Agnew 1997 A

8/23/2019 Agnew 1997 A

1/7

Virulence, parasite mode of transmission, and hostfluctuating asymmetry

P H I L I P A G N E W",#* JA C OB C. KOELLA#

" Experimental Ecolog, EHuWrich, EH-entrum W, H-8092uWrich, Siterland# Department ofoolog, Aarhus Uniersit, Uniersitetsparken B135, D-8000 Aarhus , Denmark

(agnew!aau.dk/koella!aau.dk)

SUM M ARY

Horizontally transmitted parasites are broadly predicted to be more virulent, or costly to host fitness, thanthose with vertical transmission. This is mainly because vertical transmission, from host parent tooffspring, explicitly links the reproductive interests of both parties. Underlying this prediction is a generalassumption that parasite transmission success is positively correlated with its virulence. We report resultswhere infection of larval yellow fever mosquitoes Aedes aegptiwith the microsporidian Edhaardia aedis wasexperimentally manipulated. The parasites complex life cycle allowed comparisons between estimates ofhorizontal and vertical transmission on host fitness. Our measure of virulence was the fluctuatingasymmetry (FA) of adult female wings. Hosts harbouring spores showed higher FAs than controls.Horizontally transmitting spores were associated with higher FAs than vertically transmitting spores.Furthermore, within hosts FA correlated positively with the number of horizontally transmitting spores,while no relation was seen with the number of vertically transmitting spores. A developmental mechanismuncoupling the relationship between vertical transmission and virulence is proposed.

1 . INTRODUCTION

Virulence, or cost to host fitness, can be viewed as acomposite of traits contributing towards a parasites

overall fitness. The individual traits may have positive,neutral, or negative fitness effects on the parasite(reviewed in Bull 1994). The direction and strength ofselection acting on individual traits is predicted todepend on their contribution towards overall fitnessand on their correlation with other traits (Ebert &Herre 1996). Of particular interest are correlated traitswhich act in an antagonistic manner, i.e. where aresponse to selection in one trait reduces a correlatedtraits contribution towards fitness. A widely usedassumption is the balance required between a parasitesincreased transmission success and additional costs toits host (reviewed by Anderson & May 1991). Conse-quently, selection on the parasite is often predicted to

produce intermediate virulence, which has supportfrom medical, field, and laboratory studies (reviewedby Anderson & May 1991; Bull 1994; Ewald 1994).

Comparative studies of virulence and parasite modeof transmission have been particularly illuminating inthis evolutionary scenario. Various aspects of host andenvironmental ecology act to determine upper limitsthat virulence can be maintained at. Parasites trans-mitting purely vertically from parent to offspring havetheir reproductive success explicitly linked to that oftheir hosts. Any cost to host reproductive success affects

* Author and address for correspondence.

the parasites own success and theoretically will drivethe parasite to extinction. Therefore, such parasites areoften benign (see Ewald 1994). Further support for thisnotion comes from parasites that are uniparentally

inherited. These parasites show adaptations to directlyenhance their own fitness, or that of the transmittingsex, by exploiting the non-transmitting sex (see Hurst1993). Horizontal transmission among hosts is lessconstrained by host condition: increased parasitefitness can evolve at additional costs to the host.Comparative studies of horizontal transmission, e.g.vector-borne and direct contact, follow these pre-dictions (Ewald 1994). Additionally, horizontallytransmitting ectoparasites of rock doves are morevirulent than vertically transmitting counterparts(Clayton & Tompkins 1994). An alternative approachshows that avian hosts immunocompetence increaseswith nesting ecologies more exposed to horizontally

transmitting parasites (Mller & Erritze 1996).Direct testing of the transmission modevirulence

hypothesis is difficult if parasites have only one mode oftransmission or different modes of transmission indifferent host species. A few studies have avoided thisproblem by studying parasites that use both verticaland horizontal transmission in the same host species.Phages selected for vertical transmission in theirbacterial hosts developed a more benign relationshipwith their hosts than those from the original populationselected for horizontal transmission (Bull et al. 1991). Acomparative study of nematodes parasitizing fig waspsshowed that as vertical transmission becomes more

