Wood Characteristics of Amazon Forest Types

IAWA Journal, Vol. 21 (3), 2000: 277–292 WOOD CHARACTERISTICS OF AMAZON FOREST TYPES by D.W. Woodcock1, G. Dos Santos 2 & C. Reynel3 SUMMARY The Tambopata region of the southern Peruvian Amazon supports a high diversity of both woody plants and forest types. Woods collected from low riverside vegetation, loodplain forest, clay-soil forest on an upper terrace, sandy-soil forest, and swamp forest provide an opportunity to test for signiicant differences in quantitative anatomical characters among forest types. Vessel-element length in loodplain-forest trees is signiicantly greater than in the other forest types. Speciic gravity is lower in the two early-successional associations (low riverine forest and mature loodplain forest). Vessel diameter and density do not show signiicant differences among forest types and may be responding to overall climate controls. These two characters, however, show a pattern of variation within a transect extending back from the river along a gradient of increasing substrate and forest age; in addition, sites characterized by frequent looding or presence of standing water lack vessels in the wider-diameter classes. The six characters analyzed show distributions that are, with the exception of wood speciic gravity, signiicantly nonnormally distributed, a consideration that may be important in representing characteristics of assemblages of taxa. The degree of variability seen in some of the quantitative characters shows the importance of either basing analysis on adequate sample sizes or identifying robust indicators that can be used with small samples. Key words: Quantitative wood characters, wood ecology, Amazon forest woods, fossil wood. INTRODUCTION The Tambopata region, in the southern Peruvian Amazon (13° S), is known for its biotic diversity (Conservation International, 1994) and the contributions to tropical forest botany and ecology that have been made by researchers working there (Wilson 1987; Erwin 1988; Gentry 1988; Phillips & Gentry 1994; Phillips et al. 1994a,b; Woodcock 1996; Phillips 1999; Reynel & Gentry, in press). Both the forests and the hydro1) Department of Geography, University of Hawaii, 445 Social Science, Honolulu HI 96822, U.S.A. 2) P. O. Box 5354, Pleasanton, CA 94566, U.S.A. 3) Faculdad de Ciencias Forestales, Universidad Nacional Agraria La Molina, Apdo. 456, La Molina, Lima, Peru. 278 IAWA Journal, Vol. 21 (3), 2000 Fig. 1. Map of the area around Explorerʼs Inn showing the plots mentioned in the text: 1, riverine forest; 2, mature loodplain forest; 3, clay-soil forest on upper terrace; 4, sandy-soil forest; 5, swamp forest on upper terrace. Dashed line indicates boundary between richer alluvial and sandy substrates (areas to the south and east). Elevations indicated in meters. Prepared from the 1993 1 : 100,000 national series topographic map based on 1985 aerial photographs. logic system are very little disturbed. This paper presents results of a study of the wood characteristics of the vegetation at ive sites representing distinct forest types (see Appendix). Woods were sampled in the following locations (Fig. 1): 1) An area of low, early-successional vegetation growing alongside the Rio La Torre on a sandbar deposit. Diversity is low, with dominant taxa Cecropia, Salix, and other trees typical of disturbed riverine habitats. [mean dbh ~10 cm, 8 species] 2) Mature loodplain forest growing along the La Torre on alluvial soils within a meander bend. The smooth-barked Capirona is a distinctive element of these gallery forests. There is standing water at some times of the year and severe looding Woodcock, Dos Santos & Reynel — Wood characteristics of Amazon forest 279 occurs at a recurrence interval estimated at 10–12 years. A plot for long-term ecological monitoring is near the transect (Philips & Gentry 1994; Phillips 1999). [mean dbh ~ 22 cm, 22 species] 3) Clay-soil forest on an ʻupper terraceʼ. This surface is actually an ancient loodplain of the Tambopata with an age estimated at 40,000 or more years (Salo & Kalliola 1990). Pseudolmedia laevis and P. macrophylla are dominant species at the site. [mean dbh ~ 23 cm, 21 species] 4) Sandy-soil forest. Away from the main river system, blackwater rivers drain substrates distinctly sandy in nature. The sandy-soil forest sampled here was growing alongside the Aguasnegras River approximately 6 km from the conluence of the Tambopata and the La Torre. No species were particularly prevalent. [mean dbh ~ 24 cm, 22 species] 5) Swamp forest on the upper terrace. This loodplain feature may be a meander lake or an ancient channel. The palm Mauritia lexuosa is common in these swampy areas