. Cultivated areas are frequently weeded until

the April-June rice harvest; cassava may be harvested up to 2 yr after planting. Cultivated plots on upland sites are virtually never used for

two crops in a row, so new plots are required each year. Short cropping periods make this form of land use extremely extensive. Shifting cultivation of this sort can only remain viable in areas of low population densities. Over large portions of Maranh&o, this is no longer the case.

From 1950 to 1975, land area in Maranhao utilized for temporary cropping increased threefold, while the ratio of fallowed to cropped land was cut in half (Table 1). The immediate implications of shortened fallows for crop yields--and long-term implications for site quality-are obvious. Increased population densities are the principal cause of shortened fallows throughout Maranhio. But exceptionally short fallows in the study region are a consequence of widespread conversion to pasture.


Table 1. Uses of occupied lands in Maranhio during 1950 and 1975.
Numbers represent ha (percent of total classified in
parentheses). Sources: IBGE 1956, 1979.

Land Use 1950 1975

Range 3,41541,400 (410.2%) 2,590,600 (21.7%)
Improved Pasture 110,800 0.5%) 1,218,200 (10.2%)

Temporary 3141,500 (3.6%) 1,0141,100 ( 8.5%)
Permanent 14,200 (0.2%) 111,900 ( 0.4%)

Fallow 2,4107,800 (28.0%) 4,058,100 (341.0%)

Permanent Forest
Natural 2,361,700 (27.5%) 3,0141,300 (25.2%)
Planted 41,000 ( 0.0%) 1100 ( 0.0%)

Total Classified 8,597,4100 (100.0%) 11,937,600 (100.0%)

Not Classified 9111,500 471,500

Total Occupied 9,538,900 12,4109,100


Grazing. In contrast to the rest of Maranh&o, grazing in the study region is carried out predominantly on improved pastures. This has been a recent development; in 1950, improved pastures comprised only 0.5% of Maranhio's total occupied area (Table 1). The formation of improved pastures has enabled grazing to encroach into areas once used exclusively for shifting cultivation.

Conversion to pasture is generally carried out as part of shifting

cultivation. On upland plantations of crops such as rice and maize (but rarely cassava, due to its relatively long cropping period), seeds of the African grass, hyarrhenia ruf'a (known locally as lajeado), are introduced just prior to harvest (March-May). Use of fertilizers is rare, and legume-grass associations are apparently unknown. Pastures are usually burned during the dry season, and the burn often gets out of control; wildfires are a conspicuous feature of the landscape throughout Maranh.o.

Babassu. The study region is characterized by exceptionally high coverage of babassu. Spontaneous occurrence of the palm appears to be limited only on steep hilltops (where fruit dispersal may limit colonization) and waterlogged soils (where root respiration may be inhibited). Elsewhere the palm forest virtually blankets the landscape, except where it is thinned or clearcut by shifting cultivators and ranchers. Although clearcutting of babassu is illegal in Maranhio, the practice appears to be increasingly common, especially in areas where pasture conversion is widespread.


Study Sites

Lago Verde. The ecological research was centered on an 800-ha farm in the county of Lago Verde, ca. 25 km NW of Bacabal (Fig. 9). The farm is flat and dissected by several intermittent streams. Soils are generally fertile and well-drained. The predominant soil type is Eutrophic Red-Yellow Podzolic. Approximately 75% of the farm has been converted to pasture, and the rest is covered by various ages of forest fallow.

I selected three 1-ha sites for ecological studies, representative of undisturbed primary forest, old secondary forest, and annually burned pasture. The forest sites were located in an 11-ha reserve maintained for hunting and for harvest of fallen trees as a source of lumber. Each

study site measured 80 x 125 m. For orientation, wooden stakes were used to define twenty 20 x 25 m blocks; on the forested sites, the blocks were delimited by 2-in wide paths. Community characteristics of each site were obtained by measuring abundance, frequency per block, and diameter at breast height (dbh) of all trees with dbh 10 cm. Unknown species were collected and identified in the herbarium of the Museu Paraense Emilio Goeldi in Bel~m. Community characteristics of the

forested sites are summarized in Appendices A and B.

The primary forest site is probably representative of the original vegetation that once covered the region. Radiocarbon dating of charcoal fragments found at ca. 15 cm depth indicate that the site was burned 650 250 yr before present (letter dated 4 August 1983 from Richard R. Pardi, Radiocarbon Laboratory, Queens College, Flushing, New York, USA). I believe that the site has suffered no major disturbance since. Local


informants reported that the site was still covered by structurally complex, species-rich forest when they first arrived in the region during the 1930s. Since this time, the site has been subjected to minimal disturbance. At the time of my study, emergents attained 30-35 m, and the canopy was more or less continuous at 15-25 m. Most of the emergent species were deciduous during all or part of the dry season. Vines were relatively abundant, whereas epiphytes were rare. Species richness, at 63 species (dbh >10 cm) per ha, was lower than that of upland forests in Amazonia (Anderson and Benson 1980). With relatively high abundance (20.2%), frequency (9.2%), and dominance (27.4%), babassu was the most important species on the site.

The secondary forest was originally cut and burned for shifting cultivation in 1948; the site had suffered no major disturbance since.

Emergents attained 20-25 m, and the canopy--comprised largely of a uniform stand of babassu--was relatively continuous at 10-15 m. Species

richness was 31 species (dbh >10 cm) per ha. The percentage of total species that occurred in both the primary and secondary forest sites (coefficient of community: Whittaker 1975) was 31.9%. In abundance (66.7%), frequency (21.7%), and dominance (84.1%), babassu was far and away the most important species on the site.

The pasture was established through introduction in 1974 of the exotic grass, Hyparrhenia rufa, on a site that had already undergone several cycles of shifting cultivation. The abundant understory competitors (including Lantana camara, Stachytarpheta cajanenesis, Sida spp.,


Cyperus spp., and juveniles of babassu) were annually cut and burned. The 10-15 m high canopy was comprised exclusively of babassu, which occurred at a density of ca. 100 stems per ha.

Satellite sites. In addition to the principal field site at Lago

Verde, satellite sites were established at Bel~m, Pindar6 Mirim, and Caxias (Fig. 7). The Bel6m site was used for experiments involving establishment (Chapter 4); the Pindar6 Mirim and Caxias sites were used for phenological observations (Chapter 3). Meteorological data (Fig. 11) show a wide range of rainfall at these sites. Information concerning land uses and experimental design at each site is provided in the appropriate chapters.



Babassu belongs to the cocosoid group, one of 15 major groups within the palm family (Moore 1973b). Cocosoid palms are distinguished from the other groups by their fruits, which contain a thick, generally three-pored endocarp. The principal center of cocosoid diversity is South America, where 24 of 28 genera occur (Moore 1973b). Two wellknown representatives of the group are the coconut (Cocos nucifera) and the African oil palm (Elaeis guineensis).

The taxonomy of babassu has been a source of confusion since the palm was first described well over a century ago. This confusion begins on the generic level. Although babassu is generally assigned to the genus Orbignva, the status of this and the four to five other genera in the Attalea alliance has been questioned by Wessels Boer (1965). These genera are distinguished solely on the basis of staminate flower morphology. No correlation with other characters has been observed, and staminate flower morphology runs across a separation based on other characters such as endocarp pores. Lines between genera in the Attalea alliance are further blurred by collections of intermediate staminate flower types, which have prompted the description of two new genera; one of these, Markleva, was described by Bondar (1957) as a possible intergeneric hybrid involving babassu.

A classification based solely on staminate flower morphology is of little use in the field, as most species flower only for a short time.



Thus it is no surprise that babassu is often confused with other species and even genera, notably Attalea. Oil palms commercially referred to as babassu include Orbignva agrestis in Pars, Attalea oleifera in Goias and Minas Gerais, A. eraensis in Minas Gerais, and A. Dindobassu in Bahia (Markley 1971). Botanists are not immune to generic confusion. Aboriginal dispersal was evoked to account for the disjunct occurrence of babassu in the state of Sdo Paulo (Ferri 1974, 1980), although the population has been identified as pertaining to Attalea (J. T. de M. Costa, pers. comm.).

