Genetic variability and chromosome-length polymorphisms of the witches' broom pathogen Crinipellis perniciosa from various plant hosts in South America.
ABSTRACT Crinipellis perniciosa has been classified into at least four known biotypes associated with members of unrelated plant families. In this study, genetic variability is shown for 27 C (Cacao), 4 S (Solanum), and 7 L biotype (Liana) isolates of C. perniciosa collected from different regions of Brazil and South America. The objective was to investigate the genetic variability of the pathogen in the cacao-producing region of Bahia, Brazil, and elsewhere, through microsatellite analysis, and attempt to identify possible correlations between host specificity and electrophoretic karyotypes. The PCR-banding patterns were found to vary both within and between the different biotypes, and a correlation was established between the PCR-banding patterns and the chromosomal-banding patterns of each isolate. Microsatellite and chromosomal patterns among all of the L and S biotype isolates were distinctly different from the C biotypes analysed. A higher degree of genetic and chromosomal variability was found among C biotype isolates from the Amazon in comparison with C biotype isolates from Bahia, which seems to be comprised of only two main genotypes. This finding has important implications to the current cacao-breeding programme in Brazil.
Genetic variability and chromosome-length polymorphisms
of the witches’ broom pathogen Crinipellis perniciosa from
various plant hosts in South America
Johana RINCONESa, Gabriel D. MAZOTTIa, Gareth W. GRIFFITHb, Alan POMELAc,
Antonio FIGUEIRAd, Gildemberg A. LEALd, Marisa V. QUEIROZe, Jorge F. PEREIRAe,
Ricardo A. AZEVEDOf, Gonc ¸alo A. G. PEREIRAa,*, Lyndel W. MEINHARDTg
aLaborato ´rio de Geno ˆmica e Expressa ˜o, Departamento de Gene ´tica e Evoluc ¸a ˜o, Instituto de Biologia, Universidade Estadual de Campinas,
CP 6109, Campinas, 13083-970, Sa ˜o Paulo, Brazil
bInstitute of Biological Sciences, University of Wales, Penglais, Aberystwyth, Ceredigion SY23 3DD, UK
cAlmirante Centro de Estudos de Cacau, CP55, Itajuipe, 45630-000, Bahia, Brazil
dCentro de Energia Nuclear na Agricultura, Universidade de Sa ˜o Paulo, CP 96, Piracicaba, 13400-970, Sa ˜o Paulo, Brazil
eUniversidade Federal de Vic ¸osa, Departamento de Microbiologia, CEP 36571-000, Vic ¸osa, Minas Gerais, Brazil
fDepartamento de Gene ´tica, Escola Superior de Agricultura Luiz de Queiroz, Universidade de Sa ˜o Paulo, Piracicaba 13400-970,
Sa ˜o Paulo, Brazil
gSustainable Perennial Crops Laboratory, Plant Science Institute USDA – USDA, Beltsville MD 20705
a r t i c l e i n f o
Received 17 November 2005
Received in revised form
22 March 2006
Accepted 7 May 2006
Intraspecific genetic variability
Pulsed-field gel electrophoresis
a b s t r a c t
Crinipellis perniciosa has been classified into at least four known biotypes associated with
members of unrelated plant families. In this study, genetic variability is shown for 27 C
(Cacao), 4 S (Solanum), and 7 L biotype (Liana) isolates of C. perniciosa collected from differ-
ent regions of Brazil and South America. The objective was to investigate the genetic
variability of the pathogen in the cacao-producing region of Bahia, Brazil, and elsewhere,
through microsatellite analysis, and attempt to identify possible correlations between
host specificity and electrophoretic karyotypes. The PCR-banding patterns were found to
vary both within and between the different biotypes, and a correlation was established
between the PCR-banding patterns and the chromosomal-banding patterns of each isolate.
Microsatellite and chromosomal patterns among all of the L and S biotype isolates were
distinctly different from the C biotypes analysed. A higher degree of genetic and chromo-
somal variability was found among C biotype isolates from the Amazon in comparison
with C biotype isolates from Bahia, which seems to be comprised of only two main geno-
types. This finding has important implications to the current cacao-breeding programme in
ª 2006 The British Mycological Society. Published by Elsevier Ltd. All rights reserved.
Crinipellis perniciosa (Agaricales, Tricholomataceae) is the causal
agent of witches’ broom disease (WBD) of cacao (Theobroma
cacao). This fungal pathogen is believed to have originated in
the Amazon basin and it infects plant species within the fam-
ilies Malvaceae, Solanaceae, Bignoniaceae, Bixacea, and Malpighia-
ceae (Evans 1980; Griffith et al. 2003; Griffith & Hedger 1994b,c;
* Corresponding author. Tel.: þ55 19 37886237; fax: þ55 19 37886235.
E-mail address: firstname.lastname@example.org.
0953-7562/$ – see front matter ª 2006 The British Mycological Society. Published by Elsevier Ltd. All rights reserved.
available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/mycres
mycological research 110 (2006) 821–832
Purdy & Schmidt 1996; Resende et al. 2000). Since its first
reported occurrence in coastal Ecuador in 1894, the disease
has spread to cacao plantations throughout the Americas
and Caribbean islands, causing severe economic losses (Grif-
fith et al. 2003; Pereira et al. 1996; Purdy & Schmidt 1996). At
thistimethisdiseaseis limitedto theAmericas,butis a poten-
tial threat to all tropical cacao-growing regions of the world,
where the cultivation of cacao is typically done by small land-
holders with limited technological inputs that make them
particularly susceptible to this fungal disease.