and diversity of tree species is low. The trees sampled are adjacent to a permanent monitoring plot (Philips & Gentry 1994; Phillips 1999). [mean dbh ~ 29 cm, 8 species] The analysis compares six quantitative wood characteristics (vessel diameter and density, vessel-element length, iber length, ray density, and speciic gravity) at ive sites and along one environmental gradient (site 2 transect). The question addressed is whether these wood characters differ signiicantly from one forest type to another. This information helps to distinguish characters providing information about local environments from those that may be responding to overall climate controls. The results presented add to an increasing body of literature on the variation of wood structure and properties in different environments and climate regimes that will lead to better understanding of wood ecology and more well-founded interpretations of fossil wood (Carlquist 1975, 1988; Chudoff 1976; Barajas-Morales 1985, 1987; Baas & Schweingruber 1987; Wheeler & Baas 1991, 1993; Lindorf 1994; Woodcock & Ignas 1994; Wiemann et al. 1998; Williamson 1984). The unit of analysis here is the assemblage of woody plants considered by species (or wood type) since fossil plant assemblages must generally be studied in terms of species (or type) rather than individuals represented. Most of the taxa studied can be distinguished on the basis of their wood (although told apart with dificulty, the Pseudolmedia species are here counted as distinct taxa). There are various ways of representing the wood characters. Qualitative characters can be expressed as percent occurrence (e.g., percent of taxa with homocellular rays). Quantitative characters can be averaged over an assemblage or, as is often done, expressed as percent occurrence in different categories (e.g., percent of taxa with vessels < .050 mm). There are also cases in which presence or absence of a particular characteristic may be signiicant apart from its prevalence; these characters (such as presence of storied rays) are of special interest since they may permit interpretations on the basis of one or few taxa. Quantitative wood characters present special problems because they often exhibit a high degree of variability, even among species at one location. In addition, although 280 IAWA Journal, Vol. 21 (3), 2000 often treated by means of categories, the best way to construct these categories is not necessarily evident. Thus special attention is paid to the categories in general use and derivation of categories such that they are most useful for environmental interpretations. Several studies have analyzed spatial variation in wood characters and quantitative relationships to climate (Woodcock 1994; Woodcock & Ignas 1994; Wiemann et al. 1998). Analysis of the response of wood structure to climate should ideally be based on wood loras of no less than 25 species collected from locations around the world. Dificulties in assembling the amount of material required has led to other approaches being taken. One study looked at the spatial variation in wood characters of the eastern North American tree lora by putting together range data with information from a global database for wood characters (Woodcock & Ignas 1994). Another matched species lists from 37 loras with wood characters of the corresponding species or genus and presented a multivariate representation of the wood : climate relationship (Wiemann et al. 1998). Both studies relied on the OPCN database (Wheeler et al. 1986), a global database used mainly for wood identiication; this database includes only a portion of the characters used to describe woods, presents quantitative characters in terms of categories that may not be derived in an optimal way and are in some cases overlapping, and involves other complications such as lack of information on vessel diameter and density for the ring-porous taxa and inconsistencies in measurement practices (see Wheeler 1986). MATERIALS AND METHODS Climate The area is at 12° 50ʼ S and 69° 18ʼ W. Precipitation is approximately 2400 mm annually, with the months May–October distinctly drier than the remainder of the year (three consecutive months < 100 mm; Phillips et al. 1994b). Precipitation amounts decrease, and length of the dry season increases, eastward toward the border with Bolivia. Mean annual temperature is 22–24 °C. In the months June–August, incursions of extratropical air masses can produce temperatures as low as 10 °C. Sampling Woody vegetation was sampled along 60-m line transects. Sampling extended on either side of the transect until a suficient sample size was obtained. In two cases, the transects had speciic orientations. The