Proc. R. Soc. Lond. B (1997) 264, 915 # 1997 The Royal SocietyPrinted in Great Britain

9

8/23/2019 Agnew 1997 A

2/7

10 P. Agnew and J. C. Koella Parasite irulence and host fluctuating asmmetr

important in a particular hostparasite relationshipthe less virulent the nematodes are (Herre 1993).However, most comparative studies of transmissionand virulence are not phylogenetically controlled(Harvey & Pagel 1991). This leads to the possibilitythat differences in virulence arose from differentevolutionary histories rather than due to observedpatterns of transmission. Herres (1993) work suffersleast from this potential problem by only consideringclosely related species. A further problem is that

selection for vertical transmission necessarily acts onboth host and parasite. Any response to selection couldbe due to the parasite, the host, or both. For example,Sweeney et al. (1989) selected for increased verticaltransmission success of the microsporidian Amblosporadxenoides in its mosquito host ulex annulirostris, butwere unable to attribute this to either host or parasite.Bull et al. (1991) largely controlled for this possibilityand demonstrated that the response to selection waspredominantly due to the parasite.

The hostparasite relationship between the yellowfever mosquito Aedes aegpti and the microsporidianEdhaardia aedis (Becnel et al. 1989) is suited for testing

the mode of transmissionvirulence hypothesis. Theparasite uses both vertical and horizontal transmissionat different stages in its complex life cycle (seeMaterials and Methods) and is host-species specific(Andreadis 1994). This avoids the problems statedabove. Following horizontal infection of host larvae,the parasites subsequent mode of transmission may beeither vertical or horizontal. Each mode of transmissionhas its own distinct type of spore. Vertical transmissionof the parasite requires the reproductive success asadults of females infected as larvae. In contrast,horizontally infected larvae can only transmit theparasite horizontally to other larvae if they themselves

die as larvae or pupae. Death in this manner is inducedby spore production of the parasite. This alone supportspredictions of the transmissionvirulence hypothesis.However, this binary measure of virulence (hosts alivevs dead hosts) related to transmission (vertical vshorizontal) is a coarse description of the correlationsinvolved in this hostparasite relationship. Greaterinsight can be taken from comparison between adultfemales. Many of these females harbour spores capableof vertical transmission. Also, some females will havesurvived an attempt at horizontal transmission by theparasite, and harbour spores specific for that purpose.Theories on the evolution of virulence predict that thelatter females should have experienced greater costs

during their development than the former females. Asthese costs are experienced by hosts during devel-opment, we predict that the fluctuating asymmetry(FA) of adult characteristics should differ betweenfemales based on the mode of transmission pursued bythe parasite. Some individuals harbour both sporetypes, thus permitting within-host analysis of theireffects on FA. Only females were considered as onlythey provide a potential route for vertical transmission(Hembree & Ryan 1982), and the sexes may providedifferent developmental cues for the parasite.

Fluctuating asymmetries are small randomdeviations from an otherwise bilaterally symmetrical

trait (Van Valen 1962). They are believed to arisefrom some form of stress the individual is unable tobuffer during development of the trait. These stressesmay arise from a variety of sources (Mller & Swaddle1996), including parasites (Mller 1996). In somecases, FAs themselves may have direct negative effectson components of fitness, e.g. avian flight performance(Balmford et al. 1993). Stabilizing selection on suchtraits is predicted to reduce developmental flux bycanalization (Mller & Pomiankowski 1993). More

often FA is cited as an indirect measure of fitness;numerous studies report increased FAs to correlatenegatively with other fitness traits (reviewed Mller &Swaddle 1996).

