Taxonomy and Arboretum Design

Scot Medbury

In the second half of the nineteenth century, arboreta joined natural historymuseums and zoological gardens as archetypal embodiments of the Victorianfascination with the natural world.

Grouping plants by type is a familiar practicein North American gardens where small, sepa-rate collections of maples, oaks, or other gen-era are common features. Although it is nowunusual to follow a taxonomic scheme in thelayout of an entire garden, such arrangementswere the vogue in nineteenth-century botani-cal gardens and arboreta. The plant collectionsin these gardens were frequently grouped intofamilies or genera and then planted out alonga winding pathway so that visitors encoun-tered specimens in a taxonomic sequence.Growing related plants together, in effect,

organizes a collection into a living encyclope-dia, allowing for comparison of the character-istics of species within a genus or generawithin a family. By planting related taxa in anevolutionary progression, the more compli-cated sequential taxonomic arrangement re-veals the ancestral affinities of modern floras.Good examples of this display theme are the"order beds" of herbaceous plants at gardenslike Kew and Cambridge, which have longprovided botany students with a compact syn-opsis of the plant kingdom arranged intaxonomic sequence.Despite the educational advantages, there

are significant horticultural and managementproblems that result from the application oftaxonomy to the layout of a garden. Becauseplant families tend to be ecologically heteroge-neous, they require a variety of cultural condi-

tions. This is especially problematic when theconcept is applied to a plant collection thatstrives to be all-inclusive. The arboretumprojects of the Olmsted landscape architec-tural firms illustrate some of these problemsand also exhibit how changes in plant tax-onomy were expressed in the landscape.

Historical BackgroundThe historical antecedents for arranging plantcollections taxonomically include the first Eu-ropean botanical garden, the Orto Botanica,founded in Pisa in 1543. The plants in this gar-den were grouped according to their medicinalproperties and, by the end of the sixteenth cen-tury, by morphological characteristics as well(Hill 1915). As the science of botany advancedduring the Renaissance, the practice of storingherbarium collections in a taxonomic orderdeveloped, and this probably contributed tothe practice of arranging living collections in asimilar fashion.

Following the publication of Linnaeus’ com-prehensive plant classification system in 1753,botanical taxonomy changed radically, andtaxonomic gardens quickly followed suit. Wil-liam Aiton used the Linnaean system in layingout the original nine-acre botanical garden atKew in 1760, as did the Reverend ErasmusDarwin (grandfather of Charles Darwin) whendesigning his private garden at Litchfield inEngland (Simo 1988).

14

This plan, which illustrates the entry on botanic gardens m Liberty Hyde Bailey’s 1914 edition of the StandardCyclopedia of Horticulture, reflects taxonomy’s role as the standard method of organizing plant collections mbotanic gardens.

In 1759, the French botanist Bernard deJussieu became dissatisfied with the Linnaeansystem while laying out a taxonomic garden atVersailles. Jussieu began moving plants aroundin pots in an attempt to express an arrange-ment that reflected "genealogical" relation-ships. Linnaeus himself had allowed that thiswas the goal of botanists although he had notbeen able to provide more than outlines forsuch an arrangement. Bernard’s arrangementwas further developed by his nephew AntoineLaurent de Jussieu, and his classic GeneraPlantarum fairly soon gained broad acceptancein Europe and became the basis for taxonomicarrangements in gardens.Following Jussieu’s work, three successive

systems of classification have been principallyemployed in the layout of sequentially orderedtaxonomic plant collections. In chronological

order, these were (1) the system of the Swissbotanist Augustin Pyramus de Candolle,which was based on rather different principlesthan that of Jussieu, although the main differ-ence might seem to be in the plant with whichthe sequence of flowering plants starts-Dutchman’s-pipe (Aristolochia) for Jussieu,and buttercups (Rananculus) for Candolle; (2)the system of George Bentham and JosephDalton Hooker, published in England between1862-1883, and in some ways an elaboration ofthe Candollean system; and (3) the post-Dar-winian system of Adolph Engler and KarlPrantl, published in Germany between 1887-1915 and the first widely accepted system to bebased on evolutionary progression.

