White-Sand Vegetation of Brazilian AmazoniaAuthor(s): Anthony B. AndersonSource: Biotropica, Vol. 13, No. 3 (Sep., 1981), pp. 199-210Published by: The Association for Tropical Biology and ConservationStable URL: http://www.jstor.org/stable/2388125 .Accessed: 02/06/2011 15:17

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unlessyou have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and youmay use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .http://www.jstor.org/action/showPublisher?publisherCode=tropbio. .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printedpage of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact support@jstor.org.

The Association for Tropical Biology and Conservation is collaborating with JSTOR to digitize, preserve andextend access to Biotropica.

http://www.jstor.org

White-Sand Vegetation of Brazilian Amazonia

Anthony B. Anderson1

Instituto Nacional de Pesquisas da Amazonia (INPA). Caixa Postal 478, 69.00 Manaus. AM. Brasil

ABSTRACT This paper describes a distinctive vegetation type occurring on white-sand soils which are scattered throughout the Amazon region. Although essentially homologous throughout Amazonia, white-sand vegetation is known locally by various terms: to avoid confusion, exclusive use of the single term "Amazon caatinga" is proposed. The Amazon caatinga is distinguished from other regional vegetation types by its white-sand soil, scleromorphic physiognomy, and unusual floristic composition. Possible origins of the white-sand soil include: in-situ weathering, alluvial deposition, or podzolization. The scleromorphic physiognomy of the caatinga indicates a lack of nutrients and/or periodic water deficiencies in the soil. Extreme edaphic conditions and the insularity of these sites have probably acted as powerful selective forces, generating a biota character- istically low in diversity and high in endemism. Numerous endemic species may currently face extinction due to wide- spread burning of caatinga vegetation. Repeated fires on white-sand sites result in the effective arrest of succession; subse- quent mining of these sites for sand eliminates any possibility of vegetational reestablishment. Alternative benefits of the Amazon caatinga in its intact state are considered.

RESUMO Este trabalho descreve um tipo distinto de vegetacAo que ocorre em solos de areia branca espalhados pela regiao amazonica. Apesar de ser essencialmente homologo em toda a Amazonia, esse tipo vegetacional e' conhecido localmente por varios termos: para evitar confusao, o uso exclusivo do termo "caatinga amazonica" e proposto. A caatinga amazonica se distingue de outros tipos regionais de vegetacao por seu solo de areia branca, sua fisiognomia escleromorfica, e sua rara composicio floristica. As possiveis origens do solo de areia branca incluem: intemperismo in situ, deposicao aluvial, ou podzolizacAo. A fisiognomia escleromorfica da caatinga indica uma falta de nutrientes e/ou periodicas deficiencias de agua no solo. Ex- tremas condices edaificas e a insularidade desses sitios provavelmente tem atuado como poderosos fatores seletivos, gerando uma biota caracteristicamente pobre em diversidade e rica em endemismo. Especies endemicas estao em perigo de extinco devido a queimada da caatinga nas areas da Amazonia em desenvolvimento. Repetidas queimas desta vegetacao efetiva- mente impedem a sucess-ao; a extraao de areia para construcoes elimina qualquer possibilidade de re-estabelecimento vege- tacional. Os beneficios da preservacdo ida caatinga amazonica em seu estado natural sio considerados.

SOILS COMPOSED ENTIRELY OF WHITE SAND are scattered throughout the Amazon region of Brazil. Associated with these soils is a highly distinctive vegetation, which ranges in structure from savanna to forest and is characterized by pronounced sclero- phylly, low diversity, and high endemism. Due to its distinctiveness and relative simplicity, white-sand vegetation has been the subject of considerable scrutiny by botanists. Yet observations and studies in different areas of Amazonia have produced a frag- mented knowledge of this vegetation, as well as a confusing array of terms to describe it. Mounting pressures for development of the region increase the need for information concerning vegetation on oligo- trophic soils, which have exceptionally low agricul- tural po-tential. With the objective of developing a more synthetic terminology and vision, this paper examines general aspects of white-sand vegetation in Brazilian Amazonia.

TERMINOLOGY Evergreen sclerophyllous vegetation on white sand is

lPresent address: Department of Botany, University of Florida, Gainesville, Florida 32611, U.S.A.

recognized by distinct vernacular terms in each coun- try or region where it occurs (table 1). Such terms continue to appear in the literature, despite Beard's (1955) attempt to establish a universal terminology. In Brazil, however, there is even disagreement over which vernacular term or terms should be utilized. Spruce (1908) first used the term caatinga to de- scribe the vegetation under consideration. Later, Ducke (1922) and Sampaio (1945) referred to dif- ferent structural phases of this vegetation as campina and campinarana. Other authors cite terms such as humirizal, carrascal, chavascal, and charavascal (Au- breville 1961, Pires 1974).

In the botanical literature on Brazilian Amazonia, the terms most commonly encountered are caatinga, campina, and campinarana. As caatinga also refers to a totally unrelated vegetation type in the arid North- east of Brazil, Takeuchi (1960) suggested that the term not be applied in Amazonia, and subsequent authors have concurred. But they have not agreed on which terms should take its place: Pires (1974) used the terms "low campina" (campina baixa) and "high campina" (campina alta), Lisboa (1975) proposed "Amazon campina" (campina amazonica) and "Amazon campinarana" (campinarana amazon-

BIOTROPICA 13(3): 199-210 1981 199

TABLE 1. Vernacular terms for evergreen sclerophyllous vegetation on white-sand soils in the humid tropics.

TERMS campina caatinga baixa muri bush bana varillal bajo padang campinarana caatinga alta wallaba forest cunuri varillal alto kerangas

yaguacanan

REGION Central Amazonia Northwest Amazonia Guyana Venezuela Peru Borneo (Brazil) (Brazil)

SOURCE Ducke & Black Spruce (1908) Richards Herrera Revilla Whitmore (1953) Rodrigues (1961) (1952) (1977) (1978) (1975)

ica), and botanists of the Projeto Radambrasil (1976) used the single term "campinarana" with various descriptive epithets.

The present proliferation of terms obscures a general consensus that, despite structural and floristic variations, the vegetation under consideration rep- resents a continuum with many features in common. To resolve the confusion over terminology, each of the terms most prevalent in the literature is reviewed.

CAMPINA.-This term is used in Amazonia to refer to savannas and scrubs on white sand (Ducke and Black 1953); it is also used in other regions of Bra- zil to describe structural phases of cerrado and campo (G. Eiten, pers. comm.). In Amazonia the term campina,' meaning "small field," appropriately de- scribes reduced phases of the vegetation under consid- eration, which typically occupy small and disjunct areas over much of the region. The chief disadvant- age of campina is that it requires another term to describe the less-reduced phases, which range from low to moderately tall forests.

