Genetic Diversity and Improvement inLeucaena

Potential for Improvement of LeucaenaInterspecific Hybridisation

through

C.T. Sorensson1

Abstract

The interspecific hybrid approach to leucaena improvement is reviewed. Leucaena specieshybridise readily (76%) giving breeders the opportunity to transfer genes of interest from wildspecies into breeding populations and cultivated species. Techniques have been developed tohelp make the approach practical. Several interspecific hybrids appear promising on the basisof their growth rate and combinations of traits. Aspects requiring further research, includingappropriate deployment of interspecific hybrids, are considered.

THIS paper describes the potential for improvingleucaena by breeding useful hybrids between species.It reviews:• reasons for using hybrids;• hybrids that seem promising;• technical advances that help make hybridisation

possible;• particular breeding strategies;• heterosis (hybrid vigour) and its importance to

interspecific leucaena hybrids.Leuceana species hybridise far more readily than

do most other herbaceous and semi-woody plantgenera. Ninety-one of the 120 possible interspecificcombinations among 15 leucaena species are com-patible (Sorensson and Brewbaker, in press), in con-trast to 5% compatibility when alfalfa species arecombined (Quiros and Bauchan, 1988).

Tropical tree genera, for example Eucalyptus,Erythrina and Acacia, generally tolerate crossingbetween species. Trees may be tolerant of widehybridisation either because of genetic redundancyin their DNA or because tree species often evolvedin geographical, phenological and ethologicalisolation from related species, and thus havehad little or no pressure to develop barriers tohybridisation.

1 New Zealand Forest Research Institute, Division ofBiotechnology, Private Bag 3020, Rotorua, New Zealand(Formerly at Department of Horticulture, University ofHawaii at Manoa, 3190 Maile Way, Honolulu, Hawaii96822, USA)

Reasons for Hybridising Leucaena SpeciesThe main reason for breeding interspecific hybridsis that it enables us to move genes from species thathave useful traits to others that lack them. Leucaenaleucocephala is widely-grown and has excellentqualities, including adaptability to a range of sitesand agronomic practices, drought resistance, foragepalatability/digestibility and bypass protein, highyields of leaf and wood, aggressive coppicing, woodpulping characteristics, and nitrogen fixation.However, it also has several limitations. Forexample, it sets a great deal of seed, exhibits season-ality of growth, is sensitive to attack by psyllids(Heteropsylla cubana), and yields poorly whengrown at cool sites (below about 15°C) or in acidsoils with high available aluminium to calcium ratio.Other negative aspects of L. leucocephala includeslow seedling establishment, high mimosine content,intolerance to waterlogging, and the susceptibilityof its postwood to termites.

At least half of the lesser-known species in thegenus Leucaena display superior qualities that wewould like to transfer to L. leucocephala. Forexample, four diploid species (L. collinsii, L. diversi-folia, L. esculenta and L. pallida) are highly resistantto psyllids (Sorensson and Brewbaker 1986; Brayet al. 1990). L. diversifolia is also regarded as asource of genes for cool temperature tolerance(Brewbaker et al. 1988) and acid soil tolerance(Hutton 1990). L. esculenta and L. pallida have highseedling vigour (Sorensson et al. 1994).

47

There are other reasons for attempting to breedinterspecific hybrids in the genus leucaena.l They may be adapted to a wider range of sites.l They may be highly vigorous (heterosis) and out-

yield their parents.The hybrids may be unique biochemically becauseof their genetic diversity, as expressed by com-pounds such as multimeric enzymes.

l Segregating interspecific hybrids exhibit anenormous range of phenotypes, which at leastindicate genetic recombination, if not high geneticdiversity.

l Interspecific hybrids of tropical trees generally setless seed. Seedless leucaena hybrids would beuseful in some agricultural production schemes.

l If superior interspecific hybrids were identified,they might form the basis of a commercial seedindustry.

l Interspecific hybrids could have unique com-binations of the phenotypic traits desired forhorticultural use in gardens and landscapes.

Promising Leucaena Interspecific Hybrids

The first notable interspecific hybrids to be reportedin the literature were those of L. pulverulenta xL. leucocephala in Indonesia. In contrast to L.leucocephala, which is a self-compatible tetraploid,L. pulverulenta is outcrossing. About a hundredyears ago, L. pulverulenta was introduced as a shadecrop for coffee in cool upland sites where L. leuco-cephala was relatively unproductive. Whenoccasional interspecific hybrids formed as a resultof cross-pollination by bees, they were apparentlyidentifiable by their outstanding growth rates andlow seed production (advanced generation lines thatsegregated for seedless individuals). Although thisparticular hybrid is quite susceptible to psyllids, andis therefore not appropriate for planting in manyparts of the wet tropics, it is potentially useful forarid parts of India (V. Gupta, pers. comm.).

In northern South America where leucaenas origi-nated, cultivated species like L. esculenta have beenmoved outside their original range, providing oppor-tunities for species to hybridise. There is now sub-stantial evidence that interspecific hybridisation hasplayed a significant role in forming new species, par-ticularly with regard to the four tetraploid speciesthat all appear allopolyploid (Hughes and Harristhese Proceedings). The offspring of L. esculentax L. leucocephala is one of these natural hybrids.It occurs sporadically in south central Mexico, andoften grows to impressive dimensions (Hughes andHarris 1994). The seed sterility of this natural hybridhas been verified in hybrids artificially bred inColombia and Hawaii (Sorensson 1987; Hutton

1988). Segregants among Fl L. leucocephala K8 xL. esculenta K138 produce gums (Brewbaker andSorensson, 1990) with some structural similaritiesto gum arabic.

Two other triploid interspecific hybrids that showpromise for cultivation are L. diversifolia (n = 26)x L. leucocephala and L. pulverulenta x L.diversifolia (n = 52) (Brewbaker and Sorensson,1990). Although Fl hybrids segregate some unthriftyseedlings, most are very productive. Triploids of L.diversifolia x L. leucocephala set few seeds, havereadily digestible leaf dry matter, are highly psyllid-resistant and can apparently tolerate frost(Sorensson 1987, 1993; Gutteridge and Sorensson1992). Triploids of L. pulverulenta x L. diversifoliaare essentially seedless, highly psyllid-resistant, andpotentially good wood producers, but appear tohave low forage quality (Sorensson 1993).

The hybrid L. diversifolia subsp. diversifolia (4n)x L. leucocephala has caused wide interest, and hasbeen evaluated fairly thoroughly internationally inthe Nitrogen Fixing Tree Association LeucaenaPsyllid Trials and Leucaena Seed Production Trials.This self-compatible tetraploid is based on twoparents that are relatively uniform, and is itself quiteuniform in the Fl generation. Advanced generationprogeny segregate individuals with exceptionalvigour and tolerance to cool conditions. Althoughmost of them are prolific seeders and onlymoderately psyllid-resistant (Wheeler and Brewbaker1990), a recurrent inbreeding selection programcould produce lines from this hybrid with excellentqualities for wood production. Moderate levels ofpsyllid resistance are considered to be adequate forwood production.

Perhaps the most exciting hybrid is L. pallidaK748 x L. leucocephala K636 ( = CPI 84581 xPI 443740). In replicated two-year-old forage trialsconducted against eleven species in Hawaii, andagainst six species in Brisbane, Australia, this hybridyielded more total above-ground biomass and edibledry forage than the other species (Castillo 1993; M.Austin and colleagues, unpublished). Its seedlingvigour was also superior (Sorensson et al. 1994).Total seedling weight of L. pallida x L. leuco-cephala hybrids was more than double that of L.leucocephala K636 on day 84, both in pots in theabsence of psyllids, and in the field with psyllidspresent.

This hybrid (referred to in Hawaii as KX2) is self-incompatible in the first generation (Sorensson1989a) but segregates self-compatible individuals inlater generations that can be selfed and entered intorecurrent selection programs. It usually has aspreading habit with basal branching. Like L.diversifolia (4n) x L. leucocephala ( = KX3), KX2

48

l

segregates wildly in later generations. An apparentassociation between psyllid resistance and smallleaflet dimension in KX2 becomes less evident inlater generations (Sorensson 1993). Continuing workin Hawaii by Brewbaker and colleagues aims toproduce stable KX2 lines. Current line x testerexperiments with KX2 combinations in Hawaii andAustralia indicate that certain tree x tree crossingsproduce hybrids with high specific combining abilityfor forage yield, but low proportions of leaf to stem(Sorensson et al. 1994) and high condensed tannincontent (Castillo

Technical

Several technical

1993).

Aspects of InterspecificHybridisation

aspects of interspecific hybridis-ation have received serious study.

Researchers have identified three self-compatiblespecies that could require emasculation; L. diversi-folia (4n), L. leucocephala and Leucaena sp. ‘glossy’(Sorensson 1989b). Differences in their floralanatomy (i.e. relative lengths of androecia andgynoecia) account for differences in their responseto cross-pollinations in the absence of emasculation,which have been more than 90% successful for L.diversifolia but less so (about 50%) for L. leuco-cephala (Sorensson and Sun 1990; Sorensson 1993).Three emasculation methods have been described— soap (Hutton and Gray 1959), pre-dawnemasculation (Sorensson 1988a), and bud emascu-lation (Gupta and Patil 1984).

Leucaena pollen can be manipulated using newlydeveloped techniques. Pollen desiccated over calciumchloride remains viable for 20 days when stored at–20°C, and for up to three months when storedat –75°C (Sorensson 1993). Pollen has beengerminated and grown in vitro on sucrose-agar-borate slides to assess its viability (Sorensson andNagahara 1989). Studies on radioresistance ofleucaena pollen to gamma irradation indicate that30 k rads can render pollen genetically ineffective,and irradiated mentor pollen have been applied inpollen mixtures to overcome the self-sterility of L.pallida (Sorensson 1993).

Several aspects of flowering have been studied.Flowering seasons of Leucaena species have beendescribed in northern South America (Hughes 1993),in India (Gupta 1990) and in Hawaii (Sorensson1988a). They do not differ substantially betweensites. Immature inflorescences develop at rates whichare linear and vary from species to species(Sorensson 1989b). Inflorescences of most speciesbegin meiosis when they reach 60% of their maturediameter (Sorensson 1993). We know roughly howmuch pollen leucaena species produce, and the

length of time during which styles are receptive topollen before and after anthesis (Sorensson 1988b).Most shaded branches do not flower profusely.Flowering in L. leucocephala K636 can be partiallysynchronised by timing irrigation to manage soilwater stress on Molokai, Hawaii (N. Dudley pers.comm.).

Several morphological traits of interspecifichybrids have been modelled. For example, flowercolour is shown primarily to be inherited additively,and triploids exhibit dosage effects. The latter arealso significant in leaf parameters — pinnule lengthand width, number of pinnules per leaf, pinnae pairsper leaf. With rare exception, leaf parameters areinherited on a geometric scale between parents(Sorensson 1987, 1993), apparently as a result ofadditive interaction between developmentalregulatory genes. Leaf morphology can also be usedto identify hybrids from parental-type seedlings(Sorensson 1990, 1993; Sorensson and Shelton1992). Diameters of inflorescences, shapes andnumbers of petiolar glands, and leaflet shapes ofinterspecific hybrids are primarily controlled byadditive inheritance (Sorensson 1987, 1990).

The quantity of seed produced by cross-breedingfifteen leucaena species has been measured,following nearly 60 000 floret pollinations(Sorensson and Brewbaker in press). The averagenumber of viable interspecific seeds resulting fromcross-pollination of ten florets has been tabulated(Sorensson in press). A technician pollinating about30 inflorescences of a highly compatible interspecificmating can produce about 2500 viable hybrid seedsin one day (assuming six pods per inflorescence andfifteen seeds per pod). The potential for hybrid seedproduction in orchards with bee pollination has beenpartially demonstrated by the heavy hybrid seedproduction of a single tree of diploid L. diversifoliaplanted amongst L. leucocephala (Gutteridge andSorensson 1992).

Several aspects of seed production have been con-sidered. The susceptibility of leucaena pods to seedbeetles in Hawaii has been surveyed, with L. diversi-folia (2n, 4n) being less preferred by beetles(Sorensson 1993). Visual estimation of insectinfestation in leucaena seed has been shown to beinaccurate. An inexpensive and effective seedcleaning method, in which the seed is soaked inwater and separated by floating in salt water, hasbeen successfully tested (Sorensson 1993). Freezingseeds does not reduce seed viability (Cobbina et al.1990) and is highly effective in killing insectsinfesting the seeds (N. Dudley, pers. comm.; C.Hughes, pers. comm.). In Hawaii, seeds fromLeucaena species take from about 70 to 330 daysto mature (Wheeler 1991).

49

Breeding Strategies Involving InterspecificHybridisation

There are two ways of breeding interspecific hybrids— one relies on clonal reproduction to capture hete-rosis (hybrid vigour) and the other does not.

Several tropical trees have been greatly improvedby propagating clones of their elite segregants, andlarge areas have already been planted to clonallypropagated interspecific hybrids of Eucalyptus andCasuarina. Equally impressive gains could be madeby use of clonal leucaenas, as has been partiallydemonstrated at Pondok Gedeh, Indonesia (Toruan-Mathias et al., these Proceedings) using a natura-lised line apparently derived from a hybrid of diploidL. diversifolia and L. leucocephala. Several rootedcutting techniques show promise for large-scale use(e.g. Osman, these Proceedings).

All vigorous interspecific leucaena hybridsproduce flowers, with the exception of L.pulverulenta x L. lanceolata (Sorensson 1987). Of52 different hybrids that flowered in Hawaii, 30produced some seed from open pollination.Triploids have proved difficult to use in standardbreeding programs because their chromosomes areunstable and because they tend to set little or noseed (Hutton 1985). Diploid hybrids vary greatly intheir seed production. A few vigorous diploidhybrids produced seed readily in Hawaii (L.shannonii x L. collinsii, L. diversifolia x L.collinsii, L. shannonii x L. salvadorensis, L.lanceolata x L. shannonii, L. pulverulenta x L.retusa, L. lanceolata x L. macrophylla), and itappears they could be bred further. Pollen stain-ability and seed production are generally well-correlated (Sorensson 1987, 1993).

Three tetraploid species, L. diversifolia, L. leuco-cephala and L. pallida, make up the primarybreeding pool, and are completely compatible witheach other (Sorensson and Brewbaker in press). Alltetraploid hybrids produced among the four knowntetraploid leucaena species have been reasonablyvigorous and seed productive, although a fewtetraploid lines derived from L. diversifolia x L.leucocephala in Brazil were unstable, losing between1 and 18 chromosomes (Freitas et al. 1991). Previousbreeding programs based on advanced generationprogenies have emphasised mass selection, but manynow concentrate on self-compatible segregants andinvolve inbreeding and recurrent selection.Tetraploid hybrids may be backcrossed to a parent,crossed to a third species (three-way), or evencrossed with a different interspecific hybrid(four-way).

Breeders can use several breeding tactics withinterspecific hybrids which set fertile seed.

(i)

(ii)

(iii)

(iv)

(v)

Move small numbers of genes from one speciesto another by backcrossing and recurrentselection.Mass-produce hybrid seed from particularparent trees, either species or hybrids, whosehybrid progeny have high general and specificcombining ability for yield or quality.Select elite true-breeding lines from apopulation of primarily outcrossing hybrids bycycles of selfing or sibbing, and bulk theinbreds as a synthetic line with reasonablybroad genetic diversity.Repeatedly mass select hybrid populations tocull undesirable genes but otherwise maintaingenetic diversity.Move genes from diploid species into thetetraploid breeding pool via induced polyploidyor unreduced gametes. This approach was usedin Hawaii to produce several tetraploid hybridsfrom 2n–4n or 4n–2n matings (Sorensson1988b) and could well have been an importantmechanism of allopolypioid speciation in thegenus Leucaena.

Heterosis

Heterosis is a major reason for interest in leucaena’sinterspecific hybrids, as it is in other tree groupssuch as poplars, pines and eucalypts. Heterosis, orhybrid vigour, is the phenomenon in which offspringof crosses between populations perform better thanthe average of the two populations. Heterosis isthought to result from dominance effects whenheterozygous loci, in favourable states, mask or sup-press the negative effects of mildly unfavourablealleles (which may be tightly linked to favourablealleles and thus hard to breed out).

