Department of Plant and Microbial Biology,
University of California at Berkeley,
Berkeley,1 and Roche Molecular
Systems, Alameda,2 California
Received 3 August 1999/Returned for modification 14 October
1999/Accepted 2 November 1999
For many pathogenic microbes that utilize mainly asexual modes of
reproduction, it is unknown whether epidemics are due to either the
emergence of pathogenic clones or environmentally determined increases
in the population size of the organism. Descriptions of the genetic
structures of epidemic populations, in conjunction with analyses of key
environmental variables, are able to distinguish between these
competing hypotheses. A major epidemic of coccidioidomycosis (etiologic
agent, Coccidioides immitis) occurred between 1991 and 1994 in central California, representing an 11-fold increase above the mean
number of cases reported from 1955 to 1990. Molecular analyses showed
extensive genetic diversity, a lack of linkage disequilibria, and
little phylogenetic structure, demonstrating that a newly pathogenic
strain was not responsible for the observed epidemic. Epidemiological
analyses showed that morbidity caused by C. immitis was
best explained by the interaction between two variables, the lengths of
droughts preceding epidemics and the amounts of rainfall. This shows
that the principal factors governing this epidemic of C. immitis are environmental and not genetic. An important
implication of this result is that the periodicity of cyclical
environmental factors regulates the population size of C. immitis and is instrumental in determining the size of epidemics. This knowledge provides an important tool for predicting outbreaks of
this pathogen, as well as a general framework that may be applied to
determine the causes of epidemics of other fungal diseases.
 |
INTRODUCTION |
The microbes of interest in this
report, pathogenic fungi, cause severe disease epidemics in both
animals and plants, illustrated by recently reported mass mortalities
in sea fan coral and amphibian populations (3, 19). However,
it is usually not known why fungal epidemics occur. Epidemics of many
species of microbes have their origin in the differential success of
highly fit individuals (e.g., Staphylococcus aureus
[14], Mycobacterium tuberculosis [15], Vibrio cholerae
[22], and Phytophthora sp.
[5]). Within phytopathogenic fungi, strong selection
by host resistance alleles (20) causes differential success
between fungal clones and this selection has been shown to result in
epidemics as highly fit clones emerge (7). Due to the low
rates of sexual recombination in microbe populations, such epidemics
can be identified by the characteristically high levels of nonrandom
association between alleles (genetic linkage disequilibrium
[26]). However, epidemics have other causes and may
arise due to the occurrence of favorable environmental conditions
releasing a pathogen from environmental constraints. This is typified
by the epidemicity of malaria in Central America resulting from a
cyclicity of environmental variables (4). Under these
conditions, the genetic structures of epidemic populations are not
expected to be composed of highly fit clones that have arisen due to
selection (26). To demonstrate the principal cause
underlying any epidemic of infectious disease, it is necessary to
distinguish between these environmental and genetic hypotheses. This
task may be done by comparing the population genetic structures of
epidemic pathogen isolates against predictive statistical models that
incorporate known or potentially important environmental variables. We
demonstrate this approach by using combined molecular and
epidemiological analyses to describe the factors that have resulted in
an unexplained epidemic of the pathogenic fungus Coccidioides immitis (12).
C. immitis is endemic to semiarid soils of the New World.
The fungus is dimorphic, with the parasitic phase causing a systemic infection in humans and other vertebrate hosts once arthrospores have
been inhaled (32). Molecular analysis has shown that
C. immitis consists of two genetically isolated cryptic
species, the California and non-California species, that have been
reproductively isolated from one another for an estimated 11 million
years (23, 24). These species occur with largely
nonoverlapping geographical distributions (23, 24), and
although no sexual state has ever been described for C. immitis, the non-California species has been shown to be
recombining in nature (8). Clinical isolates of C. immitis from both geographical areas have been shown to exhibit a
wide range of degrees of virulence for mice, demonstrating the
existence of virulence-determining factors (18).
Between 1991 and 1994, California experienced an epidemic of
coccidioidomycosis when the numbers of reported cases increased from a
1950 to 1990 average of 
400 cases per year to more than 4,500 cases
per year (2) (Fig. 1). More
than 80% of these cases occurred in one location, Bakersfield, Calif.
