Caribbean Journal of Science, Vol. 36, No. 1-2, 19-30, 2000 Copyright 2000 College of Arts and Sciences University of Puerto Rico, Mayagüez

Wood Identification in Historic Sites: Inferences for Colonial Trade and Modification of Vegetation on Barbuda

DAVID R. WATTERS1 AND REGIS B. MILLER2

1Section of Anthropology, Carnegie Museum of Natural History, 5800 Baum Boulevard, Pittsburgh, Pennsylvania 15206-3706. [email protected] 2USDA Forest Service, Forestry Products Laboratory, One Gifford Pinchot Drive, Madison, Wisconsin 53705-2398. [email protected]

ABSTRACT.–We report on four wood species identified from five samples taken from timbers in four historic buildings on Barbuda. Two samples were reported previously and three are new identifications. Fortuitous retention of splinter remnants from one of the older samples allowed us to verify the accuracy of its original identification. None of the species are native to Barbuda. Species of the white pine group (Pinus spp.) and the white oak group (Quercus spp.) grow in North America, greenheart (Chlorocurdium rodiei) is native to Guyana and Suriname in South America, and bulletwood (Manilkara spp.) is native to the Neo­ tropics,includingtheAntilles.Thesedatacorroborateempiricallytheimportationoftimberproductsintothe colonial British West Indies, a trade that had been studied mainly by historians working from documents. Our analysis of the current body of knowledge about human-induced modification of Barbuda‘s native vegetation concludes that the vegetation has been modified more profoundly by the island’s historic and modem inhabitants, whose occupation is brief compared to the preceding 3,500-year span of Amerindian occupation.

INTRODUCTION note sent with the samples, were included in Record’s report. They read, in part: The first scientific analysis of wood The building still contains some old, decay­ samples obtained from a colonial-era build­ ing woodwork-two large beams and some ing on Barbuda was conducted to address a lintels built in over doorways-all of the historical enigma. At issue was who con­ same species. . . .Barbuda has never pro­ duced any large enough to have yielded structed the Martello Tower, the Spanish, this timber, so it must have come from out- as was popularly supposed by Barbudans, side…(Record, 1945:2). or the English, as John Beard thought more likely. In 1944, Beard, of the Forest Depart­ Such details provide insight into the Mar­ ment in Trinidad, submitted pieces of tello Tower’s structural integrity as ob­ wood from a beam in the Martello Tower to served in 1944. A decade after Record’s ar­ Samuel J. Record, Yale University School of ticle, Wagenaar Hummelinck (1956) wrote Forestry. Record (1945) identified the wood of the architectural features he observed asDemeraragreenheart, Ocotearodiei (M. R. during his July 1955 visit, including the still Schomb.) Mez., which in 1991 was changed existing floorbeams. to Chlorocurdium rodiei (M. R. Schomb.) The Martello Tower and its adjacent gun Rohwer et al. platform are known locally as River Fort Record’s report is brief but valuable. By (BA-H2), one of ten historical archaeology linking greenheart with Demerara, part of sites identified on Barbuda during a study the colony then known as British Guiana conducted in 1978-1979 (Watters, 1980) (now Guyana), Record supported Beard’s (Figs. 1 and 2). Most of these sites date to belief that the Martello Tower was of En­ the Codrington family’s leasehold, from the glish construction. Comments by Beard 1680s until 1870, when Barbuda was used about the structure, which he wrote in a as a ”provisioning island” to supply staples 19 20 D. R. WATERS AND R. B. MILLER