We experimentally tested the mode of transmissionvirulence hypothesis with Ae. aegpti and E. aedis. Ourpredictions were that the FA of adult female mosquitowings would correlate positively with the degree ofstress hosts experienced during their development. Thedegree of stress was predicted to depend on whether theparasite pursued either vertical or horizontal trans-mission following larval infection.

2 . M ATERIALS AND M ETHODS

(a) The host and its parasite

Aedes aegpti, the yellow fever mosquito, has been ex-tensively studied due to its role as a disease vector and itssuitability for laboratory maintenance (Christophers 1960).The Rockefeller strain obtained from the Swiss TropicalInstitute (Basel) was used. Mosquitoes were maintained at28(p0.5) mC and 85(p5)% humidity with a 12 h12 hlightdark cycle.

Edhaardia aedis is an obligate intracellular microsporidianparasite with a complex life cycle (Becnel et al. 1989). Hostlarvae ingest uninucleate spores from the environment (figure1). In the mid-gut, polar tubes extend from the spores to

Figure 1. Alternative life cycles of Edhaardia aedis infectingAedes aegpti. Solid lines indicate the horizontalverticaltransmission sequence. Dashed lines indicate the horizontalhorizontal transmission sequence (see text for details).

Proc. R. Soc. Lond. B (1997)

8/23/2019 Agnew 1997 A

3/7

11Parasite irulence and host fluctuating asmmetr P. Agnew and J. C. Koella

puncture epithelial cells and initiate the infection. Furthertissues become infected by some unknown mechanism as theparasite progresses through several developmental stages.Binucleate spores are specifically produced within hostoenocyte cells. These spores, carried in the haemocoel, aresolely responsible for vertical transmission by infection of afemales developing eggs. Further development by theparasite is prevented until the eggs hatch and larvae develop.Infection is predominantly seen in fat body tissue whereuninucleate spores are produced. Vertically infected larvaerarely survive beyond pupation. The proliferation of uni-

nucleate spores results in their death. These spores are notshed during the hosts life, only being released into theenvironment following the rupture of the host cuticle.

Alternatively, binucleate spores produced in horizontallyinfected larvae may germinate and infect tissues other thandeveloping eggs. In this case, the parasite continues withdevelopment that otherwise would be seen in verticallyinfected larvae. This culminates in the production ofuninucleate spores in the fat body. If enough spores areproduced to induce the hosts death before its emergencefrom the larval site, then uninucleate spores will be releasedand available for horizontal transmission: vertical trans-mission will have been bypassed. However, if insufficientnumbers of uninucleate spores are produced, then hosts may

emerge harbouring uninucleate spores. The two spore typesare specific to their mode of transmission: uninucleate sporesare incapable of vertically transmitting the parasite via afemales eggs.

Edhaardia aedis were obtained from stock maintained byDr J. J. Becnel at the United States Department of Agri-culture, Gainesville.

(b) Experimental protocol and data collection

Two experiments were designed to produce emergingadult female Ae. aegpti harbouring infections of E. aedispursuing vertical transmission or having failed to achievehorizontal transmission from infected larvae or pupae. Wemanipulated doseifood conditions in one experiment and

doseiage at infection in the other experiment to expose theparasites potential patterns of transmission.

In both experiments uninfected mosquito eggs weresynchronously hatched. Within 6 h of hatching larvaewere separated into rearing pans measuring10 cmi10 cmi10 cm and containing 300 ml of tap water.Dead larvae and pupae were removed daily. Live pupae fromwithin treatments were transferred on a daily basis to smallbeakers of water kept under inverted mesh-bottomed beakers.Two days after emergence, adults were killed by freezing,and stored individually in 1.5 ml vials at k20 mC untilfurther investigation.

