Botanical Journal of the Linnean Society

, 2007,

153

, 401–416. With 5 figures

© 2007 The Linnean Society of London,

Botanical Journal of the Linnean Society,

2007,

153

, 401–416

401

Blackwell Publishing LtdOxford, UKBOJBotanical Journal of the Linnean Society0024-4074© 2007 The Linnean Society of London? 20071534401416Original Article

VARIABILITY OF SYNGONANTHUS MUCUGENSIS (ERIOCAULACEAE)A. C. S. PEREIRA Et al.

*Corresponding author. E-mail: borba@gmx.net

Genetic and morphological variability of the endangered

Syngonanthus mucugensis

Giul. (Eriocaulaceae) from the Chapada Diamantina, Brazil: implications for conservation and taxonomy

ANA CARINA S. PEREIRA

1

, EDUARDO L. BORBA

2

* and ANA MARIA GIULIETTI

1

1

Departamento de Ciências Biológicas, Laboratório de Sistemática Molecular de Plantas, Universidade Estadual de Feira de Santana, Rodovia BR 116, km 3, Feira de Santana, Bahia, 44031-460, Brazil

2

Departamento de Botânica, Laboratório de Sistemática, Instituto de Ciências Biológicas, Universidade Federal de Minas Gerais, Avenue Antônio Carlos, 6627, Pampulha, Belo Horizonte, Minas Gerais, 31270-901, Brazil

Received December 2005; accepted for publication October 2006

Allozymic and morphometric studies were carried out on ten populations of

Syngonanthus mucugensis

(Eriocaul-aceae), a species from north-eastern Brazil threatened by extinction. Genetic and morphological variability was lowor moderately low in all populations, being lower in populations from Rio de Contas/Catolés (

P

L

=

14.3–21.4,

A

=

1.1–1.2,

H

e

=

0.026–0.059, D2M

=

26.893–33.157) than in those from Mucugê (

P

L

=

28.6–35.7,

A

=

1.3–1.5,

H

e

=

0.078–0.164, D2M

=

28.999–45.077). A high coefficient of endogamy (

F

is

=

0.257) was found, which can be explained by thereproductive characteristics and distribution of the species. The values for genetic and morphological structuring(

F

st

=

0.512 and

A

MRPP

=

0.175, respectively) were high as a result of the differentiation between populations from thetwo areas. The mean genetic identity between populations from the two areas (0.812) was much lower than betweenpopulations from the same area (Mucugê, 0.980; Rio de Contas/Catolés, 0.997). These results indicate that we aredealing with two distinct taxa and, as a result of the nature of the morphological differences found, a new subspeciesis described for the populations of the region of Rio de Contas and Catolés,

Syngonanthus mucugensis

ssp.

riocontensis

. Such conclusions raise important implications for the conservation of

Syngonanthus mucugensis

, andwill be used in the drawing up of management plans for its conservation. © 2007 The Linnean Society of London,

Botanical Journal of the Linnean Society

, 2007,

153

, 401–416.

ADDITIONAL KEYWORDS:

allozyme – campo rupestre – endemism – morphometrics.

INTRODUCTION

The Espinhaço range is the principal mountain chainin eastern Brazil, extending for more than 1000 kmfrom north to south, from the middle of Minas GeraisState to the central-north of Bahia state. It consists oftwo main regions: the Diamantina plateau in MinasGerais and the Chapada Diamantina in Bahia. Inthese areas, the vegetation above an altitude of 900 mis known as ‘campo rupestre’, which is characterizedby the occurrence of sandy and stony soils with her-

baceous and shrubby vegetation in islands of outcrop-ping, usually quartzitic, rock. The campos rupestresare renowned for their species richness with manyendemic elements, especially at the species level(Giulietti & Pirani, 1988; Giulietti, Pirani & Harley,1997). As a result of the discontinuity between thesemountain ranges, a large proportion of the speciesthat occur there are found as disjunct populations.This disjunction has been cited as one of the main fac-tors in the differentiation of populations, leading tothe high degree of endemism found in such areas(Giulietti & Pirani, 1988; Harley, 1988; Borba

et al

.,2001; Jesus

et al

., 2001). It has been estimated that30% of the campo rupestre taxa are restricted to this

402

A. C. S. PEREIRA

ET AL

.

© 2007 The Linnean Society of London,

Botanical Journal of the Linnean Society,

2007,

153

, 401–416

vegetation type, with certain taxonomic groups, espe-cially Eriocaulaceae and Velloziaceae, showing partic-ularly high levels of species richness and endemism,and whose centres of diversity are to be found inthese formations (Giulietti

et al

., 1997). According toGiulietti & Pirani (1988), the Eriocaulaceae presentan interesting distribution pattern, also shared byother groups, characterized by endemism andmicroendemism.

The genus

Syngonanthus

Ruhl. (Eriocaulaceae) con-tains about 200 species, distributed in the Americasand Africa, with the greatest number occurring inSouth America, and with a high degree of endemism inBrazil.

Syngonanthus

and, especially, Sect.

Eulepis

,with their beautiful flower-heads, are popularlyknown as ‘sempre-vivas’ (Lazzari, 2000). Amongst thespecies of this group, which are restricted to theChapada Diamantina, is

Syngonanthus mucugensis

Giul. which, owing to its ornamental value, has beenover-exploited by local communities as an alternativesource of revenue (Giulietti

et al

., 1988, 1996).Flowering scapes are collected before the fruits havedeveloped, thus seriously affecting seed production(Giulietti

et al

., 1988, 1996). The extraction of theseplants from wild populations, together with theirrestricted distribution, is resulting in this and otherspecies of Eriocaulaceae becoming seriously threat-ened with extinction. Although other species of ‘sem-pre-vivas’ are to be found in Mucugê, it is this specieswhich presents the greatest volume of commercializa-tion in the region, and which is one of the speciesavailable for export for decorative purposes (Giulietti

et al

., 1988, 1996; Lazzari, 2000).In Mucugê, populations of this species form patches

on the tops of the mountains between 1100 and1500 m. Such populations were mapped and catego-rized [Good, Fair (Regular), Poor, Almost Extinct, andExtinct] with regard to their conservation status bythe Projeto Sempre-viva (directed from the MunicipalPark, which is maintained by the Mucugê Corpora-tion). Of the populations on record, 17.5% of thoseexamined were found to be in the last two categories.Although the Brazilian Institute for the Environment(IBAMA) has prohibited the collection and commer-cialization of this species, these activities are still con-tinuing at an unacceptable level. Unfortunately, inspite of various attempts, it has not yet been proven tobe possible to raise the plant in cultivation on a com-mercial scale that would help to diminish the threat ofits extinction in the wild, apparently because theplants must be infected by specific mycorrhiza and aresensitive to pH and soil characteristics (Paixão-Santos

et al

., 2003, 2006; Silva

et al

., 2005a, 2005b).Giulietti (1996), in the original publication of this

species, treated it as a microendemic, restricted toMucugê, on the eastern side of the Chapada Diaman-

tina. However, Lazzari (2000) recognized other areas,in Catolés and Rio de Contas, on the western sideof the Chapada, where she considered that

S

.

mucugensis

also occurred. Such an amplification ofthe species range would theoretically help to reducethe threat of its extinction. Lazzari (2000) mentionedcertain morphological differences that distinguishthese plants from those from Mucugê, notably differ-ences in the size of the leaves, spathes, scapes, andcapitula, these structures being larger in plants fromMucugê than in those from Catolés. In spite of thesedifferences, Lazzari did not propose the recognition ofthe latter plants as a new taxon, as many of the veg-etative and floral characters appeared to be similar inboth groups. She suggested that further studies wouldbe needed to elucidate this question.