To resolve intergeneric confusion in the Attalea alliance, Wessels Boer (1965) recommended reducing Orbignva and related genera to the single genus Attalea, but his suggestion has not been followed by other contemporary taxonomists (Moore 1973b, Glassman 1977). Countervailing arguments include the weight of taxonomic tradition, and the observation that Wessels Boer's scheme merely reduces the confusion to the specific level (H. E. Moore, pers. comm.). Due to the uncertain limits between genera, the possible occurrence of intergeneric hybrids, and the practical difficulty of distinguishing genera in the field, I believe that the current splitting of the Attalea alliance into five genera (Moore 1973b) is not justified. Final resolution of the confused generic limits within the Attalea alliance will require exhaustive study of the entire group. As such a study is not on the horizon at present, I shall follow taxonomic tradition and consider babassu within the genus Orbignva.


On the specific level, at least 10 names associated with babassu have been published within the genus Orbignva (Table 2). Most of this proliferation has resulted from species descriptions based on inadequate botanical specimens, a problem inherent to palm taxonomy in general (Tomlinson 1979). In the more recent taxonomic literature (e.g. Rizzini 1963; Bondar 19641; Glassman 1977, 1978), most of the names listed in Table 2 are considered synonymous, and it is generally agreed that the babassu complex consists of a maximum of three species. Due to

potential confusion among these species, I embarked on a taxonomic study of the babassu complex. My objectives were (1) to identify the species that was the subject of the ecological studies described in the following chapters and (2) to distinguish it from morphologically similar species.


Lack of adequate specimens has been the principal impediment to resolving the taxonomy of the babassu complex. To remedy this situation, I collected representatives of the complex over a substantial portion of its presently known range (Fig. 15). Due to the unusual structure and bulkiness of large palms such as those in the babassu complex, complete botanical collections require extensive field documentation; recommended procedures are described at length in Balick et al. (1982). Specimens of all collections were deposited at the Museu Paraense Emilio Goeldi in Bel~m, Brazil; duplicates were sent to other Brazilian institutions (CENARGEN, INPA, CPATU) and the New York Botanical Garden, Bronx, N.Y., U.S.A.


L, c c c

cii 61 il

41 41 41 41

cd c; t o

cd ol 1 .2 2


61 61 c;l
61 61 cd

Z 7 61 41 61 c;l 1. c

C4 61 61


I o o .

lo Ac


Si. m


80 70 60 50 40

o".. ,:GUYANA ,0


0I BRAZl L -.-o
20-0 0' -"


:i ', .-., --.
s -,,. 2

20I I I I
80 70 60 50 40

Figure 15. Distribution of specimens of Orbignya martiana and its
suspected hybrid complexes. Circles represent 0. martiana; squares represent the Orbignya-Maximiliana complex; triangles represent the 0. martiana-0. eichleri
complex. Sources: Wessels Boer (1965), Glassman (1977),
and M. J. Balick and A. B. Anderson (unpublished).


A first-hand familiarity with the taxa in the field provided a sound basis for reviewing the extensive and often contradictory literature on nomenclature. Results pertaining to the morphology and distribution of taxa in the babassu complex represent a synthesis of data already published, data obtained in the field, and morphological comparisons of specimens in the laboratory.

Results and Discussion

Examination of data on specimens obtained over a substantial area indicates that the babassu complex is comprised of one highly variable, widely distributed species and two associated species that appear to hybridize with the former. Babassu is henceforth used exclusively in reference to the principal species of the complex, which was the subject of the ecological research described in the following chapters. Nomenclature

The scientific name of babassu tentatively utilized in this study is Orbiznva martiana Barb. Rodr. (Table 2). Its valid publication by Barbosa Rodrigues (1898) represents the first taxonomic description that can be definitely attributed to babassu. Previously published names associated with babassu rAttalea speciosa (Mart.) Barb. Rodr. and Orbiinva lvdiae Drude] were based on incomplete collections, and their accompanying descriptions were either insufficient or incorrect; neither can be definitely attributed to babassu. Barbosa Rodrigues (1903) subsequently reduced O. martiana to synonymy under the new combination, O. speciosa (Mart.) Barb. Rodr. Although widely accepted in Brazil, the latter name is invalid because the same combination (Q_ speciosa Barb.


Rodr.) had been previously published by Barbosa Rodrigues (1891) in reference to another species (cf. Article 64 International Code of Botanical Nomenclature: Stafleu 1972, p. 59). Orbignva martiana is therefore the earliest, validly published name definitely referable to this taxon and it maintains precedence over subsequently published names associated with babassu (Table 2).

My use of 0. martiana is tentative, due to an older name that remains to be considered. Martius (1844) published a relatively complete description of D phalerata Mart., which he designated as the type species of the genus. Subsequent collections of this species have apparently not been obtained (Glassman 1977), and Martius' published description and illustration provide no basis for distinguishing it from 0. martiana. The lack of specimens of 0. chalerata prompted a recent (July-August 1982) expedition, sponsored by the New York Botanical Garden, to the type locality in eastern Bolivia. Final judgment

concerning the relationship of these two taxa awaits examination of the

specimens obtained.

Description and Differentiation

A technical description of babassu, based on specimens obtained to date, is provided in Appendix C. This description illustrates the extreme variability of the species. Despite its variability, babassu possesses a number of diagnostic features that together enable one to distinguish it from related species. These features include an erect, solitary habit; large, erect leaves recurved at the apex, with rigid leaflets disposed in a vertical plane and often with striations on the


outer (abaxial) surface of the sheath; large inflorescences that are either exclusively staminate or hermaphroditic, each containing a persistent, woody bract with a prominent apex; staminate inflorescences with numerous, erect branches (rachillae), each branch bearing two to four rows of staminate flowers on its lower (abaxial) surface only; staminate flowers with a small calyx, two or rarely three dentate-tipped petals, approximately 24 irregularly coiled and twisted stamens, and a pistillode; hermaphroditic inflorescences with numerous branches, each branch bearing one to rarely three pistillate flowers at the base and a variable number of staminate flowers at the tip, the latter either inviable or opening after the pistillate flowers; and a massive, oblong drupe with a thick, woody indehiscent endocarp and usually two to five oily seeds.

Due to its apparent propensity to hybridize, the distinction between babassu and at least two closely related species (Orbi~nva eichleri and Maximiliana maria) is somewhat blurred. I suspect that these latter species hybridize with babassu. Morphological comparisons (Appendices D and E) illustrate the intermediate morphology of the

purported hybrids ("Orbi~nva teixeirana" and "Markleva dahlgreniana"). Germination experiments have shown that both of the latter produce viable seeds and thus are capable of regenerating; however, they are only known to occur in association with their apparent progenitors. The distribution of babassu and its purported hybrids is shown in Figure 15. Anatomical, palynological, and electrophoretic studies of these complexes have been initiated by colleagues at the New York Botanical


Garden in the U.S. and the Centro Nacional de Recursos Gengticos (CENARGEN) in Brazil.

Distribution and Origin

Babassu has a widespread distribution in South America (Fig. 15). Dissemination of the large, heavy fruits has probably been exclusively by mammals (either living or extinct) or water. Among living mammals, rodents such as pacas (Agouti Daca) and agoutis (DasvDrocta Dunctata) are effective, short-range dispersal agents (see Chapter 3); humans and possibly monkeys disseminate the fruits over greater distances. Coevolution between hard-fruited palms such as babassu and large, Pleistocene mammals seems likely (Janzen and Martin 1982), but a paucity of paleontological evidence makes it impossible to assess the potential contribution of these now extinct animals to the current distribution of babassu.

Markley (1971) argued convincingly that the babassu palm originated on the Brazilian Shield in the state of Goias, a major center of diversity for the entire Attalea alliance. Since its origin during the Precambrian, the Brazilian Shield has undergone considerable erosion by numerous river systems. With few exceptions (discussed below), babassu areas from the Paraguay River to the Amazon Basin are connected by watersheds to the Brazilian Shield in Goi~s (Figs. 1 and 15). Throughout its range, babassu is largely confined to floodplains, river valleys, and former deltas, which lends support to the idea of dissemination by rivers. Although their high specific gravity (ca. 1.4 g fresh weight/cm3) prevents the fruits from floating, they can be


carried considerable distances by runoff' and rivers in flood. Lighter fruits would tend to be carried farther, which could explain why babassu fruits are relatively light (80-250 g fresh mass) over most of the

species' range, while the heaviest fruits known (up to 800 g fresh mass) are found in its presumed center of origin in Goias (Fonseca 19211; M. J. Balick and A. B. Anderson, unpublished data).