While previous epidemics of C. immitis had been observed in
the previous 40 years, these were attributable to dust storms and
earthquakes causing arthroconidia to become wind-borne (17, 30,
33). However, the 1991-1994 epidemic was unusual in that it was
an order of magnitude larger than those previously seen, was localized
in distribution, and persisted for 3 years (Fig. 1), compared to the
previous maximum of 16 weeks for an outbreak in 1978 (17).
Epidemiological analyses of the 1991-1994 epidemic found no
contributing factors among environmental or sociological variables
(annual rainfall, particulate matter, wind speed, occupation) and only
the expected biases of the risk factors ethnicity, sex, and diabetes
(2). On the other hand, recent molecular data had shown no
genetic variation in 25 isolates collected during the epidemic using 12 loci that were polymorphic in a Tucson population of C. immitis (10). This lack of variation was consistent
with the hypothesis that the 1991-1994 epidemic was caused by the
evolution and expansion of a single virulent C. immitis
clone. However, as argued by Burt et al. (10), an
alternative explanation for the lack of polymorphism seen in
Bakersfield at these loci is that genetic drift, occurring over the 12 million years since speciation between California and non-California
C. immitis, fixed alleles that were originally polymorphic
in the ancestral population (36). Thus, markers that are
polymorphic in non-California C. immitis are not polymorphic in California C. immitis and vice versa, giving rise to the
observed clonality in the epidemic population. Because the markers
found to be polymorphic in the non-California species are monomorphic in the California species, they cannot be used to address the questions
of clonality and recombination.
View larger version (29K):
[in this window]
[in a new window]
|
FIG. 1.
Annual number of cases of coccidioidomycosis reported
from Bakersfield (solid line) and annual rainfall (dotted line). El
Niño years are marked with an asterisk. These data were supplied
by the Division of Communicable Disease Control, California Department
of Health Services.
|
|
Therefore, to further investigate the causes behind this epidemic, we
found additional genetic markers that were polymorphic in the
California species of C. immitis and used them to reanalyze the original 25 Bakersfield isolates and 12 more clinical isolates collected from Bakersfield during the epidemic. Also included were two
representative outgroup non-California isolates from Arizona and
Guatemala. Further, we collected an expanded data set of environmental
variables and used retrospective statistical analyses to ascertain if
these factors are correlated with the epidemicity of C. immitis.
| |
MATERIALS AND METHODS |
C. immitis isolates.
C. immitis isolates
were cultured from patients suffering from respiratory and disseminated
coccidioidomycosis during the 1993-1994 phase of the epidemic by the
Kern County Public Health Laboratory. Isolates were cultured in liquid
medium, and genomic DNA was extracted as described previously
(9).
Isolation of polymorphic genetic loci.
Two classes of
genetic markers were used. Firstly, slowly evolving single-nucleotide
polymorphisms (SNPs; mutation rate, 10
9) were isolated
to provide data on the genetic diversity and reproductive mode of
C. immitis isolated from the epidemic. Secondly, rapidly evolving multiallelic short tandem repeats (STRs; mutation rate, 102 to 105) were used to increase the
statistical power for differentiation between isolates that were
identical at all SNP loci.
To search for SNPs, seven tester strains were randomly chosen from the
37 Bakersfield C. immitis isolates and genomic DNA was
amplified by PCR using arbitrary primers and low-stringency conditions.
Amplified bands were used as template DNA in a new PCR, except that 0.1 µl of [
-35S]thio-dATP (12.5 mCi ml1; 1 mCi = 37 MBq) was added and the amplicons were electrophoresed on
an MDE gel (AT Biochem, Malvern, Pa.) to reveal single-strand conformational polymorphisms (SSCPs) (8). Polymorphic
amplicons were sequenced to determine the genetic basis of the
variation, and specific primers were designed to amplify the locus. We
also searched for polymorphisms in C. immitis sequences from
the GenBank database for (i) CTS1, (ii) pyrG
(OR), (iii) ITS, (iv) BL, and (v)
SP and unpublished sequences of HSP60
(HSP) and BGL2 provided by G. Cole (Medical
College of Ohio, Toledo). Restriction endonuclease assays were designed
to score each polymorphic site. If no suitable restriction site
existed, then SSCP was used to score the polymorphism. The primer
sequences used have been published previously (16).