Experiment I consisted of 20 treatments that initiallycontained 200 larvae each. The treatments consisted of fourlarval food availabilities combined with exposure to fiveconcentrations of uninucleate spores. Larval diet wasmanipulated as 25 %, 50 %, 75 % and 100 % of astandardized ration of TetraminTM fish food. Rations pertreatment were adjusted daily allowing for mortality andpupation. Rations per larva on a 100% diet were: day 1,0.06 mg; day 2, 0.08 mg; day 3, 0.16 mg; day 4, 0.32 mg;even-numbered days thereafter, 0.64 mg. Seventy-two hoursafter hatching, larvae were transferred to 100 ml beakers ofwater and exposed to either 0, 10, 100, 1000 or 10000uninucleate spores ml" for 24 h. The spores had beenharvested from vertically infected larvae. The days foodration was also added to the beakers. After exposure, larvaewere rinsed in tap water and returned to their originalrearing trays.

Experiment II consisted of nine treatments initiallycontaining 100 larvae each. Treatments involved three larvalages of infection (24, 72 and 120 h post-hatch) and threeconcentrationes of uninucleate spores (10, 100 and 1000uninucleate spores ml"). Daily maintenance and infectionwere otherwise as in Experiment I.

Adult female wings were later removed and stuck to glassmicroscope slides. Length, from the distal end of the alula tothe peripheral tip of vein R3, was measured with imageanalysis software (NIH Image 1995) to an accuracy of0.01 mm viaa video cameralinked to a dissecting microscope.

Signed wing asymmetries in each experiment were normallydistributed (Shapiro-Wilk Wtest: Experiment I, Wl0.946,

p0.05; Experiment II, Wl0.986,p0.05) with meansnot significantly different from zero (Experiment I, t

"!*)l

k0.073,p0.9; Experiment II, t"#*l1.235,p0.2).

Some mosquitoes were measured twice, from different videoimages. Their asymmetries were replicable (intra-classcorrelation coefficients, left rl0.99, right rl0.98, F

(*,"'!l

5.274,p0.0001).

(c) Statistical analyses

The parasites life cycle gives temporal correlationsbetween the two spore types. This correlation cannot be

deduced from spores found in individuals collected as adults.In neither experiment did the numbers of the two spore typessignificantly correlate (two-tailed linear regressions of log

"!(uninucleate spores) on log

"!(binucleate spores), Experiment

I F",(#l0.009,p0.9, Experiment II F

",#$l0.659,p

0.4). Therefore, we treated them as independent variables insubsequent analyses.

Our a priori interest was in the relationship between themode of transmission pursued by the parasite and virulence.The parameters chosen to assay this were the number andtype of spores on host FA. Different experimental treatmentswere used to generate the ranges of spore productionobserved. However, there may have been a consistentrelationship between treatment, production of spores, andthe resulting asymmetry, as well as direct effects on

asymmetry. Therefore, multiple linear regressions removedtreatment effects from the log

"!-transformed (number of

spores) and square root-transformed (absolute asymmetry)data. In no case did the treatments or their interactions havesignificant effects (p0.1). The effects of residual sporenumbers on the residual asymmetries were tested by multiplelinear regression; these analyses were not sensitive to theordering of terms entered into the models.

It was not possible to subtract treatment influence from thepresence of spore. Instead, we show the effect of thetreatments on binucleate and uninucleate spore presence andon asymmetry itself. Nominal logistic regressions were usedfor treatment effects on spore presence. Dosages were log

"!-

transformed with no exposure to uninucleate spores coded aszero. A non-parametric two-way ANOVA tested for absoluteasymmetrys response to treatment conditions (Sokal &Rohlf 1995).

A one-tailed Wilcoxon ranked sums test on the data fromExperiment I compared absolute asymmetries of controls vsindividuals harbouring either type of spore. The sameanalysis, in each experiment, compared absolute asymmetriesof individuals harbouring binucleate or uninucleate spores.Each experiments result was combined for a probability ofsignificance test. Our predictions were that FA would behigher in the presence of spores and with uninucleate spores.Non-parametric analyses were used when residuals ofparametric analyses failed to give normal distributions, evenafter data transformations. The statistical package JMP(SAS Institute Inc. 1994) performed all the analyses.