CAMPINARANA.-The meaning of this term, 'false campina," provides little or no descriptive informa- tion, and its application is confusing. Local inhabit- ants of central Amazonia, where the term originated, use it to refer to a variety of upland vegetation types of soils with a high percentage of sand. For example, non-sclerophyllous forests on sandy oxisols, which are high in diversity and low in endemism, are called campinarana: such forests bear no relation to the vegetation under consideration. Ducke and Black (1953) vaguely used the term to refer to a "transi- tional" vegetation between savannas and forests in Amazonia. Botanists of the Projeto Radambrasil (1976) used the term (with various epithets) to describe sclerophyllous vegetation on white sand, non-sclerophyllous upland forests on sandy oxisols, and palm swamps on sandy alluvium. The value of the term campinarana is therefore reduced by its imprecise meaning and variable application.

CAATINGA.-This term is derived from the Tupi language: "caa" means leaf, tree, or forest; "tinga"

means white (Rodrigues 1961), possibly in refer- ence to the relatively high light-penetration charac- teristic of white-sand vegetation. Caatinga has been used by local inhabitants and botanists alike to de- scribe such vegetation in the Rio Negro region. Hence the term is etymologically appropriate and, within Amazonia, unambiguous in its application.

To undersco,re the relatedness of white-sand veg- etation in Brazilian Amazonia, it is proposed that the term caatinga be applied to this vegetation throughout the region. With descriptive epithets, the single term can be used to refer to all structural phases of this vegetation. To prevent confusion when considering other phytogeographic regions such as the Northeast of Brazil, the term "Amazon caatinga" is recommended (cf. Ducke and Black 1953). This term is already accepted by phytogeographers in Bra- zil (e.g., Eiten 1978; J. M. Piers, pers. comm.) and has gained acceptance abroad (e.g., Sarmiento and Monasterio 1975, Klinge et al. 1978).

DELIMITATION Vegetation types homologous to the Amazon caatinga have a worldwide distribution in the humid tropics (cf. Whitmore 1975). In northern South America, they are reported in French Guiana (Granville, in press), Surinam (Heyligers 1963), Guyana (Rich- ards 1952, Fanshawe 1952), and the Amazonian portions of Venezuela (Beard 1955, Klinge et al. 1978), Colombia (R. E. Schultes, pers. comm.), and Peru (Revilla 1978). In Brazil, caatinga is one of the major vegetation types of Amazonia (Pires 1974) and occurs throughout the region (Ducke and Black 1953, Aubreville 1961, Sombroek 1966), generally occupying small and disjunct areas (fig. 1). How- ever, in the Rio Negro and Rio, Branco Basins of northwestern Amazonia, caatinga covers hundreds of thousands of square kilometers (Projeto Radam- brasil 1976; J. M. Pires, pers. comm.). It is not known to occur in other regions of Brazil, apparently reaching its southernmost extension at approximately Lat. 8-9? S (Soares 1953, Lleras and Kirkbride 1978).

Caatinga is limited to soils composed entirely of

200 Anderson

FIGURE 1. Aerial view of a caatinga scrub in central Ama- zonia. Total area of the caatinga scrub is approximately 1 hectare; the surrounding caatinga woodland occupies an area of less than 50 hectares.

white sand. Where these sands grade into soils con- taining even a small percentage (< 5%) of clay within the reach of roots, caatinga gives way to other vegetation types, such as non-sclerophyllous forests or floristically distinct campos. While caatinga generally occurs on topographically intermediate to high sites (the so-called terra firme), it is also found on low sites which are subject to flooding. Even on terra firme, caatinga soils may be hydromorphic, due to local drainage patterns or the presence of an under- lying hardpan (Viera and Oliveira 1962). Other sclerophyllous vegetation types of Amazonia, which have similarities and in some cases form a continuum with the caatinga, are as follows.

RESTINGA.-This vegetation type occurs on sand dunes adjacent to the Atlantic coast of Amazonia, extending south to Rio de Janeiro. Although phy- siognomically similar to reduced phases of the Ama- zon caatinga, the two vegetation types are floristically unrelated (Pires 1974).

AMAZONIAN CAMPO.-This vegetation type has a widespread distribution in Amazonia and is extreme- ly heterogeneous (Egler 1960, Pires 1974, Eiten 1978). In comparison to caatinga, fire typically plays a more prominent role in campo, and many of its component species (e.g., Curatella americana) are fire-resistant. Campo is best distinguished from caatinga by its soils, which generally contain var-

iable amounts of clay, and its relatively cosmopolitan flora. Yet gradations occur between the two veg- etation types. For example, a floristically intermed- iate vegetation is reported on red sands in Surinam (Heyligers 1963) and southern Amazonia (Lleras and Kirkbride 1978).

A curious type of campo, known as campo rup- estre ('"rock campo"), has many similarities with the vegetation under consideration. In Amazonia, campo rupestre is mostly restricted to the peripheries of the region and occurs on rocky outcrops which often grade into sandy soils. Although their sub- strates form a continuum, campo rupestre and caatinga have distinct physiognomic and floristic characteristics. The herbs, shrubs, and small trees of campo rupestre frequently have whorled leaves and candelabra-like branching (Eiten 1978), totally unlike growth forms in caatinga. Although both vegetation types have common floristic elements, campo rupestre has many unusual species and an even higher degree of endemism.

IGAPO.-This type, often associated with sandy soils, occurs in inundated areas adjacent to streams and rivers throughout Amazonia (Pires 1974). Its rela- tively low topographic position and more cosmopoli- tan flora best distinguish igap6 from caatinga. How- ever, both vegetation types form a continuum, with numerous physiognomic and floristic similarities. Like caatinga, igapo is drained by tea-colored water, which forms the black-water streams and rivers char- acteristic of the region (cf. Klinge 1967, Janzen 1974).

Although precise delimitation is not always pos- sible, the highly characteristic soils, physiognomy, and flora of the caatinga make it one of the most distinctive vegetation types of Amazonia. These as- pects will now be examined in greater detail.

STRUCTURAL AND FUNCTIONAL ASPECTS

STRUCTURE AND PHYSIOGNOMY The various structural phases of the Amazon caatinga can be conveniently recognized as follows (fig. 2).

CAATINGA SAVANNA (caatinga savanica) .-This phase is dominated by lichens and graminoids, which generally form a sparse cover over the bare sand. Occasional clusters of woody plants may be present, but they constitute a cover of less than 10 percent.