An example of the devastating impact of the introduction
of this fungus into disease free regions can be found in the
cacao-producing region of southeastern Bahia, Brazil, where
plantations were abandoned or substituted for other crops
(Pereira et al. 1996). This has increased the rate of destruction
of the Atlantic rainforest (‘Mata Atla ˆntica’) and has caused
serious socioeconomic problems in the region. Currently the
only control mechanism is the selection of resistant plants,
which are cloned and distributed to the producers. However,
this selection is based on limited information about the
genetic variability of both the plant and the fungal pathogen.
The broad host-range of C. perniciosa has prompted several
authors to propose the following classification system based
on host specificity: (1) the C biotype infects species of Theo-
broma and Herrania (Malvaceae); (2) the S biotype affects several
members of the Solanaceae ; (3) the L biotype is found on
liana vines (especially the species Arrabidaea verrucosa – Bigno-
niaceae) and associated plant debris (Evans 1978; Griffith &
Hedger 1994b,c; Hedger et al. 1987).
The C and S biotypes cause the characteristic symptoms of
WBD on their respective hosts, whereas the L biotype gener-
ally causes no symptoms (Hedger et al. 1987). Another impor-
tant distinction among the biotypes is that while the C and S
biotypes exhibit primary homothallism (Delgado & Cook
1976; Evans 1980; Griffith et al. 2003; Griffith & Hedger 1994c),
the L biotype has an outcrossing breeding strategy (bifactorial
heterothallism), which is widespread among the agaric fungi
(Griffith & Hedger 1994a,b).
RFLP analyses of the mitochondrial DNA, ITS and IGS re-
gions of the rRNA locus has allowed the separation of the C
and S biotypes and has revealed a clonal population structure
for these biotypes that correlates with their non-outcrossing
breeding strategy (De Arruda et al. 2003; Griffith & Hedger
1994b). In contrast, the L biotypes show a non-clonal popula-
tion structure due to outcrossing and a high degree of genetic
variability (Griffith & Hedger 1994b,c; Griffith 1989). Moreover,
although the genetic variability between isolates from the dif-
ferent biotypes of C. perniciosa have been examined with RAPD
(Andebrhan & Furtek 1994; Andebrhan et al. 1999), Enterobac-
terial Repetitive Intergenic Consensus (ERIC) repetitive ele-
ment sequence-based PCR (De Arruda et al. 2003) and AFLP
(Ploetz et al. 2005), there has been no attempts to correlate
the results. In general, diversity is higher among C biotype iso-
lates from the Amazon region than between isolates from
Bahia and some isolates from the Amazon clustered together
with those from Bahia. These data indicate that the original
introductions into Bahia were probably from Amazonian C
biotype strains (Andebrhan et al. 1999; De Arruda et al. 2003).
In the present study we analysed the genetic and chromo-
somal variation of 38 isolates of C. perniciosa comprising three
different biotypes (C, S, and L) collected from different geo-
nian Ecuador. The objectives were to evaluate the genetic
from the Amazonian region and to establish whether the ge-
Field isolates of Crinipellis perniciosa were collected at different
times, geographic locations, and from different hosts. The 38
isolates examined are listed in Table 1; 27 isolates correspond
to the C biotype and each of these constitutes a single-spore
culture obtained from different basidiomes collected in the
field; four correspond to the S biotype and were obtained as
described for the C biotype; and seven correspond to the L
biotype and were obtained by isolation from stipe tissues of
basidiocarps collected in the field [including one unmated
(monokaryotic) single spore isolate, L7, lacking clamp connec-
tions, the progeny of L1]. C biotypes from other countries were
not evaluated due to phytosanitary concerns and regulations.
S biotypes were taken only from one region of Brazil as they
are only found in this region. All isolates were independently
tested for pathogenicity on their respective hosts (data not
shown). The collection sites are shown schematically in Fig
1. All isolates used in this study can be obtained from the Lab-
orato ´rio de Geno ˆmica e Expressa ˜o at UNICAMP, as well as further
information concerning the collection data.
Genotypic analysis of isolates by PCR amplification
Isolates were analysed with the microsatellite primers,
TeloA1R (CCCTAA)3according to Meinhardt et al. (Meinhardt
et al. 2002a,b) and TeloC1 (TTTACGG)3, a repeat sequence de-
rived from the Crinipellis perniciosa sequencing data. A total
of eight microsatellite primers were initially tested but only
these two are shown in this study because of their specific
results. Reactions were conducted as described by Meinhardt
et al. (Meinhardt et al. 2002a,b).
plug preparations were done as described previously (Rincones
et al. 2003). Chromosomal separations were carried out in a Con-
tour–Clamped Homogeneous Electric Field (CHEF) system (DR-II,
Bio-Rad, Herts) under the conditions described previously (Rin-
cones et al. 2003), with minor modifications. Gels were run at
(TBE)bufferataconstantcurrentof1.4 Vcm?1foratotal354 hdi-
vided into three blocks. Block 1 consisted of pulse intervals pro-
gressively increasing from 2700 s to 5000 s for 172 h; block 2
progressively increasing from 2500 s to 3000 s for 80 h; and b 3
had a constant pulse time of 2200 s for 102 h. Approximate
band sizes were calculated from the relative migration of the
chromosomes of the Schizosaccharomyces pombe size standard
822 J. Rincones et al.
of the ethidium bromide-stained pattern of the chromosomal
bands was performed using the software Image Master?VDS
(Amersham Bioscience, Little Chalfont). CHEF gels were trans-
ferred to membranes (Hybond Nþ, Amersham Biosciences) and
probed with PCR labelled-amplified fragments of: (1) ITS regions
(White et al. 1990) of the ribosomal DNA (rDNA) of C. perniciosa;
transcriptase–RNase H integrase of Tricholoma matsutake (Gen-
tion, hybridization, and washing of the membranes were
conducted according to the manufacturer’s instructions.
