This item is only available as the following downloads:
PHENOTYPIC AND MOLECULAR CHARACTERIZATION OF CERCOSPORA
SPECIES PATHOGENIC TO WATERHYACINTH
DAURI JOSE TESSMANN
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
To lone, my wife.
To my father and mother.
I thank Dr. Raghavan Charudattan, chairman of my graduate supervisory
committee, for his support, patience, guidance, and encouragement through all stages of
my graduate studies. I thank Dr. H. Corby Kistler for his guidance, counsel, and
encouragement. I thank my committee members, Dr. James W. Kimbrough, Dr. Richard
D. Berger, and Dr. Maria Gallo-Meagher for their support and assistance. My special
thanks to Dr. Gerald Benny for his assistance and advice. I also thank Dr. James F.
Preston III for his advice and assistance and Dr. David J. Mitchell for his counsel. I thank
Jim DeValerio for his assistance and friendship. I extend my thanks to Rodney Pettway,
Ellen Dickstein, Kris Beckham, and Eldon Philman for their help with DNA techniques,
fatty acid analysis, spectral analysis, and greenhouse work. I thank Dr. Erin N. Rosskopf,
Dr. Manjunath Keremane, Dr. Margaret Smither-Kopperl, Dr. Wellington Pereira, and
Dr. Lisias Coelho for their help, support, and friendship. I thank Mr. Jay Harrison for his
advice with statistical analysis. My special thanks to Dr. Jose C. Dianese and his wife
Heloisa Dianese, for having encouraged me to continue my studies and pursue this
degree. I also thank Dr. Robinson A. Pitelli for his support and friendship. I extent my
appreciation to my friends and classmates and laboratory colleagues, Dr. Jugah Kadir, S.
Chandramohan, Camilla Yandoc, Angela Vincent, Dr. Yasser Shabana, Dr. Gabriela
Wyss, and Matt Petterson, in the plant pathology department. I thank the Brazilian
government and its agency CNPq, for my scholarship, the Florida/Brazil Institute for a fee
waiver, and the Universidade Estadual de MaringA for providing me a study leave to
pursue this degree. Finally, I extend my sincere thanks to Dr. George Agrios, chairman of
the plant pathology department, and the official staff of the department.
TABLE OF CONTENTS
LIST OF TABLES .................................................................................................... vii
LIST OF FIGURES.................................................................. ..............................viii
ABSTRACT.................. .......................................................................... x
1 INTRODUCTION........................................................................................ 1
Reproductive Biology, Origin, and Dispersal of Waterhyacinth............................... 2
Biologial Control of Waterhyacinth .................................................. ...................... 5
The Species of Cercospora as Agents for Biological Control.................................. 11
R research R ationale ................................................................................................ 13
2 A MOLECULAR CHARACTERIZATION OF CERCOSPORA SPECIES
FROM WATERHYACINTH...................................................................... 15
Introduction ................................................................................................ 15
Materials and Methods ................. ........................................................... 18
Fungal Isolates..................................... ................. ............... 18
D N A Isolation ................................................................................................. 19
DNA Amplification and Sequencing ...................................... ........................ 20
Southern Blot Analysis of the 3-Tubulin Gene.................................................. 23
Phylogenetic Analysis ....................................................................................... 23
Results ........................................... ....................... ....................... 25
Identification of Isolates.................................................................................. 25
Phylogenetic Analysis ...................................................................................... 29
Discussion ................................................................. .......................... 52
3 PATHOGENIC VARIABILITY AND BIOCHEMICAL
CHARACTERIZATION OF CERCOSPORA SPECIES FROM
W ATERHYACINTH ....... ........................ ........................................................61
Introduction ....................................... ...................... ..... ......... 61
M materials and M ethods................................................................... ....................... 63
Fungal Isolates and Cultural Characteristics................................................... 63
Toxin A analysis ................................................................................................ 64
Pathogenicity Test and Virulence Analysis..................................................... 67
Fatty Acid Analysis ...................... ......................... 69
Results .......................................... .............................................. 71
Cultural Characteristics................................................................................. 71
Virulence and Toxin Analysis........................................................................ 72
Fatty Acid A analysis .............................. ........................................................... 84
Discussion ............................. ...... .................... ........................ 92
Variability of Cultural and Pathogenic Traits ................................................... 92
Fatty Acid Analysis .......................................................................................... 95
4 SUMMARY AND CONCLUSIONS...... .... .......................................................... 97
A SEQUENCE ALIGNMENTS OF ELONGATION FACTOR-la ...................... 101
B SEQUENCE ALIGNMENTS OF P-TUBULIN GENE..................................... 104
C SEQUENCE ALIGNMENTS OF HISTONE 3 GENE .................................... 108
D SEQUENCE ALIGNMENTS OF rDNA 5.8S GENE AND INTERNAL
TRANSCRIBED SPACER REGIONS............................................................ 111
LIST OF REFERENCES .................... ................. ........................... 114
BIOGRAPHICAL SKETCH......................................................................... 125
LIST OF TABLES
2-1 Designations, origins, and morphological characteristics of conidia of the
isolates of Cercospora species from waterhyacinth used in this study .............. 26
2-2 Conidial size and morphology of Cercospora piaropi and Cercospora
rodmanii recorded in the literature ........................................ .......................... 27
2-3 Sequencing results for the data sets aligned with CLUSTAL 1.7W and
analyzed using PAUP version 4.0b I.......................................... ...................... 30
2-4 Maximum likelihood comparisons of tree topologies obtained for
elongation factor-let, P-tubulin, and histone 3 sequences...................................47
3-1 Designations and geographic origin of the isolates of Cercospora piaropilC.
rodmanii analyzed in this study ............................................... ....................... 65
3-2 Mycelial growth rates of isolates of Cercospora piaropi/C. rodmanii from
w aterhyacinth, in m m day .................................................................................. 74
3-3 Geographic origin, color of pigments produced in axenic culture,
production of plant pathogenic toxins, and virulence of isolates of
Cercospora piaropi/C. rodmanii ............................................. ....................... 76
3-4 Production of cercosporin by isolates of Cercospora piaropilC. rodmanii
from waterhyacinth, in gtg per g wet mycelium.................................................... 81
3-5 Correlation matrix among some physiological traits of 55 isolates of
Cercospora piaropi/C. rodmanii from several geographical locations .............. 86
3-6 Fatty acid composition of Cercospora spp. isolates from 4-day-old mycelia........ 88
3-7 Fatty acid composition of Cercospora spp. isolates from 5-day-old mycelia........ 89
3-8 Fatty acid composition of Cercospora spp. isolates from 6-day-old mycelia........ 90
3-9 Canonical discriminant analysis of isolates of Cercospora spp. based on
fatty acid profiles with three different harvesting times of mycelia.................... 91
LIST OF FIGURES
2-1 Maps of the elongation factor-la, 0-tubulin, and histone-3 genes; and of
ribosomal DNA region (rDNA)............................................... ........................ 22
2-2 Conidia of Cercospora from waterhyacinth with truncate (A) and obconic
(B) bases (x400)............................................... ....................... ...................... 28
2-3 Tree length distribution for elongation factor-la, 1-tubulin, and histone 3
datasets based on 10,000 random trees................ ........................... 31
2-4 Results of the partition-homogeneity test implemented in PAUP*4.0b 1............. 33
2-5 Most-parsimonious tree inferred from elongation factor-I a gene..................... 34
2-6 Maximum likelihood tree inferred from elongation factor-la gene...................... 35
2-7 Neighbor-joining tree on distances derived from sequences of the
elongation factor-I a gene ........................................................ ....................... 36
2-8 Most-parsimonious tree inferred from P-tubulin gene......................................... 38
2-9 Maximum likelihood tree inferred from 0-tubulin gene.................................. .... 39
2-10 Neighbor-joining tree on distances derived from sequences of P-tubulin
gene ....... .................. .... ....................................... ........................... 40
2-11 Sequences of amplified segments of P-tubulin gene for the isolates 2619
and WH83. ............................................................................ 41
2-12 Southern blot analysis of the P-tubulin gene of Cercospora species from
w aterhyacinth ................................................................................................. 43
2-13 One of three equally parsimonious trees inferred from histone 3 gene. ................44
2-14 Maximum likelihood tree inferred from histone 3 gene...................................... 45
2-15 Neighbor-joining tree on distances derived from sequences of the histone 3
gene..................................................... .............................. .......... 46
2-16 Strict consensus of 24 most-parsimonious trees of length 246 based on
parsimony analysis of combined elongation factor-I a, P-tubulin, and
histone 3 datasets ..................... .. .......................... 49
2-17 Maximum likelihood tree inferred from the combined elongation factor-la,
P-tubulin, and histone 3 datasets.............................................. ........................... 50
2-18 Neighbor-joining tree on distances inferred from the combined elongation
factor-la, P-tubulin, and histone 3 datasets ........................ ......................51
3-1 An example of differences in colony characteristics of Cercospora
piaropilC. rodmanii from waterhyacinth after growth for 7 days on PDA............ 73
3-2 Isolates of C. piaropi/C.rodmanii plotted in a descending order by their
virulence in waterhyacinth ..................................................... ......................... 79
3-3 Ultraviolet spectra of crude extracts from the yellow pigment-producer,
isolate WHK (A), and the reddish-purple pigment-producer, isolate BA57
(B), compared to the standards cercosporin (C) and beticolin-1 (D) in ethyl
acetate. ........ ...... .......................................................... 83
3-4 Crude extracts of isolates of Cercospora piaropilC. rodmanii resolved on
thin layer chromatograms under long-ultraviolet light........................................ 85
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
PHENOTYPIC AND MOLECULAR CHARACTERIZATION OF CERCOSPORA
SPECIES PATHOGENIC TO WATERHYACINTH
Dauri Jose Tessmann
Chairman: R. Charudattan
Major Department: Plant Pathology
Phylogenetic relationships among isolates of Cercospora species pathogenic to
waterhyacinth, collected from several geographic regions of the weed, were examined by
using partial DNA sequences from three protein-coding genes, elongation factor-la, 3-
tubulin, and histone 3. Each gene included at least one intron area. In cladograms from
individual, as well as from combined datasets for 14 isolates, with the maximum
parsimony, maximum likelihood, and neighbor-joining methods, two statistically well-
supported clades were found: a major clade-grouping of isolates from Brazil, Venezuela,
Mexico, Florida (United States), South Africa, and Zambia; and a minor clade-grouping
of isolates from Texas (United States). In the grouping pattern of the phylogenetic trees,
most of the isolates have spread together with the plant host, from its center of origin in
South America. The isolates grouped together by DNA analysis had distinctive
differences in colony color, pigmentation color, and intensity, growth rate, cercosporin
production, and virulence. In addition, the rDNA region containing ITS1, ITS2, and the
5.8S gene were invariant even when compared with C. beticola as the outgroup. In the
discriminant analysis, the fatty acids methyl ester (FAME) profiles had reduced resolution
for differentiating populations and species among isolates of C. piaropi/C. rodmanii and
the level of resolution was influenced by the age of the colonies. The grouping defined
by discriminant analysis of FAME profiles had no relation to the grouping defined by
DNA analysis. Shape and dimensions of conidia were unreliable criteria for taxonomic
differentiation of isolates that composed the two clade-groupings. In addition, the isolate
that typified C. rodmanii did not show differences in DNA sequence in relation to the
other isolates grouped in the major clade, including some that had conidial size and
morphology typical of C. piaropi. Therefore, the separation of these species, besides of
not having strong phenotypic support, did not also have support from the phylogenetic
analysis. Consequently, the description of the species C. piaropi is emended herein to
include C. rodmanii as a synonym.
The use of plant pathogens as agents of biological control has been an effective
alternative to control several important weeds (Adams, 1988; Charudattan, 1988;
Charudattan, 1991; Charudattan and Walker, 1982; TeBeest, 1996; Templeton and
TeBeest, 1979). Many research efforts have been applied during the last two decades to
develop and implement biological control of the aquatic weed, waterhyacinth (Eichhornia
crassipes [Mart.] Solms), with plant pathogens (Charudattan, 1990). These efforts
included the search for highly virulent pathogens in the plant's native and adventive
ranges; evaluation of the potential of these pathogens for biological control, including
studies about their host ranges and life cycles; and development of technology to use
them as bioherbicides.
Waterhyacinth, an aquatic plant indigenous to lowland tropical South America
(Penfound and Earle, 1948; Barrett and Forno, 1982), has during the last hundred years
become one of the most noxious weeds in many tropical and subtropical regions of the
world. The methods used to control this weed include the use of chemical herbicides,
physical removal with mechanical harvesters, and biological control with insects, mites,
and plant pathogens (Harley and Forno, 1990; Charudattan, 1990; Murphy and Barrett,
1990; Gallagher and Haller, 1990; Center, 1996a). Among these control measures,
chemical herbicides and mechanical removal have been the best short-term solutions for
those areas that require immediate removal of populations of this weed. The problem is
that these measures need to be used continuously to keep infested areas under control.
Moreover, the use of mechanical removal has been very expensive, and the continued use
of herbicides on aquatic environments has raised public concern about the harmful side
effects of chemicals on wildlife, agriculture, and human health. Furthermore, many
developing countries can not afford the recurrent expenses with chemical herbicides and
they usually are less prepared than developed countries in technology and regulatory
safeguards to apply chemical herbicides in aquatic environments. For all these reasons,
the development and implementation of biological control has been considered
indispensable for a long-term solution to this weed problem (Labrada et al., 1996).
Reproductive Biology, Origin, and Dispersal of Waterhvacinth
Eichhornia crassipes is the only species in the genus Eichhornia that has become
a noxious, aggressive weed although a few other species may have similar tendencies.
Eichhornia is an unnatural or polyphyletic group (Kohn et al., 1996) that comprises eight
species of freshwater aquatics in the monocotyledonous family Pontederiaceae. This
family includes the North American pickerelweed, Pontederia cordata L. All species
belonging to the genus Eichhoria, with the exception of the exclusively African E. natans
(Beauv.) Solms, are native to the New World tropics (Barrett, 1988).
Studies of evolution and dispersal of Eichhornia species around the world have
been based mostly on polymorphism of their floral organs (Barrett, 1977; Barrett, 1988).
Flowers of E. crassipes can be divided into three sexual types that differ in the length and
position of their reproductive organs, the male stamens and the female pistils. These
types, or floral morphs, are distinguished by their long, medium, or short styles, i.e., the
prolongation of the ovary. Hence, E. crassipes is described as tristylous plant (Barrett,
From a global perspective, populations of the three style morphs have been found
only in the lowland tropical South America. Outside the native range of this plant, only
the long- and medium-style morphs have been found, with predominance of the medium-
style morph (Barrett, 1977; Barrett and Forno, 1982). The center of origin of E. crassipes
was suggested to be located in the lowland tropical South America, the Amazon basin,
because the greatest diversity of Eichhornia species has been recorded in this region
(Barrett, 1988) and this is the only region were populations of E. crassipes having three
morphs are found (Barrett and Forno, 1982).