CAATINGA SCRUB (caatinga arbustiva). -This phase is characterized by open areas of bare sand and her- baceous plants, intermixed with shrubs and small trees up to approximately 7 m tall. The latter may

White-Sand Vegetation of Amazonia 201

35 HIGH FOREST ON OXISOL CAATINGA FOREST

'E~~~~~~~~~~~~~~~~~~~~~~~~~~~"

0O 25 50 7514

15 CAATINGA WOODLAND CAATINGA SCRUB

0 ~~~~~~25 50 '5

ItETERS

FIGURE 2. Profiles of oxisol forest and caatinga forest, woodland, and scrub. Upper profile from upper Rio Negro (Rodrigues 1961); lower profile from central Amazonia (Anderson 1978). Depths of upper and lower profiles are 5 m and 2 m, respectively.

form distinctive clumps or be distributed singly, forming up to 90 percent cover (fig. 3).

CAATINGA WOODLAND (caatinga arb6rea) .-This phase has a more or less continuous cover of shrubs and trees. The canopy is often patchy and variable in height, ranging from approximately 5 to 15 m, with occasional emergents to 20 m. Light penetration is high, permitting a dense understory of shrubs and small trees with thin trunks (fig. 4).

CAATINGA FOREST (caatinga alta) .-Here the can- opy, which reaches heights of 20 to almost 30 m, is generally uniform and continuous. Lower strata tend to form a patchy cover, which permits relatively high light penetration.

Although extremely variable, the structural phases of the Amazon caatinga constitute a continuum, with a number of characteristics in common. All are more or less reduced in biomass and have relatively high light penetration. Elements typical of tropical rain forests-such as drip-tips, tree buttresses, vines, and big woody climbers-are relatively infrequent or absent. Epiphytic orchids (cf. P. I. S. Braga 1977), bromeliads (cf. M. M. N. Braga 1977), and bryo- phytes (cf. Lisb6a 1976) are often abundant and high in diversity. Where woody plants predominate, the caatinga may be characterized by thick accumu- lations of humus and litter (cf. Stark 1970, Anderson et al. 1975), which are seldom encountered elsewhere in the humid, lowland tropics.

The physiognomy of caatinga plants is likewise distinctive, with a number of characteristics which suggest physiological stress. Shrubs and small trees typically have a dwarfed and rachitic aspect, with reduced quantities of foliage, thin branches, and small crowns; occasional emergents may have excep- tionally widespread, twisted branches (fig. 5). Stump

FIGURE 3. Panorama of a caatinga scrub. Reserva Biologica do INPA/SUFRAMA, BR-174 km 45.

r _a

FIGURE 4. View of a caatinga woodland. Reserva Biol6gica do INPA/SUFRAMA, BR-174 km 45.

sprouting is common in the caatinga, especially in structurally reduced phases. Although such features are characteristic of vegetation which is periodically burned, other fire-adaptive features are absent: bark is thin and smooth, the majority of species (includ-

202 Anderson

7~~~~~~~ 7,~~~~~~~~I

FIGURE 5. Branching of Aldina heterophylla (Legumino- sae), a common dominant of caatinga woodland and scrub in central Amazonia. Reserva Biologica do INPA/SUF- RAMA, BR-174 km 45.

ing herbs) are perennial, and virtually all species are evergreen. The woody vegetation is distinctively sclerophyllous, with relatively small, shiny, coriaceous leaves which are often held erect (fig. 6).

NUTRIENT DEFICIENCIES The curious structure and physiognomy of the caa- tinga are a reflection of extreme edaphic conditions, which may result from a lack of nutrients and/or periodic water deficiencies in the sandy soils. Pro- ponents of nutrient deficiencies as the principal limiting factor (e.g., Richards 1952, Stark 1971, Janzen 1974) emphasize the extreme poverty of soils associated with the caatinga and homologous vegetation types: these soils are deficient in mineral elements and have an exceptionally low cation ex- change capacity in comparision to the more wide- spread oxisols (table 2). The oligotrophic nature of

TABLE 2. Mineral soil properties of a non-hydromorphic, white-sand entisol (Manaus-Boa Vista high- way, km 45) and a fine-textured, yellow oxisol (Manaus-Itacoatiara highway, km 30) in cen- tral Amazonia. Data presented: mean + stand- ard deviation, with number of composite samples in parentheses. Samnple taken at 0-10 cm.

Property Entisol Oxisol

pH 4.7 ?0.4 (15) 4.4 ?0.2 ( 7) C (%) 0.76 0.28 ( 5) 2.26 0.28 ( 6) N (%) 0.03 0.01 ( 5) 0.25 ?0.01 (10) P in ppm 2 ? 1 (10) 4 ? 2 ( 7) CEC in meq/100 g 1.98 ? 1.37( 5) 8.49 ? 4.13( 7) Ca in meq/100 g 0.02 ? 0.01 ( 5) 0.1 ? 0.2 ( 7) K in meq/100 g 0.05 ? 0.03 (15) 0.08 ? 0.03 ( 7) Mg in meq/100 g 0.03 ? 0.01 ( 5) 0.4 ? 0.3 ( 7) Al in meq/100 g 0.6 ? 0.3 (15) 1.6 ? 0.2 ( 7) Base

saturation (%) 7 ?3 ( 5) 7.7 ?3.3 ( 6)

FIGURE 6. Sclerophylly in Pagamea duckei (Rubiaceae), a common species of caatinga woodland and scrub in central Amazonia. Reserva Biologica do INPA/SUFRAMA, BR- 174 km 45.

these soils is largely due to their origins, which may include: in situ weathering of impoverished parent materials, such as sandstones, quartzites, or granites; alluvial deposition of quartz sands originating from the Guiana or Brazilian Shield; or podzolization due to a fluctuating water table which leaches organic matter and clay constituents (sesquioxides) from the upper profile, leaving behind degraded sand. Tenta- tive nomenclature and specific examples of white- sand soils originating by these processes are sum- marized in table 3.

The leaves of tropical white-sand vegetation ex- hibit features which appear to be responses or adap- tations to nutrient-poor soils. Foliage in the Amazon caatinga is commonly yellow or chlorotic, yet with- out signs of pathogenic infection, which may indicate a widespread response to nutrient deficiencies in these habitats. Ferri (1960) suggested that the coriaceous nature of leaves in the caatinga is due to a low supply of mineral nutrients in the soil; a similar case of scleromorphism induced by a lack of nutrients has been documented in the cerrado of central Brazil (Arens 1963, Goodland 1971). Jan- zen (1974) predicted that leaves of tropical white- sand vegetation contain exceptionally high levels of phenolics and other secondary compounds. This prediction has been substantiated by evidence from Africa (McKey et al. 1978). As a causal mechanism, janzen (1974) hypothesized that plants growing on

White-Sand Vegetation of Amazonia 203

TABLE 3. Tentative summary of the major soil types associated with caatinga vegetation in Brazilian Amazonia. Nomen- clature follows Soil Survey Staff (1975).