All isolates were genotyped at two loci containing polymorphic STRs,
locus 621r, containing an (AC)n
(n = 6 to 18) microsatellite, and locus B34,
containing a (TAA ACA AAC)n (n = 1 to 6) minisatellite (16). The sequence for locus
621r was provided by D. Carter, amplified with TAMRA-labeled
primers, and genotyped using an automated sequencer and GENESCAN
software (Applied Biosystems). Locus B34 was found as
described above and genotyped by analysis of SSCPs.
Genetic analyses.
If the epidemic was due to the spread of
an epidemic clone of C. immitis, then all loci must share a
recent common evolutionary history due to their inheritance as a single
linkage group. This clonal mode of reproduction is expected to result
in (i) low levels of genetic diversity and (ii) strong nonrandom
associations between alleles at different loci, based on the assumption
that extensive genetic recombination has not occurred between the
epidemic clone and nonepidemic individuals of C. immitis.
Due to the short time scale of the epidemic, this assumption is
probably warranted. In all subsequent statistical analyses, isolates
from outside California (1036 and 3272) were excluded due to their
outgroup status.
(i) Population genetic analyses.
The probability of sampling
a particular genotype more than once in this data set may be calculated
using the binomial expression
where G is the number of genotyped C. immitis isolates, n is the number of isolates with the
same genotype as that in question (and equals 1 for this data set), and
P is the probability of observing the original genotype
(P = 
fr, where
fr is the frequency of each allele found at a
locus). This method assumes that (i) different genotypes arise by
recombination and not mutation, (ii) mating is random, and (iii) loci
are at linkage equilibrium. Demonstration of low linkage disequilibria
in the data set validates assumptions ii and iii, and low sequence
diversity between loci suggests that mutation rates in C. immitis are not unusually high. The disequilibrium coefficients
for alleles A and B at two loci were calculated
as DAB = pAB 
pApB, where pAB is
the gametic frequency of AB and pA
and pB are the respective allele frequencies.
The disequilibrium coefficients for 78 pairs of loci were calculated,
and Fisher's exact test was used to assess the significance of the
association (37). The index of association (IA) measures the
degree of association among all loci based on the variation of the
genetic distance between individuals (6, 26). The IA was
calculated, and its significance was assessed by comparison with values
calculated from 1,000 artificially recombined data sets (8).
(ii) Phylogenetic analyses.
Alleles at each locus were
treated as phylogenetic characters with two character states. Maximum
parsimony was used to find the shortest tree that fitted the data using
the program PAUP 4.062a and the strength of branches was examined by
bootstrapping the data 1,000 times. In order to test whether the
observed data set contained more phylogenetic signal than a population
undergoing complete recombination, 500 artificially recombined data
sets were created and the lengths of their most parsimonious trees were
compared to those found for the observed data set (1). As
alleles are subject to reversals at STR loci (29),
621 and B34 were only used to measure genotypic
diversity and were not used in the genetic analysis unless specified.
Epidemiological analyses.
C. immitis morbidity data
for 1955 to 1995 in Kern County were supplied by the California
Department of Health Services, Sacramento. To account for changes in
population size over time and to highlight variation between years, the
data were transformed by calculating the relative change in C. immitis morbidity (RCM) as the numbers of cases in year
n divided by the numbers of cases in year n 1.
Generalized linear models were constructed with RCM as the dependent
variable and the following as independent variables: (i) mean annual
rainfall, (ii) the occurrence of type I El Niño southern
oscillation (ENSO) events, (iii) the annual Palmer drought severity
index (PDSI), (iv) the length of moderate to extreme droughts preceding
any particular year, (v) mean annual temperature, (vi) mean annual
10-µm particulate matter concentration (PM10), (vii) mean annual
total suspended particle concentration (TSP), and (viii) yearly 1970 to
1996 Kern County population size. Type I ENSO events (ii) are known to
be the principal determinants of above-normal rainfall in California
and are defined as an equatorial Pacific sea surface anomaly of
+2.0°C (Fig. 1) (J. Null, http://www.nws.mbay.net/cal_enso.html). Droughts (iii) are defined as having a PDSI of 1 to 2.9 for 9 months or more following National Climatic Data Center guidelines (28). Rainfall and temperature data were obtained from the
Bakersfield Meadows meteorological station, and drought severity
indicators were for the Central Valley region (California Division 5).
These data were collated from the National Climatic Data Center
databases (28). PM10 (vi) and TSP (vii) data were for the
San Joaquin valley air basin for the years 1986 to 1996 and 1965 to
1991, respectively (11).
