The Useful Plants of Tambopata, Peru: I. Statistical Hypotheses Tests with a New Quantitative Technique Author(S): Oliver Phillips and Alwyn H

Home , Batis (plant), Ficus insipida

The Useful Plants of Tambopata, Peru: I. Statistical Hypotheses Tests with a New Quantitative Technique Author(s): Oliver Phillips and Alwyn H. Gentry Source: Economic Botany, Vol. 47, No. 1 (Jan. - Mar., 1993), pp. 15-32 Published by: Springer on behalf of New York Botanical Garden Press Stable URL: http://www.jstor.org/stable/4255479 Accessed: 07/10/2009 08:24

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 unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may 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=nybg.

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

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Springer and New York Botanical Garden Press are collaborating with JSTOR to digitize, preserve and extend access to Economic Botany.

http://www.jstor.org THE USEFUL PLANTS OF TAMBOPATA, PERU: I. STATISTICAL HYPOTHESES TESTS WITH A NEW QUANTITATIVE TECHNIQUE1

OLIVER PHILLIPS AND ALWYN H. GENTRY

Phillips, Oliver(Department of Biology, WashingtonUniversity, One Brooking'sDrive, Campus Box 1137, St. Louis, MO 63130-4899, U.S.A.), and Alwyn H. Gentry (MissouriBotanical Garden,Box 299, St. Louis, MO 63166-0299, U.S.A.).THE USEFULPLANIS OF TAMBOPATA, PERU: I. STATISTICAL HYPOTHESES TESTS WITH A NEW QUANTITATIVE TECHNIQUE. Economic Botany 47(1):15-32. 1993. This paper describes a new, simple, quantitativetechnique for evaluatingthe relative usefulnessof plants to people. The techniqueis then comparedto the quantitativeapproaches in ethnobotanythat have been developedrecently. Our technique is used to calculatethe importanceof over600 speciesof woodyplants to non-indigenousmestizo people in Tambopata,Amazonian Peru. Two general classes of hypothesesare formulated and tested statistically,concerning (1) the relativeimportance of differentspecies, and (2) the importance of differentfamilies. The plantfamilies are comparedwith respectto all uses, and with respect tofive broadgroups of uses. Palms, Annonaceae,and Lauraceaewere found to be the most useful woodyplant families. On average,the 20 largest woodyplant families are most importantto mestizosfor subsistenceconstruction materials, followed in descendingorder by commercial, edible, technological,and medicinaluses.

Las plantas uitiles de Tambopata, Perui:I. Pruebas estadisticas de hip6tesis etnobotanicas con una nueva tecnica cuantitativa.En este estudiose describeuna nuevatecnica cuantitativa para la evaluaci6nde la relativautilidad de plantas a la gente. Esta tecnicase comparacon aquellas tecnicas cuantitativasrecientemente desarrolladas en etnobotdnica.Con esta tecnica nosotros estimamosla importanciaque las plantas lenosas, mds de 600 especies,tienen para los mestizos de Tambopatade la Amazonia del Peru. Estadfsticamente,se pruebandos hip6tesisgenerales concernientesa (1) la relativa importanciade especies diferentes,y a (2) la importanciade diferentesfamilias. Las familias de plantas son comparadasentre ellas en relaci6na todos los usos, y con respectoa cinco gruposamplios de usos. Se descubri6que lasfamilias lenosas mas utiles son las palmeras,Annonaceas, y Lauraceas.En terminopromedio, las 20 familias mas grandes de plantas lenosas tienen prioridadcomo materialesde construcci6nde subsistencia, seguidasen ordendescendiente por sus usoscomerciales, comestibles, tecnol6gicos, y medicinales. Key Words: quantitative ethnobotany; hypothesis tests; statistics; Tambopata, Peru; mestizo.

The fate of tropical forests and indigenous peo- nobotany has had several causes. The science is ples have recently attracted considerable popular intrinsically interdisciplinary, making it suscep- interest. Yet, paradoxically, the pace of research tible to charges of being vague and imprecise, into the indigenous plant uses and vegetation and ethnobiologists study and learn from cul- management processes that could offer alterna- tures that western science and biomedicine hold tives to the destruction is dwarfed by accelerating to be "primitive" and inferior. However, some rates of cultural and biological extinction. In spite criticism of the methods and philosophical ap- of ethnobotany's relatively high public profile, proach of ethnobotany has certainly been justi- few institutions apparently see it as a real science fiable. As several others have pointed out there worthy of significant financial support. is a lack of methodological rigor in much eth- Historically, this negative perception of ethnobotanical research (cf. Johns, Kokwaro, and Kimanani 1990; Trotter and Logan 1986), and a frequent unwillingness to define falsifiable hy- I Received 15 June 1992;accepted 13 October1992. potheses.

Economic Botany 47(1) pp. 15-32. 1993 ? 1993, by The New York Botanical Garden, Bronx, NY 10458 U.S.A. 16 ECONOMICBOTANY [VOL. 47

Partly in response to the long-standing per- the way ethnobotany is done, but it will also ception of ethnobotany as not being "scientific," enhance the image of ethnobotany among other there is now a strong movement to modify the scientists. traditional compilation-style approaches to eth- It is possible to draw a close analogy between nobotany, by developing methods that allow re- the hypothesis-testing aims of quantitative eth- searchers to quantitatively describe and analyze nobotany as outlined here, and the recent trend patterns in what they study (Prance 1991; Elvin- in systematics towards a more scientific meth- Lewis and Lewis n.d.). Quantitative and even odology. This was mainly triggered by the ap- statistical hypotheses-testing techniques have re- plication of cladistic methodology to plant sys- cently been applied to, inter alia, the following tematics, which has forced taxonomists to questions: (1) evaluating the importance of veg- reevaluate some of their assumptions about the etation to one ethnic group (e.g., Prance et al. nature of the characters they use and the way in 1987; Unruh and Alcorn 1988; Anderson and which they use them (e.g., Wiley 1981; Gilmartin Posey 1989; Balee and Gely 1989); (2) comparing 1986; Sneath 1988; Stevens 1991). The tradi- the uses of, (a) hectare forest plots (Prance et al. tional role of taxonomists as describers and com- 1987), and (b) entire or partial regional floras pilers has now largely been superseded by a mod- (Toledo et al. 1992; Bye n.d.), by different ethnic em systematics in which the construction of groups; (3) comparing the importance of different phylogenies, using "refutation by experimenta- vegetation types to one people (Boom 1990; An- tion" as a fundamental philosophical principle, derson 1990, 1991; Salick 1992); establishing the occupies a central position. We believe that relative importance of different (4) medicinal quantification, and the associated explicit hy- plant species (e.g., Adu-Tutu et al. 1979; Elvin- pothesis-testing approach, can have similar ben- Lewis et al. 1980; Friedman et al. 1986; Johns, eficial effects in ethnobotany. By attracting sci- Kokwaro, and Kimanani 1990; Trotter and Lo- entific respect, and hence more students and gan 1986) and (5) families (Moerman 1979, 1991; research funding, these approaches can help to But, Hu, and Cheung Kong 1980; Kapur et al. generate sufficient high quality information to 1992); (6) comparing the importance of different impact on conservation and development issues. plant families and uses among plants sold in a In this paper we (1) develop a simple technique peasant marketplace (Martin 1992); and (7) test- of analyzing ethnobotanical data, and (2) com- ing a model of the origins of medicinal plant use pare it with existing approaches in quantitative (Johns and Kimanani 1991). ethnobotany. We then (3) demonstrate how this In cultural anthropology, a quantitative school technique can be used to test two kinds of hy- has long been influential, and more recently there potheses that are of interest to ethnobotanists to has been interest in integrating quantitative and illustrate its potential for contributing to the de- qualitative types of research (e.g., Johnson 1978; velopment of quantitative ethnobotany. In a sec- Smith and Heshusius 1986; but see Hammersley ond paper (Phillips and Gentry 1993), we pro- 1992). Until recently, however, ethnobotanists pose several more kinds of ethnobotanical had been more reluctant to appreciate the po- hypotheses, and test them statistically with our tential significance of quantification. Indeed, the data. very term "quantitative ethnobotany" was coined as recently as 1987 (Prance et al.). Here we define ECOLOGICAL AND CULTURAL it, relatively broadly, "as the application of quan- SETTING titative techniques to the direct analysis of con- The study area is located in the southeastern temporary plant use data." From the examples Peruvian department of Madre de Dios, and in- above alone, it is clear that quantitative ethno- cludes the Zona Reservada Tambopata (ZRT) botany has the potential to contribute to a wide and surrounding area (Fig. 1). The forest-types range of important issues at the interface of sci- of the Tambopata region are representative of ence and development. A further benefit of ap- seasonal tropical moist forest in southwestern plying quantitative techniques to data analysis is Amazonia. The regional climate and forest-types that they act as a spur for conscious attempts to are described in detail by Erwin (1984), Gentry refine the methodology of data collection (Johns, (1988), Phillips (n.d.), and Reynel and Gentry Kokwaro, and Kimanani 1990). Closer attention (n.d.). to methodological issues will not only improve This ethnobotanical study is unusual in that 1993] PHILLIPS & GENTRY: QUANTITATIVE ETHNOBOTANY I 17

