Molecular Ecology (2008) 17, 2552?2561 doi: 10.1111/j.1365-294X.2008.03764.x
? 2008 The Authors
Journal compilation ? 2008 Blackwell Publishing Ltd
Blackwell Publishing Ltd
Do mosquitoes filter the access of Plasmodium 
cytochrome b lineages to an avian host?
ANDREA B.  GAGER,*?? JOS? DEL ROSARIO LOAIZA,?? DONALD C.  DEARBORN? 
and ELDREDGE BERMINGHAM?
*Department of Ecology and Evolutionary Biology, Princeton University, Princeton, NJ 08544, USA, ?Smithsonian Tropical Research 
Institute, Apartado Postal 0843-03092, Balboa, Panama, ?Department of Biology, McGill University, Montreal, Quebec, Canada H3A 
1B1, ?Department of Biology and Program in Animal Behaviour, Bucknell University, Lewisburg, PA 17837, USA
Abstract
Many parasites show fidelity to a set of hosts in ecological time but not evolutionary time
and the determinants of this pattern are poorly understood. Malarial parasites use vertebrate
hosts for the asexual stage of their life cycle but use Dipteran hosts for the sexual stage.
Despite the potential evolutionary importance of Dipteran hosts, little is known of their
role in determining a parasite?s access to vertebrate hosts. Here, we use an avian malarial
system in Panama to explore whether mosquitoes act as an access filter that limits the range
of vertebrate hosts used by particular parasite lineages. We amplified and sequenced
Plasmodium mitochondrial DNA (mtDNA) from Turdus grayi (clay-coloured robin) and
from mosquitoes at the same study site. We trapped and identified to species 123 141 female
mosquitoes and completed polymerase chain reaction (PCR) screening for Plasmodium
parasites in 435 pools of 20 mosquitoes per pool (8700 individuals total) spanning the 11
most common mosquito species. Our primers amplified nine Plasmodium lineages, whose
sequences differed by 1.72%?10.0%. Phylogenetic analyses revealed partial clustering of
lineages that co-occurred in mosquito hosts. However PAN3 and PAN6, the two primary
parasite lineages of T. grayi, exhibited sequence divergence of 8.59% and did not cluster in
the phylogeny. We detected these two lineages exclusively in mosquitoes from different
genera ? PAN3 was found only in Culex (Melanoconion) ocossa, and PAN6 was found only
in Aedeomyia squamipennis. Furthermore, each of these two parasite lineages co-occurred
in mosquitoes with other Plasmodium lineages that were not found in the vertebrate host
T. grayi. Together, this evidence suggests that parasite?mosquito associations do not restrict
the access of parasites to birds but instead may actually facilitate the switching of vertebrate
hosts that occurs over evolutionary time.
Keywords: access filter, avian malaria, host shift, host-parasite coevolution, life cycle, parasite
diversity
Received 2 December 2007; revision received 8 March 2008; accepted 13 March 2008
Introduction
Epidemiological disasters often occur when an infectious
organism invades novel, immunologically naive host species
(van Riper III et al. 1986; Lawson 1995; Hofmeyr et al. 2000).
Such shifts occur when the factors that determine host
range are altered. Thus, in order to understand the
circumstances that lead to host range shifts we must first
uncover what limits parasite-host range in the present
time. The host range of a parasite can be defined most
simply as the number of host species in which a parasite
can successfully produce infective progeny (Matthews
1998). Although it is widely held that host ranges are kept
small by coevolutionary arms races between host and
parasite (Fry 1990; Lajeunesse & Forbes 2002), there are
persistent generalist parasites (Ricklefs & Fallon 2002;
Szymanski & Lovette 2005) and there is abundant
Correspondence: Andrea B. Gager, Fax: +64-9-373-7668; E-mail:
abgager@gmail.com 
?Present address: School of Biological Sciences, University of
Auckland, 3 A Symonds Street, PB 92019, Auckland 1142, New Zealand.
P L A S M O D I U M  L I N E A G E S  I N  M O S Q U I TO  A N D  AV I A N  H O S T S 2553
? 2008 The Authors
Journal compilation ? 2008 Blackwell Publishing Ltd
documentation of parasites shifting to new hosts (Escalante
et al. 1995; Waldenstrom et al. 2002; Ricklefs et al. 2004).
Therefore, coevolution is insufficient to explain the wide
variation in parasite host range size observed in nature.
The coarsest determinants of host range have been
proposed as two filters (Combes 2001). The host range of a
parasite is made up of species that are (i) encountered by
the parasite; and (ii) compatible with the parasite (i.e. low
immune response, high resource value). In this paper, we
explore the role of mosquitoes as a potential encounter
filter that determines the access of avian malarial
parasites (Plasmodium spp.) to a neotropical avian host, the
clay-coloured robin Turdus grayi.
Recent studies of the ecology of avian malarial parasites
have used DNA sequencing of the cytochrome b gene to
identify parasites in blood samples of hundreds of bird
species from both new and old world (Bensch et al. 2000;
Fallon et al. 2003, 2004; Beadell et al. 2004;  Szymanski &
Lovette 2005; Hellgren et al. 2007; Ishtiaq et al. 2007).
Although there is no consensus about how to define species
with such data, these studies have revealed an enormous
diversity of parasites perhaps on the same order as the
number of bird species. A wide continuum of host ranges
exists in this group of parasites and this trait has no
phylogenetic pattern, i.e. specialist and generalist parasites
may be close relatives (Ricklefs & Fallon 2002). Furthermore,
comparison of host and parasite phylogenies indicates
host-switching has been common in the history of the
group of parasites: closely related parasites may be found
in distantly related host species, and distantly related
parasites can share a single host species. Thus, a paradox
exists because parasites show fidelity to a group of hosts in
ecological time but almost none in evolutionary time
(Ricklefs et al. 2004). This begs the question of what factors
restrict a parasite to its vertebrate-host range in ecological
time when, in this group of parasites, host switching has
been relatively common in history and the physiological
capability to expand host range clearly exists.