The fit of each linear regression model was assessed by inspection of
the regression r values with their associated residuals and
tested with the F statistic. Noncontributing variables were rejected in
a stepwise fashion, and interactions between contributing variables
were tested to find the most parsimonious model.
| |
RESULTS |
Genetic analyses.
All loci were biallelic and contained a
single allele within an isolate, confirming the haploid state of
C. immitis. The alleles grouped all isolates from
Bakersfield into a single clade, with the exception of isolates 1036 and 3272, which formed a separate clade with high bootstrap support
(Fig. 2). Where isolates from this study
had been typed using other genetic markers in previous studies
(10), they corresponded exactly to the assignment of C. immitis to one of two species, California or
non-California (23). In this study, all isolates from
Bakersfield were of the California species and isolates 1036 and 3272 were of the non-California species, as expected from previous data
(10, 23).
View larger version (21K):
[in this window]
[in a new window]
|
FIG. 2.
Strict consensus of 46,072 most-parsimonious
phylogenetic trees. Heavy bars signify branches with strong bootstrap
support. Values in parentheses are for analyses with the STR loci
included. Multilocus genotype designations are shown, and C. immitis isolates with identical genotypes are boxed.
|
|
Considerable allelic diversity was observed at both loci containing
STRs, with eight alleles identified for 621r and five alleles for B34 (Table 1).
Multilocus genotypic diversity was high, with a total of 34 unique
multilocus genotypes observed in the sample of 37 California isolates.
Three pairs of isolates had identical genotypes at all loci (Fig. 2).
Of these, two pairs were expected to be observed by chance due to their
sharing of alleles that had a high frequency in the data set (binomial
probability, P > 0.01) but one pair (isolates 2005 and
2267, genotype S) was unlikely to be observed in this data set
(binomial probability, P < 0.001) and could be
considered genetically identical clones. While the extensive genetic
diversity seen in the data set suggests that these isolates are not
descended from a single clonal ancestor, it is necessary to examine the
structure of the variation in order to determine the degree of mixis
within the parental population of C. immitis. Nonrandom
associations were rare, with only 4 of 78 pairs of loci (STR loci
excluded) showing significant linkage disequilibria (P < 0.05), and no statistical associations were observed after
controlling for type I error with a Bonferroni correction
(31). The multilocus IA for all loci was not significantly greater than that expected for a fully recombined population
(P = 0.28, STR loci excluded).
A second approach used to assess the population structure was
phylogenetic analysis. If the epidemic population is clonal in origin,
then the data will fit a short, well-resolved tree with minimal
homoplasy. On the other hand, if the epidemic isolates are derived from
a genetically diverse recombining population, then multiple homoplasies
will result and the tree(s) will be poorly resolved. No single tree
could be fitted to the SNP data; instead, 46,072 trees of 32 steps were
identified, all of which had low consistency indices (0.44). The strict
consensus of these trees, shown in Fig. 2, is poorly resolved, with few
internal branches and most isolates falling into a polycotomy of 31 isolates. Two C. immitis isolates of the non-California
cryptic species (1036 and 3272) were included as outgroup taxa and fell
outside the California clade with strong bootstrap support, as
expected. Randomization of alleles within California between isolates
but within each locus mimics mixis by creating artificially recombined data sets. Comparison of the observed tree length with those found for
500 recombined data sets shows that the observed tree is 16 steps
longer than the shortest possible tree length of 13 (expected under a
model of complete clonal evolution of C. immitis) but significantly shorter than those of the recombined data sets (Fig. 3). This is largely due to the effect of
the three duplicated genotypes A, S, and X. Collapsing these duplicates
into single genotypes (clone correcting [26])
eliminates the effect of resampling of individuals of these genotypes
and renders the observed tree length statistically indistinguishable
from that of the simulated randomly mating populations.
View larger version (23K):
[in this window]
[in a new window]
|
FIG. 3.
Distribution of tree lengths for 500 artificially
recombined data sets using all data (A) and clone-corrected data with
the identical genotypes A, S, and X represented only once (B).
|
|
Epidemiological analyses.