we worked with non-Indian tropical forest peo- 72 71 70 69 ple. Recently, several scholars have drawn at- -0 P U 0- tention to the need for more ethnoecological work with such peoples (e.g., Hiraoka 1992; Padoch and de Jong 1992; Parker 1989; Prance 1991), who may have adapted techniques from extinct or endangered indigenous cultures. Most of the current inhabitants of the Tam- bopata area are mestizos. The term "mestizo" covers a wide spectrum of people, from very recent migrants from the Andes, to colonists of ACIONAL MA 12- >30 years in the region who came from else- where in Madre de Dios or further north in Pe- Pto.Mwoineai 3>~~~~N ruvian Amazonia, and who, like Brazilian ca- boclos, are linked biologically and historically to *13 kilometers 13- native American cultures (Parker 1989). Eighty- 0 50 100 - three percent of our adult informants were born 72 7* 70 6 in Madre de Dios, but only about 55% had lived Fig. 1. Map of Madre de Dios: showing location more than half their lives in the immediate area of the Zona Reservada Tambopata, other protected of the newly-formed La Torre community on the areas, and principalrivers. Reproducedwith permission from Duellman and Koechlin 1. west bank of the Tambopata River. Such mo- (1991). Cocha CashuBiological Station. 2. PakitzaBiological Station. bility is not unusual among non-indigenous Am- 3. Zona Reservada Tambopata. 4. Reserva Cuzco azonians (e.g., Coomes n.d.; Padoch and de Jong Amaz6nico. 5. Reserva Lago Sandoval. 6. Santuario 1990), and implies that conditions that allow a Nacional del Heath. relatively free exchange of knowledge between mestizos about their environment have long ex- isted. iiedo-Vasquez et al. 1990). This approach explicitly ignores herbs, shrubs, small vines, and epiphytes, but the problems with identifying ster- METHODS ile vouchers make it almost impossible to do DATA COLLECTION comprehensive ethnobotanical inventories of Most of the ethnobotanical data collection was smaller plants (but see Salick 1992). We do in- done in a series of one hectare tree plots at the clude large lianas, otherwise only included in for- ZRT. The plots were originally laid out and tagged est hectare-plot ethnobotanical analyses by Boom by T. Erwin and the Smithsonian BIOLAT pro- (1989) and Paz y Mifio et al. (1991). gram (Erwin 1984). Collections of all tagged tree Ethnobotanical data were collected between and liana species -10 cm d.b.h. in the seven 1986 and 1991 by OP and numerous assistants forest plots in the ZRT were made between 1984 (see Acknowledgments). We interviewed a total and 1989 by AG, variously working with Proyec- of 29 people, to record their knowledge of some to Flora del Peru6botanists Rodolfo Vasquez, of the approximately 570 tagged plot species in Nestor Jaramillo, Percy Nuniez and Camilo Diaz, the 6.1 ha surveyed, and of 35 additional tree, and with OP (Gentry collection numbers 45 576- liana and arborescent palm species. In total, we 46250, 51064-51558 and 57535-58155, and recorded use data in 1885 independent "events." Phillips 182-290 and 620-691). Collection du- An "event" is defined as the process of asking plicates, almost all of which have been deter- one informant on one day about the uses they mined (cf. Gentry 1988; Reynel and Gentry n.d.), know for one species. Thus, if in one day we are deposited at MO and USM, and in part at encountered the same species more than once, AMAZ and CUZ. The tagged, identified trees the informants' responses to our question in each and lianas in a total of 6.1 ha, representing seven encounter were simply combined, with one ex- different forest types, were used for the ethno- ception: when the same informant gave a differ- botanical study, in a similar manner to other ent name to the same species on the same day, ethnobotany-inventory studies (Prance et al. each encounter was then recorded as a separate 1987; Boom 1989, 1990; Anderson 1991; Pi- event. (We supposed that the different identifi- 18 ECONOMICBOTANY [VOL. 47 cations indicated that the initial response could various other activities (e.g., collecting and pro- not have influenced the subsequent responses, cessing non-timber forest products for subsis- i.e., the second event was "independent" of the tence and commercial uses (>20 informants), first.) hunting (? 8), felling timber trees and/or making Eighty percent of the data were gathered by furniture, canoes etc. (-7), working in a tourist OP using a relaxed "walk-in-the-woods" ap- lodge (6), selling herbal remedies (>4), boat- proach. Initially, we worked only within the hect- driving (3)). As a mostly male team we found it are plots, recording information on local names easier to work with mestizo men than women, and uses together with stem tag numbers for later and this is reflected in the male bias in our in- cross-reference to the plot lists. In 1991, further formants (24:5). Given that our investigation is "walk-in-the-woods" data were collected by OP of uses of woody forest plants, however, this is along more than 30 km of the area's trails; in probably not a severe handicap, as most mestizo these cases, local names and uses of trees and men spend more time in primary forest than lianas that OP recognized to species in the field most women, and we predict therefore that they were asked. As the project proceeded, it became should know those plants better. However, our clear that some species had a unique and uni- impression is that women are frequently more versally known local name, especially among knowledgeable about herbaceous and non-forest heavily-used plants; for some of these well-known plants than men, and especially for medicinal species supplementary data were collected sim- uses (cf. Lewis and Elvin-Lewis 1990; Kainer and ply by asking informants to relate the uses they Duryea 1992). As far as we are aware, there are knew. no published studies that systematically compare Our interest in analyzing the data statistically the degree and kind of ethnobotanical knowledge made us conscious of the need to minimize the of women and men, in any tropical forest culture. possibility of one informant's answers directly Notwithstanding the potential methodological influencing another's. Therefore, on almost all difficulties, there is a clear need for such work to occasions, we interviewed informants individ- address this large gap in our understanding of ually. This, together with the large divergence in such cultures. the quantity and kind of uses reported even by similarly-aged informants (Phillips and Gentry PRIMARY DATA ANALYSIS 1993), gives us reasonable confidence in using Over 1 6 uses were defined by the informants. statistical techniques that require that each in- The uses are listed in Appendix 1, to give the formant not be directly influenced by other in- reader an indication of both the astonishing va- formants' responses. riety of ways in which even non-indigenous peo- In addition, we attempted to apply the follow- ple may exploit their natural resources, and to ing conditions consistently, in order to minimize show how some uses (especially some medicinal the potential incentives to embellish informa- and technological ones) are highly species-spe- tion, give "wrong answers," or to treat the pro- cific, whilst other requirements (especially con- cess as a chore: struction needs) can be satisfied by a broader range of species. The condensed definitions we (1) Make the data-gathering process as informal give to each use listed in Appendix 1 are intended as possible. to match the informants' concepts as concisely (2) Fit it in when the informants had free-time. as possible, and as far as we are able to determine (3) Compensate informants for their time, most- the categories reflect emically defined concepts ly with tools or clothing that they asked for. of functional equivalence. In general, then, the We tried to approximately equate the ma- level of resolution of the analysis approaches that terial value of such gifts with the monetary of a mestizo taxonomy of uses. The one notable and/or subsistence value of the time and la- exception is among commercial uses, where we bor given to the project by the informant. considered several separately described uses to- We also tried to work with a representative gether to simplify the analysis. However, this sample of the local people, of all ages (cf. Ap- grouping barely affects the results of the analyt- pendix 1), professions, and of both sexes. Our ical process described below because one single informants are mostly farmers along the lower use (the selling of sawn planks) dominates the Tambopata River, although many worked in commercial category. 1993] PHILLIPS & GENTRY: QUANTITATIVE ETHNOBOTANY I 19