Waterhyacinth was introduced into many countries by man for ornamental
purposes because of its beauty: an attractive plant with clusters of violet and yellow
flowers perched atop floating rosettes of bulbous green leaves. This plant is easily spread
because its rosettes of floating leaves are held together only by delicate horizontal stolons
which can easily break apart. In the newly invaded areas, this plant has been a very
effective colonizer because of its free-floating habit, its capacity for rapid vegetative
propagation, and the absence of its natural enemies (Barrett and Forno, 1982; Barrett,
1988). Because growth of waterhyacinth is directly related to the level of available
nutrients in the water in which the plant is growing (Chadwick et al., 1966; Wahlquist,
1972), the explosive growth rate of this weed has been observed mostly in water bodies in
process of eutrophication. Such waters are common particularly in habitats that are
disturbed by human activities (Labrada et al., 1996).
In the 1970s, some misconceptions about the reproductive biology of
waterhyacinth were clarified. Because early investigators did not observe seeds and
seedlings in nature, and based on the generalization that plants that grow exclusively by
vegetative propagation over long periods often lose their ability to reproduce sexually,
many investigators assumed that clones of waterhyacinth were sexually sterile and could
not regenerate from seed. However, later evidence showed the contrary. Barrett (1980a;
1980b) observed that waterhyacinth reproduces sexually both in the native and adventive
ranges of its distribution. The extent of sexual reproduction and its contribution to the
spread of the weed, however, varies greatly in different regions. A number of factors
such as climate, pollinating agents, and factors that affect seed germination have been
considered responsible to limit the efficiency of sexual reproduction. Albeit seed
production may occur in many regions, not all of these regions are favorable for seed
germination and production of seedlings. This may happen because the seeds produced
by waterhyacinth are usually deposited at the bottom of ponds and rivers where they can
remain viable for several years. For germination and seedling development, cyclic
changes in the water level of rivers and ponds must occur, causing periodic drought at
their margins. This is a typical phenomenon observed in the lowlands of tropical South
America. However, in some of the adventive areas, the absence of flooding and drought
cycles may reduce the role of seeds in the spread of this plant.
The problems caused by this plant in areas where it becomes a weed are many.
They include clogged irrigation canals, blocked waterway transport routes, water losses,
reduction in fish population, and destruction of wild life habitats (Pieterse, 1990). The
United States of America was one of the first and foremost countries that faced dramatic
problems caused by this weed and is the country that has accumulated the richest
experience to overcome these problems. Waterhyacinth was first observed in the US
after the Civil War, however the earliest authentic account details its introduction at the
1890 Cotton Centennial Exposition in New Orleans, Louisiana (Penfound and Earle,
1948). Since then and during the following decades, this plant became a very noxious
weed in the Southeast and in California, where it caused a massive negative impact on
human and economic activities in these areas. These problems were reduced only after
the mid-1970s as a result of the introduction and establishment of some the weed's
natural enemies and intensive weed-management programs with herbicides (Center,
1996b; Confrancesco, 1996; Haller, 1996). Other countries that have had long struggles
against this weed include South Africa, Australia, and India.
In recent decades, the problems caused by this weed have increased in magnitude
in many African, Asian, and Latin American countries, mainly in those areas where
human activities have increased (Labrada et al., 1996). In Africa, the rapid spread of
waterhyacinth and the increased problems observed during the last decade in Uganda,
Malawi, Ivory Coast, Benin, and Nigeria have called for urgent international action
(Charudattan et al., 1996). Even though in Brazil waterhyacinth is native to the Amazon
basin, it has been a weed problem but only in water reservoirs located near highly
populated areas (Dr. R. Pitelli, UNESP, Jaboticabal, Brazil, personal communication,
Biological Control of Waterhvacinth
Biological control strategy for waterhyacinth has been based mostly on arthropods
and plant pathogenic fungi. The most studied and used arthropods for biological control
are the weevils Neochetina bruchi Hustache and N. eichhorniae Warner (Coleoptera:
Curculionidae), the moth Sameodes albiguttalis Warren (Lepidoptera:Pyralidae), and the
mite Orthogalumma terebrantis Wallwork (Acarina:Galumidae). Of these organisms, N.
eichhorniae has been the most important control agent which, along with N. bruchi, has
been successfully introduced from South America to several countries, including USA,
Mexico, South Africa, India, and Australia (Center, 1996a). These insects cause
reduction in the growth of waterhyacinth populations as a consequence of the damage
from feeding on the laminae by adult weevils and the tunneling by larvae at the base of
petioles and into the crown. The potential to increase the damage caused by the
arthropods by the interaction with some fungi and bacteria has been recorded
(Charudattan, 1986; Charudattan et al., 1978). However, no further study with other
insect-plant pathogen associations has been reported. A better understanding of these
interactions can be extremely valuable to increase the efficacy of the biocontrol agents.
Plant pathogens were suggested as possible biocontrol organisms for
waterhyacinth as early as the 1930s. As pointed out by Martyn (1977), the first published
paper concerned with plant pathogens as controls for waterhyacinth was written by
Agharkar and Banerjee (1932) in India. They reported studies with a Fusarium species
pathogenic to waterhyacinth. However, it was not until the 1960s that a serious
consideration was given to pathogens as biological control agents for waterhyacinth. It
was then that scientists at the Commonwealth Institute for Biological Control, Bangalore,
India, discovered many new pathogens and evaluated the biological control potential of
some of them (Nag Raj, 1965; Nag Raj and Ponnappa, 1967; 1970). In 1970, another
program was initiated in Florida to control waterhyacinth with pathogens that led to a
serious effort to collect, identify, and screen pathogens in the United States and in other
Studies on the biological control ofwaterhyacinth with plant pathogens have been
based only on fungal plant pathogens. However, some bacteria belonging to
Xanthomonas and Erwinia have been reported to be associated with damage caused by
the Neochetina weevils. These bacteria have been associated with a chlorotic halo that
surrounds the weevil's feeding spots (Charudattan, 1996). The most comprehensive list
of fungi reported from waterhyacinth in different parts of the world is presented by R.
Charudattan in Gopal (1987) with more than one hundred different fungi associated with
this plant. However, for most of these associations, etiological studies are incomplete or
non-existing, and only about 10 of them have been considered important for biological
control (Charudattan, 1990; Charudattan et al., 1996).
The first disease recorded on waterhyacinth was a leaf spot caused by the fungus
Cercospora piaropi Tharp. This fungus, described and named by Tharp (1917) based on
a specimen collected in Palestine, Texas, was subsequently recorded in India
(Thirumalachar and Govindu, 1954; Vasudeva, 1963; Nag Rag, 1965), Florida (Freeman
and Charudattan, 1974), South Africa (Morris, 1990), Australia (Galbraith, 1983), Brazil
(Barreto and Evans, 1995), and Mexico (Martinez Jimenez and Charudattan, 1998). The
symptoms of this disease have been described as discrete dark brown leaf spots present
on leaves and petioles. These spots are discrete on young leaves but lead to chlorosis and
necrosis of part or the whole of older leaves and petioles. Even though this fungus did
not appear to cause appreciable damage to the waterhyacinth plants when it was first
noted, severe epidemics of the disease caused by this fungus were reported by Martyn
(1985) in Texas, USA, and Morris (1990) in South Africa.
The second disease reported in waterhyacinth was a rust caused by the fungus
Uredo eichhorniae Fragoso and Ciferri, from the Dominican Republic (Ciferri and
Fragoso, 1927). In addition, a smut, caused by Doassansia eichhorniae Ciferri, was
reported from that country almost at the same time (Ciferri, 1928). Despite the
worldwide distribution of waterhyacinth, there have been no new sightings of D.
eichhorniae from anywhere including the Dominican Republic. Presently, U. eichhorniae
has been recorded only in Argentina (Charudattan and Conway, 1975) and Southern
Brazil (Dr. R. Charudattan, unpublished field records). Uredo eichhorniae has potential
as a biocontrol agent, but its introduction into the United States has not been allowed
because the complete life cycle and consequently possible alternate host(s) are still
unknown (Dr. R. Charudattan, personal communication).
Rhizoctonia solani Kiuhn was recorded on waterhyacinth for the first time in India
(Padwick, 1946), where it caused extensive blotches and streaks, and often killed
individual plants. In the American continent, R. solani was recorded, first from the
anchoring waterhyacinth (E. azurea [Swartz] Kunth) in the Panama Canal Zone (Freeman
and Zettler, 1971) and shown to be pathogenic to E. crassipes (Joyner and Freeman,
1973). It was later found on E. crassipes, in Louisiana, USA (Freeman et al., 1982). The
fungus isolated in Panama was named later as Aquathanathephorus pendulus Tu and
Kimbrough gen. and sp. nov. based on differences on the morphology of the basida
observed through induction of the perfect stage (Tu and Kimbrough, 1978).
The first report of a zonate leaf-spot disease of waterhyacinth was by Padwick,
(1946) in India. He attributed the cause to a new fungus, naming it Cephalosporium
eichhorniae Padwick. However, this species was later considered a synonym of C.
zonatum (Rintz, 1973), and C. zonatum was reclassified to Acremonium zonatum
(Sawada) Gams (Gams, 1971). This fungus has been also recorded on waterhyacinth in
El Salvador, Panama, Florida (Freeman et al., 1973), Louisiana, USA (Rintz, 1973), and
Mexico (Martinez Jimenez and Charudattan, 1998). Rintz (1973) observed that this
fungus attacked several host plants under artificial conditions and did not cause severe
damage on waterhyacinth. Martyn (1977) observed that A. zonatum was able to control
small, young plants rather than large mature plants.
Marasmiellus inoderma (Berk.) Sing. was reported by Nag Raj (1965) as the
causal agent of thread blight on leaves and petioles of waterhyacinth in India. According
to this report, this fungus could spread quickly through plant populations of plants
because it had an abundant aerial mycelial growth. No further studies have been reported
about this pathogen on waterhyacinth.
Myrothecium roridum Tode ex Fries was also described a cause of a disease on
waterhyacinth in India (Ponnappa, 1970). Although this fungus inflicted extensive
damage to waterhycinth by necrotic lesions on leaves, it was not considered for use as a
biological control agent because of its broad host range (Ponnappa, 1970).
Several species of Bipolaris and Helminthosporium (two related genera) have
been reported to attack waterhyacinth in various countries. One species in particular, B.
oryzae (Breda de Haan) Shoemaker (=B. stenospila [Drechs.] Schoemaker) was reported
as the causal agent of severe blighting on shoots of waterhyacinth in the Dominican
Republic (Charudattan et al., 1975). A Dominican isolate of this fungus from
waterhyacinth was also pathogenic to sugarcane, rice, and bermudagrass, and therefore it
is not likely to be safe for use as a bioherbicide (Charudattan, 1996).
Alternaria eichhorniae Naj Raj and Ponnappa has been recorded in India (Nag
Raj and Ponnappa, 1970), Bangladesh (Badur-ud-Din, 1978), Indonesia
(Mangoendihardjo et al., 1977), Thailand (Rakvidhyasastra et al., 1978), and Egypt
(Shabana et al., 1995a). This fungus causes a severe leaf blight on waterhyacinth, has a
narrow host range (Nag Raj and Ponnappa, 1970; Shabana et al., 1995a), and is being
studied for development as a bioherbicide to control waterhyacinth (Shabana, 1996;1997;
Shabana et al.; 1995b; 1997).
A second Cercospora species, C. rodmanii Conway, was described on
waterhyacinth from the Rodman Reservoir, Florida, based on some phenotypic
differences in relation to C. piaropi (Conway, 1976a). Indeed, to circumscribe C.
rodmanii, Conway (1976a) emended the diagnosis of C. piaropi given by Tharp (1917).
Based on Conway's description, C. rodmanii had longer conidia than C. piaropi; conidia
of C. rodmanii had truncate bases compared to the obconic bases observed in C. piaropi;
Cercospora rodmanii was considered more virulent than C. piaropi; the former caused a
general blighting symptom on the foliage compared with more discrete leaf spots caused
by C. piaropi. Moreover, C. rodmanii had the presence of a well-developed stroma at the
base of conidiophores; and the presence of an associated pycnidial state, Asteromella sp.
However, in practice it has been difficult to distinguish between these two species based
on these diagnostic characters (Martyn, 1985; Morris, 1990; Martinez Jimenez and
The species of Cercospora as Agents for Biological Control
The isolate of Cercospora from Rodman Reservoir, named C. rodmanii, has been
the most studied pathogen for biological control of waterhyacinth. Even though, severe
epidemics of C. piaropi have been recorded more recently in Texas, USA (Martyn, 1985)
and South Africa (Morris, 1990), early reports considered this pathogen to be incapable of
causing serious damage to waterhyacinth (Nag Rag, 1965; Freeman and Charudattan,
Cercospora spp. are the most widespread pathogens of waterhyacinth.
Cercospora piaropi and C. rodmanii have been reported collectively, from every
continent in which waterhyacinth has spread (Charudattan, 1996). The host specificity of
Cercospora species on waterhyacinth has been the most important factor to consider
these pathogens as biocontrol candidates (Freeman and Charudattan, 1974; Conway and
Freeman, 1977). Cercospora rodmanii was shown to has a limited host range in a study
that involved 58 species in 22 plant families, and plants that developed symptoms showed
damage only on senescent tissues. The blighted areas showed the presence of both C.
rodmanii and a long-beaked Alteraria sp. (Conway and Freeman, 1977).
In studies of the C. rodmanii-waterhyacinth pathosystem, the infection process
originated from either fungal mycelium or conidia. In both instances the hyphae grew
into the stomata, ramified in the substomatal cavity, and invaded the surrounding tissue.
A stroma developed in the stomatal cavity and a fascicle of conidiophores arose from it
and emerged through the stomata. Primary and secondary conidia were produced on the
conidiophores (Freeman and Charudattan, 1984).
In nature, conidia produced on diseased tissue are spread by wind and serve to
disseminate C. rodmanii from disease foci. The speed and intensity of the epidemic are
directly related to the number of conidia produced, which is related to the amount of
diseased and dead tissue that is available for sporulation. Under natural conditions in
Florida, the sporulation reaches a peak during the fall and early winter (Freeman and
Disease stress within a population of waterhyacinth is manifested initially as an
overall chlorotic appearance of the plants. Numerous severely spotted or dead leaves
soon become evident on the plants. As the disease progresses, the entire population of
waterhyacinth turns brownish in appearance. At this stage, the waterhyacinth population
begins to decline (Freeman and Charudattan, 1984).
Biological control of waterhyacinth with C. rodmanii has relied mostly on the
bioherbicidal strategy, but this fungus has been released also as a classical biocontrol
agent in South Africa, Egypt, and Honduras (R. Charudattan, personal communication).
The differences between the classical and the bioherbicidal strategy are that, in the first, a
pathogen is introduced from the geographic origin of the weed into the weed's adventive
range where the control is desired. In the bioherbicidal strategy, native or exotic
pathogens are cultured in vitro on a large scale and applied in fairly high concentration to
the weed (Templeton, 1982; Charudattan, 1985). In the United States, a bioherbicidal
strategy strategy was proposed to control this weed, based on C. rodmanii, which can
provide significant levels of control when used under conditions that limit host growth
rate or in combination with other biotic and abiotic agents (Conway, 1976b; Freeman and
Charudattan 1984; Charudattan, 1986; Charudattan et al., 1985). However, even though
C. rodmanii has shown good potential to control waterhyacinth, the market for aquatic
weed control has been dominated by chemical herbicides that provide fast, economical,
and predictable control. Therefore, the incentive to develop and register C. rodmanii was
discontinued (Charudattan, 1991).