Some of leucaena’s interspecific hybrids show out-standing heterosis. Field workers may recall plotsof species that were dominated by a suspiciouslylarge and productive tree with morphology similar,but not identical, to the parental species. In perhapstwenty such cases that I have investigated, the out-standing individual has always proved to be an inter-specific hybrid. Further inspection sometimesidentified additional hybrids whose performanceswere unremarkable. It is difficult to say howsuperior these hybrids really are because they shadeout their neighbors. However, in a seedling vigourstudy (Sorensson et al. 1994), Fl hybrids of bothL. pallida x L. leucocephala and L. diversifolia xL. leucocephala had grown taller than their parents,at rates of about one cm per day, by day 84.

Several biochemical or genetic models have beendeveloped to explain heterosis. In one such model,an undesirable allele of one gene (A) is tightly linked

50

to a desirable allele of another gene (B). If the geneproduct of B uses the gene product from A as aprecursor, then providing A through interspecifichybridisation may enable the potential of allele Bto be fully expressed. This could remove a factorlimiting yield. Correcting such factors is critical tothe success of a plant breeder (Mangelsdorf 1952).

Two additional points should be noted. Firstly,in the absence of A’s gene product, the allele B,though present, confers no advantage to the parent.As an example, when plants with good pestresistance are grown in a pest-free environment theirpest resistance gives them no advantage, and mayeven carry a metabolic cost. Secondly, if heterosisis conferred by a single locus H, then it is possibleto inbreed many generations and retain heterosis,assuming the breeder carefully selects for heterosis(Mangelsdorf 1952).

Another model for heterosis involves multimericenzymes. In the simplest case with dimeric enzymes,parent 1 has an enzyme composed of identicalsubunits A and parent 2 has an enzyme composedof identical subunits B. Their hybrid has three formsof dimeric enzymes — both parental types (AA, BB)plus a unique form (AB). The situation rapidly getsmore complex when enzymes are composed of threeor four subunits. The hybrids benefit from thegreater diversity of enzymes, as well as from uniqueforms of enzymes within plants. Populations ofhybrids bred from few parent trees may validly becriticised for being ‘inherently narrowly-based’genetically (Hughes 1993). Yet, within individualplants, there may be significant diversity ofisozymes. This diversity may show up as stabilityacross environments, and probably underlies thewide environmental adaptability of L. leucocephala,which is thought to have a hybrid origin with L.pulverulenta as the maternal parent (Hughes andHarris, these Proceedings).

Polyploidy also can increase genetic diversitywithin plants, and result in heterosis by making itmore likely that alleles will exist in favourableheterozygous states. Tetraploid leucaenas which aremultiallelic could produce a range of isozyme formswithin a plant. Polyploidy also protects the genusagainst wide hybridisation (Sorensson and Brew-baker in press). It enhances chromosome pairing inFl interspecific hybrids by allowing chromosomesto preferentially pair within genomes — poorchromosome pairing within genomes could havecaused hybrid breakdown in F2 of three-waytetraploid hybrids (Sorensson 1989a).

If heterosis is not visible in hybrid plants grownin some environments, it may show up in exoticenvironments. Plant species have developedenzymatic systems which operate efficiently in the

soil-light-water regimes of their native environments,but which may be ill-adapted to exotic environ-ments. Hybrids may do better in such environments,because of the diversity of their enzymes and geneproducts. Leucaena diversifolia (4n) x L. leu-cocephala is a vigorous hybrid: grown at an exoticupland site in Hawaii under low temperatures it per-formed outstandingly better than its parents (Brew-baker et al. 1988).

Research Priorities

The most important area for new research is hybridseed production, since new hybrid germplasm canbe neither tested nor adopted without seed. Parentallines could be clonally propagated, and then pilotseed production orchards could be set up at sitesthat are relatively free of seed pests (such as easternAustralia or northern India) to try to produce massquantities of seed for distribution. Basic researchtopics might include:• how to optimise pollen transfer by bees between

trees used as males and femaleshow distant seed orchards should be from foreign

pollen sources to prevent pollen contaminationhow to synchronise floweringmanagement of tree form and spacing to optimise

and mechanise seed harvesthow to produce several genotypes of seed froma single orchard (this would mainly investigate theeffects of compatible and incompatible pollen onseed set).Since seed orchards need vegetative propagation,

more work is needed on vegetative propagationmethods, particularly on those that already showpromise (Osman, these Proceedings; Toruan-Mathiuset al., these Proceedings). Unfortunately, it appearsthat rooting ability may be genotype-specific, as itis in Eucalyptus (Toruan-Mathius pers. comm.).Clones are needed in order that yields of elite geno-types can be measured accurately, traits’ sensitivityto environmental conditions can be estimated, clonescan be used as replicates in some experimentaldesigns, and for mass propagation of seedless orgum-producing plants. Cloned stands of self-incompatible species or hybrids would remain seed-less if planted as a monoculture (Brewbaker 1988).

Three other research areas warrant attention.First, can chromosomes or chromosome segmentscontaining useful genes be transferred betweenspecies, for example by backcrossing? Perhaps thiscould be done through chromosome addition(Leucaena ‘confertiflora’ has eight more chromo-somes than L. leucocephala).

Second, in order to decide which hybrids areappropriate for cultivation at specific sites,

51

•

• •

•

researchers need to identify the qualities that makeinterspecific hybrids different from wild Leucaenaspecies. Are hybrids truly better adapted to marginalor stressful sites than parent species? Is the heterosisor combining ability of some hybrids significant ineconomic terms? Do seedless hybrids channel carboninto leaves and wood that is otherwise used up inflower and seed production? Can seedless hybridsplay a major role in revegetating damaged andfragile ecosystems, perhaps acting as biological nursetrees?

Third, recommendations are needed to guide theuse of interspecific hybrids by small landholders.Hybrids, particularly segregating ones, may not beappropriate for small landholders who cannot easilyafford to buy elite seed, and who may find itdifficult to properly manage the great phenotypicvariability of a segregating population. If weak andill-adapted segregants could be culled out in a hedge-row or protein bank forage system — perhaps atthe age of six months (Sorensson et al. 1994), thiscould leave a genetically-diverse and extremelyproductive plant population that would be quitebeneficial to a small landholder. Hybrids could beill-suited to other situations too, such as low plantdensity for wood production, ormoisture.

Acknowledgments

McIntire Stennis Forestry Funds,

limiting soil

US Dept ofAgriculture, supported this research financially. TheHawaii Natural Energy Institute partially fundedtravel to this workshop.

References

Bray, R.A., Julien, M.H. and Room, P.M. 1990. Leucaenapsyllid in Australia — the current situation. In:Napompeth, B. and MacDicken, K.G. ed., LeucaenaPsyllid: Problems and Management. Bogor, Indonesia,January 16-21, 1989. Bangkok, Thailand, F/FRED,ICRC and NFTA, 8-11.

Brewbaker, J.L. 1988. Cloning of seedless leucaenas forplantation use. Leucaena Research Reports 9, 111-112.

Brewbaker, J.L. and Sorensson, C.T. 1990. LeucaenaBentham: new tree crops from interspecific hybrids. In:Janick, J. and Simon, J. ed., Advances in New Crops.Portland, Oregon, Timber Press, 283-289.

Brewbaker, J.L., Wheeler, R.A. and Sorensson, C.T. 1988.Psyllid-tolerant highland leucaena yields. LeucaenaResearch Reports, 9, 11-13.

Castillo, A. 1993. Agronomic performance, psyllid(Heteropsylla cubana Crawford) resistance and foragequality of L. leucocephala (Lam.) de Wit, L. pallidaBenth., L. diversifolia Schlecht. and their hybrids.M.Agr.Sc. thesis, University of Brisbane, Queensland,Australia.

Cobbina, J., Kolawole, G.O. and Atta-Krah, A.N. 1990.Leucaena and Gliricidia seed viability and germinationas influenced by storage conditions. Leucaena ResearchReports, 11, 91-93.

Freitas, L.H.D., Schifino Wittman, M.T. and Paim, N.R.1991. Floral characteristics, chromosome number andmeiotic behavior of hybrids between L. leucocephala (2n= 104) and tetraploid L. diversifolia (2n = 104).Brasilian Journal of Genetics, 14(3), 781-789.

Gupta, V.K. 1990. Genetics and breeding of Leucaenaleucocephala (Lam.) de Wit. In: Pathak, P.S., Deb Roy,R. and Singh, P. ed., Multipurpose Tree Species forAgroforestry Systems. IGFRI, Jhansi, India, 99-109.

Gupta, V.K. and Patil, B.D. 1984. A simple technique ofhand emasculation in Leucaena leucocephala. LeucaenaResearch Reports, 5, 29-30.

Gutteridge, R.C. and Sorensson, CT. 1992. Frost toleranceof a hybrid of L. diversifolia x L. leucocephala inQueensland, Australia. Leucaena Research Reports, 13,3-5.

Hughes, C.E. 1993. Leucaena Genetic Resources: The OF1Leucaena Seed Collections and a Synopsis of SpeciesCharacteristics. Oxford Forestry Institute, Oxford, UK,117 p.

Hughes, C.E. and Harris, S.A. 1994. Characterisation andidentification of a naturally occurring hybrid in Leucaena.Plant Systematics and Evolution, 192, 177-197.

Hutton, E.M. 1985. Problems in breeding low-mimosinetypes in the genus Leucaena. Tropical Agriculture, 62,329-333.

-- 1988. Results from two interspecific leucaena crosses:L. leucocephala x L. esculenta and L. leucocephala xL. shannonii grown in an acid oxisol. Leucaena ResearchReports, 9, 37-39.

-- 1990. Field selection of acid-soil tolerant leucaenafrom L. leucocephala x L. diversifolia crosses in atropical oxisol. Tropical Agriculture, 67(l), 2-8.

Hutton, E.M. and Gray, S.G. 1959. Problems in adaptingLeucaena glauca as a forage for the Australian tropics.Empirical Journal of Experimental Agriculture, 27,187-196.

Mangelsdorf, A.J. 1952. Gene interaction in heterosis. In:Gowen, J.W. ed., Heterosis: a Record of ResearchesDirected Toward Explaining and Utilizing the Vigourof Hybrids, 320-329.

Quiros, C.F. and Bauchan, G.R. 1988. The GenusMedicago and the Origin of the Medicago sativa Com-plex. Alfalfa and Alfalfa Improvement,ASA/CSSA/SSSA, Madison, Wisconsin, MonographNo. 29, 107.

Sorensson, C.T. 1987. Interspecific Hybridisation in theGenus Leucaena Benth. M.Sc. thesis, Dept of Agronomyand Soil Science, University of Hawaii at Manoa.

-- 1988a. Pollinating and emasculating techniques forleucaena species. Leucaena Research Reports, 9,127-130.

-- 1988b. Utilizing unreduced gametes for productionof novel hybrids of leucaena species. Leucaena ResearchReports, 9, 75-76.

-- 1989a. Status and mechanisms of self-incompatibilityin leucaena species. Plant Cell Incompatibility News-letter, 21, 77-85. PO Box 2000, Cortland, New York13045.

52

-- 1989b. Rate of inflorescence development of leucaenaspecies. Leucaena Research Reports, 10, 80-83.

-- 1990. Shapes of pinnules of leucaena species and Flinterspecific hybrids. Leucaena Research Reports, 11,124-128.

-- 1993. Production and Characterisation of Inter-specific Hybrids of the Tropical Tree Leucaena(Leguminosae: Mimosoideae). PhD dissertation, Univer-sity of Hawaii at Manoa.

- - in press. Seed productivity of artificial hybridisationsamong fifteen species in the genus leucaena. In:Proceedings of the 1992 IUFRO conference on BreedingTropical Trees, Cali, Colombia, October 9-14.

Sorensson, C.T. and Brewbaker, J.L. 1986. Psyllidresistance of leucaena species and hybrids. LeucaenaResearch Reports, 7, 13-15.

- - in press. Interspecific compatibility among fifteen leu-caena species (Leguminosae: Mimosoideae). AmericanJournal of Botany.

Sorensson, C.T. and Nagahara, D. 1989. In vitro pollengermination of leucaena species. Leucaena ResearchReports, 10, 84-86.

Sorensson, C.T. and Shelton, H.M. 1992. Number ofpinnules per pinna used as a marker to identify seedlingsof L. diversifolia, L. leucocephala, L. pallida and theirhybrids. Leucaena Research Reports, 13, 91-94.

Sorensson, C.T., Shelton, H.M. and Austin, M.T. 1994.Seedling vigour of some leucaena spp. and their hybrids.Tropical Grasslands, 26.

Sorensson, C.T. and Sun, W.C. 1990. Production of inter-specific hybrid seed without emasculation. LeucaenaResearch Reports, 11, 129-132.

Wheeler, R.A. 1991. Guide to Management of LeucaenaSeed Orchards. Winrock International-F/FRED,Bangkok, Thailand, 18 p.

Wheeler, R.A. and Brewbaker. J.L. 1990. An evaluationof the results from the leucaena psyllid trial network.Leucaena Research Reports, 10, 11-16.

53

Systematics of Leucaena:Recent Findings and Implications for Breeding and

Conservation

C.E. Hughes and S.A. Harris1

Abstract

The status of leucaena systematics is reviewed. Considerable taxonomic confusion persistsand is shown to be hampering the efficient utilisation, genetic improvement and conservationof leucaena species. Herbarium research, morphological and molecular analyses, and recentfield exploration and collection are providing new insights into the evolution of the genus,the status of species, species relationships, the origin of tetraploid species and parentage ofnatural hybrids. Results of recent research are summarised and reveal a more complex pictureof the genus than has generally been acknowledged. The full extent, scope and importanceof human interference in the recent evolution of the genus is now becoming apparent. Resultsdemonstrate that the widespread indigenous domestication of leucaena species for pod pro-duction in Mexico accounts for recent speciation and provides a plausible scenario for theorigin of L. leucocephala in domestication. The multidisciplinary approach has been highlybeneficial, combining field and laboratory work and including both morphology and moleculesto study the genus. It has allowed significant advances to be made towards a better under-standing of the genus.

LEUCAENA is a small genus of around 17 speciesbelonging to the tribe Mimoseae. All species arenative to the New World with the greatest speciesdiversity in Mexico (13 species, 6 endemics) andnorthern Central America (6 species, 3 endemics)in seasonally dry, mainly tropical habitats. Thegenus extends north into southern Texas, USA,sporadically across the Caribbean, and into SouthAmerica as far south as Peru.

Status of leucaena systematics

The taxonomic history of leucaena is complex. Atone time it included another genus, Schleinitzia, andit was later split into three genera. However, thedelimitation of species has been the greatest sourceof difficulty and confusion. While some species arenarrowly distributed, uniform and clearly distinct,the complex patterns of morphological variation insome of the widely distributed polymorphic species

1 Oxford Forestry Institute, Department of Plant Sciences,University of Oxford, South Parks Road, Oxford OX13RB, UK

and species groups, notably the L. diversifolia, L.shannonii, L. esculenta and L. macrophylla groups,have led to difficulties in the circumscription ofspecies. Choice between specific and subspecificrank has also been a source of debate and confusion.These problems are aggravated by the considerablehuman interference known to have affected leucaenathrough its indigenous domestication in Mexico, andby known polyploidy and interspecific hybridisation.

Bentham (1842) first established the genusleucaena, initially with four species, and later withnine (Bentham 1875). Botanical exploration over thenext 50 years resulted in the description of a prolifer-ation of supposed new species and culminated inthe revisions of Standley (1922) who recognised 15species, and Britton and Rose (1928) who added 24species, bringing the total to 39. A long dormantperiod followed with only minor additions andlimited synthesis prompted by the compilation ofthe Floras of Peru (Macbride 1943), Guatemala(Standley and Steyermark 1946), Panama (Schery1950) and Novo-Galicia (McVaugh 1987). In total,some 61 different species or subspecies names havebeen ascribed to leucaena.