The distribution of number of species per use 30 - loosely approximates a log-normal distribution (Fig. 2). From the position of the "veil line" (Hubbell and Foster 1990; Preston 1948, 1962), we calculate that the informants we worked with 2-2 would tell of an additional 25 to 30 uses (mostly technological and medicinal) if we could have sampled all the large woody plant species in the Z 10 ZRT. Thus, we predict that the 29 informants know of up to 150 separate uses for the ZRT large woody plants. 0 By contrast, our broad categories of uses (Ed- 1 2 4 8 16 32 64 128 256 ible, Construction, Commerce, Medicinal, and Species pw Use Technology and Crafts) are based on an explicitly Fig. 2. Number of uses vs. species per use. The etic perspective. For example, among our data distribution approximates a truncated log-normal dis- there are instances where the ingestion of a plant tribution; the "veil-line" is the vertical axis. part is inseparably both medicinal and nutritive (cf. Etkin 1986; Iwu 1986), and in such cases we had to make a judgement to assign the use to one category and not another. These categories are ns thus, in part, artificial constructs (Prance et al. where nS equals the number of informants inter- 1987), erected for analytical convenience. viewed for species s. We first analyzed the data with a technique we Table 1 illustrates the technique with a typical developed related to other "informant-indexing" example, Clarisia racemosa R.&P. (Moraceae). or "informant-consensus" methods (Adu-Tutu Table 1A shows data from three events with one et al. 1979; Friedman et al. 1986; Johns, Kok- informant; for each event each individual use waro, and Kimanani 1990; Perez Salicrup 1992; cited by the informant is recorded. The number Trotter and Logan 1986). The differences in our of uses are totaled for each event, and each use technique from those of Adu-Tutu et al., Perez is averaged across events. These mean values are Salicrup, Trotter and Logan, and Friedman et al. transferred to Table 1B, after totalling within the result partly from the different nature of the data. use categories (two uses were recorded in the For example, apparently none of these investi- medicinal category, one in each of the other cat- gators reinterviewed the same informants about egories). Table l B lists the UViKvalues of Clarisia the same plant. Similarly, Johns, Kokwaro, and racemosa for each of the six informants inter- Kimanani developed a technique to factor out viewed, by category and by total. UVs values for the influence of a species' abundance on the like- the species, by category and total, listed in Table lihood of its being used, a refinement not directly 1C, represent the mean values of UVisfrom Table relevant to the questions we ask here. Although lB. each of the techniques is designed to address dif- This example highlights two important aspects ferent questions, all share the valuable property of the technique. Firstly, out of the 13 C. race- that they make statistical analyses of ethnobo- mosa events, one use-the commercial use of tanical data possible. sawn wood-is only mentioned once. The av- Our estimate of the use value of each species eraging process reduces its contribution to an s for each informant i, UVi,, is defined as: appropriately negligible 3% of the total UVs;thus, given a sufficient number of events and informants interviewed, the technique guarantees that

nis unimportant uses or even informants' mistakes will not unduly influence the UVs. where Uis equals the number of uses mentioned Secondly, it is clear that the answers given by in each event by informant i, and nis equals the the same person to the same question can vary number of events for species s with informant i. enormously. In events #1 and #3, the informant Our estimate of the overall use value for each mentioned just one, medicinal, use, but in event species s, UVs, is then: #2 he mentioned five uses in four broad cate- 20 ECONOMICBOTANY [VOL.47

TABLE 1. CALCULATING USE VALUES (UV5) USING THE DATA PROCESSING TECHNIQUE: CLARISIA RACEMOSA AS AN EXAMPLE.

A. Informant 1: all the C. racemosa uses mentioned in three separate events.

0 Evsnt#2 1X.00 0 1 A

: Re'->.--'t,,.,~~~~~~~~~~~~~~~~~~~~~~~...... ,..: ~~~~~~~~~~~~~~~

R B ' . ' ' X X X' ' ' g ...... R R gR R ...... f 2RR~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~x..X.

BR'f5. Use 3'f values.f4- *' o.f ' C . 'raceosa'' (UV for infoRmants 1 to 6.

B.~~~~~~~~~~~~~~~~~~~~~~~~~~X.s vaue of C. raems fo.. nomnt o6

...... gB*SggR...... g.. ..? .....-0:....0 R':i

Informant |Total tCon2 gEdi2 Corn2 MNed2 {Tec2

12 12667 0? 1 0? 10667 11

13 0o 0o 0o 0o 0 0

' Specific uses: Sawn-wood = wood sawn to build house; Edible Fruit =fr uit eaten raw; Wood: Sold =sawn wood sold by mestizos; Med,A = barkc scrapings applied as a poultice to leishmaniasis sores; Med,B =latex mixed with latex from other species and applied as a poultice to reduce swellings and bruises. 2 Categories of use: Con = construction uses; Edi =edible uses; Coin= commercial uses; Med =medicinal uses; Tec = technological uses. 1993] PHILLIPS& GENTRY:QUANTITATIVE ETHNOBOTANY I 21

TABLE 1. CONTINUED.

C. C. racemosa use values (U,) (= mean of six UV1. values).

2 1 E Clarisia Total Con 2 Edi2 Comr2 lMed2

1 T0 0

_ _ _ cel=or a s a 1 6O I _ 00 00 6. 0

0.306 0.056 0.500 0.500 UV's= 1.667 0.306 mean of

each UVi8

gories (construction, commerce, edible, and me- research effort as equally as possible across all dicinal)! There are several possible reasons for the species. this variation, but this example clearly shows how the accuracy of our estimate for UVi, will COMPARISON OF METHODS IN increase as the number of events increases. Sim- QUANTITATIVE ETHNOBOTANY ilarly, Table l B displays the contrast in averaged Although a number of different approaches responses for different informants, showing that have been taken to the collection and analysis of the same principle holds for our estimate of UV,. quantitative data in ethnobotany, those that are The observation that accuracy increases with primarily focused on direct non-market value sample size is hardly surprising, but one of the ("Consumptive Use Value" sensu McNeely et benefits of using techniques like these is that they al. 1990) can be grouped into three categories can give a quantitative description of the rela- (Table 2): (1) the importance of different plants tionship between sample size and the accuracy or uses is directly related to the percentage of of the estimate of UVs, and thereby provide valu- informants mentioning a given use (Adu-Tutu et able insight into the reliability of ethnobotanical data. 3.5 This notion is explored further in Fig. 3, in which the running mean UV, of Clarisia racemosa is plotted against the number of informants. The error bars represent normal theory 95% confidence intervals, for each UV, estimator based on 1 through 6 informants. Each distribution is based on all possible combinations of recorded UVi, values, for a given number of informants. The graph underscores the need for working with several informants, but also shows E2.5 - extapolated how the rate of improvement in our confidence of the UV, estimate rapidly decreases as the number of informants increases. This allows us to make a clear prescription about ethnobotanical Fg3.Clarisia racemosaof Informants use values. methodology: given limited time, researchers Range rep- resentsetmtsnormal theory 95% confidence intervals for the seeking to quantify the utility of a fixed number UVnnNumber based on the populationof all possible UVsstimtesfor a given informantsample size. (For of species to an ethnic group will maximize sta- oneinfrmat n= 6 possible UVs values, for two n = tistical confidence in their data by spreading their 15,forthre n= 60, for four n = 15, for five n = 6.) 22 ECONOMIC BOTANY [VOL. 47

TABLE 2. A COMPARISON OF THE APPROACHES USED TO ESTIMATE CONSUMPTIVE USE VALUE.