Microsatellite-based PCR analysis
The microsatellite PCR primer TeloC1 revealed genotypic dif-
ferences between the various isolates of the L and S biotypes,
but failed to show any difference for C biotype isolates used in
this study (Fig 2). This was particularly interesting as it gave
the same specific banding pattern for all of the C biotype iso-
lates, regardless of their origin, allowing its differentiation
from the other biotypes, therefore suggesting that this primer
could be used as a C biotype indicator (Fig 2C). The primer
TeloA1R (CCCTAA)3(Meinhardt et al. 2002a,b) was able to
Table 1 – Isolates used in this study
Collection no. Host Locationa
C (Bahia) C01*
Itajuı ´pe, BA, Br.
Ilhe ´us, BA, Br.
Belmonte, BA, Br.
Ilhe ´us, BA, Br.
Santo Amaro, BA, Br.
Itabuna, BA, Br.
Itabuna, BA, Br.
Itabuna, BA, Br.
Itabuna, BA, Br.
Aiquara, BA, Br.
Inema, BA, Br.
Ibirataia, BA, Br.
Itagiba, BA, Br.
Ilhe ´us, BA, Br.
Itabuna, BA, Br.
Itabuna, BA, Br.
Itapebi, BA, Br.
C (Amazon) C19
Tabatinga, AM, Br.
Marituba, PA, Br.
Ouro Preto, RO, Br.
Belem, PA, Br.
Altamira, PA, Br.
Medicila ˆndia, PA, Br.
Ariquemes, RO, Br.
Ji-Parana, RO, Br.
Alta Floresta, MT, Br.
Rio Pomba, MG, Br.
Coimbra, MG, Br.
Rio Pomba, MG, Br.
Poc ¸os de Caldas, MG, Br.
San Carlos, NA, Ec.
Pichilingue, RI, Ec.
San Carlos, NA, Ec.
Pichilingue, RI, Ec.
Pichilingue, RI, Ec.
Pichilingue, RI, Ec.
San Carlos, NA, Ec.
*Reference isolate used for the genome project (www.lge.ibi.unicamp.br/vassoura)
a Origin of isolates: Br., Brazil; BA, Bahia State, except for C05, all C (Bahia) isolates were collected within a region of approximately 90 km2of
Southeastern Bahia; the States of Amazonas(AM), Mato Grosso (MT), Rondo ˆnia (RO),and Para ´ (PA) correspond to the Brazilian Amazon; the State
of Minas Gerais (MG) is located in Southeastern Brazil; Ec., Ecuador; NA, Napo Province in Amazonian Ecuador; RI, Los Rios Province in Coastal
Ecuador (Fig 1).
Genetic variability in C. perniciosa823
separate most of the biotype C, L and S isolates into specific
groups (Fig 3B–C). The C-biotype isolates from Bahia were
separated into two groups with the TeloA1R primer, CG1
(comprising isolates C01, C02, C03, C04, C08, C09, C11, C14,
C15 and C17) and CG2 (comprising isolates C05, C06, C07,
C10, C12, C13, C16, and C18). The nine Amazon isolates were
separated into seven groups with the three isolates from
Para C20, C22 and C23 representing a single group (Fig 3A).
All of the L and S biotype isolates showed genetic variations
with the TeloA1R primer (Fig 3B–C). In the case of the L biotype
isolates, this primer revealed a somewhat similar banding
patterns for L1 and its progeny L7 indicating that alterations
in a given population could possibly be monitored with this
primer. All PCR amplicons for the different primers ranged
from 3000 bp to 500 bp in size.
Fig 4A shows the four different karyotypes obtained for the
18 C biotype isolates (C01-C18) collected in various regions
of Southeastern Bahia: C01–C18 (Table 1). Most of the isolates
from this region could be grouped into two different
Fig 1 – Maps detailing the sites for the collection of the isolates of Crinipellis perniciosa listed in Table 1. A. Map of South
America detailing the sites for the collection of isolates of C. perniciosa in Ecuador. B. Map of Brazil shows the sites of isolate
collection in the Brazilian Amazon and Southeastern Brazil. C. Map of Bahia shows sites of isolate collection in the south-
eastern cacao-growing region of this state. The shaded area represents the cacao-growing region in the state of Bahia.
824 J. Rincones et al.
karyotypes that were designated as CG1 (comprising isolates
C01, C02, C03, C04, C09, C11, C14, and C17) and CG2 (compris-
ing isolates C05, C06, C07, C10, C12, C13, C16, and C18). The
other two karyotypes were exhibited by isolates C08 and C15
and they are both very similar to karyotype CG1: in the case
of isolate C08, its karyotype showed an extra band of approx-
imately 4.5 Mb in size (arrow on Fig 4A) while the karyotype of
isolate C15 presented a slightly smaller fifth band (arrow on
Fig 4A). In contrast, all nine isolates collected in the Brazilian
Amazon (C19–C27, Table 1) exhibited different karyotypes,
varying in band number and sizes, as shown in Fig 4D.
