Genetic Resources and Crop Evolution 39: 149-158, 1992.(Q) 1992 Kluwer Academfc Publishers. Printed in the Netherlands.

Genetic diversity in the Chamaecytisus proliferus complex(Fabaceae: Genisteae) in the Canary Islands in relation to in situconservation

J. Francisco-Ortega!.2, M. T. Jackson!,3, J. P. Catty! & B. V. Ford-Lloyd!(School of Biological Sciences, The University of Birmingham, Edgbaston, Birmingham, B152TT, UK;2Present address: Department of Plant Biology, The Ohio State University, 1735 Neil Avenue, Columbus,OH 43210-1293, USA; 3Present address: Plant Genetic Resources Centre, International Rice ResearchInstitute, P.O. Box 933, 1099 Manila, Philippines

Received 12 June 1992; accepted 13 November 1992

Key words: Chamaecytisus proliferus, diodiversity escobon, fodder legumes, isozymes, Macaronesia,tagasaste

Summary

Electrophoresis was carried out for six isozyme systems on 175 accessions of the seven morphologicalforms of Chamaecytisus proliferus (L. fil.) Link (Fabaceae: Genisteae) from the Canary Islands. Previousstudies have shown that there is clear differentiation into morphological and ecological forms. Theunderlying genetic variation as visualised by studying isozymes does not reflect this. While substantialgenetic diversity can be identified, this was found to be as much within as between morphological forms.This genetic diversity was found to decrease from east to west, and reflects the pattern of variationremaining after colonisation of the individual islands, which is assumed to have progressed in the samedirection. Subsequently, adaptive radiation has given rise to the overlying morphological variation asexhibited in the seven morphological forms.

In terms of in situ conservation, overall genetic diversity can be easily conserved in the abundantpopulations occurring in the east of the archipelago, while more attention is required for cons~rvationof the much rarer morphological forms found in the west, despite their relative lack of isozymediversity.Abbreviations: ACO-aconitase, ADH-alcohol dehydrogenase, IDH-isocitrate dehydrogenase,MDH -malate dehydrogenase, PGD-6-phosphogluconate dehydrogenase, PGM -phosphoglucomutase,CA-cluster analysis, PCA-principal component analysis

Introduction

Among a broad group of endemics, which arelocally used as forage species in the CanaryIslands, is tagasaste (Chamaecytisus proliferus(L. fil.) Link var. palmensis (Christ) Hansen &Sunding) (Fabaceae: Genisteae) which is widelycultivated in La Palma, La Gomera, Tenerife,Gran Canaria and EI Hierro. It has also nowachieved importance in some areas of New

Zealand and Australia where it has beenunder cultivation since late in the last century(Francisco-Ortega et al., 1991). Within the genusChamaecytisus, tagasaste is the only species whichis cultivated. Furthermore it represents the mostimportant non-ornamental cultivated species withits centre of diversity and origin in the CanaryIslands.

Chamaecytisus proliferus forms a complex ofseven morphological forms (Table 1), each with a

150

Table 1. The seven morphological forms, of the Chamaecytisus proliferus complex in the Canary Islands after Acebes-Ginoves et al.

(1991)

Taxon Common name Island Ecology

North: 700-1300 m, cliff belt withheath (Erica arborea)

Wild in La Palma,cultivated elsewhere

Typical tagasaste North: 700-1300 ffi, cleared areas ofheath belt and laurel (Laurus azorica)wood

La PalmaWhite tagasaste

White escobon of Tenerife Tenerife

North: 1300-1200 m, south:1000-2000 m,pine (Pinus canariensis) forest

North: 700-1300 m, cleared areas ofheath belt and laurel wood

North: 1300-2200m, south:700-2200 m, pine forest

La Gomera andTenerife

Narrow-leaved escobon

White escobon of Gran Canaria Gran Canaria North: 700-1300 m, cleared areas ofheath belt and laurel wood