In this work, the variability and differentiationbetween populations from different localities wereevaluated using genetic and morphological analyses,with the intention of clarifying the taxonomy of thespecies and, consequently, its actual distribution. Suchinformation will aid in the diagnosis of the conserva-tion status of the species, with a view to developing amanagement plan for its conservation. With this aim,an investigation of genetic variability was carried outusing allozyme markers, and a morphological analysiswas performed using a multivariate morphometrictechnique; the results from both analyses were corre-lated. Until now, no such studies on Eriocaulaceaehave been performed, but they have already beenemployed in the Espinhaço range for other groups,such as Orchidaceae and Cactaceae, producing resultsthat have been proven to be useful in taxonomy, con-servation, and for an understanding of evolutionaryprocesses and biogeographical patterns (Borba

et al

.,2001, 2002; Jesus

et al

., 2001; Lambert, Borba &Machado, 2006a; Lambert

et al

., 2006b). This study ispart of a project, developed by our group, on the con-servation and management of species of Eriocaulaceaeand other groups of plants in the Chapada Diaman-tina, which are threatened by extinction as a result ofover-collecting; it brings together studies on demogra-phy, reproductive biology, variability, propagation, andethnobotany.

MATERIAL AND METHODS

S

PECIES

AND

POPULATIONS

SURVEYED

Plants of

S. mucugensis

(Fig. 1) used for this studywere sampled from ten natural populations at threelocalities in the Chapada Diamantina, Bahia: Mucugê(six), Rio de Contas (three), and Catolés (one) (Fig. 2,Table 1). Great care was taken to collect individualsfrom different clones, because this species canreproduce vegetatively. A total of 340 individuals was

VARIABILITY OF

SYNGONANTHUS MUCUGENSIS

(ERIOCAULACEAE)

403

© 2007 The Linnean Society of London,

Botanical Journal of the Linnean Society,

2007,

153

, 401–416

sampled, 200 of which were used in morphometricanalysis. The total sample was used in allozymeelectrophoresis. Leaves were collected for the geneticanalysis, and inflorescences, spathes, and scapes formorphometric analysis. Vouchers are deposited inthe herbarium of Universidade Estadual de Feira deSantana (HUEFS, Table 1).

A

LLOZYME

STUDY

Small sections of leaf tissue were crushed in 0.3 ml ofgrinding buffer [100 ml Tris/HCl 0.1 mol l

−

1

, pH 7.0;6.846 g saccharose; 0.6 g polyvinylpyrrolidone (PVP);0.0372 g ethylenediaminetetraacetic acid (EDTA);0.145 g bovine serum albumin (BSA); 0.13 g sodium

Figure 1.

Syngonanthus mucugensis

; individual of Mucugê population. A, Individual with young inflorescences. B,Capitulum. Scale bar, 5 mm.

Table 1.

Populations of

Syngonanthus mucugensis

occurring in the Chapada Diamantina, Brazil, used in this study.Vouchers are deposited in the herbarium of the Universidade Estadual de Feira de Santana (HUEFS)

Municipality, population Name

Individuals (

N

)

Location VoucherMorphometrics Genetics

Mucugê, Projeto Sempre Viva 1 M1 21 30 12

°

59

′

48.6

″

S, 41

°

20

′

40.2

″

W

Pereira & Oliveira 195

Mucugê, Projeto Sempre Viva 2 M2 20 30 12

°

59

′

45.6

″

S, 41

°

20

′

20.5

″

W

Pereira & Oliveira 197

Mucugê, Serra do Leobino M3 19 62 12

°

58

′

55.3

″

S, 41

°

26

′

15.1

″

W –Mucugê, near Proj. Sempre Viva M4 20 33 13

°

01

′

11.4

″

S, 41

°

19

′

59

″

W –Mucugê, Gobira M5 20 62 13

°

05

′

30.7

″

S, 41

°

22

′

09.6

″

W

Borba

et al.

1820

Mucugê, Serra da Tesoura M6 20 32 13

°

08

′

32.2

″

S, 41

°20′14.9″W Borba et al. 1921Rio de Contas, Itoibira 1 R1 20 31 12°23′47″S, 41°52′00″W Pereira et al. 005Rio de Contas, Itoibira 2 R2 20 20 13°23′43″S, 41°51′47″W Pereira et al. 003Rio de Contas, Itoibira 3 R3 20 20 13°23′55.5″S, 41°51′51″W Pereira et al. 004Catolés, Campos do Virassaia C1 20 45 13°20′38″S, 41°50′15″W Miranda et al. 560

404 A. C. S. PEREIRA ET AL.

© 2007 The Linnean Society of London, Botanical Journal of the Linnean Society, 2007, 153, 401–416

diethylcarbamate (DIECA); 0.6 g borax; 100 µl β-mer-captoethanol; modified from Sun & Ganders, 1990].The extracts were absorbed on Whatman No. 3 paperand applied to a Sigma starch gel. Three buffersystems were used: system 1 electrode: boric acid0.3 mol l−1, NaOH 0.06 mol l−1, pH 8.0; gel: Tris0.01 mol l−1, pH 8.5 (modified from Shaw & Prasad,1970); system 2 electrode: histidine 0.065 mol l−1

adjusted to pH 6.5 with citric acid; gel: electrode bufferdiluted 1 : 4 (modified from Stuber, Goodman &Johnson, 1977); system 3 electrode: lithium hydroxide0.05 mol l−1, boric acid 0.0935 mol l−1, EDTA0.0059 mol l−1, pH 8.0; gel: electrode solution diluted1 : 10 (modified from Ridgway, Sherburne & Lewis,1970). Standard horizontal electrophoresis was per-formed until the inner marker (bromophenol blue)reached 9 cm from the application site using the

following running conditions: system 1, 25 and 13 mA;system 2, 150 V; system 3, 25 mA.

Eleven enzymatic systems gave sufficient resolutionfor reading and were used: buffer system 1: esterase(EST; EC 3.1.1.1), acid phosphatase (ACP; EC 3.1.3.2),leucine aminopeptidase (LAP; EC 3.4.1.1), hexokinase(HK; EC 2.7.1.1), 6-phosphogluconate dehydrogenase(6PGD; EC 1.1.1.44), glucose-6-phosphate dehydroge-nase (G6PD; EC 1.1.1.49); buffer system 2: malatedehydrogenase (MDH; EC 1.1.1.37); buffer system 3:phosphoglucomutase (PGM; EC 2.7.5.1), phosphoglu-coisomerase (PGI; EC 5.3.1.9), isocitrate dehydroge-nase (IDH; EC 1.1.1.42), shikimic dehydrogenase(SKDH; EC 1.1.1.25). The staining procedures weresimilar to, but slightly adjusted from, those of Alfenaset al. (1991; SKDH), Brune, Alfenas & Junghans(1998; LAP, ACP, EST, HK, and G6PD), Corrias et al.

Figure 2. Map of the Chapada Diamantina, north-eastern Brazil, showing the localities of populations of Syngonanthusmucugensis.

100 km

VARIABILITY OF SYNGONANTHUS MUCUGENSIS (ERIOCAULACEAE) 405

© 2007 The Linnean Society of London, Botanical Journal of the Linnean Society, 2007, 153, 401–416

(1991; IDH, 6PGD, and PGI), and Soltis et al. (1983;PGM and MDH).

Enzymatic systems showing more than one locuswere numbered in ascending order from the locus withthe lowest mobility. The alleles were numbered accord-ing to their mobility relative to the allele of a standardindividual of S. mucugensis present in all gels and des-ignated as 100. The allelic frequencies were deter-mined by manually counting the banding patterns ofthe homozygotes and heterozygotes stained in the gels.The genetic variability for each population was esti-mated using the following parameters: proportion ofpolymorphic loci (PL; 0.95 criterion), mean number ofalleles per locus (A), and observed (Ho) and expected(He) mean heterozygosity per locus. Departures fromthe expected mean heterozygosity under Hardy–Wein-berg (HW) equilibrium were tested using χ2, with acorrection for small samples according to Levene(1949). Partitioning of genetic diversity between con-specific populations was estimated by F statistics (Fis,the inbreeding coefficient, measures the decrease inheterozygosity as a result of nonrandom mating withina population; Fst, the fixation index, measures the dif-ferentiation between populations; Wright, 1978). Clus-ter analysis was performed using the genetic distancematrix [Nei’s (1978) unbiased genetic distance] of thepopulations employing the unweighted pair-groupmethod of arithmetical averages (UPGMA) (Sneath &Sokal, 1973). All analyses were performed using theBIOSYS 1.0 software package (Swofford & Selander,1989), except for the cluster analysis which was per-formed in Statistica 5.5 (Statsoft, 2000).