8/23/2019 Agnew 1997 A

4/7


3 . RESULTS

Increased dosages of uninucleate spore concen-tration significantly increased the presence of bothbinucleate and uninucleate spores in adult females ofExperiment I (table 1 a, b). Higher food availabilities

Table 1. Experiment I. Analses of treatments on the presenceof both spore tpes and on absolute asmmetr

(a) Nominal logistic regression of treatmentson binucleate spore presence

source d.f. # p

food 1 2.019 0.155dose 1 8.957 0.003interaction 1 5.927 0.015r#l0.31, nl1204

(b) Nominal logistic regression of treatmentson uninucleate spore presence

source d.f. # p

food 1 2.997 0.0830

dose 1 13.792 0.0002interaction 1 0.650 0.4201r#l0.31, nl1204

(c) Non-parametric two-way ANOVA oftreatments on absolute asymmetry

source d.f. m.s. H p

food 3 65 263 6.415 n.s.dose 2 32 766 3.221 n.s.interaction 10 55 909 5.495 n.s.error 1080 95 748

Table 2. Experiment II. Analses of treatments on the

presence of both spore tpes and on absolute asmmetr

(a) Nominal logistic regression of treatmentson binucleate spore presence

source d.f. # p

age 1 10.787 0.001dose 1 0.171 0.679interaction 1 1.633 0.201r#l0.26, nl161

(b) Nominal logistic regression of treatmentson uninucleate spore presence

source d.f. # p

age 1 0.031 0.861dose 1 0.658 0.417interaction 1 0.856 0.347r#l0.16, nl161

(c) Non-parametric two-way ANOVA oftreatments on absolute asymmetry

source d.f. m.s. H p

age 2 414 0.306 n.s.dose 2 416 0.307 n.s.interaction 4 1477 1.090 n.s.error 120 1382

Figure 2. Mean (js.e.) absolute asymmetries of femalemosquito wings grouped by the type of spores that individualsharboured. (a) Experiment I. Control s spore presence onabsolute asymmetry; Wilcoxon ranked sums: #

["]l9.741,

pl0.002. Binucleate s uninucleate presence on absoluteasymmetry; Wilcoxon ranked sums: #

["]

l3.272,pl0.071.(b) Experiment II. Binucleate s uninucleate presence onabsolute asymmetry. Wilcoxon ranked sums: #

["]l3.238,

pl0.072. Combined probability test of binucleate anduninucleate presence on absolute asymmetry from Experi-ments I and II: #

[%]l10.569,p0.05.

interacted with higher dosages to increase binucleatespore presence. Lower food availability contributed toa lower presence of uninucleate spores. In ExperimentII, where larvae experienced high food conditions,increasing age at infection significantly increased thepresence of binucleate spores (table 2a). The lowerrange of dosages had no significant effect on either

spore types presence (table 2 a, b). In neither ex-periment did treatment conditions affect asymmetry(table 1c, 2 c).

Within both experiments, individuals harbouringuninucleate spores tended to show more asymmetrythan those with binucleate spores (figure 2 a, b), andover both experiments this difference was significant.

Once treatment effects were removed from thenumber of each type of spore and on asymmetry therewere significant and positive correlations betweenuninucleate spore number and asymmetry in bothexperiments (figure 3 a, table 3a, b). In contrast,neither experiment showed a relationship between


8/23/2019 Agnew 1997 A

5/7


Figure 3. Correlations between residual log"!

(spore number)and residual N(absolute asymmetry) once the effects of theother spore type are removed. (a) Binucleate spores. (b)

Uninucleate spores. Experiment I filled diamonds, Experi-ment II open diamonds.