riverside transect (transect 1) ran parallel to the river in the woody vegetation just adjacent to the river on a sandbar deposit. The loodplain forest transect (transect 2) extended back from the river across the ridgeand-swale topography within a meander bend and was thus aligned with a gradient of increasing substrate and forest age. For the lower-diversity riverside and swamp forests, sample sizes were small but included all the dominant tree species. In all but site 1, minimum diameter of the trees sampled was 10 cm. Palm species, which are prominent in all but site 1, were not sampled. This omission is due to the dificulty of making anatomical sections of palm stems and the absence of characters known to Woodcock, Dos Santos & Reynel — Wood characteristics of Amazon forest 281 relate to climate (although presence /absence of palm species, and their prevalence, may be signiicant climatically). Wood specimens, one per tree, were obtained by means of a 12-mm increment borer and thus represent the outermost wood. The cores were sealed and kept frozen until green volume could be measured for the speciic-gravity determinations. Voucher herbarium specimens are in the collection of the Herbarium of the Department of Forest Science, Universidad Nacional Agraria-La Molina (MOL), along with duplicate wood slides. The cores were sectioned using standard methods. Vessel diameter is mean tangential diameter of an average of no fewer than 30 vessels with all vessels in a given area measured (rather than being sampled randomly). Vessel frequency is the average of 5 1-mm2 ields situated randomly (with all pores counted). Vessel-element and iber length are the average of 30 measurements taken randomly and were obtained from macerated material. Ray density is the number of rays bisecting a line perpendicular to the ray axis and represents the average of 5 measurements. Speciic gravity was determined as dry weight over green volume, with volume determined by water displacement. Analysis of variance was performed by means of the Tukey-Kramer HSD test with the signiicance level set at .05. The histograms and the information in Table 1 are for the region as a whole (and thus have duplicate occurrences eliminated). Distributions were tested for normality using the Shapiro-Wilk W test. RESULTS AND DISCUSSION .28 .24 .20 .16 .12 .08 .04 0- a • • • b • – 25 – 20 • – 15 – 10 number vessel diameter (mm) Vessel diameter The ive sites do not differ signiicantly with respect to mean vessel diameter (Fig. 2a). However, trees with vessels wider than .175 mm are absent from 1 and 5, the –5 | | | | | 1 2 3 4 5 site | | | 0.00 0.05 | | 0.10 | | 0.15 | | | | 0.20 0.25 vessel diameter (mm) | | 0.30 Fig. 2. a) Comparison of vessel diameter at the ive sites. Values are averaged by species. Box plot shows the median, the boundary between the middle and outer quartiles (ends of box), and range (bars). b) Frequency histogram for vessel diameter (all sites). 282 IAWA Journal, Vol. 21 (3), 2000 riverside and swamp-forest transects. A characteristic common to these two sites is presence of standing water for signiicant periods during the year. Vessel diameter has a direct inluence on volumetric low of water through the tree and has been linked to both water availability (decreasing diameter with increasing aridity) and need (increasing diameter with increasing tree size and height; Carlquist 1988). The physiological models of Aloni (1991) relate vessel size to gradients of auxin within the tree. In this view, shorter-statured vegetation has narrower vessels because of the limited gradients of auxin that can occur over shorter distances – a good explanation for the absence of wider-vesseled taxa at site 1. Vessel diameter is often broken down into categories convenient for macroscopic identiication: vessels > .200 mm (visible to the naked eye), vessels > .100 mm (visible with a handlens), vessels < .100 mm (visible only microscopically). Absence (or very low occurrence) in the Tambopata assemblages of taxa with vessels < .050 mm is probably signiicant ecologically. Taxa with vessels in this diameter category are represented in the dry tropics (Barajas-Morales 1985; Lindorf 1994; Woodcock, submitted) and in the latewood of many middle-latitude ring-porous taxa but appear to be very rare in the wet tropics (Baas, personal communication). Analyses of the spatial variability of diffuse-porous taxa with vessels in this diameter category have shown relationships to temperature parameters (Woodcock & Ignas 1994; Wiemann et al. 1998). Taxa in the middle and large diameter classes (> .100 mm, > .200 mm) are wellrepresented in the Tambopata assemblages. Wide vessels appear in