Escobon of southernGran Canaria

Gran Canaria North: 1300-200 m, south:1300-2000 m, pine forest

I. C. pro/iferusssp. pro/iferusval. hierrensis(pit.) Aceb.

2. C. pro/iferusssp. pro/iferusval. pa/mensis (Christ)Hans et Sund

3. C. pro/iferusssp. pro/iferusval. ca/derae Aceb.

4. C. pro/iferus

ssp. pro/iferus5. C. pro/iferus

ssp. angustifo/ius(Ktoei) Kunkel

6. C. pro/iferusssp. pro/iferusval. canariae(Christ) Kunkel

7. C. pro/iferusssp. meridiana/is Aceb.

are mirrored by isozyme patterns, and proposalsregarding evolutionary development of thespecies within the Canaries are made. Effectiveconservation strategies are proposed.

Materials and methods

Isozyme analysis was conducted on 175 accessionsrepresenting samples of wild, cultivated and semi-cultivated populations of C. proliferus (Francisco-Ortega & Jackson, 1990; Francisco-Ortega, 1992).They represent the complete range of ecogeo-graphical variation found in this species complex.Seeds were collected in 1989 (Francisco-Ortegaet al., 1990) and samples are now held in CentroNacional de Recursos Fitogeneticos (Alcala deHenares, Spain).

Enzyme extraction was carried out individuallyfrom at least ten mature embryos from eachaccession using the protocol of Nagamine et al.(1989). Horizontal starch gel electrophoresis wasconducted following procedures given by Kephart(1990) using a 0.135 M tris 0.043 M citric acidsolution as electrode buffer.

distinct ecological requirement (Acebes-Ginoves,1990; Acebes-Ginoves et al., 1991; Francisco-Ortega, 1992) and associated with specific life-zones of each island (Table 1). The species followsthe pattern of other Genisteae, and has beenreported as tetraploid with 2n = 4x = 48, 50, 52

(Cristofolini, 1991).The value of isozymes for diversity and

evolutionary studies at the infraspecific levelhas often been reviewed (e.g. Brown, 1990;Crawford, 1990). Isozyme studies have been ofparticular use in assessing patterns of plantevolution and speciation in islands (e.g. Wendel& Percy, 1990; Aradhya et al., 1991; Walters &Decker-Walters, 1991) where rapid speciationthrough adaptive radiation and vicariance has ledto the development of high levels of morphologi-cal variation within usually narrow geographicalboundaries.

In this paper we examine patterns of isozymevariation in relation to the complete range ofdistribution and morphology of C. proliferus sensulata. Genetic variation within and between islandshas been studied to confirm whether patternsdeduced by studying morphology and ecology

151

Six enzyme systems were selected after apreliminary survey of 12 enzymes, and these wereaconitase (ACO), alcohol dehydrogenase (AD H),isocitrate dehydrogenase (IOH), malate dehydro-genase (MDH), 6-phosphogluconate dehydro-genase (PGD), and phosphoglucomutase (PGM).Staining for IOH, MDH and PGM were reportedby Nagamine et al. (1989) and ACO, ADH andPGD by Wendel & Weeden (1989). Staining forACO was modified by the elimination of agarose.

Zones of enzyme activity were identified numeri-cally according to relative migration, those zonesof activity producing the most anodal electro-phoretic variants having the lowest numericaldesignation. Within each zone of activity, bandswere also numbered from slower to fastermigration in the gel. Genetic interpretation ofisozyme phenotypes was based on a comparisonof observed patterns of variation and knowledgeof protein quaternary structures and subcellular

localisation previously reviewed by Weeden &Wendel (1989) and Crawford (1990). This hasled to the definition of putative isozymes andallozymes and to the determination of accessionallele frequencies.

Ordination of results was carried out byPrincipal Component Analysis (PCA). Eachaccession was regarded as an OTU and prior tothe actual analyses, allele frequency values peraccession were transformed (arcsine) and stan-dardised to zero mean and standard deviation of1.0. The statistical package CLUSTAN (Wishart,1987) was used.

