relationship between microfilarial intake and mortality

331

Density-dependent processes in the transmission of human onchocerciasis: relationship between microfilarial intake and mortality of the simuliid vector M. G. BASANEZ13*, H. TOWNSON 2 , J. R. WILLIAMS 3 , H. FRONTADO 4 , N. J. VILLAMIZAR 4 and R. M. ANDERSON 3 1

Department of Biology, Imperial College of Science, Technology and Medicine, Prince Consort Road, London SW7 2BB, UK 2 Liverpool School of Tropical Medicine, Pembroke Place, Liverpool L3 5QA, UK 3 Wellcome Centre for Epidemiology of Infectious Disease, Department of Zoology, University of Oxford, South Parks Road, Oxford OX\ 3PS, UK *Centro Amazonico para Investigacion y Control de Enfermedades Tropicales, C.A.I.C.E.T. Puerto Ayacucho, Estado Amazonas, Venezuela (Received 7 August 1995 ; revised 24 January and 28 March 1996; accepted 28 March 1996) SUMMARY

In order to construct an analytical model of onchocerciasis transmission, it is necessary to elucidate the functional relationships of the various population rate processes taking place within the human and vector hosts. Two previous papers have explored the evidence for density-dependent regulation in relation to microfilarial intake by, and larval development within, the Simulium host. This paper investigates the survivorship of wild-caught blackfly samples fed on subjects with different intensities of Onchocerca volvulus microfilarial infection. Analyses were based on data for Guatemalan S. ochraceum s.l. (possessing a well-developed cibarial armature), West African iS. damnosum s.l. (forest species), and South Venezuelan S. guianense (the latter two lacking a toothed cibarium). The mean survival times of samples of the 3 species, kept under laboratory conditions, decreased as parasite intake increased, the rate of mortality being dependent on the fly's age (measured as time post-feeding) and on the worm load acquired. An empirical, timedependent hazard function was fitted to observed death rates/fly/day which rose very shortly after engorgement, declined subsequently, and rose again throughout the extrinsic incubation period of the parasite. The parameters of this hazard model were all positively correlated with the density of microfilariae in the bloodmeal. Expressions of survivorship and life-expectancy as explicit functions of time post-feeding and mean parasite intake were derived. The average expectation of life at engorgement for uninfected flies in the laboratory was estimated to be around 1 week for both, armed and unarmed blackflies. Residual life-expectancy decreased with time post-feeding and microfilarial load in both categories of vectors. This decline (resulting from age- and parasite-dependent mortality rates) was much more pronounced in those species lacking a toothed fore-gut. Whilst a fraction of heavily infected 5. ochraceum was able to survive the latent period of the parasite, being therefore potentially capable of transmitting the infection, equivalent worm loads in 5. guianense resulted in a drastic reduction of the expectation of infective life. These results provide additional evidence to support the hypothesis that, in the case of intrinsically susceptible vectors, unarmed simuliids are more efficient at low microfilarial loads, when the transmission rate from human to vector host is higher, and parasite-inducedflymortality is negligible. The opposite takes place in armed flies, which perform poorly at low parasite burdens and better at heavier loads, with little parasite-induced vector death. Key words: Onchocerca volvulus, Simulium, cibarial armature, age-dependent fly survival, parasite-induced vector mortality.

INTRODUCTION

Vector life-expectancy is an important factor influencing the success of parasite transmission. In the case of arthropod-borne pathogens, some insect hosts must survive longer than the time taken by the parasite to develop to the infective stage post* Corresponding author. Department of Zoology, University of Oxford, South Parks Road, Oxford OX1 3PS. Tel: 0186S 281221; Fax: 01865 281245. E-mail: [email protected].

ingestion in an infected feed, in order to be transmitted to the vertebrate host during successive bloodmeals. This requirement, termed the 'longevity factor' within the definition of vectorial capacity, is one determinant of the basic or case reproductive number of vector-borne diseases (Macdonald, 19526; Garrett-Jones, 1964; Dye, 1994). The original definition of vectorial capacity (Macdonald, 19526; Garrett-Jones, 1964), used in many subsequent studies of the dynamics of vectortransmitted infections, was based on a series of