Table 3. Multiple regressions of residual spore numbers onresidualN(absolute asmmetr)(a) Experiment I

source d.f. s.s. F p

binucleate 1 0.0003 0.058 0.811uninucleate 1 0.0328 7.457 0.009interaction 1 0.0030 0.672 0.417error 41 0.1806r#l0.20, nl45. Wl0.973,p0.5

(b) Experiment II

source d.f. s.s. F p

binucleate 1 0.0006 0.290 0.598uninucleate 1 0.0377 17.757 0.001interaction 1 0.0002 0.099 0.757error 16 0.0340r#l0.60, nl20. Wl0.926,p0.1

binucleate spore number and asymmetry, nor anyinteraction between the spore types (figure 3 b, table3a, b).

4 . DISCUSSION

Adult female mosquitoes from Experiment Iharbouring either type of spore showed higher wing FAthan control mosquitoes. Treatment conditions hadeffects on the presence of both spore types but notdirectly on FAs, but could still have had some effect viacorrelations with spore production. However, theserelationships are likely to be context-dependent as thecorrelation of dosage with spore presence in the two

experiments showed. The presence of uninucleatespores in both experiments was associated with higherFAs than that of binucleate spores. Thus, FA increasedin the presence of either spore type, but was consistentlyhigher in association with horizontally transmittinguninucleate spores than vertically transmittingbinucleate spores.

All developmental stages of the infection shouldimpose some metabolic cost to their host as micro-sporidia have no mitochondria (Sprague et al. 1992).Spore production in particular should be costly tohosts. A classic symptom of microsporidioses is whitepatches of hypertrophied cells following their rupture

after intracellular spore production. Other costs mayinvolve direct competition between host and parasitefor protein. This is often a limiting resource in aquaticenvironments, especially for non-carnivorous species,and is strongly implied in the evolution of adult femalemosquite feeding ecology. Protein is a key constituentin the multi-layered walls of both spore types (Va!vra &Maddox 1976). Beyond their structural role, proteinsare vital to almost all metabolic processes, especiallyduring the digestion and formation of tissues atmetamorphosis (Chapman 1969).

Interpretation of these results relies on our as-sumption that FA is a measure of cost to the host. Aside

from its wide use as a measure correlating negativelywith fitness traits, we have no direct evidence that wingFA is costly to the host. Wing symmetry, however, is atrait likely to have experienced stabilizing selection fora long period of time leading to canalization andreduction of developmental flux (Mller &Pomiankowski 1993). The flat relationship betweenwing size and FA in both experiments supports thisnotion (linear regressions: Experiment I, F

","!*'l

0.421,p0.5; Experiment II, F","#*l0.325,p0.5).

Biomechanical insect models show that the lift gen-erated per wing is proportional to the square of winglength (Brodsky 1994). Therefore, even small asym-metries will require correction to maintain steady

flight, with a cost to flight performance or energeticefficiency. Wing and tail asymmetries are reported asdetrimental to avian flight performance (Balmford etal. 1993).

Further exploration of correlations between viru-lence and mode of transmission can be taken fromwithin hosts. For parasites using vertical and horizontaltransmission the correlations between virulence andassociated traits can pose acute problems. Selectionacting on traits favouring one mode of transmissionmay produce a correlated response in virulence thatdetrimentally affects the parasites success in the othermode of transmission (Bull 1994). Spore production by


8/23/2019 Agnew 1997 A

6/7


E. aedis is an example of this problem. Horizontaltransmission success is known to increase with uni-nucleate spore densities in the aquatic environment(Hembree 1982; Hembree & Ryan 1982). A directcorrelation between binucleate spore density andvertical transmission success has not been demonstrateddirectly. However, they are not motile and rely onbeing carried around in the hosts haemocoel to theoocytes, suggesting that increased densities will lead toearlier and greater probabilities of transmission success.

Therefore, we predict there is a positive selectionpressure for increased spore production common toboth modes of transmission. Against this, increasedspore production should also entail greater costs to thehost. Selection in the two modes of transmission for costto the host are opposite; high virulence is required tokill the host for horizontal transmission, while it shouldbe minimized for vertical transmission.