Order Suborder Origin Example Source

Weathering in situ of par- Serra do Cachimbo Soares (1953) ent material

Entisols Aquents, Psamments Fluvial transport & deposi- Manaus G. Ranzani (pers. comm.) tion

Spodosols Aquods, Humods Podzolization S. Paulo de Olivenca Sioli & Klinge (1961)

poor soils are under increased selective pressure to evolve anti-herbivore defenses. Yet Davies et al. (1964) showed that plants grown in nitrogen- or phosphorous-deficient soils produce higher concen- trations of phenolics than whefl grown in non-defi- cient soils. This finding suggests that increased occur- rence of secondary compounds in white-sand vege- tation may simply represent an acclimational response (St. John and Anderson, in press).

The oligotrophic nature of white-sand soils has pronounced effects on nutrient cycling. Thick ac- cumulations of slowly decomposing litter provide an exclusive substrate for development of a root mat, which hardly penetrates the mineral soil and can be pulled off the surface like a rug. Release of a biotic compound such as ethylene or an auxin in decom- posing litter has been shown to induce root growth (St. John and Machado 1978). The root mat con- sequently attains a pronounced development in trop- ical white-sand vegetation. Klinge and Herrera (1977) found that roots may comprise over 60 per- cent of the total biomass of an Amazon caatinga; in oxisol forests, the figure is typically 20 percent. The root mat in white-sand forests and woodlands is usually 10-30 cm thick, significantly (P -- 0.01) thicker than its counterpart on oxisols, according to data from Stark and Jordan (1978). The thickness of the root mat, combined with an exceptional abun- dance of vesicular-arbuscular (T. V. St. John, pers. comm.) and ectomycorrhizae (Singer 1979), results in increased absorbing surfaces which maximize nu- trient uptake. Radio-tracer experiments by Stark and Jordan (1978) in white-sand forest in Venezuela demonstrated the effectiveness of the root mat in pre- venting nutrient losses. Hence, although mineraliza- tion of litter is slow, the high area of absorbing sur- faces in the root mat increases nutrient uptake while minimizing losses through the porous soil. By pro- viding a circuit for increased nutrient uptake and leak-free cycling, the thick root mat appears to be a key factor in enabling forests of moderately high biomass to develop on soils which are virtually sterile.

WATER DEFICIENCIES Proponents of water deficiencies as the principal limiting factor would argue that the pronounced scleromorphism of the caatinga represents a xero- morphic response to physiological drought. Reduced phases of white-sand vegetation are characterized by comparatively high surface temperatures and low relative humidity during the day (Schulz 1960), which may contribute to soil-moisture deficits through increased evaporation. Excessive drainage through the porous sands may periodically result in a drop of the water table below the zone of root penetration (cf. Sioli 1960, Sombroek 1966). The unstable nature of the water table in caatinga has been noted in the upper Rio Negro region, where rainfall is exceptionally high throughout the year: a few days without rain can cause dramatic fluc- tuations (E. Medina, pers. comm.). In this same region, Ferri (1960) found a lack of water stress in the leaves of caatinga plants and concluded that water deficiencies are not a limiting factor in these habitats. However, his conclusion was based on a few days of measurements carried out during the rainy season; short periods of water stress during the drier months may be sufficient to exert a controlling in- fluence on the vegetation (E. Medina, pers. comm.).

Working in homologous vegetation in Sarawak, Brunig (1970) found a number of characteristics which appear to have adaptive value by enhancing the cooling of leaves, thereby minimizing transpira- tion losses necessary to keep leaves cool. These char- acteristics include: small leaf sizes, which increase convectional cooling; shiny leaf surfaces, which re- duce heat load by increasing the reflectance of radia- tion; steep inclination of leaves, which reduces heat load from incident radiation; and lack of canopy roughness, which reduces heat load by decreasing the interception of incident radiation. In comparison with other vegetation types, the Amazon caatinga is also characterized by relatively small leaves, shiny leaf surfaces, steep inclination of leaves, and lack of canopy roughness: these features may have similar

204 Anderson

adaptive value in minimizing transpiration losses (cf. Medina et al. 1978).

Present lack of experimental evidence makes it impossible to determine the relative importance of nutrient and water deficiencies in the Amazon caa- tinga. Although their relative importance remains uncertain, these deficiencies appear to be consider- ably greater in the caatinga than in other vegeta- tion types of the Amazon region. Under such dis- tinctive selective forces, it is no surprise that the caatinga is unique not only in structural and function- al terms, but in its biological composition as well.

BIOGEOGRAPHY The unusual flora of the Amazon caatinga has long been noted by botanists (e.g., Spruce 1908). Studies of this vegetation in different regions (cf. Egler 1960, Rodrigues 1961, Pires and Rodrigues 1964, Anderson et al. 1975) provide some general insights concerning their floristic co'mposition. Families dom- inant in caatinga forest and woodland include Le- guminosae, Euphorbiaceae, Sapotaceae, Guttiferae, Rubiaceae, and Myristicaceae. Important, high-diver- sity components of forests on oxisols-such as Mor- aceae, Lecythidaceae, and Loganiaceae (Strychnos) -are absent or extremely rare in caatinga, while palms may be locally abundant but are invariably low in diversity. Examples of dominant families in structurally reduced phases are Melastomataceae, Rubiaceae, Chrysobalanaceae, Myrtaceae, Malphigia- ceae, and Vochysiaceae. Terrestrial herbaceous families include Cyperaceae, Gramineae, Eriocaulaceae, Xyri- daceae, and Schizaeaceae; Cyperaceae are more abun- dant than Gramineae, in contrast to most Amazonian campos (except campo rupestre). Lichens (Cladonia spp., Parmelia spp.) are a common feature in struc- turally reduced caatinga. Families of the caatinga which rarely occur in other lowland vegetation types of Amazonia include Cyrillaceae, Ericaceae, and Lis- socarpaceae.