Observed seasonal rainfall and
occurrence of type I ENSO events are shown in Fig. 1. No periodicity is
obvious for the 1955-1996 morbidity data, suggesting that the ENSO
influence on central Californian rainfall is not a primary factor in
determining whether epidemics of C. immitis occur. This lack
of ENSO influence is borne out by analysis of variance of the El
Niño years that occurred between 1955 and 1993, during which
period no association was seen between the eight ENSO events and the
mean RCM (F4,30 = 0.792, P = 0.540).
Multivariate regression excluded from consideration the variables mean
annual rainfall, mean annual temperature, ENSO, PDSI, PM10, TSP, and
population size. However, a significant correlation was observed
between the length of drought and RCM (r = 0.439, P = 0.004). Post hoc analysis identified an outlying data point with
high influence (Studentized residual = 0.64). This data point has
been inflated by the transformation used here and corresponds to an
unusually low incidence of coccidioidomycosis in 1967 rather than an
epidemic in 1968 (Fig. 1). Exclusion of this data point and reanalysis
normalized the residuals and increased the fit of length of drought and
RCM, as well as introducing mean annual rainfall as a significant
variable (multiple r = 0.673; rainfall, P < 0.01; drought, P < 0.001).
The regression equation y = 0.295x1 + 0.001x2 0.16 was used to predict relative
change in coccidioidomycosis in the past time series as a function of
the variables length of drought (x1) and
rainfall (x2). The modeled change in RCM is
shown in Fig. 4 together with that
observed. The regression explains 45% of the variation in the yearly
series and successfully predicts the 1991-1994 epidemic. The model
predictions differ markedly from observations only during 1966 to 1975, where the RCM is underestimated, and 1978. The period 1966 to 1975 is
unusual in that there were a number of wet years in rapid succession
and it is possible that factors others than those included in this
model predominated during this time period. The unexplained peak in
infections in 1978 is due to a singular event, a windstorm that blew
through Kern County on 20 December 1977, exposing a large number of
people and resulting in a spike of infections that were recorded for the subsequent year (30).
View larger version (17K):
[in this window]
[in a new window]
|
FIG. 4.
Relative change in the number of cases of
coccidioidomycosis in Bakersfield from 1956 to 1996 (solid line). The
dotted line shows the modeled change using the regression equation
y = 0.295x1 + 0.001x2 0.16 with the variables length of drought
(x1) and mean annual rainfall
(x2). The asterisk beside the number one
signifies the outlier data point.
|
|
| |
DISCUSSION |
Here, we have shown that a lengthy and persistent epidemic of
coccidioidomycosis is attributable to environmental, and not genetic,
causes. We can rule out the amplification of a virulent clone of
C. immitis as a cause of the epidemic because isolates of
C. immitis collected during the epidemic are genetically
diverse and have undergone extensive interlocus recombination. While it may be argued that the genetic diversity in this population may have
accrued due to our isolation of markers containing hypervariable SNPs,
we can show that this is not the case. Sequencing of the SNPs
demonstrated that the polymorphic sites were rare (>1% of nucleotides
were polymorphic in Bakersfield genomes), and sequencing of multiple
isolates showed that no more than two nucleotides were ever found at a
single polymorphic site. Moreover, genotyping of these loci in
non-California C. immitis showed that all SNP loci had
drifted to fixation (unpublished data), a situation that would not be
expected to have occurred if these loci were hypervariable. Therefore,
it appears that the 1991-1994 epidemic was not caused by the emergence
of a single clonal lineage of C. immitis. Rather, the
epidemic was due to infection by multiple progeny from a parental fungal population that was close to panmixia. It is apparent that (as
among its sister taxon, non-California C. immitis)
recombination is occurring in nature between individuals of the
California type of C. immitis. No sexual reproduction (the
production of ascospores and meiosis) has ever been directly observed
in C. immitis, and it is reasonable to assume that when sex
does occur, it is rare (8). For unlinked loci, the medium
time to genetic equilibrium in a randomly mating population with a
recombination rate (c) of 0.01 is 69 generations and that
for c = 0.001 is about 693 generations (13).
Given that the generation frequency of C. immitis is, at the
most, 1 to 2 per year (J. W. Taylor, personal communication), we
would not expect genetic equilibrium to be established within the time
scale of this epidemic. This rules out the evolution of a virulent
C. immitis clone as a cause of the 1991-1994 epidemic.
However, there is also direct evidence for clonal propagation, and
subsequent infection, by C. immitis within this data set.