Where no conflict occurs between selective forcesacting on spore production in horizontal transmission,there were positive and significant correlations betweenthe number of uninucleate spores and FA in bothexperiments. However, conflict occurs between trans-

mission success and virulence of vertical transmission.No significant correlations were found between FA andbinucleate spore number in either experiment. Theseanalyses were performed on residuals once treatmenteffects were removed from spore numbers and asym-metry. Additionally, the results are not explainable byvolumetric differences in the two spore types ornumerical arguments in total spore number, asinteractions between the spore types were not sig-nificant.

We propose that the differential spore costs ex-perienced by the host lie within an adaptive de-velopmental mechanism of the parasite. Uninucleate

spores are only produced in the fat body cells (Becnelet al. 1989). These cells perform numerous functionsbut are particularly active at the time of pupation andhave a major role in protein metabolism (Christophers1960). Disruption of this cell population probably hasseveral developmental consequences for the fitness ofthose individuals surviving to adulthood. This suggestsa possible causal link between stress and FA, as thecuticle is generally between 5075% protein (Chap-man 1969). In contrast, binucleate spores are onlyproduced in oenocytes (Becnel et al. 1989). Thesesecretory cells are predominantly involved in thetiming of larval developmental events (Christophers1960). Disruption of these cells seems most likely to

alter larval rates of development, but has little directeffect on adult characteristics.

Previous results have reported reduced fecundityand longevity of E. aedis-infected adult Ae. aegptifemales (Becnel et al. 1995; Hembree 1982). Oocytedamage during binucleate spore germination wasinterpreted as a possible cause (Becnel et al. 1995).However, female blood-feeding behaviour declinessharply as uninucleate spore number increases, but isunaffected by binucleate spore number (Koella &Agnew 1997). This can account for both reducedfecundity and longevity. Uninucleate spore presence inadults represents E. aedis caught in transition between

its mutually incompatible modes of transmission. Wesuggest that reductions in a range of adult fitnesscharacteristics arise from a suite of effects resulting inthe damage caused to host fat body tissue byuninucleate spore production, and that E. aedis showsadaptation to specifically avoid such effects whenpursuing vertical transmission.

In summary, we present relationships between aparasites mode of transmission and its effects on hostdevelopmental stability. Horizontal transmission is

associated with higher virulence than vertical trans-mission. Where transmission and virulence are not inconflict there are positive correlations between sporenumber and cost to the host. When conflict occurs,with vertical transmission, the parasite appears tomaintain its fitness by partially uncoupling thecorrelation between spore number and cost. Themechanism for uncoupling this relationship may lie inspecific developmental events of the parasite in hosttissues of different functional roles.

We thank Jimmy Becnel andcolleagues forsupplying parasitematerial and advice, Werner Rudin for supplying mos-quitoes, and anonymous referees for comments that improved

an earlier manuscript.

REF ERENCES

Anderson, R. M. & May, R. M. 1991 Infectious diseases ofhumans: dnamics and control. Oxford University Press.

Andreadis, T. G. 1994 Host range tests with Edhaardia aedis(Microsporida: Culicosporidae) against northern Nearticmosquitoes. . Inertebr. Pathol. 64, 4651.

Balmford, A., Jones, I. L. & Thomas, A. L. R. 1993 Onavian asymmetry: evidence of natural selection forsymmetrical tails and wings in birds. Proc. R. Soc. Lond.

B 252, 245251.

Becnel, J. J., Garcia, J. J. & Johnson, M. A. 1995 Edhaardiaaedis (Microspora: Culicosporidae) effects on the repro-ductive capacity of Aedes aegpti (Diptera: Culicidae). .

Med. Entomol. 32, 549553.Becnel, J. J., Sprague, V., Fukuda, T. & Hazard, E. I. 1989

Development of Edhaardia aedis (Kudo, 1930) N.G., N.Comb. (Microsporida: Amblyosporidae) in the mosquitoAedes aegpti (L.)(Diptera: Culicidae). . Protoool. 36,119130.