In order to appreciate the progression ofplant families in taxonomic gardens, it is firstnecessary to understand the placement of the

15

gymnosperms and subdivisions of the an-giosperms (i.e., monocotyledons and dicotyle-dons) within each of these classification

systems. The sequence of dicot families is es-pecially important, for although the pre-Dar-winian systems of Candolle and Bentham andHooker began with the polypetalous (many-petalled) buttercups and magnolias, Engler andPrantl’s dicot sequence commenced with thewillows and birches, whose apetalous (petal-less) flowers they considered to be more primi-tive. Both the Candolles (Augustin as well ashis son Alphonse) and Bentham and Hookerplaced the gymnosperms between the mono-cots and dicots because they thought that gym-nosperms really were very complex organisms,just like dicotyledonous trees, whereas Englerand Prantl placed them first, as the purportedprogenitors of the angiosperms.The sequence of families in Jussieu’s,

Candolle’s, and Bentham and Hooker’s sys-

tems was not intended to show evolutionaryprogression. However, they did attempt to re-flect their authors’ general ideas of the progres-sion of morphological complexity. Jussieu’sarrangement, as far as can be ascertained,forms a basically linear sequence, but theCandolles and Bentham and Hooker were ada-mant that plant relationships did not follow alinear sequence, although the printed pageforced such a sequence on them. Most plantclassification systems appearing after the workof Charles Darwin and Alfred Russell Wallacehave been predicated on an understanding ofdescent and evolution and, therefore, havetried to establish "evolutionary" relationshipsamong plants. In the first of these phylogeneticsystems, such as Engler and Prantl’s, plantfamilies were placed in a sequence beginningwith the most primitive plants and endingwith the most advanced. This is still the casetoday, although a veritable forest of evolution-

Contour plan of the Derby Arboretum, 1839. The arboretum path that winds around the peripherys concealedfrom the mam walk at the center by thick evergreen shrubbery and six- to ten-foot berms. From } C. Loudon’scatalog, The Derby Arboretum, published m London, 1840.

16

ary "trees" has been produced. Each treepurports to show the complexity of the rela-tionships between plant families that cannotbe accomodated by the linear sequence of theprinted book.

The Derby ArboretumIt was to the Candollean system that theEnglish author and garden designer JohnClaudius Loudon looked when laying out theDerby Arboretum in England in 1839, the mostinfluential of the taxonomically arranged Brit-ish gardens. Early in his career Loudon had be-come intrigued by the novel marriage ofscience and landscape beauty that a taxonomicgarden presented (Simo 1988). In 1803 heseized upon Jussieu’s system as the organizingstructure for a large arboretum and flower gar-den at Scotland’s Scone Palace. In 1811 he rec-ommended a similar "living museum" for thecity of London, with plantations arranged bythe Linnaean system in one area and byJussieu’s system in another. Neither the SconePalace nor the London garden materialized asenvisioned. But with the taxonomic design forDerby, Loudon brought the arboretum into anew era, where it joined the natural historymuseum and the zoological garden as an arche-typal embodiment of the Victorian fascinationwith the natural world.The Derby Arboretum was designed to be

viewed in a prescribed sequence. This conceptdrew on the eighteenth-century English tradi-tion of emblematic landscape gardens suchas Stourhead, where statuary and classicaltemples, as they were revealed sequentially tothe viewer, were intended to call up specificideas and allusions, usually from classical his-tory or poetry. At Derby, however, a new para-digm was evoked, that of science. The pathswere designed to follow, in sequence, the"natural order" of the plant collections.The main walk at Derby is on a central axis

that brings visitors to a seating area in themiddle of the park. The tree collection wasplanted along a secondary walk that takes aserpentine course around the park’s perimeter,allowing visitors to enter the park, experience

the entire collection, and then leave by thesame gate without retracing their steps.Loudon employed the "gardenesque" style(which he created and advocated) when plant-ing the arboretum, displaying the trees singlywith sufficient room for each specimen to de-velop without touching others. Such carefulplanning notwithstanding, the arboretum wasintended to be torn up and replanted every fewdecades, in order to remove outsized trees andto permit the addition of new taxa (Loudon1840).The Derby Arboretum greatly impressed

both the American landscape architectFrederick Law Olmsted, Sr., and his friend andmentor, Andrew Jackson Downing, America’sfirst native-born professional landscapedesigner and most influential transmitter ofcontemporary English design for Americanuse. Both men, when given the opportunity todesign public parks, included taxonomic arbo-reta in their proposals, drawing heavily onLoudon’s writings and his seminal design forthe Derby Arboretum.