Mosquitoes have the potential to act as an external factor
that determines the access of malarial parasites to a pool of
vertebrate hosts in ecological time but could change readily
over evolutionary time, resulting in the changes seen in
phylogenetic analyses of vertebrate-host use. Malarial
parasites have a two-stage life cycle, with sexual reproduction
occurring in a mosquito host and asexual reproduction
occurring in a vertebrate host. Mosquitoes are thus critical
components of this system for two reasons (i) because
parasite sexual reproduction occurs in the mosquito, this
life stage is integral to the reproductive isolation of parasite
lineages; and (ii) behavioural and ecological aspects of
mosquitoes (e.g. seasonality, geographical range, circadian
patterns of behaviour) can determine the vertebrate hosts
to which the parasites are exposed. Yet most recent studies
of associations between avian malarial parasites and their
vertebrate hosts have neglected the role of mosquitoes,
primarily because of the logistical difficulties of assessing
parasites? use of mosquito hosts. In this study, we explore
the prevalence, diversity, and phylogeny of avian malarial
parasites in common neotropical mosquito species and
in a neotropical avian host, T. grayi, to test whether the
vertebrate-host range of parasites is filtered by the ecology
and behaviour of their invertebrate host.
Turdus grayi is a common tropical resident bird species
in Central America. The population at our study site at the
Smithsonian Tropical Research Institute in Gamboa, Panama,
is regularly infected by two parasite lineages in the genus
Plasmodium, PAN3 and PAN6 (unpublished data). Other
avian species at this site are also infected with Plasmodium
spp., though those patterns of host association are not well
described to date. Using PCR and DNA sequencing of
parasite cytochrome b sequences, we examine the distribution
and phylogenetic relationships of Plasmodium lineages
(including PAN3 and PAN6) in the 11 most common
mosquito species in Gamboa. If PAN3 and PAN6 co-occur
in a particular mosquito species, then this ?access filter?
could explain their co-occurrence in T. grayi. If PAN3 and
PAN6 occur in different mosquito species, the access filter
is less likely to explain their co-occurrence in an avian
host. The strongest evidence against a mosquito-based
access filter would be if particular mosquitoes carry
multiple Plasmodium lineages with only a subset of those
lineages occurring in a particular avian host species, such
as T. grayi.
Materials and methods
Study system
This study was conducted at the Gamboa Field Research
Station of the Smithsonian Tropical Research Institute
in central Panama. The habitat of the station includes
grasslands and secondary humid tropical forest bordering
the Panama Canal. Turdus grayi, the clay-coloured robin, is
a year-round tropical resident and one of the most common
species in the study area (Ridgely & Gwynne 1989).
The genus Plasmodium (Apicomplexa: Haemosporidia)
is paraphyletic and includes parasites found in lizards and
birds, which together form a sister group to the genus
Haemoproteus (Bensch et al. 2000; Perkins & Schall 2002;
Ricklefs & Fallon 2002; Beadell et al. 2004). The genus
Plasmodium also includes the more distantly related group
of parasites found in mammals, including those that cause
malaria in humans (Escalante et al. 1995,  1998; Rathore
et al. 2001). The development of techniques for the molecular
identification of Haemosporidian parasites (including
the genera Plasmodium, Haemoproteus, Leucocytozoon and
Hepatocystis) has revealed a species diversity which was
previously obscured by morphological similarities between
2554 A .  B .  G A G E R  E T  A L .
? 2008 The Authors
Journal compilation ? 2008 Blackwell Publishing Ltd
species (Bensch et al. 2000; Ricklefs & Fallon 2002; Beadell
et al. 2004; Ishtiaq et al. 2007), though there remains the
difficulty of reconciling the historical species concepts. All
malarial parasites have a two-stage life cycle; the asexual
stage occurs in a vertebrate host and the sexual stage in an
invertebrate host in the order Diptera (flies, mosquitoes
and relatives). Mosquitoes in the genus Culex are thought
to be the most common hosts for avian Plasmodium parasites
(Valki?nas 2004) although species in the genera Aedeomyia
(Crewe 1976; Gabaldon et al. 1981), Aedes (Eyles 1951)
and others have also been identified either in natural or
experimental studies.
Mosquito samples
Our goal was to sample host-seeking female mosquitoes
that thus would have the potential to transmit Plasmodium
to avian hosts. We collected mosquitoes over an eight-month
period from August 2003 to March 2004, using eight Center
for Disease Control (CDC) incandescent light traps, two
CDC ultraviolet light traps, and two carbon dioxide
arbovirus traps; the use of multiple trap types helps broaden
the sampling of a mosquito community. Traps were set out
bimonthly for 12 h each time (6 p.m?6 a.m) for a total of
more than 2300 trap-hours. Following each 12-h trapping
period, collection bags were removed from traps and placed
in a freezer within two hours of trap closing. Individual
mosquitoes were then sorted by sex and the females were
sorted to species based on morphology. To avoid sequencing
Plasmodium DNA from parasites picked up in blood meals
but unable to develop (i.e. parasites incompatible with that
particular mosquito host), visibly gravid females and
mosquitoes with visible undigested blood meals were
removed from the study. Individual mosquitoes were then
put in pools of 20 individuals per cryotube and frozen until
DNA extraction. Dr James Pecor of the US Army Walter
Reed Biosystematics Unit confirmed species identification.
Mounted specimen collections from this project have been
deposited in the Natural History Museum at the University
of Costa Rica and in the Entomology Department at the
University of Panama.