environments ranging from very dry to wet in the tropics (Barajas-Morales 1985) and in the earlywood of mid-latitude ring-porous species. Diffuse-porous woods with vessels ~.150 mm and greater are found only in the tropics. Analyses based only on diffuse-porous species suggest an decrease in percent of taxa with vessels < .100 mm going toward the tropics (Woodcock & Ignas 1994) and a signiicant positive relationship to Mean Annual Range of Temperature (Wiemann et al. 1998). The distribution of the species averages for all sites displays a considerable degree of nonnormality (Fig. 2b). The characteristics of this distribution, together with results of an ongoing study of vessel diameter distributions (Woodcock, in preparation), suggest that vessels > .150 may be a more natural category for representing the wide-vesseled taxa . Vessel diameter also shows a small but signiicant decrease along transect 2 (r = -.38, p = .04; Fig 3a). Within this meander, material was deposited sequentially as the river moved outward, leaving a series of ridges and swales that can be seen in the ine relief visible at the surface. The transect thus extended back from the river within a gradient of increasing substrate and forest age. The decrease seen in vessel diameter may correspond to lower water availability away from the river and in the older part of the stand where there is more competition for water. An alternative explanation may be the increasing prevalence of small trees, which are often narrow-vesseled, in the older part of the stand where a subcanopy element begins to be evident. Vessel density, which is inversely but nonlinearly related to vessel diameter in this sample, shows the opposite pattern (r = .53, p < .01; Fig. 3b). Woodcock, Dos Santos & Reynel — Wood characteristics of Amazon forest 200 a 0.20 vessel density (mm -2) vessel diameter (mm) 0.25 283 b 150 0.15 100 0.10 0.05 0.00 0 10 20 30 40 50 distance 60 50 0 0 10 20 30 40 distance 50 60 Fig. 3. Variation in a) vessel diameter and b) vessel density along transect 2 (loodplain forest). X-axis is the distance along the transect. Vessel density Vessel density is among the most variable of quantitative anatomical characters. In the woods studied here, values at one site range from 1–2 to > 150 mm-2 and the coeficient of variation is the highest of all the quantitative variables analyzed (Table 1). No signiicant differences were found among the ive sites with respect to this measure (Fig. 4a). Vessel density, like vessel diameter, has been linked with degree of mesomorphy/xeromorphy (Carlquist 1988). Table 1. Values of the quantitative variables for the region as a whole.* Mean Coeficient of variation Distribution signiicance nonnormal Vessel diameter .122 mm (n = 74) .40 + Vessel density 18 mm -2 (n = 74) 1.43 + Vessel-element length .728 mm (n = 71) .53 + Fiber length 2.76 mm (n = 73) .31 + Ray frequency 11 mm -1 (n = 74) .51 + Speciic gravity .539 (n = 71) .30 – *) Duplicate occurrences deleted. IAWA Journal, Vol. 21 (3), 2000 100 10 - a • • b • • – 25 – 20 • – 15 – 10 number vessel density (mm-2 ) 284 –5 | | 1 2 | | | 3 4 5 site ||||||||||||||||||||||||||||||||||||||||| 0 10 25 40 55 70 85 100 125 150 175 200 vessel density (mm-2 ) Fig. 4. a) Vessel density at the ive sites (log scale). Plots as in Fig. 2a. b) Frequency histogram for vessel density (all sites). The large number of gradations represented is to show the frequency categories employed in the OPCN database. Four vessel-density categories are generally employed in wood identiication: vessels < 5 mm-2, < 20 mm-2, 20– 40 mm-2, and > 40 mm-2 (with ring-porous taxa not included). Prevalence of taxa in these categories shows a coherent spatial pattern, and corresponding signiicant correlations to temperature and precipitation, in the eastern North America tree lora (Woodcock & Ignas 1994). Wiemann et al. (1998), looking at globally distributed loras, found that these same vessel density categories showed climate sensitivity but were not among their best predictors. Average vessel density in the Tambopata assemblages is 16.6 mm-2. Reference to the frequency distribution, however, shows the prevalence of taxa in the lower frequency categories (Fig. 4a) and demonstrates how unrepresentative mean values may be. The distribution is signiicantly nonnormal and shows a better it to a lognormal curve. Diffuse-porous taxa with vessels < 5 mm-2 are not found above a latitude of ~30° in North America (Woodcock & Ignas 1994) and appear be an indicator of tropical or near-tropical conditions. These taxa are not, however, restricted to wet environments, as is shown by the occurrence of stem-succulent taxa like Erythrina or Bombacaceae