There are three important exceptions to the links between the

Brazilian Shield and babassu's present range. The first is its disjunct occurrence in Suriname and Guyana (Fig. 15). Until the early Pleistocene, the eastern portions of the Brazilian and Guayana Shields were connected, which probably accounts for the close floristic ties between the two regions (Maguire 1970). The Guianas represent a second center of diversity of the entire Attalea alliance (Wessels Boer 1965).

A second exception is babassu's occurrence in the Brazilian states of Maranhao and Piaui' (Fig. 1). Markley (1971) suggested that the Tocantins River may have once served as a link for dissemination of the palm to these states. The Tocantins has its headwaters in the Brazilian Shield and presently drains into the Amazon River. During the

Cretaceous, however, it may have flowed across the northern part of Maranhao and built up the huge delta that presently covers most of the northern third of the state. It seems unlikely that this delta could have been built by the relatively insignificant rivers of Maranhao that exist today (Pindar6, Grajau, Mearim, Itapecuru, and Parnai"ba), most of which have their headwaters only a short distance from the delta. Uplift of the Serra do Gurupfi in western Maranhao (probably during the


lower Tertiary) may have forced the Tocantins to change to its present course.

A final anomaly in babassu's distribution is its disjunct occurrence in the Brazilian state of Ceari (Fig. 1). Here the palm is confined to isolated and geologically very old uplifts, and there are no river systems that run between these structures and the highlands of Goigs. The uplifted areas to which babassu is confined in Ceard have long served as favorable sites for human settlement and agriculture, due to their relatively cool temperatures, high rainfall, and fertile soils. Markley (1971) argued that babassu's occurrence in Cearg is due to dissemination by humans. Early aboriginal groups may have introduced fruits obtained from the Parnafba valley ca. 200-300 km distant.


In autecological studies of plants, the various components of

reproduction (e.g., phenology, pollination, dispersal, predation, germination) are often analyzed as separate entities. Yet these events are

obviously interdependent. The timing of flowering, for example, may be related to competition for animal pollinators (e.g., Rathcke in press). Tight interdependence between the often conflicting demands of seed dispersal, escape from seed predators, germination, and ultimately establishment has been repeatedly documented (e.g., Stebbins 1971, Harper 1977, Silvertown 1981, Willson 1983).

This chapter commences the life history of babassu with an analysis of the palm's reproduction under a range of ecological conditions. Thus, comparative phenological observations were conducted in the major ecological zones of Maranhao where babassu forests occur. The widespread changes in land uses currently underway in Maranhao are likely to have a profound impact on the reproductive biology of local populations of the palm (e.g., see Moore 1977, Janzen 1978a). I therefore carried out studies of babassuls phenology, floral biology, dispersal, predation, germination, and seed fate under a range of land uses associated with babassu.





I carried out monthly phenological observations of babassu from January 1981 to March 1982. All adult palms at each site were labelled and mapped. On the intensive study sites at Lago Verde, the sample consisted of all adults in a 0.5-ha area; in each of the satellite sites (Pindar6 Mirim and Caxias), 60 adults were randomly selected for observation. The former site had been recently converted to pasture, whereas

the latter was in fallow following its use for shifting cultivation in 1979-80. These sites are representative of current management practices

of babassu stands in the seasonally wet savanna (Pindar Mirim) and dry woodland (Caxias) ecological zones of Maranhao. On the site at Pindar6 Mirim, leaves of babassu had been harvested for thatch ca. 2 yr prior to initiation of the study.

To permit observations of leaf production and loss, the palms were climbed and green leaves were labelled by spraying paint on the undersides of the rachises. To enhance the visibility of newly emerged leaves, ca. 10 basal leaflets per side (representing ca. 5% of the total) were removed from marked leaves. I calculated leaf production per month by counting the newly formed (unlabelled) leaves that had completely expanded in the crown; loss per month was determined by counting the number of labelled leaves that had fallen to the ground. Fallen leaves were cut in half after being counted so as to prevent being recounted in subsequent months, and the leaves of adjacent palms were


painted with different colors and/or patterns to avoid confusing the sources of fallen leaves.

As in other palms, the reproductive structures (inflorescences and infructescences) in babassu are relatively few, large, and permanent. In addition to obtaining separate data on each structure, I could thus deduce its phenological activity during the ca. 1-mo interval since the

last observation. To prevent confusion among several flowering and fruiting structures borne simultaneously on a palm, phenological observations were accompanied by a measurement of the compass bearing of each structure in relation to the palm stem. In the case of functionally female inflorescences (containing at least one pistillate flower and a

variable number of staminate flowers that were usually inviable), I distinguished the following categories: (1) inflorescence in flower (pistillate flowers yellow); (2) infructescence with immature fruits (fruits tan to brown and not falling); (3) infructescence with mature fruits (fruits falling); and (4) infructescence with no fruits (all fruits fallen). In the case of male inflorescences (containing exclusively staminate flowers), I distinguished the following categories:

(1) inflorescence in flower (flowers intact or falling); (2) inflorescence in flower during the past month (flowers fallen, inflorescence axes yellow to light tan); and (3) inflorescence in flower prior to past

month (inflorescence axes tan).

On the three study sites at Lago Verde, fruit production per month was measured from September 1980 to April 1982. Mature infructescences were harvested, and the number of fruits produced per infructescence was


determined by counting fruits and/or calyces (in the case of abscised fruits). A subsample of ca. 30 fruits from each infructescence was airdried and weighed to the nearest gram.

Considerable variability in the ratio of male to female inflorescences (sex ratio) was observed among the palms in the phenological study. To compare sex ratio variability on the population level, I measured the distributions of inflorescence types in six populations. Distributions were obtained by determining the percentage of each inflorescence type per palm. The inflorescences tend to remain attached

for several years following anthesis; my counts thus provided a longterm picture of reproductive behavior. Data were obtained from two geographically isolated populations located in Tocantin6polis, Goias, and Pirapora, Minas Gerais. Additional data were obtained from four populations located within my phenological study sites at Lago Verde, Pindar4 Mirim, and Caxias; all of the latter sites occur within the more or less continuous band of babassu stands that stretches across northcentral Maranhao (Fig. 12).

Floral Biology

To determine duration of flowering in individual inflorescences, I carried out daily phenological observations from 1-15 March 1982 on 38

palms on the secondary forest site and 15 palms on the pasture site. Palms with bracts about to open were selected for observation. Flowering was judged to begin upon opening of the bract. Proportion of

flowers that ultimately became fruits was determined through examination


of a random sample of 10 infructescences obtained in each of the two habitats.

Studying pollination in babassu required recognition of female inflorescences before they opened. As no morphological differences could be discerned between the two inflorescence types prior to anthesisq I distinguished them by opening a small slit in the bract and

examining the flowers within. In female infloreseences, the slit was subsequently sealed with tape to prevent contamination.

To determine whether self-pollination occurs in babassu, sleeves were designed to exclude external sources of pollen from female inflorescences. Each sleeve was made of light-weight canvas in the form of a tapered cylinder with a length of ca. 2.5 m and a maximum diameter of ca. 0.5 m. Circular wires were sewn on the inside of the sleeve to provide -support, and a small polyethylene window was placed in the middle to allow observation of the inflorescence. Entry by insects was prevented by painting bands of insect trapping adhesive over the base and tip of the inflorescence bract; one end of the sleeve was placed over each band and securely shut with wire.

Small bags were designed to isolate individual pistillate flowers from pollen sources. To reduce chances of contamination, the entire inflorescence was covered with a sleeve prior to opening. Each flower was individually bagged after the inflorescence had opened, and the sleeves were subsequently removed. Time of pollination was estimated by

exposing pistillate flowers to external sources of pollen at different periods following opening of the bract. To test whether parthenogenesis


occurs in babassu, flowers were left bagged throughout the flowering period, and I removed the pistils from others; both sets of flowers were subsequently checked for fruit set.