Fig 5A shows the karyotypes obtained for all isolates of the
S and L biotypes. The CG1 karyotype is shown for comparison
purposes. All isolates examined for these two biotypes
exhibited different karyotypes, varying in band number and
A densitometric analysis of the ethidium bromide-stained
banding pattern was performed for all karyotypes. Two of
these analyses are shown as examples (Fig 4B–C) to illustrate
that the relative intensity of the ultraviolet fluorescence was
higher for some bands. This result suggests that each of these
bands represent at least two chromosomes, which could be
homologous or heterologous chromosomes of similar sizes.
Results from the densitometric analysis were taken into
consideration when calculating total genome sizes for each
isolate (Table 2). Table 2 lists the approximate sizes of the
bands and summarizes the electrophoretic karyotypes for
each isolate, grouped by biotypes and geographic origin.
Fig 5B illustrates the results of some of the Southern hybrid-
izations; however, all hybridization results are summarized
in Table 2. Southern hybridization of blots from the pulsed-
field gels with the ITS probe shows that the rDNA repetitions
were usually clustered to a single chromosome, independent
of the biotype (Fig 5B, Table 2). Isolate L6 was the only excep-
tion with the ITS probe hybridizing to two chromosomal
bands (Fig 5B, Table 2). The rDNA repetitions were usually
found on one of the two largest chromosomes, except in
isolate C25, where the ITS probe hybridized to the third chro-
mosomal band (Fig 5B, Table 2).
The probe containing a sequence similar to a reverse
transcriptase–RNase H integrase hybridized to most bands of
every isolate examined with the exception of the smaller
bands in some isolates (Fig 5B, Table 2). This result suggests
CG1CG2 C19C20 C21C22C23 C24C25C26MC27
L2 L3L4 L5 L6L7S1S2S3MS4
Fig 2 – Microsatellite primer based analysis with the primer TeloC1. A. The banding profile for the seven L biotype isolates
amplified with the TeloC1 primer. B. PCR banding pattern of the four S biotype isolates amplified with primer TeloC1. C.
Comparison of nine out of the 11 C biotype groups, CG1 and CG2 correspond to the two chromosomal groups from Bahia
and the other isolate designations are as listed in Table 1; isolates C08 and C15 showed the same amplicon band pattern
seen for CG1. The isolate designations are as stated in Table 1. M indicates the 1 kb DNA standard from Invitrogen.
Genetic variability in C. perniciosa825
that retrotransposons are widespread throughout the genome
of all isolates examined.
In this study, we report genetic and chromosomal-level vari-
ability of isolates within and among three different biotypes
of the Crinipellis perniciosa species complex, collected from dif-
ferent geographic locations. This constitutes one of the most
comprehensive studies reported to date for this species with
regards to the number of isolates from different basidiomes,
geographic distance between collection points for the various
isolates, and number of different biotypes examined. We also
present for the first time the electrophoretic karyotypes of S
and L biotypes of C. perniciosa. This study is the first in a series
of reports that will attempt to form a cohesive understanding
of genetic variability found in the various biotypes of this
fungal pathogen, which will be used to formulate how the
variability affects the pathogen–host interaction as well as
RAPD fingerpriniting by Andebrhan et al. (1999) of isolates
collected at the beginning of the WBD outbreak in Bahia, sug-
gested that there had been two discrete introductions of the
pathogen (i.e. two independent genotypes), a hypothesis
that correlates with the presence of only two different karyo-
types among more recently collected samples (Rincones et al.
2003) and only two populations detected in Bahia through
AFLP analysis (Ploetz et al. 2005). This evidence clearly shows
the stability of the C genotypes in Bahia, which have remained
almost unchanged from the time of their introduction. Due to
this natural genetic stability of the clonal populations of the C
and S biotypes and their non-outcrossing reproductive strat-
egy, it is possible to supplement the genetic variability
M L1L6 L2 L3L4L5 L7MS1S2 S4S3
Fig 3 – Microsatellite primer-based analysis with the primer TeloA1R. A. Comparison of the banding patterns for C
biotype isolates from the Amazon C19 to C27 and the representative chromosomal groups from Bahia CG1 and CG2 and
the two CG1 subgroups, C08 and C15 amplified with the microsatellite primer TeloA1R. B. PCR banding profiles for the
seven L biotype isolates. C. Banding profile for the four S biotype isolates from Minas Gerais. The isolate designations are
as listed in Table 1. M indicates the 1 kb DNA standard from Invitrogen.
826J. Rincones et al.
analysis with chromosomal analysis via electrophoretic
With a single exception, all C biotype isolates collected in
Southeastern Bahia (C01 to C18) possessed eight chromo-
somes and a total genome size of approximately 30 Mb;
only one isolate (C08) possessed nine chromosomes with an
estimated total genome size of 34.8 Mb. In fact, C biotype iso-
lates collected in this region were very homogeneous in their
karyotypes, with eight isolates presenting karyotype CG1 and
another eight isolates presenting karyotype CG2 (Table 2).