spp., which typically have few, wide vessels, where conditions are dry or very dry. But because the latter taxa represent a specialized adaptation and are limited in their occurrence, it is still probably true that taxa with sparse vessels are prevalent only in the wet tropics. Vessel-element length Vessel-element length is signiicantly greater at site 2 than at the other sites (Fig. 5a). Carlquist (1988) reports that longer vessel elements are typical of more mesic environments and presents a number of functional explanations. The mature loodplain forest at site 2 may experience the most favorable hydrologic conditions of all the sites, being close to the river and having soils that are sandy with a signiicant alluvial component yet not permanently waterlogged (as in the swamp forest). 3.2 2.8 2.4 21.6 1.2 .8 .4 0- a • b • • • – 25 – 20 – 15 • – 10 number 1.8 1.6 1.4 1.2 1.8 .6 .4 .2 0- 285 –5 | | 1 2 | | | 3 4 5 site | 0.0 | | 0.5 | | 1.0 | | 1.5 | | 2.0 | | 2.5 | vessel element length (mm) | 3.0 c d • • • • • – 30 – 20 number iber length (mm) vessel element length (mm) Woodcock, Dos Santos & Reynel — Wood characteristics of Amazon forest – 10 | | | | | 1 2 3 4 5 site | 0.0 | | 0.5 | | 1.0 | | 1.5 | | 2.0 iber length (mm) | | 2.5 | | 3.0 Fig. 5. a) Vessel-element length at the ive sites. b) Frequency histogram for vessel-element length. c) Fiber length at the ive sites. d) Frequency histogram for iber length (all sites). A general relationship between vessel-element length and precipitation has been recognized by Lindorf (1994) for six low-latitude loras from around the world. Values for four of the Tambopata sites are within the range of that for cloud forest with precipitation of 2000 mm, whereas values for site 2 exceed all those listed by Lindorf. Fiber length Sites 1 and 3 differ signiicantly in iber length (Fig. 5c). It is not possible to cite many comparative studies or associate a clear functional signiicance with this character. Barajas-Morales (1985) found that dry forest had signiicantly lower iber length than wet forest in Mexico. The values for dry forest (613 mm annual precipitation) in that study approximate those found at site 1. It seems possible that longer ibers may be associated with stiffness and thus be more typical of higher-statured vegetation. Carlquist (1988), however, notes that loristic considerations may be important in explaining the variability in this character. IAWA Journal, Vol. 21 (3), 2000 30 24 18 12 60- a b • • • • • – 20 – 15 – 10 number ray density (mm -1) 286 –5 | | | | | 1 2 3 4 5 site | 0 | | | 5 | | 10 | | 15 | 20 | ray density (mm -1) | 25 | | 30 Fig. 6. a) Ray density at the ive sites. b) Frequency histogram for ray density (all sites). Ray density The sites do not differ signiicantly with regard to ray density (Fig. 6a). The functional signiicance of dimensions and density of rays is dificult to assess. Wiemann et al. (1998) did not ind signiicant climate sensitivity for the ray-density categories used in the OPCN database (< 4 mm-2, > 12 mm-2). The modal frequency seen here is lower than that reported for a large sample of woods from all over the world (Metcalfe & Chalk 1983). 1.2 1.8 .6 .4 .2 0- a • • • b • • – 20 – 15 – 10 number speciic gravity Speciic gravity Values of wood speciic gravity at the two riverine sites (1, 2) are signiicantly lower than those at the upland forest or swamp-forest sites (3, 4; Fig. 7a). Speciic gravity is a measure of the amount of cell wall material present in wood (Panshin –5 | | | | | 1 2 3 4 5 | | | | | | | | | | | | 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 site speciic gravity Fig. 7. a) Wood speciic gravity at the ive sites. b) Frequency histogram for wood speciic gravity (all sites). Woodcock, Dos Santos & Reynel — Wood characteristics of Amazon forest 287 & de Zeeuw 1980), and thus the biomass content, and has been used in interpreting fossil wood (Wheeler et al. 1995). Generally, wood speciic gravity can be expected to be low in wet, tropical environments and higher where conditions are drier and /or colder (Chudoff 1976). Here it appears that faster-growing, early successional vegetation has lower speciic gravity wood. Saldarriaga (1989) has found that secondary succession in the northern Amazon is also associated with an increase in wood speciic gravity of the trees represented. It thus appears that this character may be useful in providing information about successional status of the taxa or forest types represented (as in Wheeler et al. 1995), especially if analyzed in conjunction with information about paleosols or depositional environment. Speciic gravity is one of the only variables studied with a distribution that is