MORPHOMETRIC ANALYSIS

Forty-two continuous and discontinuous flower char-acters were measured (Table 2). Owing to the minutesize of the flowers, they were dissected and thendrawn with the aid of a stereomicroscope equippedwith a camera lucida, and measurements were thenmade from the drawings for greater precision. Dis-criminant analysis was conducted for all characters.The standardized coefficients for canonical variablesresulting from discriminant analysis were used toidentify the characteristics that contributed most sig-nificantly to the resulting patterns observed. Clusteranalysis for the populations was carried out using theMahalanobis generalized distance calculated from thepooled residual covariances within the group matrixand UPGMA as the clustering algorithm. The medianvalues of the Mahalanobis generalized distance werecalculated from the individuals to the centroid of theirpopulation (D2M), and a nonparametric varianceanalysis was carried out using the Kruskal–Wallistest. A multiresponse permutation procedure (MRPP)analysis was used to calculate the average within-

group distance (mean Euclidean distance, ED) for allpopulations and the chance-corrected within-groupagreement (AMRPP) between populations. The D2M andED values were used as measurements of morpholog-ical variability, and AMRPP as a measurement of themorphological differentiation of conspecific popul-ations, and correlated with Fst (Borba et al., 2002;Lambert et al., 2006a, 2006b). The two indices ofmorphological variability are essentially different, asD2M is more affected by the form and ED is moreaffected by the size of the characters (Lambert et al.,2006a, 2006b). Discriminant analysis and clusteranalysis were carried out using Statistica 5.5, MRPPwas run using PCOrd 4.10 (McCune & Mefford, 1999),and variance analysis was performed using BioEstat3.0 (Ayres et al., 2003). For all characters, a univariateanalysis was also performed of the average values ofthe pooled individuals of Mucugê and the pooled indi-viduals of Rio de Contas and Catolés by the t-testusing Statistica 5.5.

CORRELATION ANALYSIS

An analysis of Spearman’s nonparametric correlationwas carried out between the genetic variability (He)and morphological variability (ED and D2M). Thisanalysis was run in Statistica 5.5. The Mahalanobisgeneralized distance matrix was compared withmatrices of Nei’s distance (Nei, 1978) and the geo-graphical distance of the populations using the Manteltest, with the Monte Carlo option of PCOrd (1000 ran-domizations). The pair-wise geographical distancesbetween the populations were computed with geodeticdistances on WGS84 ellipsoid, calculated using theINVERSE 2.0 program (National Geodetic Survey,2002).

RESULTS

GENETIC VARIABILITY

Using 11 enzymatic systems, 14 loci were obtainedthat presented good resolution and were used in thisstudy (Table 3). Three loci were monomorphic for allthe populations studied (PGI-1, MDH-1, and EST-3),but most were polymorphic. PGM, ACP, and LAP pre-sented the highest polymorphism, with three alleles.All populations, except R2 and C1, presented threealleles in at least one of the most polymorphic loci.Two diagnostic loci were found (PGI-2 and IDH), pre-senting fixed exclusive alleles for populations fromMucugê and for populations from Rio de Contas/Catolés. Five loci (6PGD, LAP, HK, EST-1, and EST-2)did not present activity for the Rio de Contas/Catoléspopulations. The populations M3, M5, and M6 pre-sented specific rare alleles in the systems 6PGD,

406 A. C. S. PEREIRA ET AL.

© 2007 The Linnean Society of London, Botanical Journal of the Linnean Society, 2007, 153, 401–416

Table 2. Morphological characters used in the morphometric analysis of ten populations of Syngonanthus mucugensisoccurring in the Chapada Diamantina, Brazil. Values presented for the pooled populations of Mucugê and of Rio de Contasand Catolés are means ± standard deviation (minimum–maximum). Values in millimetres, except scape length, presentedin centimetres

Morphological character Mucugê Rio de Contas/Catolés

Involucral bractsNumber of series of bracts* 11.8 ± 1.6 (9.0–16.0) 9.7 ± 1.5 (7.0–16.0)Bract length of the outermost series* 4.9 ± 0.6 (3.6–6.7) 2.4 ± 0.4 (1.7–3.3)Bract apex angle of the outermost series* 35.4 ± 8.9 (14.0–53.0) 42.4 ± 6.2 (23.0–61.0)Bract width of the outermost series at 1/5 of its length* 0.8 ± 0.2 (0.3–1.3) 0.7 ± 0.1 (0.3–0.9)Bract width of the outermost series at 2/5 of its length* 1.3 ± 0.2 (0.7–1.8) 0.9 ± 0.1 (0.7–1.2)Bract width of the outermost series at 3/5 of its length* 1.6 ± 0.2 (1.0–2.2) 1.0 ± 0.1 (0.8–1.3)Bract width of the outermost series at 4/5 of its length* 1.5 ± 0.2 (0.9–2.0) 1.0 ± 0.1 (0.6–1.3)Bract length of the innermost series* 9.4 ± 1.0 (6.4–12.0) 6.5 ± 0.6 (4.8–8.0)Bract width of the innermost series at 1/5 of its length* 1.5 ± 0.4 (0.7–2.3) 1.0 ± 0.2 (0.6–1.4)Bract width of the innermost series at 2/5 of its length* 1.4 ± 0.4 (0.7–2.2) 1.1 ± 0.2 (0.7–1.6)Bract width of the innermost series at 3/5 of its length* 1.0 ± 0.3 (0.6–1.9) 0.9 ± 0.1 (0.7–1.3)Bract width of the innermost series at 4/5 of its length* 0.9 ± 0.2 (0.6–1.6) 0.8 ± 0.1 (0.4–1.0)

Pistillate flowerSepal length 3.0 ± 0.3 (2.3–3.7) 3.1 ± 0.4 (0.3–3.8)Sepal width at 1/5 of its length* 0.34 ± 0.1 (0.2–0.6) 0.27 ± 0.1 (0.1–0.8)Sepal width at 2/5 of its length* 0.5 ± 0.1 (0.3–0.7) 0.4 ± 0.1 (0.2–0.5)Sepal width at 3/5 of its length* 0.61 ± 0.1 (0.4–0.9) 0.55 ± 0.1 (0.4–0.9)Sepal width at 4/5 of its length 0.6 ± 0.1 (0.3–1.0) 0.6 ± 0.1 (0.4–0.9)Length of the connate part of the petals* 1.7 ± 0.4 (0.8–2.3) 1.5 ± 0.3 (0.7–2.2)Length of the free part of the petals* 2.28 ± 0.3 (1.3–3.3) 2.42 ± 0.3 (1.7–3.2)Petal width at 1/5 of its length* 0.19 ± 0.0 (0.1–0.3) 0.16 ± 0.0 (0.1–0.3)Petal width at 2/5 of its length 0.2 ± 0.0 (0.1–0.3) 0.2 ± 0.0 (0.1–0.3)Petal width at 3/5 of its length† 0.2 ± 0.1 (0.1–0.3) 0.2 ± 0.1 (0.1–0.4)Petal width at 4/5 of its length† 0.2 ± 0.1 (0.1–0.3) 0.2 ± 0.1 (0.1–0.4)Length of the free part of the styles† 2.9 ± 0.6 (1.2–4.2) 2.7 ± 0.4 (1.0–3.6)Length of the free part of the appendages† 1.0 ± 0.3 (0.1–1.7) 1.0 ± 0.2 (0.3–1.5)Column length 1.1 ± 0.3 (0.4–1.7) 1.2 ± 0.2 (0.0–1.8)