Use values indexed Use values assigned Aspect/technique by informants by researchers Uses counted Methods explicit Yes Unlikely (because values No (simply counting and results po- are assigned a posterio- uses cannot account tentially repro- ri by researchers) for unequal levels of ducible? research effortl) Objectivity of use Relatively objective Relatively subjective Relatively subjective values generated Distribution of use Continuous Discrete Discrete values generated Negative data use- Yes (negative data is Yes (negative data may No able in analysis?2 used directly to calcu- help differentiate "mi- late use values) nor"/"major" uses) Statistical compari- Yes (score from each No No sons possible be- species or informant tween uses? represents one observation) Statistical compari- Yes (score from each in- No No sons possible be- formant represents tween species? one observation) Statistical compari- Yes (knowledge of well- Yes, but less precise Yes, but less precise sons possible be- studied species can be than A; no such study than A; no such tween infor- compared) known study known mants? Statistical compari- Yes, and such compari- Yes, but such compari- Yes, but interpretation sons possible be- sons could be valid sons are less valid is difficult because no tween ethnic even with different re- with different re- two studies involve groups, plant searchers searchers an equal level of re- communities? search effort Valid comparisons Yes, the relative impor- Unlikely (researchers Unlikely (one commer- possible between tance of each is di- value-judgements may cial use is unlikely to subsistence and rectly determined obscure the compari- be of equivalent im- commercial uses? from the informants' son) portance to one sub- responses sistence use) Speed of data col- Most time-consuming Less time-consuming Potentially the least lection and anal- time-consuming ysis

This problem is redundant in Moerman's (1991) study because he analyzed a plant use universe, not a sample. 2 "Negative data," defined as informant either not recognizing plant or recognizing it but knowing no uses. al. 1979; Friedman et al. 1986; Johns, Kokwaro, Table 2 compares the three groups of methods and Kimanani 1990; Kainer and Duryea 1992; for some selected important attributes. The one Perez Salicrup 1992; Trotter and Logan 1986; significant drawback to the approach of indexing this study); (2) use values are assigned by the of use values by informant response is that it is researchers (Piiiedo-Vasquez et al. 1990; Prance more time-consuming than other methods. It is et al. 1987; Turner 1988), and (3) the uses in clearly easier and quicker to subjectively assign separate categories of different taxa are listed as importance values, or simply count the number part of the taxon's activity signature (Balee and of uses recorded, than it is to calculate an im- Gely 1989; Hunn 1982) or simply counted (e.g., portance value for each use for each informant Anderson 1991; Paz y Mifio et al. 1991; Toledo and/or species, but with access to spreadsheet et al. 1992; Unruh and Alcorn 1988). software the averaging process is not unduly time- 1993] PHILLIPS& GENTRY:QUANTITATIVE ETHNOBOTANY I 23 consuming. More importantly, this averaging informants; and (e) investigate the relationship process incorporates more information about the between informant knowledge and age. relative importance of species than a simple count For each hypothesis test we used a modified of the number of uses per species can do. (For version of the raw UV, data. We merged use data comparative purposes, using our data, we re- of congeneric species when those species consis- gressed each species' UV, on the number of uses tently shared the same or similar mestizo names per species. The r2 value of the regression equa- and uses (cf. e.g., Adu-Tutu et al. 1979:323). For tion is 0.563; i.e., simply totalling the number of example, all Salacia (Hippocrateaceae) liana spe- uses per species explains 56.3% of the variance cies ? 10 cm d.b.h. share the same names "sapote in UV, values; the remaining 43.7% is accounted de liana" and "chuchuhuasillo" and uses (edible for by the process of indexing each use by its fruits, and bark as an ingredient in a drink that relative importance.) has an invigorating effect): we calculated the The most important advantages of the infor- modified use value of each species that comprises mant-indexing method are related to the range this one "folk-species" from our data on all Sal- of detailed statistical analyses of ethnobotanical acia sapote de liana and chuchuhuasillo species. data it permits, and the increased likelihood of The principal reason for making this modifica- obtaining significant results. This is because it tion is that it creates taxa that more closely ap- makes efficient use of the research effort. (More proximate mestizo concepts of their environ- information is contained in continuous data than ment than do botanical species alone. A further in discrete data; furthermore, the approach al- benefit is the improvement in quantity and qual- lows inclusion of negative data.) It is also clearly ity of ethnobotanical data for each taxon. Thus, the closest to satisfying established scientific from more than 605 species with data (mean methodology-both because the method of as- number of events per botanical species = 3.16, signing importance is explicit and relatively in- mean number of informants per botanical spe- dependent of the researcher (so that the results cies = 2.05), we create 380 folk-species (mean are potentially reproducible by different re- number of events per folk-species = 4.96, mean searchers), and because it permits the testing of number of informants per folk-species = 2.59). specific, falsifiable null hypotheses (Popper 1963). Henceforth we use the term "species" as short- Indeed, null hypotheses can be tested in a broad hand for "folk-species." sweep of botanical, ecological and human categories interesting to ethnobotanists, e.g., between COMPARISONSOF SPECIES uses, taxa, informants, ethnic groups, and plant = two species are equally useful (for each pair- communities. To our knowledge, until now this Ho wise broad approach has never been used to test hy- comparison) potheses other than those relating to the popu- Each species' use value (UV,) can be compared larity and/or effectiveness of different medicinal statistically with any other species' use value(s) plants (Adu-Tutu et al. 1979; Friedman et al. by the non-parametric Kruskall-Wallis test 1986; Johns, Kokwaro, and Kimanani 1990). (equivalent to the Wilcoxon rank sum and Mann- Whitney U-tests for pair-wise comparisons), using each species' set of UViSvalues as data. (Be- HYPOTHESIS TESTS cause we collected ethnobotanical data for as To demonstrate the potential of this broad ap- many as possible of the 570 plot tree and liana proach for analyzing ethnobotanical data, we (a) species, we had relatively few informants for most apply use values generated by the specific tech- species, and so could not compare the use values nique described in this paper to make compar- of most species this way.) Table 3 illustrates the isons between the utility of different species to results for species with nine or more informants mestizo people. We then use modified use values each (for each of these, folk-species and botanical as a basis for comparing (b) the utility of different species are equivalent). The null hypothesis is plant families to mestizos. In a second paper rejected at the 5% level for 20 out of 28 com- (Phillips and Gentry 1993), we compare (c) the parisons. Note that, by chance alone, we would contributions of ecological, physiognomic, and expect the null hypothesis to be rejected for 5/100 phylogenetic factors to determining a species' x 28 = 1.4 comparisons. utility, and (d) the relative knowledge of different It is worth noting that, of the species compared 24 ECONOMIC BOTANY [VOL. 47

TABLE 3. USE VALUES OF EIGHT SPECIES COMPARED. (Read downwards for first species in each pairwise comparison; shaded squares indicate the column species has a higher use value than the row species)

D1 EF1 F1 I1 J 01 Pi S1

D XXXXX ~ ~ 3 0 t~ 25 4

xxxxx ** * *S*

B 16.2 XXXXX 14.2 0.4 6. 6 14.9 16. 2 12. 6

xxxxx *** **

lF 0.6LZ SE4m XX 2 411 0.6 xxx*** xxxx1 *7 I13. 04ggg 11. IXXXXX 131 171 1414 7*4

** **g *l...l l XXXX X .....xxxx

01 11.6 14. 3 74...... 5... * xxxxx..** ***

All values are chi-squaredvalues for pairwisecomparisons Significancelevels for the chi-squared values: ***= P < 0.001, ** =P < 0.01 * =P < 0.05, s = P < 0.10; all pairwise tests are Kruskall-Wallisnon-parametric comparisons. (Wilcoxon rank sum or Mann-WhitneyU test generateidentical P-values for pairwisecomparisons.) Note that these are P-values for individual pairwise comparisons.The trueP-values for the most significantcolumn or cell wouldbe subjectto a Bonferronicorrection.

1 Species with ?9 informants: D= Dipteryx odorata (Fab: Papsiionoid), UVs 1.70, ni 9. E= Euterpeprecatoria (Palm), UVs = 4.30, ni 14. F = Ficus insipida (Moraceae)? UVs 2.17, n 9. I= Irzartea delto,dea (Palm), UVs 4.41, ni 11. J= Jessenia bataua (Palm), UVs = 3.12, n, = 9. O0= Oenocarpues mapora (Palm), UVs 281,2 ni 12. P= Pouteria macrophylla (Sapotaceae), UVs 1.00, nz 9. S= Scheelea phalerata (Palm), UVs 2.67, ,1l = 12. 1993] PHILLIPS& GENTRY:QUANTITATIVE ETHNOBOTANY I 25