Abundance and behavior of insect visitors to both inflorescence types were determined by direct observation and collection. Sticky traps were used to monitor insects between observations. Insect specimens were identified at the Museu Paraense Emiltio Goeldi in Bel6m; unidentified specimens were sent to specialists in Brazil and the U.S.

To determine whether wind pollination occurs in babassu, sleeves were designed to exclude insect visitors from female inflorescences while permitting penetration of wind-borne pollen. The sleeves were made out of nylon netting with a ca. 0.5-mm mesh, sufficient to exclude all observed insect visitors. Dispersal of pollen by wind was monitored through the use of pollen traps, which were distributed at regular intervals on both the secondary forest and pasture sites. The trap consisted of a microscope slide covered with a thin coat of vaseline; the slide was placed inside a horizontally oriented section of 15-cmwide PVC pipe secured to a stake or stem. After a 1 -wk exposure, the slides were removed, covered, and subsequently examined in the laboratory.

Fruit Biology

Dispersal. I monitored removal of babassu fruits in the pasture and secondary forest, as well as on a nearby field. This latter site had been cut, burned, and planted at the end of 1980. During the course of the study, the site remained virtually abandoned except for sporadic


harvesting of cassava, and dense regrowth became established. At each of the three sites, 10 one-layered clusters of 25 fruits each were set up, with a stake marking the center of each custer. Fruit movement was measured at 1- to 2-wk intervals. The search for dispersed fruits was conducted to a distance of 4 m from the stake. To label fruits, a 14-mm staple was driven into the apex of each fruit. To eliminate human odor, fruits were handled with gloves that had been thoroughly rubbed with leaves and soil. I tested potential effects of the labelling technique

on dispersal agents by setting up four pairs of fruit clusters (25 fruits per cluster) on the secondary forest site. Dispersal was arbitrarily defined as movement of more than 50 cm. No significant difference (p = .19) was found in dispersal of labelled and unlabelled fruits, according to the randomization test for matched pairs (Siegel 1956).

Predation. To monitor colonization by seed-predaceous beetles, six

fruit clusters were set up on the secondary forest site. I prevented dispersal by stapling the apices of the fruits to a nylon line (10

fruits per line) attached to a stake. Twenty fruits were gathered from each cluster at intervals of 0, 109 209 /09 80, and 160 d following initiation of the experiment (27 October 1981). The fruits gathered from each cluster were isolated in closed sacks to prevent further predator colonization. After air drying, the fruits were opened to determine total number of seeds, number of attacked seeds, and number of beetle larvae per fruit.

Seed fate. To monitor the fate of seeds over time under various environmental conditions, five fruit clusters were set up in each of


three sites: secondary forest, pasture, and field. I prevented dispersal by stapling the fruits to nylon lines attached to a stake. Twenty fruits per pile were gathered at the outset of the experiments (27 October 1981) and at 3-mo intervals during 1 yr.

I obtained a long-term (ca. 18 mo) measure of seed fate from fruits used in the irrigation experiment (described in Chapter 4). At the end of this experiment, 20 fruits were gathered from each of five weeded (sun) and five non-weeded (shade) plots that had not been irrigated. In both seed fate experiments, gathered fruits were stored in closed sacks,

air-dried, and opened to determine numbers of eaten, defective (aborted or rotted), germinated, and quiescent (viable but not germinated) seeds.

Germination. An experiment was set up on the secondary forest site to monitor germination over time and to test for between-palm variability in fruit germination. Recently f allen f ruits (100 f rom each of 10 palms) were dehusked to inhibit dispersal and planted in 40 Plots (25 fruits per plot, four plots per palm). At ca. 1-mo intervals, I measured germination by checking the base of each fruit for emergence of one or more embryonic axes (i.e., plumule, cotyledonary node, and radicle).

I conducted additional experiments to test the effects of fire, herbivory, and shading on germination of babassu fruits. These experiments were established on a cleared, periodically weeded site ca. 50 m from secondary forest on well-drained soil. The fruits used in each of these experiments (100 per treatment) were obtained from a single palm.

Unless otherwise specified, fruits were dehusked and placed on plots shaded with suspended frames upon which babassu leaves were tied. At 1-


wk intervals, the frames were lifted to check for germination. To test the effects of fire, I exposed fruits to each of the following treatments: (1) a prolonged, high-intensity fire, simulating a forest fallow burn; (2) a rapid, low-intensity fire, simulating a pasture burn; and

(3) no fire. To test the effect of consumption of the fruit mesocarp by

herbivores, I compared germination in fruits that had been dehusked (epicarp and mesocarp removed, simulating the effect of herbivory) and fruits that had not been dehusked. To test the effect of shading, I planted 100 fruits on an exposed plot and 100 fruits on a plot shaded as

described as above.



Leaf turnover. Figure 16 shows data on leaf turnover of babassu at the wet (Pindar Mirim), intermediate (Lago Verde), and dry (Gaxias) sites; similar data from the primary forest, secondary forest, and pasture sites are shown in Figure 17. At all sites, leaf production peaked during the two rainy seasons and leaf loss peaked during the single dry season of observations. As in palms generally, leaf production in babassu is synonymous with growth; the rainy season is, therefore, the growing season for the palm. Over a 1-yr period after initiation of observations, no difference was detected among leaf production rates per palm at the wet (_x = 3-9) intermediate (5i = 4.1 ), and dry (_x = 3.7) sites. By contrast, Duncan's Multiple Range test revealed significant differences (p :S.05) in rates of leaf production per palm at the three Lago Verde sites; the order was pasture (Rc = 5.1) > primary forest


1981 1982J F M A MJ J A S O N D J FM
+ 1.0 (N= 60)

O _0 0 5.

+1.0 (N= 143

Z +0.5

o -0.5
o 1.0H
g -1.5
U- +1.0 (N 60)
< +0.5


Figure 16. Mean monthly gain and loss of leaves per palm on three
sites in Maranhao. Wet site = Pindare Mirim, intermediate
site = Lago Verde, and dry site = Caxias. N= number of
palms observed per site. ND = no data.


1981 1 1982J F M A M J J A S 0 N D J F M

+1.0 (N=21)
+ 0 N

-< -0.5

X -I.5

+1.0 N=75)
O +0.5
O = -i = -=
o -0.5
0 -1.0
CL -I.5
La_ +1.0 -(N= 44)
+0-5- 1 .0

Figure 17. Mean monthly gain and loss of leaves per palm on three
study sites at Lago Verde. N = number of palms observed
per site. ND = no data.


4-4) > secondary forest (R = 3.7). These results can probably be attributed to differential shading of adult palms on the three sites. Shading was lowest in the pasture. Potentially high shading in the primary forest may have been reduced by the relatively pronounced canopy

roughness (which thus permitted greater lateral penetration of light) and the emergent status of many of the adult palms. In the secondary forest, shading was highest because the dense, uniform babassu stand formed a closed, interlocking canopy.

Except at the wet site, leaf production and loss rates were approximately equal after 1 yr of observations; net gain for this period averaged .05 leaves per palm, which is virtually steady state. Assuming steady state conditions, one can divide the number of leaves borne per palm (R = 18.2, n = 199) by annual production per palm (R = 4.1 n = 199) to obtain an average leaf lifespan of 4.4 yr. At the wet site, production rates were far greater than loss rates; the net gain per year was 3.3 leaves per tree. The low losses were probably due to leaf harvesting for thatch 2 yr prior to initiation of the study. Yet gross leaf production (X_ = 3-9) was not significantly different from the combined mean obtained from the other sites (3Z = 4.1 ). I thus suspect that the single defoliation had little effect on leaf production rate at the wet site.