main in other sites as well. Watters and Miller surmised that the timbers used in Barbuda’s colonial-era buildings could be from imported woods. This article presents the results of our in­ terdisciplinary research on the wood samples from Barbuda’s historic sites and discusses the inferences those findings have for importation of timber products. We augment this research by drawing to­ gether data from other disciplines to ex­ plore the contexts of colonial structures and to summarize the evidence for the alter­ ation of Barbuda’s native vegetation during the prehistoric and historic eras. Such an interdisciplinary approach to research has been used sparingly by Caribbean histori­ calarchaeologists. Deagan’s (1987,1995) re- search on the archaeology of Spanish colo­ nization in the Caribbean is a notable exception. FIG. 1. Four of ten historical archaeology sites on This is the second article to argue the Barbuda yielded wood samples: Highland House (BA- merit of using an interdisciplinary ap­ H1); River Fort (BA-H2); SpanishPoint (BA-H3); Gun proach for studying Caribbean biogeo­ Shop Cliff (BA-H4). graphical and paleoecological issues, and of explicitly incorporating archaeological (e.g., hides, charcoal, wood, lime, rope, and data in such research. Materials from his­ livestock) to the Codrington’s sugar estates toric sites are the focus of this paper; the onAntigua (Tweedy, 1981). Two sites, High- first article (Watters et al., 1999) incorpo­ land House (BA-H1) and Codrington Castle rates specimens from prehistoric sites. (BA-H7), have been reported in detail (Wat­ ters and Nicholson, 1982; Watters, 1997). MATERIALS AND METHODS The Demerara attribution for the Mar­ tello Tower’s greenheart samples bears di­ Two samples were obtained from in situ rectly on the topic of the importation of structural timbers, a floor joist within the timber products into the British West Indies MartelloToweratRiverFort(BA-H2) anda during the colonial era. The investigation of window lintel in the Gun Shop Cliff build­ trade has been largely within the scope of ing (BA-H4). The third sample came from a historians working with documentary evi­ displaced beam lying in the interior of the dence. Identifying wood samples collected Spanish Point "Castle" structure (BA-H3). directly from historic structures provides Samples consisted of large splinters pried the means to investigate the timber trade and split from the individual structural empirically. members atthesesites. Table 1 summarizes Record’s (1945) intriguing report was the collection data and Figures 3-5 document impetus for the additional collection of the sampled timbers. The specimens were wood samples from three historic sites (BA- submitted to the USDA ForestService, For­ H2, BA-H3, BA-H4) by Watters in 1983 and est Products Laboratory (FPL) in October 1984 (Fig. 1). Securing an additional sample 1984. from the Martello Tower would confirm or In 1970, FPL obtained the Samuel James refute Record’s identification. Watters had Record wood collection and associated files observed timbers at the Martello Tower from Yale University. Miller searched the and lintels at the Gun Shop Cliff site in collection and files while writing this paper 1979, and suspected that wood might re- in hopes of finding, evidence of Record’s BARBUDA'S HISTORIC WOODS 21

FIG. 2. The Martello Tower and adjoining gun platform at River Fort (BA-H2). FIG. 3. Sampled in situ joist from the interior of the Martello Tower at River Fort (BA-H2). FIG. 4. Displaced timber sampled at the Spanish Point structure (BA-H3). FIG. 5. Sampled in situ lintel from the southwest window at Gun Shop Cliff (BA-H4). 22 D. R. WATTERS AND R. B. MILLER

TABLE 1. Sample collection data and wood identifications.