Ducke and Black (1953) provided an exhaus- tive list of species which occur exclusively in caatinga and are endemic to specific regions, such as the upper Rio Negro. Due to a high incidence of endemic sp-ecies, the floristic composition of this vegetation varies considerably from one region to another, as indicated by comparisons between distant caatinga sites (Anderson 1978). Yet a number of species ex- clusive to caatinga have wide geographic ranges and hence serve as indicators of this vegetation over large areas of Amazonia. Examples include: Cephalo- stemon gracile (Rapateaceae), an herb common on humid sites in eastern Amazonia: Gaylussacia anma-

zonica (Ericaceae), a low shrub widespread through- out eastern and central Amazonia; Glycoxylon ino- phyllum (Sapotaceae) and Pagamea duckei (Ru- biaceae), shrubs or small trees often dominant on caatinga sites in central Amazonia; Mauritia carana (Palmae), a large palm characteristic of hydro- morphic white sands throughout the Rio Negro Basin; Lissocarpa benthami (Lissocarpaceae), Hevea pauciflora var. coriacea (Euphorbiaceae), and Laden- bergia aomazonensis (Rubiaceae), tree species typical of caatinga sites in western Amazonia (Ducke and Black 1953).

Many plant species in the caatinga are restricted to a limited range of habitats. Anderson (1978) found that the majority (54.5 %) of vascular ter- restrial species in a central Amazon caatinga occur exclusively in this vegetation type. A moderate pro- portion of the species also occurs in upland forests on oxisols (23.6%) and igapo (20.0%), which are generally oligotrophic habitats; relatively few (2.6% ) are known from the nutrient-rich varzea, a vegeta- tion type located adjacent to white-water rivers. While the habitat range of terrestrial species tends to be narrow, epiphytes such as orchids generally occur in a wide range of habitats (P. I. S. Braga, pers. comm.), probably because they are not limited directly by soil conditions.

Although the caatinga has been less studied by zoologists, a number of animals are presently known to be restricted to these habitats. A species of titi monkey endemic to western Amazonia, Callicebus torquatus, is found almost exclusively on white-sand sites (Kinzey and Gentry 1979). Two bird species, Neopelma chrysocephalum and Xenopipo atronitens (Pipridae), are limited to white-sand vegetation in Brazil and Venezuela (D. Oren, pers. comm.). In central Amazonia, Brown and Benson (1977) re- ported two endemic subspecies of the passion flower butterfly, Heliconius hermathena (Heliconiini), which are restricted to caatinga.

Plant communities in the caatinga have a com- paratively reduced species richness (fig. 7), with a pronounced tendency toward dominance by one of a few species (Takeuchi 1960, Anderson et al. 1975). Animal communities show a similar trend (cf. Jan- zen 1974). In caatinga woodland and scrub in central Amazonia, avifaunal diversity as measured by the information index value (in ln) was approximately 2.3, compared to values ranging from 3.5 to nearly 4.0 in adjacent forest on oxisol; the dominant species on the caatinga site comprised almost 40 percent of the individuals sampled (D. Oren, pers. comm.). Low diversity is probably in part a function of in- sularity, as well as of extreme environmental condi-

White-Sand Vegetation of Amazonia 205

200 0J High forest

EN Igapo forest

g Caatinga woodland

150

w 0 w a-

o 100

w

z

50-

0

FIGURE 7. Number of species per hectare in Amazonian vegetation types. DBH - 10 cm, except for site correspond- ing to leftmost bar, where DBH - 15 cm. Data for left- most bar from Prance et al. (1976), for middle two bars from Black et al. (1950), and for rightmost bar from An- derson (unpublished data).

tions. In central Amazonia, where the caatinga has a scattered occurrence (fig. 8), species richness is low compared to the upper Rio Negro region, where this vegetation covers immense areas (Ducke and Black 1953). Biogeographers (MacArthur and Wil- son 1967) have shown that area and degree of isola- tion account for most of the variation in species num- bers on islands: a similar relationship appears to ex- ist in the caatinga.

Area and degree of isolation likewise appear to influence the biological composition of these habitats. On scattered caatinga sites occupying small areas ( < 100 ha), most of the avifauna is comprised of generalist species with a wide geographic distribu- tion; extensive caatinga sites ( > 100 ha) contain numerous habitat specialists with restricted geo- graphic ranges (Oren, in press). As is the case on islands, small and/or isolated caatinga sites appear to select for plant species with long-distance disper- sal mechanisms (cf. Macedo and Prance 1977). In a caatinga in central Amazonia, Macedo (1977) found that 75.7 percent of the species investigated have the potential for long-distance dispersal, including 59.5 percent which are bird-dispersed. Floristic compari- sons by Anderson (1978) of three caatinga sites in central Amazonia revealed a high overlap of genera and species, indicating the effectiveness of dispersal mechanisms over long distances and formidable bar- riers such as the Rio Negro.

2'30

KM-50

~ZF- I

KM-40

240

0 1 2 3 LEGEND SCALE I 125 000

KM 0 CAATINGA (Scrub or savanna pkem)

FIGURE 8. Occurrence of caatinga savanna and scrub in an area of central Amazonia.

The insular nature of the caatinga has further sig- nificance in rapidly developing areas such as Manaus, where this vegetation is systematically burned (fig. 9). At burned sites, many species and certain char- acteristic elements of the original flora-such as lichens, bryophytes, bromeliads, and orchids- disap- pear completely and are replaced by a monotonous secondary growth known as capoeira (Ducke and Black 1953, Anderson 1978). As caatinga vegeta- tion is rapidly destroyed over wider areas, recoloniza- tion by the original biota becomes increasingly un- likely, due to lower probabilities of dispersal from undisturbed sites. The high incidence of geographic endemics suggests that numerous species may face extinction.

The destruction of caatinga vegetation in areas of Amazonia designated for development raises the question of whether these habitats will recuperate in time, and how much time will be required. While there are no definitive answers, clues may be found by examining the origin and successional status of non-forest vegetation on white sand.

206 Anderson

ORIGIN AND SUCCESSION The incidence of non-forest vegetation in the Ama- zon region is surprisingly high. Pires (1974) esti- mated that over 226,000 square kilometers of Bra- zilian Amazonia are covered by savanna or scrub. The origin of such vegetation is uncertain despite much speculation in the literature (cf. Hills 1969, Sarmiento and Monasterio 1975). Although white- sand savanna and scrub are probably produced by a variety of factors, two appear to be of overriding significance: position and stability of the water table, and fire.

White-sand soils can be characterized as either hydromorphic or non-hydromorphic, depending on whether the water table remains within a meter of the surface over most of the year. Hydromorphic and non-hydromorphic soils support distinct floristic as- sociations, even where they occur adjacent to one another (Heyligers 1963, Anderson 1978). On hy- dromorphic so,ils, the position of the water table ap- pears to be a crucial factor in determining vegeta- tion structure. In his exhaustive study of white-sand vegetation in Surinam, Heyligers (1963) found that savanna is situated in areas where the water table remains constantly at the surface, while less-reduced scrub occurs on slightly more-elevated sites, where the water table is deeper and root systems apparently attain greater development. Stability of the water table may also determine vegetation structure in these habitats. In white-sand savanna and scrub ("bana") in Venezuela, Herrera (1977) found that the water table drops from the surface to over 100 cm depth during periods of low rainfall; simultaneous measure- ments under white-sand forest ("yaguacanan") showed negligible variation. A fluctuating water table may exert a detrimental effect on the vegeta- tion through decreased water availability and/or increased leaching within the root zone.

In contrast, Heyligers (1963) found that on non- hydromorphic sites, vegetation structure bears little relation to po,sition of water table or its degree of fluctuation. A more potent factor in determining vegetation structure on these sites is fire. In central Amazonia, the presence of charcoal and ceramic shards at a number of reduced caatinga sites led Prance and Schubart (1977) to conclude that the present structure of the vegetation is due to fire. In Malaysia, white-sand savanna and scrub ("padang") are produced by fire (Whitmore 1975), and similar habitats in Surinam are still frequently burned by Indians during the dry season (Heyligers 1963, Stark 1970). Nor is vegetation on hydromorphic sands im- mune: Indians in Surinam occasionally burn such

sites to facilitate capture of tortoises (Heyligers 1963).

In northern South America, white-sand forests or woodlands destroyed by fire are initially recolon- ized by typical secondary elements such as Cecropia, Vismia, Byrsonima, and Pteridium (Ducke and Black 1953, Heyligers 1963). Although initial colonization is rapid, subsequent growth is considerably slower than on oxisols. Ten years after. a burn, Heyligers (1963) observed that most of the initial colonizers had died, and that scattered white-sand species such as Dirnorphandra conjugata were gradually becom- ing reestablished. On a site burned 30 years previous- ly, he found a scrub dominated by Dirnorphandra; 50 years after a burn, a white-sand woodland had developed. In the continued absence of fire, Hey- ligers concluded that forest would eventually develop on these sites.

Although regeneration does occur when white- sand vegetation is subjected to fire, succession appears to be comparatively prolonged, as indicated by the slow growth of secondary colonizers following a burn. Succession may be even more prolonged if a burn is widespread. As was mentioned above, fires which destroy white-sand vegetation over large areas reduce the probability of dispersal from undisturbed sites and thus may prolong succession. Intensity of the burn is an additional factor. Heyligers (1963) found that following a light to moderate burn, a number of fire-resistant species such as Dimorphandra serve as "regeneration centers," providing a focal point for establishment of colonizing species; intensive burns destroy these centers and thus appear to, prolong suc- cession. Repeated burns have a similar effect. Based on observations in Brazil and Surinam, Stark (1970) suggested that four or five burns of what was initially white-sand forest produce sites which are devoid of vegetation and severely depleted in nutrients. The long-term effects of what were apparently widespread, intensive, or repeated burns can be seen today. Radio- carbon dating of charcoal remains found on bare- sand sites in central Amazonia indicates that the vegetation was burned 800-1100 years ago (Prance and Schubart 1977). Lack of regeneration during this time implies that succession has been effectively arrested on these sites.

The above considerations indicate that white- sand vegetation has a low resilience to externally imposed stresses such as fire. Above-ground biomass and litter store an exceptionally large proportion of total nutrients in these ecosystems (Herrera et al. 1978). Burning of the vegetation is likely to cause considerable losses of released nutrients, due to the low storage capacity of the soil. On non-hydromorphic

White-Sand Vegetation of Amazonia 207

sites, the thick surface root mat is relatively suscepti- ble to destruction by fire (fig. 9), leading to further nutrient losses. Failure of rapid reestablishment fol- lowing widespread, intensive, or repeated burns may thus result in a depletion of soil nutrients below levels necessary for plant growth. Attempts to re- establish productive forests on such sites are likely to fail, as indicated by trial planting on degraded white-sand soils in Malaysia (Mitchell 1963). With- out a viable nutrient storage component, succession is effectively arrested and a biological desert is pro- duced.

THE VALUE OF THE AMAZON CAATINGA A low resilience to external stresses, in addition to un- favorable soil conditions, indicates that the Amazon caatinga has little potential for agriculture. Indeed, similar vegetation in Malaysia is referred to as "ker- angas," meaning "forest which, when cleared, will not grow rice" (Whitmore 1975:132). At present, their lack of agricultural value appears to justify the use of caatinga sites as a source of sand for road- building and construction projects throughout Ama- zonia (fig. 10). The value of the caatinga in its in- tact state has not been considered when calculating the cost of extracting its sand.

An accurate knowledge of the extent and dis- tribution of these habitats in Amazonia is a pre- requisite for determining viable alternatives for land use. The Amazon caatinga has never been accurately mapped, although satellite images now provide the means to do so. The high reflectance of the vegeta- tion and soil, as well as the relative uniformity of the canopy, produces a distinctively pale tone and fine texture on aerial photographs (Whitmore 1975), making these habitats amenable to mapping. Such a map would be useful for two reasons. First, by in- dicating sites of high endemism, it would provide a criterion for the location and delimitation of bio- logical reserves in Amazonia. Second, by indicating sites of low agricultural potential, it would provide a criterion for the location and delimitation of focal points for development within the region. This last criterion could have been usefully applied prior to the establishment of an extensive agricultural zone on the Manaus-Boa Vista highway (BR-174 km 35- 105): a high incidence of caatinga vegetation faith- fully attests to the overall impoverished state of soils in this zone.

The value of white-sand habitats ultimately de- pends on what alternative benefits they can provide. Carefully managed forests homologous to the Amazon

9 _

: 7 ~~~~~~~~~~~~~~. ~~~~~~. .. ... ..

10

FIGURES 9-10. Figure 9. Burned caatinga, with surface root mat and litter reduced to ash (BR-174 km ca. 25). Figure 10. Caatinga site mined for sand, with cement factory in background (AM-070 km ca. 75).

caatinga are used as a continuing source of timber in Malaysia (Whitmore 1975). The recreational value of white-sand vegetation within the Bako National Park in Sarawak is increasingly appreciated by urban visitors (Whitmore 1975). The relative simplicity of the caatinga makes it ideal for educational and research purposes, as indicated by the intensive use of one such site in central Amazonia (Reserva Biol6gica do INPA/SUFRAMAI, BR-174 km 45). Finally, the unusual flora and fauna in these habitats are of value not only for aesthetic reasons. With the implementation of man-made ecosystems throughout the humid tropics, organisms adapted to high-stress environments such as the Amazon caatinga are likely to become increasingly important as a genetic re- source.