Flowering and fruiting. Data on inflorescence production for both

sets of sites are shown in Figures 18 and 19. Like leaf production, flowering of babassu peaked during the rainy season and was lowest during the dry season. The similarity of inflorescence and leaf Dheno-


1981 '1982J FMAMJ J AS ONDJ FM

(N= 60)
-j 0 o U

C 0.5

(N= 143)

o 0-L


Z 1.0



Figure 18. Mean monthly production of male and functionally female
inflorescences per palm at three sites in Maranh~o. Wet
site = Pindar& Mirim, intermediate site = Lago Verde,
and dry site = Caxias. N = number of palms observed per


1982J F MA MJ J AS 0 N D J FM


w .
z O -(N=75)
0 0

0 0.5
z 1.5

O : 0(N:44) 7
O -- t-


Figure 19. Mean monthly production of male and functionally female
inflorescences per palm at three study sites at Lago
Verde. N = number of palms observed per site.


logies is probably due to the physical association of' the two structures; each inflorescence is attached to the inner (adaxial) surface of' a leaf' sheath. However, a ca. 1-yr lag occurred between a leaf's em ergence in the crown and the opening of' its associated inflorescence. Earlier inflorescence expansion was apparently prohibited by the tight packing of' newly emerged leaves near the top of' the crown. Only continued growth of' the palm during the following rainy season loosened this packing sufficiently to permit inflorescence expansion.

Dissection of' crowns showed that each leaf' base contains a single inflorescence. Under suitable conditions, all inf'lorescences ultimately

develop to anthesis, thus resulting in an inf'lorescence-to-leaf' production ratio of' 1. Production data over the entire period of' observations revealed high ratios f'or the dry site at Caxias (1.02) and the pasture at Lago Verde (1.04). (The possibility of' obtaining inflorescence-toleaf' production ratios greater than 1 results from counting these structures simultaneously. A more accurate ratio would be obtained by comparing production of' leaves in a given year and production of' their corresponding inflorescences during the following year.) Shading probably caused intermediate ratios on the primary forest (.92) and secondary forest (.89) sites at Lago Verde. The low ratio at the wet site (.69) can be attributed to the leaf' harvest 2 yr previously. Babassu's apparent response to defoliation was to increase inflorescence abortion rather than to decrease leaf' production.

No difference in the phenology of' male and female inf'lorescences was detected. Approximately 20% of' total inf'lorescences were female in


all populations observed, despite high variability (0-74%) among individual palms.

Variability on the population level was analyzed by comparing the distributions of female inflorescences per palm in six populations (Fig. 20). Despite a wide range of ecological conditions, the four contiguous populations (forest and pasture at Lago Verde, wet site at Pindarg Mirim, and dry site at Caxias) showed no differences in distributions of female inflorescences per palm, according to the Kolmogorov-Smirnov Test (Siegel 1956). However, distributions in the two disjunct populations (northerly site in Goigs and southerly site in Minas Gerais) were significantly different from the first four populations, as well as from each other (p .01 in all cases). These results suggest that the distribution of female inflorescences per palm in populations of babassu is governed by genetic factors. Potential implications of this finding are

discussed in Chapter 7.

Fruit production. At all sites, frequencies of immature infructescences peaked as flowering ended in April-June and were lowest when fruiting peaked in October-December (Figs. 21-22). This pattern was consistent despite the wide range of climatic conditions among sites (Fig. 13). According to phenological observations, there was a ca. 9-mo lag between the flowering peak (January-March, Figs. 18-19) and the fruiting peak (October-December, Figs. 23-24). Observations on individual palms confirmed that fruits require an average of 9 mc to reach maturity.


40 (N=75) (N=41)



40 (N=60) (N=60)

30Cn 2010

CL 0 I I I I i F1

40 (N=25) (N=25)



0 I 1 7
0 20 40 60 80 100 0 20 40 60 80 100

Figure 20. Distributions of female inflorescences per palm in six
populations of babassu. Forest and pasture sites = Lago
Verde; wet site = Pindar Mirim; dry site = Caxias;
northerly site = Tocantin6polis, Goias; and southerly site = Pirapora, Minas Gerais. N = number of palms observed per site.


1981 1982J F M A M J J A S 0 N D J FM

cr 0


-(N= 143)
O---=--- -=-----------0 -'- -- 7-rw z

z Li

O 1.0
w 0.5 DRY SITE

LL 0per palm at three sites in Maranho. Wet site = Pindar
.Mirim, int numbermediate site = Lago Verde, and dry site
Caxias. N = number of palms observed per site.


1981 I 1982J F M A M J J A SONDJ FM

i~~~o.~ ~ LllIU iTDLI

n 0.5


a 0.5 -PASTURE
L w

1.0 E





Figure 22. Mean monthly number of immature and mature inflorescences
per palm at three study sites at Lago Verde. N = number
of palms observed per site.




)- 200-

a 0- rf l 1 r


- 6005

IL 400- -200
20 1 F

o0 n n ] l l
J F M A M J JASONDJ FMA 1981 1982

Figure 23. Monthly fruit production per ha on three study sites
at Lago Verde. Values represent air-dried mass.


7 -(N=24)

-~ 4


a. 3 (N = 75)

0 8 (N=44)
a: 7
F- 5


1981 1982

Figue 2. Man mnthy fuitproduction per palm on three study
stsat Lago Verde. Values are air-dried mass. N
numbr ofpalm persite.


Fruit yields per ha and per palm (Table 3) were highest in the pasture,, probably due to the moderate density and relative openness of' the babassu stand on this site. Per-ha yields were lowest in the primary forest and intermediate in the secondary forest. This is probably due to differences in palm densities, which were low in the primary forest (48 mature palms/ha) and high in the-secondary forest (150 mature palms/ha). Conversely, per-palm yields were intermediate in the primary forest and low in the secondary forest, probably due to increased shading of mature palms on the latter site.

All three sites showed substantial increases in yields between the 1980-81 and 1981 -82 fruiting seasons. On all sites the principal source of these higher yeilds was an increase in the number of palms that bore fruit; increased numbers of infructescences per fruiting palm constituted the second most important source (Table 4). Floral Biology

Fruit set. In the secondary forest, 78.4+% (standard deviation= 24.3%) of flowers became fruits, 19.9% (24.3%) either aborted or were not fertilized, and 1.7% (1.7%) were apparently attacked by the larvae of an unidentified weevil. In the pasture, the figures were 82.1% (20.7%)v 17.1% (18.8%), and 0.8% (2.1%) respectively. These numbers indicate high breeding success of babassu in both habitats.

Parthenogenesis. Fruit set consistently failed to occur in flowers that were not exposed to pollen sources (n = 31 on three inflorescences), as well as in flowers in which the stigmas had been removed


Table 3. Fruit yields of babassu per ha and per palm during 1980-81 and 1981-82 on three sites. Values are
oven-dried mass in kg.

1980-81 1981-82 1980-81 1081-82

Primary Forest 4l73 732 9.9 15.2

Secondary Forest 1328 1650 8.8 10.9

Pasture 1356 2866 15.4 32.6


Table 4. Sources of increased fruit yields of babassu per
ha between 1980-81 and 1981-82 on three sites.
Numbers represent percent of total increase in

Primary Forest Secondary Forest Pasture Yield per Infructescence -6.5 -26.5 +12.3

Number of Infructescences +39.4 +35.6 +24.7
per Fruiting Palm

SUBTOTAL +32.9 +9.1 +37.0
(Yield per Fruiting Palm)

Number of Fruiting Palm +67.1 +90.9 +63.0

TOTAL 100.0 100.0 100.0


(n = 8 on one inflorescence). These results indicate that parthenogenesis does not occur in babassu.

Self-pollination. Inflorescences in babassu are generally either male (containing exclusively staminate flowers) or female (containing at least one pistillate flower and a variable number of staminate flowers that are usually inviable). The former are far more numerous than the latter. During a 3-wk period in March 1981, 21 (12%) of 173 inflorescences examined contained pistillate flowers. Of these 21 inflorescences, only three (14%) contained staminate flowers that shed pollen and thus appeared to be viable. Consequently, most inflorescences that contain pistillate flowers are exclusively female in a functional sense.

In a pistillate flower bagging experiment, fruit set occurred only in flowers exposed 0-48 h after opening of the inflorescence; subsequently exposed flowers did not set fruit. Thus pistillate flowers appear to be receptive for a maximum of 48 h after inflorescences open. In the three female inflorescences with functional staminate flowers that I observed, pollen release invariably began more than 48 h after opening of the inflorescence. In one of these inflorescences, fruit set did not occur when external sources of pollen were excluded by bagging. Thus pollen transfer from viable staminate flowers to receptive pistillate flowers on the same inflorescence appears to be prevented by protogyny.

Pollen transfer from male inflorescences to receptive female inflorescences on the same palm likewise appears to be rare. On a given palm, simultaneous opening of the two inflorescence types is unlikely.