These two karyotypes are identical to those previously
reported in Bahia, comprising only four isolates that were
included in this work (Rincones et al. 2003) and these two
chromosomal groups (CG1 and CG2) are supported by the
microsatellite primer, TeloA1R, which separated all of the
C-biotype isolates from Bahia into two groups that were
identical to the chromosomal groups. The two additional kar-
yotypes for this region represented by isolates C08 and C15
are remarkably similar to karyotype CG1. C15 differs from
CG1 in the size of the fifth band, which appears slightly
smaller, whereas isolate C08 differs from CG1 with an addi-
tional band of approximately4.5 Mb.Furthermore,
CG1C19 C20C21 C22C23C24C25 C27C26
Fig 4 – CLPs between isolates of the C biotype of Crinipellis perniciosa from Bahia-Brazil. A. Four different karyotypes obtained
for the 18 C biotype isolates collected in Southeastern Bahia (C01-C18, Table 1). B. Densitometric analysis of the ethidium
bromide-stained pattern of karyotype CG1, numbers represent the integrated optical density (IOD) for the respective peaks.
C. Densitometric analysis of the ethidium bromide-stained pattern of the karyotype of isolate C08, numbers represent
the IOD of the respective peaks. CG1: karyotype of the C biotype group 1 consisting of isolates C01, C02, C03, C04, C09, C11,
C14, and C17. CG2: karyotype of the C biotype group 2 consisting of isolates C05, C06, C07, C10, C12, C13, C16, and C18. Sizes
correspond to the relative migration of the three large chromosomes of Schizosaccharomyces pombe (strain 972h, Bio-Rad).
Genetic variability in C. perniciosa 827
polymorphisms associated with the hybridization of the
rDNA between these four karyotypes from Bahia lends sup-
port to the CG1 origin of C08 and C15. The data showed
that this marker hybridized to the second largest chromo-
some (4.6 Mb) in karyotypes CG1, C08 and C15 and the largest
chromosomal band (5.3 Mb) in the CG2 karyotype (Fig 5B,
Table 2). These apparent subgroups of CG1 could not be
detected by any of the microsatellite markers tested, which
is a limitation of PCR-based analysis, especially in the
detection of chromosomal rearrangements that can only be
revealed by pulsed-field gel electrophoresis.
These subgroups of CG1 are possibly the result of very
recent differentiation events and the chromosomal differ-
ences between isolates C08 and C15 in comparison to CG1
are subtle and could have arisen from the CG1 karyotype in
a single event: either a deletion or a translocation of a portion
of the fifth chromosome in the case of isolate C15; and a dupli-
cation in the case of isolate C08. It is also probable that band 2
S2S3 S4 L1 L2L3 L4L5 L6 L7
S01ITS L02ITSRT L06ITS
Fig 5 – Analysis of the CLPs between isolates of the S and L biotypes of Crinipellis perniciosa from South America. A. CLPs
between isolates of the S and L biotypes of the Crinipellis perniciosa species complex together with reference karyotype CG1.
Sizes correspond to the relative migration of the three large chromosomes of Schizosaccharomyces pombe (strain 972h, Bio-
Rad). B. Examples of the Southern hybridization of the gels shown in Figs 4A, 4D and 5A with the two probes tested. ITS:
probe consisting of the ITS-amplified region of the rRNA locus of reference isolates C01. RT: probe consisting of a PCR-am-
plified insert of a genomic clone from isolate C01 showing high sequence similarity (BLASTx, E value 9E-13) to a reverse
transcriptase–RNase H integrase of Tricholoma matsutake (GenBank accession no. AY661428). Isolate designations are as in
Table 1. Results of all hybridization experiments are summarized in Table 2.
828J. Rincones et al.
of the CG1 karyotype represents two chromosomes, and
a small portion of one of these chromosomes was translo-
cated, thus generating two different sized chromosomes in
isolate C08. The densitometric analysis of isolate C08, which
shows bands 2 and 3 are less intense than bands 1 and 4 (Fig
4C), supports this latter hypothesis and suggests that the
CG1 karyotype could have doublets at chromosomes 1, 2,
and 3. This would increase the total genome size of the CG1
isolates to 40.1 and the total number of chromosomes to 10.
Although thisis a large increasein relationto the size reported
previously, the total genome size determined by the Feulgen
(32.98?7.94 Mb) (Rincones et al. 2003). Furthermore, bioinfor-
matic analyses of the shotgun assemblies have generated
a 37 Mb estimation of the total genome size (unpublished
data), which also supports the hypothesis of a larger genome.
The microsatellite analysis of the nine C biotype isolates
collected in the Brazilian Amazon region (C19 to C27, Table
2) with the PCR primer TeloA1R separated all of them into dis-
crete groups, which corresponds to chromosomal variability
except for isolates C20, C22 and C23 (Figs 3A, 4D).