Following Wendel & Percy (1990) total variation(Ht) was divided into within and between morpho-logical forms and island components (Hs and Dstrespectively). The unbiased genetic distance andunbiased genetic identity (Nei, 1987) were alsocalculated. A similarity matrix of genetic distancesand a dendrogram were obtained after a Cluster

1 2 3 4 5 6 7 8 9 10 12 13 1415

I I I IADH' ---1- 1---2 ADH2--3

4t t t -ACO2

Analysis (CA) (Unweighted Pair Group Arith-metic Average Clustering Method = UPGM) usingNTSYS (Rohlf, 1988).

Results

PGMI-2 (-0.39), ACO2-5 (-0.36), ACOI-3(0.18) and PGD2-2 (0.11). The six alleles with thehighest eigenvector values along the second com-ponent were ACOI-4 (0.31), ACO2-2 (-0.28),ACO2-1 (-0.20), ACOI-2 (-0.13), PGD2-2(-0.12) and ADH2-0 (-0.10).

A scatter diagram with scores on the first twocomponents is illustrated in Fig. 2. Results suggestthat patterns of variation for some of the allelesare related to island provenance. They indicate thatallozyme ACO2-1 was very common in La Palmawhereas alleles ACOI-5 and ACO2-5 were almostexclusive to Gran Canaria. Accessions of escobonfrom Gran Canaria tended to have high negativevalues along the first component whereas popu-lations from the rest of the archipelago hadpositive scores. Furthermore, accessions of wildtagasaste from La Palma and of escobon from ElHierro had negative values along the second com-ponent whilst escobons collected in Tenerife hadpositive scores.

There was no clear within island differentiationof morphological forms. For instance populationsof both types of tagasaste coincided in this PCAscatter. A similar situation was observed for thetwo morphological forms from both Tenerife andGran Canaria, which did not show distinct scoresalong the first two components.

Genetic diversity

Ten putative loci were identified for the sixenzymes studied and these are illustrated in Fig. 1.Two zones of activity (ACOI and ACO2) wererevealed for ACO system. Five alleles weredetected for ACOI and four for ACO2. Two zonesof activity were observed for ADH (ADHI andADH2). Two alleles were identified for ADHl,one of which was regarded as null. Five alleleswere recognised at ADH2. Alleles ADHl-l andADH2-1

had nearly the same electrophoreticmobility resulting in some overlapping of theirregions of activity. Allele ADH2-2 was observed inall individuals. There was one zone of activity forIDH with a total of eight alleles. Two clear zones(MDHI and MDH2) of enzyme activity wereobserved for the MDH system. Some activity wasalso detected between the MDH2 zone and theorigin. However, banding profiles in this zone werenot clear and therefore were not included in theanalysis. Only two phenotypes were detected forMDH. Two zones of activity (PGMI and PGM2)were identified for PGM system. Four alleles weredetected at PGMI and two at PGM2. Two zonesof activity were identified for PGD (PGDI andPGD2). Seven different bands were detected atPGD 1 zone whilst only three bands were found atPGD2. A genetic interpretation of the bandingpatterns on PGD 1 was not possible and thereforethese data were not included in the multivariateanalysis.

As a result of the genetic interpretation of thesesix isozyme systems, allele frequencies were calcu-lated for accessions and for morphological forms.These are summarised for each morphologicalform in Table 2.

Principal Component Analysis

Results shown in Tables 2-4 indicate that thefour escobons from Gran Canaria and Tenerife-La Gomera not only possess a greater number ofalleles per locus but also show the highest valuesof total gene diversity. The results suggest thatfor the ten loci studied, both escobon of EI Hierroand the wild tagasastes from La Palma were lessvariable than the other morphological forms,suggesting that germplasm from the eastern islands(i.e. Gran Canaria, Tenerife and La Gomera) ismore variable than that from the western islands(i.e. La Palma and EI Hierro).

A partition of the gene diversity within andbetween morphological forms is shown in Table 5.Most of the genetic variation arises as a resultof the gene diversity within morphological forms(Gst = 0.126) and there is only limited isozymedifferentiation amongst the seven morphologicalforms.