In 21 female inflorescences observed, overlap of receptivity (0-48 h after bract opening) with pollen release from a male inflorescence on the same palm occurred only once.

The possibilities of inducing self-pollination in babassu are unknown. As inflorescences containing pistillate flowers are either functionally female or (more rarely) protogynous--and opportunities for pollen transfer between inflorescences on the same palm are rare --selfpollination is probably unusual under natural conditions.

Insect pollination. Presentation of complete data on abundance-and

behavior of insect visitors to inflorescences of babassu awaits identification of the specimens sent to specialists. A high diversity and abundance of insects were observed visiting male infloresences, which are odoriferous and produce copious pollen. Conversely, relatively low insect diversity and abundance were found on female inflorescences, the pistillate flowers of which give off no noticeable scent and do not produce nectar. The only insect that occurred in abundance on both inflorescence types was an undescribed species of Mystrops. This ca. 2mm long nitidulid beetle was found in sticky traps placed among the leaves of palms that were not bearing inflorescences, which suggests that it lives in the crowns of babassu. Prior to opening of either inflorescence type, the beetles were observed to gather in the crown by the hundreds or sometimes thousands. In a sample of flowers obtained from an open male inflorescence, I consistently found more than one beetle per flower; subsequent calculations produced an estimate of 48,300 staminate flowers on the inflorescence. The beetles were ob-


served to feed on pollen and probably use both flower types as breeding sites. Although they were far less abundant on female inflorescences, more than one beetle was consistently observed entering each pistillate flower. Entry was made either along the petal margins or, more commonly, between the basally fused stigmas at the tip of the flower. Pollen of babassu was found in isolated collections of Mystroa sp. obtained from female inflorescences that contained no viable staminate flowers, indicating pollen transport from another inflorescence. The beetles were observed in both secondary forest and pasture, and similar specimens of Mystrops were collected from the crown of a babassu palm in Pirapora, Minas Gerais, indicating that the association is widespread.

Wind pollination. Transport of babassu pollen by wind was repeatedly confirmed by visual observations in both the pasture and secondary forest sites. Preliminary analysis of pollen trapping data indicates a higher abundance of wind-borne pollen in the open pasture than in the closed secondary forest. Variances in quantities of pollen trapped in both habitats were extremely high. Insect exclusion from one female inflorescence in each habitat resulted in 57% fruit set in the secondary forest (total = 180 flowers) and 92% fruit set in the pasture (total = 167 flowers). These results indicate that wind pollination occurs in both habitats but may be more effective on the relatively open pasture site.

Fruit Biology

Dispersal. The proportion of babassu fruits removed > 4 m was significantly higher (p < .01 one-way ANO VA) in the secondary forest


than in the field or pasture (Fig. 25). In the secondary forest, fruits were probably removed by pacas (Aouti paca) and agoutis (2,asyprocta punc ta ta). In the field and pasture, removal was effected either by domesticated animals or by runoff.

Predation. The only non-human seed predator of babassu observed at my study sites was a species of bruchid beetle, Pachyjrners nucleorum. Attack of fruits and seeds by this predator (Fig. 26) occurred only after fruit abscission (0 d). The rate of colonization increased sharply at 10 d and leveled off at 40 d. Although ca. 70% of fruits were colonized, only ca. 40% of seeds were destroyed; an average of 1.9 out of a total of 3.1 seeds per fruit escaped predation.

Among fruits collected 40 d or longer after abscission, 20.1%

suffered predation of all seeds. Colonization of these fruits by P. nucleorum was approximately the same regardless of the number of seeds per fruit (Table 5). Because each larva was limited to one seed, however, lower proportions of seeds were eaten as the number of seeds per fruit increased (Table 5). These results suggest that selection by seed predaceous beetles favors many-seeded fruits in babassu.

Seed fate. In my experiment on seed fate over time (Fig. 27), the percentage of defective (aborted or rotted) seeds did not vary significantly among sites. After 1 yr, most of the seeds on the secondary forest and pasture sites had succumbed to predation; removal of palm trees that harbor seed-predaceous bruchid beetles probably resulted in significantly lower levels of predation in the field (p _".05, one-way ANO VA). Largely due to differences in predation, percentage of quies-





I..i. I 5I0




Figure 25. Removal of babassu fruits on three study sites at Lago
Verde. Shaded portions of bars show percentage of fruits
removed >4 m.






- /
Ll / --+ FRUITS


0 40 80 120 160

TIME (d)

Figure 26. Attack of babassu fruits and seeds by the hruchid beetle,
Pachymerus nucleorum, over time. Bars represent 1 standard


Table 5. Attack of babassu fruits and seeds by the seedpredaceous beetle, Pachymerus nucleorum.

Total N Colonized Total N Colonized
N % N %

1 16 10 62.5 16 10 62.5

2 63 50 79.4 126 67 53.2

3 124 86 69.4 372 157 42.2

4 113 85 75.2 452 198 43.8

5 23 19 82.6 115 37 32.2
6 4 3 75.0 24 6 25.0



80 60

40 20

1-00 ir80 60

Cj 40 ~-20



80 60



1981 1982

Figure 27. Seed fate over time in three study sites at Lago Verde.


cent seeds was significantly greater in the field after 1 yr (p S .05, one-way ANOVA). The relative lack of shading in the pasture probably accounted for the significantly lower percentage of germinated fruits in this habitat (p : .01, one-way ANOVA).

In the experiment on long-term (18 mo) seed fate, the percentage of fruits with 2 1 germinated seed(s) was significantly higher (p '_' .01,V Chi-square test) on non-weeded plots (62%) than on weeded plots (1%). Conversely, the percentage of fruits with : l quiescent seed(s) was significantly higher (p .05, Chi-square test) on weeded plots (78%) than on non-weeded plots (28%). I suspect that these latter seeds were quiescent rather than dormant; their germination was probably low due to

insufficient moisture on the non-weeded plots. These results suggest that under relatively exposed conditions, babassu seeds can maintain their viability for extended periods.

Germination. Figure 28 shows babassu's cumulative germination over time, beginning at the peak of fruiting during the latter part of the dry season. Most germination occurred ca. 3 mo after fruit fall; high soil moisture conditions prevalent during the rainy season are probably necessary for successful germination. Although this seasonal pattern of

germination appears to be relatively constant in babassu, germination success of fruits obtained from different palm trees varied, considerably and in some cases significantly (Fig. 29). As all fruits were planted under identical conditions, these results suggest that observed intraspecific variability in babassu's germination may be under genetic control.


50- 20


3 0 ....

.. .. .. .


20 0....0.. 50..00..5
......T.M (days) ...

Figure. 28. Cumulativ germiatio ofbbas.futsoertme. ntr
vals ~ ~ ~ ~ ~ ..... rersnz tnaderr hddbr ersn
rainfall. ~ ~ ~ ~ ~ ~ ~ .. Daa.eeoband.rm.4Noebr.91. o2
April 1982..








9 10

0 20 40 60 80 100

Figure 29. Germination in fruits obtained from 10 babassu
palms. Each hatched bar represents one palm;
intervals represent 1 standard error.


Table 6. Germination of babassu fruits in four experiments. Asterisks indicate significant
differences (* for p -S05, ** for p .01).

Treatment Control

Removal of Mesocarp 54 34

Intensive Burn 10 54

Moderate Burn 58 54

Shading 66 0


The abundance of' seedlings and juveniles on burned sites has led to

the suggestion that fire promotes germination in babassu (Smith 1974). Removal of' the mesocarp by dispersal agents and shading could also stimulate germination in the palm. Results of' experiments designed to test these hypotheses are shown in Table 6. 1 found that a moderate fire, such as occurs annually in pastures, had no effect on germination success. However, fruits exposed to an intensive fire, such as occurs when forest fallow is burned in shifting cultivation, exhibited significantly reduced germination (p,-: .01, Chi-square test). Removal of the mesocarp, simulating the effects of dispersal agents, was found to promote germination of babassu (p ...05). Likewise, germination was significantly higher in shaded fruits than in fruits left in the sun (p <.O1). This final response probably results from enhancement of germination by higher soil moisture levels in the shade.