3 Are TABLE 5. FAMILIES WITH FAMILY USE VALUES LOWER THAN EXPECTED. 2.5- Number of Family Family species use value1'2 Z 2 -Ann < ~~~~~~~~~~~~~~~~~~~Lau Mys Bignoniaceae 15 0.424 s 15- Fab:C > ~~~~~~~~~~~Bur Nyctaginaceae 14 0.433 s a) i - - or Fab:M Sap llu Rubiaceae :D ~~~Fab:PChr 16 0.525 s I ~~~~~~Mel Flacourtiaceae 10 0.575 bps Rub Fla Apocynaceae 13 0.669 .5 -Eieycl Euphorbiaceae 22 0.734 * Melastomataceae 11 0.945 0 Chrysobalanaceae 13 1.019 -1.96 0 1.96 Fabaceae (Papilionoids) 29 Number of Individual-Case Standard Errors from Regression Line 1.152 Fabaceae (Mimosoids) 38 1.179 Fig. 4. Family use values vs. standard error of their Moraceae 54 1.184 s residuals. Sapotaceae 25 1.202

I Null hypothesis: the average total use value of species in a given here, palms clearly dominate (two, Iriartea del- family is not significantly more or less than the predicted mean total FUV of families of equivalent size. toidea R.&P. and Euterpe precatoria Mart. are 2 H0 rejected at the following levels: s = P < 0.10, * P < 0.05. the most useful species of all to mestizos). The pattern of dominance of palms is confirmed by our familial analyses (COMPARISON OF PLANT COMPARISONSOF PLANT FAMILIES FAMILIES, below). There is clear potential for ap- = on average,a species in a given is plying this technique to other kinds of inter-spe- Ho family no more or less useful than predictedfor cific comparisons. Thus, with sufficiently focused species in families of equivalentsize data, the relative importance of different plants (for eachfamily) as medicinal species, or even as remedies or preventatives for one important health problem, In order to evaluate the importance of different could be statistically evaluated. Information of plant families to mestizos, we analyze our use- this kind may be helpful in setting priorities for value data with a regression technique modified conservation, and would clearly be an important from Moerman (1991). (We could also have used factor in discriminating between candidate spe- Kruskall-Wallis or Mann-Whitney comparisons cies for applications as diverse as pharmacolog- to test pair-wise null hypotheses, in the same way ical screening, or promotion in traditional med- that we compared species, using as separate ob- icine, agroforestry, or enrichment planting. servations each UVs value, as opposed to each UVi, value. We use regression analysis here to demonstrate some of the choice in analytical technique available when working with use value TABLE 4. FAMILIES WITH FAMILY USE VALUES data.) HIGHERTHAN EXPECTED. For each family, Moerman regressed the total Number of Family number of Native American medicinal species Family species use valuel"2 in North America on the total number of species, Clusiaceae 11 1.261 and calculated residuals for each family. Families Burseraceae 12 1.353 with positive residuals (i.e., above the regression Fabaceae (Caesalpinoids) 16 1.555 line) have more uses than predicted by the num- Meliaceae 12 1.559 ber of species in the family, families with nega- Myristicaceae 18 1.686 tive residuals have fewer. Unlike Moerman, we Lauraceae 62 1.847** have only a sample of each family's set of Annonaceae 38 1.976** UV, values, so we Arecaceae 11 2.978** modified the technique to calculate individual-case standard errors of departure for l Null hypothesis: the average total use value of a species in a given each Family Use Value (FUV) from the regres- family is not significantly more or less than the predicted mean total FUV of families of equivalent size. sion, where FUV = Sum(UVs)/(number of spe- 2 Ho rejected at the following levels: * = P < 0.01. cies). For a given family size, residuals for in- 26 ECONOMICBOTANY [VOL. 47

TABLE 6. FAMILY USE VALUES (CONSTRUCTION TABLE 7. FAMILY USE VALUES (COMMERCIAL USES). USES).

Number of Family Number of Family Family species use value '2,3 Family species use value',2,3

Moraceae 54 0.174 *** - Flacourtiaceae 10 0 - Bignoniaceae 15 0.202 s - Chrysobalanaceae 13 0 s - Fabaceae (Mimosoids) 38 0.248 ** - Hippocrateaceae 16 0 * - Apocynaceae 13 0.259 - Rubiaceae 14 0 * - Rubiaceae 16 0.301 - Moraceae 54 0.026 ** Sapotaceae 25 0.392 - Euphorbiaceae 22 0.039 * Euphorbiaceae 22 0.396 - Sapotaceae 25 0.104 - Melastomataceae 11 0.409 - Bignoniaceae 15 0.111 - Burseraceae 12 0.424 - Fabaceae (Mimosoids) 38 0.151 - Nyctaginaceae 14 0.433 - Apocynaceae 13 0.154 - Flacourtiaceae 10 0.500 - Arecaceae 11 0.265 + Fabaceae (Papilionoids) 29 0.537 - Melastomataceae 11 0.318 + Chrysobalanaceae 13 0.669 + Clusiaceae 11 0.364 + Clusiaceae 11 0.682 + Burseraceae 12 0.364 + Arecaceae 11 0.791 + Fabaceae (Papilionoids) 29 0.371 + Myristicaceae 18 0.848 s + Meliaceae 12 0.481 s + Fabaceae (Caesalpinoids) 16 0.852 s + Fabaceae (Caesalpinoids) 16 0.489 * + Meliaceae 12 1.034 s + Annonaceae 38 0.495 **+ Annonaceae 38 1.143 ** + Myristicaceae 18 0.512 **+ Lauraceae 62 1.205 *** + Lauraceae 62 0.615 * +

i Null hypothesis: the average construction use value of species in a i Null hypothesis: the average commercial use value of a species in a given family is not significantly more or less than the predicted mean given family is not significantly more or less than the predicted mean construction FUV of families of equivalent size. commercial FUV for families of equivalent size. 2 Ho rejected at the following levels: s = P < 0.10, * P < 0.05, 2 Ho rejected at the following levels: s = P < 0.10, * = P < 0.05, **= p < 0.01,*** = P < 0.001. **= P < 0.01,*** = P < 0.001. 3 + and - indicate that FUVis respectively above/below the predicted 3 + and - indicate that commercial FUVis above/below the predicted value for a family of equivalent size. value for a family of equivalent size.

dividual-case predictions of more than 1.96 Balick 1989; Pinheiro and Balick 1987; Prance standard errors from the regression line are sig- et al. 1987) is confirmed here statistically, illus- nificantly different than expected from the num- trating the need for effective conservation of this ber of species in the family; i.e., the null hy- supremely useful family. In fact, species for spe- pothesis is rejected. cies, palms are between half and three orders of In Fig. 4, the significance of the departure of magnitude more important to mestizos than one each FUV from its predicted value is displayed third of all the other woody plant families ana- graphically, for families with ten or more woody lyzed. species. Tables 4 and 5 list families with FUV Second, most of the families that stand out as higher and lower than predicted, together with being exceptionally useful are important mostly significance levels. The null hypothesis is reject- because of their inmportanceas timber-both as ed at the 5% level for four, at the 10% level for subsistence uses for rural families (the main item eight, out of 20 families; there is also a strong in the construction category) and for commerce. suggestion that the residuals would have been Averaged across the 20 most speciose families, significant for several other families had the spe- construction uses contribute 47% to total use val- cies sample been larger. Table 5 lists the families ue, with 20% from commerce, 19% from edibles, in rank order, according to the fraction of their 8% from technology, and just 6% from medicin- FUV, that reflects construction (Table 6), com- als. The relative importance of these destructive mercial (Table 7), edible (Table 8), technological uses to mestizos parallels Piiiedo-Vasquez et al.'s (Table 9), and medicinal uses (Table 10). (1990) findings with ribereiios in northeast Peru. The tables highlight several interesting points. At both localities, timber harvesting has been First, the importance of forest palms to neotrop- practiced in a non-sustainable way, and at both ical peoples that is often remarked upon (e.g., localities establishment of firm land or resource 1993] PHILLIPS & GENTRY: QUANTITATIVE ETHNOBOTANY 1 27