The differentiation of the species C. piaropi and C. rodmanii, based on the
characters used for their description, has been difficult in practice and has caused
scientific and regulatory (plant quarantine) problems and consequently has prevented to
some extent the development and implementation of biological control of waterhyacinth.
In addition, the species C. piaropi has been recorded in several regions of the world,
including the center of origin of waterhyacinth and its adventive areas, whereas C.
rodmanii has been recorded only in Florida. In view of the fact that these species were
circumscribed based on their morphology and virulence, the possibility arises that, like
many other fungal species, the DNA-based techniques may help to reconcile the two
species. Hence, the objective of this research was to 1) delineate phylogenetically
informative characters based on a collection of isolates of C. piaropi and C. rodmanii
from several geographic origins; 2) characterize the partial DNA sequences of three
evolutionary conserved genes; and 3) assess the validity of the two Cercospora spp.
reported on waterhyacinth. It was hypothesized that this approach will allow the
redefinition of the species based on the phylogenetic species concept and to provide some
insights about the biogeography of these pathogens of waterhyacinth. Another objective
was to determine the extent of variability in some phenotypic traits, including cultural
features, phytotoxins, and virulence among a population of isolates of C. piaropi and C.
rodmanii cultured under standard conditions. Finally, an attempt was made to determine
the usefulness of fatty acid methyl ester profiles to discriminate isolates, populations, or
species of Cercospora from waterhyacinth.
A MOLECULAR CHARACTERIZATION OF CERCOSPORA SPECIES FROM
Two Cercospora species have been studied and used as biocontrol agents for the
control of the aquatic weed waterhyacinth (Eichhornia crassipes [Mart.] Solms): C.
piaropi described by Tharp (1917) from a Texas specimen and C. rodmanii described by
Conway (1976a) from a Florida specimen. The species C. rodmanii was described by
Conway (1976a) who examined decaying waterhyacinth plants collected at the Rodman
Reservoir, Florida, where it caused severe epidemics of a Cercospora leaf-spot disease.
He observed that the Rodman isolate had some differences in conidial morphology and in
disease symptomatology in relation to what was described for C. piaropi by Tharp (1917).
Indeed, to circumscribe these two species of Cercospora on waterhyacinth, Conway
(1976a) emended the original description of C. piaropi, given by Tharp (1917), and
differentiated these species based on the observation that conidia of C. rodmanii were
longer than those of C. piaropi; conidia of C. rodmanii had a truncate base while those of
C. piaropi had an obconic base; C. rodmanii was more virulent than C. piaropi and
caused general blighting symptom on the foliage compared to the more discrete leaf spots
caused by C. piaropi; C. rodmanii had the presence of a well-developed stroma at the
base of conidiophores; and the C. rodmanii specimens examined had the presence of an
associated pycnidial state, Asteromella sp. (Conway, 1976a). Based on the Rodman
isolate, a bioherbicidal strategy was proposed to control waterhyacinth with the
expectation that this pathogen can provide significant levels of control when used under
conditions that limit host growth rate or in combination with other biotic and abiotic
agents (Conway, 1976b; Freeman and Charudattan, 1984; Charudattan, 1986;
Charudattan et al., 1985).
Since its original description from Texas, C. piaropi has been recorded also in
India (Thirumalachar and Govindu, 1954; Vasudeva, 1963; Nag Rag; 1965), Florida,
USA (Freeman and Charudattan, 1974), Australia (Galbraith, 1983), South Africa
(Morris, 1990), Brazil (Barreto and Evans, 1995), and Mexico (Martinez Jimenez and
Charudattan, 1998), whereas the species C. rodmanii has been recorded only in Florida.
In fact, it is difficult to distinguish the two species, as observed in the reports of Martyn
(1985) and Morris (1990). The problem is that the variability in size and shape of conidia
sometimes encompasses the description of both species, and the differences in disease
symptoms between these two species are not distinguishable in practice.
The species definition in Cercospora Fres. has been based upon morphological
criteria and host affiliation (Chupp, 1953; Ellis, 1971). However, in addition to the
problems that such criteria cause for species identification, it is still unknown whether
they are appropriate to circumscribe species defined by phylogenetic relationships.
Moreover, even though those Cercospora species with a known teleomorphic stage have
been linked to Mycosphaerella, in the phylum Ascomycota (Sivanesan, 1984), it is still
unknown whether the form-genus Cercospora is in fact a monophyletic group.
Molecular markers have been a valuable source of new diagnostic characters for
studies of fungal taxonomy and evolution (Bruns et al., 1991; Kohn, 1992). In fact, DNA
sequences of conserved genes can be abundant sources of phylogenetically informative
characters, which are appropriate for phylogenetic analysis of fungal taxa and to define
species according to the phylogenetic species concept (O'Donnell and Cigelnik, 1997;
O'Donnell et al., 1998a).
DNA sequence analysis of internal transcribed spacers regions (ITS) of the
ribosomal DNA have been used to distinguish species and groups of species in a number
of taxonomic studies of fungi (Chen et al., 1996; Crawford et al., 1996; Harrington, 1998;
Roy et al., 1998). In addition, in recent studies, protein-coding genes were appropriate for
phylogenetic analysis of fungi at the species level. Such genes may have little or no
variation in their amino acid sequences but their third codon positions and intron regions
appear to have a high rate of nucleotide substitutions. Protein-coding genes have
advantages as molecular markers over ribosomal genes in that they offer a large number
of unlinked sources of phylogenetic information (Geiser et al., 1998a). Indeed, these loci
may show different levels of resolution for the different groups of fungi, and as seen in
Fusarium, the concurrent analyses of more than one independent locus increases the
strength of the conclusions by offering different lines of evidence for a phylogenetic
hypothesis (O'Donnell and Cigelnik, 1997; O'Donnell et al., 1998a; O'Donnell et al.,
The objective of this study was to determine the phylogenetic relationships among
isolates of Cercospora species pathogenic to waterhyacinth, from several geographical
origins, based on partial DNA sequences from three protein-coding genes: elongation
fator-1 a, P-tubulin, and histone 3; and the ribosomal DNA regions containing ITS I,
ITS2, and the 5.8S gene. The hypotheses tested were whether C. rodmanii and C. piaropi
are two distinct species according to the phylogenetic concept of species, and whether
genetic divergence, if found, is related to the geographical origin of the isolates.
Materials and Methods
The isolates used in this study were recovered from symptomatic leaves of
waterhyacinth, and monocultures were obtained from hyphal tips. Their designations and
geographic origins are presented in the Table 2-1. These isolates have been preserved in
the fungal collection of the Biological Control of Weeds Laboratory of the Plant
Pathology Department, University of Florida.
For isolate identification, a test tube containing 1 g of autoclaved, wet, seeds of
ryegrass (Lolium multiflorum Lam.) was inoculated with a 5 cm3 mycelial plug removed
from a 7-day-old culture on potato dextrose agar (PDA) plate. The tube was kept at room
temperature with 12 h light. Production of conidia was observed 7 to 14 days after
inoculation. Since the isolates sporulated irregularly, this procedure needed to be
repeated to identify the isolates. A natural substratum, such as ryegrass seed, was
preferred because most of the isolates did not sporulate in axenic cultures even when
grown in several media. Observations and conidial measurements were made with a light
microscope. In addition, the isolates were grown on V-8 agar in petri plates for
observation of cultural attributes, including the production of the typical reddish-purple
pigment, and for determination of the occurrence of the Asteromella conidial state by
flooding the plates with distilled water (Conway, 1976a).
Roux bottles containing 90 ml of potato dextrose broth (PDB) (Difco, Detroit,
MI) were inoculated with six 5-mm3 mycelial plugs excised from the margin of an 8-day-
old culture on PDA plate (Difco, Detroit, MI). After 5-6 days of growth, the contents of
the flasks were filtered through sterile cheesecloth, squeezed dry, and rinsed three times
with sterile deionized water. Mycelium was placed into 13-ml plastic tubes, stored for 24
h in a -800C freezer, and lyophilized for 24-48 h. The dry mycelium was then mixed with
liquid nitrogen, ground to a fine powder, and combined with DNA extraction buffer that
consisted of a 1:1:0.4 volume of Nuclei Lysis Buffer (0.3 M sorbitol, 0.1 M Tris, and 20
mM EDTA, at pH 7.5), DNA isolation buffer (0.2 M Tris at pH 7.5, 50 mM EDTA, and
0.2 mM cetyltrimethylammonium bromide), and 0.5% Sarkosyl (Koenig et al., 1997;
Ten milliliters of the extraction buffer were combined with approximately Ig of
ground mycelium in 15-ml tubes, and the tubes were placed in a 65C water bath for 60
minutes. The contents of the tubes were then mixed by inversion, and I ml of the
solution was transferred to a sterile, 1.5-ml microcentrifuge tube. Five hundred
microliters of chloroform:octanol (24:1) solution was added to each tube. The solution
was mixed thoroughly by inversion. The solution was then centrifuged for 10 minutes at
12,000 g in a microcentrifuge, at room temperature. The supernatant was transferred to
sterile 1.5-ml tubes and treated with 5 il of a suspension containing 20 mg RNAse
(Sigma Chemical Company, St.Louis, MO) per ml for 30 minutes at 370C. Following the
RNAse treatment, 5 pl of a suspension of 20 mg Proteinase K (Sigma Chemical
Company, St. Louis, MO) per ml were added and allowed to remain in solution for 20
minutes at 37'C. One volume of ice-cold isopropanol was then added, and the tubes were
shaken until the DNA was visible as a white precipitate. After 60 minutes in a -20C
freezer, the tubes were centrifuged at 10,000 g for 5 minutes and the DNA pellet was
washed with 100 pl of 70% ethanol three times and allowed to dry. Then, the pellet was
suspended in 100 .l of TE buffer (10 mM Tris pH = 7.6; 1 mM EDTA) and samples were
placed at 40C until the DNA was dissolved. The method described here is a modified
version of the DNA extraction method used by Koenig et al. (1997) and Rosskopf (1997).
Samples that were difficult to resuspend were subjected to a LiCI treatment.
Three hundred microliters of ice-cold 4M LiCI solution were added to each tube, and the
tubes were placed on ice for 30 minutes before centrifugation at 12,000 g for 10 minutes
at 4C. The supernatant was transferred to a sterile 1.5-ml tube containing 600 pt of
isopropanol. This solution was mixed, and the tubes were kept at room temperature for
30 minutes. After centrifugation at 12,000 g for 10 minutes at 4C, the supernatant was
discarded and 100 pl of ice-cold 70% ethanol was added to the tubes. After
centrifugation at 12,000 g for 5 minutes, the ethanol was discarded and the DNA was air-
dried in a hood. TE buffer pH = 7.6 (100 ptl) was added to the tubes and they were placed
in a water bath at 650C until the pellet was dissolved. Concentration and purity of DNA
was estimated spectrophotometrically.
DNA Amplification and Sequencing
Approximately 200 ng of template DNA per 100 l reaction mixture was used for
DNA amplification by the polymerase chain reaction (PCR). Primers for amplification of
P-tubulin and histone 3 gene sequences were based on those used by Glass and
Donaldson (1995). The internal transcribed spacer regions (ITS) of the nuclear ribosomal
repeat were analyzed with primers ITS4 and ITS5 (White et al., 1990). Gene maps with
the primer position and sequences are shown in Figure 2-1. For amplification of
elongation factor-la gene sequences, a set of primers was designed with the software
PC/Gene (IntelliGenetics Inc., Mountain View, Ca) based on Aureobasidium pullulans
EF-la gene sequence (GenBank accession no. APU19723). All primers for protein-
coding genes span at least one intron area. The primers were synthesized at either the
University of Florida Interdisciplinary Center for Biotechnology Research
Oligonucleotide Synthesis Laboratory (Gainesville, FL) or by Gibco BRL, Gaithersburg,
MD). PCR was performed using final concentrations of the components in the reaction
mixture as follows: 20 mM Tris-HCI (pH 8.4); 50 mM KC1; 3 mM MgC12; 1 JM of each
primer; 200 pM each of dATP, dCTP, dGTP, and dTTP; and 2.5 U of Taq polymerase
(Gibco-BRL, Gaithersburg, MD) per 100 pl of reaction mixture. A GeneAmp 9600
(Perkin-Elmer Applied Biosystems, Foster City, CA) was used for the amplification. For
amplification of the elongation factor-la, j-tubulin and histone 3 gene sequences, the
cycling conditions used an initial denaturation step of 1 min at 940C, with 35 cycles of
940C for 45 seconds, 620C for 30 seconds, and 720C for 45 seconds. The last cycle
included a 10-minute incubation at 720C and then storage at 40C. For amplification of
the rDNA region, the annealing temperature was dropped to 550C. PCR fragments were
cleaned using PCR Preps Kit (Promega INC., Madison, WI) according to the
manufacturer's instructions. Sequences were obtained with a Perkin-Elmer Applied
Elongation factor- la
c~ III, ___~b
Nuclear rDNA genes
I 18S 5.8 28S
ITSI ITS2 4
Bt2a 5' GGTAACCAAATCGGTGCTGCITTC 3'
Bt2b 5' ACCCTCCGTGTAGTGACCCTTGGC 3'
EFIa 5' ATCAACCTCGTCGTTATCGGCCACG 3'
EFIb 5' TCAGACTTCACGTTGTCGAGGACCC 3'
H3-la 5' ACTAAGCAGACCGCCCGCAGG 3'
H3-lb 5' GCGGGCGAGCTGGATGTCCTT 3'
ITS4 5'TCCTCCGCTTATTGATAT 3'
ITS5 5' GGAAGTAAAAGTCGTAACAAGG 3'
Figure 2-1. Maps of the elongation factor-lo, 0-tubulin, and histone-3 genes; and of
ribosomal DNA region (rDNA). Position of primers used for PCR amplification and
sequencing are indicated by arrows. Shaded boxes denote protein-coding sequences
(exons), and cross-hatched boxes denote introns. The P-tubulin and histone 3 maps are
from Neurospora crassa (Glass and Donaldson, 1995), and elongation factor-la map is
from Aureobasidium pullulans (Thomewell et al., 1995). The rDNA region map was
based on White et al. (1990).
Biosystems model 373A or 377 automated DNA sequencer (Perkin-Elmer, Foster City,
California, CA) in the DNA Sequencing Core Laboratory of the University of Florida.
Southern Blot Analysis of the 3-tubulin Gene
The possibility that isolates may have multiple copies of thep-tubulin gene was
tested through Southern blotting analysis. Approximately 5 tg of DNA was digested
with at least 10 units of Pst I, EcoR I, and Hind III restriction enzymes (Gibco-BRL,
Gaithersburg, MD) and incubated for at least 4 h. These restriction enzymes were chosen
based on analysis of the restriction map of the 3-tubulin gene of Colletotrichum
graminicola (Ces.) G.W. Wils. (GenBank accession no. M34492) using the University of
Wisconsin Genetics Computer Group program. Restriction fragments were separated by
electrophoresis in 0.8% agarose in Tris-borate-EDTA (TBE) buffer at pH 7.0. Gels were
run at 50 V for approximately 18 h Lambda phage DNA digested with Hind II was
used for size markers. DNA was detected by staining gels in a solution of ethidium
bromide (0.5 ptg ml') followed by UV trans-illumination. DNA was transferred to Zeta-
probe GT membrane (Bio-Rad Laboratories, Hercules, CA) using the capillary transfer
method. The DNA transfer proceeded for at least 16 h and DNA was immobilized by UV
cross-linking and hybridized to a 32P-labeled DNA probe. The DNA probe was a 380 bp-
PCR product of the p-tubulin gene from the isolate 28-1, amplified with the primers Bt2a
and Bt2b and labeled using the RadPrime DNA Labeling System (Gibco-BRL,
Gaithersburg, MD), according to procedures provided by the manufacturer.