not signiicantly nonnormal (Fig. 7b, Table 1). CONCLUSIONS This study shows that some wood characters known to have climate sensitivity (e.g., vessel density and density) show little variation among forest types of the Tambopata Region and thus may be subject to overall climate controls. Other characters such as speciic gravity vary signiicantly among the forest types and can give information about local environment or ecology (low-speciic gravity taxa as indicators of early successional taxa or associations). In the case of speciic gravity, the relationships found are understandable in terms of what is known about the variability of wood properties, i.e., the lower speciic gravity of early successional taxa. For other characters showing differences among the forest types (vessel-element and iber length), more study is clearly needed to understand the site-to-site variation. A positive aspect of these results is that information about forest type and aspect and depositional environment may be inherent in the fossil wood record. On the other hand, local variability in wood characters from site to site within the same general region may, like differential preservation, be a complicating factor for paleoclimatic interpretations until more information is obtained. The results presented here add another dimension to our knowledge of the variability of wood characters in nature and point out the paucity of information that still exists about wood characters, their occurrence and functional signiicance. As seen here and has been reported elsewhere (Metcalfe & Chalk 1983), wood characters, although perhaps normally distributed in individual taxa, are often nonnormally distributed when assemblages of taxa are considered. Average values may thus misrepresent assemblage characteristics. Reliance on modal categories or percent of taxa in various categories are two ways of dealing with this problem. Identiication of indicators that are suficiently robust to be applied to small samples would also be a positive development. Wiemann et al. (1998) present the most complete model available to date of response of wood characters to climate. Their validation tests showed good results for Mean Annual Temperature, with precipitation much less well predicted. Of the quantitative characters considered here, vessel diameter and density and ray density (rep- 288 IAWA Journal, Vol. 21 (3), 2000 resented in terms of the OPCN categories) were included in their analyses but only percent of taxa with vessels < .100 appears in the inal equations for Mean Annual Range of Temperature and Cold Month Mean Temperature. These researchers suggest that the degree of variation inherent in quantitative characters (within the tree, etc.) may mean that qualitative characters are more reliable for the purposes of paleoclimatic reconstruction. A signiicant part of the problem with the quantitative variables may, however, be related to limitations of the databases used, inconsistencies in measurement (such as that mentioned by Wiemann et al. 1998, for vessel den-sity), and problems in representation (derivation of the frequency categories). Thus indications of the precipitation sensitivity of the quantitative variables (Carlquist 1988; Woodcock 1994; Woodcock & Ignas 1994) should not be discounted as yet. There are multiple types of representations that could be used to express relationships between wood characters and climate. One way is to represent wood characters and climate as continuous variables and analyze for linear relationships (Woodcock & Ignas 1994; Wiemann et al. 1998). This approach is logical given the desirability of quantitative estimates of paleoclimates and the possibility of expressing even qualita-tive variables in quantitative form (percent of taxa with a given character) when assemblages of taxa are considered. It does, however, depend on adequate sample sizes and assumes that differential preservation is not introducing signiicant bias. Another possibility would be to identify particular characters whose occurrence or prevalence is signiicant climatologically: for example, diffuse-porous woods with vessels > .100 mm as an indicator of tropical or near-tropical conditions (Mean Annual Temperature above a certain value); vessels < .050 mm as an indicator of signiicant seasonality in temperature or precipitation; vessel density < 5 mm-2 in taxa not stem-succulent as an indicator of wet tropical conditions (precipitation above a certain amount). An array of such characters, considered together, might yield quite precise climatic information and permit interpretations based on smaller samples. It might be possible to employ this approach in conjunction