The first two components from PCA accountedfor almost 30% of the total variation. The sixalleles most responsible for the variation alongthe first component were, with eigenvector valuesin brackets, PGM1-3 (0.44), ACO1-5 (-0.41),

156

is genetically close to both kinds of tagasaste.However, when the analysis was carried out on anisland basis, escobon of EI Hierro was more relatedto the forms found in Tenerife and La Gomera(Fig. 4) than to La Pal~a. These two differentCA outcomes for germplasm from this islandconfirmed results from PCA (Fig. 2) as this islandhad an intermediate position between La Palmaand Tenerife-La Gomera.

0.12

0.

0.00

Discussion

Studying isozyme diversity in the C. proliferuscomplex has indicated the existence of a high levelof variability. None of the isozymes was mono-morphic and on average there were more than twoalleles per locus. Among the ten putative geneticloci studied, four of them (ACOl, ACO2, PGD2and PGMl) clearly showed variation related tothe geographical distribution of the species. Highvalues of variation were detected within islandsand within morphological forms, which suggeststhat genetic differentiation in C. proliferus is low.However, both PCA and CA based on Nei'sgenetic distance illustrated among island differ-

Fig. 4. Clustering dendrogram (UPGMA method) forChamaecytisus proliferus populations based on Nei's unbiasedgenetic distance and island provenance for ten variable isozymeloci. Provenance is coded as follows: I = EI Hierro; 2 = LaPalma; 3 = La Gomera; 4 = Tenerife; 5 = Gran Canaria.

ences, and those alleles responsible for amongisland differentiation could be identified. Discrimi-nation of any of the morphological forms was notpossible by any of these analyses which suggests agreater genetical divergence of the complex amongislands than among morphological forms.

The two morphological forms which exist ineach island did not show any isozyme differen-tiation even though clear ecological and morpho-logical differences exist between those forms foundwithin each of La Palma, Tenerife and GranCanaria (Acebes-Ginoves, 1990; Francisco-Ortega,1992).

Furthermore, it was found that populations ofescobon from Tenerife and escobon from GranCanaria which grow under very similar ecologicalconditions (e.g. along the bottom of small ravinesof the pine forest where soils are extremely sandy)and which are similar morphologically have differ-ent patterns of allozyme variation for ACOl,ACO2 and PGMI. These results, which aresimilar to those obtained when studying phenoliccompounds (Francisco-Ortega, 1992), do indicatethat genetic differentiation between islands hasoccurred.

Fig. 3. Clustering dendrogram (UPGMA method) for theseven morphological forms of Chamaecytisus pro/iferus basedon Nei's unbiased genetic distance for ten isozyme loci.Morphological forms coded in Table 1.

157

(Lowrey & Crawford, 1985) in the HawaiianIslands, and for Dendroseris (Crawford et al., 1987)in the Juan Fernandez Islands. It was found thatthese island endemics were highly variable formorphological and ecological traits but showedlimited isozyme differentiation.

Strategies and priorities for in situ conservationof the species should take into consideration bothlevels of genetic variation as well as populationabundance for each morphological form. Ecologi-cal studies (Francisco-Ortega, 1992) show thatthe most common morphological forms occur inGran Canaria (i.e. escobon of southern GranCanaria and white escobon of Gran Canaria) andTenerife-La Gomera (i.e. narrow-leaved escobon).These three forms are extremely common in theseislands, where they produce extensive scrubs. Fromthe genetic diversity perspective, they should begiven the highest priority for conservation becausevariation is greatest. It is fortunate therefore thatthese morphological forms (with the exception ofwhite escobon ofTenerife) cannot be considered asendangered by any external factor.

The most rare forms exist in the wild inEl Hierro (i.e. escobon of El Hierro), La Palma(i.e. typical tagasaste and white tagasaste) andTenerife (i.e. white escobon of Tenerife). It isbelieved that these morphological forms shouldactually receive the greatest attention in terms ofconservation as these are the most rare morpho-logical forms within the complex. The establish-ment of the Canarian network of wild reserves(Anon., 1987) has facilitated the conservationin situ of most of the populations of these threemorphological forms.