Karyotypes for the C biotype Amazon isolates differed
chromosomes, sizes, and total genome size. These isolates
had eight to ten chromosomes and their total genome sizes
varied from 30.8–39 Mb (Table 2). The ITS probe hybridized
in most cases to band 2, with the exceptions of isolate C19,
in which the ITS probe hybridized to the largest chromosome,
and isolate C25, in which the rDNA chromosome was located
on band 3 (Fig 5B, Table 2). The size of the rDNA chromosome
for the Amazonian C biotype isolates varied considerably
more, from 3.8 and 5.0 Mb, in comparison with the Bahia iso-
latesin which thesize of the rDNAchromosomewas either 4.6
or 5.3 Mb. This variation in the size of the rDNA chromosome
accumulate over time and would result in alterations that per-
sist in new clonal populations. The isolates C20, C22 and C23,
which were grouped together by the TeloA1R primer, are from
the same state and two of the isolates are from locations in
close proximity, suggesting that the similarity observed could
have been derived from a common ancestor. This agrees with
the findings of Ploetz et al. (2005), who identified genetic ho-
mogeneity in the isolates of specific geographic regions. We
detected only slight chromosome variations (Fig 4D) among
these isolates, which supports this possibility, and this could
be a situation analogous to the subgroups of CG1 found in
Table 2 – Estimated sizes and number of chromosomes of the 38 isolates examined of the Crinipellis perniciosa species
complex. The chromosomal bands were numbered arbitrarily according to size in decreasing order
Genetic variability in C. perniciosa829
Bahia. If so, this could suggest that chromosomal rearrange-
ments are the first step in population differentiation. Repro-
duction of the C. perniciosa species complex is strictly sexual
via meiosis within basidia (Delgado & Cook 1976; Griffith &
Hedger 1994c), and in agreement with the meiotic mainte-
nance hypothesis proposed by Kistler and Miao (1992), in
Bahia, where the fungus has been established for at least
15 y, only a very low chromosomal variation (chromosome
length polymorphisms – CPLs) could be detected. Although
the meiotic maintenance hypothesis is supported by observa-
tions in several fungal species (Kistler & Miao 1992; Po ¨ggeler
et al. 2000; Zolan 1995) there are reports of sexually reproduc-
ing fungi, such as Coprinus cinereus (Zolan et al. 1994), Septoria
tritici (McDonald & Martinez 1991), Leptosphaeria maculans
(Plummer & Howlett 1993), Ophiostoma ulmi (Dewar & Bernier
1995), and Pythium sylvaticum (Martin 1995), with extensive
CLPs that are generated and maintained through meiotic
processes without apparent reduction in fertility. The C and
S biotypes exhibit primary homothallism (non-outcrossing)
and any polymorphisms that arise could in principle be per-
petuated by autofertilization, which could explain the origin
of C08 and C15. If so, further analysis of isolates from Bahia
may reveal additional subgroups that have originated from
In addition to the TeloA1R microsatellite primer that corre-
lates with chromosomal variation, a C-biotype specific PCR
primer (TeloC1) was found (Fig 2). Data derived from this
primer show that the C S and L biotypes appear to be distinct
groups within the species complex, thus reinforcing the dif-
ferentiation into the various biotypes.
All four of the S biotype isolates and the seven L biotype
isolates examined revealed distinctly different microsatellite
groups and karyotypes. (Figs 3B–C, 5A, Table 2). The S biotype
showed the most variation, with karyotypes that exhibited
between eight and 12 chromosomes and their total genome
sizes varied from 30.4–46.1 Mb. In the case of the L biotype,
the karyotypes presented between eight and ten chromo-
somes and the genome sizes ranged from 30.1 and 40.7 Mb.
Except for the presence of bands smaller than 2.7 Mb in
some isolates of the L biotype, no significant differences
were observed between the karyotypes of the three biotypes
that would allow a general correlation of karyotype with
host specificity. However, the limited chromosomal-level ge-
netic variability observed in the C biotype isolates from Bahia
in comparison with C biotypeisolatesfrom theAmazonandto
the S biotype isolates found in the neighbouring state of Minas
Gerais, Brazil, support the hypothesis that C biotypes were in-
troduced into Bahia from the Amazon region (Andebrhan et al.
Southern hybridization of the CHEF gels blots for the S
biotype revealed that the ITS probe hybridized to the first or
second largest band, but the size of this band varied only
from 4.8–5.0 Mb among the four isolates (Fig 5B, Table 2).
The situation was very similar in the L biotype, with the
rDNA chromosomes varying only from 4.8–5.2 Mb in size,
with a single remarkable exception, isolate L6, in which the
ITS probe hybridized to bands 1 and 3 (5.3 and 4.3 Mb, respec-
tively). Isolate L6 was the only one of the 38 isolates examined
that exhibited two rDNA chromosomes (Fig 5B, Table 2). This
situation could have arisen from a duplication of part of the
rDNA chromosome; however, our data are too limited to allow
any conclusions on this matter. Polymorphism in the rDNA
chromosome has been reported in numerous species of fungi
and may be produced by a number of mechanisms (Zolan
1995). CLPs among L biotype isolates were detected even
between isolates collected only a few meters apart. These iso-
lates belonged to different somatic compatibility groups but
may have been related as evidenced by shared mating type
factors, as was the case with isolates L6 and L7 collected
from the same vine in Bosque Viejo, Pichilingue Province,
Ecuador, which shared two mating type factors (A9 and B10)
and thus may have been sib-related (Griffith & Hedger
1994b). A comparison of the karyotypes of the parental isolate
L1 and its monokaryotic progeny L7 shows that some of the
bands from the parental isolate (L1) are conserved but differ-
ent-sized bands arise in the progeny (L7: bands 1, 3 and 5;
Fig 5A, Table 2).
The Southern hybridization of the CHEF gel blots to a geno-
mic probe showing high sequence similarity to a reverse tran-
scriptase from C01 marked most of the chromosomal bands of
all the isolates analysed (Fig 5B, Table 2). In addition, prelimi-
nary gene expression analysisof thissameclone throughDNA
microarray technology suggests that the presence of cocoa
extracts activate these genes (M. Sabha, pers. comm.), and
this could be a mechanism for introducing genetic variation,
given the fact that this fungus must undergo meiosis in order
to produce basidiospores (Delgado & Cook 1976; Evans 1980).
The role of transposable elements in the reorganization of
a fungal genome through ectopic recombination or simply
by their activation has already been reported in several fungal
species, such as Fusarium oxysporum (Davie `re et al. 2001), Mag-
naporthe grisea (Dobinson et al. 1993), Schizosaccharomyces
pombe (Levin et al. 1990), Saccharomyces cerevisiae (Boeke
1989), and Neurospora crassa (Kinsey & Helber 1989). A similar
mechanism might be at work in C. perniciosa and several fam-
ilies of transposable elements have already been detected in
the genome of this phytopathogen (Araujo et al. 2004; Pereira
et al. 2003).
This study shows that genetic variation between isolates
within a given biotype and between biotypes can be deter-
mined with microsatellite-based PCR combined with chromo-
somal analysis. Further studies of additional isolates of this
pathogen, along with detailed mapping of locations and hosts
from Bahia and from the entire Amazon region would allow
for a more complete genetic variability evaluation of the fun-
gus. Future studies are now needed to utilize this information
together with pathogenicity data to develop an overall picture
of how genetic variability within this species complex effects
the pathogen–host interaction.
Furthermore, our results suggest that the chromosomal
rearrangements observed in C. perniciosa are generated
through meiotic processes, which may involve the presence
of multiple copies of retrotransposons. Therefore, it is impor-
tant that isolates comprising as many C biotype genetic
variability groups/karyotypes as possible should be taken
into consideration when assessing resistance of cacao clones.
Our results indicate the fragility of the current breeding pro-
gramme being conducted in Brazil. All cacao clones selected
have been challenged against only the two main genotypes
of C. perniciosa present in Bahia, which may be a very limited
830 J. Rincones et al.
strategy that would render the culture very susceptible to new
introductions from the Amazon. Therefore, it is imperative
that a new concept for this programme be considered and
that phytosanitary measures are strengthen in order to pre-
vent the entrance of new C biotype isolates into Bahia, thus
possibly extending the time for resistance breakdown (Bartley
1986) of the newly-planted resistant cacao clones.
The authors wish to thank Maricı ´lia De Arruda, Karina Grama-
cho, Paulo Albuquerque, Ju ´lio Cascardo, and Robert Weigart
Barreto for kindly providing isolates used in this study, to
Maricene Sabha for the use of unpublished data, to Eduardo
Formighieri for bioinformatics support, and for the ongoing
corporate support given by Fazenda Almirante Cacao (Mars)
and Cargill. We also thank two anonymous reviewers for sug-
gestions that generally improved this manuscript. This work
was supported by FAPESP (Fundac ¸a ˜o de Amparo a ` Pesquisa
do Estado de Sa ˜o Paulo, Nos. 00/10545-4 amd 02/09280-1)
CNPq (Conselho Nacional de Pesquisa e Desenvolvimento,
Nos. 68.0032/01-0; 47.1609/03-0) and SEAGRI (Secretaria de
Agricultura do Estado da Bahia).
r e f e r e n c e s
Andebrhan T, Figueira A, Yamada MM, Cascardo J, Furtek DB,
1999. Molecular fingerprinting suggests two primary outbreaks
of witches’ broom disease (Crinipellis perniciosa) of Theobroma
cacao in Bahia, Brazil. European Journal of Plant Pathology 105:
Andebrhan T, Furtek DB, 1994. Random amplified polymorphic
DNA (RAPD) analysis of Crinipellis perniciosa isolates from dif-
ferent hosts. Plant Pathology 43: 1020–1027.
Araujo EF, Cascardo J, Carazzolle M, Pereira GAG, Queiroz MV,
2004. Boto, um elemento transponı ´vel putativo da classe II em
Crinipellis perniciosa. Ana ´lise e distribuic ¸a ˜o em diferentes bio ´-
tipos e regio ˜es geogra ´ficas. Anais da XXIV Reunia ˜o de Gene ´tica de
Microrganismos, 2004, Gramado, Rio Grande do Sul. Universidade
Federal de Rio Grande do Sul, Porto Alegre, 125.
Bartley BGD, 1986. Cacao. Breeding for Durable Resistance In
Perennial Crops. In: Theobroma cacao. Food and Agriculture Or-
ganization of the United Nations, Rome, p. 25–42.
Boeke JD, 1989. Transposable elements in Saccharomyces cerevisiae.
In: Berg DE, Howe MM (eds), Mobile DNA. American Society for
Microbiology, Washington, DC, pp. 335–374.
Davie `re JM, Langin T, Daboussi MJ, 2001. Potential role of tans-
posable elements in the rapid reorganization of the Fusarium
oxysporum genome. Fungal Genetics and Biology 34: 177–192.
De Arruda MCC, Miller RNG, Ferreira MASV, Felipe MSS, 2003.
Comparison of Crinipellis perniciosa isolates from Brazil by ERIC
repetitive element sequence-based PCR genomic fingerprint-
ing. Plant Pathology 52: 236–244.
Delgado JC, Cook AA, 1976. Nuclear condition of basidia, basid-
iospores, and mycelium of Marasmius perniciosus. Canadian
Journal of Botany 54: 66–72.
Dewar K, Bernier L, 1995. Inheritance of chromosomes-length
polymorphisms in Ophiostoma ulmi (sensu lato). Current Genetics
Dobinson KF, Harris RE, Hamer JE, 1993. Grasshopper, a long
terminal repeat (LTR) retroelement in the phytopathogenic
fungus Magnaporthe grisea. Molecular Plant-Microbe Interactions
Evans HC, 1978. Witches’ broom disease of cocoa (Crinipellis
perniciosa) in Ecuador. I. The fungus. Annals of Applied Biology
Evans HC, 1980. Pleomorphism in Crinipellis perniciosa, causal
agent of Witches’ broom disease of cocoa. Transactions of the
British Mycological Society 74: 515–526.