Staminate flowerSepal length* 3.1 ± 0.3 (2.1–3.6) 3.0 ± 0.2 (2.2–3.4)Sepal width at 1/5 of its length* 0.6 ± 0.1 (0.3–0.9) 0.5 ± 0.1 (0.3–0.8)Sepal width at 2/5 of its length* 0.8 ± 0.1 (0.4–1.1) 0.6 ± 0.1 (0.4–0.9)Sepal width at 3/5 of its length* 0.7 ± 0.1 (0.4–1.1) 0.6 ± 0.1 (0.4–1.1)Sepal width at 4/5 of its length* 0.5 ± 0.1 (0.3–0.8) 0.4 ± 0.1 (0.3–0.7)Length of the connate part of the petals 1.3 ± 0.3 (0.8–2.8) 1.3 ± 0.1 (0.9–1.7)Length of the free part of the petals* 1.7 ± 0.3 (0.4–2.5) 1.9 ± 0.2 (1.4–2.4)Petal width at 1/5 of its length* 0.6 ± 0.1 (0.3–1.1) 0.5 ± 0.1 (0.3–0.7)Petal width at 2/5 of its length* 0.7 ± 0.1 (0.3–1.3) 0.6 ± 0.1 (0.4–0.8)Petal width at 3/5 of its length 0.6 ± 0.1 (0.3–0.9) 0.6 ± 0.1 (0.4–0.8)Petal width at 4/5 of its length* 0.43 ± 0.1 (0.1–0.8) 0.36 ± 0.1 (0.2–0.6)Petal apex angle* 46.6 ± 10.6 (24.0–77.0) 42.0 ± 8.6 (21.0–60.0)Filament length 3.3 ± 2.5 (2.2–30.5) 3.4 ± 0.4 (2.7–4.6)Anther length* 0.7 ± 0.2 (0.4–2.3) 0.6 ± 0.1 (0.3–1.0)Scape length* 45.4 ± 9.7 (9.6–72.0) 30.5 ± 5.7 (19.8–46.0)Spathe length* 69.6 ± 9.1 (50.0–92.3) 39.1 ± 4.5 (27.5–53.0)

Statistically significantly different mean values in the t-test: *P < 0.01; †P < 0.05.

VARIABILITY OF SYNGONANTHUS MUCUGENSIS (ERIOCAULACEAE) 407

© 2007 The Linnean Society of London, Botanical Journal of the Linnean Society, 2007, 153, 401–416

Table 3. Allele frequencies at 14 allozymic loci in ten populations of Syngonanthus mucugensis occurring in the ChapadaDiamantina, Brazil. See Table 1 for the names of the populations

Locus M1 M2 M3 M4 M5 M6 R1 R2 R3 C1

PGI-1100 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000(N) 28 29 62 32 62 31 31 19 19 46PGI-2100 1.000 1.000 1.000 1.000 1.000 1.000 – – – –107 – – – – – – 1.000 1.000 1.000 1.000(N) 28 29 62 32 62 31 31 19 19 46IDH100 1.000 1.000 1.000 1.000 1.000 1.000 – – – –105 – – – – – – 1.000 1.000 1.000 1.000(N) 28 29 62 32 62 31 31 19 19 46SKDH94 – – – – 0.016 – – – – –100 1.000 1.000 1.000 1.000 0.984 1.000 1.000 1.000 1.000 1.000(N) 28 29 62 32 62 31 31 19 19 27PGM82 0.036 – – – – – 0.032 – 0.026 –100 0.768 1.000 1.000 1.000 0.960 0.984 0.968 0.868 0.789 0.739110 0.196 – – – 0.040 0.016 – 0.132 0.184 0.261(N) 28 29 62 32 62 31 31 19 19 446PGD100 1.000 1.000 0.982 1.000 1.000 1.000 – – – –110 – – 0.018 – – – – – – –(N) 28 15 56 24 43 31 31 19 19 46ACP85 – – 0.185 0.141 0.032 – – – – –100 0.571 0.655 0.637 0.578 0.758 0.677 0.903 0.737 0.594 0.685115 0.429 0.345 0.177 0.281 0.210 0.323 0.097 0.263 0.406 0.315(N) 28 29 62 32 62 31 31 19 16 46LAP78 – – – – – 0.059 – – – –100 0.250 0.632 0.700 0.625 0.438 0.941 – – – –125 0.750 0.368 0.300 0.375 0.563 – – – – –(N) 8 19 10 8 16 17 31 19 19 46HK100 1.000 1.000 1.000 1.000 1.000 1.000 – – – –(N) 28 29 62 32 62 31 31 19 19 46MDH-171 – – – 0.094 – – 0.065 0.053 – –100 1.000 1.000 1.000 0.906 1.000 1.000 0.935 0.947 1.000 1.000(N) 28 29 62 32 62 31 31 19 19 46MDH-2100 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000(N) 28 29 62 32 62 31 31 19 19 46EST-1100 0.571 0.845 0.778 1.000 0.938 1.000 – – – –105 0.429 0.155 0.222 – 0.063 – – – – –(N) 28 29 36 28 40 16 31 19 19 46EST-2100 0.500 0.818 0.903 0.788 0.591 0.400 – – – –110 0.500 0.182 0.097 0.212 0.409 0.600 – – – –(N) 17 22 31 26 33 15 31 19 19 46

408 A. C. S. PEREIRA ET AL.

© 2007 The Linnean Society of London, Botanical Journal of the Linnean Society, 2007, 153, 401–416

SKDH, and LAP, respectively. The allele PGM-82 wasfound in only three populations, M1, R1, and R3,always at low frequency.

All populations presented low degrees of geneticvariability, considering the parameters analysed. Thepercentage of polymorphic loci (PL, 0.95 criterion)ranged from 14 to 35%, the mean number of alleles perlocus (A) between 1.1 and 1.5, and the mean expectedheterozygosity (He) from 0.026 to 0.164 (Table 4). Ingeneral, populations from Rio de Contas/Catolésshowed the lowest variability.

Of the ten populations, only one (M2) did not showsignificant deviation from the expected values in HWequilibrium in all loci, and two (R2 and C1) showedsignificant deviation from these equilibria in all loci.No locus was in equilibrium in all populations, evenconsidering the two regions separately, except ACP inMucugê. Three loci were not in equilibrium in any ofthe populations (6PGD, MDH-1, and SKDH), althoughthese were polymorphic in only one (SKDH and 6PGD)

and three (MDH-1) populations. All the polymorphicloci that were not in equilibrium showed a deficiencyof heterozygotes, except EST-2 in population M1,which was reflected in the elevated Fis (Table 5).

MORPHOLOGICAL VARIABILITY

The morphological variability, represented by themedian Mahalanobis generalized distance (D2M) andthe mean Euclidean distance (ED), showed that theMucugê populations contained higher variability lev-els than those of Rio de Contas and Catolés, with theexception of population M3, which showed the lowestvariability of any of the Mucugê populations (Table 4).An analysis of variance confirmed the greater variabil-ity of the Mucugê populations and indicated that theRio de Contas and Catolés populations showed a sim-ilar degree of morphological variability. Spearman’stest showed that there was a correlation between thegenetic (He) and morphological variability, as given by

EST-3100 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000(N) 1 1 2 1 2 1 8 8 3 25

Locus M1 M2 M3 M4 M5 M6 R1 R2 R3 C1

ACP, acid phosphatase; EST, esterase; G6PD, glucose-6-phosphate dehydrogenase; HK, hexokinase; IDH, isocitratedehydrogenase; LAP, leucine aminopeptidase; MDH, malate dehydrogenase; N, sample size; 6PGD, 6-phosphogluconatedehydrogenase; PGI, phosphoglucoisomerase; PGM, phosphoglucomutase; SKDH, shikimic dehydrogenase.

Table 3. Continued

Table 4. Genetic variability at 14 allozymic loci and morphological variability based on the morphometric analysis of 42morphological characters in ten populations of Syngonanthus mucugensis occurring in the Chapada Diamantina, Brazil.See Table 1 for the names of the populations. Standard deviations in parentheses. A locus was considered to be polymorphicif the frequency of the most common allele did not exceed 0.95

Population N A PL Ho He D2M ED

M1 23.9 (2.4) 1.4 (0.2) 35.7 0.144 (0.065) 0.164 (0.062) 44.371 8.390M2 24.8 (2.2) 1.3 (0.1) 28.6 0.095 (0.042) 0.108 (0.049) 45.077 8.146M3 49.5 (5.6) 1.4 (0.2) 28.6 0.069 (0.041) 0.110 (0.051) 28.999 5.537M4 26.8 (2.6) 1.4 (0.2) 28.6 0.081 (0.053) 0.113 (0.055) 36.755 7.251M5 49.4 (5.3) 1.5 (0.2) 28.6 0.108 (0.056) 0.115 (0.051) 42.213 8.387M6 25.6 (2.5) 1.3 (0.1) 21.4 0.077 (0.051) 0.078 (0.045) 39.986 8.620R1 18.6 (4.0) 1.2 (0.1) 14.3 0.014 (0.010) 0.026 (0.015) 33.157 7.081R2 11.8 (2.4) 1.2 (0.1) 21.4 0.011 (0.008) 0.053 (0.032) 26.893 7.356R3 11.2 (2.4) 1.2 (0.2) 14.3 0.027 (0.023) 0.061 (0.042) 28.539 6.421C1 26.9 (5.7) 1.1 (0.1) 14.3 0.016 (0.014) 0.059 (0.040) 27.467 6.254

A, mean number of alleles per locus; D2M, median Mahalanobis generalized distance of the individuals to the centroid ofthe population; ED, mean Euclidean distance between the individuals of the population; He, expected mean heterozygosityper locus ( Nei, 1978; unbiased estimate); Ho, observed mean heterozygosity per locus ( Nei, 1978; unbiased estimate); N,mean sample size per locus; PL, percentage of polymorphic loci.