with other representations, as a kind of check, or as a way of looking at the response to precipitation, which has thus far proved dificult to model. ACKNOWLEDGMENTS This study beneited from assistance of faculty and staff in the Department of Forest Science, National Agrarian University-La Molina; Marcia Koth and staff of the US Fulbright Ofice; Max Gunther and staff of Peruvian Safaris; the Peruvian Instituto de Recursos Naturales; and Tania Durt and David Taylor. Comments of Michael Wiemann are also appreciated. The senior author wishes to acknowledge fellowship support from the Bunting Institute of Radcliffe College. This research was carried out with support from a grant from the Fulbright Commission for Educational Exchange between the US and Peru and NSF grant ATM-07899. REFERENCES Aloni, R. 1991. Wood formation in deciduous hardwood trees. In: A.S. Raghavendra (ed.), Physiology of Trees: 175–197. Wiley, New York. Woodcock, Dos Santos & Reynel — Wood characteristics of Amazon forest 289 Baas, P. & F.H. Schweingruber. 1987. Ecological trends in the wood anatomy of trees, shrubs, and climbers from Europe. IAWA Bull. n. s. 8: 245–274. Barajas-Morales, J. 1985. Wood structural differences between trees of two tropical forests in Mexico. IAWA Bull. n.s. 6: 355–364. Barajas-Morales, J. 1987. Wood speciic gravity in species from two tropical forests in Mexico. IAWA Bull. n.s. 8: 143–148. Carlquist, S. 1975. Ecological strategies of xylem evolution. University of California, Berkeley. Carlquist, S. 1988. Comparative wood anatomy: systematic, ecological, and evolutionary aspects of dicotyledon wood. Springer-Verlag, New York. Chudoff, M. 1976. Density of tropical timbers as inluenced by climatic life zones. Commonwealth Forestry Review 55: 203–217. Conservation International. 1994. The Tambopata-Candamo Reserved Zone of southeastern Perú: a biological assessment. Washington, D.C. Erwin, T.L. 1988. The tropical forest canopy – the heart of biotic diversity. In: E.O. Wilson & F. Peter (eds.), Biodiversity: 123–129. National Academy Press, Washington, D.C. Gentry, A.H. 1988. Changes in plant community diversity and loristic composition on environmental and geographical gradients. Ann. Missouri Bot. Gard. 75: 1–34. Lindorf, H. 1994. Eco-anatomical wood features from a very dry tropical forest. IAWA J. 15: 361–376. Metcalfe, C.R. & L. Chalk. 1983. Anatomy of the dicotyledons, 2nd ed. Vol. II: Wood structure and conclusion of the general introduction. Clarendon Press, Oxford. Panshin, A. J. & C. de Zeeuw. 1980. Textbook of wood technology, 4th ed. McGraw-Hill, New York. Phillips, O.L. 1999. Long-term environmental change in tropical forests: Increasing tree turnover. Environm. Conserv. 23: 235–248. Phillips, O.L. & A.H. Gentry. 1994. Increasing turnover through time in tropical forests. Science 263: 954–958. Phillips, O.L., A.H. Gentry, C. Reynel, P. Wilkin & C. Galvez-Durand B. 1994a. Quantitative ethnobotany and Amazonian conservation. Conserv. Biol. 8: 225–248. Phillips, O.L., P. Hall, A.H. Gentry, S.A. Sawyer & R. Vasquez. 1994b. Dynamics and species richness of tropical rain forests. Proc. National Acad. Sci. 91: 2805–2809. Reynel, C. & A.H. Gentry. In press. La lórula de Tambopata. Missouri Bot. Gard., St. Louis, Missouri. Saldarriaga, J.G. 1987. Recovery following shifting cultivation: A century of succession in the Upper Rio Negro. In: D.F. Jordan (ed.), Amazonian rain forests: Ecosystem disturbance and recovery. Springer-Verlag, New York. Salo, J.S. & R. J. Kalliola. 1991. River dynamics and natural forest regeneration in the Peruvian Amazon. In: A. Gómez-Pompa et al. (eds.), Rain forest regeneration and management: 245–256. Vol. 6 of Man in the Biosphere series. UNESCO and Cambridge University Press, Paris, Cambridge. Wheeler, E.A. 1986. Vessels per square millimeter or vessel groups per square millimeter? IAWA Bull. n.s. 7: 73–74. Wheeler, E.A. & P. Baas. 1991. A survey of the fossil record for dicotyledonous wood and its signiicance for evolutionary and ecological wood anatomy. IAWA J. 12: 271–332. Wheeler, E.A. & P. Baas. 1993. The potentials and limitations of dicotyledonous wood anatomy for climatic reconstructions. Paleobiology 19: 486– 497. Wheeler, E.A., J. McClammer & C.A. LaPasha. 1995. Similarities and differences in dicotyledonous woods of the Cretaceous and Paleocene San Juan Basin, New Mexico, USA. IAWA J. 16: 223–254. 290 IAWA Journal, Vol. 21 (3), 2000 Wheeler, E.A., R.G. Pearson, C.A. LaPasha, T. Zack & W. Hatley. 1986. Computer-aided wood identiication reference manual. North Carolina Agricultural Research Series Bulletin 474, Raleigh, NC. Wiemann, M.C., E.A. Wheeler, S.T. Manchester & K.M. Portier. 