Two patients were infected by C. immitis isolates with
identical multilocus genotypes, 2005 and 2267. The patients who
contributed these isolates lived and worked in the same town, but the
C. immitis isolates were collected by separate clinicians.
This is strong evidence that these people had been infected by the
local dispersion of asexually produced C. immitis spores
from a single fungal individual and is the first data demonstrating this occurrence.
Statistical analyses discarded several variables as factors
contributing to the epidemic, including the long-term population increase in Bakersfield, mean annual temperature, and TSP. It has long
been recognized that coccidioidomycosis is a seasonal disease, with
rates of new cases peaking in October to November (21, 34),
the dustiest months of the year. It has also been observed that the
number of cases is sometimes greatest after a heavy winter rainfall
(34). Our analysis found the first statistical evidence to
link the amount of rainfall with the number of cases of
coccidioidomycosis. Unexpectedly, the occurrence of ENSO was not found
to be a significant factor in determining the number of cases of
coccidioidomycosis in this region. Rather, the number of cases was
related to the length of the drought preceding rainfall, showing that
the timing of rainfall, and not simply its amount, is important in
determining infection rates. Most significantly, the drought preceding
the 1991-1994 epidemic was the most sustained seen since 1956, implicating it in the development of this uncharacteristically large
epidemic. How the drought length may affect the growth of C. immitis is a matter of conjecture. It is known that C. immitis is a poor competitor on nutrient media and is easily
overgrown by common soil fungi but is resistant to dessication and high temperatures (35). We can hypothesize that this particular
extended drought suppressed fungal competitors relative to C. immitis to the extent that constraints on the growth of C. immitis were released when the 1992 ENSO increased rainfall in
California. However, this is a purely speculative scenario and further
studies of the natural soil ecology of C. immitis are
necessary to further this argument. We conclude that a fortuitous
conjunction of climatic variables appears to have allowed the epidemic
to occur. Further analyses across the geographic range of C. immitis will demonstrate the generality of this finding and
establish whether the predictive model developed here is of use as an
early warning system for epidemics of this particular fungal pathogen.
The emergence of aggressive fungal pathogens and the consequent
epidemics in their host organisms are currently receiving much
attention. It has been demonstrated that a chytrid,
Batrachochytrium dendrobatidis (25), is
responsible for a pandemic in amphibian populations and that
Aspergillus sydowii is causing mass mortality of sea fan
corals in the Caribbean (19). However, in both cases, it is
not known whether environmental change impacting host fitness or
increases in virulence of the pathogen are to blame (3, 27).
That a recent Phytophthora epidemic on British alder
(Alnus sp.) was due to the emergence of an aggressive hybrid
(5) shows the potential for damaging genetic change in
fungal species. However, our data show equally how environmental
factors have to be accounted for. Here, we have shown the practical
utility of using well-characterized loci and population genetic
analysis in conjunction with multivariate analyses to answer
biologically relevant epidemiological questions. These approaches will
aid in dissecting the factors behind epidemics of fungal populations,
as well as in elucidating fundamental features of microbial natural history.
We thank R. Talbot for supplying the isolates; G. Cole and D. Carter for the information on unpublished sequences; D. Geiser, D. Greene, and S. Kroken for comments and assistance; and J. J. Bull
and F. J. Ayala for comments.
This work was supported by grants to J. W. Taylor from the NIH and
the Novartis Agricultural Discovery Institute, Inc.
| 1.
|
Archie, J. W.
1989.
A randomization test for phylogenetic information in systematic data.
Syst. Zool.
38:239-252[CrossRef].
|
| 2.
|
Arsura, E.,
J. Caldwell,
R. Johnson,
H. Einstein,
G. Welch,
R. Talbot, and H. Affentranger.
1996.
Coccidioidomycosis epidemic of 1991: epidemiologic features, p. 98-107.
In
H. E. Einstein, and A. Catanzaro (ed.), Coccidioidomycosis. National Foundation for Infectious Diseases, Washington, D.C.
|
| 3.
|
Berger, L.,
R. Speare,
P. Daszak, and D. Green.
1998.
Chytridiomycosis causes amphibian mortality associated with population declines in the rain forests of Australia and Central America.
Proc. Natl. Acad. Sci. USA
95:9031-9036[Abstract/Free Full Text].
|
| 4.
|
Bouma, M. J., and C. Dye.
1997.
Cycles of malaria associated with El Nino in Venezuela.
JAMA
278:1772-1774[Abstract].
|
| 5.
|
Brasier, C. M.,
D. E. L. Cooke, and J. M. Duncan.
1999.
Origin of a new Phytophthora pathogen through interspecific hybridization.
Proc. Natl. Acad. Sci. USA
96:5878-5883[Abstract/Free Full Text].
|
| 6.
|
Brown, A. H. D.,
M. W. Feldman, and E. Nevo.
1980.
Multilocus structure of natural populations of Hordeum spontaneum.
Genetics
96:523-536[Abstract/Free Full Text].
|
| 7.
|
Brown, J. K. M., and C. G. Simpson.
1994.
Genetic analysis of DNA fingerprints and virulences in Erysiphe graminis f. sp. hordei.
Curr. Genet.
26:172-178[CrossRef][Medline].
|
| 8.
|
Burt, A.,
D. A. Carter,
G. L. Koenig,
T. J. White, and J. W. Taylor.
1996.
Molecular markers reveal cryptic sex in the human pathogen Coccidioides immitis.
Proc. Natl. Acad. Sci. USA
93:770-773[Abstract/Free Full Text].
|
| 9.
|
Burt, A.,
D. A. Carter,
G. L. Koenig,
T. J. White, and J. W. Taylor.
1995.
A safe method of extracting DNA from Coccidioides immitis.
Fungal Genet. Newsl.
42:23.
|
| 10.
|
Burt, A.,
B. M. Dechairo,
G. L. Koenig,
D. A. Carter,
T. J. White, and J. W. Taylor.
1997.
Molecular markers reveal differentiation among isolates of Coccidioides immitis from California, Arizona and Texas.
Mol. Ecol.
6:781-786[CrossRef][Medline].
|
| 11.
|
California Air Resources Board.
1998.
California Air Resources Board, Air Quality Data Review Section
Sacramento.
|
| 12.
|
Centers for Disease Control.
1994.
Update: coccidioidomycosis California, 1991-1993.
Morbid. Mortal. Weekly Rep.
43:421-423[Medline].
|
| 13.
|
Crow, J. F., and M. Kimura.
1970.
An introduction to population genetics.
Harper & Row, Publishers, New York, N.Y.
|
| 14.
|
De Sousa, A.,
S. Sanches,
M. L. Ferro,
M. J. Vaz, and H. De Lancastre.
1998.
The intercontinental spread of a drug-resistant methicillin-resistant Staphylococcus aureus clone.
J. Clin. Microbiol.
36:2590-2596[Abstract/Free Full Text].
|
| 15.
|
Edlin, B. R.,
J. I. Tokars,
M. H. Grieco,
J. T. Crawford,
J. Williams, and S. D. Holmberg.
1992.
An outbreak of multidrug-resistant tuberculosis among hospitalized patients with the acquired immunodeficiency syndrome.
N. Engl. J. Med.
326:1514-1521[Abstract].
|
| 16.
|
Fisher, M. C.,
G. L. Koenig,
T. J. White, and J. W. Taylor.
1999.
Primers for genotyping single nucleotide polymorphisms and microsatellites in the pathogenic fungus Coccidioides immitis.
Mol. Ecol.
8:1082-1084[CrossRef][Medline].
|
| 17.
|
Flynn, N. M.,
P. D. Hoeprich, and M. M. Kawachi.
1979.
An unusual outbreak of wind-borne coccidioidomycosis.
N. Engl. J. Med.
301:358-361[Medline].
|
| 18.
|
Friedman, L.,
S. E. Smith,
W. G. Roessleb, and R. J. Berman.
1956.
The virulence and infectivity of twenty-seven strains of Coccidioides immitis.
Am. J. Hyg.
64:198-210[Medline].
|
| 19.
|
Geiser, D. M.,
J. W. Taylor,
K. B. Ritchie, and G. W. Smith.
1998.
Cause of sea fan death in the West Indies.
Nature
394:137-138[Medline].
|
| 20.
|
Gentzbittel, L.,
S. Mouzeyar,
S. Badaoui,
E. Mesties,
F. Vear,
D. Tourvieille De Labrouhe, and P. Nicolas.
1998.
Cloning of molecular markers for disease resistance in sunflower, Helanthus annuus L.