VARIABILITY OF SYNGONANTHUS MUCUGENSIS (ERIOCAULACEAE) 409

© 2007 The Linnean Society of London, Botanical Journal of the Linnean Society, 2007, 153, 401–416

the median Mahalanobis generalized distance(r = 0.672; P = 0.033), although such a correlation wasnot shown to be statistically significant (r = 0.394;P = 0.259) when the mean Euclidean distance wasused as a coefficient of variability (Fig. 3).

GENETIC AND MORPHOLOGICAL STRUCTURE

Considering all populations, S. mucugensis showed ahigh Fst, mainly as a result of the loci PGI-2 and IDH,which were diagnostic in the identification of the

Table 5. F statistics (Wright, 1978) at 14 allozymic loci and the chance-corrected within-group agreement (AMRPP) of themultiresponse permutation procedure (MRPP) analysis of 42 morphological characters in ten populations of Syngonanthusmucugensis occurring in the Chapada Diamantina, Brazil. Calculated using all populations pooled and the populations ofthe two regions separately

Locus

Fis Fst

All pop. MC R/C All pop. MC R/C

PGI-2 – – – 1.000 – –IDH – – – 1.000 – –SKDH 1.000 1.000 – 0.015 0.013 –PGM 0.523 0.460 0.553 0.115 0.132 0.0626PGD 1.000 1.000 – 0.016 0.015 –ACP 0.254 0.041 0.667 0.050 0.034 0.064LAP −0.004 0.004 – 0.368 0.205 –MDH-1 1.000 1.000 1.000 0.055 0.079 0.031EST-1 0.606 0.606 – 0.234 0.182 –EST-2 0.020 0.020 – 0.291 0.149 –Mean 0.257 0.144 0.651 0.512 0.134 0.061AMRPP 0.175 0.115 0.021

ACP, acid phosphatase; EST, esterase; IDH, isocitrate dehydrogenase; LAP, leucine aminopeptidase; MC, Mucugê; MDH,malate dehydrogenase; 6PGD, 6-phosphogluconate dehydrogenase; PGI, phosphoglucoisomerase; PGM, phosphoglucomu-tase; R/C, Rio de Contas/Catolés; SKDH, shikimic dehydrogenase.

Figure 3. Representation of the values of morphological variability (D2M, median Mahalanobis generalized distance ofthe individuals to the centroid of the population; ED, mean Euclidean distance between the individuals of the population)and genetic variability (He, mean expected heterozygosity) in ten populations of Syngonanthus mucugensis occurring inthe Chapada Diamantina, Brazil. See Table 1 for the names of the populations.

D2M (L)

ED (R)

mean expected heterozygosity

M1M2

M3

M4

M5

M6

R1

R2

R3

C1

M1

M2

M3

M4

M5

M6

R1

R2

R3C1

5.0

5.5

6.0

6.5

7.0

7.5

8.0

8.5

9.0

24

28

32

36

40

44

48

0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18

med

ian

Mah

alan

obis

dis

tanc

e

mea

nE

uclid

ean

dist

ance

410 A. C. S. PEREIRA ET AL.

© 2007 The Linnean Society of London, Botanical Journal of the Linnean Society, 2007, 153, 401–416

populations by locality. This fact became apparentwhen Fst was calculated separately for each locality,this value being greater for the Mucugê populations(Table 5). In the same way, the AMRPP values, taking allthe populations together, were greater (AMRPP = 0.175)than when the populations were analysed separately,with the greatest values shown by those from Mucugêrelative to those from Rio de Contas and Catolés. Thehigher Fst and AMRPP values for Mucugê indicated agreater genetic and morphological differentiationbetween the populations here than for the other local-ities. The differentiation of population M1 was mainlyresponsible for the high value of Fst in this locality,whereas M3 was responsible for the high value ofAMRPP.

PHENETIC RELATIONSHIPS

The genetic identity between the Mucugê populationswas greater than 0.94, whereas that between the Riode Contas and Catolés populations was over 0.99(Table 6), indicating a slightly greater similaritybetween the populations of the latter region, as can beseen in the dendrogram (Fig. 4A).

The classification matrix obtained from discrimi-nant analysis showed a percentage of correct classifi-cations of individual plants, in relation to the

population from which they came, of 80–100%.Observed errors of classification arose exclusivelybetween populations from the same locality, and thegreater part of these occurred between populationsfrom Rio de Contas/Catolés, demonstrating greatersimilarity between these populations (Table 7).

The analysis of canonical variables showed, in thefirst axis, two distinct groups: one formed by theMucugê populations and the other by those from Riode Contas and Catolés. In the second axis, populationsM5 and M6 were differentiated from the remainder ofthose from Mucugê (Fig. 5A). The third axis providedevidence of the distinctiveness of population M3(Fig. 5B). The characters responsible for the differ-ences shown in the discriminant analysis were, in thefirst axis, mainly those related to the length and widthof the involucral bracts, showing a negative and posi-tive correlation with the axis, respectively; in thesecond axis, the differentiation was mainly a resultof the length and width of the sepals and petals in thestaminate flowers, showing a negative and positivecorrelation, respectively; in the third axis, the differ-entiation was mainly caused by characters related tostyle and column size in pistillate flowers, both corre-lated positively with the axis. The difference betweenpopulations was statistically significant with P < 0.01,except for R2 and R3 (P = 0.0437).

Table 6. Matrix of mean genetic identity (Nei, 1978; unbiased estimate) between ten populations of Syngonanthusmucugensis from the regions of Mucugê (MC) and Rio de Contas/Catolés (R/C), Chapada Diamantina, Brazil. Minimumand maximum values are given in parentheses

Region No. of populations MC R/C

MC 6 0.980 (0.943–0.999) –R/C 4 0.812 (0.756–0.833) 0.997 (0.991–1.000)

Table 7. Classification matrix of the individuals in the discriminant analysis of 42 morphological characters in tenpopulations of Syngonanthus mucugensis occurring in the Chapada Diamantina, Brazil. See Table 1 for the names of thepopulations

Population % correct M1 M2 M3 M4 M5 M6 R1 R2 R3 C1

M1 95.2 20 – – 1 – – – – – –M2 90 2 18 – – – – – – – –M3 100 – – 19 – – – – – – –M4 100 – – – 21 – – – – – –M5 89.5 – – – – 17 2 – – – –M6 95 – – – – 1 19 – – – –R1 80 – – – – – – 16 1 1 2R2 85 – – – – – – 1 17 1 1R3 95 – – – – – – 1 – 19 –C1 90 – – – – – – 1 – 1 18

Total 92 22 18 19 22 18 21 19 18 22 21

VARIABILITY OF SYNGONANTHUS MUCUGENSIS (ERIOCAULACEAE) 411

© 2007 The Linnean Society of London, Botanical Journal of the Linnean Society, 2007, 153, 401–416

The majority of the 42 morphological variablesshowed statistically significantly different mean val-ues between the two areas (Mucugê and Rio de Con-tas/Catolés) for P < 0.01 (N = 30) or P < 0.05 (N = 34)in the t-test (Table 2). In these cases, the populationsfrom Rio de Contas and Catolés generally presentedlower values than those from Mucugê.