Babassu's phenology showed remarkable constancy over a wide range of ecological conditions. The general pattern was a pronounced peak in leaf production and flowering during the wet season, while leaf loss and fruiting peaked during the dry season. Despite its dry-season peak, flowering within a given population occurred in most months. Male inflorescences were generally more abundant than female inflorescences, and I could detect no seasonal variation in sex ratios within the populations observed.


Leaf and inflorescence production in babassu are strongly correlated due to the close physical association of the two structures. Nevertheless, most phenological studies of palms in their natural habitats (e.g., Mora Urpi and Solis 1980, Bullock 1981, Myers 1981, Beach in press) have dealt exclusively with reproductive components. To my

knowledge, studies of both leaf and inflorescence production in palms have been limited to cultigens under plantation conditions (e.g., Child 1974, Hartley 1977). In the African oil palm, for example, leaf production occurs throughout the year, which results in nearly continuous flowering (Corley et al. 1976); in babassu, leaf production and flowering tend to be seasonal. Likewise, the minimum annual rate of leaf production in oil palm recorded in the literature is ca. 18 leaves (Corley et al. 1976), compared to a mean of ca. four leaves in babassu. Probably as a result of its high rate of leaf production, minimum leaf lifespan in wild stands of oil palm in Nigeria is only 2.1 yr (Zeven 1967); in babassu, mean leaf lifespan is 4.4 yr. I believe that this latter figure has important implications for management of babassu under shifting cultivation, a topic explored further in Chapter 7.

As has been documented in many other species (e.g., Harper 1977, Janzen 1978b, Pifiero and Sarukhan 1982), fruit yields in babassu vary considerably from year to year. As is the case in oil palm (Hartley 1977), I suspect that climatic factors such as rainfall may have a significant effect on annual variability in babassu's fruit yields. The influence is probably most pronounced at the time of inflorescence differentiation rather than at fruit set. Examination of developmental


series shows that inflorescence types (male or female) can be clearly distinguished up to 12 leaves prior to the newest emerged leaf. Given an average production of ca. four leaves per year, this means that sex determination occurs at least 3 yr prior to leaf expansion in the crown, 4 yr prior to flowering, and nearly 5 yr prior to fruiting. Detecting possible relationships between climate and yield will therefore require

long-term studies of babassu's phenology. Pollination

My results indicate that reproduction in babassu is predominantly or perhaps exclusively via outcrossing, and that pollination is effected by nitidulid beetles and wind. Such a combination of pollination systems has been observed in other species of palms (Uhl and Moore 1977, Moore and Uhl 1982). The coconut (Cocos nucifera) appears to be pollinated by either honey bees or wind in Hawaii (Shodt and Mitchell 1967).

The oil palm was thought to be exclusively wind-pollinated (Hartley 1977), but a recent study (Syed 1979) has implicated weevils as pollinators within the palm's natural range in West Africa and thrips on the Malay peninsula. Likewise, Myers (1981) suggested that both wind and insects effect pollination in Raphia taedigera in Costa Rica. Nitidulid beetles are rarely cited as pollinators (Faegri and van der Pijl 1979),

but their role in the pollination of palms has either been documented (Essig 1971) or implicated (Dransfield 1982). Wind pollination in palms has been reported since antiquity, according to Faegri and van der Pijl (1979).


In babassu, insect pollination is probably predominant in forests, while wind pollination is likely to be enhanced in open habitats such as

pastures. Neither system is exclusive of the other, and both may have come into play during babassu's evolutionary history. Predominance of

male inflorescences; small (20 pm in diameter), monosulcate pollen (personal observation); and lack of nectaries all point to pollination

by wind (Moore and Uhl 1982). Yet the strong scent produced by babassu's staminate flowers and the bright inner surface of its inflorescence bracts suggest adaptation to insect pollinators as well. The combination of two pollination systems clearly enhances the palm's adaptability to a wide range of ecological conditions. Dispersal

The fruits of many palms are fleshy and colored, and obviously adapted for dispersal by animals (Corner 1966, van der Pijl 1969, Kiltie

1981). Birds are probably the principal dispersal agents of palms (McKey 1975), although mammals such as rodents and primates (including humans) also feed on their fruits and disperse their seeds (Burtt 1929, Janzen 1971).

The dispersal of babassu fruits by animals other than humans occurs almost exclusively within forested habitats, where pacas (Agouti paca)

and agoutis (Dasyrocta punctata) are abundant. These rodents carry fruits to less exposed micro-sites (often > 4 m distant, Fig. 25), where the starchy mesocarp is partially or completely consumed and the fruits abandoned. Scatterhoarding of babassu fruits by these rodents apparently does not take place. Other mammals that I observed or were reported


by local inhabitants to feed on the mesocarp (but not the seeds) of babassu fruits include porcupines (Ooendu prehensilis and Coendu sp.), spiny rats (Proechimys longicaudatus and Mesomys hispidus), and peccaries (Taassu tajacu and T. pecari, both locally extinct). None of these species appears to have a significant role in dispersal of babassu fruits.

My field observations indicate that water-mediated dispersal is a significant factor on relatively open sites such as pastures. Here soil

compaction results in increased surface runoff, by which fruits were observed to be carried distances of ca. 25 m. The potential role of water-mediated dispersal is far greater. Dissemination by rivers largely accounts for the present distribution of babassu over nearly half a

continent (see Chapter 2).

Although it remains undocumented, the role of humans in the dispersal of babassu is potentially significant. As discussed in Chapter 2, the present occurrence of babassu in the Brazilian state of Ceara may be due to dissemination by indigenous peoples. Today settlers throughout the Amazon region commonly transplant babassu to areas where it did not

formerly occur. On a local scale in Maranhiio, gathering of babassu fruits is a widespread activity, particularly among women. Intensity of fruit gathering is especially high on relatively open, accessible sites

such as pastures. Although the seeds from almost all gathered fruits are ultimately consumed, some accidental dispersal occurs during gathering and transport. More significantly, perhaps, is the role of human gatherers of babassu fruits as agents of selection. People prefer-


entially gather large, many-seeded fruits for extraction of the oil-rich seeds; smaller, few-seeded fruits are left to regenerate.

Although rodents, runoff, and people play significant roles as

dispersal agents, babassu's massive fruits (usually >100 g) are relatively immobile. This immobility constitutes a crucial limiting factor in the palm's ecology and largely accounts for its current distribution

(see Chapter 2). Reduced dispersal is the price babassu pays for enhanced protection from seed predators. Predation

Of 28 insect species known to feed on babassu (Table 7), nine are potential seed predators. Most of the latter feed on stored seeds; the only seed-predaceous insects under natural conditions are two species of

bruchid beetle: Pachymerus nucleorum and Carbruchus lipismatus. Of these, only the former was present at all of my study sites in Maranhao. Popularly known in Brazil as bicho do coco, the larvae of this beetle prey on the seeds of a wide variety of palms, including economic species such as carnauba (Copernicia prunifera), licuri (Syagrus coronata), piassava (Attalea funifera), tucum (Astrocaryum vulgare), coconut (Cocos nucifera), and African oil palm (Elaeis guineensis). The penetration of

babassu fruits by this beetle was described by Bondar (1936). Ovipositing beetles remain in the crown of babassu during the day; at night

they deposit eggs on the surfaces of fallen fruits. The odor of the fermenting fruit mesocarp probably attracts ovipositing beetles (see Janzen 1971). Such attractants appear to persist for a maximum of 40 d after fruit abscission, after which colonization of babassu fruits by


+ + + +
0 4)
43 4)

.f.1 CO 4->







0 CL

4J C/I


0 0 cd
L. 0 4-) 0 Cd
cd 0 0 a 0
0 0. :s 4.3 "-f 0 m -H c
(D Go L4 0 >
.0 4J r. 9.. :$I
cd ca :3 r-4 a :s (D

w ca (a Cd
L. co Go .0 4.3 r. V m
a4 0 :s :3 0 a 0 0
4.) ca -H L. L :3 0 a Cd
cu r-I 0 u 0 m L. 0 0 .,4 -0
0 -H L. a 0 (D v 4-1 fn
0 0 > r-4 0 a) q 0 4)
a) -H > 0 4 0 0 0 0 L4 0
0 cd 0
a) cd &0 I'd -H -0 cd -H o 4) Cd 0
44 0 ca a) CA O14 -H -4 1: r-4 u z v 94
0 q Cd 10 cd .0 a (D 0 r4
Cd -I V L. -H a 0 >% 8 ca -0
E-1 cd 0 q C\J 0 401 M ---T -H U) %0 El- 00 4 ON.0 0 0 10 CJ 0 M
Lo S4 a .0 -H m L4 a a] -H 4)
0 0 m a 43 Cd L. 8
E- .4.3 .0 co 0 m L. (D L.
cl. ul r4 v 4 0 f. 0 0) -4 w
0 q a4 0 w pa 0 0 u Cl
a C4 r-I
0 0 0