Theor. Appl. Genet.
96:519-525[CrossRef].
|
| 21.
|
Jinadu, B. A.
1995.
Valley Fever Task Force report on the control of Coccidioides immitis.
|
| 22.
|
Karaolis, D. K. R.,
J. A. Johnson,
C. C. Bailey,
E. C. Boedeker,
J. B. Kaper, and P. R. Reeves.
1998.
A Vibrio cholerae pathogenicity island associated with epidemic and pandemic strains.
Proc. Natl. Acad. Sci. USA
95:3134-3139[Abstract/Free Full Text].
|
| 23.
|
Koufopanou, V.,
A. Burt, and J. W. Taylor.
1998.
Concordance of gene genealogies reveals reproductive isolation in the pathogenic fungus Coccidioides immitis.
Proc. Natl. Acad. Sci. USA
95:8414[Free Full Text].
|
| 24.
|
Koufopanou, V.,
A. Burt, and J. W. Taylor.
1997.
Concordance of gene genealogies reveals reproductive isolation in the pathogenic fungus Coccidioides immitis.
Proc. Natl. Acad. Sci. USA
94:5478-5482[Abstract/Free Full Text].
|
| 25.
|
Longcore, J. E.,
A. P. Pessier, and D. K. Nichols.
1999.
Batrachochytrium dendrobatidis gen. et sp. nov., a chytrid pathogenic to amphibians.
Mycologia
91:219-227.
|
| 26.
|
Maynard-Smith, J.,
N. H. Smith,
M. O'Rourke, and B. G. Spratt.
1993.
How clonal are bacteria?
Proc. Natl. Acad. Sci. USA
90:4384-4388[Abstract/Free Full Text].
|
| 27.
|
Morell, V.
1999.
Are pathogens felling frogs?
Science
5415:728-731.
|
| 28.
|
National Climatic Data Center.
1998.
Vol. I.
National Climatic Data Center, Asheville, N.C.
|
| 29.
|
Orita, M.,
H. Iwahana,
H. Kanazawa,
K. Hayashi, and T. Sekiya.
1989.
Detection of polymorphisms of human DNA by gel electrophoresis as single-strand conformation polymorphisms.
Proc. Natl. Acad. Sci. USA
86:2766-2770[Abstract/Free Full Text].
|
| 30.
|
Pappagianis, D., and H. Einstein.
1978.
Tempest from Tehachapi takes toll or Coccidioides conveyed aloft and afar.
West. J. Med.
129:527-530[Medline].
|
| 31.
|
Rice, W. R.
1989.
Analyzing tables of statistical tests.
Evolution
43:223-225[CrossRef].
|
| 32.
|
Saubolle, M. A.
1996.
Life cycle and epidemiology of Coccidioides immitis, p. 1-9.
In
H. E. Einstein, and A. Catanzaro (ed.), Coccidioidomycosis. National Foundation for Infectious Diseases, Washington, D.C.
|
| 33.
|
Schneider, E.,
R. A. Hajjeh,
R. A. Spiegel,
R. W. Jibson,
E. L. Harp,
G. A. Marshall,
R. A. Gunn,
M. M. McNeil,
R. W. Pinner,
R. C. Baron,
R. C. Burger,
L. C. Hutwagner,
C. Crump,
L. Kaufman,
S. E. Reef,
G. M. Feldman,
D. Pappagianis, and S. B. Werner.
1997.
A coccidioidomycosis outbreak following the Northridge, California, earthquake.
JAMA
277:904-908[Abstract].
|
| 34.
|
Smith, C. E.,
R. R. Beard,
H. G. Rosenberger, and E. G. Whiting.
1946.
Effect of season and dust control on coccidioidomycosis.
JAMA
132:833-838.
|
| 35.
|
Swatek, F. E., and D. T. Omieczynski.
1967.
Coccidioidomycosis.
The University of Arizona Press, Tucson.
|
| 36.
|
Taylor, J. W.,
D. J. Jacobson, and M. C. Fisher.
1999.
The evolution of asexual fungi: reproduction, speciation and classification.
Annu. Rev. Phytopathol.
37:197-246[CrossRef][Medline].
|
| 37.
|
Weir, B. S.
1996.
Genetic data analysis II.
Sinauer Associates, Inc., Publishers, Sunderland, Mass.
|
![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
Top
Abstract
Introduction