The structure observed in the analysis of canonicalvariables was reflected in the dendrogram of theMahalanobis distance between populations, providingfurther evidence of the large morphological differenti-ation between the Mucugê and Rio de Contas/Catoléspopulations, and the greater similarity betweenpopulations from the latter region when compared

Figure 4. Dendrograms showing the phenetic relationships between ten populations of Syngonanthus mucugensis occur-ring in the Chapada Diamantina, Brazil. Constructed using the matrix of genetic distances (Nei, 1978; unbiased estimate)based on 14 allozymic loci (A) and the matrix of the Mahalanobis generalized distance based on 42 morphological characters(B) with the unweighted pair-group method of arithmetical averages (UPGMA) as clustering algorithm. See Table 1 forthe names of the populations.

Mahalanobis Generalized distance / UPGMA

0 20 40 60 80 100 120 140

C1

R3

R2

R1

M3

M6

M5

M2

M4

M1

Nei (1978) genetic distance / UPGMA

0,00 0,05 0,10 0,15 0,20 0,25

C1

R3

R2

R1

M6

M5

M4

M3

M2

M1

412 A. C. S. PEREIRA ET AL.

© 2007 The Linnean Society of London, Botanical Journal of the Linnean Society, 2007, 153, 401–416

Figure 5. Analysis of canonical variables. Representation of the scores on the first three canonical axes using 42 morpho-logical characters in ten populations of Syngonanthus mucugensis occurring in the Chapada Diamantina, Brazil. A,Canonical axes 1 and 2. B, Canonical axes 1 and 3. See Table 1 for the names of the populations.

VARIABILITY OF SYNGONANTHUS MUCUGENSIS (ERIOCAULACEAE) 413

© 2007 The Linnean Society of London, Botanical Journal of the Linnean Society, 2007, 153, 401–416

with those from the former (Fig. 4B). It was also foundthat population M3 could be differentiated from theother Mucugê populations, the latter forming twosubgroups.

Mantel’s test demonstrated a significant correlationbetween the Mahalanobis generalized distance andNei’s genetic distance (r = 0.711; P = 0.003), andbetween Nei’s genetic distance and the actual geo-graphical distance between populations (r = 0.711;P = 0.001). However, no significant correlation wasobserved between the Mahalanobis generalized dis-tance and the geographical distance (r = 0.038;P = 0.342).

DISCUSSION

STRUCTURE AND VARIABILITY

From the parameters of variation considered, the pop-ulations of S. mucugensis from Rio de Contas/Catolésshowed low values in relation to the mean valuesreported for plants with similar characteristics (mono-cotyledons, herbs, endemic species, plants with wind-dispersed seeds and sexual reproduction). The Mucugêpopulations, however, showed values that were lowbut compatible with those found by Hamrick & Godt(1990). Some species belonging to families related tothe Eriocaulaceae, such as the Poaceae, also show lowindices of genetic variability (Novak, Mack & Soltis,1991; Wang, Wendel & Dekker, 1995a, 1995b; Godt,Johnson & Hamrick, 1996). The low variability inthese cases could be a result of the genetic markerused; therefore, other genetic markers with greaterresolution in detecting variability should be employedto verify whether the low indices that were originallyfound are maintained.

The endogamy coefficients (Fis) obtained were high,considering that the species is monoecious and thatthere is no temporal overlap in flowering periodbetween pistillate and staminate flowers in the samecapitulum (Ramos, Borba & Funch, 2005). However,this situation was not observed between flowers ofdifferent capitula from the same individual, whichfavours geitonogamy, as the plants are self-compatible(Ramos et al., 2005). S. mucugensis is pollinatedmainly by insects that fly short distances between vis-its (Diptera, Fig. 1B), which results in a large numberof geitonogamous crosses or crosses between closelyrelated individuals. Considering that seed dispersal isnormally over short distances, the resulting pattern ofdistribution of individuals is of family clusters withina population (Ramos et al., 2005). Such conditionsfavour endogamy and could explain the high Fis valuesobtained.

The high values obtained for the genetic (Fst) andmorphological structure for this species, consideringall the populations studied, indicate a large overall

differentiation between populations from the twoareas, as can be seen from a comparison of the struc-ture values obtained when analysing the two localitiesseparately. The genetic structure of the Mucugê pop-ulations is influenced principally by that of populationM1, as a result of an inversion in the frequency of alle-les 100 and 125 in the LAP enzyme system. Withregard to the morphological structure, population M3stands out, showing the greatest differentiation fromthe others from Mucugê, as well as displaying theleast morphological variability. This population,although not the smallest visited, was found to besubstantially degraded, mainly as a result of indis-criminate collecting.

With the use of allozyme markers for species ofAcianthera (formerly Pleurothallis subgenus Acian-thera, Orchidaceae; Borba et al., 2001), and for Pro-teopsis argentea (Asteraceae) (Jesus et al., 2001), bothrestricted to the Cadeia do Espinhaço, recent studieshave found moderate to high levels of genetic struc-ture, associated with a disjunct geographical distribu-tion of their populations, also reflected in theirmorphological structure (Borba et al., 2002). Speciesendemic to this mountain range, with small and frag-mented populations, such as P. argentea, have shownreduced genetic diversity within and increased differ-entiation between populations. In this species, twovery close populations in the Serra do Cipó, separatedby less than 2 km, show frequent inverted alleles andalleles restricted to only one of the populations (Jesuset al., 2001). By contrast, Borba et al. (2001) found anunexpectedly high level of genetic variability in manypopulations studied, and a low to moderate differenti-ation between populations of the same species. InDiscocactus zehntneri Britton & Rose (Cactaceae),another species restricted to the Cadeia do Espinhaço,Machado (2005) also found high Fst and AMRPP values,reflecting genetic and morphological differentiation ofits populations. These were associated with a north/south disjunction in the geographical distribution.Similar results have been obtained by Lambert et al.(2006a, 2006b) for species of Melocactus, anothergenus of Cactaceae.

TAXONOMIC CIRCUMSCRIPTION AND CONSERVATION

This study is the first to involve populations from theChapada Diamantina with an east/west disjunction,the municipalities of Rio de Contas and Abaíra(Catolés is a district of the latter) being situated on thewestern side of the Cadeia do Espinhaço, with Mucugêlocated in the east. These three localities are charac-terized by their floristic richness, with a large numberof endemic species (Harley & Simmons, 1986;Stannard, 1995; Menezes & Giulietti, 2000; Zappiet al., 2000; Souza, 2001).

414 A. C. S. PEREIRA ET AL.

© 2007 The Linnean Society of London, Botanical Journal of the Linnean Society, 2007, 153, 401–416

The low level of genetic identity between popula-tions from the two areas, representing two mountainranges of the Chapada Diamantina, is less than themean values found for conspecific populations andclose to the values recorded for populations of closelyrelated congeneric species (Crawford, 1989; Van derBank, Van der Bank & Van Wyk, 2001). Such data,together with the morphological results, which alsoindicate differentiation between the two areas,suggest that we are dealing with two taxa that aredistinct, but related. Owing to the nature of themorphological differences between the populations ofthe two localities, mainly restricted to size and not tothe form of vegetative and reproductive characters, wepropose subspecific status for the Rio de Contas andCatolés populations of S. mucugensis. A formaldescription of the new subspecies follows.

SYNGONANTHUS MUCUGENSIS SSP. RIOCONTENSIS A.C.S.PEREIRA & GIUL., SSP. NOV.

Type: Brazil, Bahia State, Rio de Contas, Serra doItoibira, 19.iv.2003, Pereira et al. 05 (holotype:HUEFS!).

Diagnosis: A subspecie typica bractea involucraliexteriori interiorique breviore angustioreque foliospatha et anthera brevioribus differt.

Description: Leaf 2.5–4.0 cm long. Spathe 3.7–5.3 cmlong. Scape 28–46 cm long. Involucral bracts, outer1.5–3 mm long, 0.7–1.4 mm wide; inner 6–8 mm long.Styles, free part 1–3.5 mm long. Anthers 0.2–0.8 mmlong.

Etymology: Refers to the municipality of Rio deContas in which the type population is located.

Geographical distribution: This subspecies is endemicto the campos rupestres of the Chapada Diamantina,Bahia state, Brazil, being restricted to the municipal-ities of Rio de Contas, in the Serra do Itoibira, andAbaíra, in the fields of Virassaia in the district ofCatolés.