Acknowledgements

This work was carried out thanks to a personalgrant (JFO) from the Ministerio de Educacion yCiencia, Programa Nacional de Becas de Forma-cion de Personal Investigador en el Extranjero,Spain (grant no. PG88 42044506). The Inter-national Board for Plant Genetic Resources pro-vided financial support for germplasm collection.Our gratitude is also to A. Arroyo-Hogdson, M.Fernandez-Galvan and A. Santos-Guerra (CentroRegional de Investigacion y Tecnologia Agrarias,Tenerife) for their help.

The lack of genetic discontinuity betweenmorphological forms also shows that the speciesis allogamous. All the allelic variation whichwas identified for anyone island was, in someinstances, also to be found in just one popu-lation from that island. Both Gottlieb (1981) andCrawford (1990) suggested that in outcrossingspecies there are usually similar allelic frequenciesin separate populations because of gene flow bothwithin and between populations. Only in thosecases where some populations were isolated fromeach other would some alleles be present at higherfrequencies. A high level of allogamy for C pro-liferus was previously found by Webb & Shand(1985) and Woodfield & Forde (1987) based onreproductive biology and morphology respectively.

Of some importance is the question of howthe biodiversity of the species is structured inthe Canary Islands. Data from ecology andmorphology indicate that variation decreases fromeast to west (Francisco-Ortega, 1992). Populationsfrom the islands closer to the African continent(i.e. Tenerife and Gran Canaria) show greatermorphological variation and wider ecologicaladaptation. Similar conclusions are drawn fromthe isozyme analysis. Genetic diversity in popu-lations from Gran Canaria and Tenerife-LaGomera was larger than that observed in La Palmaand El Hierro. Also, the only islands with uniquealleles were those of Gran Canaria and Tenerife-La Gomera. Patterns of genetic diversity acrossthe archipelago indicate that colonisation followedan east-west path where now, the oldest morpho-logical forms, and the greatest diversity occurs inGran Canaria.

It is clear that morphological differentiationhas taken place and that morphological variantsfound in each island are likely to be derived froma common ancestor originating in proximity tomainland Africa. Indeed two species (i.e. C. mollisand C. pulvinatus) are found in the Atlas Moun-tains in Morocco. Localised ecological adaptationhas produced the morphological variants found ineach island through adaptive radiation. This hasnot been reflected yet in isozyme variation, wherethe patterns of genetic variation arising from theinitial colonisation is still present.

Isozyme variation in C. proliferus in theCanary Islands follows a similar pattern to Bidens(Helenurm & Ganders, 1985) and Tetramolopium

158

References distribution of tagasaste (Chamaecytisus proliferus (L. fil.)Link ssp. palmensis (Christ) Kunkel). A fodder tree from theCanary Islands. J. Adelaide Bot. Gard. 14: 67-76.

Gottlieb, L. D., 1981. Electrophoretic evidence and plantpopulations. Progr. Phytochem. 7: 1-46.

Helenurm, K. & F. R. Ganders, 1985. Adaptive radiation andgenetic differentiation in Hawaiian Bidens. Evolution 39:753-765.

Kephart, S. R., 1990. Starch gel electrophoresis of isozymes:a comparative analysis of techniques. Amer. J. Bot. 77:693-712.

Lowrey, T. K. & D. J. Crawford, 1985. Allozyme divergenceand evolution in Tetramolopium (Compositae: Astereae) onthe Hawaiian Islands. Syst. Bot. 10: 64-72.

Nagamine, T., J. P. Catty & B. V. Ford-Lloyd, 1989. Pheno-typic polymorphism and allele differentiation of isozymes infodder beet, multigerm sugar beet and monogerm sugar beet.Theor. Appl. Genet. 77: 711-720.

Nei, M., 1987. Molecular Evolutionary Genetics. ColumbiaUniversity Press, New York.

Rohlf, F. J., 1988. NTSYS-pc, Numerical Taxonomy andMultivariate Analysis System. Exeter Software, New York.

Walters, T. W. & D. S. Decker-Walters, 1991. Patterns ofallozyme diversity in the West Indies cycad Zamia pumila(Zamiaceae). Amer. J. Bot. 78: 436-445.