1998. Dicotyledonous wood anatomical characters as predictors of climate. Palaeogeography, Palaeoclimatology, Palaeoecology 139: 83–100. Williamson, G.B. 1984. Gradients in wood speciic gravity of trees. Bull. Torrey Bot. Club 111: 51–55. Wilson, E.O. 1987. The arboreal ant fauna of Peruvian amazon forests: irst assessment. Biotrópica 19: 245–251. Woodcock, D.W. 1994. Occurrence of woods with a gradation in vessel size across a ring. IAWA Bull. n.s. 15: 377–385. Woodcock, D.W. 1996. Tambopata y la ecologia de los bosques tropicales. Bol. Soc. Geogr. de Lima 108: 24–52. Woodcock, D.W. & C.M. Ignas. 1994. Prevalence of wood characters in eastern North America: What characters are most promising for interpreting climates from fossil wood? Amer. J. Bot. 81: 1243–1251. Appendix — Tambopata taxa. Numbers in brackets indicate site occurrences for taxa found at more than one location. Site Species Family 1 1 1 1 1 1 1 1 Cecropia icifolia Warburg ex Snethlage Enterolobium schomburgkii (Bentham) Bentham [1, 2] Ficus insipida Willdenow subsp. insipida [1, 2] Margaritaria nobilis L. f. Miconia calvescens DC. Salix humboldtiana Willdenow Sapium glandulosum (L.) Morong Visma angusta Miquel Cecropiaceae Leguminosae Moraceae Euphorbiaceae Melastomataceae Salicaceae Euphorbiaceae Guttiferae 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 Enterolobium schomburgkii (Bentham) Bentham [1, 2] Erythrina ulei Harms Eugenia lorida DC. Ficus insipida Willdenow [1, 2] Ficus mathewsii (Miquel) Miquel Guatteria cf. olivacea R.E. Fries Inga semialata (Vell. Conc.) C. Martius [2, 4] Inga sp. Minquartia guianensis Aublet Ocotea sp. Perebea angustifolia (Poeppig & Endicher) C.C. Berg Pourouma cecropiifolia C. Martius [2, 3] Pourouma guianensis Aublet subsp. guianensis Sylogene caulilora (Miquel & C. Martius) Mez Symphonia globulifera L. f. Tabernaemontana lavicans Willdenow ex Roemer & Schultes Leguminosae Leguminosae Myrtaceae Moraceae Moraceae Annonaceae Leguminosae Leguminosae Olacaceae Lauraceae Moraceae Cecropiaceae Cecropiaceae Myrsinaceae Guttiferae Apocynaceae Woodcock, Dos Santos & Reynel — Wood characteristics of Amazon forest 2 2 2 2 2 2 ?Tetragastris Theobroma cacao L. subsp. sphaerocarpum (A. Chev.) Cuatrec. Trichilia quadrijuga H. & B. subsp. quadrijuga Unonopsis veneiciorum (C. Martius) R.E. Fries Virola calophyllum Warburg Xylopia cuspidata Diels Burseraceae Sterculiaceae Meliaceae Annonaceae Myristicaceae Annonaceae 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 Casearia javitensis H. B.K. Cecropia sciadophylla C. Martius Ceiba cf. pentandra (L.) Gaertner Endlicheria bracteata Mez Guarea glabra M. Vahl Hevea guianensis Aublet [3, 4] Jacaranda copaia (Aublet) D. Don Licania britteniana Fitsch Naucleopsis ternstroemiilora (Mildbraed) C.C. Berg Nectandra lucida Nees Pourouma cecropiifolia C. Martius [2, 3] Pourouma minor Benoist [3, 4] Pouteria hispida Eyma Pouteria torta (C. Martius) Radlkofer Pseudolmedia laevis (R. & P.) J. F. Macbride Pseudolmedia macrophylla Trecul Quiina lorida Tulasne Quiina sp.? Simarouba cf. amara Aublet indet. indet. Flacourtiaceae Cecropiaceae Bombacaceae Lauraceae Meliaceae Moraceae Bignoniaceae Chrysobalanaceae Moraceae Lauraceae Cecropiaceae Cecropiaceae Sapotaceae Sapotaceae Moraceae Moraceae Quiinaceae Quiinaceae? Simaroubaceae Moraceae? Lauraceae? 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 Abarema jupunba (Willdenow) Britton & Killip Calycophyllum spruceanum (Benth.) Hooker f. ex Schumann Casearia cf. decandra Jacquin Conceveiba guianensis Aublet Cordia scabrifolia DC. Dialium guianensis (Aublet) Sandwith Hevea guianensis Aublet [3, 4] Inga cf. chartacea Poeppig Inga cf. semialata (Vell. Conc.) C. Martius [2, 4] Iryanthera juruensis Warburg [4, 5] Licania cf. britteniana Fritsch Micropholis guyanensis (A. DC.) Pierre Nectandra cuspidata Nees Neea sp. Pourouma minor Benoist [3, 4] Roucheria punctata (Ducke) Ducke Sloanea fragrans Rusby Tachigali peruviana (Dwyer) Zarucchi & Herendeen Leguminosae Rubiaceae Flacourtiaceae Euphorbiaceae Boraginaceae Leguminosae Euphorbiaceae Leguminosae Leguminosae Myristicaceae Chrysobalanaceae Sapotaceae Lauraceae Nyctaginaceae Cecropiaceae Linaceae Elaeocarpaceae Leguminosae 291 292 IAWA Journal, Vol. 21 (3), 2000 4 4 4 4 Virola calophylla Warburg Virola sebifera Aublet indet. [pink wood – Aspidosperma?] indet. [with included phloem] Myristicaceae Myristicaceae Apocynaceae? 5 5 5 5 5 5 5 5 Brosimum lactescens (S. Moore) C.C. Berg Iryanthera juruensis Warburg [4, 5] Licaria armeniaca (Nees) Kostermans Maquira coreacea (Karsten) C.C. Berg Nectandra sp. Pseudolmedia sp indet. indet. Moraceae Myristicaceae Lauraceae Moraceae Lauraceae Moraceae Myristicaceae? Leguminosae