Representative specimens examined: BRAZIL. Bahia:Abaíra, Catolés, Serra do Bicota, campos do Virassaia,próximo à lapa do vaqueiro, 13°20′38″S, 41°50′15″W,22.iv.2003, Miranda et al. 560 (HUEFS); Cabaceira,Riacho Fundo, atrás da Serra do Bicota, 13°23′S,41°51′W, 25.x.1993, Ganev 2725 (HUEFS); Catolés,Campos do Virassaia, 13°21′S, 41°51′W, 30.xii.1993,Ganev 2665 (HUEFS). Rio de Contas, Serra do Itoi-bira, 13°23′43″S, 41°51′47″W, 19.iv.2003, Pereira et al.02 (HUEFS); 13°23′43″S, 41°51′47″W, 19.iv.2003,Pereira et al. 03 (HUEFS); 13°23′43″S, 41°51′47″W,

19.iv.2003, Pereira et al. 04 (HUEFS); 13°23′43″S,41°51′47″W, 19.iv.2003, Pereira et al. 05 (HUEFS);Trilha Catolés-Caimbola, próximo ao cruzamento parao Bicho, 23.iii.1999, Nascimento & Tabarelli 142(HUEFS); Caiambola, Serra da Mesa, 13°23′S,41°52′W, 19.iv.2003, Giulietti et al. 2404 (HUEFS);13°23′S, 41°51′W, 19.iv.2003, Giulietti et al. 2421(HUEFS).

Notes: Syngonanthus mucugensis ssp. riocontensisA.C.S.Pereira & Giul. is similar to the typical subspe-cies, differing mainly by the length of the leaf, spathe,and scape, length and width of the outer and innerinvolucral bracts, and length of the free part ofthe style and anther; these are all shorter inS. mucugensis ssp. riocontensis.

The taxonomic implications of recognizing a distincttaxon for the populations on the western range of theChapada Diamantina will directly affect the conserva-tion status of S. mucugensis. We suggest that a planfor the conservation and management of the speciesshould consider the two subspecies separately. Thelevels of diversity between populations of the samespecies should be used as a criterion for the prioriti-zation of the conservation of populations (Chamber-lain, 1998), in order to ensure, for species with high Fst

values, the preservation of allelic and genetic diversityimplicit in protecting a larger number of populations(Hamrick et al., 1991). The Mucugê populationsexhibit greater genetic and morphological variabilityand a greater variability of structure than the Rio deContas/Catolés populations, mainly as a result of pop-ulations M1 and M3. Therefore, it is necessary to pro-tect more populations of the former, covering a greaterarea, in order to conserve the genetic variability andevolutionary potential of the subspecies.

Other strategies related to management should alsobe considered. Population M3 should be monitored,and its improvement with genotypes of populationsM2 and M4 should be considered. These are locatednearby and possess greater variability. The same couldbe performed for population M6 in relation to popula-tion M5. The higher values of variability found in thepopulations of Mucugê may also be explained by theconservation efforts in that locality. The more variablepopulations are situated in the Parque Municipal deMucugê (M1 and M2), created specifically to preservethe ‘sempre-vivas’. Population M5 is one of the mostabundant populations and is consequently sought bycollectors; in spite of being outside the park limits, it issituated in a frequently monitored area.

Conversely, the populations of S. mucugensis ssp.riocontensis present less variability and differentia-tion. Thus, as a result of this apparent homogeneity, aunique population may contain a large proportion ofthe genetic variability of the subspecies. However, we

VARIABILITY OF SYNGONANTHUS MUCUGENSIS (ERIOCAULACEAE) 415

© 2007 The Linnean Society of London, Botanical Journal of the Linnean Society, 2007, 153, 401–416

must take into consideration the occurrence of twopossibilities: (1) the genetic variability of the subspe-cies is naturally low; and (2) the studied populationsare genetically depauperate. In both situations, werecommend a search for new populations and anassessment of their variability. If the second hypothe-sis is confirmed, the genetic improvement of the pop-ulations should be considered. However, improvementof the populations of this subspecies using individualsfrom the populations of Mucugê should never be con-sidered. Conversely, if a naturally low variability isconfirmed, a project that ensures the conservation ofthis new taxon should be designed, because this con-dition makes these plants vulnerable to indiscrimi-nate collection and to habitat destruction by humanactivities.

ACKNOWLEDGEMENTS

We thank Raymond M. Harley for improvements tothe manuscript, and Delmar L. Alvim, Euvaldo R.Júnior, Francisco H. F. Nascimento, Oremildes A.Oliveira, Roy A. Funch, and Sabrina M. Lambert forhelping on field excursions. This work was supportedby a grant from Fundo Nacional do Meio Ambiente(FNMA #75/2001). ACSP received a fellowship fromConselho Nacional de Desenvolvimento Científico eTecnológico, Brazil (CNPq). ELB and AMG weresupported by a grant from CNPq (PQ2 and PQ1A,respectively).

REFERENCES

Alfenas AC, Peters I, Brune W, Passador GC. 1991. Eletro-forese de proteínas e isoenzimas de fungos e essênciasflorestais. Viçosa: Universidade Federal de Viçosa.

Ayres MM, Ayres JR, Ayres DI, Santos AS. 2003. Bioestat3.0. Aplicações estatísticas nas áreas das ciências biológicas emédicas. Belém: Sociedade Civil Mamirauá.

Borba EL, Felix JM, Solferini VN, Semir J. 2001. Fly-pollinated Pleurothallis (Orchidaceae) species have highgenetic variability: evidence from isozyme markers. Annalsof Botany 88: 419–428.

Borba EL, Shepherd GJ, van den Berg C, Semir J. 2002.Floral and vegetative morphometrics of five Pleurothallis(Orchidaceae) species: correlation with taxonomy, phylogeny,genetic variability and pollination systems. Annals of Botany90: 219–230.

Brune W, Alfenas AC, Junghans TG. 1998. Identificaçõesespecíficas de enzimas em géis. In: Alfenas AC, ed. Eletro-forese de isoenzimas e proteínas afins: fundamentos e apli-cações em plantas e microorganismos. Viçosa: UniversidadeFederal de Viçosa, 201–328.

Chamberlain JR. 1998. Isozyme variation in Calliandra cal-othyrsus (Leguminosae): its implications for species delimi-

tation and conservation (in population biology). AmericanJournal of Botany 85: 37–47.

Corrias B, Rossi W, Arduino P, Cianchi R, Bullini L.1991. Orchis longicornu Poiret in Sardinia: genetic, morpho-logical and chorological data. Webbia 45: 71–101.

Crawford DJ. 1989. Enzyme electrophoresis and plantsystematics. In: Soltis DE, Soltis PS, eds. Isozymes in plantbiology. Portland: Dioscorides Press, 146–164.

Giulietti AM. 1996. Novas espécies no gênero SyngonanthusRuhl. (Eriocaulaceae) para o Brasil. Boletim de Botânica daUniversidade de São Paulo 15: 63–72.

Giulietti AM, Pirani JR. 1988. Patterns of geographic distri-bution of some plant species from the Espinhaço Range,Minas Gerais and Bahia, Brazil. In: Vanzolini PE, HeyerWR, eds. Proceedings of a workshop on neotropical distribu-tion patterns. Rio de Janeiro: Academia Brasileira deCiências, 39–69.

Giulietti AM, Pirani JR, Harley RM. 1997. Espinhaçorange region. Eastern Brazil. In: Davis SD, Heywood VH,Herrera-Macbryde O, Villa-Lobos J, Hamilton AC, eds.Centres of plant diversity: a guide and strategy for their con-servation, Vol. 3, The Americas. Cambridge: World WildlifeFoundation and IUCN, 397–404.

Giulietti AM, Wanderley MGL, Longhi-Wagner HM,Pirani JR, Parra LR. 1996. Estudos em ‘sempre-vivas’:taxonomia com ênfase nas espécies de Minas Gerais, Brasil.Acta Botanica Brasilica 10: 329–377.

Giulietti N, Giulietti AM, Pirani JR, Menezes NL. 1988.Estudos em sempre-vivas: importância econômica do extrat-ivismo em Minas Gerais, Brasil. Acta Botanica Brasilica 1(Suppl. 2): 179–193.