Webb, C. J. & J. E. Shand, 1985. Reproductive biologyof tree lucerne (Chamaecytisus palmensis, Leguminosae).New Zealand J. Bot. 23: 597-606.

Weeden, N. & J. F. Wendel, 1989. Genetics of plant isozymes.In: D. E. Soltis & P. S. Soltis (Eds.), Isozymes in PlantBiology, pp. 46-72, Dioscorides Press, Portland, Oregon.

Wendel, J. F. & R. G. Percy, 1990. Allozyme diversity andintrogression in the Galapagos Islands endemic Gossypiumdarwinii and its relationship to continental G. barbadense.Biochem. Syst. & Ecol. 18: 517-528.

Wendel, J. F. & N. F. Weeden, 1989. Visualisation andinterpretation of plant isozymes. In: D. E. Soltis & P. S.Soltis (Eds.), Isozymes in Plant Biology, pp. 46-72,Dioscorides Press, Portland, Oregon.

Wishart, D., 1987. CLUSTAN User Manual, 4th edn.University of St Andrews, St Andrews, Scotland.

Woodfield, D. R. & M. B. Forde, 1987. Genetic variabilitywithin tagasaste. Proc. New Zealand Grassland Assoc. 48:103-108.

Acebes-Ginoves, J. R., 1990. Contribucion al Estudio de losGeneros Chamaecytisus Link y Dorycnium Mill. en elArchipielago Canario. (Ph.D. Dissertation, Universidad deLa Laguna, Tenerife).

Acebes-Ginoves, J. R., M. del Arco-Aguilar & W. Wildpretde La Torre, 1991. Revision taxonomica de Chamaecytisusproliferus (L. fil.) Link en Canarias. Vieraea 20: 191-202.

Anonymous, 1987. Legislacion del Suelo y OrdenacionTerritorial. Consejeria de Politica Territorial del Gobiernode Canarias, Santa Cruz de Tenerife.

Aradhya, K. M., D. Mueller-Dombois & T. A. Ranker, 1991.Genetic evidence for recent and incipient speciation in theevolution of Hawaiian Metrosideros (Myrtaceae). Heredity67: 129-138.

Brown, A. H. D., 1990. The role of isozyme studies in molecularsystematics. Austral. Syst. Bot. 3: 39-46.

Crawford, D. J., 1990. Plant Molecular Systematics. JohnWiley, New York.

Crawford, D. J., T. J. Stuessy, T. J. & M. Silva, 1987. Allozymedivergence and the evolution of Dendroseris (Compositae:Lactuceae) on the Juan Fernandez Islands. Syst. Bot. 12:435-443.

Cristofolini, G., 1991. Taxonomic revision of Cytisus Desf. sect.Tubocytisus DC. (Fabaceae). Webbia 45: 187-219.

Francisco-Ortega, J., 1992. An Ecogeographical Study withinthe Chamaecytisus proliferus (L. fil.) Link Complex(Fabaceae: Genisteae) in the Canary Islands. (ph.D. Disser-tation, The University of Birmingham, United Kingdom).

Francisco-Ortega, J. & M. T. Jackson, 1990. Tagasaste andEscobon (Chamaecytisus proliferus (L. fil.) Link) in theCanary Islands. A Summary of an EcogeographicalSurvey and Collection of Plant Genetic Resources Spon-sored by IBPGR during June to August 1989. InternationalBoard for Plant Genetic Resources, Rome, IBPGRReport 90/1.

Francisco-Ortega, J., M. T. Jackson, A. Santos-Guerra & M.Fernandez-Galvan, 1990. Genetic resources of the fodderlegumes tagasaste and escobon (Chamaecytisus proliferus(L. fil.) Link sensu lato) in the Canary Islands. FAO/IBPGRPl. Genet. Resources Newslett. 81/82: 27-32.

Francisco-Ortega, J., M. T. Jackson, A. Santos-Guerra & M.Fernandez-Galvan, 1991. Historical aspects of the origin and