Site Sample Datecollected Identification Present samples BA-H2 insitu May30,1984 rodiei River Fort floorjoist (greenheart) BA-H3 displaced January23,1983 Pinus(speciesof Spanish Point beam white pine group) BA-H4 insitu January16,1983 Quercus (speciesof Gun Shop Cliff windowlintel white oak group) Previous samples BA-H2 beam 1944 Chlorocardium rodiei River Fort (byBeard) (greenheart) BA-H1 timber 1958 Manikara(speciesof HighlandHouse (byHarris) bulletwood) contact with John Beard. Miller found the noted that the transverse surface of BA-H4 original letters from Beard and Record and was distinctly ring porous with wide rays three large splinters in their original brown and an abundance of tyloses in the vessels envelope. (pores). Examination of the tangential sec­ In November 1984, Miller examined and tion distinctly revealed two sizes of rays; identified the three wood samples (Table 1) uniseriate rays and rays over 10 cells wide. following the standard wood identification These features, coupled with a dendritic procedures used at FPL. The transverse latewood vessel pattern and thin-walled surface was smoothed using a hand held latewood vessels that are not distinc with microtome knife and examined using a 14X the 14X hand lens, confirmed the identifi­ hand lens. Miller then cut thin (20-50 µm) cation as a species in the white oak group radial and tangential sections using a hand (subgenus Lepidobalunus, also known as held microtome knife. Sections were placed Leucobalanus of the genus Quercus. on a microscope slide, followed by a few BA-H3wasidentifiedas a species in the drops of a 50:50 solution of glycerine/ white pine group (subgenus Strobus, sec­ ethanol, and a cover slip. The slide was tion Strobus, subsection Strobi of the genus placed on a hot plate to boil off the ethanol, Pinus). Axial and horizontal resin ducts and allowing the glycerine to replace the air in ray tracheids were present. The ray trac­ the section. The slide was examined under heids did not contain denticulations (out- a light microscope at approximately 100 to growths from the horizontal walls typical 600 times magnification. To identify the of the red and yellow pines). The cross-field Watters samples and the recently rediscov­ pitting was large and window-like and ered Beard samples, Miller used textbooks only 1-2 pits were observed per cross-field (Panshin and deZeeuw, 1980; Hoadley, BA-H2 was identified as greenheart 1990), dichotomus keys and descriptions (Chlorocardium rodiei). Although greenheart (Hess 1946, 1948 Kribs, 1968; Kukachka, lacks oil cells and septate fibers typical of 1981), computer-assisted wood identifica­ the family , the large vessel-ray tion programs (Miller et al., 1987 [IDENT8], pits, typical greenish brown heartwood, Ilic 1987 [CSIROID]), and the wood and mi­ unilaterally paratracheal parenchyma, and croscope slide collections at FPL (Miller, oil droplets in the ray cells were present. In 1999). addition the sample was diffuse porous with nearly exclusively solitary medium- RESULTS sized pores. Wood identifications The three large splinters from Beard (Fig. 6) that Record (1945) had identified as Table 1 lists the identifications of the greenheart (Chlorocurdium rodiei) were ex­ three samples submitted in 1984. Miller amined. Miller confirmed the identification BARBUDA’S HISTORIC WOODS 23

FIG. 6. Three greenheart wood samples from the Mar­ tello Tower submitted by John Beard to Samuel J. Re- cord in 1944. Photograph courtesy of USDA Forest Ser­ vice, Forest Products Laboratory.

FIG. 7. A page in John Beard’s letter with his plan- view sketch of the Martello Tower and gun platform (”redoubt”). Photograph courtesy of USDA Forest Ser­ vice, Forest Products Laboratory. as greenheart and compared each to tification to genus. Harris notes that only sample BA-H2 and to vouchered samples one species, M. bidentata, is present in the from the FPL collections. Although all the Lesser Antilles (see also Pennington, 1990). samples were similar, Miller could not de­ Harris (personal communication, 1999) termine that the specimens obtained by does not recall which building at Highland Beard and by Watters were from the same House contained the timber or whether it timber. Record (1945:2) had reported ver­ was in situ or dislodged. batim the portion of Beard’s original letter concerning the Martello Tower, but did not reproduce a planview drawing of the Mar­ DISCUSSION tello Tower and gun platform (the ”re- Geographic distributions of identified woods doubt”) that Beard had incorporated in his letter (Fig. 7). Knowing the natural geographic distri­ WoodatHighlandHouse bution of woods and comparing them to wooden artifacts found in archaeological A reference to a fifth historic wood sites can be very helpful in understanding sample, identified from another site, was the use of wood by past societies and cul­ discovered during a review of the scholarly tures. Such data can provide information literature about Barbuda. Geographer regarding local vegetation, trade routes, David Harris obtained a specimen, in 1958, and foreign contacts. Accurate identifica­ from a construction timber in a building at tion of the wood is imperative for any hy­ Highland House (BA-H1) in the Highlands pothesis, interpretation, or inference de- (Fig. 1). He described it as ”. . . a very rived therefrom. heavy, durable wood prized for build­ In the United States, there are over 10 ing. . .” (Harris, 1965:110). Harris reported species in the white oak group (Little, the specimen as bulletwood (Manilkara), 1979); worldwide there are many more. The and he specifically credits the Scientific Of­ various species cannot be separated based ficer, Timber Development Association, solely on wood anatomy. The most com­ Ltd. in London, with having made the iden­ mon and commercially important species 24 D. R. WATERS AND R. B. MILLER in North America is white oak (Quercus bulletwood is used for heavy construction alba), which ranges throughout the eastern and ship frames where resistance to decay United States (Little, 1979). Because it is fungi and termites is important. very strong and decay resistant, white oak No species of Manilkara is listed by Fran­ is commonly used for shipbuilding and for cis et al. (1994) as having been collected on tight cooperage, which were two commodi­ Antigua or Barbuda. However, Francis ties in high demand during the colonial era (personal communication, 1999) believes (Alden, 1995). that some species of Manilkara grew on An­ The species of the white pine group (sub- tigua, probably in the upland habitat, al­ genus Strobus, section Strobus, subsection though he did not find it nor did his talks Strobi) are native to North America (includ­ with residents find anyone who knew of it. ing Mexico) and north temperate regions of He regards Barbuda as too dry to grow M. Asia, but not to Europe. In North America, bidentata, but does not rule of the possibility the two commercially important species are that other species of Manilkara might have western white pine (Pinus monticola) and grown there. eastern white pine (Pinus strobus) (Little, 1979). The former grows mainly in the Sources for Barbuda’s imported timbers mountains of the Pacific Northwest, while the latter grows mostly in the Lake States, One of the three wood samples analyzed Northeast, and the Appalachian Moun­ for this article documents with certainty the tains. In the historic era, eastern white pine movement of timber to Barbuda from was prized highly for furniture, general South America. Given the very restricted construction, and ship masts because of its natural geographic distribution of green- large size, availability, and ease of working heart, the Martello Tower timber must have (Alden, 1997). originated in South America, with the more Greenheart is native to Guyana and a likely source being the Demerara sector of small part of Suriname. A hard, heavy the British Guiana colony. wood, it is very resistant to decay fungi, The second wood is a species of the white termites, and marine borers. It is used in pine group. North America is the more marine and ship construction, docks, vats, logical source for the Barbuda specimen, and other places where durability is re­ based on proximity and the documented quired (Gerald et al., 1996). In Guyana, trade in wood products. Since in the colo­ wooden buildings are built with green- nial era commercial wood products were heart. shipped through the seaports on the Atlan­ Bulletwood or balata is the common tic coast, it follows that the trees being name for any species of Manilkara, a large logged would have been the species that genus in the Sapotaceae. According to Pen­ grew in the eastern part of that continent; nington (1990), Manilkara is a tropical or that is, eastern white pine (Pinus strobus). subtropical genus of shrubs to large trees Thus, we suppose that the BA-H3 sample is growing in Africa, Asia, and America. eastern white pine, a commonly available There are 30 species in the Neotropics, wood that was the species of choice for eight of which currently grow in the West many uses. Indies. One species, M. bidentata (A. DC.) The source of the third wood sample, a Chev., grows in the West Indies, Panama, species of the white oak group, is more and northern South America. Within the problematic because the natural geographic West Indies, it grows today in the Domini- distribution of white oaks includes North can Republic, Puerto Rico, Virgin Islands, America and Europe, and the various spec­ and Windward Islands, and occurs in semi- cies cannot be separated solely on wood evergreen forests (Pennington, 1990). This anatomy. Neither continent can be ruled species has been exploited heavily during out as a source for the white oak lintel in this century as a source of commercial the Gun Shop Cliff building. If the source balata, a product of the latex, and for its was North America, as seems probable very durable timber. As with greenheart, based again on trade records and proxim- BARBUDA’S HISTORIC WOODS 25 ity, then the BA-H4 wood is most likely One construction timber remained in 1958, eastern white oak (Quercus alba). from which Harris obtained his bulletwood A Neotropical origin for the Highland sample; but neither lintels nor timbers were House bulletwood timber reported by Har­ seen by Watters in the course of five visits ris is in accord with the geographic distri­ to Highland House since 1978. bution of Manilkara, but a more definitive Beard’s contention that the Martello source remains problematic. Harris’s (1965: Tower was of English (more properly Brit­ 110) suggestion that this timber may be M. ish) construction was correct. Documents bidentata is derived from that species’ oc­ indicate River Fort’s Martello Tower was currence within the Antilles; today the built in the 1740s when Admiral Knowles, sources nearest to Barbuda are the Virgin based at the English Harbour naval facility Islands and Windward Islands. If the speci­ on Antigua, constructed a system of forti­ men was M. bidentata, as Harris suggests fications on Barbuda (Watters, 1997:278- (although the Scientific Officer identified it 279). One joist (the sample) and two lintels only to genus), the timber must have been (over doorways into magazines) were in imported because that species is unlikely place in 1984; one fallen joist rested on the ever to have grown on Barbuda. Presently, floor. If Beard’s statement (Record, 1945) is eight species of Manilkara grow in the An­ correct, that the beams and lintels were of tilles, and Francis believes that at least one the same species, then the wood in the grew formerly on Antigua, where no speci­ magazine lintels should be greenheart as mens have been collected in modern times. well. Thus, Antigua cannot be ruled out as a can­ The Spanish Point structure (BA-H3), didate for the source of Barbuda’s High- sometimes referred to as a “Castle,” has de­ land House timber, especially in view of teriorated so extensively that its original the Codrington family’s connection to both islands (Watters, 1980, 1997). height cannot be determined. It is circular in planview and its date of construction is uncertain. No building is shown at this lo- Preservation and chronology of the sampled cation on the oldest maps of Barbuda historic sites (dated between 1750 and 1785). The first Barbuda’s historic structures have British Admiralty chart of the island, pro­ steadily deteriorated since being aban­ duced in 1814, depicts a rectangular, not doned, with mainly the durable materials circular building. A visitor in 1850 confirms (stone and brick masonry) being preserved. it was circular and indicates it probably Structural elements of wood were pre- was still in use at that time since he found served in only four of ten historic sites (Fig. it locked (Watters, 1980:128). The wood 1). It is unlikely that old wood will be sample from Spanish Point came from a found in other structures on Barbuda, with beam dislodged from the building’s stone- the possible exceptions of the unstudied work (Fig. 4), and therefore has less reliable abandoned wells and disused animal wa­ provenience than the two in situ samples tering pens that are dispersed across the from the River Fort and Gun Shop Cliff landscape. Identifiable charred wood may sites. be preserved below ground, in subsurface The Gun Shop Cliff building (BA-H4) re­ contexts at both historic and prehistoric lates to a brief episode in the 1890s of phos­ sites. phate mining in a sea cave in the northeast Extant structures at Highland House in­ escarpment of the Highlands, an enterprise clude seven buildings, a cistern, and a wa­ that took place 20 years after the Codring­ ter catchment, all built of stone. There are tons surrendered their lease (Anonymous, no remnants of two timber buildings re­ 1928; Martin-Kaye, 1959:38-40; Watters, portedly built about 1750. The Codringtons 1980). Although long ago said to be a ”ru­ built Highland House as a retreat; historic ined building’’ (Earle, 1921:1), the standing artifacts indicate it was used from the 1720s stone walls are mainly intact and four until 1800 (Watters and Nicholson, 1982). wooden window lintels were observed in 26 D. R. WATTERS AND R. B. MILLER

1978, when Watters first visited the site. Spanish Point, Gun Shop Cliff, and High- Two lintels remained in place in 1992. land House. There are alternatives, how- The four historic sites yielding the five ever. The greenheart joist at the Martello wood samples span about 175 years, dur­ Tower may have come from lumber stock ing and after the Codrington leasehold, at the English Harbour naval facility that from as early as initial construction at Admiral Knowles commanded. Other pos­ Highland House in the 1720s, to the Gun sibilities include illicit trade with the Shop Cliff building from the phosphate nearby Dutch islands or war booty result­ mining venture of the 1890s. The timber ing from attacks launched against the with the most secure date for its time of use French islands. Shipwrecks provided an al­ in construction is the in situ joist from the ternative source for wood, and since white River Fort’s Martello Tower built by Admi­ oak, white pine, and greenheart were used ral Knowles in the 1740s. in ship construction, the possibility exists that the foreign timbers on Barbuda came Colonial trade in timber products from such wrecks. Nicholson (1994:61-62) identifies 145 ships that foundered on Bar­ Throughout the historic period, the Brit­ buda’s dangerous reefs, and the salvaging ish North American colonies actively of stranded vessels is well documented in traded with the British West Indies in a va­ the records of Barbuda (Tweedy, 1981:148- riety of timber products, shipped from 160). ports along the eastern seaboard to the Barbuda exported wood products at the sugar islands (Pares, 1956:37-91). same time it imported foreign woods. Tweedy (1981:128-130) documents the All the sugar colonies needed vast quantities shipment of wood to Antigua; to the Cod­ of North American lumber for making hogs- rington sugar estates, which received fire- heads and barrels to ship their sugar and rum. . . .Hence the reliance on North America wood, charcoal, timbers, shingles, rafters, for supplying staves and hoops and even the joists, handles for tools, spokes and shafts lumber, shingles and frames for building for carts, staves, hoops, and wheelbarrows; houses.(Carrington,1996:163). and to the naval facility at English Harbour that purchased periodically firewood and The volume of trade is impressive. During timbers. Sale of wood products was a rela­ a 3-year period (1771-1773), four major tively minor component of the accounts for wood products shipped by the American Barbuda, but it is noteworthy that the is- colonies totaled (in millions) nearly 76.7 land’s greatest profit in wood products co­ board feet of lumber, 59.6 shingles, 58 incided with the American Revolution, staves, and 4.7 hoops (Carrington, 1988: when timber imports from the normal sup- Table 30). The American Revolution ad­ pliers were interdicted (Tweedy, 1981:129). versely affected the West Indies sugar in­ dustry because Great Britain prohibited the importation of timber products from the re- BARBUDA’S VEGETATION, PAST AND belling North American colonies. PRESENT Although Barbuda was not a sugar- producing island, it was an integral part of Several disciplines have contributed to the extensive holdings of the Codrington our knowledge of Barbuda’s past and pres­ family in the West Indies, which included ent vegetation. Data in this section are com­ sugar plantations on Antigua and St. Kitts piled from studies by botanists, geogra­ (Watters, 1997). Importation of a wide phers, anthropologists, and historians. range of products for the sugar estates and The Europeans and Africans who settled their subsequent transshipment to Barbuda Barbuda in the 17th century did not en- is verified through historic documents counter a pristine environment because the (Tweedy, 1981). Thus, trade probably ac­ island’s vegetation had been affected by counts for the foreign woods represented in some 3,500 years of Amerindian prehistory. the construction materials at River Fort, The presence of Archaic Age Amerindians BARBUDA’SHISTORICWOODS 27 by about 4000 B.P. is confirmed by radio- occupation in the skilled worker category of carbon dates on shell artifacts, but a more the Barbudm slavelists (Tweedy, 1981:184). intensive occupation (based on numbers Tall trees are scarce on Barbuda. Occa­ and sizes of archaeological sites) took place sionally an individual of respectable height during the Ceramic Age, especially from emerges from the canopy on the marginal 1500to500B.P.(ca.A.D.500-1500)(Watters plain, where its roots have tapped a reliable et al., 1992). water source, but clusters of taller trees are Archaeobotanical data compiled from generally restricted to moisture-retaining prehistoric sites in the Caribbean region by sinkholes, as at Darby Sink Cave (Harris, Newsom’s (1993) pioneering research are 1965:38). enhancing our knowledge of the paleoen­ Although shallow soils and low rainfall vironments encountered by Amerindians retard growth, the paucity of large on other islands. For Barbuda, we lack di­ trees on Barbuda usually is attributed to rect evidence of past vegetation in the logging in the past. ”During this period, forms of wood charcoal, charred , or virtually every tree large enough to use for other floral components from archaeologi­ construction materials was felled” (Francis cal contexts. Swidden (slash and burn) cul­ et al., 1994:l). Historical records document tivation of manioc, the hallmark of Ceramic a time when logging was intensive. Dennis Age agricultural practices in the Caribbean Reynolds, the Codrington family’s man­ (Petersen, 1997), would have impacted veg­ ager on Barbuda, claimed that ”. . .good etation in localized areas, notably in the vi­ timber trees were scarce because Colonel cinity of the Highlands where Amerindian Martin had cut most of them down.. .” villages were concentrated (Watters et al., (Tweedy, 1981:130). Reynolds‘ letter, writ- 1992; Harris, 1965:91). Amerindians used ten in 1787, refers to Samuel Martin who, wood for tools, fishtraps, firewood, dwell­ with William Byam, subleased Barbuda ings, and canoes. from the Codringtons for 15 years starting In the historic era, plantation agriculture in 1746. This was the only time the Cod­ did not take hold on Barbuda because of rington family gave up control of Barbuda the island’s shallow soils and insufficient before surrendering the lease in 1870 (Wat­ rainfall. Yet, it was tried repeatedly. ters, 1997). Reynolds’ statement indicates that Martin exploited the sublease to full Outsiders have often attempted commercial advantage for short term gain; Martin and agriculture, and have usually failed. Sugar, Byam also were responsible for construct­ indigo, and cotton plantations had the same stunted lifespan in the eighteenth century ing the two timber buildings at Highland that coconut, irrigated vegetable, and cotton House (Watters and Nicholson 1982). projects had in the twentieth. (Berleant- The implication of Reynolds’ statement is Schiller,1991:44). that good timber trees had endured on Bar­ buda for at least the first sixty years of the In contrast with the sugar islands of the Codrington lease, This further suggests Lesser Antilles where extensive deforesta­ that the Codringtons may have husbanded tion took place to provide lands for cane- the island’s timber resources by practicing fields (Kimber, 1988), Barbuda’s vegetation selective cutting, an enlightened attitude was altered mainly by other factors. Harris that would accord well with the idea of (1965) found that clearing land for subsis­ Barbuda being a ”provisioning island,” tence crops, felling trees for charcoal burn­ providing staples to the Codrington sugar ing, and the introduction of grazing ani­ estates on a sustainable, long term basis. mals and alien were the primary This argument is tenuous, but it raises the agents in modifying Barbuda’s vegetation, issue of whether Barbuda’s forests, and es­ and that the rate of modification acceler­ pecially its larger trees, really were logged ated throughout the colonial era. Tweedy as intensively during the initial part of the (1981:145-148) documented the building historic era as is usually supposed. and repair of boats, which required the cut­ Scientific observations on Barbuda’s veg­ ting of larger trees. The shipwright was an etation during the historic era are scarce 28 D. R. WATTERS AND R. B. MILLER because botanists and collectors rou­ Barbuda’s native vegetation was modi­ tinely bypassed the island. Howard’s com­ fied in the prehistoric era and was altered pilation (1975:373) of the early botanical significantly in the historic era, and the im­ collectors working in the Lesser Antilles pacts continue today through free-ranging lists only one visit to Barbuda, by Henri du animals, charcoal burning, and small plot Ponthieu in 1786-1787. While there, du cultivation (Berleant-Schiller, 1983). Al­ Ponthieu wrote to William Codrington in though modified, Barbuda’s vegetation England and his letters provide important was less severely impacted than the forests information on the architectural features of Antigua, where clearance for the sugar and social conditions on Barbuda in the industry was far more intensive and ex­ 1780s (Tweedy, 1981:170; Watters, 1997). tensive (Harris, 1965:29-35; Francis et al., Beard (1949) applied the term Evergreen 1994:3). Bushland to Barbuda’s vegetation. Harris (1965) established the agents and diverse activities that had altered Barbuda’s veg­ CONCLUSIONS etation, which he termed Evergreen Wood- land. The introduction to this work (Harris, Our analysis has confirmed that the four 1965:1-4), in which he presents his observa­ timber samples collected by Beard and tions while traversing Barbuda’s landscape Watters at three historic sites on Barbuda from coast to coast, is the written account belong to three imported woods. The natu­ that best evokes a feeling for the vegetation ral geographic ranges for greenheart (Chlo­ one encounters. Stoddart et al. (1973)report rocardium rodiei) and for the species of the on the island’s unusual mangrove vegeta­ white pine group (Pinus spp.) and white tion situated well inland from all current oak group (Quercus spp.) never encom­ coastlines. Howard’s (1979, 1988, 1989a, passed Barbuda. Humans were the agent of 1989b)Flora of the Lesser Antilles compiles movement of these woods to Barbuda. A Barbuda’s terrestrial species. fourth wood, Manilkara spp., collected by The records of Beard, Harris, and Harris at Highland House, appears not to Howard were incorporated into a woody be native to Barbuda, although its geo­ plant survey of Barbuda (and Antigua), graphic distribution encompasses various conducted by the USDA Forest Service, In­ Antillean islands and its range may have ternational Institute of Tropical Forestry included Antigua, the island nearest Bar­ (IITF) in 1991 (Francis et al., 1994). One buda. hundred twenty-seven woody species, in­ We have drawn upon various disciplines cluding 65 believed to be native, were com­ to explore the impacts of humans on past piled for Barbuda (Francis et al., 1994:2 and environments on Barbuda, and each source Table 4). has contributed to our understanding of Francis et al. (1994:Table 2) segregated those changes. However, to date these dis­ woody species they observed during the ciplines (botanical studies excepted) have IITF survey into three ”Life Form’’ catego­ not generated sufficient data to test these ries (medium to large tree, small tree, and observations in a rigorous, empirical man­ shrub). Only 24 species (19%) were as- ner. Yet, Barbuda, because it has well pre- signed to the medium to large tree cat­ served subsurface stratigraphy in its ar­ egory, and of those only four were native chaeological sites and a range of aquatic trees: Zanthoxylum flavum Vahl (West In­ habitats (swamps, salt ponds), holds prom­ dian satinwood), Bursera simaruba (L.) Sarg. ise for generating such data in the future, (gumbo-limbo), Bucida buceras L. (oxhorn through archaeobotanical research on mac­ bucida), and Tabebuia heterophylla (DC.) robotanical remains, phytoliths, and pollen Britt. (species in the roble group “roble”). (in prehistoric and historic sites), and These four species could have provided the through paleobotanical research (e.g., cor­ resources suitable for making many of the ing of Barbuda’s inland mangrove wood products that Tweedy (1981:128-130) swamps). All of these studies attest to the lists as Barbuda’s exports to Antigua. importance of “cross fertilization’’ among BARBUDA’S HISTORIC WOODS 29 disciplines, a point well worth recalling in this age of ever increasing specialization in scientific fields.

Acknowledgments. -Waters' fieldwork in 1978-1979 was supported by a Fulbright- Hays Doctoral Dissertation Research Abroad Fellowship (DHEW:OE) and an Andrew Mellon Pre-Doctoral Fellowship at the University of Pittsburgh; his 1983 and 1984 research was funded by the Carnegie Museum of Natural History’s M. Graham Netting Research Fund and Clapp Chari­ table and Educational Trust. The authors are grateful to Ans Dirkmaat for translating Wagenaar Hummelinck’s article from Dutch, Jennifer Brown for manuscript preparation, Morris Nedd and Cathy Wat­ ters for field assistance, Desmond Nichol­ son (Museum of Antigua and Barbuda) and the Barbuda Council for arranging and au­ thorizing the field research, and Richard A. Howard (Arnold Arboretum, Harvard Uni­ versity) for putting the authors in contact with one another. We also appreciate the responses of our colleagues David Harris, John Francis, and Dick Howard to ques­ tions we posed, and the insightful com­ ments made by three reviewers. Finally, we pay tribute to Samuel J. Record for his fore- sight in preserving the Martello Tower wood samples and his correspondence with John Beard, a commendation we be­ lieve is truly appropriate because Dr. Re- cord died in February 1945, only three months after receiving and identifying Beard’s specimens. Record’s (1945) article on the Martello Tower wood sample and the announcement of his death are pub­ lished in the same issue of Tropical Woods, the journal he edited.

LITERATURE CITED 30 D. R. WATTERS AND R. B. MILLER