TABLE 8. FAMILY USE VALUES (EDIBLE USES). TABLE 9. FAMILY USE VALUES (TECHNOLOGICAL USES). Number of Family Family species use value1,2,3 Number of Family Family species use value 12,3 Myristicaceae 18 0 s - Flacourtiaceae 10 0 - Lauraceae 62 0 *** - Nyctaginaceae 14 0 s - Melastomataceae 11 0 s - Lauraceae 62 0.011 **- Nyctaginaceae 14 0 * - Meliaceae 12 0.015 - Meliaceae 12 0.019 - Burseraceae 12 0.019 - Sapotaceae 25 0.025 s - Bignoniaceae 15 0.022 - Myristicaceae 18 0.046 - Euphorbiaceae 22 0.030 - Fabaceae (Papilionoids) 29 0.050 - Rubiaceae 16 0.037 - Fabaceae (Mimosoids) 38 0.061 - Apocynaceae 13 0.058 - Flacourtiaceae 10 0.075 - Fabaceae (Caesalpinoids) 16 0.073 - Apocynaceae 13 0.077 + Clusiaceae 11 0.080 - Annonaceae 38 0.083 + Chrysobalanaceae 13 0.088 - Fabaceae (Caesalpinoids) 16 0.088 + Fabaceae (Papilionoids) 29 0.095 - Bignoniaceae 15 0.089 + Melastomataceae 11 0.218 - Rubiaceae 16 0.098 + Annonaceae 38 0.254 - Euphorbiaceae 22 0.120 + Moraceae 54 0.593 * + Arecaceae 11 0.131 + Sapotaceae 25 0.673 * + Clusiaceae 11 0.136 + Fabaceae (Mimosoids) 38 0.727 ** + Moraceae 54 0.185 ** + Arecaceae 11 1.529* + Chrysobalanaceae 13 0.282 s + Burseraceae 12 0.446 ** + I Null hypothesis: the average edible use value of a species in a given family is not significantly more or less than the mean edible precicted I Null hypothesis: the average technological use value of a species in FUV of families of equivalent size. a given family is not significantly more or less than the predicted mean 2 H0 rejected at the following levels: s = P < * = < 0.10, P 0.05, technological FUV of families of equivalent size. ** = p < 0.01. 2 Ho rejected at the following levels: s = P < 0.10, * = P < 0.05, 3 + and - indicate that edible FUVis above/below the predicted value ** = p < 0.01. for a family of equivalent size. 3 + and - indicate that technological FUV is above/below the predicted value for a family of equivalent size. tenure is needed to encourage sustainable levels of extraction in the future. the least important, would be one of the most Third, however, some forest plant families re- important families by technique (a) (it is a large main important sources of non-timber forest family), and of near-average importance using products (NTFPs) to mestizos, in the edible, me- technique (b) or (c) (since the more abundant and dicinal and technological categories, as well as frequent species tend to be more useful). Ulti- leaf thatch in the construction category. More- mately, of course, the choice of technique de- over, the relative importance of the forest as a pends on the precise nature of the hypotheses source of medicine is probably underestimated being investigated. by our emphasis on the most speciose tree and large vine families, and possibly by the bias to- CONCLUSIONS wards male informants. NTFP collection in the We have shown how a relatively simple data- area is often relatively non-destructive (cf. Phil- processing technique can be applied to large eth- lips n.d.). nobotanical data sets, to generate a quantitative We should point out that our definition of FUVs use value index. The index helps to make a num- (i.e., the average of the family's set of UV, values, ber of important kinds of ethnobotanical ques- similar to Prance et al.'s definition ( 1987)) is only tions tractable, using statistical, hypothesis-test- one of several possible ways to measure family ing analyses. Here, we used use values to compare usefulness based on UV, values. Thus, familial the relative usefulness of different species and importance might also be defined as: (a) the sum families of woody plants in Amazonian Peru; in of the component UV, values; or, the mean of subsequent papers we will further explore the the UV, values weighted by each species' (b) rel- technique's potential for testing other hypotheses ative density, or (c) relative frequency. For ex- about patterns and processes in plant use. ample, Moraceae, whose FUV ranks as one of We hope that students and practicing ethno- 28 ECONOMICBOTANY [VOL.47

(1991). OP was funded by grants to the Tambopata Flora Study Group TABLE 10. FAMILY USE VALUES (MEDICINAL (1986), and by Cambridge University grants (1988); and by an NSF USES). Doctoral Dissertation Improvement Award (BSR-900105 1), a WWF-US Garden Club of America Award (1991), and a Conservation Intemational Number of Family Plants Program Grant (1990-1992), whilst a doctoral student at Wash- Family species use value"12'3 ington University. AG is a Pew Scholar in Conservation and Environ- ment and a Senior Curator at the Missouri Botanical Garden; fieldwork Nyctaginaceae 14 0 s - and herbarium work with the Proyecto Flora del Peru is also supported Flacourtiaceae 10 0 - by the Mellon Foundation and the MacArthur Foundation. Fabaceae (Mimosoids) 38 0 ** - Bignoniaceae 15 0 s - LITERATURE CITED Clusiaceae 11 0 - Annonaceae 38 0 ** - Adu-Tutu, M., K. Asanti-Appiah, D. Lieberman, J. B. Melastomataceae 11 0 - Hall, and M. Elvin-Lewis. 1979. Chewing stick Sapotaceae 25 0.008 * - usage in southern Ghana. Economic Botany 33: Meliaceae 12 0.009 - 320-328. Lauraceae 62 0.016 s - Anderson,A. B. 1990. Extractionand forest man- Chrysobalanaceae 13 0.038 - agementby ruralinhabitants in the Amazon estu- Fabaceae (Caesalpinoids) 16 0.054 + ary. Pages 65-85 in A. B. Anderson,ed., Alterna- Rubiaceae 16 0.089 + tives to deforestation.Columbia University Press, Fabaceae (Papilionoids) 29 0.099 + New York. Burseraceae 12 0.100 + 1991. Forestmanagement strategies by rural Apocynaceae 13 0.122 + inhabitantsin the Amazon estuary.Pages 351-360 Euphorbiaceae 22 0.149 + in A. G6mez-Pompa,T. C. Whitmore,and M. Had- Moraceae 54 0.207 * + ley, eds., Rain forestregeneration and management. Arecaceae 11 0.261 * + UNESCO, Paris. Myristicaceae 18 0.279 ** + , and D. A. Posey. 1989. Managementof a tropical scrub savanna by the GorotireKayap6 of I Null hypothesis: the average medicinal use value of a species in a given family is not significantly more or less than the predicted mean Brazil.Advances in Economic Botany 7:159-173. medicinal FUV of families of equivalent size. Balee, W. A., and A. Gely. 1989. Managed forest 2 Ho rejected at the following levels: s = P < 0.10; * = P < 0.05, succession in Amazonia: the Ka'apor case. Ad- ** = p < 0.01. vances in Economic Botany 7:129-158. 3 + and - indicate that medicinal FUV is above/below the predicted value for a family of equivalent size. Balick, M. J., and H. T. Beck. 1989. Useful palms of the world: a synoptic bibliography.Columbia University Press, New York. botanists will see our approach as part of the Boom, B. M. 1989. Use of plant resourcesby the challenge to develop a science that is both ca- Chacabo.Advances in Economic Botany 7:78-96. pable of testing specific hypotheses, and of pro- 1990. Useful plants of the PanareIndians of viding better quality information about how and the VenezuelanGuayana. Advances in Economic why people use plants. If ethnobotany can move Botany 8:57-76. But, P. P., S. Hu, and Y. CheungKong. 1980. Vas- from the scientific sidelines to center-stage, we cular plants used in Chinese medicine. Fitoterapia will believe it make a greater contribution to the 51:245-264. conservation and ethical use of biological and Bye, R. n.d. Ethnobotanyof the Mexican dry trop- cultural diversity. ical forests. In S. Bullockand H. A. Mooney, eds., TropicalDry Forests.Cambridge University Press, ACKNOWLEDGMENTS Cambridge,England. We are indebted to the residents of the community of La Torre and Coomes,0. n.d. Rain forest extraction,agroforestry, the surrounding area for generously sharing their knowledge of the Tam- and resourcedepletion: an environmentalhistory bopata forest with us. Michel Alexiades, Claudia Galvez-Durand Bes- from the northeasternPeruvian Amazon. Paper nard, Flor Chavez, Gavin Nicholson, and Paul Wilkin all helped with plant use data collection, and were present during some of our walks and presentedto the XVI InternationalCongress of the interviews. Camilo Diaz, Nestor Jaramillo, Percy Nufiez, and Rodolfo LatinAmerican Studies Association, April 6, 1991, Vasquez helped collect voucher specimens in the inventory plots. We Washington,DC. also owe special thanks to the Tambopata Reserve Society, Max Gunther, Duellman, W., and J. Koechlin von Stein. 1991. Re- Marcia Morrow, and the staffat the Explorer's Inn for invaluable logistical help. We thank Michel Alexiades, Walter Lewis, Eugene Shultz, Paul serva Cuzco Amaz6nico, Peru: biological investi- Wilkin, and two anonymous reviewers for their constructive advice on gation, conservation,and ecotourism. Occasional ways to improve the text, and for provoking us to confront some philo- paper, Museum of Natural History, University of sophical problems with our approach and analysis. Rogerio Castro helped Kansas, 142:1-38. Lawrence,KS. to translate the abstract, and Stanley Sawyer made several valuable sug- gestions on statistical methodology. We thank Bill Duellman and Linda Elvin-Lewis, M., J. B. Hall, M. Adu-Tutu, Y. Afful, Trueb for permission to reproduce the map from Duellman and Koechlin K. Asante-Appiah,and D. Lieberman. 1980. The 1993] PHILLIPS& GENTRY:QUANTITATIVE ETHNOBOTANY I 29