Griffith GW, 1989. Population structure of the cocoa pathogen
Crinipellis perniciosa (Stahel) Sing. PhD Thesis, University of
Wales, Aberystwyth, UK.
Griffith GW, Hedger JN, 1994a. Dual culture of Crinipellis perniciosa
and potato callus. European Journal of Plant Pathology 100: 371–
Griffith GW, Hedger JN, 1994b. Spatial distribution of mycelia of
the liana (L-) biotype of the agaric Crinipellis perniciosa (Stahel)
Singer in tropical forest. New Phytologist 127: 243–259.
Griffith GW, Hedger JN, 1994c. The breeding biology of biotypes of
the witches’ broom pathogen of cocoa, Crinipellis perniciosa.
Heredity 72: 278–289.
Griffith GW, Nicholson JN, Nenninger A, Birch RN, Hedger JN,
2003. Witches’ brooms and frosty pods: two major pathogens
of cacao. New Zealand Journal of Botany 41: 423–435.
Hedger JN, Pickering V, Aragundi JA, 1987. Variability of popula-
tions of the witches’ broom disease of cocoa (Crinipellis
perniciosa). Transactions of the British Mycological Society 88: 533–
Kinsey JA, Helber J, 1989. Isolation of a transposable element from
Neurospora crassa. Proceedings of the National Academy of Science,
USA 86: 1929–1933.
Kistler HC, Miao VPW, 1992. New modes of genetic change in
filamentous fungi. Annual Review of Phytopathology 30: 131–152.
Levin HL, Weaver DC, Boeke JD, 1990. Two related families of
retrotransposons from Schizosaccharomyces pombe. Molecular
and Cellular Biology 10: 6791–6798.
Martin F, 1995. Meiotic instability of Pythium sylvaticum as dem-
onstrated by inheritance of nuclear markers and karyotype
analysis. Genetics 139: 1233–1246.
McDonald BA, Martinez JP, 1991. Chromosome length polymor-
phisms in a Septoria tritici population. Current Genetics 19: 265–
Meinhardt LW, Wulf NA, Bellato CM, Tsai SM, 2002a. Telomere
and microsatellite primers reveal diversity among Sclerotinia
sclerotiorum isolates from Brazil. Fitopatologia Brasileira 27: 211–
Meinhardt LW, Wulf NA, Bellato CM, Tsai SM, 2002b. Genetic
analyses of Rhizoctonia solani isolates from Phaseolus vulgaris
grown in the Atlantic Rainforest region of Sa ˜o Paulo, Brazil.
Fitopatologia Brasileira 27: 259–267.
Pereira JFI, Araujo EF, Cascardo J, et al., 2003. Ana ´lise de seque ˆn-
cias putativas de transcriptase reversa em Crinipellis perniciosa.
In: Anais do 49?Congresso Nacional de Gene ´tica, 2003, A´guas de
Lindo ´ia, Sa ˜o Paulo. Sociedade Brasileira de Gene ´tica, Ribeira ˜o
Preto, p. 1156.
Pereira JL, de Almeida LC, Santos SM, 1996. Witches ´ broom
disease of cocoa in Bahia: attempts at eradication and con-
tainment. Crop Protection 15: 743–752.
Ploetz RC, Schnell RJ, Ying Z, et al., 2005. Analysis of molecular
diversity in Crinipellis perniciosa with AFLP markers. European
Journal of Plant Pathology 111: 317–326.
Plummer KL, Howlett BJ, 1993. Major chromosomal length
polymorphisms are evident after meiosis in the phytopath-
ogenic fungus Leptosphaeria maculans. Current Genetics 24:
Po ¨ggeler S, Masloff S, Ku ¨ck U, 2000. Karyotype polymorphism
correlates with intraspecific infertility in the homothalic as-
comycete Sordaria macrospora. Journal of Evolutionary Biology 13:
Genetic variability in C. perniciosa 831
Purdy LH, Schmidt RA, 1996. Status of cacao Witches’ broom:
biology, epidemiology and management. Annual Review of
Phytopathology 34: 573–594.
Resende MLV, Gutemberg BAN, Silva LHCP, et al., 2000. Crinipellis
perniciosa proveniente de um novo hospedeiro, Heteropterys acu-
tifolia, e ´ patoge ˆnico a T. cacao. Fitopatologia Brasileira 25: 88–91.
Rincones J, Meinhardt LW, Vidal BC, Pereira GA, 2003. Electro-
phoretic karyotype analysis of Crinipellis perniciosa, the causal
agent of witches’ broom disease of Theobroma cacao. Mycologi-
cal Research 107: 452–458.
White TJ, Bruns T, Lee S, Taylor J, 1990. Amplification and direct
sequencing of fungal ribosomal RNA genes for phylogenetics.
In: Innis MA, Gelfand DH, Sninsky JJ, White TH (eds), PCR
protocols: A Guide to Methods and Applications. Academic Press,
San Diego, pp. 315–322.
Zolan ME, 1995. Chromosome-length polymorphism in fungi.
Microbiological Reviews 59: 686–698.
Zolan ME, Heyler NK, Yeager-Stassen N, 1994. Inheritance of
chromosome-length polymorphisms in Coprinus cinereus.
Genetics 137: 87–94.
832J. Rincones et al.