Godt MJW, Johnson BR, Hamrick JL. 1996. Genetic diver-sity and population size in four rare southern Appalachianplant species. Conservation Biology 10: 796–805.

Hamrick JL, Godt MJ. 1990. Allozyme diversity in plantspecies. In: Brown AH, Clegg MT, Kahler AL, Weir BS, eds.Plant population genetics, breeding and genetics resources.Sunderland, MA: Sinauer, 43–63.

Hamrick JL, Godt MJ, Murawski DA, Loveless MD. 1991.Correlations between species traits and allozyme diversity:implications for conservation biology. In: Falk DA, HolsingerKE, eds. Genetics and conservation of rare plants. New York:Oxford University Press, 75–86.

Harley RM. 1988. Evolution and distribution of Eriope (Labi-atae) and its relatives in Brazil. In: Vanzolini PE, Heyer WR,eds. Proceedings of a workshop on neotropical distributionpatterns. Rio de Janeiro: Academia Brasileira de Ciências,71–120.

Harley RM, Simmons NA. 1986. Florula of Mucugê –Chapada Diamantina, Bahia, Brazil. Kew: Royal BotanicGardens.

Jesus FF, Solferini VN, Semir J, Prado PI. 2001. Localgenetic differentiation in Proteopsis argentea (Asteraceae), aperennial herb endemic in Brazil. Plant Systematics andEvolution 226: 59–68.

Lambert SM, Borba EL, Machado MC. 2006a. Allozymediversity and morphometrics of the endangered Melocactusglaucescens (Cactaceae), and investigation of the putative

416 A. C. S. PEREIRA ET AL.

© 2007 The Linnean Society of London, Botanical Journal of the Linnean Society, 2007, 153, 401–416

hybrid origin of M. × albicephalus (M. ernestii × M. glauce-scens) in north-eastern Brazil. Plant Species Biology 21: 93–108.

Lambert SM, Borba EL, Machado MC, Andrade SCS.2006b. Allozyme diversity and morphometrics of Melocactuspaucispinus (Cactaceae) and evidence for hybridization withM. concinnus in the Chapada Diamantina, north-easternBrazil. Annals of Botany 97: 389–403.

Lazzari LRP. 2000. Redelimitação e revisão de Syngonanthussect. Eulepis (Bong ex. Koern) Ruhland-Eriocaulaceae.Unpublished DPhil Thesis, Universidade de São Paulo.

Levene H. 1949. On a matching problem arising in genetics.Annals of Mathematical Statistics 20: 91–94.

Machado MC. 2005. O gênero Discocactus Pfeiff. (Cactaceae)no estado da Bahia, Brasil: variabilidade morfológica, vari-abilidade genética, taxonomia e conservação. UnpublishedMSc Thesis, Universidade Estadual de Feira de Santana,Bahia.

McCune B, Mefford MJ. 1999. PCOrd – multivariate analy-sis of ecological data, Version 4.10. Gleneden Beach: MJMSoftware.

Menezes NL, Giulietti AM. 2000. Campos Rupestres. In:Mendonça MP, Lins LV, eds. Lista vermelha das espéciesameaçadas de extinção de Minas Gerais. Belo Horizonte:Fundação Biodiversitas e Fundação Zoo-Botânica de BeloHorizonte.

National Geodetic Survey. 2002. INVERSE, Version 2.0.URL http://www.ngs.noaa.gov/ [accessed on 5 January2005].

Nei M. 1978. Estimation of average heterozygosity and geneticdistance from a small number of individuals. Genetics 89:583–590.

Novak SJ, Mack RN, Soltis DE. 1991. Genetic variation inBromus tectorum (Poaceae): population differentiation in itsNorth American range. American Journal of Botany 78:1150–1161.

Paixão-Santos J, Dornelles ALC, Silva JRS, Rios AP.2003. Germinação in vitro de Syngonanthus mucugensisGiulietti. Sitientibus Série Ciências Biológicas 3: 120–124.

Paixão-Santos J, Dornelles ALC, Silva JRS, SantanaJRF, Lima-Brito A. 2006. Ajuste do meio MS para o cultivo‘in vitro’ de Syngonanthus mucugensis Giulietti, espécieameaçada de extinção. Sitientibus Série Ciências Biológicas6: 36–39.

Ramos COC, Borba EL, Funch LS. 2005. Pollination inBrazilian Syngonanthus (Eriocaulaceae) species: evidencefor entomophily instead of anemophily. Annals of Botany 96:387–397.

Ridgway GJ, Sherburne SW, Lewis RD. 1970. Polymor-phism in the esterases of Atlantic herring. Transactions ofthe American Fisheries Society 99: 147–151.

Shaw CR, Prasad R. 1970. Starch gel electrophoresis ofenzymes – a compilation of recipes. Biochemical Genetics 4:297–320.

Silva JRS, Lima-Brito A, Santana JRF, Dornelles ALC.2005a. Efeito da sacarose sobre o enraizamento e desenvolvi-mento ‘in vitro’ de Syngonanthus mucugensis Giul. Sitienti-bus Série Ciências Biológicas 5: 56–59.

Silva JRS, Paixão-Santos J, Rios AP, Santana JRF, Dor-nelles ALC. 2005b. Estudo da germinação e morfologia dodesenvolvimento pós-seminal de Syngonanthus mucugensisGiul. ‘in vitro’. Sitientibus Série Ciências Biológicas 5: 60–62.

Sneath PHA, Sokal RR. 1973. Numerical taxonomy. SanFrancisco: Freeman.

Soltis DE, Haufler CH, Darrow DC, Gastony GJ. 1983.Starch gel electrophoresis of ferns: a compilation of grindingbuffers, gel and electrode buffers and staining schedules.American Fern Journal 73: 9–27.

Souza ER. 2001. Aspectos taxonômicos e biogeográficos dogênero Calliandra Benth. (Leguminosae-Mimosoideae) naChapada Diamantina, Bahia, Brasil. Unpublished MScThesis, Universidade Estadual de Feira de Santana, Bahia.

Stannard BL. 1995. Flora of the Pico das Almas, ChapadaDiamantina, Bahia, Brazil. Kew: Royal Botanic Gardens.

Statsoft Inc. 2000. STATISTICA for Windows (Computer Pro-gram Manual). Tulsa, OK: Statsoft.

Stuber CW, Goodman MM, Johnson FM. 1977. Geneticcontrol and racial variation of β-glucosidase isozymes inmaize (Zea mays L.). Biochemical Genetics 15: 383–394.

Sun M, Ganders FR. 1990. Outcrossing rates and allozymevariation in rayed and rayless morphs of Bidens pilosa.Heredity 64: 139–143.

Swofford DL, Selander RB. 1989. BIOSYS-1: computer pro-gram for the analysis of allelic variation in population genet-ics and biochemical systematics. Champaign, IL: IllinoisNatural History Survey.

Van der Bank H, Van der Bank M, Van Wyk B. 2001. Areview of the use of allozyme electrophoresis in plant sys-tematics. Biochemical Systematics and Ecology 29: 469–483.

Wang R, Wendel JF, Dekker JH. 1995a. Weedy adaptationin Setaria spp. I. Isozyme analysis of genetic diversity andpopulation genetic structure in Setaria viridis. AmericanJournal of Botany 82: 308–317.

Wang R, Wendel JF, Dekker JH. 1995b. Weedy adaptationin Setaria spp. II. Genetic diversity and population geneticstructure in Setaria glauca, S. geniculata and S. faberii(Poaceae). American Journal of Botany 82: 1031–1039.

Wright S. 1978. Evolution and the genetics of populations, Vol.4. Variability within and among natural populations.Chicago, IL: University of Chicago Press.

Zappi DC, Lucas E, Stannard BL, Pirani JR, Queiroz LP,Atkins S, Hind N, Giulietti AM, Harley RM, Mayo SJ,Carvalho AM. 2000. Biodiversidade e conservação naChapada Diamantina, Bahia: Catolés, um estudo de caso. In:Araújo EL, Moura AN, Sampaio EVSB, Gestinari LMS,Monteiro JMT, eds. Biodiversidade conservação e uso suste-ntável da flora do Brasil. Recife: Imprensa Universitária,87–89.