dental health of chewing-stick users of southern tive reserves. Acre, Brazil. Economic Botany 46: Ghana:preliminary findings. Journal of Preventa- 408-425. tive Dentistry 6:151-159. Kapur, S. K., A. K. Shahi, Y. K. Sarin, and D. E. ,and W. H. Lewis. n.d. New conceptsin med- Moerman. 1992. The medicinalflora of Majouri- ical and dental ethnobotany.In R. E. Schultesand Kirchi forests (Jammuand KashmirState), India. S. von Reis, eds., Ethnobotanytoday. Dioscorides Journalof Ethnopharmacology36:87-90. Press, Portland,OR. Lewis, W. H., and M. Elvin-Lewis. 1990. Obstetrical Erwin, T. 1984. TambopataReserved Zone, Madre use of the parasiticfungus Balansia cyperiby Am- de Dios, Peru: history and description of the re- azonianJivaro women. EconomicBotany 44:131- serve. Revista Peruanade Entomologia27:1-8. 133. Etkin, N. L. 1986. Multidisciplinaryperspectives in McNeely, J. A., K. R. Miller, W. V. Reid, R. A. Mit- the interpretationof plantsused in indigenousmed- termeier,and T. B. Werner. 1990. Conservingthe icine and diet. Pages2-29 in N. L. Etkin,ed., Plants world's biological diversity. World Bank, W.R.I., in indigenous medicine and diet. Redgrave Pub- I.U.C.N., C.I., and W.W.F., Washington,DC. lishing Company,Bedford Hills, NY. Martin, G. J. 1992. Searchingfor plants in peasant Friedman,J., Z. Yaniv, A. Dafni, and D. Palewitch. marketplaces.Pages 212-223 in M. J. Plotkin and 1986. A preliminaryclassification of the healing L. M. Famolare,eds., Sustainableharvesting and potential of medicinal plants, based on a rational marketing of rain forest products. Island Press, analysis of an ethnopharmacologicalfield survey Washington,DC. among Bedouinsin the Negev Desert, Israel.Jour- Moerman, D. E. 1979. Symbols and selectivity: a nal of Ethnopharmacology16:275-287. statisticalanalysis of Native Americanmedical eth- Gentry, A. H. 1988. Changes in plant community nobotany. Journal of Ethnopharmacology1:1 1 1- diversityand floristiccomposition on environmen- 119. tal and geographicalgradients. Annals of the Mis- 1991. The medicinal flora of native North souri BotanicalGarden 75:1-34. America:an analysis.Journal of Ethnopharmacol- Gilmartin,A. J. 1986. Experimentalsystematics to- ogy 31:1-42. day. Taxon 35:118-119. Padoch, C., and W. de Jong. 1990. Santa Rosa: the Hammersley,M. 1992. What's wrong with ethnog- impact of the forest productstrade on an Amazo- raphy?Routledge, London and New York. nian place and population.Advances in Economic Hiraoka, M. 1992. Caboclo and riberenoresource Botany 8:151-158. managementin Amazonia: a review. Pages 134- , and . 1992. Diversity, variation,and 157 in K. H. Redford and C. Padoch, eds., Con- change in riberenioagriculture. Pages 158-174 in servation of Neotropical forests. Columbia Uni- K. H. Redfordand C. Padoch, eds., Conservation versity Press, New York. of Neotropicalforests. Columbia University Press, Hubbell, S., and R. Foster. 1990. Structure,dynam- New York. ics, and stand equilibriumstatus of old-growthfor- Parker, E. P. 1989. A neglectedhuman resourcein est on BarroColorado Island. Pages 522-541 in A. Amazonia:the Amazon caboclo. Advancesin Eco- H. Gentry,ed., FourNeotropical forests. Yale Uni- nomic Botany 7:249-259. versity Press, New Haven, CT. Paz y MintoC., G., H. Balslev, R. Valencia R., and P. Hunn, E. 1982. The utilitarianfactor in folk biolog- Mena V. 1991. Lianasutilizadas por los indigenas ical classification.American Anthropologist 84:830- Siona-Secoya de la Amazonia del Ecuador. Re- 847. portes Tecnicos 1. Ecociencia,Quito, Ecuador. Iwu, M. M. 1986. Empiricalinvestigations of dietary Perez Salicrup,D. R. 1992. Evaluaci6nde la inten- plants used in Igbo ethnomedicine.Pages 131-150 sidad de uso de arbolesde la selva huimedaen dos in N. L. Etkin, ed., Plants in indigenousmedicine comunidadesde la regionde Los Tuxtlas,Veracruz. and diet. RedgravePublishing Company, Bedford Biology thesis. U.N.A.M., Mexico. Hills, NY. Phillips, 0. L. B. n.d. The potential for harvesting Johns, T., J. 0. Kokwaro,and E. K. Kimanani. 1990. fruit in tropical rainforests:new data from Ama- Herbal remedies of the Luo of Siaya District, Ke- zonian Peru. Biodiversityand Conservation1. nya: establishingquantitative criteria for consen- , and A. H. Gentry. 1993. The useful plants sus. Economic Botany 44:369-381. of Tambopata,Peru: II. Additionalhypothesis test- , and E. K. Kimanani. 1991. Test of a chemical ing in quantitativeethnobotany. Economic Botany ecological model of the origins of medicinal plant 47:33-43. use. Ethnobotany3: 1-10. , C. Reynel, and A. Gentry. n.d. Los tipos de Johnson, A. W. 1978. Quantificationin culturalan- bosqueen Tambopata.In C. Reyneland A. Gentry, thropology.Stanford University Press, Stanford, CA. eds., Florade Tambopata.Missouri Botanical Gar- Kainer, K. A., and M. L. Duryea. 1992. Tapping den, St. Louis, MO. women's knowledge:plant resourceuse in extrac- Piftedo-Vasquez,M., D. Zarin, P. Jipp, and J. Chota- 30 ECONOMICBOTANY [VOL. 47

Inuma. 1990. Use-values of tree species in a com- Smith, J. K., and L. Heshesius. 1986. Closingdown munal forest reserve in northeastPeru. Conserva- the conversation:the end of the quantitative-qual- tion Biology 4:405-416. itative debate among educational inquirers.Edu- Pinheiro,C. UrbanoP., and M. J. Balick. 1987. Bra- cational Researcher15:4-12. zilian palms: notes on their uses and vernacular Sneath, P. H. 1988. The phenetic and cladistic ap- names, compiled and translatedfrom Pio Correa's proaches. Pages 252-273 in D. Hawksworth,ed., "Dicionariodas PlantasUteis do Brasile das Ex6- Prospectsin systematics.Academic Press, London. ticas Cultivadas,"with updatednomenclature and Stevens, P. F. 1991. Characterstates, morphological added illustrations.Contributions from the New variation,and phylogeneticanalysis: a review. Sys- York BotanicalGarden, Vol 17. New York Botan- tematic Botany 16:553-583. ical Garden,Bronx, NY. Toledo, V. M., A. I. Batis, R. Bacerra,E. Martinez, Popper,K. R. 1963. Conjectureand refutations:the and C. H. Ramos. 1992. Productsfrom the trop- growth of scientific knowledge.Harper and Row, ical rain forests of Mexico: an ethnoecological New York. approach. In M. Plotkin and L. Famolare, eds., Prance, G. T. 1991. What is ethnobotany today? Non-wood productsfrom tropical rain forests. Con- Journalof Ethnopharmacology32:209-216. servationInternational, Washington, DC. , W. Balee, B. M. Boom, and R. L. Carneiro. Trotter, R. T., and M. H. Logan. 1986. Informant 1987. Quantitativeethnobotany and the case for consensus:a new approachfor identifyingpoten- conservation in Amazonia. ConservationBiology tially effective medicinal plants. Pages 91-112 in 1:296-310. N. L. Etkin,ed., Plantsin indigenousmedicine and Preston, F. W. 1948. The commonness, and rarity, diet. RedgravePublishing Company. Bedford Hill, of species. Ecology 29:254-283. NY. 1962. The canonical distribution of com- Turner, N. J. 1988. "The importance of a rose": monness and rarity:part 1 and 2. Ecology43:185- evaluating the cultural significance of plants in 215, 410-432. Thompson and Lilloet Interior Salish. American Reynel, C., and A. H. Gentry. n.d. Flora de Tam- Anthropologist90:272-290. bopata.Missouri Botanical Garden, St. Louis, MO. Unruh,J., and J. Alcorn. 1988. Relative dominance Salick, J. 1992. Amuesha forest use and manage- of the usefulcomponent in youngmanaged fallows. ment: an integrationof indigenoususe and natural Advances in Economic Botany 5:47-52. forest management.Pages 305-332 in K. H. Red- Wiley, E. 0. 1981. Phylogenetics.John Wiley, New ford and C. Padoch, eds., Conservationof Neo- York. tropical forests. Columbia University Press, New York.

APPENDIX 1. MESTIZO USES OF TAMBOPATA WOODY PLANTS.

Number of Infor- Usel (defined by informants) Spp.2 mants3 C: Sawnwoodfor floors, walls 223 8 0: Commercein unfinishedconstruction products (e.g., planks,posts; wood for parquet,plywood) 171 10 E: Raw fruit eaten 144 16 C: Roundwood for beams in house construction 122 5 E: Hunterswait close to fruitingtree for game 49 5 C: Roundwood for house posts (postes,horcones) 42 6 T: Fuelwood 35 3 M: Remedy for brokenbones 17 4 C: Sawnwoodfor furniture 16 3 M: Remedy for hernias 13 3 C: Thatch for housing and temporarybuildings 12 6 C: Sawnwoodfor house posts (postes,horcones) 12 5 M: Remedy for rheumatismand/or muscle-aches 11 2 0: Commercein edibles (fruits,nuts, palm hearts) 10 9 M: Remedy for diarrhea(chronic & acute not always distinguished) 10 5 M: Remedy for fever (sometimes specifiedas malaria) 10 4 T: Resin burnt for light when keroseneunaffordable or unavailable 10 3 1993] PHILLIPS& GENTRY:QUANTITATIVE ETHNOBOTANY I 31

APPENDIX 1. CONTINUED.

Number of Infor- Use' (defined by informants) Spp.2 mants3 E: Cooked seeds eaten 9 6 0: Commercein finishedconstruction products (e.g., fumiture,canoes) 9 5 C: Canoes made from hollowed trunk 9 4 M: Remedy for swellingsand bruises 9 3 0: Commercein technology/craftitems (e.g., necklaces,charcoal, cured skins) 9 3 M: Remedy for patico (=sore spots in childrens'mouths, presumablya gum infection) 9 2 E: Soaked fruit eaten 8 11 E: Palm heart eaten raw/cooked 7 10 E: Beetle larvae (surf)from rottingtrunk eaten raw/cooked 7 7 E: Edible oil extractedfrom fruit 7 7 0: Commercein medicines 7 6 C: Roundwood/split-trunkfor fence-postconstruction 7 5 E: Immatureseed eaten raw 7 5 M: Remedy for bronchitis 7 2 E: Matureseed eaten raw 6 8 M: Remedy for cutaneousleishmaniasis 6 5 M: Remedy for colds (resfrio) 6 4 M: Preventshairloss 6 4 M: Generaltonic (sometimes specifiedas blood or digestive system) 6 3 E: Fruitjuice (refrescos)extracted 6 2 T: Barkused for tanning skins 6 1 C: Split-trunkfor partitionwall construction 5 7 M: Remedy for liver-complaints(possibly includinghepatitis) 5 5 T: Rope for carryingheavy loads and/or for tying 5 5 T: Barkashes mixed with clay to make ceramics 5 1 T: Resin used as glue, mixed with clay to make ceramics 5 1 C: Split-trunkfor outside house-wallconstruction 4 10 T: Bait for fishing 4 6 C: Sawnwoodfor canoe construction 4 6 T: Wood used to make tool-handles(axes, hammers,brushes, etc.) 4 3 E: Tree producesfruit eaten by pigs 4 3 T: Canoe paddles 4 2 M: Headacheremedy 4 2 T: Press for sugar-caneprocessing (trapiche) 4 2 C: Split-trunkfor floor construction 3 6 C: Split-trunkfor roofingand floor supports 3 5 M: Topical antibiotic (e.g., treatingfungal skin infections) 3 4 M: Remedy for coughs 3 3 T: Seeds used for necklaces 3 2 T: Aerial traps made to protectlivestock from vampire bats 3 1 M: Wound-healing 2 6 E: Barkused for flavouringalcohol 2 4 M: Remedy for kidney-complaints 2 4 M: Aphrodisiac(enhances sexual desire and/or performance) 2 3 T: Dyes 2 3 M: Remedy for cancer(intemal cancersonly) 2 3 T: Latex used to waterproofcloth and canvas 2 3 C: Split-trunkfor roof-guttersconstruction 2 3 E: Beetle larvae (surf)from seed eaten raw/cooked 2 2 T: Buttress-rootsections used as boards for washinggold (bateas) 2 2 E: Fruit fermentedfor alcoholic drink 2 2 E: Latex drinkable 2 2 M: Local anaesthetic 2 2 32 ECONOMICBOTANY [VOL. 47

APPENDIX 1. CONTINUED.

Number of Infor- Usel (defined by informants) Spp.2 mants3

M: Purge 2 2 M: Love charm 2 1 T: Mortar to remove chaff from rice 2 1 T: Soap 2 1 M: Remedy for varicose veins 2 1 M: Anthelmintic 1 4 C: Trunk for foot-bridge construction 1 4 T: Charcoal 1 3 C: Temporary canoes (tarapotos) 1 3 M: Remedy for anaemia 1 2 M: Candiru removal (causes the fish to be excreted from orifice) 1 2 M: Remedy for eye cataracts 1 2 T: Fruit used as a football 1 2 T: Leaves used as hand-fans 1 2 E: Seeds processed to make chocolate 1 2 T: Spear for fishing 1 2 M: Remedy for witchcraft (counteracts dano) 1 2 M: Antiseptic (e.g., treats pustulent abscesses) 1 1 T: Arrow strings for childrens' toys 1 1 M: Remedy for asthma 1 1 E: Cooked fruit eaten 1 1 E: Edible fungi collected from fallen/cut tree 1 1 E: Edible macaw chicks taken from nest in tree 1 1 M: Remedy for carnosidad (loss of vision due to growths over the eye, including cataracts) 1 1 T: Fibre used in weaving 1 1 T: Fish poison 1 1 E: Fruit made into jam 1 1 T: Garden mulch 1 1 T: Leaves used as brooms 1 1 T: Liana cut for drinking water 1 1 T: Sections cut to make living fences 1 1 T: Ornamental plants 1 1 E: Palm heart fermented for alcoholic drink 1 1 M: Penis extender 1 1 T: Petiole used as leverage to adjust tension in canoe-making 1 1 M: Prevents tuberculosis 1 1 C: Rafts 1 1 T: Rat poison 1 1 T: Resin used as wood sealant (e.g., for canoe caulking) 1 1 E: Seed for cooking oil 1 1 T: Spathes or seedling leaves used for carrying and wrapping food 1 1 T: Spathes used as temporary drinking vessels 1 1 C: Split-trunk for shelving 1 1 C: Split-trunk for paths 1 1 C: Split-trunk for temporary building 1 1 T: Stem cut and decorated at Carnival time 1 1

l All uses are subsistence uses except when specified as commerical: C = Construction uses; E = Edible uses; 0 = Commercial uses; M = Medicinal (and "Magic") uses, including both remedial and preventative uses; T = Technological and Craft uses. 2 All tree and liana species -10 cm d.b.h. in 6.2 ha of forest plots at Tambopata are included, together with 35 other non-plot woody species. 3 We interviewed 29 mestizo informants (aged approximately from 5 to 67, mean age = 34 years, median age = 35 years, including 20 men, three women, and six children and youths).