North American ExamplesNorth America’s first botanical gardens wereplanted without particular attention to taxo-nomic or other thematic arrangements. Thecontinent’s first proposal for a taxonomicallyarranged garden appears to have been made in1839, for Nova Scotia’s Halifax Public Garden,followed closely by a design by Downing for aDerby-like arboretum in Boston’s Public Gar-den, probably in 1841 (Zaitzevsky 1982). But itwas Olmsted and Calvert Vaux’s inclusion of ataxonomic arboretum in their 1858 "Green-sward" plan for New York’s Central Park thatbecame the most significant early proposal,since it inaugurated eighty years of involve-ment in taxonomic arboretum design by theOlmsted firms.As with the Derby Arboretum, the forty-

acre Central Park Arboretum was designed tobe a self-contained and sequential experience.Also like the Derby, its plan followedCandolle’s system of classification. Since the

Derby Arboretum had been criticized by

17

Downing for its "peculiarity of design," a ref-erence to the use of "scattered single trees andshrubs" (Downing 1850), Olmsted and Vaux’splanting plans for the Central Park Arboretumavoided the aesthetic shortcomings of suchspotty, gardenesque planting by displaying treespecies both as specimens and en masse(Zaitzevsky 1982).Olmsted and Vaux not only attempted to

make the taxonomic arrangement appear pic-turesque but also tried to place families wherethey would grow best. Thus they attempted toreconcile one of the major problems of taxo-nomic arrangements. Strict adherence to taxo-nomic groups and a fixed, linear sequence offamilies may locate plants on unsuitable sites,where they will not flourish. After all, mem-bership in a botanical genus or family implieslittle or nothing about a particular species’ cul-tural requirements or preferences in habitat.Species within the same genus may originatein such widely dissimilar habitats as bog anddesert, as occurs, for instance, within the ge-nus Pinus.Given this formidable problem, Olmsted

and Vaux did their best to bring each family"into a position corresponding to its naturalhabitats," in some locations winding the pathsto achieve this. Nevertheless, in their designthey were preoccupied with preserving the bo-tanical sequence rather than concerned withthe habitat preferences and performance of in-dividual species.Olmsted’s 1858 plan for the Central Park

Arboretum never came to fruition, but fifteenyears later he was presented with an evengreater opportunity, this time in Boston. Theresult was the Arnold Arboretum, NorthAmerica’s quintessential taxonomically ar-ranged plant collection. The Arnold has devel-oped out of the collaboration and foresight of avariety of institutions and individuals, amongthem Charles Sprague Sargent, its first directorand, with Olmsted, co-designer.Olmsted and Sargent chose Bentham and

Hooker’s classification as the taxonomic guidefor their planting plan, which, althoughBentham and Hooker’s Genera Plantarum was

published after the appearance of Darwin’sOrigin of Species (1859), did not embraceDarwin’s views. The trees were set out bygenera, ordered according to Bentham andHooker’s sequence. Every species to beincluded was planned for in advance, whichrequired modifications later when unantici-pated species and subspecific taxa (subspecies,varieties, and formae) were acquired. As withthe Candollean system, Bentham and Hooker’sclassification begins with the magnolias andtheir relatives, which were assembled at theentrance to the then one-hundred-and-thirty-acre arboretum. The rest of the collection thenfollowed according to sequence, although thistime it was to be viewed from a winding car-riage road instead of a pedestrian path, a sen-sible innovation given the size of the property.The design also arranged species geographi-

cally within each generic group. The plants ofNorth America were the first to be encoun-

tered, followed by those of Europe, and finallythose of Asia. This created considerable com-

plexity in the layout. To add to this complex-ity, the species within each continental

subgroup were placed in the sequence in whichthey appeared in Bentham and Hooker’s book.Because the main collection was intended to

be permanent, specific places for individualspecimens and groves were designated on theplan. Early studies for the distribution ofplants placed related species on both sides ofthe road, as Olmsted had done in the CentralPark Arboretum plan. But in the final Arnoldplan, species groups were assembled on oneside only, with the next genus appearing acrossthe road, and so forth, in staggered fashion.The Bentham and Hooker sequence was fol-

lowed quite closely in Olmsted and Sargent’splan. Only one major genus, Salix, appears tohave been placed out of sequence and that wasdue to cultural necessity. The moisture-lovingwillows were planted in wet ground near thearboretum entrance, far from their proper placeat the end of the dicot sequence. Bentham andHooker placed the conifers after the dicots;consequently, Olmsted deployed the dicotsalong winding roads so as to terminate at an

18

In the Bentham and Hooker sequence, comfers followed the dicotyledons. This 1991 photo by Rdcz andDebreczy captures part of the Arnold Arboretum’s collection at just over the century mark.

existing stand of native hemlocks. Nearby, hecreated a pinetum for cultivated conifers.Unlike the plan for Central Park, where

families containing mostly shrubs were inter-spersed in proper sequence among the treefamilies, Sargent arranged the shrubs at theArnold in a separate fruticetum (from theLatin frutex, meaning shrub), also arrangedin a progressional sequence according to

Bentham and Hooker (Gamboni and Hamburg1983).

In a few cases, strict adherence to the taxo-nomic scheme resulted in poor performanceamong various groups of plants. For instance,

the plan called for magnolias to be planted nearthe entrance, but some tender species have hadto be sited elsewhere, where warmer microcli-mates prevail. Similarly, the flowering cherrieshad been established in one of the coldest areasin the arboretum. In subsequent years, the se-quence has been modified when necessary toaccomodate the cultural requirements of theplants.The Arnold Arboretum undoubtedly had an

enormous impact on the development ofAmerican gardens that followed, including theNew York Botanical Garden and the BrooklynBotanical Garden, where parts of the perma-

II§U<»

*S

~o

g§I’a

~cj

00

ISM

2022a

§M

’C!

Ia

I8

§u

~go

o~CjCO

!~~i"&q

i5oo

20

nent collections were laid out taxonomically.Olmsted maintained a profound interest in thecreation of arboreta throughout the remainderof his career, producing arboretum plans forthe city of Rochester, New York, Stanford Uni-versity, and other institutions. Olmsted’s lastcommission, the Biltmore estate in NorthCarolina, included an ambitious proposal forwhat would have been the world’s greatest col-lection of trees and shrubs, arranged taxonomi-cally along a sinuous nine-mile drive. Thecollections policy for the Biltmore arboretumwas the broadest imaginable: every woodyplant from the world that might be hardy, cul-tivars included, was to be acquired andplanted, whether it was currently in cultiva-tion or not. (Throughout much of the ArnoldArboretum’s history, its collections policy hasfocused on hardy species of woody plants, by-passing most cultivars.) Such comprehensive-ness ultimately proved to be the Biltmorearboretum’s undoing. Because the layout of thecollection was determined by a plant classifi-cation system (in this case, Bentham andHooker’s), it was necessary to know in advancehow many hardy tree and shrub species wouldbe represented in each genus so that adequatespace could be allocated in the proper sequen-tial location. Due to an incomplete knowledgeof temperate floras (especially of Asiatic

regions) and widespread synonymy in thenursery trade, compilation of such a masterplanting list was a daunting task, as it wouldbe today. This impasse no doubt played a partin the collapse of the arboretum project at theturn of the century.As successors to the senior Olmsted’s prac-

tice, the Olmsted Brothers firm continued atradition of making taxonomic plans for arbo-reta and influenced other landscape architectsto do the same. The firm was commissioned to

generate plans for the Missouri Botanical Gar-den, the Holden Arboretum in Ohio, the origi-nal Rancho Santa Ana Botanic Garden insouthern California, and the University ofPennsylvania’s Morris Arboretum. Unfortu-nately, all of these designs either wentunexecuted or no longer exist.

The Missouri Botanical Garden project isnotable in that it roughly coincided with thepublication of Engler and Prantl’s classifica-tion system, the first system based on Darwin-ian ideas of evolution to achieve wide use.William Trelease, the garden’s first director,decided to make use of both the Bentham andHooker and the Engler and Prantl systemswhen engaging the Olmsteds to lay out twonew geographic collection areas. The Benthamand Hooker was chosen for the American col-lection for its familiarity among botanists. Alarger garden devoted to the "universal flora"followed the Engler and Prantl system becauseit illustrates evolutionary affinities amongplants. Both gardens were to contain meresynopses of their respective floras. In this way,the designers avoided the horticultural prob-lems that have plagued other taxonomicallyarranged gardens because representative spe-cies from a particular family or genus could beselected based upon their horticultural com-

patibility. In addition, the designers did nothave to wrestle with the planning issues aris-ing from a comprehensive collections policy,such as those that confounded the Biltmore

project.In 1936, the Olmsted Brothers produced the

firm’s last taxonomic arboretum plan, for theUniversity of Washington Arboretum in Se-attle. The Engler and Prantl system was exclu-sively applied here to a collection intendedfrom the outset to be comprehensive in scope.Following the post-Darwinian system of

Engler and Prantl, the taxonomic sequence be-gan with Gingko biloba, the most primitivehardy gymnosperm, followed next by the coni-fers, the woody monocots, and finally by thedicots. The dicot sequence was initiated not bythe magnolias as in Bentham and Hooker’s sys-tem, but with apetalous families like the wil-low and birch.

Despite the aesthetic and intellectual appealof their strikingly rendered plans on paper, theOlmsted Brothers’ last arboretum design re-vealed a sharp decline in conceptual quality.The firm appears to have been copying aspectsof the Arnold Arboretum plan merely out of

21

custom, without reexamining the theoreticalbasis for arranging plants in a taxonomic se-quence. Incredibly, the critical interrelation-ship between the botanical sequence and thepedestrian circulation was omitted, therebystripping the design of the educational ele-ments that justify the use of a taxonomic se-quence in the first place. Fortunately, theUniversity of Washington Arboretum hassince developed independently of the OlmstedBrothers’ plan for it, with greater sensitivitygiven to the physical characteristics of the siteand the cultural requirements of the plants.

DisadvantagesWith few exceptions, North American botani-cal gardens founded after the 1930s have es-chewed sequential taxonomic arrangements infavor of geographic, ecological, or strictly aes-thetic schemes (or combinations thereof). Gar-dens continue to present small displaysorganized by evolutionary sequence, such asthe Prehistoric Glen at Honolulu’s Foster Bo-tanical Garden or the Plant Families garden atthe North Carolina Botanical Garden, but thecomprehensive application of taxonomy togarden design is virtually forgotten today.

For purposes of comparing related plants,the approach works reasonably well with her-baceous perennials, where the plants are smalland the flowers within reach-a substantialamount of diversity can be encompassedwithin a relatively small area. Woody plantslend themselves less successfully to a sequen-tial taxonomic treatment. The flowers are of-ten out of reach and the plants are spacedfarther apart. This makes comparisons of floralcharacters difficult. However, there remainsthe advantage of comparison of overall formas, for instance, in the collections of maplesand lindens.

A redrafted version of the Olmsted Brothers’taxonomic plan for the University of WashingtonArboretum, which employed the classificationsystem of Engler and Prantl From the Bulletin ofthe University of Washington Arboretum, Volume1, Number 1, December 1936.

22

A view ot the Arnold Arboretum s linden (Tilia) collection m 1991 A taxonomic scheme offers the opportumtyfor side-by-side compansons of overall form. Photo by Rdcz and Debreczy.

New developments in taxonomy also poseproblems for sequentially arranged taxonomiccollections of woody plants. While a herba-ceous garden can be torn out and replanted fol-lowing acceptance of a new taxonomic system,such a drastic approach is impractical in a ma-ture arboretum. Nevertheless, woody collec-tions that follow an obsolete classification

system are anachronisms, worth maintainingfor their historical interest but lacking in someof the educational values that originally led to

the use of a taxonomic sequence. At the sametime, it is clear that no planting sequence cando justice to natural relationships, whether asunderstood in 1850 or 1993. From the point ofview of teaching natural relationships, somemeans of organizing a collection is better thannone.