Following alignment with CLUSTAL W1.7 (Thompson et al., 1994), the measure
of the phylogenetic signal for each dataset was estimated by the skewness of the tree-
length distributions (Hillis and Huelsenbeck, 1992; Hillis et al., 1993), implemented in
PAUP version 4.0b1 (Swofford, 1997). Phylogenetic relationships among taxa for
individual and combined datasets were inferred based on the maximum parsimony,
maximum likelihood, and neighbor-joining methods, implemented by PAUP (Swofford,
1997) and MacClade Version 3.01 (Maddison and Maddison, 1992), and performed with
all DNA characters unweighted. All of these methods of phylogenetic inference have
been shown similar accuracy (Hillis et al., 1994). In the maximum parsimony analyses,
trees were obtained using the stepwise addition option in heuristic search with random-
addition sequences, and alignment gaps were treated as missing characters. For neighbor-
joining analysis, a distance matrix was generated using the Jukes-Cantor procedure. For
maximum likelihood analysis, a 2:1 transition/transversion rate was assumed. Clade
stability was estimated with 1000 bootstrap replications (Hillis and Bull, 1993)
implemented in PAUP (Swofford, 1997) and by decay indices (Bremer, 1988) calculated
with TreeRot (Sorenson, 1996). Other measures, including tree length and consistency
and retention indices, were calculated with PAUP.
The partition-homogeneity test (PHT) option in PAUP (Swofford, 1997) was
used to determine whether the elongation factor-la, P-tubulin, and histone 3 datasets
were in conflict, with 1000 replicates. For the PHT, only phylogenetically informative
characters were used. Gaps were considered missing data for all analyses. The Kishino-
Hasegawa test was used to compare alternative constrained and unconstrained tree
topologies using PAUP (Swofford, 1997).
This study used the phylogenetic species concept (Nixon and Wheeler, 1990) as
used by O'Donnell and Cigelnik (1997), which recognizes species as the smallest group
of populations or lineages diagnosable by a unique combination of fixed apomorphies.
Identification of Isolates
Some isolates showed characteristics of conidial size and morphology that
encompassed the description of both Cercospora species described as pathogens of
waterhyacinth, C. piaropi and C. rodmanii. The isolates WH9BR, WHK, BA57, 175-102,
and 2619, from Florida; 62-2, 62-4, and 34 from Brazil; WHV from Venezuela; 2867
from Mexico; and 400 from Zambia (Tables 2-1; 2-2) were in this category. The
presence of conidia with both obconic and truncate bases (Figure 2-2) is not a feature
related with the geographic origin of the isolates. However, some isolates showed only
conidia with obconic base and conidial size concordant with the description of C. piaropi,
such as the isolates WH83 and RR29 from Florida; TX15, TX18, and TX20, from Texas;
10, 28-1 and 18-2, from Brazil; and MX3 from Mexico. The fact that all isolates showed
conidia with obconic bases and some isolates showed conidia with both, obconic and
truncate bases confused the species identification. The isolate WH9BR, which was
collected by Conway and deposited as a type culture of C. rodmanii also showed conidia
with obconic and truncate bases. In general, conidia with truncate bases were longer than
conidia with obconic bases and the range of conidial length for most isolates was more
close to the range described for C. piaropi. In addition, none of these isolates produced
the Asteromella conidial state in V-8 agar cultures flooded with distilled water.
Table 2-1. Designations, origins, and morphological characteristics of conidia of the isolates of Cercospora species from
waterhyacinth used in this study.
Isolate Geographic original Conidial size (um)b Morphology of conidial baseb
WH9BR' Florida, USA 55- (148) -232 x 3.0- (3.8) -4.5 Truncate/obconic
RR29 Florida, USA 44- (132)-196 x 3.0- (4.0) -4.5 Obconic
WH83 Florida, USA 33- (115) -183 x 3.0- (3.4) -4.0 Obconic
WHK Florida, USA 35- (132)-243 x 3.5- (4.0) -5.0 Truncate/obconic
BA57 Florida, USA 31-( 91)-148 x 2.5- (3.5) -4.0 Truncate/obconic
175-102 Florida, USA 48- (127) -205 x 3.0- (3.4) -4.5 Truncate/obconic
2619 Florida, USA 35- ( 92) -253 x 2.5- (3.5) -4.0 Truncate/obconic
TX15 Texas, USA 28- ( 76) -142 x 2.5- (3.0) -3.5 Obconic
TX18 Texas, USA 33- (109)-157 x 2.0- (3.0) -3.5 Obconic
TX20 Texas, USA 35-( 84)-167 x 2.5- (2.8) -3.5 Obconic
62-4 Northeast Brazil 39- (117) -211 x 2.5- (3.4) -4.0 Truncate/obconic
62-2 Northeast Brazil 35- (118)-175 x 2.0- (3.2) -4.0 Truncate/obconic
10 Southeast Brazil 35- (147)-219 x 3.0- (4.0) -4.5 Obconic
28-1 Southwest Brazil 28-( 81)-177 x 2.5- (3.5) -4.5 Obconic
18-2 Southwest Brazil 22-( 69) -158 x 2.0- (2.7) -3.5 Obconic
34 South Brazil 24-(115)-175 x 2.5- (3.0) -3.5 Truncate/obconic
WHV Venezuela 55-(141)-203 x 2.0-(3.0)-3.5 Truncate/obconic
2867 Mexico 44- (139)-225 x 2.5- (3.6) -4.5 Truncate/obconic
MX3 Mexico 42- (108)-163 x 3.0- (3.8)- 4.0 Obconic
114 South Africa NDd ND
400 Zambia 26- (119) -187 x 2.5- (3.7) -4.0 Truncate/obconic
a The isolates RR29, BA57, 175-102, TX15, TX18, and TX20 were collected for this study; the isolates 114 and 400 were obtained from Dr. M.
Morris, Stellenbosch, South Africa; and the other isolates were obtained from the culture collection of the Biological Control of Weeds
Laboratory of the University of Florida. Additional details of geographic origin are available from Dr. R. Charudattan.
b Based on conidia produced by cultures grown on autoclaved ryegrass seeds.
C. rodmanii culture type collected by K. Conway.
d ND = not determined due to lack of sporulation.
Table 2-2. Conidial size and morphology of Cercospora piaropi and Cercospora rodmanii recorded in the literature.
Species Geographic Conidial size (tm) Morphology of
origin conidial base
C. piaropi (Tharp, 1917)a Texas, USA 80-140 x 3 Truncate/obconicb
C. piaropi (Chupp, 1953) USA 25-140 x 2-3.5 Truncate
C. piaropi (Thirumalachar and Govindu, 1954) India 66-120 x 2-4.2 Truncate
C. piaropi (Vasudeva, 1963) India 80-140 x 3 Truncate
C. piaropi (Freeman and Charudattan, 1974) Florida, USA 55-121 x 3.3-4.4 Truncate
C. piaropi (Morris, 1990) South Africa 50- (140) -270 x 1.5- (2.5) -3.5 Truncate/obconic
C. piaropi emendedd by Conway ) -25-( 95)-220 x 2.0- (3.5) -5.0 Obconic
C. rodmanii (Conway, 1976) Florida, USA 66- (172)-374 x 3.0- (4.0) -5.0 Truncate
a Original species description.
b The species diagnostic given by Tharp (1917) describe only truncate conidia; however, Conway (1976a) examined the holotype
and observed mostly obconic conidia.
Original species description.
Figure 2-2. Conidia of Cercospora from waterhyacinth with truncate (a) and obconic (B)
The isolate 114 from South Africa did not produce condia in any medium,
substrate, or condition tested. It was identified based on the collector's information.
The aligned DNA sequences of the elongation factor-la, p-tubulin, and histone 3
datasets provided enough phylogentically informative characters to infer relationships
among the isolates of Cercospora sp. from waterhyacinth (Table 2-3; Appendices A, B,
From 431 characters of elongation factor- Ia dataset, 14 characters were
parsimony-informative and three were variable characters that were parsimony-
uninformative. From the 380 characters of P-tubulin dataset, six characters were
parsimony-informative and 64 variable characters were parsimony-uninformative. From
309 characters of the histone 3 dataset, 17 characters were parsimony-informative and
124 variable characters were parsimony-uninformative. However, the aligned DNA
sequences of the 5.8 rDNA gene and is ITS flanking regions for eight of the isolates were
invariant in this region even when compared with the outgroup, C. beticola (Table 2-3;
The tree-length distribution analysis, based on 10,000 replicates, showed strongly
significant skewed tree-length distributions for the elongation factor-la, p-tubulin, and
histone 3 datasets (Figure 2-3), with gl values of -2.817, -2.063, and -3.230 (P=0.01),
respectively. This analysis indicated that these datasets were significantly nonrandom
and potentially informative about phylogeny (Hillis et al., 1993; Hillis and Huelsenbeck,
1992). The partition homogeneity test (Farris et al., 1995; Huelsenbeck et al., 1996) was
Table 2-3. Sequencing results for the data sets aligned with CLUSTAL 1.7W and
analyzed using PAUP version 4.0bl.
Total Variable Parsimony- Parsimony-
Data set characters characters informative uninformative
Elongation factor-la 431 17 14 3
P-Tubulin 380 70 6 64
Histone 3 309 141 17 124
5.8S rDNA and ITS regions 586 6 0 6
5000 Elongation factor- la
E 2000 -
'C c'cl N 0r m '
N 0 CR 0 C '/ ~ ~ N
.n 1. 1
Tree Length (steps)
Figure 2-3. Tree length distribution for elongation factor-la, P-
tubulin, and histone 3 datasets based on 10,000 random trees.
' 1- -
applied to decide whether these three datasets could be combined for phylogenetic
analysis. With the same 14 isolates from each dataset as ingroups and C. beticola as the
outgroup, this test rejected the null hypothesis of homogeneity in the phylogenetic signal
among datasets (P=0.002). Thus, these datasets should not be combined for phylogenetic
analysis (Figure 2-4). According to this test, the datasets can not be combined whether
the actual sum of the tree length of individual datasets are less than the shortest tree
generated when the dataset are combined. In this study, elongation factor-la, P-tubulin,
and histone 3 had tree lengths of 17, 78, and 20 steps, respectively. Thus, the sum of
these tree lengths (= 115 steps) was lower than the tree length of the shortest tree
generated when these dataset were combined (116 steps).
The most-parsimonious tree inferred from a segment of 431 bp of elongation
factor-la dataset (Figure 2-5), for 16 isolates of C. piaropi/C. rodmanii, showed a strong
grouping (100% bootstrap support, decay index = 8) for the isolates 10, 28-1, and 62-2
from Brazil; 114 from South Africa; 2867 and MX3 (Mexico), WH83, WH9BR, WHK,
BA57, 2619, and 175102 from Florida, USA; 400 from Zambia; and WHV from
Venezuela. In addition, this tree showed a strong grouping (100% bootstrap support,
decay index = 5) for the isolates TX20 and TX 18, from Texas, USA. The grouping of the
isolates 62-2, WH83, WH9BR, WHK, and BA57 inside the major group did not resolve
with strong support (64% bootstrap support and decay index = 1). The phylogenetic tree
inferred through maximum likelihood method showed the same topology as the tree
inferred through parsimony (Figure 2-6). For the neighbor-joining tree, even though it
showed a more branched topology than the other two trees, the only two clades with
S200 P = 0.002
1 0 I ,
110 111 112 113 114 115 116 117 118 119 120
Sum of Tree Lengths
Figure 2-4. Results of the partition-homogeneity test implemented in PAUP*4.0b 1.
The arrow indicates the summed length of 115 steps of the single most
parsimonious tree inferred from the actual elongation factor-la, P-tubulin, and
histone 3 datasets. The vertical bars show the random distribution of the sum of tree
lengths obtained from 1000 random repartitions of the combined datasets.
) ( --WH83
8 64 WH9BR
5 -- TX18
Figure 2-5. Most-parsimonious tree inferred from elongation factor- la gene (length = 17,
consistency index = 1.0, retention index = 1.0, rescaled consistency index = 1.0).
Bootstrap replication percentages and decay indices (in parentheses) are indicated above
the nodes. Edge length is indicated below the nodes and branches.
Figure 2-6. Maximum likelihood tree inferred from elongation factor-la gene. Bootstrap
replication frequencies are indicated above the nodes.
Figure 2-7. Neighbor-joining tree on distances derived from sequences of the elongation
factor-la gene. Bootstrap replication frequencies are indicated above the nodes.
strong support (100% bootstrap) grouped the same isolates as parsimony and maximum
likelihood tree (Figure 2-7). Those clades placed internal to the major clade did not have
strong bootstrap support. Actually, by default PAUP did not show bootstrap values lower
The most-parsimonious tree inferred from a segment of 380 bp of the p-tubulin
dataset (Figure 2-8), for 21 isolates of C. piaropi/C. rodmanii, also showed a major group
containing the isolates MX3 and 2867 from Mexico; 62-4, 28-1, 18-2, 34, 10, and 62-2
from Brazil; WHK, 175102, BA57, WH9BR, WH83, and RR29 from Florida, USA; 114
from South Africa; 400 from Zambia, and WHV from Venezuela. Also, this tree showed
a group containing the isolates TX15, TX18, and TX20, from Texas, USA. The
statistical support for these groupings were 77% bootstrap and decay index = 1 for the
major group, and 71% bootstrap and decay index = 2 for the minor group. According to
this tree, the isolate 2619 from Florida was distinct from all other members. This
happened because the nucleotide sequence of the intron area of the p-tubulin gene in this
isolate accumulated more sequence differences than the other isolates (Figure 2-11).
The maximum likelihood tree inferred from p-tubulin dataset had exactly the
same topology of the maximum parsimony tree (Figure 2-9). This tree showed 71 and
72% bootstrap support for the two major clades, which grouped the same isolates such as
the two major groupings in the parsimony tree. Also the neighbor-joining tree inferred
from this dataset showed a more branched topology; however, only two clades showed
more than 50% bootstrap support (Figure 2-10). The major group, with 71% bootstrap
support, contained the same isolates as the major group in parsimony and maximum
2 1 TX20
Figure 2-8. Most-parsimonious tree inferred from (3-tubulin gene (length = 74,
consistency index = 0.973, retention index = 0.818, rescaled consistency index = 0.796).
Bootstrap replication percentages and decay indices (in parentheses) are indicated above
the nodes. Edge length is indicated below the nodes and branches.
Figure 2-9. Maximum likelihood tree inferred from P-tubulin gene. Bootstrap replication
percentages are indicated above the nodes.
Figure 2-10. Neighbor-joining tree on distances derived from sequences of P-tubulin
gene. Bootstrap replication percentages are indicated above the nodes.