The ensuing taxonomic confusion was radicallysimplified by Brewbaker and Ito (1980), Brewbaker(1987) and Sorensson and Brewbaker (1994) whoquestioned the validity of the numerous binomialsand accepted first 10, and subsequently 16, speciesas legitimate, establishing a temporary workingtaxonomy for the genus. In a brief synoptic paperon the genus for Mexico, Zarate (1984a) mentionedseveral new taxa and these have now been publishedvalidly in a new revision of leucaena for Mexico(Zárate 1994). Despite several recent minortaxonomic clarifications (Hughes 1988, 1991; Zárate1987a,b; Pan 1988), no complete taxonomic revisionhas been published since that of Britton and Rose(1928).

In cultivated taxa and their wild relatives, theefficient use, improvement and conservation ofbiological diversity must be guided by biology ofthe taxa concerned (Falk and Holsinger 1991),understanding (i) the phylogeny (ii) the amount anddistribution of genetic variation (iii) the effectivedesign of suitable sampling strategies (iv) effectiveconservation methods.

Successful long-term taxon management dependson knowledge of the genetics and demography ofthe taxon, enabling the design of biologically soundmanagement strategies. Such data are increasinglyimportant in the development of integrated con-servation strategies, combining population andtaxon management with in-situ and ex-situcollections.

Leucaena systematics remain very confused. Thereis no good guide for identifying species, and accuratespecies distribution maps have only recently beencompiled (Hughes 1993). Several new species havebeen discovered and await formal description whileseveral undescribed species have been used inbreeding and improvement. The origins of theknown polyploid species are still unconfirmed, andthe accuracy and status of several published namesis still under debate. As an ever wider spectrum ofspecies and hybrids is brought into use and breeding,it will become more difficult to identify speciesproperly. Lack of a clear taxonomic frameworkhampers other research, germplasm acquisition, treeimprovement and genetic conservation programs sothey inevitably follow sub-optimal strategies. Thetaxonomy of the genus is urgently in need ofrevision.

The OFI leucaena research program:background and objectives

Early work on leucaena at the Oxford Forestry Insti-tute (OFI) concentrated on assembling seedcollections of some of the lesser-known andpotentially valuable species from Central America

(Hughes 1986, 1988, 1991). Collections were laterexpanded to include the complete spectrum ofleucaena species (Hughes 1993).

Since 1990, research has focused on the sys-tematics of the genus with the overall objective ofcompletely revising the taxonomy of leucaena,including species descriptions, keys to the identifi-cation of species, botanical illustrations, distributionmaps, a clear understanding of species relationships,interspecific hybridisation, the origin of the knownpolyploid species and the phylogeny of the genus.

Here we present the main results of the first fouryears of the OFI leucaena systematics research, anddiscuss their implications for leucaena utilisation,improvement and conservation. Detailed results arepresented and discussed elsewhere (Hughes 1991,1993; Harris et al. 1994a,b,c; Hughes and Harris,1994). Research is continuing to complete ataxonomic revision.

Methods

Herbarium and botanical database

More than 2700 botanical specimens from 20Mexican, U.S. and European herbaria have beenexamined and logged onto a botanical database,BRAHMS (Filer 1993). This has allowed accuratespecies distribution maps to be produced for the firsttime (Hughes 1993). Accurate distribution maps notonly underpin the taxonomy but are the basis ofefficient utilisation, breeding and conservation. Themaps have made it possible to make preliminaryassessment of environmental tolerances for differentspecies (Hughes 1993), to design efficient andaccurate sampling strategies for germplasm collec-tions, and to plan collection expeditions and in situgenetic conservation (Hellin and Hughes 1994).

Exploration and field collection

Botanical material has been collected during fieldexploration throughout leucaena’s range. Severalnew taxa have been discovered as well as numerousextensions to species distributions. All known taxahave been explored and observed in the field, andcomplete flowering and fruiting material gathered,forming a working collection for taxonomicdescription and morphological analysis. In addition,we have obtained seed, dried leaf material for DNAextraction, fixed flowers, wood and bark samples,bruchid seed predators and photographs, and gainedinformation on ethnobotany and conservationstatus. This unique collection has also formed thefoundation for molecular analysis of the genus.

55

Field trials and morphology

Morphology has been re-assessed, both using thelarge collection of botanical specimens assembledat Oxford, and on a living collection established byCONSEFORH in Honduras in 1989. Previous workon leucaena systematics had relied on a limited rangeof often unreliable taxonomic characters such as leafsize, shape and number of pinnae and leaflets, podsize and pubescence. Work is in progress to iden-tify a wider range of reliable diagnostic charactersfor the genus. Bark, inflorescences, extrafloral nec-taries, seeds, seedlings, pollen and flowers areproviding good characters for analysis.

Results

1. Structure of the genus and status of speciesand subspecies

Molecular analysis

Over the past decade, molecular genetics has startedto make a significant contribution to general under-standing of (i) phylogenetic relationships; (ii) theamount and distribution of genetic variation; (iii)hybrid parentage; (iv) the genetic stability of in situand ex situ collections. However, the vast majorityof such studies has been conducted in herbaceouscrops such as cereals and grain legumes, with veryfew studies of tree crops and even fewer of tropicaltrees.

Molecular analysis of leucaena has been carriedout using a suite of techniques and differentmolecules. Chloroplast DNA (cpDNA) and nuclearribosomal DNA (rDNA) have been used to examinespecies relationships, the origins of known tetraploidspecies (Harris et al. 1994a,c) and natural hybridparentage (Hughes and Harris 1994). Isozymes andseed storage proteins have been used to look atspecies relationships and the distribution of geneticvariation within and between species andpopulations (Harris et al. 1994b; Chamberlain 1993).Intact total DNA is readily extracted from eitherfresh or dried leaves of all leucaena species. Har-vesting of leaves into plastic bags and drying withsilica gel, according to the method of Chase andHills (1991), has been successfully used to obtainDNA from leucaena trees in the field, overcomingthe need for seed (Hughes and Harris 1994). Chloro-plast DNA, rDNA, isozymes and seed storageproteins show amounts of variation that enablepatterns of genetic variation to be analyzed at alllevels, from the overall structure of the genus to theidentification of putative hybrids and the structureof individual populations. Analysis of cpDNA andrDNA data has allowed a suite of species-specificmolecular markers to be identified for the majorityof species. Chloroplast DNA has been found to bematernally inherited.

Considerable progress has been made towards abetter knowledge of the structure and evolution ofthe genus, relationships between species (Fig. 1), andthe status of species and subspecies. The importanceof indigenous domestication and the occurrence ofinterspecific hybridisation in leucaena are betterunderstood. A provisional delimitation of validspecies and subspecies and their synonyms ispresented in Table 1. Some of the principal resultsare summarised below and discussed for each of themain species groups.

Figure 1 Chloroplast DNA phylogeny of the genusLeucaena, generated via a parsimony approach usingcharacter data derived from the analysis of 14 restrictionenzymes and 8 mung bean chloroplast DNA probes thatcovered approximately 85% of the choloroplast genome.

Large-leaflet leucaena complex

The large-leaflet morphological complex of leucaenacomprises four species L. lanceolata, L. macro-phylla, L. multicapitula and L. trichodes. There hasbeen some debate over how the members of thiscomplex are related, and how many species andsubspecies it includes (Zárate 1984a; Brewbaker1987), but morphology and cpDNA evidence con-firm the four species above.

Brewbaker (1987) placed L. multicapitula intosynonymy with L. trichodes, although this wasreversed by Sorensson and Brewbaker (1994). Thereis good morphological and molecular evidence that

56

Table 1. Leucaena taxonomy, species names, synonymy and chromosome numbers (2n).

Recognised species andauthorities

Recognised subspecies Synonyms Chromosome number (2n) notes

L. collinsii B&R collinsii L. esculenta (Mot. etSesse ex ADC) Benth.ssp collinsii S. Zárate

52, 56

zacapana C.E. Hughes ?Previously confused withL. diversifolia.

L. confertiflora Zàrate Two subspecies c. 112ined. recognised by

Zàrate (1994)

L. cuspidata Standley

L. diversifolia(Schlecht.) Benth.

Possibly two subspecies. ?

diversifolia L. laxifolia Urban 104L. trichandra (Zucc.)Benth.

stenocarpa (Urban)S. Zàrate

L. stenocarpa Urban 52L. standleyi B&R A widespread polymorphic taxonL. revoluta B&RL. guatemalensis B&RL. molinae Standley andWilliams

L. esculenta esculenta L. confusa B&R 52(Moc. & Sesse ex ADC) L. doylei B&RBenth.

matudae Zàrate ?

paniculata (B&R)S. Zàrate

L. dugesiana B&RL. oaxacana B&RL. paniculata B&RL. pallida B&R

104Confusion remains over thecorrect name. L. pallida hasbeen widely used

pueblana (B&R)C.E. Hughes comb.ined.

? Unpublished combination forTehuacan Valley endemic.

L. greggii S. Watson Rhyncoleucaena greggii 56(S. Watson) B&R

L. involucrataS. Zirate

? Discovered in 1991 by OFI inSonora, Mexico; allied to theesculenta complex

L. lanceolata S. Watson Zàrate recognises twosubspecies lanceolataand sousae but notclearly defined

L. microcarpa RoseL. brandegeei B&RL. pubescens B&RL. rekoi B&RL. sonorensis B&RL. cruziana B&RL. palmeri B&RL. purpusii B&RL. sinaloensis B&RL. nitens M.A. Jones

52A widespread and variablespecies

57

Table 1. Leucaena taxonomy, species names, synonymy and chromosome numbers (2n). (continued)

Recognised species andauthorities

Recognised subspecies Synonyms Chromosome number (2n) notes

L. leucocephala (Lam.) leucocephala L. glauca (Willd.) 104de Wit Benth. = the shrubby or Hawaiian

varieties

glabrata (Rose) S.Zàrate

L. glabrata Rose 104= the giant or Salvador type

L. macrophylla Benth. macrophylla L. macrocarpa RoseL. houghii B&R

52Doubtfully distinct from L.trichodes

nelsonii (B&R) S. Zàrate L. nelsonii B&R ?

L. multicapitula Schery 52Doubtfully distinct at specieslevel from L. trichodes and L.macrophylla

L. pulverulenta(Schlecht.) Benth.

Acacia pulverulentaSchlecht.

56

L. retusa Benth. Caudoleucaena retusa 56(Benth.) B&R

L. salvadorensisStandley ex. B&R

L. shannonii J.D. Smith 56subsp. Previously confused with L.salvadorensis (Standley) shannonii and with the SalvadorS. Zàrate type of L. leucocephala (subsp.

glabrata)

L. shannonii J.D. Smith shannonii

magnifica C.E. Hughes

52, 56

?

L. trichodes (Jacq.)Benth.

L. canescens Benth. 52L. bolivarensis Britton & Part of the L. macrophylla/L.Killip multicapitula complexL. colombiana Britton &Killip

L. sp nov 1 ? Unnamed; discovered byCONSEFORH in 1990 inHonduras; in the L. shannoniicomplex

58

it is a distinct species. Zarate (1984a) recognised twosubspecies within both L. lanceolata (with materialfrom southern Mexico assigned to subsp. sousae)and L. macrophylla (with lowland coastal materialfrom Guerrero and Oaxaca attributable to subsp.nelsonii). The cpDNA results show two plastometypes in L. macrophylla corresponding to thesubspecies and supports their recognition (Harris etal. 1994a). Further evidence for the distinction oftwo subspecies within L. macrophylla comes frommorphology, geography and because they growradically differently in trials (Stewart et al. 1991).Patterns of morphological variation within L.lanceolata are more complex and appear to followa clinal pattern from NW to SE, which suggests con-tinuous variation rather than two recognisablesubspecies. The more complex pattern of variationwithin L. lanceolata is supported by the cpDNAdata.

The shannoni complex

The shannonii complex comprises three species —L. salvadorensis, L. shannonii (with two subspeciesshannonii and magnifica) and a new, as yetundescribed, species from northern Hondurasdesignated L. sp nov 1. We do not know how themembers of this complex are related and study hasbeen hampered by the confusion surrounding thenames of some of the taxa.

For example, L. salvadorensis (Hellin and Hughes1994) had been placed into synonymy with L. leuco-cephala subsp. glabrata (Brewbaker 1980) or alter-natively called a subspecies of L. shannonii (Zarate1987b). However, detailed studies of morphology,growth characteristics and distribution show that L.salvadorensis might be treated as a species in its ownright (Hughes 1988; Hellin and Hughes 1994). ThecpDNA, seed storage protein and isozyme dataclearly separate it from both L. shannonii and L.leucocephala (Harris et al. 1994a; Chamberlain1993). There is now general consensus that L. salva-dorensis should be treated as a valid species.

The discovery in 1991 of a new species, Leucaenasp nov 1, by the CONSEFORH project in northernHonduras, is another sign of the incomplete stateof leucaena systematics. This fine tree is clearlydistinct and appears to be most closely related toL. salvadorensis, based on its morphology andmolecular evidence (Fig. 1) (Harris et al. 1994a;Chamberlain 1993).

Hughes (1991) described L. shannonii subsp. mag-nifica based on a range of morphological differencesfrom typical L. shannonii. It is a narrowly restrictedendemic from southeastern Guatemala. ChloroplastDNA and isozyme variation support the recognitionof two subspecies within L. shannonii, and suggest

that there is very little within-population geneticvariation for subsp. magnifica compared to otherspecies in the shannonii complex (Chamberlain1993).

The diversifolia complex

L. diversifolia sensu lato is the most widespreadspecies in the genus, occurring through the highlandsof south-central Mexico and Central America as farsouth as Nicaragua. Apart from L. leucocephala,it is the most widely known and planted species out-side its native range. L. diversifolia was studied byPan (1985) who confirmed the distinction of diploidand tetraploid cytotypes and designated them asseparate subspecies trichandra and diversifoliarespectively (Pan 1988). The diploid subspecies wasdescribed as L. diversifolia subsp. stenocarpa byZárate (1984a, 1994). The tetraploid subspeciesdiversifolia was found at that time to be restrictedto a small area in central Veracruz in Mexico andwas hypothesised to be an autopolyploid, i.e. der-ived from diploid L. diversifolia. The L. diversifoliacomplex sensu Pan (1985) also includes L. pul-verulenta and L. esculenta subsp. paniculata (= L.pallida sensu Brewbaker 1987). The nomenclatureof this tetraploid remains controversial (Brewbaker1987; Zárate 1984a, 1994). We discuss this taxonfurther in relation to the esculenta group.

From recent exploration, it appears that thetetraploid subspecies diversifolia is more widespreadthan believed by Pan (1985). It occurs from Hidalgo,south along a narrow belt on the wet slopes of theSierra Madre Oriental facing the Gulf, then throughVeracruz, northern Oaxaca and Chiapas in Mexico,into northern Guatemala in Huehuetenango (Hughes1993). Populations from northern Oaxaca weredescribed by Zárate (1984a, 1994) as L. brachycarpa,but are apparently indistinguishable morphologicallyfrom L. diversifolia subsp. diversifolia. However,recently collected material from these areas has notyet been verified as tetraploid. In the analysis ofcpDNA, leucaena accessions identified as L. diver-sifolia s.l. were found in two separate and robustlysupported plastome clades (Fig. 1). The majorityof the populations examined fell into groupsaccording to their morphological subspecies. ThecpDNA phylogeny casts doubt on Pan’s hypothesisthat tetraploid L. diversifolia is an autotetraploidderivative of diploid L. diversifolia. Instead it sug-gests that it is a segmental allotetraploid, with L.pulverulenta as the most likely maternal progenitorof the tetraploid (Harris et al. 1994a). This fits wellwith present-day distributions (L. pulverulenta andtetraploid L. diversifolia are found sympatrically inparts of central Veracruz around Misantla) and withthe close morphological affinities of these twospecies as recognised by Zárate (1984a).

59

The nomenclature of the L. diversifolia group iscomplex and unresolved. Tetraploid L. diversifoliais probably of allopolyploid origin and is moreclosely related to L. pulverulenta than to the diploid,so it seems illogical to maintain these two taxa assubspecies; they should be treated as separatespecies.

Both morphological and molecular analysis showthat L. diversifolia s.l. is one of the most variabletaxa in the genus. Molecular analysis of twentypopulations indicated that the two cytotypes can bereadily distinguished, and that both contain highlevels of genetic diversity, the diploid being morevariable than the tetraploid. There is tremendousmorphological variation across the disjunct rangeof diploid L. diversifolia so it is possible that thespecies may need to be split taxonomically. This isbeing postponed until detailed morphometric studiesare complete.

Many of the introductions of L. diversifolia out-side the Neotropics have been of the Hawaiianaccession K 156, a self-compatible tetraploid derivedfrom seed collected in Veracruz (Brewbaker 1987).This accession is likely to have a narrow geneticbase, but there is tremendous variation availablewithin L. diversifolia, so a much broader geneticbase could be examined. Breeders might incorporateimproved adaptability, productivity, product qualityand psyllid tolerance into L. diversifolia, either fordirect planting or for use as a parent in the pro-duction of hybrids. Seed source will probably be ofcritical importance in L. diversifolia, justifyingdetailed provenance trials for this species.

The esculenta group

The esculenta group, comprising three taxa, L.esculenta subsp. esculenta, L. esculenta subsp.paniculata and L. esculenta subsp. matudae, hasbeen investigated most actively by Zárate (1984a)who proposed this designation of subspecies. Zárate(1994) described a fourth taxon L. involucratabelonging to this group. Pan (1985) proposed thatthe tetraploid L. esculenta subsp. paniculata was anamphidiploid of L. esculenta subsp. esculenta anddiploid L. diversifolia. The cpDNA data would sup-port a hybrid origin with subsp. esculenta as thematernal parent (Harris et al. 1994a). As mentionedabove, the nomenclature of this taxon is underdebate (Pan 1985; Brewbaker 1987; Zarate 1984a).Considerable morphological and molecular varia-tion has been demonstrated within subsp.paniculata. Recent field work suggests that this maybe partly accounted for by the occurrence of a dis-tinct taxon, endemic to the Tehuacan Valley inPuebla and Oaxaca which is clearly referrable to thebasionym L. pueblana.

Zarate (1984a) identified a distinct taxon endemicto the Balsas Depression in Guerrero and designatedit a subspecies of L. esculenta subsp. matudae. Thisname has now been formally published (Zárate1994) but apparently leucaena researchers have notwidely accepted it as valid (Brewbaker 1987). Thetaxon’s morphology and molecular analysis link itto the esculenta group. However, subsp. matudaeis readily separated from typical esculenta by a suiteof diagnostic morphological and molecularcharacters and by its unique growth form in trials(Stewart et al. 1991), and may be more appropri-ately treated as a distinct species.

The last taxon, L. involucrata, was discovered inremote mountains in north-central Sonora.Although allied to the esculenta group, this taxon’slocation is more than 1500 km north of the nearestknown occurrence of other members of the group.The finding adds to the confusion and debate aboutthis group, and the problem of assigning rank totaxa. One reason for the extreme complexity of theesculenta group may be the particular attention paidto the group during its indigenous domestication.Clearly, these taxa have been transported, cultivatedand even selected over several thousand yearsthroughout south-central Mexico.

Leucaena cuspidata and L. confertiflora

These are two of the least known leucaena taxa andconsequently their taxonomy and relationships arevery poorly understood. Brewbaker (1987) hasquestioned the validity of L. cuspidata as a separatespecies. Our work suggests that both L. cuspidataand L. confertiflora are valid species, not onlycrucially important to the understanding of theevolution of the genus but also of considerableinterest in leucaena research and use.

Leucaena cuspidata is a very distinctive specieswith large woody pods, glossy cuspidate leaflets andunusual floral bracts. It occupies a basal positionon the cpDNA phylogeny (Fig. 1) (Harris et al.1994a) and is clearly a valid species. It is distributedin the Mexican states of Hidalgo, Queritaro and SanLuis Potosi.

Leucaena confertiflora was first discovered in1974 in the village of San Pedro Chapulco in Puebla,Mexico by Sergio Zárate and Bob Reid. At that timeZarate (1984a) emphasised its affinities to L.cuspidata, treating it as a subspecies of cuspidata.L. confertiflora has now been formally described,and clearly represents a valid species in its own rightwith many interesting attributes. It is a self-compatible tetraploid (Sorensson and Brewbaker1994) but further work will be required to find outits origin and affinities. Preliminary cpDNA evi-dence does not support a close association betweenL. confertifora and L. cuspidata.

60

Leucaena greggii and L. retusa

Although Britton and Rose (1928) placed L. greggiiand L. retusa in two segregate monotypic genera(Rhyncoleucaena and Caudoleucaena respectively)all subsequent authors have treated them as part ofthe genus Leucaena (Zárate 1984a; Brewbaker 1987;Hughes 1993). Morphologically L. greggii and L.retusa are separated from the rest of the genus bytheir yellow flowers held on long peduncles; longcaudate, exserted or short pointed floral bracts;thickened woody pods with longitudinal or obliquetransverse seed alignment; and small erect peg-shaped extrafloral nectaries (Hughes 1993). Toemphasise their separation from the rest of thegenus, these species are geographically andecologically isolated in northern Mexico and Texas(Hughes 1993), where they withstand regular frostand snow with minimum temperatures down to–15ºC (Glumac et al. 1987). L. retusa fails tonodulate when inoculated with strains of Rhizobiawhich effectively nodulate other species of Leucaena(Halliday and Somasegaran 1983), and both speciesshow very poor growth and survival in field trialscompared to other Leucaena species (Brewbaker1987; Stewart et al. 1991).

Evidence from cpDNA implies that these twospecies are closely related, and supports the viewthat they are distinct from other members of thegenus and that the group is primitive. Other datasuggest that L. cuspidata, another northerly dis-tributed taxon, is also primitive. Overall, theevidence suggests that leucaena originated atnortherly latitudes and migrated south, radiatingfurthest in south-central Mexico.

Leucaena leucocephala

Although L. leucocephala is used throughout thetropics, its origin and natural distribution remainunknown. Brewbaker and his colleagues (Brewbaker1983; Pan 1985; Pan and Brewbaker 1988) specu-lated that L. leucocephala was an amphidiploidbetween two sympatric leucaena species such asdiploid L. diversifolia and L. shannonii. However,they presented no firm evidence of possibleparentage. So far, no natural populations of thespecies have been located anywhere in Mexico orCentral America. As a result, this species’ germ-plasm has been collected from cultivated material,and only limited genetic variation has been located.We need a clear understanding of the origin of L.leucocephala to understand why it lacks geneticvariation.

Two subspecies are recognised. Subsp. leu-cocephala is a shrubby taxon that occurs mainly inthe Yucatan peninsula and the Isthmus of Tehuan-tepee, while subsp. glabrata is a more arborescent

taxon corresponding to the giant, ‘Peru’ or‘Salvador’ types (Zárate 1987a). The two subspeciesare readily distinguished using morphology orisozyme multi-enzyme phenotypes. The natural dis-tribution of this important tetraploid species hasbeen obscured by human interference. Recentexploration and molecular analysis are providingnew insights into the origin of L. leucocephala.Patterns of infraspecific genetic variation within thespecies are being found, but the picture is still notcomplete (Harris et al. 1994a,b,c).

A third variant has been discovered in the high-lands of Huehuetenango, Guatemala, in the uppervalley of the Rio Cuilco, where it is cultivated bylocal people. It has small pods and glabrous leavesand is also distinct in its isozyme phenotype (Harriset al. 1994b). As far as we know this variant hasnot been collected or used in Leucaena breedingefforts.

Evidence from the chloroplast genome shows thatL. leucocephala is closely associated with L. pul-verulenta along with tetraploid L. diversifolia(Harris et al. 1994a) (Fig. 1). This evidence seemsto show that L. pulverulenta is the maternal parentof L. leucocephala. That would mean L. leu-cocephala originated in, or near, the distribution ofL. pulverulenta on the eastern coast of Mexico. Thepaternal parent is unknown and cannot be identi-fied from cpDNA (which is known to be maternallyinherited). Initial analysis of ribosomal nuclear DNAindicates that the paternal parent would lack a sitefor the restriction enzyme Stu-I within its rDNA.Further work, using rDNA, randomly amplifiedpolymorphic DNA (RAPD) or sequencing, areneeded to pin-point the paternal parent. It has beenshown that hybrids formed following human inter-ference when Leucaena species were being domesti-cated in Mexico. This has led Harris et al. (1994c)and Hughes and Harris (1994) to speculate that L.leucocephala may have originated in domestication.

Leucaena species are cultivated and used as minorfood plants to produce pods in many parts ofMexico (Standley 1922; Whitaker and Cutler 1966;Zárate 1984b; McVaugh 1987; Casas 1992). We arenow discovering how many species are usedindigenously, and over what geographic spread andtime span. We can deduce the resulting degree ofdomestication and its importance to the evolutionof the genus (Harris et al. 1994c; Hughes and Harris1994). At least 13 species of Leucaena are currentlyused for human food in different parts of Mexicoand northern Guatemala. Archaeological evidenceshows that unripe pods and seed have been harvestedand used for several thousand years in the TehuacanValley and other areas (Smith 1967). Species suchas L. esculenta and L. leucocephala have been trans-ported over long distances and are widely cultivated

61

and marketed throughout central Mexico. The trans-port, use and cultivation of a wide range ofLeucaena species continues to the present day.

This long history of use and importance ofLeucaena species in indigenous communities meansthat local people have intimate knowledge ofdifferent species in terms of pod characteristics,season of pod production and flavour. Localresidents in Oaxaca cull the sterile L. leucocephalax L. esculenta hybrids, calling them ‘Guaje macho’,or ‘male’ leucaena (Hughes and Harris 1994). Inmany parts of Oaxaca and Puebla, farmers culti-vate three or more species to obtain pods year-round, with a variety of qualities. The pods are soldin markets throughout southern and central Mexico.

Given this rich and detailed ethnobotanical back-ground it is quite plausible that L. leucocephalaarose in domestication during the last few thousandyears following cultivation of one or both of theparental species somewhere in eastern coastalMexico. The Huastec region of eastern coastalMexico in north-central Veracruz is one of the areaswhere early agriculture first developed (MacNeish1965). An artificial origin would account for the lackof known natural populations and the limitedgenetic variation found in L. leucocephala, althoughtwo separate origins would have to be postulatedto account for the two known subspecies. Atpresent, L. leucocephala is extensively distributedin cultivation throughout Mexico, probably becauseof its favourable pod production characteristics. Ityields abundant pods more or less continuously allyear, and with high seed set per pod. Pods are easyto harvest and both pod and seed are sweet com-pared to those of the more bitter L. esculenta group.Such a tree, arising in cultivation, would have beenimmediately noticed and seized upon for widercultivation and use. The continued attempts tocultivate it for pod production today, at or beyondits site limits in the colder and drier parts of centralMexico, are witness to this drive and interest.

Genetic variation within L. leucocephala is likelyto be very limited and there is little value in furthertesting of varieties within that species. Below, wediscuss possible future natural or semi-naturalhybridisation leading to the formation of newspecies.

2. Interspecific Hybrids

Sorensson and Brewbaker (1994)’ have thoroughlyinvestigated the possibility of artificially crossingspecies with Leucaena. They have shown that thereare few genetic barriers to initial interspecifichybridisation within the genus. The production ofartificial interspecific hybrids has been the main

focus of leucaena breeding efforts to date (Brew-baker and Sorensson 1990; Sorensson 1992 and theseProceedings). However, the occurrence of naturalleucaena hybrids has received only sporadicattention, despite its obvious importance concerningthe possible hybrid origins of the known tetraploidspecies.

Interspecific hybridisation is accepted as animportant way of generating evolutionary noveltyin the plant kingdom (Anderson 1949; Stebbins1959; State 1975; Grant 1981). Recent data haveconfirmed that stabilisation of hybrids can lead to:(a) the origin of new homoploid hybrid derivative

taxa (Gallez and Gottlieb 1982; Rieseberg et al.1990);

(b) the introgressive origin of new intraspecific taxa(Rieseberg et al. 1990; Abbott et al. 1992);

(c) the origin of new allopolyploid (amphidiploid)species (Soltis and Soltis 1990; Ashton andAbbott 1992).

There are many unconfirmed reports of naturalhybrids in the genus. From Mexico and CentralAmerica, these reports include L. leucocephala xL. esculenta, L. diversifolia x L. leucocephala, L.pulverulenta x L. leucocephala (Sorensson andBrewbaker 1994) and L. pulverulenta x L. diver-sifolia (Zárate 1982). In addition, hybrid originshave been proposed for L. esculenta subsp.paniculata (= L. pallida) (Pan 1985), L. leuco-cephala and L. diverstfolia (Harris et al. 1994a).Zárate (1984a) suggested a hybrid origin for L.lanceolata subsp. sousae.

One of these hybrids has been thoroughly inves-tigated. Hughes and Harris (1994) were able to showthat hybrids of L. leucocephala subsp. glabrata xL. esculenta subsp. esculenta occurred in six statesin south-central Mexico. While investigating thishybrid we discovered several important aspects ofleucaena evolution and implications for future useof Leucaena species. The hybrid’s identity was con-firmed beyond reasonable doubt from a com-bination of geographical, morphological andmolecular evidence (Hughes and Harris 1994).Molecular evidence showed that L. leucocephala wasthe female parent in all the hybrids tested.

This hybrid has now been found over a wide areaof south-central Mexico in the states of Chiapas,Oaxaca, Guerrero, Morelos, Puebla and Hidalgo.In all cases the plants occur in disturbed areas intowns or villages where the parental species are cul-tivated for pod production. This suggests that thehybrid is the result of human interference wherespecies have been brought into artificial sympatrythrough cultivation, and thus may be described assemi-natural. It is largely or completely sterile andall trees must therefore have arisen as separate F1

62

hybrids. The widespread occurrence of this sterileF1 hybrid raises the possibility that speciation couldoccur through chromosome doubling and the for-mation of a fertile allohexaploid, reproductivelyisolated from each of the parents (Ashton andAbbott 1992).

We have also observed and collected a further sixputative hybrid combinations, some of which arecommon and widespread while others are rare andrestricted in Mexico and Central America. Work isin progress to confirm their identities. In all cases,these additional hybrids are also the result of humaninterference, such as transport and cultivation forpod production, which has brought together spe-cies that would not naturally occur in the sameplace.

Although there are few published reports of con-firmed natural hybrids, it is becoming clear thathybridisation has been a major mechanism for thegeneration of new species in Leucaena, largely (if nottotally) induced by human interference. Leucaenaspecies are increasingly being spread and used in cul-tivation and breeding. They are being planted inclose proximity in mixed species evaluation trials,in arboreta and in agroforestry plantings, both inCentral America and elsewhere. In Mexico and Cen-tral America, L. leucocephala is cultivated so widelythat it is unusual to find Leucaena species with noleucocephala trees within the natural populations.Inevitably the resulting spontaneous hybridisation,and the evolution of new taxa, will present bothopportunities and hazards.

New hybrids could turn out to be useful in treeplanting programs, but the seed collected in suchareas will not have guaranteed genetic integrity,creating further confusion and sub-optimal seedmaterial for planting. Aggressive new taxa couldarise which are potentially bad weeds (Abbott 1992).In the case of natural populations, introduction ofalien species could result in genetic pollution ofnative germplasm. It is of fundamental importancethat we should be able to identify the hybrids whicharise. To do so, we will need clear documentationof species introductions, informat ion onmorphology and species-specific molecular markers.

Discussion, Conclusions and Future WorkThe natural populations of the genus Leucaena havenow been more thoroughly explored, collected andinvestigated than those of almost any other tropicalwoody genus. The information provides a basis forour rapidly improving knowledge of the systematicsof the genus. New taxa are still coming to light in

the more isolated and under-collected parts ofMexico and Central America, probably with morestill to be found. The work has shown that the genusis more complex, in terms of number of taxa, thanwas previously acknowledged, and has highlightedthe importance of continuing human interferenceand interspecific hybridisation in the rapid recentevolution of new species.

Traditionally, forest genetic resources have beenassessed by examining a combination of morpho-logical and agronomic traits, but these mainlyexhibit continuous variation and their effectivenesshas been questioned by several authors (Gottlieb1977; Brown 1979). On the other hand, the appli-cation of biochemical and molecular techniques hasprovided a set of powerful tools for studying geneticdiversity within and among species and plantpopulations. These tools have generated useful dataduring the first phase of molecular analysis of thegenus. For example, analysis of proteins and DNAprovides important information.