Brodsky, A. K. 1994 he eolution of insect flight. OxfordUniversity Press.

Bull, J. J. 1994 Perspective: Virulence. Eolution 48,14231437.

Bull, J. J., Molineux, I. J. & Rice, W. R. 1991 Selection ofbenevolence in a hostparasite system. Eolution 45,875882.

Chapman, R. F. 1969 he insects: structure and function.London: The English Universities Press.

Christophers, S. R. 1960 Aedes aegpti (L.) the yellow fevermosquito.Itslifehistory,bionomicsandstructure.CambridgeUniversity Press.

Clayton, D. H. & Tompkins, D. M. 1994 Ectoparasitevirulence is linked to mode of transmission. Proc. R. Soc.Lond. B 256, 211217.

Ebert, D. & Herre, E. A. 1996 The evolution of parasiticdiseases. Parasitol. oda 12, 96101.

Ewald, P. W. 1994 Eolution of infectious disease. New York:Oxford University Press.

Harvey, P. H. & Pagel, M. D. 1991 he comparatie method in


8/23/2019 Agnew 1997 A

7/7


eolutionar biolog. Oxford Series in Ecology and Evolution.Oxford University Press.

Hembree, S. C. 1982 Dose-response studies of a new speciesof per os and vertically transmittable microsporidianpathogen of Aedes aegpti from Thailand. Mosq. es 42,5561.

Hembree, S. C. & Ryan, J. R. 1982 Observations on thevertical transmission of a new microsporidian pathogen ofAedes aegpti from Thailand. Mosq. es 42, 4955.

Herre, E. A. 1993 Population structure and the evolution ofvirulence in nematode parasites of fig wasps. Science, Wash.

259, 14421445.Hurst, L. D. 1993 The incidences, mechanisms and evol-

ution of cytoplasmic sex ratio distorters in animals. Biol.Re. 68, 121194.

Koella, J. C. & Agnew, P. 1997 Blood-feeding success of themosquito Aedes aegptidepends on the transmission route ofits parasite Edhaardia aedis. Oikos. (In the press.)

Mller, A. P. 1996 Parasitism and developmental instabilityof hosts: a review. Oikos 77, 189196.

Mller, A. P. & Erritze, J. 1996 Parasite virulence andhost immune defence: host immune response is related tonest re-use in birds. Eolution 55, 20662072.

Mller, A. P. & Pomiankowski, A. 1993 Punctuated

equilibria or gradual evolution: fluctuating asymmetryand variation in the rate of evolution. . theor. Biol. 161,359367.

Mller, A. P. & Swaddle, J. P. 1996 Deelopmental stabilitand eolution. Oxford University Press.

NIH Image 1.52: http:\\rsb:\\info.nih.gov\nih-image\

SAS Institute Inc. 1994 3.1.5. SAS Institute Inc. Cary, NC.Sokal, R. R. & Rohlf, F. J. 1995 Biometr, the principles and

practise of statistics in biological research. San Francisco: W. H.Freeman and Co.

Sprague, V., Becnel, J. J. & Hazard, E. I. 1992 Taxonomy

of phylum Microspora.ritical Re. Microbiol. 18, 285395.Sweeney, A. W., Doggett, S. L. & Gullick, G. 1989

Laboratory experiments on infection rates of Amblosporadxenoides (Microsporidia: Amblyosporidae) in the mos-quito ulex annulirostris. . Inertebr. Pathol. 53, 8592.

Van Valen, L. 1962 A study of fluctuating asymmetry.Eolution 16, 125142.

Va!vra, J. & Maddox, J. V. 1976 Development of the Micro-sporidia. In omparatie pathobiolog, vol. 1 (ed. L. E. J.Bulla & T. C. Cheng), pp. 87109. New York: PlenumPress.

Receied 5 August 1996; accepted 5 September 1996.

Proc R Soc Lond B (1997)

Documents

Agnew 1997 A