1 GCAGACCATCTCTGGCGAGCACGGTCTCGACAGCAACGGTG ....... 42
I 1 l l l l l l l I I I I I l i I I Il l i i l It I l II
2 GCAGACCATCTCCGGCGAACATGGCCTCGACGGCTCCGGCGTGTATGTGC 51
43 ....... ........ ....... .. ............ .........CTACA 47
52 AGCAGATCGCAATGGATAAATGGAGCAGCGACTGACGTCGTG GGTACA 101
48 ATGGCAGCTCCGAGCTTCAGCTCGAGCGCATGAGCGTTTACTTCAACGAG 97
1 1i1 i i I I I I I I Il l l l l lill i l l l l i I l l l l I l l l l i
102 ATGGCACGTCTGACCTCCAGCTCGAGCGCATGAACGTCTACTTCAACGAG 151
98 GTTCG....... TGGCCCGAACTCCAAACCTTCCGAGATGTCCACAACGC 140
II II II I I I I I IllI I lIII
152 GTACGCCCGCATTGAGCAGAGCACCAAACTGCTCGAACTCGAGCTGACGC 201
141 GTCTCTTGGTTCATACGGACCCACTGACCGCCTCTCCGGCTTCCGGCAA 190
I I I l l l i l i l l l l l
202 G ........................... AACTGCACAGGCTTCCGGCAA 223
191 CAAGTACGTTCCTCGCGCCGTCCTCGTCGATCTTGAGCCCGGTACCATGG 240
1 l i l l l1 I I I I I I I l il i l l l l i lI l lI 1 1 I I 1 1 1 1 1 11i l
221 CAAGTATGTCCCACGTGCCGTCCTCGTCGATTTGGAGCCTGGCACCATGG 273
241 ATGCTGTCCGTGCTGGTCCTTCGGTCAGCTTTTCCGCCCCGACAACTTC 290
I I I I I I I I I I I I I I i I I i l l l i l l l l l l l l l l i lI I I I Il l l
274 ATGCCGTCCGCGCTGGTCCATTCGGCCAGCTTTTCCGCCCAGACAACTTC 323
291 GTTTTCGGCCAGTCTGGTGCTGGCAACAACTGGGCCAAGGGTCACTACAC 340
I I lI l l Il l l I I i I I i l I l l l l ll l l l l l lI ll i I I
324 GTCTTCGGCCAGTCCGGCGCCGGAAACAACTGGGCCAAGGGTCACTAAAC 373
Figure 2-11. Sequences of amplified segments of P-tubulin gene for the isolates 2619
(upper line) and WH83 (lower line). Vertical lines show matching of nucleotides between
aligned sequences. Sequences inside the boxes correspond to exon areas.
likelihood trees. However, in this tree the isolate 2619 grouped with the isolates from
Texas (62% bootstrap support).
The hypothesis that the differences observed in DNA sequences of P-tubulin gene
of isolate 2619 were due to the presence of extra copies of this gene in its genome was
disproved through Southern blot analysis (Figure 2-12). Southern blots showed the
presence of a single copy of p-tubulin gene in all eight isolates screened, including the
isolate 2619. As predicted in the restriction map of 0-tubulin, the DNA probe hybridized
to two DNA fragments digested with Pst I, to one DNA fragment digested with EcoR I,
and to two DNA fragments digested with Hind II.
The Figure 2-13 shows one of the three most-parsimonious trees inferred from a
segment of 309 bp of the histone 3 dataset, for 14 isolates of C. piaropi/C. rodmanii,
which also showed a strong grouping (96% bootstrap support, decay index = 4) for the
isolates WH83, WH9BR, WHK, and 2619 from Florida, USA; 2867 and MX3 from
Mexico; 10, 62-2, and 28-1 from Brazil; 114 from South Africa; WHV from Venezuela;
and 400 from Zambia. In addition, this tree also showed a strong grouping (100%
bootstrap support, decay index = 11) for the isolates TX18 and TX20 from Texas. The
same two well-supported groupings were observed in the maximum likelihood tree
(Figure 2-14) and in the neighbor-joining tree inferred from this dataset (Figure 2-15). In
the neighbor-joining tree only two groupings had strong support (84 and 100% bootstrap).
The statistical comparison between the topologies of the trees inferred from
parsimony analyses showed significant differences among them (Table 2-4). The
elongation factor-la tree differed significantly (P=0.0175) from and the P-tubulin tree
but not from histone 3 tree; however when the P-tubulin tree was constrained by inserting
Pst I EcoR I Hind III
i ig ..r
12 3 4 5 6 7 8 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
Figure 2-12. Southern blot analysis of the P-tubulin gene of Cercospora species from
waterhyacinth. Total DNA of isolates 2619 (lane 1), 28-1 (lane 2), WH83 (lane 3),
WH9BR (lane 4), 62-2 (lane 5), WHK (lane 6), 400 (lane 7), and WHV (lane 8) were
digested with Pst I, EcoR I, and Hind III. Digested DNA was probed with a 32P-labeled
380 bp PCR amplification of the P-tubulin gene from the isolate 28-1.
Figure 2-13. One of three equally parsimonious trees inferred from histone 3 gene (length
= 147, consistency index = 0.63, retention index = 0.963, rescaled consistency index =
0.956). Bootstrap replication percentages and decay indices (in parentheses) are indicated
above the nodes. Edge length is indicated below the nodes and branches.
Figure 2-14. Maximum likelihood tree inferred from histone 3 gene. Bootstrap replication
frequencies are indicated above the nodes.
Figure 2-15. Neighbor-joining tree on distances derived from sequences of the histone 3
gene. Bootstrap replication percentages are indicated above the nodes.
Table 2-4. Maximum likelihood comparisons of tree topologies obtained for elongation factor-la, P3-tubulin, and
histone 3 sequences a.
Treeb Length In LC Difference SDe Tf Pg
(steps) In Ld difference
EF-la (Fig. 2-5) 17 -698.49004 best
tub2 (Fig. 2-8) 28 -754.20175 55.71171 23.36445 2.3845 0.0175*
Constrained tubh 20 -721.40525 22.91521 19.91336 1.1507 0.2505
H3 (Fig. 2-13) 20 -721.40525 22.91521 19.91336 1.1507 0.2505
a Results were obtained with the Kishino-Hasegawa test implemented with PAUP*4.0b 1.
bMost parsimonious tree from each dataset.
d Difference in log likelihood compared to that of the best tree.
SStandard deviation of log likelihood.
g Probability of getting a more extreme T-value with the two-tailed test under the null hypothesis of no
difference between the two trees.
h Isolate 2619 was placed in the same clade containing the isolates WH83, WHK, WH9BR, 10, 62-2, 28-1,
WHV, 114,400, MX3, and 2867.
* Significant at P-0.05.
the isolate 2619 inside the major clade, it was no longer significantly different.
Therefore, the differences in tree topologies were due to differences in nucleotide
sequences of the intron area of the P-tubulin gene in the isolate 2619 as shown in the
Even though these three datasets should not be combined because they have
heterogeneous phylogenetic signals, the most parsimonious tree inferred from the
combined datasets showed basically the same topology of the trees inferred from
individual datasets. A strict consensus of the 24 most parsimonious trees shown in Figure
2-16, belongs to the same two well-supported groupings of isolates observed in the trees
of individual datasets, with 100% bootstrap support for the minor clade grouping of the
isolates from Texas and 99% bootstrap support for the major clade grouping of the
isolates from the other origins. With combined data, the isolate 2619 was grouped within
the major clade. The topology of the maximum likelihood tree of combined data was the
same as the maximum parsimony tree (Figure 2-17); and the neighbor-joining tree, as in
the individual datasets, showed a more branched topology but with only two well
supported clades, which grouped the same isolates like the other trees (Figure 2-18).
All datasets analyzed separately or combined, through maximum parsimony,
maximum likelihood or neighbor-joining methods, suggested that the isolate WH9BR,
identified as C. rodmanii, was not distinct from the other members of the major clade that
grouped isolates from several geographical locations, including some isolates that showed
conidial size and morphology in concordance with the description of C. piaropi, such as
WH83, RR29, 10, 28-1, 18-2, and MX3. In addition, the characteristics and dimensions
.oo r TX20
Figure 2-16. Strict consensus of 24 most-parsimonious trees of length 246 based on
parsimony analysis of combined elongation factor-la, P-tubulin, and histone 3 data sets.
Bootstrap replication percentages are indicated above the nodes.
Figure 2-17. Maximum likelihood tree inferred from the combined elongation factor-la,
P-tubulin, and histone 3 datasets. Bootstrap replication percentages are indicated above
Figure 2-18. Neighbor-joining tree on distances inferred from the combined elongation
factor-la, P-tubulin, and histone 3 datasets. Bootstrap replication percentages are
indicated above the nodes.
of conidia were unreliable criteria for taxonomic differentiation of isolates that composed
the two groupings defined by the phylogenetic analysis.
More than 3000 species names have been validated in Cercospora (Pollack,
1987). The circumscription of species in this genus has been based on host affiliation and
conidial morphology of the species (length, width, and morphology of bases and tips) and
conidiophores (length, diameter, geniculation, and fasciculation) (Chupp, 1953; Ellis,
1971). As rationalized by Ellis (1971), "it has been customary for plant pathologists and
mycologists to describe as new any Cercospora found on a host plant for the first time".
Chupp (1953) and Ellis (1971), discussed the reliability of the criteria used for species
circumscription in Cercospora. Some species in this genus have wide host ranges and the
size of conidia and conidiophores can have variations induced by changes in
environmental conditions, especially humidity. These factors complicate the taxonomy
and identification of species belonging to this genus. The study reported here shows that
the criteria used by Conway (1976a) for differentiating the species C. rodmanii from C.
piaropi are not adequate to provide a clear identification of these species.
This study inferred phylogenetic relationships for a population of isolates
belonging to the complex C. piaropi/C. rodmanii and obtained from several geographical
locations. The phylogenetic trees inferred from three independent lines of evidence were
concordant in showing that the population of isolates of C. piaropi/C. rodmanii from
waterhyacinth grouped in two well-supported clades: a major clade, which grouped
isolates collected in the center of origin of waterhyacinth (Brazil) and in several of the
host's adventive areas; and a minor clade, which grouped isolates only from Texas. The
characteristics and dimensions of conidia were unreliable criteria for differentiation of
isolates that composed these two groups, and the isolate WH9BR that is the type culture
of C. rodmanii was not differentiated, based on DNA sequencing data, from the isolates
that had morphological features of C. piaropi. Thus, the differentiation of the species C.
piaropi and C. rodmanii based on phenotypic traits, besides being not clear in some
situations was also not supported by the DNA sequencing data. Based on the finding that
isolates of these species grouped together in a well-supported clade, they should be placed
in the same species according to the phylogenetic species concept (Nixon and Wheeler,
1990; O'Donnell and Cigelnik, 1997).
The differentiation between the species C. piaropi and C. rodmanii is difficult.
Martyn (1985) and Morris (1990) noted that C. piaropi and C. rodmanii appear to be
closely related, both in morphology and disease symptomatology. In addition, in the
emended description of C. piaropi, Conway (1976a) described conidia of C. piaropi as
acicular with obconic bases, but earlier descriptions refer to them as obclavate to acicular
(Thirumalachar and Govindu, 1954), truncate, tapering towards the tip (Tharp, 1917;
Freeman and Charudattan, 1974), or as truncate (Chupp, 1953). Morris (1990) observed
that C. piaropi conidia were mostly acicular with truncate bases, with a few conidia being
obconic. In reference to the size of conidia, Freeman and Charudattan (1984) observed
that frequently both long and short conidia were produced on dead leaf tissue when leaves
infected by C. piaropi were incubated under moist conditions. In this study, the
possibility of having the isolates mixed was ruled out because all cultures were obtained
as monocultures from hyphal tips, and this procedure was done twice.
Therefore, to avoid problems of identification and also problems of
communication among scientists and quarantine officials, I propose a second emendation
to the description of the species C. piaropi and to consider the species C. rodmanii as its
synonym. Cercospora piaropi should remain as the valid name based on the principle of
priority of publication of the International Code of Botanical Nomenclature (Greuter et
The two groups defined by the phylogenetic analysis as major and minor clade
groupings could be considered to represent two different species of Cercospora,
according to the phylogenetic concept of species (Nixon and Wheeler, 1990; O'Donnell
and Cigelnik, 1997). However, until additional criteria are available to delimit
Cercospora species, these groups must be considered to be taxonomically conspecific.
Wang et al. (1998) also considered conspecific the two groups found in C. zeae-maydis,
identified through AFLP analysis and DNA sequencing of the 5.8S ribosomal DNA and
the ITS regions. Indeed, differently from this study, Wang et al. (1998) found that the
DNA sequences of the 5.8S ribosomal DNA and the ITS regions had eight informative
characters which supported the differentiation of the two groups observed through AFLP
analysis. The authors interpreted these two groups as being sibling species.
The study reported here is a contribution for the taxonomy of species in
Cercospora. However, many questions still need to be answered, such as whether the
Cercospora species defined by host affiliation and morphological criteria correspond to
the distinct lineages identified by cladistic analysis, and whether the form-genus
Cercospora is a monophyletic group. Studies to determinate the correlation between
lineages defined by host affiliation and lineages defined by molecular markers, as well as
their common ancestry, have been presented for other groups of fungi and have been very
insightful. For example, molecular studies identified several clonal lineages with
independent evolutionary origins in Fusarium oxysporum Schlechtend.:Fr.f.sp. cubense
(E.F.Sm) Snyder & Hans, a taxon defined by its pathogenicity to bananas (Koenig et al.,
1997; O'Donnell et al., 1998b). In addition, these pathogens of banana could be as
closely related to pathogens of other hosts. In the anther smut fungus Microbotryum
violaceum (Pers.:Pers.) Deml. and Oberw. correlation was found between haplotypes,
defined by partial DNA sequence of y-tubulin gene and their respective host species of
origin (Garr et al., 1997).
The utility of using elongation factor-la, P-tubulin, and histone 3 genes for
phylogenetic studies at the species level or below has been demonstrated in previous
work. DNA sequences of the P-tubulin gene were used in phylogenetic studies in
Fusarium and in the Gibberellafujikuroi (Sawada) Wollenw. species complex
(O'Donnell et al., 1998a; O'Donnell and Cigelnick, 1997) and in Aspergillus section
Fumigati (Geiser et al., 1998a). DNA sequences of the histone 3 gene were used to
differentiate Fusarium species associated with conifers (Donaldson et al., 1995). More
recently, DNA sequences of elongation factor-1 a were used for phylogenetic studies in
Fusarium oxysporum (Schlecht.) Snyd. and Hans. (O'Donnell et al., 1998b). In this
study, these three loci were shown to be appropriate for phylogenetic inferences in
Cercospora. These loci can provide diagnostic tools needed to investigate species
boundaries and to correlate the host range and biogeography of these fungi.
The question of whether independent datasets should be analyzed separately or
combined and analyzed simultaneously is still controversial. Miyamoto and Fitch (1995)
defend that independent datasets should rarely be combined but should be kept separate
for phylogenetic analysis because their independence increases the significance of
corroboration. According to de Queiroz et al. (1995), assuming that the goal is to
uncover the true phylogeny of the entities in question, arguments to combine data based
on the notion that one should use the "total evidence" available, or that the combined
analysis gives the tree the greatest descriptive and explanatory power, are not compelling.
However, to combine datasets can enhance detection of real phylogenetic groups. On the
other hand, if there is heterogeneity among datasets with respect to some property that
affects estimation of phylogeny, then combining the data can give misleading results. In
this study, even though according to the statistical test the three datasets should not be
combined, the phylogenetic tree inferred from the combined data did not show misleading
results (Figures 2-16; 2-17; 2-18).