DNA extraction and sequencing
DNA was extracted from pools of 20 mosquitoes each,
following the procedure of Oskam et al. (1996). The proced-
ure was modified by using a macerating centrifuge with
ceramic beads for 45 s instead of grinding the mosquitoes
with mortar and pestle. Mosquitoes were macerated in
800 ?L fly lysis buffer (50 mm NaCl, 10 mm EDTA pH 8.0,
50 mm Tris-HCl pH 7.4, 1% Triton X-100 and 10 mm DTT)
followed by 3 h of incubation at 60 ?C with 200 ?g/mL
proteinase K. The samples were then put through four
freeze-thaw cycles (30 s at ?80 ?C in liquid nitrogen and 1
minute at 65 ?C) to ensure the rupture of all membranes.
Samples were centrifuged for 15 min at 12 000 ? g and the
supernatant was eluted and stored at ?20 ?C for PCR. The
supernatant was thawed and diluted 1:5 in sterile, distilled
water just prior to use in PCR.
We amplified a 360-bp section of the Plasmodium cyto-
chrome b gene using the primer pairs (621) 5?-AAAAATACCY
TTCTATCCAARTC-3? and (983) 5?-CAYCCAATCCATAA
TAAAGCAT-3? modified from Richard et al. (2002) to be
ambiguous at a total of three bases. This primer pair has
been shown to amplify a wide range of cytochrome b
sequences from old world and new world birds in the
genera Plasmodium and Haemoproteus (Richard et al. 2002;
Ricklefs & Fallon 2002). Amplifications were carried out in
50 ?L reactions containing 0.2 mm of each dNTP, 0.4 ?m of
each primer, 0.1 mm MgCl, 5.0 ?L Qiagen PCR 10? buffer,
1.5 units of Taq polymerase (Qiagen) and 5 ?L of diluted
DNA. PCR was initiated by a 3-minute denaturing step
followed by 35 cycles of 94 ?C for 30 s, 50 ?C for 40 s and
72 ?C for 1 minute with a final extension step at 72 ?C for
3 min. PCR products were visualized on a 2% agarose gel.
Products that were faint were re-amplified by taking a
small plug from the agarose gel, re-suspending the DNA
in 100 ?L distilled, sterile water at 65 ?C and using 5 ?L of
this suspension in a second amplification; there were no
apparent differences in parasite identity or diversity between
samples that amplified strongly in the first PCR vs. those
that required re-amplification. PCR products were cleaned in
2% agarose gels and sequenced directly on an MJ BaseStation
Automatic sequencer (using the same forward and reverse
primers used for amplification). For validation, all PCR
products were sequenced in both directions. Sequences
were edited and aligned in the program sequencher 4.1
(Gene Codes Corporation).
Phylogenetic analyses
Phylogenetic analyses of amplified cytochrome b sequences
were carried out using maximum likelihood (ML) and
neighbour joining (NJ) methods with the paup 4.0b10
software package (Swofford 2002). For the ML analysis we
used modeltest 3.5 (Posada & Crandall 1998; using the
script of B. Weir, http://www.rhizobia.co.nz/phylogenetics/
modeltest.html) to compare the likelihood scores of trees
(hierarchical log-likelihood tests) from 56 different models
of nucleotide substitution. We determined that the best
model, given the sequence data, included a six-step transition
rate matrix and nucleotide base frequencies calculated
from the data, a gamma shape parameter of 0.8874 and a
proportion of invariable sites of 0.5984. This corresponds to
a generalized time-reversible model of evolution with
gamma-distributed and invariant sites (GTR + G + I).
The tree generated with the above parameters had only
a slightly higher log-likelihood score and an identical
P L A S M O D I U M  L I N E A G E S  I N  M O S Q U I TO  A N D  AV I A N  H O S T S 2555
? 2008 The Authors
Journal compilation ? 2008 Blackwell Publishing Ltd
topology to a model without invariable sites (GTR + G).
The GTR  is the most complex model of nucleotide evolution
in the hierarchical analysis and allows for a unique value
for each nucleotide transition rate. This level of complexity
is warranted in this case because of the extreme AT bias in
Plasmodium mitochondrial genes (Bahl et al. 2003), and
GTR models have been used previously in phylogenetic
analysis of malarial parasites (Perkins & Schall 2002). NJ
analysis was carried out with distance measures from
the ML parameters determined by modeltest 3.5 as
described above. All parasites have been designated as
Plasmodium by phylogenetic affinity to published avian
Plasmodium cytochrome b sequences. Phylogenetic analysis
was performed on the nine sequences amplified from
screened mosquitoes in this study along with two previously
published cytochrome b sequences of parasite species with
known mosquito host species. These additional sequences
were from P. relictum [National Centre For Biotechnology
Information (NCBI) Accession No. AF069611] and P. elongatum
(NCBI Accession No. AY099032) both of which are transmitted
by Culex species in the subgenus Culex (Work et al. 1990;
McConkey et al. 1996; Fonseca et al. 1998; Nayar et al. 1998).
We plotted the number of novel parasite sequences identified
against the number of individual mosquitoes screened in
order to determine which mosquito species was associated
with the greater number of parasite lineages and whether
the cumulative number of parasite lineages associated with
each mosquito species appeared to be approaching an
asymptote.
Results
Identification of Plasmodium cytochrome b lineages in 
pools of mosquitoes
During the eight-month study period, we trapped a total
of 128 713 female mosquitoes, of which 123 141 (95.7%)
were identified to species. The remaining 5572 individuals
could be identified only to subgenus (n = 5489 Culex
(Melanoconion) sp., n = 78 Culex (Culex) sp., n = 5 Haemagogus
spp.). Of 30 unambiguous species in the sample, 11
species comprised 97.2% of the fully identified individuals
(Fig. 1). Using PCR with parasite-specific primers for
cytochrome b, we screened a total of 435 pools of 20
mosquitoes each (8700 individuals total) from the 11
most common species collected. Twenty of the 435 pools
screened were positive for parasite presence: 11 from
Aedeomyia squamipennis and nine from Culex (Melanoconion)
ocossa. A higher percentage of pools of Ad. squamipennis
(11/39 = 28.2%) than Cx. (M.) ocossa (9/76 = 11.8%) were
PCR-positive for parasite presence (?2 = 4.51, d.f. = 1,
P = 0.034). Ad. squamipennis has previously been reported
as a host species for avian Plasmodium (Gabaldon et al.