If we examine all the known species in the genuswe should find a range of nuclear DNA species-specific markers so that we can unambiguouslyidentify hybrids and determine the origin of thetetraploid species. We are currently searching forRAPD species-specific markers and testing for theiroccurrence in known hybrid material. Promisingdata have been obtained.

The true potential of the genus as a source ofagroforestry trees is only now becoming apparent.New and valuable species and genetic diversity havebeen found, and locations for on-going germplasmcollection pin-pointed. A sound base for genetic con-servation is also being built - for the first time itis possible to assess what to conserve and where con-servation is needed.

Recent research on leucaena systematics confirmsthat sound taxonomy is essential for the wise use,improvement and conservation of leucaena geneticresources. A new taxonomic revision is being pre-pared to tackle the problems of species identificationthat underlie all leucaena research and use.

To be accepted and useful, the revision must notonly be accurate systematically, but must promptgeneral consensus and be user-friendly. Consensusis essential to avoid continued confusion and on-going debate over species names. Stable nomencla-ture is needed to support all other research on thegenus. If we present the results of systematicresearch before completing the new taxonomicrevision, there should be time for further discussionand debate. Alongside a formal taxonomic revision,a user-friendly guide to the identification of speciesis essential, so that it can be widely distributed andused by non-botanists.

63

Acknowledgments

Funding for these studies was provided by theForestry Research Programme of the OverseasDevelopment Administration, ODA, UK throughresearch schemes R.4524 (Intensive Study ofLeucaena Genetic Resources in Mexico and CentralAmerica) and R.4727 (A Molecular Study of theGenetic Resources of the Genus Leucaena). Theconsiderable support of botanical and forestryorganisations throughout Mexico and CentralAmerica in all field work is gratefully acknowledged.Particular thanks are extended to staff at theNational Herbarium (MEXU) of the UniversidadNational Autonóma de Mexico in Mexico City andthe ODA-assisted Forest Improvement and Conser-vation project, CONSEFORH, in Honduras.Thanks are also due to Sergio Zárate in Mexico andCharles Sorensson in Hawaii, who through their freeexchange of ideas on leucaena systematics have con-tributed to our understanding of the genus.

References

Abbott, R.J. 1992. Plant invasions, interspecific hybrid-isation and the evolution of new taxa. Trends in Ecologyand Evolution, 7, 401-405.

Abbott, R-J., Ashton, P.A. and Forbes, D.G. 1992.Introgressive origin of the radiate groundsel, Seneciovulgaris L. var. hibernicus Syme. Heredity, 68, 425-435.

Anderson, E. 1949. Introgressive Hybridisation. NewYork, John Wiley.

Ashton, P.A. and Abbott, R.J. 1992. Multiple origins andgenetic diversity in the newly arisen allopolyploid speciesSenecio cambrensis Rosser (Compositae). Heredity, 68,25-32.

Bentham, G. 1842. Notes on Mimoseae, with a shortsynopsis of species. Journal Botany (Hooker), 4,416-417.

-- 1875. Revision of the suborder Mimoseae. Trans-actions of the Linnaean Society of London, 30, 335-668.

Brewbaker, J.L. 1980. What is giant Leucaena? LeucaenaResearch Reports, 1, 43-44.

-- 1983. Systematics, self-incompatibility, breedingsystems and genetic improvement of Leucaena species.In: Leucaena Research in the Asia-Pacific Region,Workshop Proceedings. Singapore, 17-22.

- - 1987. Species in the genus Leucaena. LeucaenaResearch Reports, 7, 6-20.

Brewbaker, J.L. and Ito, G.M. 1980. Taxonomic studiesof the genus Leucaena. Leucaena Research Reports, 1,41-42.

Brewbaker, J.L. and Sorensson, C.T. 1990. Leucaena: newtree crops from interspecific hybrids. In: Janick, J. andSimon, J., ed., Advances in New Crops. Oregon. TimberPress.

Britton, N.L. and Rose, J.N. 1928. Mimosaceae. NorthAmerican Flora, 23, 121-131.

Brown, A.D.H. 1979. Enzyme polymorphisms in plantpopulations. Theoretical Population Biology, 15, l-42.

Casas, F. A. 1992. Etnobotanica y procesos de domes-ticacion en Leucaena esculenta (Mot & Sesse ex DC)Benth. MSc thesis, UNAM, Mexico City, Mexico.

Chamberlain, J.C. 1993. The Taxonomy and PopulationStructure of Leucaena Benth. and Gliricidia H.B.K.DPhil thesis, Cambridge, University of Cambridge,210 pp.

Chase, M.W. and Hills, H.H. 1991. Silica gel: an idealmaterial for field preservation of leaf samples for DNAstudies. Taxon, 40, 215-220.

Falk, D.A. and Holsinger, K.E. 1991. Genetics and Con-servation of Rare Plants. Oxford, Oxford UniversityPress.

Filer, D.L. 1993. BRAHMS: Botanical Research andHerbarium Management System, Reference Manual,Version 3.1. Oxford, University of Oxford.

Gallez, G.P. and Gottlieb, L.D. 1982. Genetic evidencefor the hybrid origin of the diploid plant Stephanomeriadiegensis. Evolution, 36, 1158-1167.

Glumac, E.L., Felker, P. and Reyes, I. 1987. A com-parison of cold tolerance and biomass production inLeucaena leucocephala, L. pulverulenta and L. retusa.Forest Ecology and Management, 18, 251-271.

Gottlieb, L.D. 1977. Electrophoretic evidence and plantsystematics. Annals Missouri Botanical Garden, 64,161-180.

Grant, V. 1981. Plant Speciation. New York, ColumbiaUniversity Press.

Halliday, J. and Somasegaran, P. 1983. Nodulation,nitrogen fixation and Rhizobium strain affinities in thegenus Leucaena. In: Leucaena Research in the Asia-Pacific Region. Workshop Proceedings, Singapore.Ottawa, IDRC, 27-32.

Harris, S.A., Hughes, C.E., Ingram, R. and Abbot, R. J.1994a. A phylogenetic analysis of Leucaena(Leguminosae: Mimosoideae). Plant Systematics andEvolution, 191, l-26.

Harris, S.A., Hughes, C.E., Abbott, R. J. and Ingram,R. 1994b. Genetic variation in Leucaena leucocephala(Lam.) de Wit. (Leguminosae: Mimosoideae). SilvaeGenetica, 43, 159-167.

Harris, S.A., Chamberlain, J.L. and Hughes, C.E. 1994c.New insights into the evolution of Leucaena Bentham(Leguminosae: Mimosoideae). In: Pickersgill, B. andPolhill, R.M., ed., Advances in Legume Systematics.Kew, Royal Botanic Gardens, in press.

Hellin, J.J. and Hughes, C.E. 1994. Leucaena saiva-dorensis: utilization and conservation in CentralAmerica. Misc Publ. Siguatepeque, Honduras,CONSEFORH.

Hughes, C.E. 1986. A new Leucaena species fromGuatemala. Leucaena Research Reports, 7, 110-113.

— 1988. Leucaena salvadorensis: status, distribution andgenetic conservation. Leucaena Research Reports, 9,99-105.

— 1991. Two new subspecies of Leucaena (LeguminosaeMimosoideae) from Guatemala. Kew Bulletin, 46,

547-557.— 1993. Leucaena Genetic Resources: the OFI seed col-

lections and a synopsis of species characteristics. Oxford,Oxford Forestry Institute, 117 pp.

64

Hughes, C.E. and Harris, S.A. 1994. Identification andcharacterisation of a naturally occurring hybrid inLeucaena (Leguminosae: Mimosoideae). PlantSystematics and Evolution, 192, 177-197.

Macbride, J.F. 1943. Flora of Peru. Family Leguminosae.Botanical Series Field Museum of Natural History, 8,98-100.

MacNeish, R.S. 1965. The origins of AmericanAgriculture. Antiquity, 39, 87-94.

McVaugh, R. 1987. Flora Novo - Galiciana. A descriptiveaccount of the vascular plants of western Mexico.Volume 5, Leguminosae. Ann Arbor, University ofMichigan Press, 786 pp.

Pan, F.J. 1985. Systematics and Genetics of the Leucaenadiversifolia complex. PhD thesis, University of Hawaii.

— 1988. Comparison of diploid and tetraploid Leucaenadiversifolia. Quarterly Journal of Chinese Forestry, 21,89-98.

Pan, F.J. and Brewbaker, J.L. 1988. Cytological studiesin the genus Leucaena. Cytologia, 53, 393-399.

Rieseberg, L.H., Beckstrom-Sternberg, S. and Doan, K.1990. Helianthus annus subsp. texanus has chloroplastDNA and nuclear ribosomal genes of Helianthus debilissubsp. cucumerifolius. Proceedings National Academyof Sciences USA, 87, 593-597.

Schery. 1950. Flora of Panama. Leucaena. Annals of theMissouri Botanical Garden, 37, 302-304.

Smith, C.E. 1967. Plant Remains. In: Byers, D.S., ed.,The Prehistory of the Tehuacan Valley. Vol. 1. Environ-ment and Subsistence. Austin, USA, University of TexasPress.

Soltis, D.E. and Soltis, P.S. 1990. Isozymes in PlantBiology. London, Chapman and Hall.

Sorensson, C.T. 1992. Strategies for small scale Leucaenahybrid seed production. Proceedings IUFRO ConferenceS.2.02.08, Cali, Colombia, November 1992, in press.

Sorensson, C.T. and Brewbaker, J.L. 1994. Interspecificcompatibility among 15 Leucaena species (Leguminosae:Mimosoideae) via artificial hybridizations. AmericanJournal of Botany. 81(2), 240-247.

State, C.A. 1975. Hybridization and the Flora of theBritish Isles. London, Academic Press.

Standley, P.C. 1922. Trees and shrubs of Mexico. Con-tributions to the U.S. National Herbarium, 23, 366-369.

Standley, P.C. and Steyermark, J.A. 1946. Flora ofGuatemala, Part V. Fieldiana Botany, 24, 46-48.

Stebbins, G.L. 1959. The role of hybridization in evolution.Proceedings American Philosophical Society, 103,231-251.

Stewart, J.L., Dunsdon, A.J. and Hughes, C.E. 1991.Early biomass production of Leucaena species inComayagua, Honduras. Leucaena Research Reports, 12,101-104.

Whitaker, T.W. and Cutler, H.C. 1966. Food plants ina Mexican market. Economic Botany, 20, 6-16.

Zárate, S.P. 1982. Las especies de Leucaena Benth. deOaxaca con notas sobre la sistematica del genero paraMexico. MSc thesis, UNAM, Mexico.

— 1984a. Taxonomic revision of Leuceana Benth. fromMexico. Bulletin of the International Group for theStudy of Mimosoideae, 12, 24-34.

— 1984b. Domestication incipiente del ‘Guaje Zacatzin’.Resumenes 9 Congreso Mexicano de Botanica, MexicoCity, Sociedad Botanica Mexicana.

— 1987a. Taxonomic identity of Leucaena leucocephala(Lam.) de Wit with a new combination. Phytologia, 63,304-306.

— 1987b. Clarification of the taxonomy of Leucaenasalvadorensis Standley ex Britton & Rose. AnnalsMissouri Botanical Garden, 74, 449.

— 1994. Revision del genera leucaena en Mexico. AnalesInst. Biol. Univ. Auton. Mexico, Ser. Bot. 65(2), 83-162.

65

Leucaena Germplasm Collections, Genetic Conservationand Seed Increase

C.E. Hughes1, C.T. Sorensson2, R. Bray3 and J.L. Brewbaker2

Abstract

The status of leucaena germplasm collections is reviewed, showing that leucaena has beenmore thoroughly explored and collected than most tropical woody genera. Despite comprehensivegermplasm collecting, genetic conservation has been neglected and the status of the geneticresources of leucaena species in situ is shown to be very degraded. The paper describes exsitu genetic conservation programs and reveals that only a small fraction of the genetic variationavailable in the genus is currently conserved. Finally, seed production for planting is discussed,demonstrating the need for expanded efforts in seed increase and deployment of improvedgenetic material.

Exploration of Leucaena Genetic Resources

Historical spread of Leucaena germplasm

The genetic resources of leucaena have been underscrutiny for at least the last two millennia as a sourceof useful plants. Archaeological evidence suggeststhat Aztec, Mixtec, Maya, Toltec and Zapotecpeoples used leucaena seeds and pods as humanfood up to two or three thousand years ago in partsof south-central Mexico (Smith 1967; Zárate 1984a;Casas 1992). Thus leucaena seed was already beingcollected and transported within Mexico in pre-Colombian times and there is evidence to suggestthat the distributions of widely used species suchas L. esculenta and L. leucocephala were greatlyextended by this process of indigenous domesti-cation. Seed of a wide range of Leucaena speciescontinues to be transported within Mexico today bylocal people who are still cultivating trees for podproduction (Whitaker and Cutler 1966; Zárate1984b; Casas 1992; Harris et al. in press; Hughesand Harris these Proceedings).

1 Oxford Forestry Institute, Department of Plant Sciences,University of Oxford, South Parks Road, Oxford, OX13RB, UK2 Department of Horticulture, College of Tropical Agri-culture, University of Hawaii at Manoa, 3190 Maile Way,Honolulu, Hawaii 96822, USA3 CSIRO, 306 Carmody Road, St. Lucia, Brisbane,Queensland 4067, Australia

Spread of leucaena around the globe to the OldWorld occurred as early as the sixteenth centurywhen Spanish colonists introduced one species, L.leucocephala subsp. leucocephala to the Philippines.It is not clear whether this introduction wasdeliberate or accidental (Brewbaker 1987; Zárate1987). The same species was also in cultivation inEuropean botanic gardens in the first part of theeighteenth century or earlier. This subspecies of L.leucocephala, which corresponds to the shrubby orcommon variety, continued its spread throughoutthe tropics and is now extensively naturalised andweedy in many tropical countries (Hughes and Styles1989; Cronk and Fuller 1995). Recent isozymestudies indicate that this subspecies is representedby a single self-pollinated variety oufside of its nativerange in the Americas.

Several other species of Leucaena were introducedinto Old World cultivation as shade trees over coffeeor cacao about 100 years ago. For instance, L.pulverulenta and L. diversifolia were introduced byDutch foresters to Indonesia at the end of thenineteenth century (Djikman 1950), and L. diversi-folia was introduced to West Africa in Cameroonand Ivory Coast, and probably also to Jamaica inthe Caribbean.

Modern germplasm collections

When leucaena became an important forage plantin the 1960s, systematic germplasm collections

66

started. Since that time, three major germplasmcollections of Leucaena species have been assembledby the University of Hawaii (USA), CSIRO (Aus-tralia) and OFI (UK). These collections aresummarised in Table 1 and described in detail below.As well as these three international collections thereare many national leucaena germplasm collections,which vary in size and level of activity, and webriefly discuss these also.

University of Hawaii

In 1962, the University of Hawaii (UH) startedassembling the first germplasm collection ofleucaena, initially based on miscellaneous seedacquired world-wide from the various non-nativesources where leucaena was cultivated. Early col-lections were dominated by the one species ofgreatest interest at that time, L. leucocephala, andby seed from non-native sources. Organised effortsto explore and collect seed from the naturalpopulations of leucaena began in 1967 when Brew-baker led a UH team to Mexico and other coun-tries in Latin America. For the first time this col-lection gathered a range of Leucaena species otherthan L. leucocephala. Germplasm collections by UHresearchers, led by Brewbaker, continued intermit-tently over the next 20 years. Major expeditions in1977, 1978, 1985 and 1988 sampled natural popu-lations throughout Latin America (Brewbaker andSorensson 1994). By 1988 a large collection -almost 1000 accessions, representing most taxa -had been assembled and grown in Hawaii. It wasstill dominated by L. leucocephala which accountedfor 541 out of 967 accessions.

Seedlots are documented with basic passport dataheld on a spread sheet database using the ‘K’number series running chronologically from K1 toK967. Seed from the Hawaii collection has been dis-tributed for testing and includes the well-known andwidely planted K8 and K636 varieties of L. leuco-cephala and K156 variety of L. diversifolia. Thiscollection also forms the core of several other largegermplasm collections (approx. 90% of the UnitedStates Department of Agriculture (USDA) collectionand a significant proportion of the National SeedStorage Laboratory (NSSL), International LivestockCentre for Africa, Ethiopia (ILCA) and IndianGrassland and Fodder Research Institute (IGFRI)collections). Identities are verified from livingcollections in Hawaii.

CSIRO

Between the mid 1950s and the mid 1970s CSIROassembled a largely opportunistic collection of 200accessions, mainly L. leucocephala, from non-native

sources. Extensive targeted collections were made,principally in Mexico in the late 1970s and early1980s by R.Reid (working with S. Zárate) mainlyof species other than L. leucocephala. Passport datafor the collections is documented under the CPI(Commonwealth Plant Introduction) numberingsystem, and stored by the Australian TropicalForages Genetic Resource Centre in a UNIX-baseddata management system, designed for the purpose.

Oxford Forestry Institute

In the mid 1980s, OFI started to assemble a newleucaena seed collection, for two reasons. Firstly,several little known and potentially valuable specieshad been discovered in Guatemala and Honduras,apparently having been overlooked by previous col-lection expeditions. Secondly, OFI realised thatseveral of these species were already severelydegraded and under threat of substantial geneticerosion or even extinction (Hughes 1986, 1988;Hellin and Hughes 1994). The main objective of theOFI collection program was to assemble all Leucaenaspecies, but to concentrate on the lesser-knownspecies. From 1984 to 1992, a series of annual expedi-tions to Mexico and Central America madecollections in collaboration with the Forest Authori-ties and tree seed banks in the regions. These col-lections are now complete and are documented byHughes (1993). Identities are verified from botanicalvoucher specimens deposited in major herbaria inMexico, USA and Europe.

National Collections

National collections or collections at regional centressuch as ILCA or Centro International de Agri-cultura Tropical (CIAT), are largely derived fromthe three large international collections listed above.As well, several contain small numbers of uniqueaccessions from their local areas. A good exampleis the diploid L. diversifolia collection RSB0l fromIndonesia (Oka 1990). Within Mexico and CentralAmerica several agencies maintain seed collectionsof local species based on bulk collections fromnatural populations. The Forest Tree Seed Banksin Guatemala, Honduras and Nicaragua are notablein this respect. They regularly collect bulk seed fromnatural stands of a wide range of native Leucaenaspecies, for direct use in tree planting projects.

Seed collection strategies

The international leucaena germplasm collectionshave used different sampling strategies because theyhave radically differing objectives. Methods of col-lecting agricultural crop germplasm are not the same

67

as methods of collecting germplasm for forest treeimprovement.

The UH and CSIRO collections were madeessentially as for any crop germplasm. The collec-tion expeditions were short, they sampled widelyacross localities and included small but variablenumbers of parent trees. The resulting seedlots weretermed ‘accessions’. An accession can thus beprogeny from a single parent tree or from severaltrees, but rarely includes more than ten parents. Seedfrom individual trees cannot be identified, exceptfor self-fertile tetraploid species. This approachmakes little attempt to select particular trees withinnatural populations, and does not assemblerepresentative population samples, but it allows cost-effective sampling of as many sites as possible withinthe time available. It ensures coverage of as broada range of environments as possible by sampling inrelation to soil changes and microclimate. Unrepli-cated arboretum-style trials are used to initiallyscreen the many accessions collected. Further evalu-ation and breeding then relies on seed increase fromsuperior accessions.

In contrast, forest tree improvement programs(OFI collections) are largely directed to out-crossingspecies. They follow highly structured samplingstrategies and collect seed from natural populations.During the longer seed collection expeditions, largenumbers of parent trees are sampled across largepopulations, avoiding neighbouring trees. Bulk col-lections termed ‘provenances’ are assembled. Thesecollections usually include a minimum of 25 parenttrees and often up to 50-60, and are expected tobe genetically diverse and amenable to subsequentselection and breeding. The seed source can beidentified and provenance variation can be used asa starting point in any improvement program. Inmany cases, individual tree seedlots are also collectedat the same time. Such an approach, with main-tenance of family identity, allows family-based half-sib progeny trials (out-crossing species) and seedorchards to be established. It aims to maintain amuch stricter level of control over pedigree and amuch broader genetic base within provenances.

Status of Leucaena germplasm collections

Table 1 summarises the numbers of accessions inthe main germplasm collections of leucaena (asimplified taxonomy has been used to summarisegermplasm, with no subspecies included). The manyhybrid accessions in the UH collection are not inthis list, but hybrids in the NSSL and ILCA col-lections are placed in the spp. category. In the OFIcollection, the first numeral indicates the numberof provenances held and the second numeral thenumber of individual tree or half-sib familyaccessions.

Across all the collections there is clearly signifi-cant duplication, roughly estimated here as morethan 25%. This means that the total number oforiginal or unique accessions is almost certain to beconsiderably less than 3000. Internal duplication isessentially zero in all collections. In the table, dupli-cation refers to cases where the same seedlot isrepresented in two or more collections, whichhappens when material is exchanged and distributed.(Figures are broad estimates, not based on detailedinvestigations.) For example, the USDA collectionis largely derived from the UH collection, so dupli-cation is probably over 90% between these two. TheUH and CSIRO collections also have some seedlotsin common. Duplication can also arise whendifferent organisations independently collectmaterial from the same site and even the samepopulation. We do not know how often this happensnor how important it is.

The new OFI collections have diminished thedomination by L. leucocephala, but collections fromthis species still make up about 50% of the totalin Table 1. Many of the L. leucocephala accessionsare known to be genetically uniform and thereforealmost redundant, because of the marked lack ofgenetic variation that has been found within thespecies, and particularly within the shrubby sub-species leucocephala. Since the advent of the psyllid,many of the early collections of L. leucocephalahave become obsolete, especially now that moderncollections include a wider range of species and moregermplasm from natural populations. In contrast,the much higher level of genetic variation knownto exist in species such as L. diversifoiia and othersis only sparsely represented in the collections (Table1). A few species such as L. confertiflora and L.cuspidata are represented by only a handful ofaccessions.

There are limited quantities of seed for many ofthe accessions, in all but the OF1 collection. Forexample, more than 500 of the 967 accessions heldby the University of Hawaii have only 100 or fewerseeds in stock. Although CSIRO aims to maintain200 g of seed available for all its accessions, it hassucceeded only for self-fertile types. A similarsituation applies to the USDA and some othercollections. The NSSL collections may be regardedas being in permanent long-term storage and are ‘notaccessible’. This means that most of the accessionslisted in Table 1 need a major program ofregeneration to make them available for widespreaduse and testing. Larger seed quantities are availablefor the remainder of the accessions and for themajority of the OFI collection.

Considerable confusion surrounds the identity ofmaterial in many collections, partly because thereare several numbering systems. The three major

68

Table 1. Major Leucaena germplasm collections showing numbers of accessions.

Species UH USDA4 OFI1 C S I R O N S S L 5 ILCA IGFRI 3 TOTAL

collinsiiconfertifloracuspidatadiversifoliaesculentagreggiilanceolataleucocephalamacrophyllamulticapitulapallida2

pulverulentaretusasalvadorensisshannoniitrichodesspp.total% duplication

37 19 5(113) 10 - 4 11 1942 - 5(18) 4 - - - 24

- 1 5(20) 2 - 1 - 24118 53 19(168) 27 1 21 29 41?55 23 3(48) 13 2 3 10 15432 7 3(50) 9 - 1 30 12947 35 7(83) 41 1 3 35 245

541 493 19(78) 590 125 102 268 219717 6 4(30) 12 86

3 - 2(20)19

—2

2 2520 12 6(50) 15 - 10 1 10818 14 3(33) 45 4 6 10 13018 4 1(10) 13 2 12 60

3 1 6(123) — — 1 10 13833 23 6(130) 12 1 4 55 25823 4 2(50) 1 11 104- 14 3(92)

1 5- 5 13 - 124

967 709 1116 815 140 174 496 4417?10 ?10 0 ?30 ?100 ?100 ?100 ?>25

UH = University of Hawaii; USDA = United States Department of Agriculture; OFI = Oxford Forestry Institute; CSIRO = Com-monwealth Scientific and Industrial Research Organisation; NSSL = National Seed Storage Laboratory, USA; ILCA = InternationalLivestock Centre for Africa, Ethiopia; IGFRI = Indian Grassland and Fodder Research Institute, India1 OFI: first numeral indicates number of provenances, second numeral number of individual tree or half-sib family accessions2 L. pallida is sometimes referred to as L. esculenta subsp. paniculata3 From Gupta (1990)4 G.A. White, USDA, pers. comm. 19925 S.A. Eberhart, NSSL, pers. comm. 1988. Long-term, essentially permanent storage, ‘not accessible’.

independent international germplasm collectionseach have their own numbering systems, and severalnational and other collections also have their ownnumbering systems, even though their material ispartially derived from the three major collectionsor from each other. To confuse identification evenmore, no-one has accurately assessed levels of dupli-cation across collections, nor levels of redundancyin the collections due to low seed quantities oruniformity of genetic material. Clearly, individualcollections need to be rationalised and the majorcollections need to be catalogued, with cross-referencing of identities and duplicate collections,to produce a World Germplasm Catalogue forleucaena.

Practical aspects of leucaena seed collection andstorage

In general, Leucaena species produce large quantitiesof viable seed at an early age. Most species are lessprolific, less precocious and more seasonal than L.leucocephala in their seed production. A few species,such as L. salvadorensis, may not produce signifi-cant quantities of seed until 3-4 years of age. Seed

collection and extraction methods are straight-forward. For most species, it is critical to time seedcollection to coincide with peak ripeness. Seedpredators can cause major losses in seed collectedfrom the natural populations and seed beetles arealso a problem in other areas, such as in Hawaii.Some areas, such as Australia, apparently have noseed predation problem. Effective methods forseparating beetle-infested seed from good seed havebeen developed using flotation in salt water(Sorensson 1993) and gravity table methods (Hughes1993).

Leucaena seed retains high viability for severaldecades when stored under normal conditions of8-10% moisture content at +4°C. Seed viability inlong-term storage at lower temperatures (–20°C)has not been tested but, given the hard seed coatof most leucaena species, seed can probably bestored safely for long periods. Seed stored underpoor conditions of high moisture content and tem-perature can rapidly lose viability. While mostorganisations with large seed collections of Leucaenaspecies have access to adequate seed storagefacilities, cold storage facilities are still inadequatein some countries.

69

Leucaena Genetic Conservation

Status of Leucaena genetic resources

Leucaena species are distributed mainly in theseasonally dry tropical and upland forests of Mexicoand Central America. More than 98% of thetropical dry forest type in this region has alreadydisappeared (Nations and Kramer 1983) and lessthan 0.08% lies in any form of biological reserve(Janzen 1986). Tree cover is thus reduced to frag-mented remnants, scattered trees in fence lines,around houses and in inaccessible areas such as steepgullies or cliffs. Even these last remnants of forestare being continually degraded by further clearanceof land for agriculture, dry season fires andextensive overgrazing in some areas.

Without exception, Leucaena species have sufferedsome level of genetic degradation. This means thatall the germplasm collections outlined above havebeen assembled from disturbed populations. Fewpristine populations of Leucaena species remain.Leucaena species are susceptible to grazing and manyhave been progressively reduced because of the hugeincrease in grazing pressure since goats and cattlewere introduced to Mexico by the Spanish in the six-teenth century. This is particularly apparent for themid elevation and highland species of the Mexicanmesetas including L. retusa, L. greggii, L. cuspidata,L. pallida and L. confertiflora. These species are nowoften restricted to steep gullies and other areas thatgrazing animals cannot reach.

For a few species, particularly those withrestricted distributions, degradation has now reachedcritical levels. Active measures are needed to con-serve the genetic resources of these species. Forexample, L. shannonii subsp. magnifica, a taxonrestricted to a small area of south-east Guatemala,is now apparently reduced to fewer than 400individuals (Hughes 1986, 1991). Similarly, L. salva-dorensis, a species native to restricted areas ineastern El Salvador, southern Honduras andnorthern Nicaragua, has been reduced to small,scattered and fragmented stands (Hughes 1988;Hellin and Hughes 1994). Seed collections of L.salvadorensis carried out in these areas have beenforced to cover large areas to include sufficientparent trees; for example, inclusion of sixty parenttrees required collection over more than 100 km2

in northern Nicaragua in 1991.

Humans have also interfered with the geneticresources of Leucaena species by the processesinvolved in indigenous domestication. Many speciesof Leucaena are now protected and managed, theirseed is harvested and, in large parts of south-centralMexico, cultivated. Levels of genetic diversity inthese populations have not been widely investigated

but might be expected to be lower than in corres-ponding natural populations. Casas (1992) presentedevidence that pod morphology showed reducedphenotypic diversity in cultivated populations of L.esculenta and interprets this as a sign of pastselection by local people for favourable podcharacteristics.

Ex situ genetic conservation

Ex situ genetic conservation refers to conservationof species outside the natural ecosystem in whichthey originally evolved. Traditionally it involveslong-term seed storage, and different types of livingcollections including arboreta, botanic gardens, con-servation stands and seed production areas. Atpresent there are few programs specifically directedtowards ex situ genetic conservation of Leucaenaspecies, so at present the genetic variation in thegenus is not well conserved in any form of ex situconservation.

All but one of the large seed collections ofleucaena are aimed at short or medium term storage.These ‘working collections’ are kept under normalseed storage conditions of low humidity and tem-perature where viability remains high for 30 or moreyears. The only substantial collection in long-termstorage, at very low temperatures, is held by theUSDA National Seed Storage Laboratory, NSSL,at Fort Collins, USA (see Table 1). This collectionincludes 140 accessions, of which 125 are of L. leu-cocephala, all of which are held jointly by the UHand USDA regional working collections. Thus onlya small fraction of the genetic diversity in the genusis represented in long-term storage and most spe-cies are not included at all. Further, the materialin long-term storage bears no relation to the con-servation status of species in their native ranges; themost threatened taxa are not represented. Smallercollections of leucaena are held in other long-termstores such as those of the National Bureau of PlantGenetic Resources, New Delhi, India and theCATIE long-term seed store in Costa Rica. Theproblems of regeneration of long-term leucaena seedcollections have not been tackled so far.

As with most of the seed collections, the ex situliving collections of Leucaena species are notspecifically directed at long-term genetic conser-vation. There are many arboretum-style collections,in Australia, Brazil, Colombia, Ethiopia, Hawaii,Honduras, India, Indonesia, Taiwan and otherplaces. In every case these are working collectionsor unreplicated evaluation trials, not conservationcollections. All are relatively insecure because theylack resources and support for maintenance and aresusceptible to sporadic disasters such as fire,

70

hurricane or urban development. The amount ofgenetic diversity that can be conserved is limitedagain by lack of resources such as land; in generalspecies are represented by few trees per plot and fewaccessions or provenances. Finally, the geneticintegrity of seed collected from mixed-species livingcollections cannot be guaranteed, due to high levelsof interspecific compatibility and hybridisation.

There are a few examples of ex situ plantings inthe form of conservation stands or seed productionareas or seed orchards where larger amounts ofgenetic variation can be included in isolated blocksguaranteeing production of pure seed. A goodexample of this type of approach is the ex situ seedorchard program being undertaken in Honduras bythe CONSEFORH project for L. salvadorensis(Ponce, these Proceedings).

Thus at present the genetic variation in Leucaenaspecies is not well conserved in ex situ seedcollections or living collections.

In situ genetic conservation

In situ genetic conservation refers to conservationof species within the ecosystem in which theyoriginally evolved. Traditionally it has been confinedto protected areas such as national parks and naturereserves. As indicated above, only a small fractionof the total dry tropical forest of Mexico and CentralAmerica is protected in any form of biologicalreserve. Most species of Leucaena are not presentin any of the small and scattered reserves that doexist. There are a few exceptions. For example, L.lanceolata occurs in the Chamela reserve in Jalisco,Mexico, and L. retusa is found in the Big BendNational Park in Texas, USA. Biological reservescannot therefore be expected to make more than avery minor contribution to conservation of Leucaenagenetic resources even at species level, let alone con-servation of intraspecific genetic diversity.

Farmer-based conservation

Now that the limitations of ex situ conservation andtraditional in situ biological reserves for crop geneticresources have been realised, farmer-based con-servation is seen as making an important potentialcontribution to the overall conservation effort(Brush 1991; Cohen et al. 1991). Farmer-based con-servation may be considered as a form of in situconservation through use. Although only advocatedso far for crop plants, it appears to have consider-able potential for agroforestry trees such as Leucaenaspecies (Hughes 1993; Hellin and Hughes 1994). Thefact that farmers have been actively conserving awide range of Leucaena species over the centuriesby protecting, managing and cultivating trees, indi-cates the potential of farmer-based conservation for

leucaena. For many species, such as L. diversifolia,L. esculenta, L. pallida and L. salvadorensis, therelative abundance of trees in the present day isentirely due to farmer protection and cultivation.Conservation through use can operate over largeareas and conserve large amounts of genetic varia-tion. It minimises the effects of sporadic disasterssuch as fires and requires only limited investmentof resources because the process becomes self-sustaining. As mentioned above, indigenous domes-tication may have led to some reduction in geneticvariation but little compared to the drastic lossesof genetic variation in species of Leucaena fromnatural vegetation and current ex situ conservation.A strategy for farmer-based conservation of one spe-cies, L. salvadorensis, was elaborated by Hellin andHughes (1994) and appears to have the potential tobe applied to most, if not all, Leucaena species.Active support by donors to tree planting organi-sations, especially non-government organisations, inMexico and Central America could help to promotewider use of local leucaena species in situ with sub-stantial and cost-effective conservation benefits.

Genetic conservation strategy

Current reliance on short or medium term seedstorage and active breeding or arboretum collectionsfor genetic conservation needs to change in favourof measures specifically directed to genetic conser-vation. Farmer-based conservation in agroforestrysystems provides a viable alternative in situ conser-vation method for leucaena. In situ and ex situ con-servation are compatible and complementary andneed to be used in tandem to provide a reliablegenetic conservation strategy for leucaena.Expanded ex situ conservation programs, that incor-porate more genetic variation into long-term seedstorage and viable conservation stands, are neededto overcome the limitations of current ex situ efforts.

Seed Increase for Deployment

Tree seed supply is a critical factor in the successof agroforestry and other tree planting programs(Turnbull 1983). Abundant seed has been animportant factor contributing to the spread andwidespread adoption of L. leucocephala. For L.leucocephala, early and ample seed productionthrough self-pollination, giving true-bred material,meant that seed was spread extremely easily fromfarmer to farmer. Seed of most other Leucaenaspecies, most of which are self-sterile, is not avail-able in bulk or semi-bulk quantities at present andproblems of producing pure seed are much greaterfor out-crossing species. Now that interest is focusedon other species and hybrids, demand for their seedis likely to rise, but limited quantities of seed are

71

available. As a result, adoption of less-knownspecies is not likely to be widespread.

Seed collections for routine planting, at least ofself-incompatible species, should not be made fromtrials or arboreta. Research in Hawaii has shownthat there are few quantitative genetic barriers tointerspecific hybridisation in Leucaena (Sorenssonand Brewbaker 1994). This means that when speciesare brought into close proximity in cultivation, ashappens in trials and arboretum collections, therewill be many opportunities for the spontaneousproduction of hybrids. Therefore the geneticintegrity of seed collected from such areas cannotbe guaranteed, except for the self-pollinated speciesand hybrids. Further, the small plots in trials orarboreta contain too few parent trees to provide anadequate genetic base for routine seed productionactivities.

The natural populations in Mexico and CentralAmerica provide the only immediately availablesource of bulk seed supplies for most species. Foresttree seed banks in Nicaragua, Honduras andGuatemala regularly supply bulk seed of a range ofnative Leucaena species. In the medium term, localseed self-sufficiency is a pre-requisite for success innon-industrial forestry. Careful thought must begiven to the distribution of new species, hybrids andimproved or select genetic material. It is notable thatdespite considerable effort to breed improvedmaterial of leucaena, only a handful of select self-compatible varieties of L. leucocephala and the bredCunningham variety have been successfully deployedfor widespread use, because these varieties producelarge amounts of seed in small plots. For provenout-crossing species and interspecific hybrids, thereis a real and immediate need for seed sources whichare soundly based, well adapted and from knownprovenances.

Seed increase of the self-fertile polyploid speciessuch as L. leucocephala and L. diversifolia subsp.diversifolia presents no major obstacles. However,for out-crossing species, broadly-based compositeseed orchards need to be established to obtain goodseed production. A small seed orchard of 100 treeswas established in Australia with considerablesuccess, using a mixture of L. pallida accessions toproduce composite seed. However, it was difficultto produce seed of the single superior diploid L.diversifolia line CP146568, due to scant seed pro-duction. Seed orchards of L. salvadorensis havebeen established in Honduras (Ponce, theseProceedings) and Nicaragua following the BreedingSeedling Orchard design, BSO (Gibson 1993). Inthis, up to 50 half-sib families are planted at closespacing in a replicated design, with early assessmentand progressive selective thinning to wide spacing

for seed production. So far these trees haveproduced little seed.

The problem of seed increase was clearlyacknowledged within the leucaena psyllid trial (LPT)program, coordinated by the University of Hawaii.In 1989, a follow-up leucaena seed production (LSP)program established seed production areas for arange of psyllid tolerant species and hybrids in fivecountries, namely India, Indonesia, Philippines,Taiwan and Thailand. Most of the seed producedfrom the LSP program has been from the self-fertilepolyploids such as the K636 variety of L. leu-cocephala. In the LSP, two orchards were estab-lished at each site, usually with cross sterile taxa.Seed was supplied for more than 500 trees, includinga minimum of 10 accessions, and it was recom-mended that orchards be established at some dis-tance from the LPT trials. Success of the LSP wasmixed and seed beetle infestations limited produc-tion of viable seed at some sites. The particularproblems associated with production of hybrid seedneed special attention and have been discussed bySorensson (1992). Seed production of sterile triploidsutilising cloned self-incompatible female parents wasdiscussed as a viable commercial option by Brew-baker and Sorensson (1990).

Conclusions and Priorities for Researchand Development

Germplasm collections

The genetic resources of Leucaena species have nowbeen more thoroughly explored and collected thanalmost any other tropical woody genus. There arethree major international germplasm collectionswhose evaluation and conservation should be ofmuch higher priority than new collection programs.

There is considerable confusion surrounding thedifferent collections. Their status and differentnumbering systems need to be clarified through thecompilation of a cross-referenced world germplasmcatalogue covering all collections.

Current germplasm collections need to bereorganised to avoid or clarify duplication andreduce large holdings of redundant material, mainlyof L. leucocephala.

Genetic conservation

Programs to conserve the genetic resources ofleucaena need to be significantly expanded. Agenetic conservation strategy needs to be developedfor all Leucaena species. It should employ com-plementary in situ and ex situ methods.

Indigenous farmers have successfully conservedmany species. Farmer-based conservation offers

72

greater potential than traditional in situ conservationin biological reserves. It could be greatly expandedto cover most, if not all species of Leucaena.

There will be significant benefits for conservationif breeding and conservation are made separateobjectives of ex situ living collections and seedstorage. Genetic conservation needs to be consideredin its own right, not as a by-product of evaluationand breeding.

Seed production for planting programs

Seed increase and the planting out of new speciesand hybrids need to be greatly expanded.Appropriate systems need to be designed andadopted to produce hybrid seed and seed of out-crossing species. Out-crossing species are the normin forest trees, so there is considerable forestryexperience in seed orchard design which needs tobe adapted to leucaena seed production.

Seed increase of self-pollinating species andhybrids (always polyploid) should seek to employmulti-line technologies to broaden the genetic basein growers’ fields.

Acknowledgments

The authors gratefully acknowledge financialsupport for the OFI Leucaena Program from theForestry Research Programme of the UK OverseasDevelopment Administration. Charles Sorensson’sparticipation at the workshop was supported byACIAR and USAID through the Forest ResourcesManagement Program (II) in cooperation with theUSDA Forest Service.

References

Brewbaker, J.L. 1987. Leucaena: a multipurpose tree genusfor tropical agroforestry. In: Steppler, H.A. and Nair,P.K., ed., Agroforestry: A Decade of Development.ICRAF, Nairobi, Kenya.

Brewbaker, J.L. and Sorensson, C.T. 1990. New tree cropsfrom interspecific Leucaena hybrids. In: Janick, J. andSimon, J.E., ed., Advances in New Crops. Timber Press,Portland, Oregon, USA, 283-289:

— 1994. Domestication of lesser-known species of thegenus Leucaena. In: Leakey, R. and Newton, A., ed.,Tropical Trees—the Potential for Domestication. Insti-tute of Terrestrial Ecology, Edinburgh, Scotland.

Brush, S.B. 1991. A farmer-based approach to conservingcrop germplasm. Economic Botany, 45, 153-165.

Casas, F.A. 1992. Etnobotanica y procesos dedomestication en Leucaena esculenta (Mot. & Sessk exA.D.C.) Benth., MSc Thesis, UNAM, Mexico City,Mexico.

Cohen, J.I., Alcorn, J.B. and Potter, C.S. 1991. Utilizationand conservation of genetic resources: international

projects for sustainable agriculture. Economic Botany,45, 190-199.

Cronk, Q.C.B. and Fuller, J.L. 1995. Plant Invaders: Thethreat to Natural Ecosystems. Chapman and Hall.

Djikman, M.J. 1950. Leucaena—a promising soil-erosioncontrol plant. Economic Botany, 4, 337-349.

Gibson, G.L. 1993. Genetic conservation of native hard-woods in the degraded dry forests of Honduras. In:Resolving Tropical Forest Resource Concerns throughTree Improvement, Gene Conservation and Domesti-cation of New Species. Proceedings IUFRO Conference,Cali, Colombia, October 1992.

Gupta, V. 1990. Genetics and breeding of L. leucocephala.Unpublished, IGFRI, Jhansi, India.

Harris, S.A., Chamberlain, J.L. and Hughes, C.E. Newinsights into the evolution of Leucaena Benth.(Leguminosae: Mimosoideae). In: Pickersgill, B. andPolhill, R.M., ed., Advances in Legume Systematics,Royal Botanic Gardens, Kew. In press.

Hellin, J.J. and Hughes, C.E. 1994. Leucaena salva-dorensis: utilization and conservation in CentralAmerica. Miscellaneous Publication, CONSEFORH,Siguatepeque, Honduras.

Hughes, C.E. 1986. A new Leucaena from Guatemala. Leu-caena Research Reports, 7, 110-113.

— 1988. Leucaena salvadorensis status, distribution andgenetic conservation. Leucaena Research Reports, 9,99-105.

-- 1991. Two new subspecies of Leucaena (Leguminosae:Mimosoideae) from Guatemala. Kew Bulletin, 46,547-557.

-- 1993. Leucaena Genetic Resources: The OF1 Seed Col-lections and a Synopsis of Species Characteristics. OxfordForestry Institute, Oxford, UK.

Hughes, C.E. and Styles, B.T. 1989. Risks and benefitsof woody legume introductions. In: Stirton, C.H. andZarruchi, J.L., ed., Advances in Legume Biology.Monographs in Systematic Botany, Missouri BotanicalGarden, 29, 505-53 1.

Janzen, D.H. 1986. Guanacaste National Park: TropicalEcological and Cultural Restoration. Editorial, Univer-sidad Estatal a Distancia, San Jose, Costa Rica.

Nations, J.D. and Kramer, D.I. 1983. Central Americantropical forests: positive steps for survival. Ambio, 12,232-238.

Oka, I.N. 1990. Progress and future activities of theLeucaena psyllid research program in Indonesia. In:Napompeth, B. and MacDicken, K.G., ed., LeucaenaPsyllid: Problems and Management, WinrockInternational.

Smith, C.E. 1967. Plant Remains. In: Byers, D.S., ed., ThePrehistory of the Tehuacan Valley, Vol. 1. Environmentand Subsistence. Austin, USA, University of Texas Press.

Sorensson, C.T. 1993. Production and characterisation ofinterspecific hybrids of the tropical tree leucaena(Leguminosae: Mimosoideae). PhD thesis, University ofHawaii, USA.

— 1994. Seed productivity of artificial hybridizationsbetween fifteen species in the genus leucaena. In:Resolving Tropical Forest Resource Concerns throughTree Improvement, Gene Conservation and Domestica-tion of New Species. Proceedings IUFRO Conference,Cali, Colombia, October 1992, 359-365.

73

Sorensson, C.T. and Brewbaker, J.L. 1994. Interspecific Zárate, S.P. 1984a. Taxonomic revision of Leucaenacompatibility among 15 Leucaena species (Leguminosae: Benth. from Mexico. Bulletin of the International GroupMimosoideae) via artificial hybridization. American for the Study of Mimosoideae, 12, 24-34.Journal of Botany, 81(2), 240-247.

Turnbull, J.W. 1983. Tree seed supply — a critical factorfor the success of agroforestry projects. In: Burley, J.and von Carlowitz, P., ed., Multipurpose Tree Germ-plasm, ICRAF, Nairobi, Kenya.

Whitaker, T.W. and Cutler, H.C. 1966. Food plants ina Mexican market, Economic Botany, 20, 6-16.

- -1984b. Domesticacion incipiente del ‘Guaje Zacatin’.

Resumenes 9º Congreso Mexicano de Botanica, MexicoCity, Sociedad Botanica Mexicana.

— 1987. Taxonomic identity of Leucaena leucocephala(Lam.) de Wit with a new combination. Phytología, 63,304-306.

74

LEUCNET: Exploiting Opportunities for Improvement inLeucaena

J.L. Brewbaker’

Abstract

LEUCNET, the Leucaena Network, is proposed as a core group of scientists and institutionsworking to improve the productivity and utility of Leucaena species. This paper discusses net-working and strategic research activities, suggesting that LEUCNET should perform inter-national collaborative trials with careful statistical analysis; should collect solid benchmarkeddata as the basis for cooperative research; should encourage the planting, demonstration andconservation of superior leucaena germplasm; should use state-of-the-art technologies in com-puter spreadsheets, electronics and genetics; should emphasise communication and rapidexchange of information; and should aim for efficient use of funds in support of tropicalagroforestry.

LEUCNET is a convenient abbreviation of ‘TheLeucaena Network’, a proposed core group ofscientists and institutions sharing a common interestin improving the productivity and utility ofLeucaena species. This definition is adapted fromthe Winrock publication describing forestry net-works (Adams and Dixon 1986). LEUCNET’sprimary objective is to encourage the planting ofimproved leucaena trees, based on well-designedexperimental trials by scientists. This is not an easytask and cannot simply be left to growers. Byforming a network and cooperating towards acommon goal, researchers may pool data obtainedin limited experiments so they can make area-widerecommendations to growers.

Communication

Between 1980 and 1992, the series of LeucaenaResearch Reports (LRR) was a major vehicle forcommunication among cooperating ‘leucaen-ologists’. Through the generous contribution of thegovernment of Taiwan, LRR was printed and dis-tributed to members of the Nitrogen Fixing TreeAssociation (NFTA) - more than 1400 people in1992. Since 1980, LRR published more than 500

1 University of Hawaii, 3190 Maile Way, Honolulu, Hawaii96822

papers in 13 volumes from authors in more than50 countries. Abstracts were added in 1992, toencourage translations. In 1992, administrative cut-backs at NFTA prevented further publication ofLRR. I strongly urge this conference to consider theincorporation of LRR into LEUCNET, and seekappropriate sponsors. If LEUCNET were to joinINTERNET and similar electronic networks, com-munication could be accelerated with costsdiminished.