Taking in account the limitations in the number of samples analyzed, the
phylogenetic analysis presented in this study provides some insights about a
biogeographic hypothesis for Cercospora on waterhyacinth. The grouping of isolates of
C. piaropi/C. rodmanii from different geographical origins in a same clade, including the
isolates collected in the center of origin of waterhyacinth (Brazil and Venezuela),
suggests that the fungi may have been spread together with the plant host, from its center
of origin. It has been documented that waterhyacinth was dispersed, mostly by man, from
the lowlands of tropical South America to many tropical and subtropical regions of the
world (Barrett, 1977; 1988). In addition, the occurrence of a distinct population of the
pathogen only in a single region outside the center of origin of the host plant (Texas
isolates), suggests that this population may have adapted to waterhyacinth from other
host(s). Thus, a multiple origin of Cercospora pathogens of waterhyacinth cannot be
Biogeographic hypotheses from phylogenetic evidence has not been proposed for
species in Cercospora, but have been proposed for a few other groups of fungi. For the
Gibberellafujikuroi species complex, gene trees inferred from P-tubulin and 28S rDNA
supported a phylogeny consistent with species radiations in South America, Africa, and
Asia (O'Donnell et al., 1998a). These analyses placed the American clade, with 12
species, as a monophyletic sister-group of an African-Asian clade which had 12 and 8
species, respectively. The biogeographic hypothesis proposed in their study reflected
vicariant events associated with the fragmentation of the super-continent Gondwana.
Otherwise, other groups of fungi have a more complex biogeographic history, such as the
genus Lentinula that includes cultivated shiitake mushrooms (Hibbett et al., 1998). Here,
based on phylogenetic analysis of the 5.8S rDNA gene and its ITS flanking regions,
Hibbett et al. (1998) hypothesized that the present distribution of Lentinula in four
continents must result from some combination of vicariance, dispersal, and extinction.
Data from this study suggest a dispersal hypothesis for the Cercospora species
from waterhyacinth following the path of the plant host from its center of origin and in
combination with possible adaptations of native populations of Cercospora in the
adventive areas of the plant host. However, further studies are needed to address this
question, including a greater number of isolates sampled from different areas in the center
of origin and the adventive areas of waterhyacinth.
The large unexpected difference in nucleotide sequences observed in the segment
of P-tubulin gene of the isolate 2619 compared with the other isolates may be due to the
presence of extra copies of this gene in the genome of this isolate, as was observed by
Tsai et al. (1994) in the genome of EpichloE species. These authors interpreted this
finding as being the result of inter-specific hybridization among species belonging to this
genus. However, since the Southern blot analysis of the P-tubulin gene disproved the
hypothesis of extra copies of this gene in the genome of the isolate 2619 (Figure 2-12),
the reasons of such differences in nucleotide sequences are not clear. One possibility is
the occurrence of a history of recombination in this isolate. For instance, based on the
principle that full compatibility among gene genealogies indicates complete asexuality,
and incompatibility indicates mixis, Koufopanou et al. (1997) concluded that the
incompatibility among five protein-coding-gene genealogies inferred from isolates of the
human fungal pathogen Coccidioides immitis Rixford and Gilchrist was an indication of
sexual recombination in the population of this fungus. The same approach was given by
Geiser et al. (1998b) to determine the occurrence of history of recombination in
Aspergillusflavus Link, based on the lack of concordance among gene genealogies from
five protein-coding genes among a population of isolates of this fungus.
This analysis clearly demonstrated the utility of the elongation factor-1 a, 3-
tubulin, and histone 3 genes for phylogenetic analysis of closely related species of fungi.
All three genes harbor considerable phylogenetically informative variation that can
provide diagnostic tools needed to investigate boundaries of fungal species and even
biogeography. Phylogenetic reconstruction based on independent loci, such as the
protein-coding genes used in this study, can be very useful for biological control. It can
provide tools to delimite the species used as biocontrol agents, to track them in the field,
and to understand aspects of their origin and reproductive biology.
I propose a second emendation to the species description of C. piaropi, a species
originally described by Tharp (1917) as follows:
Cercospora piaropi sp. nov.
Spots ovate, grayish-tan centered with purplish-black borders somewhat raised
above, brighter above than below, 1.5-3 x 3-5 mm in diameter, or larger by
confluence; conidiophores epiphyllous, fasciculate but very few in each fascicle,
sparse, bright brown with yellowish apices, denticulate, sometimes branched,
pluriseptate, 100-125 x 3.5-4.5 pim; conidia hyaline, truncate at base, upward
attenuate, pluriseptate at maturity, 80-140 x 3pm.
On living leaves on Piaropus crassipes (Mart.) Britton, Palestine, Texas, Oct.30,
1914, 1. M. Lewis & B. C. Tharp.
Conway (1976a) emended the description given by Tharp (1917) as follows:
Cercospora piaropi Tharp emend. Conway
Leaf spot ovate, dark brown, later with a grayish-tan center with dark brown
borders, 1.5-3 x 3.5 mm, larger by confluence; fruiting amphigenous; stromata
lacking or a few brown cells; conidiophores borne singly or in fascicles of two to
nine, dark brown, multiseptate, not branched, sympodial, 55-200 x 2.5-5 pm; conidia
hyaline, acicular, straight to mildly curved, multiseptate, base obconic, 25-(95)-220 x
SPECIMEN: Deposited as the National Fungus Collection (BPI).
CULTURE: Deposited at the Florida Division of Plant Industries, Gainesville,
The new species, C. rodmanii, was described by Conway (1976a) as follows:
Cercospora rodmanii Conway sp. nov.
Maculae nigrae punctulatae ad rodundas, 1-3 mm latas, phyllum chloriticum et
petiolus chloriticus, extremus phylli mortuus; conidiophora fasciculata, 3-12,
amphigena, brunnea, sympodialia, orientia ex stromate, emergentia per stoma; 84-
(145)-284 x 4-(4.5)-5 lim; conidia hyalina, truncata, acicularia, multiseptata, 66-
(172)-374 x 3-(4)-5 pm.
Pycnidia brunnea, ostiolata, globosa 80-95 x 80-110 jm, sub stomate, deinde
erumpentia; ostiola 30-40 x 25-30 lm; conidia hyalina, in forma bacillorum 2-3.5 x
HABITAT: In phyllis Eichhornia crassipes (Mart.) Solms.
Leaf spots black, punctate to circular (1-3 mm diam), leaf and petiole chlorotic,
tip of leaf necrotic, conidiophores amphigenous, 3-12 in each fascicle, brown
sympodial, arising from a well developed stroma, emerging through the stroma, 84-
(145)-284 x 4-(4.5)-5 gm; conidia hyaline, truncate at base, acicular, multiseptate,
66-(172)-374 x 3-(4)-5 gm.
ASSOCIATED STATE: Asteromella pynidia dark brown, ostiolate, globose, 80-
95 x 80-110 [tm, substomatal, later erumpent, ostiole 30-40 x 25-30 Rim; condia
hyaline, bacilliform 2-3.5 x 1-1.5 lim.
TYPE SPECIMEN: Deposited at the National Fungus Collection (BPI).
TYPE CULTURE: Florida Division of Plant Industries, Gainesville, Florida,
HABITAT: On leaves of waterhyacinth (Eichhornia crassipes [Mart.] Solms).
Collected by K.E. Conway, Rodman Reservoir, Orange Springs, Florida.
Herein I propose a second emend of C. piaropi as follows:
Cercospora piaropi Tharp emend. Tessmann and Charudattan
Leaf spot grayish-tan, dark brown to black, ovate, punctate to circular, 1-3 x 3-5
mrn in diameter, or larger by confluence, leaf and petiole chlorotic, tip of leaf
necrotic; conidiophores amphigenous, stroma lacking, or with few brown cells or
well-developed; conidiophores borne singly or in fascicles of two to twelve, brown
or dark brown, multiseptate, sometimes branched 55-284 x 2.5-5 pm; conidia
hyaline, truncate at base or obconic, acicular, multiseptate, 25-374 x 2-5 Rm.
Associated state: Asteromella pycnidia dark brown, ostiolate, globose, 80-95 x 80-
110 gtm, substomal, later erumpent, ostiole 30-40 x 25-30 im; conidia hyaline,
bacilliform 2-3.5 x 1-1.5 jim.
Specimen: Deposited at the National Fungus Collection (BPI).
Culture: Deposited at ATCC.
PATHOGENIC VARIABILITY AND BIOCHEMICAL CHARACTERIZATION OF
CERCOSPORA SPECIES FROM WATERHYACINTH
Waterhyacinth (Eichhoria crassipes [Mart.] Solms), an aquatic plant indigenous
to lowland tropical South America (Perfound and Earle, 1948; Barrett, 1988), was spread
worldwide by man. Its free-floating habit and capacity for rapid vegetative propagation
has enabled it to become one of the most noxious aquatic weeds in many tropical and
sub-tropical regions of the world. Problems caused by this weed include clogged
irrigation canals, blocked waterway transport routes, water losses, and reduction in fish
populations in reservoirs (Pieterse, 1990). The strategies used to control this weed have
included mechanical removal, chemical control with herbicides, and biological control
with pathogens and insects (Charudattan, 1986; 1990). Among the pathogens studied as
biocontrol agents, the species Cercospora piaropi Tharp and C. rodmanii Conway have
been shown to decrease waterhyacinth biomass, and in some instances to cause
substantial decline of waterhyacinth populations (Charudattan et al., 1985; Freeman and
Charudattan, 1984; Martyn, 1985; Morris, 1990).
The species C. piaropi was described by Tharp (1917) in Texas, and almost sixty
years later, Conway (1976a) emended the description of C. piaropi and described a new
species, C. rodmanii. The new species was described based on plant specimens collected
at the Rodman reservoir, Florida, where a severe epidemic of a Cercospora leaf-spot
disease caused the decline of the population of waterhyacinth in that area. This new
species was differentiated from C. piaropi mainly based on conidial morphology, disease
symptomatology, and on C. rodmanii being a more aggressive pathogen than C. piaropi.
This idea was reinforced by the early reports that described C. piaropi as a pathogen not
able to cause serious damage to waterhyacinth (Nag Rag, 1965; Freeman and
Charudattan, 1974) and because the epidemics recorded at the Rodman Reservoir was the
first devastating epidemic of a Cercospora leaf spot on waterhyacinth ever noticed.
However, C. piaropi also was recorded later to cause serious damage to waterhyacinth in
Texas (Martyn, 1985) and in South Africa (Morris, 1990), even though these authors had
difficulties to differentiate this species from C. rodmanii. Indeed, the differentiation
between these two species with Conway's criteria has not been an easy task, as discussed
in Chapter 2. Until the taxonomic status of C. piaropi and C. rodmanii are changed, as
discussed in Chapter 2, this study will refer to these species as the complex C. piaropi/C.
rodmanii complex rather than referring to them as individual species.
The occurrence of pathogenic variability, even though it was presumed to exist,
has not been identified in C. piaropi/C. rodmanii. In addition, isolates of C. piaropi/C.
rodmanii, as many other Cercospora species, show a great variation in their cultural
features, such as mycelial color, pigmentation color and intensity, and growth rate. The
extent these factors are related to pathogenicity or virulence of these isolates is unknown.
Such information would be very important to optimize large-scale inoculum production,
which is an important step in the bioherbicide strategy to control this weed (Charudattan
et al., 1985; Charudattan, 1986). In Cercospora spp., two colored secondary metabolites
have been identified as phytopathogenic toxins: a red-purple compound, called
cercosporin, which was identified in several species in this genus (Assante et al., 1977;
Jenns et al., 1989; Lynch and Geoghegan, 1977; Upchurch et al., 1991), and a yellow
compound, identified only in Cercospora beticola, which corresponds to a group of six
toxins, named beticolins and formerly known as Cercospora beticola toxin (CBT) (Milat
and Blein, 1995).
The objective of this study was to determine the extent of variation in virulence
and in some physiological characteristics, including the production of phytopathogenic
toxins, among a population of C. piaropi/C. rodmanii isolates collected in several
geographic locations. An additional objective was to determine the extent of variation of
fatty acid methyl ester profiles (FAME), with the purpose to find useful biochemical
markers to differentiate isolates, populations, and species among a collection of isolates
of C. piaropi/C. rodmanii. FAME profiles have been used as biochemical characters to
address taxonomic issues in fungi at species and sub-species levels (Augustyn et al.,
1990; Bentivenga and Morton, 1996; Berger et al., 1991; Graham et al., 1995; Johnk and
Jones, 1993; Malfeito-Ferreira et. al. 1989; Martinez et al., 1991; Nemec et al., 1997; da
Silva et al., 1998; Stahl and Klug, 1996; Viljoen and Kock, 1989). The knowledge of
pathogen variability can provide valuable information for biological control programs to
select the most effective strains, to infer the stability of these strains, and to define the
boundaries of species and populations.
Materials and Methods
Fungal Isolates and Cultural Characteristics
The origin, designation, and the name of the collectors of the isolates used in this
study are listed in Table 3-1. All isolates were obtained from symptomatic leaves of
waterhyacinth and monocultures were obtained from mycelial tips. These isolates have
been preserved in the fungal collection of the Biological Control of Weeds Laboratory of
the Plant Pathology Department, University of Florida.
Fungal cultural characteristics were studied on plates containing 10 ml of PDA.
Three replicates for each isolate were inoculated with 4-mm-diameter mycelial plugs
removed from the margin of 7-day-old colonies growing in 9-cm-diameter plates
containing 20 ml of PDA. Inoculated plates were incubated in a growth chamber at a
temperature of 2420C with 12 h light, and radial growth was measured 8 days later. The
mean diameter was calculated as the average of two diagonal measurements of colony
diameter on each plate. Colony morphology, pigment diffusion, and mycelial
pigmentation were also recorded at 8 and 14 days after inoculation. The experiment had a
completely randomized design with three replicates for each isolate and was repeated
Toxin analysis was based on the protocol developed by Milat and Blein (1995).
Isolates were cultured in 50-ml test tubes containing 15 ml of V8 broth medium (200 ml
V8 juice, 3 g CaCO3, and 800 ml of tap water; sterilized). Each tube was inoculated with
eight, 4-mm-diameter, mycelial plugs removed from the margin of 7-day-old colonies
growing in 9-cm-diameter plates containing 20 ml of PDA. After incubation for 14 days
at 2520C under constant light, the mycelium was separated from the broth by filtration
using a double-layered cheese cloth, and blended for about 5 sec in a Waring blender in
ethyl acetate (20 ml/g of wet mycelium). The crude extract was separated from mycelial
debris and resolved by thin-layer chromatography using pre-coated TLC plates of silica
Table 3-1. Designations and geographic origin of the isolates of Cercospora piaropilC.
rodmanii analyzed in this study.