The most serious drawback to a taxonomic

arrangement, however, deals with horticul-tural issues. The point has already been madethat taxonomic groups above the species level

23

often contain plants from widely dissimilarhabitats. The varying degrees of sun and shadetolerance as well as the differing nutritionaland moisture requirements found amonggroups of related species cause problems whenthese plants are grown together under similarconditions. Many plants will simply die whenplaced in the wrong spot. Others will strugglefor years in a sickly or stunted condition andconsequently form poor examples of the aver-age size, growth, or appearance of a particularspecies.There are ways of avoiding some of the

aforementioned problems of sequential taxo-nomic arrangements. Planning a synoptic col-lection rather than a complete one affords theopportunity of choosing plants based on easeof culture and other factors. Diversity of ter-rain also permits greater flexibility, as the lin-ear sequence can be bent to site taxonomic

groups in the positions best suited to the ma-jority of their member species. Species clusterscomposed of the often smaller generic unitrather than of families have also tended to bemore successful, since one is most likely tofind a tolerable site for a smaller group of

species. As at the Arnold Arboretum, curatorsmust make exceptions to a hard-and-fast

pursuit of any scheme.

AcknowledgmentsThis article originated m the author’s M.S. thesis at theUniversity of Washington, where he investigatedthe application of taxonomy to the display of plantcollections. An earlier version was published in ThePubhc Garden, July 1991. The author thanks PeterDel Tredici, Karen Madsen, Richard Schulhof, StephenSpongberg, and especially Peter Stevens for theircomments and suggestions during the preparation ofthis version. The assistance of Sheila Connor at theArnold Arboretum Archives, Joyce Connolly at theOlmsted National Historic Site, and Paul Ledvma atthe Manuscripts Division, Library of Congress, is alsogratefully acknowledged.

Archives

Arboretum Documents Collection, Miller Library,Center for Urban Horticulture, University ofWashington, Seattle.

Archives of the Arnold Arboretum, Jamaica Plain, MA.

Olmsted Associates Records, Manuscripts Division,Library of Congress, Washington, DC.

Olmsted National Historic Site, National Park Service,Brookline, MA

Sources and Suggested Readings

Connor, Sheila, and B. June Hutchmson 1979/1981."The Original Design and Permanent

Arrangement of the Arnold Arboretum asDetermined by FLO[lmsted] and CSS[argent]:A Chronology." Unpublished paper. Archivesof the Arnold Arboretum.

Downing, A. J. 1850. "The Derby Arboretum." TheHorticulturist 5: 266-267.

Gambom, J. F., and B. T. Hamburg. 1983. "The ArnoldArboretum and the History of the ShrubCollection." Unpublished paper. Archives ofthe Arnold Arboretum.

Hill, A. W. 1915. "The History and Functions ofBotanic Gardens." Annals of the MissouriBotamcal Garden 2:185-240.

Loudon, J. C. 1840. The Derby Arboretum. London:Longman.

Simo, M L. 1988. Loudon and the Landscape NewHaven: Yale University Press.

Zaitzevsky, C. 1982. Fredenck Law Olmsted andthe Boston Park System Cambridge, MA:Harvard University Press.

Scot Medbury is a Beatrix Farrand Scholar in the Ph.D.program in environmental planning at the Departmentof Landscape Architecture, University of California atBerkeley.

24

Map of Hiroshima. The star marks the hypocenterof the atomic bomb blast. Circles indicate thedrstances 1 kilometer, 2 kilometers, 3 kilometersfrom the hypocenter. Adapted from Hiroshima byHiromi Tsuchida, Kosel Pubhshing Co., 1985.

0 Eucalyptus tree (Eucalyptus sp.)

0 Camphor tree (Cinnamomum camphora)

0 Gmkgo tree (Ginkgo biloba)

^J Pine tree (Pinus thunbergii)

@ Willow tree (Salix sp ) /