Biotechnology and plant breeding networks

To make best use of the limited funds for leucaenaresearch, LEUCNET must harness the new genetictechnology. Clonal technologies will permit universaldistribution of environmentally attractive seedlesstriploid hybrids (Sorensson and Brewbaker 1994).Genome mapping of conventional markers andisozymes will be supplemented by molecular markerssuch as RAPDs (randomly amplified polymorphicDNA). These may be suitable for genetic taggingof major genes that influence such traits as cold andacid tolerance. Isozymic phenotypes have shown thatthe common Leucaena (L. leucocephala) is a singlegenotype introduced to many countries, emphasisingthe fragile base on which much leucaena researchhas formerly been based (Sun 1992). Antisense DNAresearch promises to block mimosine synthesis andcreate leucaenas of value as food for nonruminant

75

animals (including humans) and fish. Above all, thegreat ease of exploiting hybridisation promisesalmost unlimited genetic diversity for leucaenaimprovement in agroforestry systems (Brewbaker1993c; Sorensson and Brewbaker 1994).

The data in Figure 1 illustrate the large geneticadvances made in corn breeding in USA, based onregional, replicated trials. Similar yield incrementsshould be possible in Leucaena species throughimproved genetics and management. The remark-able genetic advance with corn (3.5% per year) hasbeen based on germplasm with wide adaptability andlow genotype-environment (G x E) interaction.Such information is available only through net-worked, cooperative trials. The bad years of droughtand flood reduce corn’s gains to 2.9% per year(Fig. 1).

Figure 1. Corn yields in USA 1960-1993.

If comparable gains in forestry yields are possible,they may partially counteract the current loss oftropical forests at a rate of 15 million ha per year,or about 10% in a decade (or 100% in a century).Apart from a few conserved areas, there will be notropical forests left in 100 years time unless thehuman race acts responsibly. Asia and the Pacificnow have less than 300 million ha of forest left, aloss of more than half since 1940. Only in the Pacificislands and Australasia are continuing lossesnegligible.

Work on a genus like leucaena has broader goalsand a much narrower financial base than work oncorn. In leucaena, we seek everything fromimproved fuelwood yields to high quality fodder,parquet flooring and tempeh, focused largely on theless-developed countries. By networking (Adams and

Dixon 1986), scientists and institutions can optimisetheir efficiency in using limited resources to achievetheir common goals. Preliminary unreplicatedprovenance trials of multipurpose tree species, andstrategic or benchmarked research, can lay avaluable foundation for collaborative networkedtrials (Brewbaker 1986b).

Benchmarking

An experiment in a carefully chosen site serves asa benchmark that predicts what will happen onanother but similar site. Impressive site-matchingresearch has been done with Australian trees for useabroad. Effective agrotechnology transfer has oftenresulted from carefully benchmarked research of thistype. Concern about genotype-environment inter-actions (Brewbaker 1984) may encourage site-specific evaluations, but only as a follow-up to thebenchmark studies.

Many questions about leucaena can be answeredat one or two locations by one or two scientists,acting as a benchmark for extrapolation to otherlocations. Classic examples are the hybridisationinterfertility of Leucaena species (Sorensson andBrewbaker 1994), or the reaction to psyllid attack(Glover 1988). The altitudinal range of an accessionis important benchmark advice from the plantcollector, since Leucaena species are clearly dis-tinguished by variations in cold tolerance. Bench-mark data can include the variations in germplasmwhen stressed by soil and environmental factors; thedigestibilities of leucaena species and lines inlivestock; and genotype responses to pathogens andpests. There is always the proviso that racialvariation in pathogen or pest may alter area-widerecommendations. Ultimately, models andalgorithms may be used to predict area-wideperformance.

Leucaena species must be viewed increasingly asvaluable crops that need environments improvedwith fertilizers, weed management, exclusion ofherbivores and people, irrigation and ‘tender lovingcare’. Growers should be able to use benchmarkeddata from trials at our experiment stations, includingyield estimates from experimental algorithms. Beforewe establish network trials, therefore, a thoroughevaluation of the opportunities of benchmark trialsmust be made.

Types and designs of experiments

Research involves not only the collection of databut also astute statistical analysis. New tools areavailable, notably the spreadsheet programsQuattroPro, Excel, Lotus123 and others (Brewbaker1993a, 1993b, 1994). New techniques, such as cluster

76

or principal component analysis, are being used tomap molecular markers on chromosomes, charac-terise systematic relationships and follow tree growthin diverse environments. Members of LEUCNETshould aim for impeccable experimental statisticsthat fully exploit these methods (Matheson 1990).

It is easy to overdesign and complicate tree andfodder experiments with leucaena. A simple exper-iment that precedes a multi-location experiment, sayby one year, can save important time and funds.Questions that can be answered in advance includethe best months for seeding and transplanting, thebest methods of weed and local pest control, thebest entries to include for replication, the bestspacing, and the nature and severity of bordereffects.

When surveying leucaena research, a useful sim-plification is to categorise experiments into one oftwo classes: either Observational Trials (OT) orYield Trials (YT). OT are customarily unreplicatedtrials that are analysed as completely randomiseddesigns (CRD) using sampling error. YT arereplicated and often multi-location trials. They arecommonly randomised complete block (RCB)designs with sample data taken to generate bothinteraction and sampling error variances.

Observational Trials (OT), also called ‘obser-vational nurseries’, are versatile experiments thatmay be of many types. They are often conductedto ‘set the stage’ for YT but at minimal time andexpense. OT are performed primarily to convincethe investigator of the worth of a hypothesis underconditions not requiring replication, for example itis a tree or a shrub; it does or does not resist psyllids.OT are often unreplicated trials of randomisedentries at a single location with small plots, grownfor a short duration. Sampling data, for examplebetween trees within plots, provide error variancesthat show these to be research experiments, notdemonstrations. For example, a typical OT exper-iment comparing five species with four accessionseach, in plots where ten data trees are sampled, pro-vides 180 degrees of freedom for sampling error,or three times that number if the OT is repeatedin three locations. This therefore need not be con-sidered a simple experiment, as it has adequatestatistical tests.

All new accessions, breeding lines, ideas formanagement and for treatment should first be testedin unreplicated OT. One of the most wastefulexperiences of networks has been the conducting ofreplicated trials of untested germplasm or treatmentsat many locations. Funding and common sense mustnot allow this to happen with leucaena.

Yield trials (YT) are commonly designed torigorously evaluate genotype by environment (G xE) interactions. For most quantitative genetic models

it is essential to have individual tree data to pro-vide both sampling and interaction errors (Brew-baker 1994). RCBs are the standard experimentaldesigns, and repetition in time or space creates com-bined experiments, normally a form of split-plotdesign. Lattices are preferable if there are manyentries, say 21 or more, but missing values must berare. We are still considerably ignorant aboutleucaena, and must not focus solely on inter-nationally networked yield trials. Repetition andreplication of trials is expensive, time consuming andoften wasteful, especially when benchmarking canadequately answer many questions.

The augmented randomised complete block(ARCB) design is ideally suited to both YT and OTexperiments, permitting analysis of variance basedon experimental error variances (from replicateinteraction). As an example, Van Den Beldt (citedby Brewbaker 1993a, Ex.12c) planted a YT to exa-mine spacing of leucaena, using 12 different spacingregimes ranging from 2500 to 80 000 trees/ha. Fourspacings were replicated four times, and the othereight inserted as augments, at two per replicate (Fig.2). The unreplicated treatments add significantauthority to the conclusion that 2.5 year oldleucaena trees are shorter when planted in denserstands, with a regression coefficient of 0.61 m reduc-tion per 10 000 trees (Fig. 2). However, the trial wasabout half the size (and cost) of a fully replicatedexperiment.

1

I

Figure 2. Heights of leucaena (x 8) at increasing popula-tion densities.

Principal components analysis (PCA) is an oldtechnique now available for computers in simpleforms of cluster analysis, and combined withanalysis of variance in designs such as fractional

factorials. PCA can be extremely powerful in multi-location trials (e.g. Federer and Murty 1987).

A typical networked leucaena yield trial might bea fodder or green manure trial with 3-5 harvestsannually. Plots can be very small (e.g. two 3 mrows), and locations and replications can be limitedto two or three each, with 20-40 entries in two-yearRCBs or lattice designs. Data are best analysed asa strip-block, recognising the probability ofsystematic variations in yields of consecutiveharvests. Typically, the Mixed Model should beapplied, with locations and replicates as randomvariables and with varieties and harvests as fixedvariables. A data set might have the followingsources of variance (following Brewbaker 1993a,Ex 9b): locations, replicates in locations, harvests,varieties, interactions between harvests x locations,varieties x locations, harvests x varieties, harvestsx varieties x locations, and three differentestimates of error.

Individual tree data (samples) would add a fourthestimate of error here, and could be extremelyvaluable in RCBs as a basis for variance componentanalysis to determine a more cost-effective choiceof numbers of replicates and samples for futuretrials. The great importance of applying errors basedon the Random Model should be emphasised, withF tests based on variance ratios of the main effectsagainst interaction effects. In our experience inHawaii, if the Fixed Model is used, conclusions areinevitably different and often misleading whenextrapolated. Since the primary goal of bench-marked or networked trials is to extrapolate theinformation gained to other times and locations, itis imperative that the Random Model be applied.Spreadsheets are excellent for all such experimentsbecause of their elegant simplicity and cost-effectiveness.

Networked experiments in Hawaii

Germplasm collections of the genus Leucaena werestarted in 1961 in Hawaii, and the 967 accessionswere grown at least once to maturity in OT plots.Plot sizes varied, for example single-tree progeny,selfed, 10 trees; single-tree progeny, half-sib, 16trees; composites, more than 20 trees. Our first twomajor plantings, 1963-1 and 1963-2, included 20trees each of 86 accessions of 4 species in a singlereplicate each. Height and dbh (diameter at heightof 1.3 m) data were taken annually for six yearsat Waimanalo (1963-1), and fodder yields andassociated quality data were taken from two harvestson Kauai (1963-2).

From this experiment-station trial emerged thesuperior arboreal accessions of L. leucocephaladesignated K8, K28, K29, K67 and K72. All were

of the ‘giant’ type, uniform selfed lines, that arenow dispersed and grown rather widely. Most otherlines were exempted from YT, but seeds of all wereretained.

More than 50 leucaena trials have been conductedwith this broad germplasm since the early 1960s inthe Hawaii network, and many results have beenpublished in LRR (Brewbaker 1987). Most trialsgave us experience that helped us plan or providebetter materials for later trials. A primary factorin their success has been convenience of field accessby scientists on a weekly basis over these threedecades. Two major series of trials extending out-side the State of Hawaii involved networkevaluations of wood yields (Brewbaker 1986a,MacDicken and Brewbaker 1988) and of psyllidresistance in forage trials (Glover 1988).

Lessons learned in Hawaii• Leucaenas are like crops, not like pine trees; field

trials can be completed in 2 years.Small annual plantings are greatly preferable tolarge, occasional trials.Designs created in the lab rarely work in the field.Never replicate a new accession or a treatmentnot already tested.Choose carefully the months to seed and totransplant.Weeds must be controlled because youngleucaenas do not compete well with them.Young leucaenas need sun and moisture.Young leucaenas thrive on high pH (> 5.5) andhigh phosphate.Missing trees or plots are to be expected, but upto 25% missing does not affect yield.Borders are essential in yield trials.Fodder and wood yields are highly correlatedwithin species.One-year and four-year wood yields are highlycorrelated within species.Replicates and trees per plot for selfing familiescan be about half those for outcrossing families.Leucaenas thrive on high densities (> 10 000/ha);yields drop at lower densities.Scientist involvement is essential from time ofplanting.Benchmarked experiments save time and money.Augmenting saves time and money.

Leucaena, an extraordinary genus

Most foresters do not think of Leucaena species asrespectable trees. Few have had the opportunity toobserve trials with 10 m growth in 30 months(Fig. 2). Few have harvested 8-year old trees ofK636, as we did in January 1994, that averaged 16 m

78

•

••

•

•

••

•

••

•

•

•

•

••

in height and 34 cm in diameter (ranging to 46 cm),with a density of 0.56 and a richly coloured heart-wood free of defects. These are definitely respectabletrees. Nonetheless, to my knowledge there are nocommercial lumber, pulp, or woodfuel plantationsof leucaena in the world at present. This is mainlybecause of the perception that all leucaenas look likethe scrubby common cultivar found worldwide, fullof seeds and hardly tree-like.

A primary intent of this workshop was to discussthe wide range of germplasm now available forresearch in this genus, and particularly the lesser-known species (Brewbaker and Sorensson 1994). Thecollections of provenances recently assembled by theOxford Forest Institute, and the array of inter-specific hybrids, offer something for everyone.These new species and hybrids must be brought tothe attention of foresters worldwide, and there isno substitute for demonstration plantings. This isan extraordinary genus of tropical multipurposelegume trees, and there is no reason for modestyin our description or expectation of them (Anon.1984).

Acknowledgments

Tree improvement research at the University ofHawaii has been largely supported by grants fromthe US Department of Energy, through the HawaiiNatural Energy Institute, and by McIntyre-Stennisfunds of the US Department of Energy.Collaborators on this paper were Michael Austinand Wei Guo Sun.

ReferencesAdams, N. and Dixon, R.K., ed. 1986. Forestry Networks.

Proceedings of Network Workshop, F/FRED, WinrockInternational, Petit Jean, Arkansas, 112 p.

Anon. 1984. Leucaena: Promising Forage and Tree Cropfor the Tropics, 2nd edition. National Academy Press,Washington, DC.

Brewbaker, J.L. 1984. Genotype-environment interactionsand agrotechnology transfer. In: Hawaii Institute of

Tropical Agricultural Research Extension Service. AMultidisciplinary Approach to Agrotechnology Transfer.Hawaii Institute of Tropical Agricultural ResearchExtension Service 16, Honolulu, Hawaii.

— 1986a. Performance of Australian acacias inHawaiian nitrogen-fixing tree trials. In: Turnbull, J.W.,ed., Australian Acacias in Developing Countries.ACIAR Proceedings, Canberra, Australia, 16, 180-184.

— 1986b. Multipurpose tree species provenance trials.In: Adams, N. and Dixon, R.K., ed., Forestry Networks.Winrock International, Arkansas, 15-24.

— 1987. Leucaena: a genus of multipurpose trees fortropical agroforestry. In: Steppler, H.A. and Nair,P.K.R., ed., Agroforestry; A Decade of Development.ICRAF, Nairobi, Kenya, 289-323.

— 1993a. Experimental Design on a Spreadsheet. Brew-baker, Kailua, HI, 185 p. (Includes DOS-based disk.)

— 1993b. Biometry on a Spreadsheet. Brewbaker,Kailua, HI, 133 p. (Includes DOS-based disk.)

— 1993c. Tree improvement for agroforestry systems.In: Callaway, M.B. and Francis, C.A., ed., CropImprovement for Sustainable Agriculture. University ofNebraska Press, 132-156.

— 1994. Quantitative Genetics on a Spreadsheet. Brew-baker, Kailua, HI. (in press).

Brewbaker, J.L. and Sorensson, C.T. 1994. Domesticationof the lesser-known species of the genus Leucaena. In:Leakey, R.R.B. and Newton, A.C., ed., Tropical Trees:The Potential for Domestication. London, HMSO (inpress).

Federer, W.T. and Murty, B.R. 1987. Uses, limitationsand requirements of multivariate analysis in inter-cropping experiments. In: MacNeil, I.D. and Humphrey,G.J., ed., Biostatistics. D. Reidel, Amsterdam, 264-283.

Glover, N. 1988. Evaluation of Leucaena species for psyllidresistance. Leucaena Research Reports, 9, 5-18.

MacDicken, K. and Brewbaker, J.L. 1988. Growth ratesof five tropical nitrogen fixing fuelwood species. Journalof Tropical Forest Science, 1, 85-93.

Matheson, A.C. 1990. Designing experiments for MPTgenotype evalutions. In: Glover, N. and Adams, N., ed.,Tree Improvement of Multipurpose Species. WinrockMPT Network Technical Series, 2, 67-100.

Sorensson, C.T. and Brewbaker, J.L. 1994. Interspecificcompatibility among 15 Leucaena species (Mimosoideae:Leguminosae) via artificial hybridisations. AmericanJournal of Botany, 81, 240-247.

Sun, W.G. 1992. Isozyme polymorphism in the leguminousgenus Leucaena. MS Thesis, University of Hawaii, 137 p.

79