Gainesville, FL, USA
Rodman Reservoir, FL, USA
Rodman Reservoir, FL, USA
Kissimee, FL, USA
Suwannee River, FL, USA
Marion Co., FL, USA
Marion Co., FL, USA
Marion Co., FL, USA
Marion Co., FL, USA
Marion Co., FL, USA
Marion Co., FL, USA
Sarasota Co., FL, USA
Sarasota Co., FL, USA
Sarasota Co., FL, USA
Sarasota Co., FL, USA
Lee Co., FL, USA
Leon Co., FL, USA
Lake Conroe, TX, USA
Lake Conroe, TX, USA
Lake Conroe, TX, USA
Lake Conroe, TX, USA
Rodman Reservoir, FL, USA
Rodman Reservoir, FL, USA
Rodman Reservoir, FL, USA
Rodman Reservoir, FL, USA
Rodman Reservoir, FL, USA
Rodman Reservoir, FL, USA
Gainesville, FL, USA
Gainesville, FL, USA
Gainesville, FL, USA
Vero Beach, FL, USA
Minas Gerais, Brazil
Minas Gerais, Brazil
Minas Gerais, Brazil
Designation Geographic origin Collector Date
46-4 Pernambuco, Brazil R. Charudattan 1997
49-1 Pernambuco, Brazil R. Charudattan 1997
49-2 Pernambuco, Brazil R. Charudattan 1997
16-1 Corumba, Mato G. do Sul, Brazil R. Charudattan 1995
18-2 Corumba, Mato G. do Sul, Brazil R. Charudattan 1995
28-1* Rio Verde, Mato G. do Sul, Brazil R. Charudattan 1995
10* Sao Paulo City, SP, Brazil R. Charudattan 1996
34 Rio Grande do Sul, Brazil R. Charudattan 1997
2867* Mexico R. Charudattan 1994
2943 Mexico R. Charudattan 1994
MX3* Mexico R. Charudattan 1997
WHV* Venezuela R. Charudattan 1982
114* South Africa M. Morris ?
279 South Africa M. Morris ?
400* Zambia M. Morris ?
a Designation corresponds to notation of isolates used in this study. Isolates labeled with
an asterisk (*) were included in the fatty acid analysis.
gel without fluorescent indicator (E. Merck, Darmstadt, Germany). A total of 5 pl of
crude extract was spotted in the TLC plate for each isolate. Chromatograms were
developed using chloroform/methanol/water (80:20:2, v/v) as elution mixture. The
standards used were cercosporin from Sigma Chemical Co.(St. Louis, MO), and
beticolin-1, obtained from Dr. L. Milat (INRA, Dijon Cedex, France). In addition, the
ultraviolet spectrum of the crude extract was determined in a spectrophotometer and
cercosporin was quantified in each sample through reading of its absorbance in
spectrophotometer at 473 nm. The amount of cercosporin was calculated using the
formula: absorbancece at 473 nm/31,455)534] gg, where 31,455 equaled the molar
extinction coefficient at 473 nm and 534 equaled the molecular weight of cercosporin.
This formula contained a corrected extinction coefficient with ethyl acetate as solvent,
since in the original formula, used by Jenns et al. (1989) and Velicheti and Sinclair
(1994), 5 N KOH was used as solvent. This correction was needed because the
maximum absorption of cercosporin in 5 N KOH was 480 nm and in ethyl acetate was
473 nm. Each sample corresponded to an individually growing isolate. The experiment
had three replicates and was performed twice in a randomized block design.
Pathogenicity Test and Virulence Analysis
Waterhyacinth plants for virulence tests were vegetatively propagated from plants
collected from Lake Alice, located on the campus of the University of Florida. The plants
were propagated in pots with tap water supplemented (0.05% w/v) with chelated iron
(Keel-Iron, NaFe EDTA 5%; Chase & Company, Sanford, FL). For the pathogenicity
tests and screening for virulence, daughter ramets (=offsets or clones) of waterhyacinth,
with 2 leaves, and with approximately the same root mass, were placed in pots containing
125 ml of tap water with the supplement described above. Each pot had one plant and the
volume of the liquid in the pots was maintained at the initial level by daily additions of
Due to irregular, or lack of, sporulation in culture, the inoculum was prepared
from mycelium grown in still liquid cultures. Isolates were cultured in 250-ml flasks
containing 50 ml of V8 broth medium (200 ml V8 juice, 3 g CaCO3, and 800 ml of tap
water; sterilized). Each tube was inoculated with eight 4-mm-diameter mycelial plugs
removed from the margin of 7-day-old colonies grown in 9-cm-diameter plates containing
20 ml of PDA. After inoculation for 14 days at 2530C with 12 h light, the mycelium
was separated from the broth by filtration using a double-layered cheese cloth, and
blended for about 5 sec in a Waring blender at a concentration of 80 mg mycelium per ml
of water. The suspension was amended with 0.5% Metamucil (psyllium mucilloid;
Procter & Gamble, Cincinnati, OH) and 0.05% Silwet L-77 (polyalkyleneoxide modified
heptamethyltrisiloxane, 0.02% v/v, OSI Corp., Loveland Ind., Inc., Greeley, CO).
A bioassay was developed which consisted of immersing waterhyacinth leaves,
attached to plants, in a suspension of inoculum. Control leaves were immersed in a
solution containing only the amendments. This procedure was used after other
inoculation procedures, such as foliar spraying and droplet deposition on leaves, were
evaluated and found to be unsuitable. The inoculated and control plants were placed in a
dew chamber in darkness, at 2520C for 12 h, and then held in a greenhouse for 2 wk.
This experiment was conducted in a quarantine greenhouse. Disease severity was
assessed 7 and 14 days after inoculation. Disease was rated using a rating scale, where: 0
= no symptoms; I = less than 1% of the lamina surface with spots; 2 = greater than 1%
and less than 10% of the lamina surface with spots; 3 = greater than 10% and less than
25% of the lamina surface with spots; 4 = greater than 25% and less than 50% of the
lamina surface with spots; 5 = greater than 50% and less than 75% of the lamina surface
with spots; 6 = greater than 75% of the lamina surface with spots; and 7 = dead lamina.
The treatments were arranged in a randomized block design with four replications
(pots) per isolate, each with one 2- to 3-leaved plant per pot. Statistical analyses of
pathogenic variability, growth rate, and cercosporin production were done with the GLM
and CORR procedures of SAS (SAS Institute, Cary, NC).
Fatty Acids Analysis
Fourteen isolates of C. piaropi/C. rodmanii, representing different geographical
locations, and three other species of Cercospora (outgroups) were included in this study
(Table 3-1). All fungi were grown in 80 ml of modified TSB (trypticase soy broth with
10 g dextrose per liter; BBL Microbiology Systems, Becton Dickinson and Co.,
Cockeysville, MD) in 250-ml flasks in a slow-shake culture (130 rpm) for 4, 5, and 6
days at 242C in dark before harvest and fatty acid extraction. Four, 4-mm diameter
plugs removed from the periphery of 7-day-old cultures in PDA were used as initial
inoculum. Cultures were harvested using a side-arm Erlenmeyer flask fitted with a
Buchner funnel and filter attached to a vacuum pump. A nylon-type filter (polypropylene,
mesh opening of 105 .m) was used to prevent the sample from becoming contaminated
with paper fibers. After harvest, the fungal mycelium was transferred to 15-ml
polypropylene tubes and kept in a freezer at -700C until lyophilization.
Fifty-gram (dry weight) of samples of fungal tissue were placed in clean screw-
cap test tubes (13 by 100 mm; with Teflon cap liners), 2 ml of a saponification reagent
(45g sodium hydroxide in I liter of 50% methanol) was added and the mixture was
homogenized with a vortex mixer for 10 sec. The homogenate was then saponified at
1000C in a water bath for 5 min, homogenized in a vortex mixer for 10 sec, and kept in a
water bath for 25 min at 100C, then cooled in a room-temperature water bath. To
methylate the liberated fatty acids, 2 ml of 54% 6 N HC1 in methanol was added to each
tube. Sub-samples were then placed in an 800C water bath for 10 min and immediately
cooled to room temperature. To extract fatty acid methyl esters from the aqueous phase,
1.25 ml of 50% hexane-50% methyl tert-butyl ether was added to each tube, and the tubes
were rotated end-over-end for 10 min. Next, the aqueous phase (bottom of tube)
contained fungal debris was removed with a Pasteur pipette, and 3.0 ml of 1.2 NaOH in
H20 was added to each tube; the tube was then rotated end-over-end for 5 min. Finally,
the organic phase (top of tube) contained the fatty acid methyl esters was removed from
the tubes and placed in a crimp-top gas chromatography vial.
Fatty acid extracts were analyzed by gas-liquid chromatography with the
Microbial Identification System (MIS) developed by Microbial I.D., Inc. (Newark, DE).
This system was designed to analyze the fatty acid composition of unknown bacteria and
to identify them by matching the fatty acid composition of an unknown organism to one
of the fatty acid profiles in its computer database of known organisms. The MIS consists
of a chromatographic unit chromatographh, integrator, and autosampler) coupled to a
computer system. The gas-liquid chromatograph is equipped with a 25 m by 0.2 mm
phenyl-methyl-silicone-fused capillary column (Hewlett-Packard, Wilmington, DE) and a
flame ionization detector. Data from chromatographic analysis are sent to the computer
system, where fatty acids are identified on the basis of their retention times relative to
known standards, and quantified relative to other fatty acids in the sample on the basis of
peak width and area data. The system is calibrated with known fatty acid standards when
it is started and after every 10th sample. Results of each sample analysis are printed out in
a fatty acid composition report and also stored on a hard disk within the computer.
Samples were run through the gas chromatography column for 38 min, long enough for
fatty acids up to 28 carbons long to pass through.
Fatty acids profiles of individual isolates were based on analysis of the cellular
fatty acid content of three independently grown cultures in each of the three harvest days.
The value for each fatty acid in a given profile was the mean from all analyzed cultures of
that isolate. Concentrations of each fatty acid was expressed as a percentage of the total
fatty acid content. To determine if the fatty acid compositions of the isolates were
statistically different, discriminant analysis, implemented by SAS (SAS Institute, Cary,
NC), was used to test the hypothesis that isolate x(fal, fa2, fa3...fa) = y(fal, fa2, fa3...fan)
= ...= z(fal, fa2, fa3... fa), where fa is fatty acid (Stahl and Klug, 1996).
The isolates compared in this study formed colonies on PDA with colors that
ranged from pale- to dark gray, and pale pink to pink mycelium. The colonies produced
reddish purple or yellow compounds, with different intensities. In addition, some isolates
showed dark gray mycelia and these did not produce pigments (Figure 3-1). Reddish
purple pigment was produced by isolates with all the mycelial colors described above,
while only isolates with-pale pink mycelium produced yellow pigment. The growth rate
among the isolates ranged less than one fold, from 2.17-2.34 to 4.00-4.29 mm per day
Virulence and Toxin Analysis
A total of 55 isolates of C. piaropi/C. rodmanii were used in the virulence
testing. Significant differences (P<0.05) were observed in the ability of the isolates to
cause disease (i.e., difference in virulence) among the isolates (Table 3-3). The most
virulent isolates necrosed almost 75% of the leaf area in 14 days after inoculation. Most
isolates showed intermediate levels of virulence, and some isolates were nonpathogenic
to waterhyacinth. There was no relation between the degree of virulence and the
geographical origin of the isolates. Highly virulent and nonpathogenic isolates were
found for collections from inside and outside South America, the geographical center of
origin of waterhyacinth.
The ranking of the isolates for level of virulence was conserved between the
experiments, with some variation due to uncontrolled factors (Figure 3-2). The disease
grades of the third experiment were lower than the grades of the first and second
experiment. This variation is due to influence of external temperature, since the third
experiment was run at a cooler temperature (May) compared to the first and second
All highly virulent isolates produced purple pigments while the yellow-pigment
producers were weak to moderate in virulence. The nonpigment-producers were
nonpathogenic (Table 3-3).
Figure 3-1. An example of differences in colony characteristics of Cercospora piaropilC.
rodmanii from waterhyacinth after growth for 7 days on PDA.
Table 3-2. Mycelial growth rates of isolates of Cercospora piaropilC. rodmanii from
waterhyacinth, in mm day .
Isolate Ist Experiment 2nd Experiment
BC31 2.30 Imno 2.25 nopqrs
TX18 2.30 Imno 2.17 opqrs
SGS49 2.30 Imno 2.29 mnopqrs
2619 2.25 mno 2.17 opqrs
175104 2.25 mno 2.25 nopqrs
BA57 2.25 mno 2.21 opqrs
RR22 2.25 mno 2.08 pqrs
RR29 2.21 mno 2.17 opqrs
TX20 2.21 mno 1.92 s
BA55 2.21 mno 2.17 opqrs
RR31 2.17 mno 2.08 pqrs
62-2 2.13 no 1.96 rs
SGS25 2.08 o 2.04 qrs
RR27 2.08 o 2.04 qrs
z Values are the means of three replicates of each isolate. Means followed by the same
letter in each column do not differ according to Tukey's Honestly Significant Difference
Table 3-3. Geographic origin, color of pigments produced in axenic culture, production of plant pathogenic toxins, and virulence of
isolates of Cercospora piaropi/C. rodmanii.
Isolate Origin Pigment colorw Cercosporinx Diseasey
Isolate Origin Pigment color Cercosporin Disease
16-1 Southwest Brazil Np 0.00 w
18-2 Southwest Brazil Np 0.00 w
175-99 Florida, USA Np 0.00 w
Control -- 0.00 w
NT = not tested.
Np = non-pigmented.
w Observed in 10-day-old cultures grown on PDA plates at 251 C and 12 h light.
x Production of Cercospora toxins in cultures, detected in thin-layer chromatograms, where plus sign (+) means presence of
cercosporin and minus (-) sign means cercosporin was not detected. The toxin beticolin-l was not detected in any sample. The
procedures for culture, extraction, and detection by TLC are explained under Material and Methods. This experiment was done
Y Based on disease reaction on waterhyacinth leaves 14 days after inoculation, where: 0, no symptoms; 1, less than 1% of the lamina
surface with spots; 2, greater than 1 and less than 10% of the lamina surface with spots; 3, greater than 10 and less than 25% of the
lamina surface with spots; 4, greater than 25 and less than 50% of the lamina surface with spots; 5, greater than 50 and less than 75%
of the lamina surface with spots; 6, greater than 75% of the lamina surface with spots; and 7, dead lamina. Data were combined from
three experiments conducted in a quarantine greenhouse.
z Means followed by the same letter do not differ according to Tukey's Honestly Significant Difference procedure (P=0.05).
7 B* Experiment 1
6 -I s A Experiment 3
-e n --- Average
5 -A ** m *0
A AAAA A A A A
A #A A
1 AA A UM A
Figure 3-2. Isolates of C. piaropi/C.rodmanii plotted in a descending order by their virulence in waterhyacinth. The experiments 1, 2,
and 3 were performed in the periods of 11 to 25/June/1997, 17 to 31/July/1997, and 27/April to 1 l/May/1998, respectively.
The toxin cercosporin was present in the crude extract of most isolates analyzed
(Table 3-3); however toxin beticolin-1 was not detected in the samples analyzed. In one
isolate (114) cercosporin was detected but this isolate was nonpathogenic. This isolate, in
culture, showed a reddish purple pigment of very low intensity. Cercosporin was not
detected by the TLC method in nonpigmented isolates. However, it was detected in low
levels by spectrophotometric analysis (Table 3-4).