ACKNOWLEDGEMENTS Many of the ideas presented in this paper have come from discussions with W. W. Benson, P. I. S. Braga, L. Coelho, S. B. Hecht, C. F. Jordan, P. L. Lisb6a, E. P. Lleras, E. Medina, D. C. Oren, J. M. Pires, H. L. Popenoe, G. T. Prance, G. Ranzani, J. Revilla, W. A. Rodrigues, H. 0. R.

208 Anderson

Schubart, R. Singer, H. Sioli, and T. V. St. John. I especially wish to thank G. Eiten, J. J. Ewel, D. H. Janzen, W. G. Kinzey, and G. T. Prance for reviewing the manuscript. R. Sylvester Bradley and J. Dobereiner kindly assisted in ob- taining some of the soil analyses of samples from the white-

sand site presented in table 2; most of the data for the oxisol site are compliments of T. V. St. John. Support while preparing the manuscript was provided by fellowships from the Graduate Council of the University of Florida and the Organization of American States.

LITERATURE CITED ANDERSON, A. B. 1978. Aspectos floristicos e fitogeogr'aficos de campinas e campinaranas na Amazonia Central. Tese de

M. Sc., Instituto Nacional de Pesquisas da Amazonia e Universidade do Amazonas, Manaus. G. T. PRANCE, AND B. W. P. DE ALBUQUERQUE. 1975. A vegetacao lenhosa da campina da Reserva Biologica

INPA-SUFRAMA (Manaus-Caracarai, Km. 62). Acta Amazonica 5: 225-246. ARENS, K. 1963. The dwarfed plants of the "cerrado" fields as flora adapted to mineral deficiences in the soil. In, M. G.

Ferri. (Ed.). Simp6sio sobre o cerrado, pp. 285-303. Edgar Blucher, SAo Paulo. AUBREVILLE, A. 1961. Etude ecologique des principales formations vegetales du Bresil. Centre Tech. Foret Trop., Nogent-

Sur-Marne, France. 268 pp. BEARD, J. S. 1955. The classification of tropical American vegetation types. Ecology 36: 89-99. BLACK, G. A., T. DOBZHANSKY, AND C. PAVAN. 1950. Some attempts to estimate species diversity of trees in Amazon-

ian forests. Bot. Gaz. 111: 413-425. BRAGA, M. M. N. 1977. Anatomia foliar de Bromeliaceae da campina. Acta Amazonica 7(3: Suplemento): 1-74. BRAGA, P. I. S. 1977. Aspectos biologicos das Orchidaceae de uma campina da Amazonia Central. Acta Amazonica 7(2:

Suplemento): 1-89. BROWN, K. S., AND W. W. BENSON. 1977. Evolution in modern Amazonian non-forest islands: Heliconius hermathena.

Biotropica 9: 95-117. BRUNIG, E. F. 1970. Stand structure, physiognomy and environmental factors in some lowland forests in Sarawak. Trop.

Ecol. 11: 26-43. DAVIES, R. I., C. B. COULSON, AND D. A. LEWIS. 1964. Polyphenols in plant, humus, and soil. IV. Factors leading to in-

crease in biosynthesis of polyphenol in leaves and their relationship to mull and mor formation. J. Soil Sci. 15: 310-318.

DUCKE, A. 1922. Plantes nouvelles ou peu connues de la region amazonienne. Archos. Jard. bot., Rio de J. 3: 2-269. , AND G. A. BLACK. 1953. Phytogeographic notes on the Brazilian Amazon. Anais Acad. bras. Cienc. 25: 1-46.

EGLER, W. A. 1960. Contribusoes ao conhecimento dos campos da Amazonia. I - Os campos do Ariramba. Bolm. Mus. para. 'Emilio Goeldi' (Bot.) 4: 1-37.

EITEN, G. 1978. Delimitation of the cerrado concept. Vegetatio 36: 169-178. FANSHAWE, D. 1952. The vegetation of British Guiana: a preliminary review. Inst. Pap. Commonw. For. Inst. No. 29, Ox-

ford. FERRI, M. G. 1960. Contribution to the knowledge of the Rio Negro "Caatinga" (Amazon). Bull. Res. Coun. Israel

(Bot.), 8 D, 3-4: 195-208. GRANVILLE, J. J. DE (in press). Rain forest and xeric flora refuges in French Guiana during the late Pleistocene and Hol-

ocene. In, G. T. Prance. (Ed.). Proc. Assoc. Trop. Biol. Symp., Caracas, Columbia Univ. Press, New York. GOODLAND, R. 1971. Oligotrofismo e aluminio no Cerrado. In, M. G. Ferri. (Ed.). III Simp6sio sobre o cerrado, pp. 45-

60. University of Sao Paulo, SAo Paulo. HERRERA, R. 1977. Soil and terrain conditions in the International Amazon Project at San Carlos de Rio Negro, Vene-

zuela: correlation with vegetation types. In, E. F. Brunig. (Ed.). Transactions of International MAB-IUFRO work- shop on tropical rainforest ecosystem research, pp. 182-188. Univ. Hamburg, Hamburg-Reinbek. , C. F. JORDAN, H. KLINGE, AND E. MEDINA. 1978. Amazon ecosystems. Their structure and functioning with par- ticular emphasis on nutrients. Interciencia 3: 223-232.

HIEYLIGERS, P. C. 1963. Vegetation and soil of a white-sand savanna in Suriname. N. V. Noord-Hollandsche Uitgevers Maatschappij, Amsterdam. 148 pp.

HILLS, T. L. 1969. The savanna landscapes of the Amazon Basin. McGill Univ. Savanna Res. Ser. No. 14, Montreal. 38 pp. JANZEN, D. H. 1974. Tropical blackwater rivers, animals, and mast fruiting by the Dipterocarpaceae. Biotropica 6: 69-

103. KINZEY, W. G., AND A. H. GENTRY. 1979. Habitat utilization in two species of Callicebus. In, R. W. Sussman. (Ed.).

Primate ecology: problem-oriented field studies, pp. 89-100. John Wiley and Sons, N.Y. KLINGE, H. 1967. Podzol soils: a source of blackwater rivers in Amazonia. In, H. Lent. (Ed.). Atas do simp6sio sobre a

biota amazonica, Vol. 3, pp. 117-125. Conselho Nacional de Pesquisas da Amazonia, Rio de Janeiro. I AND R. HERRERA. 1977. Composite root mass in tropaquods under Amazon caatinga stands in southern Vene- zuela. IV Symp. Trop. Ecol., Panama. , E. MEDINA, AND R. HERRERA. 1978. Studies on the ecology of the Amazon caatinga forest in Southern Vene- zuela: 1. General features. Acta cient. Venez. 28: 270-276.