1981). This is a first report of Cx. (M.) ocossa as a mosquito
host for avian Plasmodium. The remaining nine species
screened had no positive pools using this primer pair
(Table 1). These included Culex nigripalpus, a species
previously reported as host for avian Plasmodium in other
parts of its range (Nayar et al. 1998).
Fig. 1 The total number of female mosquitoes from the 11 most common species caught in traps from August 2003 through September 2004.
Twenty additional species were identified in smaller numbers and were not PCR-screened for parasites. These included Anopheles
punctimacula (n = 72), An. triannulatus (n = 853), Coquillettidia venezuelensis (n = 338), Culex (Culex) coronator (n = 118), Cx. (Cx.) declarator
(n = 56), Cx. (Cx.) quinquefasciatus (n = 507), Cx. (Melanoconion) conspirator (n = 80), Cx. (M.) interrogator (n = 114), Cx. (M.) seteki
(n = 281), Cx. (M.) taeniopus (n = 94), Cx. (M.) tecmarsis (n = 4), Limathus durhami (n = 13), Mansonia flaveola (n = 6), M. pseudotitillans (n = 75),
Aedes (Ochlerotatus) angustivittatus (n = 110), Aedes (Ochlerotatus) fulvus (n = 12), Aedes (Ochlerotatus) serratus (n = 475), Psorophora ciliata
(n = 163), Sabethes cyaneus (n = 13) and Wyeomyia abebela (n = 9).
2556 A .  B .  G A G E R  E T  A L .
? 2008 The Authors
Journal compilation ? 2008 Blackwell Publishing Ltd
Diversity and phylogenetic analysis of parasite 
cytochrome b lineages
The mosquito samples contained nine unique cytochrome
b sequences of Plasmodium (designated PAN1?PAN9;
GenBank accession nos. EU600217?EU600225), differing
from one another by 1.72%?10% (uncorrected-p distance
measure; Table 2). One pool of Ad. squamipennis yielded a
sequence with double peaks indicative of mixed infection.
The forward and reverse sequences were identical, and the
pattern of double peaks corresponded perfectly to a mixed
infection by PAN1 and PAN6, each of which had been found
by itself in different pools of the same mosquito species.
Four of the nine Plasmodium lineages have also been
found in birds at this study site GenBank accession nos.
EU600226?EU6600230: two lineages (PAN1 and PAN5) were
found when we conducted a preliminary screen of five
individuals each of blue-grey tanager Thraupis episcopus,
variable seedeater Sporophila americana, and red-legged
honeycreeper Cynanerpes cyaneus; two other lineages, PAN3
and PAN6 were found frequently in a larger screening of
Turdus grayi (unpublished data). We have designed more
specific primers to detect PAN3 and PAN6 in T. grayi
(unpublished data), and we have begun using these in
conjunction with the degenerate primers used in this study
of mosquitoes. To date, we have sequenced > 100 Plasmodium-
positive individuals of T. grayi, and all but one of the
infected birds contained only PAN3 or only PAN6 or a
mixed infection of the two lineages, i.e. PAN3 and PAN6
are the common lineages in T. grayi, sometimes occurring
together and sometimes separately (unpublished. data). In
mosquitoes, PAN3 was found in two pools of Cx. (M.) ocossa
and PAN6 was found in six pools of Ad. squamipennis
(Table 3).
Table 1 The mosquito species screened, the number of pools
screened per species and the number of pools found to be PCR-
positive for presence of parasites
Mosquito species
No. of pools screened 
(20 females per pool)
No. of pools 
PCR positive
Aedeomyia squamipennis* 39 11
Anopheles albimanus 10 0
Culex (Cx.) nigripalpus* 63 0
Culex (M.) erraticus 51 0
Culex (M.) ocossa 76 9
Coquillettidia nigricans 32 0
Mansonia dyari 40 0
Mansonia indubitans 14 0
Mansonia titillans 52 0
Uranotaenia apicalis 32 0
Uranotaenia lowi 10 0
*Species previously identified as hosts for avian Plasmodium 
include Ad. squamipennis (Crewe 1976) and Culex (Cx.) nigripalpus 
(Nayar et al. 1998).
Table 2 Uncorrected-p distances between Plasmodium cytochrome b sequences amplified from two mosquito species, Ad. squamipennis and
Culex (Melanoconion) ocossa. PAN2, 3 and 7 were identified in pools of Cx. (M.) ocossa. PAN1, 4, 5, 6, 8 and 9 were identified in Ad. squamipennis
PAN1 PAN1
PAN2 0.081 PAN2
PAN3* 0.099 0.024 PAN3*
PAN4 0.100 0.078 0.089 PAN4
PAN5 0.097 0.075 0.095 0.017 PAN5
PAN6* 0.086 0.075 0.086 0.037 0.040 PAN6*
PAN7 0.074 0.047 0.061 0.074 0.077 0.066 PAN7
PAN8 0.061 0.068 0.082 0.086 0.083 0.070 0.056 PAN8
PAN9 0.086 0.075 0.092 0.034 0.040 0.037 0.066 0.064 PAN9
P. relictum 0.080 0.053 0.071 0.077 0.074 0.063 0.054 0.025 0.063 P. relictum
P. elongatum 0.080 0.066 0.072 0.089 0.086 0.069 0.066 0.040 0.074 0.017
*PAN3 and PAN6 are the Plasmodium lineages found in T. grayi (unpublished data).
Fig. 2 The cumulative numbers of unique Plasmodium cytochrome
b lineages sequenced from pools of individual mosquitoes of two
mosquito species, Aedeomyia squamipennis and Culex (Melanoconion)
ocossa.
P L A S M O D I U M  L I N E A G E S  I N  M O S Q U I TO  A N D  AV I A N  H O S T S 2557
? 2008 The Authors
Journal compilation ? 2008 Blackwell Publishing Ltd
A total of six unique Plasmodium lineages were found
after screening 39 pools (780 individuals) of Ad. squamipennis,
and three lineages were found in 76 pools (1520 individuals)
of Cx. (M.) ocossa. In neither mosquito species does it
appear that the rate of discovery of new sequences is
decreasing significantly, i.e. approaching an asymptote,
which would indicate that the total parasite sequence
diversity is almost uncovered (Fig. 2). It is clear that the
rate of discovery is higher in Ad. squamipennis than in Cx.
(M.) ocossa.
Figure 3 shows the results of ML phylogenetic recon-
structions of the nine sequences we found in mosquitoes
in Panama and the two sequences from GenBank of
Plasmodium species with known mosquito-host species.
This analysis reveals phylogenetic clustering, although
incomplete, among the parasites identified from the same
mosquito species. Similar relationships were seen in a
neighbour-joining tree. In both NJ and ML trees, PAN2
and PAN3, both from Cx. (M.) ocossa, are sister lineages
with high bootstrap support (98% in both analyses).
PAN7 clusters with these two lineages although with a
bootstrap value less than 50%. Four of the parasites from
Ad. squamipennis, PAN4, 5, 6 and 9, form a monophyletic
clade with high bootstrap values in both NJ and ML (83
and 74%, respectively). However, PAN1 and PAN8, also
from Ad. squamipennis pools, are found outside of this
clade. PAN8 was found in a third grouping, in both NJ
and ML phylogenies, with P. relictum and P. elongatum,
both of which have Culex (Culex) species as hosts. By this
measure, PAN2 and 3 are the closest sequences to PAN7
in the group. The uncorrected-p distances between the
cytochrome b sequences amplified from mosquitoes are
shown in Table 3.
Discussion
Sharing a vertebrate host but not sharing a mosquito host
The two Plasmodium parasites PAN3 and PAN6 are both
found in Turdus grayi at Gamboa, Panama, sometimes
co-occurring in the same individual bird. When we
surveyed mosquitoes at Gamboa, we detected both PAN3
and PAN6 at this study site but in two different mosquito
species ? in fact, in different mosquito genera. The cytochrome
b lineage PAN3 was found in two PCR pools of Culex
(Melanoconion) ocossa, and PAN6 was found in six PCR
pools of Aedeomyia squamipennis. These data imply that the
co-occurrence of two parasite lineages in the same vertebrate
host is not explained by access being provided by a shared
mosquito host. This co-occurrence in an avian host of
congeneric malarial parasites transmitted by different
mosquito genera contrasts with the most widely studied
Plasmodium-mosquito-vertebrate host system, namely human
malaria. In that system, multiple parasite species that share
a vertebrate host are transmitted by mosquitoes in the
same genus or even the same species (Oaks et al. 1991).
Because our sampling of mosquito species was not
exhaustive, it is important to consider what pattern of host
distribution might exist but remain undetected. Therefore,
we address the likelihood that, with further screening of
mosquito individuals and species, PAN3 and PAN6 would
be found in additional mosquito species and, possibly, in
the same mosquito species. PAN3 was found in Cx. (M.)
ocossa, which belongs to one of the most speciose mosquito
groups in the neotropics (Pecor et al. 1992; Hutchings
et al. 2005). We trapped but did not screen four other
(less common) species of Cx. (Melanoconion) (Fig. 1), and
additional species of this subgenus have been collected in
the study area (Heinemann & Belkin 1978); thus the
opportunity certainly exists for PAN3 to infect multiple
Culex (Melanoconion) species in this geographical region.
PAN6 was found in Aedeomyia which, in contrast to Culex,
is monotypic in the Americas, having only a single species,
Ad. squamipennis, found throughout much of Central and
South America. It is possible that Ad. squamipennis is the
only host for PAN6, because it would be an unprecedented
taxonomic leap for a parasite to include Aedeomyia and
other genera in its Dipteran-host range. For instance,
human malaria species are transmitted by different
mosquitoes in different geographical regions, but all are
members of the genus Anopheles (Oaks et al. 1991; Donnelly
et al. 2002; Krzywinski & Besansky 2003). Similarly, in
avian malarial systems the vectors of Plasmodium forresteri,
P. elongatum, P. juxtanucleare, P. hermani and P. kempi are all
members of the same subgenus, Culex (Culex) (Christensen
et al. 1983; Telford et al. 1997). Therefore, we believe that
because PAN3 and PAN6 were identified in different
mosquito genera, they will not be found to co-occur in
Table 3 Number of pools of Aedeomyia squamipennis and Culex (M.)
ocossa in which each Plasmodium lineage was detected
Plasmodium lineages Aedeomyia squamipennis
Culex (M.) 
ocossa
PAN1* 2 0
PAN2 0 5
PAN3** 0 2
PAN4 1 0
PAN5* 1 0
PAN6** 6 0
PAN7 0 2
PAN8 1 0
PAN9 1 0
**denotes lineages found in extensive screen of Turdus grayi at this 
site (unpublished data); *denotes lineages found in preliminary 
screen of five individuals of three other bird species at this site 
(unpublished data).
2558 A .  B .  G A G E R  E T  A L .
? 2008 The Authors
Journal compilation ? 2008 Blackwell Publishing Ltd
mosquito species regardless of how much additional
sampling is conducted.
An important implication of finding PAN3 and PAN6
exclusively in different mosquito species is that the parasites
themselves should then be considered different species,
rather than haplotypes of the same species. This group of
parasites undergoes sexual recombination and reproduction
in the mosquito; if they do not encounter one another during
the sexual stage then they must be considered reproductively
and evolutionarily isolated units. The implication that PAN3
and PAN6 are indeed reproductively isolated is consistent
with three lines of genetic evidence:
1 The genetic distance between PAN3 and PAN6 (8.2%
uncorrected-p) is much higher than the average pairwise
distances between Plasmodium cytochrome b sequences
from many other avian studies (Ricklefs et al. 2005). 
2 This genetic distance is also greater than the average
cytochrome b distance among three most closely related
human malarial parasites that are known to be different
species: P. vivax, P. malariae and P. ovale (Escalante et al.
1998). 
3 Work with avian Plasmodium has shown that lineages
with cytochrome b distances as little as 1% exhibit no
nuclear gene recombination, meaning that these lineages
are reproductively isolated (Bensch et al. 2004).
 Finding two lineages exclusively in different mosquitoes
safely implies reproductive isolation, but finding two
lineages that share a mosquito host does not necessarily
mean that they are the same species. For example, some
lineages of Plasmodium overlapped in mosquito hosts
but had different avian hosts and appeared to represent
genetically distinct lineages: PAN6 (present in T. grayi) and
PAN5 (absent in T. grayi) differed by 4.02% genetic distance
despite both being found in Ad. squamipennis. Likewise,
PAN3 (present in T. grayi) and PAN2 (absent in T. grayi)
differed by 2.37% genetic distance despite both being
found in Cx. (M.) ocossa. The finding of closely related para-
site lineages sorting into different avian-host species in
a community of birds has also been reported elsewhere
(Ricklefs et al. 2005). This highlights the need for multiple
lines of evidence when considering host-parasite coevolution
and parasite taxonomy.
Fig. 3 Maximum likelihood tree of nine
Plasmodium cytochrome b sequences obtained
from pools of mosquitoes from our Gamboa
site and of published sequences of P. relictum
and P. elongatum (both transmitted by Culex
(Culex) sp.). PAN3 and PAN6 are the two
lineages commonly found in Turdus grayi.
Likelihood parameters were determined
using modeltest 3.5 and correspond to
the GTR + G + I model with nucleotide
frequencies determined by the data.
Relationships were similar in a neighbour-
joining tree.
P L A S M O D I U M  L I N E A G E S  I N  M O S Q U I TO  A N D  AV I A N  H O S T S 2559
? 2008 The Authors
Journal compilation ? 2008 Blackwell Publishing Ltd
Diversity and phylogenetic relationships of parasites 
found in mosquitoes
Despite using only a single primer pair we found a surprising
diversity of parasite lineages in only two mosquito species.
Three lineages were found in Cx. (M.) ocossa and six
additional lineages in Ad. squamipennis. Both the number of
unique parasite sequences and the percentage of infected
pools were higher in Ad. squamipennis than in Cx. (M.)
ocossa. On average, each pool that we sampled yielded
approximately 0.23 new sequences in Ad. squamipennis and
0.05 in Cx. (M.) ocossa, suggesting that in total Ad. squamipennis
harbours many more parasites than Cx. (M.) ocossa. Given
that four of the nine sequences recovered in this study were
sequenced only once it is likely that many more sequences
would be uncovered with further screening of mosquito
pools. This is despite the fact that many mosquitoes were
negative for Plasmodium in our PCR screens. This low rate
of infection is consistent with studies based on microscopic
examination of mosquito salivary glands, which have
found infection rates as low as 1.9% for avian Plasmodium
(Reeves et al. 1954) and < 1% for human Plasmodium
(Charlwood et al. 2003, Shililu et al. 2004). This raises the
methodological point that we amplified Plasmodium DNA
from whole mosquitoes rather than from salivary glands.
If any of the amplified parasite lineages were unable to
develop into infective sporozoites in these mosquitoes (as
has been seen in some studies of experimental infections,
e.g. Grieco et al. 2005), then hosts would not be exposed to
them. This possibility does not change the fact that PAN3
and PAN6 were found exclusively in mosquitoes of different
genera, and it also seems unlikely that none of the seven
other lineages was capable of developing into sporozoites.
Phylogenetic analysis of parasite sequences recovered
from pools of mosquitoes showed that the majority of
parasites found in a mosquito species exhibit a degree of
monophyly but that this monophyly is incomplete because
distantly related parasites can also be found in the same
mosquito host. In Ad. Squamipennis, four of the six unique
Plasmodium sequences formed a monophyletic clade that
was highly supported. Similarly in Cx. (M.) ocossa, PAN2
and PAN3 are each other?s closest relatives. Finally, P. relictum
and P. elongatum, the additional two Plasmodium species
that have been found in Culex (Culex) mosquitoes, also
cluster together although with relatively lower support.
Finding a clade of related but reproductively isolated
Plasmodium lineages in a single mosquito species raises
interesting questions about the dynamics of speciation in
this system, because it suggests the possibility that speciation
could have occurred sympatrically. Furthermore, the occur-
rence of a clade of related parasites in a single mosquito
species is very unlike host associations of these parasites in
birds, where there is less congruence between bird and
parasite phylogenies (Ricklefs et al. 2004). This suggests
that parasites are more restricted to their mosquito hosts
than to their vertebrate hosts.
We now return to the original question of whether parasites
that share a vertebrate host do so because of an encounter
filter imposed by their mosquito host. Two lines of evidence
from this study suggest that parasite?mosquito associations
do not restrict the access of parasites to birds but rather
seem to be the mechanism that facilitates the switching of
avian hosts that we know has occurred in the evolutionary
history of this system. First, parasite lineages found in T.
grayi were found in different mosquito species. Second, the
two parasite lineages of T. grayi were found in their respective
mosquito species with several other parasite lineages that
were not found in T. grayi. This implies that the distribution
of parasites in T. grayi is not a simple reflection of the lineages
that are afforded access to T. grayi by mosquitoes. Instead,
vertebrate hosts appear to be frequently exposed to a
diverse array of Plasmodium lineages, only some of which
are regularly successful at inhabiting a particular verte-
brate species. Thus, the frequent encounter that mosquitoes
seem to create between apparently incompatible hosts and
parasites may be the mechanism for Plasmodium?s switching
of vertebrate hosts, rather than mosquitoes acting as an
access filter that prevents such switching.
Acknowledgements
We thank Rosemary Grant, Peter Grant, Martin Wikelski, Michaela
Hau and Robert Ricklefs for discussions and for comments on
previous drafts. Funding was provided by Princeton University,
the Smithsonian Tropical Research Institute and a National Science
Foundation Doctoral Dissertation Improvement Grant to A.B.G.
References
Bahl A, Brunk B, Crabtree J et al. (2003) PlasmoDB: the Plasmodium
genome resource. A database integrating experimental and
computational data. Nucleic Acids Research, 31, 212?215.
Beadell JS, Gering E, Austin J et al. (2004) Prevalence and differential
host-specificity of two avian blood parasite genera in the
Australo-Papuan region. Molecular Ecology, 13, 3829?3844.
Bensch S, Perez-Tris J, Waldenstrom J, Hellgren O (2004) Linkage
between nuclear and mitochondrial DNA sequences in avian
malaria parasites: multiple cases of cryptic speciation? Evolution,
58, 1617?1621.
Bensch S, Stjernman M, Hasselquist D et al. (2000) Host specificity
in avian blood parasites: a study of Plasmodium and Haemoproteus
mitochondrial DNA amplified from birds. Proceedings of the
Royal Society of London Series B-Biological Sciences, 267, 1583?1589.
Charlwood JD, Pinto J, Sousa CA et al. (2003) ?A mate or a meal? ?
Pre-gravid behaviour of female Anopheles gambiae from the
islands of Sao Tome and Principe, West Africa. Malaria Journal,
2, Article 9.
Christensen BM, Barnes HJ, Rowley WA (1983) Vertebrate host
specificity and experimental vectors of Plasmodium (Novyella)
kempi Sp-N from the Eastern Wild Turkey in Iowa. Journal of
Wildlife Diseases, 19, 204?213.
2560 A .  B .  G A G E R  E T  A L .
? 2008 The Authors
Journal compilation ? 2008 Blackwell Publishing Ltd
Combes C (2001) Parasitism: the Ecology and Evolution of Intimate
Interactions. University of Chicago Press, Chicago.
Crewe SM (1976) Aedeomyia, a possible new vector of avian
malaria. Annals of Tropical Medicine and Parasitology, 70, 477?478.
Donnelly MJ, Simard F, Lehmann T (2002) Evolutionary studies of
malaria vectors. Trends in Parasitology, 18, 75?80.
Escalante AA, Barrio E, Ayala FJ (1995) Evolutionary origin of
human and primate malarias ? evidence from the circumsporozoite
protein gene. Molecular Biology and Evolution, 12, 616?626.
Escalante AA, Freeland DE, Collins WE, Lal AA (1998) The evolution
of primate malaria parasites based on the gene encoding
cytochrome b from the linear mitochondrial genome. Proceedings
of the National Academy of Sciences, USA, 95, 8124?8129.
Eyles DE (1951) Studies on Plasmodium gallinaceum. 1. Characteris-
tics of the infection in the mosquito, Aedes aegypti. American Jour-
nal of Hygiene, 54, 101?112.
Fallon SM, Bermingham E, Ricklefs RE (2003) Island and taxon
effects in parasitism revisited: avian malaria in the Lesser
Antilles. Evolution, 57, 606?615.
Fallon SM, Ricklefs RE, Latta SC, Bermingham E (2004) Temporal
stability of insular avian malarial parasite communities. Proceedings
of the Royal Society of London, Series B: Biological Sciences, 271,
493?500.
Fonseca DM, Atkinson CT, Fleischer RC (1998) Microsatellite
primers for Culex pipiens quinquefasciatus, the vector of avian
malaria in Hawaii. Molecular Ecology, 7, 1617?1619.
Fry JD (1990) Trade-offs in fitness on different hosts ? evidence
from a selection experiment with a phytophagous mite. American
Naturalist, 136, 569?580.
Gabaldon AU, Ulloa G, Pulido J (1981) Distribucion geografica,
ecologia y etologia de Aedeomyia squamipennis, importante
vector natural de malaria aviaria en Venezuela. Boletin de la
Direccion de Malariologia y Saneamiento Ambiental, 21, 103?113.
Grieco JP, Achee NL, Roberts DR, Andre RG (2005) Comparative
susceptibility of three species of Anopheles from Belize, Central
America, to Plasmodium falciparum (NF-54). Journal of the American
Mosquito Contrological Association, 21, 279?290.
Heinemann SJ, Belkin JN (1978) Collection records of the project
?Mosquitoes of Middle America.? 10. Panama, including the
Canal Zone (PA, GG). Mosquito Systematics, 10, 119?196.
Hellgren O, Waldenstrom J, Perez-Tris J et al. (2007) Detecting shifts
of transmission areas in avian blood parasites ? a phylogenetic
approach. Molecular Ecology, 16, 1281?1290.
Hofmeyr M, Bingham J, Lane EP, Ide A, Nel L (2000) Rabies in
African wild dogs (Lycaon pictus) in the Madikwe Game
Reserve, South Africa. Veterinary Record, 146, 50?52.
Hutchings RSG, Sallum MAM, Ferreira RLM, Hutchings RW
(2005) Mosquitoes of the Jau National Park and their potential
importance in Brazilian Amazonia. Medical and Veterinary
Entomology, 19, 428?441.
Ishtiaq F, Gering E, Rappole JH et al. (2007) Prevalence and diversity
of avian hematozoan parasites in Asia: a regional survey. Journal
of Wildlife Diseases, 43, 382?398.
Krzywinski J, Besansky NJ (2003) Molecular systematics of
Anopheles: From subgenera to subpopulations. Annual Review of
Entomology, 48, 111?139.
Lajeunesse MJ, Forbes MR (2002) Host range and local parasite
adaptation. Proceedings of the Royal Society of London, Series B:
Biological Sciences, 269, 703?710.
Lawson M (1995) Rabbit virus threatens ecology after leaping the
fence. Nature, 378, 531?531.
Matthews BE (1998) An Introduction to Parasitology. Cambridge
University Press, Cambridge.
McConkey GA, Li J, Rogers MJ et al. (1996) Parasite diversity in an
endemic region for avian malaria and identification of a parasite
causing penguin mortality. Journal of Eukaryotic Microbiology, 43,
393?399.
Nayar JK, Knight JW, Telford SR (1998) Vector ability of mosquitoes
for isolates of Plasmodium elongatum from raptors in Florida.
Journal of Parasitology, 84, 542?546.
Oaks SC, Mitchell VS, Pearson GW, Carpenter CCJ (1991) Malaria:
Obstacles and Opportunities. A report of the Institute of Medicine.
National Academy Press, Washington, DC.
Oskam L, Schoone GJ, Kroon CCM, Lujan R, Davies JB (1996)
Polymerase chain reaction for detecting Onchocerca volvulus in pools
of blackflies. Tropical Medicine and International Health, 1, 522?527.
Pecor JE, Mallampalli VL, Harbach RE, Peyton EL (1992) Catalog
and illustrated review of the subgenus Melanoconion of Culex
(Diptera: Culicidae). Contributions of the American Entomological
Institute, 27, 1?228.
Perkins SL, Schall JJ (2002) A molecular phylogeny of malarial
parasites recovered from cytochrome b gene sequences. Journal
of Parasitology, 88, 972?978.
Posada D, Crandall KA (1998) modeltest: testing the model of
DNA substitution. Bioinformatics, 14, 817?818.
Rathore D, Wahl AM, Sullivan M, McCutchan TF (2001) A phylo-
genetic comparison of gene trees constructed from plastid,
mitochondrial and genomic DNA of Plasmodium species.
Molecular and Biochemical Parasitology, 114, 89?94.
Reeves WC, Herold RC, Rosen L, Brookman B, Hammon WM (1954)
Studies on avian malaria in vectors and hosts of encephalitis in
Kern County, California.2. Infections in mosquito vectors. American
Journal of Tropical Medicine and Hygiene, 3, 696?703.
Richard FA, Sehgal RNM, Jones HI, Smith TB (2002) A comparative
analysis of PCR-based detection methods for avian malaria.
Journal of Parasitology, 88, 819?822.
Ricklefs RE, Fallon SM (2002) Diversification and host switching in
avian malaria parasites. Proceedings of the Royal Society of London,
Series B: Biological Sciences, 269, 885?892.
Ricklefs RE, Fallon SM, Bermingham E (2004) Evolutionary
relationships, cospeciation, and host switching in avian malaria
parasites. Systematic Biology, 53, 111?119.
Ricklefs RE, Swanson BL, Fallon SM et al. (2005) Community
relationships of avian malaria parasites in southern Missouri.
Ecological Monographs, 75, 543?559.
Ridgely RS, Gwynne JA (1989) A Guide to the Birds of Panama, with
Costa Rica, Nicaragua, and Honduras, 2nd edn. Princeton University
Press, Princeton, NJ.
Shililu J, Ghebremeskel T, Seulu F et al. (2004) Seasonal abundance,
vector behavior, and malaria parasite transmission in Eritrea.
Journal of the American Mosquito Contrological Association, 20,
155?164.
Swofford DL (2002) PAUP* Phylogenetic Analysis Using Parsimony (*and
Other Methods). Sinauer Associates, Sunderland, Massachusetts.
Szymanski MM, Lovette IJ (2005) High lineage diversity and host
sharing of malarial parasites in a local avian assemblage. Journal
of Parasitology, 91, 768?774.
Telford SR, Nayar JK, Foster GW, Knight JW (1997) Plasmodium
forresteri n. sp., from raptors in Florida and southern Georgia:
its distinction from Plasmodium elongatum morphologically
within and among host species and by vector susceptibility.
Journal of Parasitology, 83, 932?937.
P L A S M O D I U M  L I N E A G E S  I N  M O S Q U I TO  A N D  AV I A N  H O S T S 2561
? 2008 The Authors
Journal compilation ? 2008 Blackwell Publishing Ltd
Valki?nas G (2004) Avian Malarial Parasites and Other Haemosporidia.
Taylor and Francis Publishing, London.
Van Riper C, Vanriper SG, Goff ML, Laird M (1986) The epizootiology
and ecological significance of malaria in Hawaiian land birds.
Ecological Monographs, 56, 327?344.
Waldenstrom J, Bensch S, Kiboi S, Hasselquist D, Ottosson U (2002)
Cross-species infection of blood parasites between resident and
migratory songbirds in Africa. Molecular Ecology, 11, 1545?1554.
Work TM, Washino RK, Vanriper C (1990) Comparative susceptibility
of Culex tarsalis, Anopheles franciscanus, and Culiseta inornata
(Diptera, Culicidae) to Plasmodium relictum (Haemosporidia,
Plasmodiiae). Journal of Medical Entomology, 27, 68?71.
Andrea Gager is interested in the ecology and evolution of disease
in human and non-human animals. Jose del Rosario Loaiza
studies the ecology of human and avian malaria vectors and is
now a PhD student at McGill University and the Smithsonian
Tropical Research Institute. Don Dearborn is a behavioural ecologist
with interests in host-parasite interactions, reproductive strategies,
and conservation biology. Eldredge Bermingham?s research focuses
on molecular population genetics and historical biogeography,
primarily of neotropical vertebrates and butterflies.