The isolates showed significant differences (P=0.05) in cercosporin production,
according to the quantification through absorption at 473 nm with the aid of a
spectrophotometer (Table 3-4). Even though the two experiments were run at the same
conditions, some variation was observed among them. The differences in cercosporin
production among the isolates ranged up to 17 fold in the first experiment and 12 fold in
Figure 3-3 shows the ultraviolet spectra of crude extracts for typical yellow and
reddish purple isolates, compared with the spectra of the standards for cercosporin and
beticolin-1 in ethyl acetate. The absorption spectrum of the crude extract of the yellow
pigment-producer had a maximum peak at 473 nm (Figure 3-3 A). However, it also
showed two extra peaks at 447 and 504 nm compared to the spectrum of the reddish
purple isolate (Figure 3-3 B), which was very close to the spectrum of pure cercosporin
(Figure 3-3 C). Beticolin-1 was expected to be detected through the UV spectrum,
having the maximum absorption at 435 nm (Figure 3-3 D); however, its presence was not
clear in the samples analyzed with a spectrophotometer. The isolates that were the
strongest cercosporin producers were those with more intense reddish purple
pigmentation in culture.
Table 3-4. Production of cercosporin by isolates of Cercospora piaropilC. rodmanii from
waterhyacinth, in .tg per g wet mycelium.
Isolate 1" Experiment 2nd Experiment
BA57 0.919 az 0.813 ab
175101 0.895 ab 0.747 abc
62-2 0.887 ab 0.945 a
2619 0.862 bc 0.626 bcde
62-4 0.750 bcd 0.759 ba
2702 0.737 bcde 0.664 bcd
WH9R 0.727 bcde 0.658 bcd
RR29 0.682 bcdef 0.659 bcd
SGS25 0.671 bcdefg 0.617 bcde
2703 0.668 bcdefgh 0.542 defg
61-5 0.620 cdeefghi 0.520 defgh
RR31 0.592 defghij 0.557 cdef
RR27 0.592 defghij 0.527 defgh
175102 0.566 defghij 0.296 ijklmo
BA55 0.499 efghijkl 0.333 hijklm
175108 0.456 fghijklm 0.284 ijklmno
SGS49 0.449 fghijklm 0.389 fghijk
2704 0.444 fghijklm 0.345 ghijklmn
TX20 0.431 fghijklmn 0.197 klmnop
175104 0.422 ghijklmn 0.413 fghijk
BA59 0.420 hijklmno 0.363 fghijkl
WHV 0.419 hijklmno 0.510 defg
MX3 0.394 ijklmnop 0.449 efghi
46-4 0.389 ijklmnop 0.217 jklmnop
RR22 0.385 ijklmnop 0.387 fghijk
10 0.359 jklmnopq 0.274 ijklmnop
67-1 0.359 jklmnopq 0.537 defg
BC31 0.358 jklmnopq 0.251 ijklmnop
WHK 0.357 jklmnopq 0.249 ijklmnop
2867 0.353 jklmnopq 0.278 ijklmnop
28-1 0.351 jklmnopq 0.246 jklmnop
69-1 0.335 klmnopq 0.283 ijklmno
WH9BR 0.332 klmnopqr 0.291 ijklmno
WH83 0.322 klmnopqr 0.258 ijklmnop
TX18 0.286 Imnopqrs 0.265 ijklmnop
2943 0.283 Imnopqrs 0.300 ijklmno
RR24 0.245 mnopqrs 0.242 jklmnop
SGS35 0.242 mnopqrs 0.205 klmnop
J6 0.239 mnopqrs 0.260 ijklmnop
62-1 0.234 mnopqrs 0.167 Imnop
70-12 0.228 mnopqrs 0.183 Imnop
49-1 0.222 mnopqrs 0.159 mnop
Isolate 1st Experiment 2nd Experiment
J2 0.211 mnopqrs 0.249 ijklmnop
RR4 0.206 mnopqrs 0.166 Imnop
LJ37 0.188 nopqrs 0.129 op
65-2 0.186 nopqrs 0.192 klmnop
49-2 0.171 opqrs 0.222 jklmnop
114 0.170 opqrs 0.137 nop
279 0.168 pqrs 0.121 op
400 0.162 pqrs 0.131 op
34 0.153 pqrs 0.125 op
TX2 0.115 qrs 0.179 Imnop
SGS32 0.084 rs 0.118 op
TX15 0.062s 0.109 op
16-1 0.055s 0.11l op
18-2 0.054s 0.105 op
17599 0.054s 0.081 p
z Values are the means of three replicates of each isolate. Means followed by the same
letter in each column do not differ according to Tukey's Honestly Significant Difference
P 1.6 C
0 -- <^ I --- --- --I --- ^ = = -
240 280 320 360 400 440 480 520 560 600
Figure 3-3. Ultraviolet spectra of crude extracts from the yellow pigment-producer,
isolate WHK (A), and the reddish-purple pigment-producer, isolate BA57 (B), compared
to the standards cercosporin (C) and beticolin-1 (D) in ethyl acetate.
Based on the presence of yellow metabolites in some cultures of isolates of C.
piaropi/C.rodmanii, it was hypothesized that the yellow color could be related to the
toxins beticolins, formerly called the Cercospora beticola toxin (CBT). However, the
analyses based on thin-layer chromatograms, resolved under long-UV light and having a
standard beticolin-l, did not show the presence of beticolins in the samples, but only
cercosporin with an Rf value of 0.61 (e.g., isolate WHK in the Figure 3-4 A). According
to Milat and Blein (1995), the six beticolin toxins would be expected to have R values
below that of cercosporin, as shown by the standard for beticolin-1, which had an R f
value of 0.46. The Figure 3-4 B shows the thin-layer chromatogram of a crude extract
from an isolate that produced a typical reddish-purple pigment having a band
corresponding to cercosporin.
The virulence of the isolates was positively correlated with their ability to produce
cercosporin, which ranged from 72 to 89% (P<0.0001; Table 3-5) and negatively
correlated with mycelial growth rate, which ranged from -55 to -61% (P<0.0001).
Cercosporin production and mycelial growth were negatively correlated (-50 to -62%;
P<0.003) as was the correlation between mycelial growth and virulence (-55 to -0.66%;
Fatty Acid Analysis
The isolates of C. piaropi/C.rodmanii analyzed contained the fatty acids palmitic
acid (16:0), oleic acid (18:1w9c), stearic acid (18:0), and the unresolved mixtures, named
sum feature 4 (16: lo7cl5iso20H) and sum feature 6 (18:2co6,9c/18:0 anteiso). In
addition, two of the outgroup species also contained the fatty acid myristic acid (C14:0).
The mean fatty acid composition with standard deviation for each isolate from three
1 2 3 4 1 2 3
Figure 3-4. Crude extracts of isolates of Cercospora piaropilC. rodmanii resolved on thin
layer chromatograms under long-ultraviolet light. A, lanes 1 to 4, standard cercosporin,
standard beticolin-1, nonpigmented isolate 17599, and the typical yellow pigment-
producer isolate WHK; B, lanes 1 to 3, standard cercosporin, standard beticolin-1, and
the typical purple pigment-producer isolate BA57.
Table 3-5. Correlation matrix among some physiological traits of 55 isolates of Cercospora piaropi/C. rodmanii from several
Virulence Virulence Virulence Cercosporin Cercosporin Mycelial Mycelial
production production growth growth
1st 2nd 3rd t 2nd 1st 2nd
experiment experiment experiment experiment experiment experiment experiment
2nd experiment (0.0001)
Virulence 0.92 0.91
3" experiment (0.0001) (0.0001)
Cercosporin production 0.78 0.72 0.89
1st experiment (0.0001) (0.0001) (0.0001)
Cercosporin production 0.80 0.73 0.85 0.87
2"d experiment (0.0001) (0.0001) (0.0001) (0.0001)
Mycelial growth -0.60 -0.61 -0.66 -0.62 -0.54
1i" experiment (0.0001) (0.0001) (0.0001) (0.0001) (0.0001)
Mycelial growth -0.55 -0.55 -0.61 -0.57 -0.50 0.81
2 experiment (0.0001) (0.0001) (0.0001) (0.0001) (0.0003) (0.0001)
different culture ages are presented in the Tables 3-6, 3-7, and 3-8. The isolates of C.
piaropilC. rodmanii did not differ in the number and kind of fatty acids present but
differed in the relative concentration of each type. The relative concentration of an
individual fatty acid ranged from less than 1% of the total fatty acid content to over 50%.
The results of canonical discriminant analysis of the fatty acid profiles showed an
accentuated effect of the age of the cultures in the resolution of taxa (Table 3-9). FAME
profiles from 4-day-old mycelium differentiated more of the isolates of C. piaropi/C.
rodmanii than the profiles from 5- and 6-day-old mycelia. The differentiation of isolates
of C. piaropi/C. rodmanii using 4-day-old mycelia did not relate with geographical origin
of the isolates. In addition, some of the species of Cercospora used as the outgroup did
not differentiate from the isolates of C. piaropi/C. rodmanii. Basically, the isolates of C.
piaropi/C. rodmanii could not be differentiated based on fatty acid profiles from the 5-
and 6-day-old mycelia; with the exception of the isolate 2619 at the 6-day-old mycelium
The FAME profile of the isolate WH9BR, which was collected and identified by
Conway (1976a), who described it as C. rodmanii, was significantly different (P=0.05)
from only one isolate of C. piaropi/C. rodmanii when 6-day-old mycelia were used. It
was not significantly different from any isolates of C. piaropi/C. rodmanii when 5-day-
old mycelia were used. In addition, the FAME profile of isolate WH9BR was not
significantly different from other two outgroup Cercospora species when 5- and 6-day
old mycelia were used.
Table 3-6. Fatty acid composition of Cercospora spp. isolates from 4-day-old mycelia.
Isolate / species % Of total fatty acid content (mean SD)
14:0 Sum feature 4a 16:0 Sum feature 6b 18:1 0)9c 18:0
WH83 0.00 0.00 0.79 0.01 19.36 + 3.11 49.00 + 6.10 24.66 + 1.83 6.06 0.96
WHK 0.00 + 0.00 0.74 + 0.05 24.92 2.03 39.54 2.38 27.41 0.23 7.21 0.88
WH9BR 0.00 + 0.00 0.97 0.13 26.00 5.54 39.58 8.59 26.72 1.98 6.60 1.16
2619 0.00 0.00 1.00 + 0.20 22.77 1.38 45.39 2.43 25.19 + 0.52 5.40 0.41
TX18 0.00 +0.00 0.83 0.08 19.33 + 2.16 47.65 3.08 28.42 0.89 3.78 0.32
TX20 0.00 0.00 1.03 0.12 18.79 1.33 51.92 1.43 24.63 0.99 3.62 0.67
2867 0.00 0.00 1.06 0.22 18.39 2.44 48.07 1.24 27.35 1.40 4.96 + 0.43
MX3 0.00 + 0.00 0.86 0.14 16.29 2.28 53.59 3.43 24.18 1.38 4.60 0.63
62-2 0.00 + 0.00 1.35 + 0.28 23.08 5.90 42.56 + 7.05 26.60 + 0.51 6.02 0.55
10 0.00 0.00 0.94 0.19 20.89 2.86 45.42 4.17 25.64 1.56 6.80 0.84
28-1 0.00 0.00 0.79 0.05 16.46 0.99 53.46 1.48 24.85 + 0.51 4.03 + 0.19
400 0.00 0.00 0.96 0.08 16.97 + 0.46 54.63 + 0.62 23.28 0.34 3.58 0.39
114 0.00 0.00 0.83 + 0.09 15.00 + 0.49 53.51 + 1.29 25.75 1.34 4.38 0.64
WHV 0.00 0.00 1.15 0.11 15.63 0.27 50.61 1.22 29.31 1.05 3.31 0.01
C. beticola 0.24 0.03 0.87 0.11 25.43 + 0.49 35.24 0.36 30.26 0.57 7.71 0.69
C. oenotherae 0.35 0.04 0.97 0.09 25.59 0.26 40.89 + 0.71 25.69 0.51 6.63 0.67
Cercospora sp.c 0.00 0.00 0.56 0.02 25.41 1.73 27.74 2.40 37.61 1.38 8.60 + 0.74
C. gossypina 0.00 0.00 0.61 +0.08 19.24 1.38 51.78 2.62 21.33 1.17 6.91 0.25
a Sum feature 4 is an unresolved mixture of 16:1 ow7c15 iso 20H;
b Sum feature 6 is an unresolved mixture of 18:2 co6, 9c/18:0 anteiso.
c From Jasminum sp.
Table 3-7. Fatty acid composition of Cercospora spp. isolates from 5-day-old mycelia.
Isolate / species % Of total fatty acid content (mean SD)
14:0 Sum feature 4a 16:0 Sum feature 6b 18:1 o9c 18:0
WH83 0.00 + 0.00 0.80 0.03 18.84 1.44 49.03 + 2.36 25.84 2.26 5.18 2.03
WHK 0.00 0.00 0.83 0.10 23.72 0.73 41.06 1.81 27.57 0.61 6.83 0.48
WH9BR 0.00 0.00 1.01 0.12 22.16 3.38 43.93 6.38 26.76 2.30 6.15 0.81
2619 0.00 0.00 1.14 0.10 23.28 2.41 43.71 3.61 26.28 1.27 5.48 0.29
TX18 0.00 0.00 0.79 0.05 19.00 0.68 48.41 1.73 28.17 1.02 3.49 0.29
TX20 0.00 0.00 0.98 0.06 18.58 0.18 52.78 0.38 24.17 0.43 3.49 0.21
2867 0.00 0.00 1.04 0.10 19.11 2.56 48.47 + 3.48 26.75 0.06 4.64 0.87
MX3 0.00 t 0.00 0.90 0.03 18.81 3.39 48.85 7.79 25.58 3.14 5.65 1.65
62-2 0.00 0.00 1.19 0.10 20.67 0.95 44.65 :0.50 27.16 0.57 5.73 0.32
10 0.00 +0.00 1.08 0.38 20.23 1.90 44.71 +t2.97 26.95 +0.49 6.52 0.38
28-1 0.00 t0.00 0.85 +0.04 15.73 + 0.64 54.41 + 1.58 24.84 0.84 3.95 0.12
400 0.00 + 0.00 1.04 0.08 16.16 0.83 52.81 2.36 23.24 1.27 3.54 0.21
114 0.00 0.00 0.88 0.12 16.34 1.41 53.22 1.85 24.98 0.59 4.21 0.52
WHV 0.00 0.00 1.18 0.08 15.29 0.42 50.77 0.69 29.55 1.13 3.22 0.10
C. beticola 0.24 0.02 0.86 + 0.01 26.10 + 0.95 34.10 + 0.79 30.40 0.57 8.13 1.13
C. oenotherae 0.36 +0.09 1.07 +0.07 25.36 7.37 39.11 i 12.63 28.80 4.10 5.13 1.30
Cercospora sp.C 0.00 0.00 0.94 0.45 25.69 2.33 32.06 6.85 34.36 6.22 7.14 i 1.23
C. gossypina 0.00 +0.00 0.58 0.03 20.71 1.03 48.06 2.55 23.13 1.04 7.19 0.20
a Sum feature 4 is an unresolved mixture of 16:1 w7c 15 iso 20H;
b Sum feature 6 is an unresolved mixture of 18:2 06, 9c/18:0 anteiso.
c From Jasminum sp.
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EVXC7G0XL_UZQYBM INGEST_TIME 2013-02-14T18:04:59Z PACKAGE AA00012989_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC