Ii Food Agric 1996,70,481-486

{anessa E Plumb*, Christopher D Woodt, Cliff Gay

atural Resources Institute, Central A venue, Chatham Maritime, Chatham, Kent, ME4 4TB, UK

lean Hanson and J Hero Heering

International Livestock Centre for Africa, PO Box 5689, Addis Ababa, Ethiopia

~eceived 14 December 1994; revised version received 17 May 1995; accepted 5 October 1995)

Abstract: Aqueous acetone extracts of mature leaves from 10 accessions of twospecies of Sesbania plants were prepared and analysed by high-performanceliquid chromatography (HPLC). The HPLC phenolic profiles were used todevelop a classification rule to distinguish between accessions. Out of the 44mature leaf samples analysed, 27 had distinctive HPLC phenolic profiles andwere successfully assigned to the correct accession. Seedlings of these accessionscou1d be classified to the correct species after 3 months' growth. After 3 months'growth, 8 out of 15 samples could be assigned to the correct accession and, at 6months, 7 out of 20. However, when accessions with very similar phenolic pro-files were grouped together, there was a 100% success rate for classifying thecorrect accession at 3 months and 80% success rate at 6 months. It was con-cluded that the method could distinguish between some accessions, or groups ofaccessions, of Sesbania after 3 months' growth. HPLC phenolic profiling ispotentially a relatively rapid, but not uniformly reliable, method for classifyingSesbania accessions.

Key words: Sesbania, Sesbania goetzei, Sesbania sesban, accessions, leaves,HPLC, phenolic profiles.

INTRODUCTION predicting price and country of origin of black tealiquors (McDowell et al 1991) and in distinguishingbetween different cultivars of apple (McRae et al 1990).

Sesbania is a genus of leguminous shrubs and treeswhich has been investigated for their use as multi-purpose plants (Anon 1993). Sesbania sesban (L) Merrillvar. nubica Chiov and S goetzei Harms show promise asfodder trees and both are native to sub-Saharan Africa.The International Livestock Centre for Africa (ILCA)has a germplasm collection of 300 accessions of 20species of Sesbania. There are large differences bothbetween and within species in their agronomic proper-ties (Tothill et al 1990; Hanson 199-1). There is a need,in various areas of research, for a rapid technique toclassify plant accessions at an early stage, rather thanwait for them to grow to maturity. The aim of thisstudy was to determine whether a distinctive phenolic

address: Natural Resource Management Ltd., profile can be used to distinguis? between young plants_.~ ,-,r from different Sesbania accessions and the stage of

.-, n' ""';("'h the nrofile can be used.

The use of chromatography in cultivar identificationwas reviewed by Morgan (1989) who concluded thatthese techniques offer considerable potential in chemo-taxonomic characterisation. High-performance liquidchromatography (HPLC) is well suited to such purposesas it produces quantitative data and offers a high degreeof discrimination not obtainable through morphologicalexamination. Polyphenol profiles obtained by HPLChave been used to identify cultivars of poinsettia(Stewart et at 1980) and geranium florets (Asen andGriesbach 1983). HPLC data can be analysed stat-istically and such analyses have been used to indicatevarietal differences and responses to different environ-ments in sorghum (Mueller-Harvey and Dhanoa 1991),

?~

Species Age ILCAaccession

code

Site" No

S sesbanvar nubica

3 months 2024 DZDZZDZDZDDDZSDZSDZSDZSDZZDDZZDZZDZDZDZ

2 ..;-."2. ~2222222212222212222222122222222222222

10865

1503620246 months s.

ttPSt

10865

150362024

10865Mature

15021

15022

15036

Sgoetzei 3 months 12771278

1500712771278

6 months

150071277Mature

Leaf samples of five accessions of each of two Sesbaniataxa (S sesban var nubica and S goetzei) were collectedfrom mature trees grown at three different sites (Zwai,Debre ?eit and Shola) in Ethiopia. Table 1 gives theorigins and botanical names for the accessions. Her-barium specimens have been retained at ILCA. Sampleswere obtained in duP.Iicate from different trees except intwo cases when only single trees were available (see'Table 2). Seedlings of some of the accessions weregrown at both Zwai and Debre Zeit. Leaves weresampled after 3 and 6 months (see Table 2).

The leaf samples were freeze-dried and ground topass through a 1 mm sieve and sent to the UK. Hager-man (1988) found that 700 mllitre-1 aqueous acetoneextracted more tannins than aqueous or acidic meth-anol, hence this was the preferred solvent. Samples wereextracted as described by Wood et al (1994) except thatthey were homogenised for 1 min. It had been foundthat extraction of extractable phenols was completedwithin one minute (Powell C J unpublished data).Extracts were centrifuged at 2000 g for 10 min and thenfiltered through a 0.45.um millipore filter beforeanalysis by HPLC. The extract (20 .ul) was analyseddirectly by HPLC using the method of Mueller-Harveyet al (1987).