L. V 0 0
43 (D V2 cil

z C)


4J + + +

4-) :3
a :3 U3 L4
ca 0 ca
0 L4 S
.rj ca -H r-I
;.4 4z ca a
:3 q a
ul U2 0 Q ca m
ul 0 Cd U3
U3 co a z m
:s o Cd (d r-I m 0 m m 9. 0 m
4.3 0 -H 0 4.3 -H m 0 a PC 0 r-q
010 0 m 0 u 0 0 z N 0 0 =
13 0 -H m 0 0 -ri 0 x 4) -H 0 in
=s m Q > 'I > :3 0 L,

4) -H r-I 0 4-) co a m -H 0 .0 = 0 in
ca 4) m r-I co .11 m 0 S. 0 0 a L4 0 0 0 0
0 N ca (D 0 > 0 0 S. 4J 0 01 0 0 0 a
. x 0 > 10 0) c El r-q 0 r--q -H > = 4) L.
0 0 -H M 0 (D 0 A C 0 ca 0) ca as
E-4 M :14 =4 0 01 0 rn I
Cd CA mi
0 -H r. 0 rq
(0 'a -H a 1-1 -I r-I .
-.T r-A -,q Ln r-I ko t--- cd 00 ON C:) C\j 0 m X:r Ln %.o t- 0 w 0 S, 0 4.) N 5.. C4 CM CM N CV -0 CM
4J 0 Cd 0) >1 H
CL C pq 0 04 04 4.)
0 0 z -H



beetle larvae apparently ceased (Fig. 26). Approximately 15-18 d following oviposition, tiny (0.75 x 0.10 mm) larvae emerge and proceed to the base of' the fruits. Penetration takes place through the germination

pores, which offer minimal resistance. Prior to abscission, fruits are attached at the bases and their germination pores are thus protected, which explains why fruits are only attacked after they fall. Approximately 2-3 d after penetration begins, the larvae reach the seeds. Observations of fruit colonization revealed a maximum of one larva per seed chamber (locule), suggesting that late arrivals are probably cannibalhzed (see Janzen 1971); movement between seed chambers is prohibited by the thickness of the inner endocarp walls. One seed provides sufficient food for a larva's complete development. Adults emerge ca. 2.5 mo after oviposition by burrowing a ca. 8-mm-wide exit hole through the endocarp.

In a closely related palm (Scheelea rostrata) in Costa Rica, Bradford and Smith (1977) found that ca. 60% of the fruits lost all their seeds to a predaceous bruchid beetle (Caryobruchus buscki) or to rodents; in babassu, the total figure was ca. 20% (excluding potential losses to humans). A key component of this high predator escape is the predominance of many-seeded fruits in babassu (Table 5); in contrast, most (ca. 95%) fruits in the Costa Rican population of S. rostrata were one-seeded (Bradford and Smith 1977). In both species, the proportion of seeds eaten by predators declined as the number of seeds per fruit increased. Thus, in S. rostrata mean seed losses to predators in one-, two-, and three-seeded fruits were 61%, 46%, and 40%, respectively


(Bradford and Smith 1977); the corresponding figures for babassu--62%, 53%, and 42% (Table 5)--are remarkably similar. The trend suggests that in both species, selection by seed predators favors many-seeded fruits. However, such selection probably operates within strict constraints. Bradford and Smith (1977) found that in S. rostrata, individual seeds in one-seeded fruits contained 46% and 123% higher energy content than those in two- and three-seeded fruits,, respectively. Seedlings from one-seeded fruits should therefore have a competitive advantage, In addition, seedlings emerging from the same many-seeded fruit compete intensively with each other, further increasing the competitive advantage of seedlings from one-seeded fruits. The number of seeds per fruit thus represents a compromise between opposing selective forces.

The endocarp of babassu is exceptionally thick (R 49.8 mm,, s.e.

1.1) and resistant to crushing (Fig. 30) compared to other species of palms. It would thus appear to provide effective deterrence against seed predators. Yet the figures cited above indicate that levels of predation in one- to three-seeded fruits are almost identical in babassu and S. rostrata, despite the latter's comparatively thin endocarp (4-5 mm,, according to Bradford and Smith 1977). As described above, seedpredaceous beetles penetrate the fruits of babassu via the germination pores, where the endocarp is least resistant; entry is effected in 2-3 d

(Bondar 1936). The thick endocarp of babassu thus appears to provide little deterrence against the palm's most effective seed predator. Conversely, the endocarp completely deters other potential predators such as rodents and squirrels; only monkeys (e.g., Cebu~ apla and


Jessenia bataua (20)

Astrocaryum cf. mocrocalyx (34)

Socrotea exhorrhiza (35)

Iriartea ventricosa (32)

Mauritia flexuoso (7)

Scheelea sp. (19)

Phytelephas microcarpa (17)

Orbignya martiana (6)

I I i 1 I i i I I i I I I I I I I 1Ill
0.05 0.1 0.5 1.0 5.0 10.0
FORCE TO BREAK (t) Figure 30. Force required to break palm fruits. Intervals represent +
1 standard deviation; values in parentheses are sample sizes.
Data for Orbignya martiana from Fonseca (1924); data for all
other species from Kiltie (1982).


Aotus trikingatus, according to local informants) successfully prey on babassu by pulling off immature fruits and consuming the liquid endosperm within (see Izawa and Mizuno 1977). Thus seed predators either are completely deterred or successfully bypass babassuls thick endocarp. I suspect that there is no predator currently exerting sufficient selective pressure to account for the evolution of this extraordinary structure. Coevolution with large, Pleistocene mammals (Janzen and Martin 1982) thus provides the most satisfactory explanation for the origin of

the thick, massive fruits characteristic of babassu. Germination

The results of my experiments indicate that germination in babassu

is least successful on disturbed sites such as pastures and shifting cultivation plots, where shading is minimal, fires either frequent or intensive, and mesocarp-re moving dispersal agents rare. While germination success of babassu is thus reduced, its peculiar mode of germination confers upon the palm an extraordinary capacity to recover from the stresses inherent to these sites.

Germination in babassu occurs when one or more embryonic axes emerge from the germination pores at the base of the fruit. Each axis

enters the soil and penetrates to a variable depth, where it subsequently differentiates into roots and leaves (Fig. 31). This so-called

cryptogeal mode of germination is found throughout the Attalea alliance and in distantly related palm genera such as Acanthococos and Sabal (Corner 1966), as well as in many other plant families (Jackson 1974). In plants that exhibit cryptogeal germination, the shoot arises from


"' i:

-7; /

igure 31. Cryptogeal germination in babassu. From left to right,
seedlings were 1-, 4-, 8-, and 14-wk old. Scale = 15 cm.


below ground even though the seed germinates on the surface. In the case of babassu, the underground meristem in seedlings and stemless juveniles enables them to regenerate on sites that are subject to cutting and burning. Many species exhibiting cryptogeal germination occur in fire-prone habitats, and Jackson (1974) proposed that it represents an evolutionary response to fire. However, I suspect that babassu--as well as many of its relatives in the Attalea alliance-- evolved in moist, forested habitats where natural fires were infrequent (see Chapter 2). The presence of cryptogeal germination in these species is more likely due to the general lack of secondary growth throughout the family. Lateral expansion of the palm stem is accomplished by so-called establishment growth (Tomlinson and Zimmermann 1966), during which the stem takes on a highly unstable, obconical shape. One of the means by which palms provide support for this structure is to bury their shoot axes underground.

Regardless of the selective pressures that brought about its evolution, there is no doubt that cryptogeal germination plays an important role in the current proliferation of babassu on sites subjected to human disturbances.