LISBOA, P. L. 1975. Observaco'es gerais e revisaio bibliografica sobre as campinas amazonicas de areia branca. Acta Ama- zonica 5: 211-223.

LISBOA, R. C. L. 1976. Brioecologia de uma campina amaz6nica. Acta Amazonica 6: 171-191. LLERAS, E., AND J. H. KIRKBRIDE, JR. 1978. Alguns aspectos da vegetacao da Serra do Cachimbo. Acta Amazonica 8: 51-

65.

White-Sand Vegetation of Amazonia 209

MACARTHUR, R. H., AND E. 0. WILSON. 1967. The theory of island biogeography. Princeton Univ. Press, Princeton. 203 pp-

MACEDO, M. 1977. Dispersio de plantas lenhosas de uma campina amazonica. Acta Amazonica 7 (1: Suplemento): 1-69. , AND G. T. PRANCE. 1977. The dispersal of plants in Amazonian white sand campinas. Brittonia 30: 203-215.

MCKEY, D., P. G. WATERMAN, C. N. MBI, J. S. GARTLAN, AND T. T. STRUHSAKER. 1978. Phenolic content of vegetation in two African rain forests: ecological implications. Science, N.Y. 202: 61-64.

MEDINA, E., M. SOBRADO, AND R. HERRERA. 1978. Significance of leaf orientation for leaf temperature in an Amazonian sclerophyll vegetation. Rad. and Environ. Biophys. 15: 131-140.

MITCHELL, B. A. 1963. Forestry and tanah beris. Malay. Forester 26: 160-170. OREN, D. C. (in press). Testing the refugium model for South America: a hypothesis to evaluate discrepancies in refugia

number across taxa. Iz, G. T. Prance. (Ed.). Proc. V Assoc. Trop. Biol. Symp., Caracas, Columbia Univ. Press, New York.

PIRES, J. M. 1974. Tipos de vegeta~Ao da Amazonia. Mus. Par. Emilio Goeldi, Publicac6es avuls. 20: 179-202. , AND J. S. RODRIGUES. 1964. Sobre a flora das caatingas do Rio Negro. Anais do XIII Congresso da Soc. Bot. do Brasil: 242-262.

PRANCE, G. T., AND H. 0. R. SCHUBART. 1977. Nota preliminar sobre a origem das campinas abertas de areia branca do baixo Rio Negro. Acta Amazonica 7: 567-570.

- W. A. RODRIGUES, AND M. F. DA SILVA. 1976. Inventario florestal de um hectare de mata de terra firme km 30 da Estrada Manaus-Itacoatiara. Acta Amazonica 6: 9-35.

PROJETo RADAMBRASIL. 1976. Folha Na. 19 Pico da Neblina. Levantamento de Recursos Naturais Vol. 11: 1-374. REVILLA, J. 1978. Comunidades vegetales de Mishima, Rio Nanay, Loreto, Peru. Tese de Grad., Univ. Nacional Mayor de

San Marcos, Lima. RICHARDS, P. W. 1952. The tropical rain forest. Cambridge Univ. Press, Cambridge. 450 pp. RODRIGUES, W. A. 1961. Aspectos fitossociol6gicos das caatingas do Rio Negro. Bolm. Mus. para. 'Emilio Goeldi' (Bot.)

15: 1-41. SAMPAIO, A. J. DE. 1945. Fitogeografia do Brasil. Editoria Nacional, Sao Paulo. 372 pp. SARMIENTO, G., AND M. MONASTERIO. 1975. A critical consideration of the environmental conditions associated with the

occurrence of savanna ecosystems in tropical America. Iz, F. B. Golley and E. Medina. (Eds.). Tropical ecological systems: trends in terrestrial and aquatic research, pp. 223-250. Springer-Verlag, N.Y.

SCHULZ, J. P. 1960. Ecological studies of rain forest in northern Suriname. N. V. Noord-Hollandsche Uitgevers Maats- chappij, Amsterdam. 267 pp.

SINGER, R. 1979. Litter decomposition and ectomycorrhiza in Amazonian forests. 1. A comparison of litter decomposing and ectomycorrhizal Basidomycetes in latosol-terra-firme rain forest and white podzol campinarana. Acta Amazon- ica 9: 25-41.

SIOLI, H. 1960. Estratificaado radicular numa caatinga baixa do alto Rio Negro. Bolm. Mus. para. 'Emilio Goeldi' (Nov. Ser.) 10: 1-9.

- AND H. KLINGE. 1961. Uber Gewasser und Boden des Brasilianischen Amazonasgebietes. Die Erde 92: 205-219. SOARES, L. DE C. 1953. Limites meridionais e orientais da area e ocorrencia da floresta amazonica em territorio brasileiro.

Revta. bras. Geogr. 15: 3-95. SOIL SURVEY STAFF. 1975. Soil taxonomy. A basic system for making and interpreting soil surveys. U.S. Dep. Agric.,

Washington. 754 pp. SOMBROEK, W. G. 1966. Amazon soils, a reconnaissance of the soils of the Brazilian Amazon region. Centre for Agricul-

tural Publications and Documentation, Wageningen. 192 pp. SPRUCE, R. 1908. Notes of a botanist on the Amazon and Andes, Vol. 1, A. R. Wallace. (Ed.). MacMillan and Co., Lon-

don. 518 pp. STARK, N. 1970. The nutrient content of plants and soils from Brazil and Surinam. Biotropica 2: 51-60.

1971. Nutrient cycling I: Nutrient distribution in some Amazonian soils. Trop. Ecol. 12: 24-50. AND C. F. JORDAN. 1978. Nutrient retention by the root mat of an Amazonian rain forest. Ecology 59: 434-437.

ST. JOHN, T. V., AND A. D. MACHADO. 1978. Evidencia da acao de micro-organismos na ramificacao de raizes. Acta Ama- zonica 8: 9-11. , AND A. B. ANDERSON. (In Press.) A re-examination of plant phenolics as a source of blackwater rivers. Trop. Ecol.

TAKEUCHI, M. 1960. A estrutura da vegeta~Ao na Amazonia. III. A mata da campina na regiao do Rio Negro. Bolm. Mus. para. 'Emilio Goeldi' (Nov. Ser.) Bot. 8:1-13.

VIERA, L. S., AND J. P. OLIVEIRA FILHO. 1962. As caatingas do Rio Negro. Bolm. Tec. Inst. agron. N. 42: 7-32. WHITMORE, T. C. 1975. Tropical rain forests of the Far East. Clarendon Press, Oxford. 282 pp.

210 Anderson