Examination of the HPLC chromatograms of themature leaf samples indicated that there were sevenmajor peaks that could be used to characterise theaccessions. The peaks were selected and assigned on thebasis of their retention times. For statistical analysisstandardised peak heights were used where

standardised peak height

= observed peak height1278

15007total peak height for all seven peaks

Similar results could be achieved using peak areas.Linear discriminant functions were obtained using

canonical variate analysis (CV A) by means of15358

15367

.D, Debre Zeit; Z, Zwai; S, Shola.TABLE 1

Botanical names and origins of selected Sesbania accessions

Botanical name Origin Accession codeGENST A T computer software. The discriminant func-tions produced by CV A on the data from mature leaveswere used as a classification rule. The mature leaf andseedling samples were distinguished and classified onthe basis of this rule. Correlation was also used as aclassification tool, the correlations being computed forall seedling samples with each mature profile and seed-ling samples classified as the mature accession withwhich it was most highly positively correlated. Variancecomponents for each of the seven HPLC peaks werecalculated using the REML technique to quantify varia-

S sesban (L) Merril varnubica Chiov

EthiopiaIndia (?)UgandaRwandaUgandaTanzaniaTanzaniaEthiopiaEthiopiaKenya

20241086515021150221503612771278

15007.1535815367

S goetzei Harms subsp

multifloraS goetzei Harms subsp

goetzei

between the 10 accessions and between samples foraccession..

found to be important in discriminating between the 10accessions. These functions are of the following form:

RESULTS

chromatograms (at 260 nm) typical of Sesbaniavar nubica and S goetzei are given in Fig 1. Theof the peaks with the same retention times used

.were compared and found to be closelywithin species, indicating

the peaks were the same compounds. Spectra forsix were, however, found to be different for S

YI = al x peak 1 + a2 x peak 2 ...+ a7

x peak 7 + constant

Y2 = bl x peak 1 + b2 x peak 2 ...+ b7

x peak 7 + constant

Y3 = Cl x peak 1 + C2 x peak 2

x peak 7 + constant

+ C7

and the coefficients are provided in Table 3.Table 4 gives the mean values of the coefficients

obtained from the first three discriminant functions. Thefirst canonical variate separates S sesban var nubica

For the mature leaves linear discriminant functionsThe first three functions were

AU(a)

08000

0.6000

20.4000

76

0.20003,~

)~~U1 I0.0000 ~o~ ~ \) ?ft:::::::::~~ = 50~~

id

AU

0.8000

0.6000

0.4000

0.2000

0.0000 50lime (minI

chromatograms of S sesban var nubica and S goetzei at 260 nm showing numbers allocated to peaks used forVertical scale is the absorption in absorption units (AU), horizontal scale gives the retention time on the HPLC

1277. (b) S sesban var nubica ILCA accession code 15021.

6010

between accessions of fodder tree specIes. 1 hIS ..auiabenefit plant breeders and forage agronomists and, ulti-mately, livestock owners whose benefit would be fromimproved feed availability and quality. The project hasdemonstrated that two species of Sesbania can be distin-guished by analysing leaf extracts from young plantsand comparing them to mature leaf extracts. Correctlyclassifying plants by accession using this. method hasproved to be more difficult as some accessions havesimilar phenolic profiles. Larger numbers of accessionswould probably increase this problem. However, themethod has potential to distinguish between someaccessions, or groups of similar accessions. Grouping isthe first step to forming defined core collections of agro-nomic potential from large germ plasm collections.Phenolic profiles could be used for such grouping and,with information on agronomic properties and feedingvalues, help identify improved accessions.

REFERENCES

""II"V'IJI",/'...""-,,,~ ,= using in vitro culture. Paper presented at the

bania Workshop, 9-12 September 1991, Nairobi,McDowell I, Feakes J, Gay C 1991 Phenolic

black tea liquors as a means of predictingof origin. J Sci Food Agric 55 627-641.

McRae K B, Lidster P D, DeMarco A C, Dick A Jparison of the polyphenol profilescorrespondence analysis. J Sci Food Agric 50 329-342.

Morgan A G 1989 Chromatographic applications inidentification Plant Varieties Seeds 2 35-44.

Mueller-Harvey I, Dhanoa M S 1991 Varietalamong sorghum crop residues in relation to theirHPLC fingerprints and responses to differentJ Sci Food Agric 57 199-216.

Mueller-Harvey I, Reed, J D, Hartley R D 1987isation of phenolic compounds, including fiavanoidstannins, of ten Ethiopian browse species by highformance liquid chromatography. J Sci Food Agric 391

Stewart R N, Asen S, Massie D R, Norris K H 1990identification of poinsettia cultivars by HPLC analysistheir fiavanol content. Biochem System Ecol8 119-125.

Tothill J C, Reed J, Tsehay A 1990 Genetic resourcesfodder quality in Sesbania. In: Perennial SesbaniaAgroforestry Systems (Proceedings of a workshop heldNairobi, Kenya, 27-31 March 1989), ed MacklinD O. Nitrogen Fixing Tree Association, Waimanalo,USA.

Wood C D, Tiwari B N, Plumb V E, Powell C J, Roberts B T,Sirimane V D P, Rossiter J T. Gill M 1994 Interspecies '.ferences and variability with time of protein precipitationactivity of extractable tannins, crude protein, ash and drymatter content of leaves from 13 species of Nepalese foddertrees. Journal of Chemical Ecology 203149-3162.

Anon 1993 Sesbania-another string to the farmer's bow.ILCA News/12 1.

Asen S, Griesbach R 1983 High pressure liquid chromato-graphic analysis of flavanojds in geranium florets as anadjunct for cultivar identification. J Am Sac Hart Sci 108845-850.

Hagerman A E 1988 Extraction of tannin from fresh and pre-served leaves. J Chem Eca/14 453-461.