Parasitology (1996), 113, 331-355 Copyright © 1996 Cambridge University Press

M. G. Basanez and others

simplifying assumptions concerning the population biology of the vector species. Among them is the premise that once flies become infective they remain so for life. This accounts for the sensitivity of parasite transmission success to small changes in vector life expectancy (Birley, Walsh & Davies, 1983), and possible fluctuations of the case reproductive number above and below unity in value (the level required for parasite persistence) according to seasonal changes in vector survival rates. Another assumption is that vector mortality is unaffected by insect host age, parasite load or infection status (Dye, 1992). The significance of such simplifications on the estimation of transmission success has been examined using data on vector longevity from field and laboratory observations. There is apparently conflicting evidence concerning the relevance of both age-dependent survival and parasite-induced mortality (the two may be related) for given vectorparasite associations. Increasing death rates with insect age have been observed in laboratory cohorts or in wild-caught population samples maintained in the laboratory (Kershaw, Chalmers & Duke, 1954; Kershaw, Chalmers & Lavoipierre, 1954; Lavoipierre, 1961; Townson, unpublished observations), as well as in the field (Gillies & Wilkes, 1965; Samarawickrema, 1967; Clements & Paterson, 1981), whilst constant survival rates have been recorded in natural settings (Russell & Rao, 1942; Macdonald, 1952a; Lines, Wilkes & Lyimo, 1991). Similar contradictions have emerged from studies of parasite-induced effects. Experimental infection studies, and some field observations of mosquito hosts of malaria and filariasis, suggest that there is increased mortality of infected insects, with higher death rates being experienced by those individuals more heavily burdened (Sinton & Shute, 1938; Wharton, 1957; Lavoipierre, 1958; Christensen, 1978; Gad, Maier & Piekarski, 1979). The time postinfection at which parasite-induced vector mortality is more likely to occur seems to be related to the presence of those parasite stages involved in invasion of, or migration through, the insect's body, i.e. ookinete penetration of mid-gut epithelium; oocyst rupture and subsequent migration of sporozoites in plasmodia-infected mosquitoes (Klein et al. 1982, 1986; Maier, Becker-Feldman & Seitz, 1987), and microfilarial passage out of the abdomen, active infective larvae (L3) movement within the fly, or L3 release through the mouth-parts in filariasis vectors (Kershaw, Lavoipierre & Chalmers, 1953; Townson, 1971; Lindsay & Denham, 1986; Ellrott, 1987). On the other hand, studies of naturally infected, wildcaught mosquitoes, either fail to find conclusive evidence of parasite-induced effects (Krafsur & Garrett-Jones, 1977; Lines et al. 1991), find it indirectly (Lyimo & Koella, 1992), or suggest that their operation may affect only the very small

332

proportion of the population harbouring high parasite loads (Laurence, 1963; Samarawickrema & Laurence, 1978). For onchocerciasis vectors, constant survival rates calculated from the proportion of parous flies in wild population samples, have been adopted in order to compare the longevity of savanna and forest Sitnulium damnosum populations, and hence to make inferences about their ability to transmit Onchocerca volvulus (Le Berre, 1966; Le Berre et al. 1964). The same assumption underlies estimations of vectorial capacity for different populations of several members of the S. damnosum complex (Birley et al. 1983 ; Dye & Baker, 1986; Renz, Barthelmess & Eisenbeiss, 1987). Although this premise has been very rarely tested, the analysis of laboratory studies of blackfly mortality could prove useful in elucidating the impact of age and parasite load on fly death rates. Concerning the latter, feeding experiments carried out with various simuliid species, in which the flies have been engorged on carriers with different intensities of O. volvulus infection, result in higher mortalities of those insects fed on the most heavily infected subjects. Increased death rates are associated with ingestion of large numbers of microfilariae and occur predominantly within the first 24 h post-engorgement. In simuliids without the protective effect afforded by a well-developed fore-gut armature, e.g. S. damnosum s.I. (Duke, 1962 a, 1966); S. metallicum s.l. (Omar & Garms, 1977); 5. guianense (as 5. pintoi) (Takaoka et al. 19846), and S. exiguum s.l. (Duke, 1970; Collins et al. 1992 a), the phenomenon is more pronounced than in those species with prominent cibarial teeth, e.g. S. ochraceum s.l. (De Leon & Duke, 1966; Omar & Garms, 1975); S. haematopotum (Takaoka et al. 1984a); S. oyapockense s.l., and 5 . incrustatum (as »S. limbatum)

(Shelley et al. 1987; Shelley, 1991). A similar pattern has been recorded for vectors of lymphatic filarial worms with or without armed cibaria (Zahedi & White, 1994). In two previous papers evidence for densitydependent processes in the transmission dynamics of onchocerciasis has been explored in relation to microfilarial uptake by, and larval development within, the blackfly host (Basanez et al. 1994, 1995). The overall aim of these and the present study is to identify those population rate parameters which control the dynamics of O. volvulus life-cycle in Simulium in order to construct an analytical model of onchocerciasis transmission. Parasite-induced vector mortality is thought to play an important part in regulating such dynamics (Dietz, 1982). The demonstration of its existence and the investigation of the conditions under which it is more likely to occur are relevant to control strategies that rely mainly or solely on measures against the microfilarial stage of the parasite. Ivermectin-based chemotherapy campaigns are increasingly being adopted in West

333

Parasite-induced mortality in onchocerciasis vectors

African and Latin American endemic countries often in the absence of anti-vectorial programmes. This paper focuses on the results of experimental work carried out to examine the survivorship of simuliid vectors infected per os with varying microfilarial burdens of O. volvulus (the 'risk factor')• In these studies the variable of interest is the survival time of blackfly population samples exposed to different dosages of the parasite. First, non-parametric tests are used to compare the patterns of survival observed in groups of flies fed on subjects with increasing intensities of skin infection. These methods make no assumptions about the underlying distribution of survival times that would originate from specific processes of hazard (mortality) and survivorship among these flies. Second, the daily mortality rates experienced by these groups of insects, with differing infection loads, are calculated. Resulting trends between death rate and time postfeeding are quantitatively explored through parametric models. Third, the life-expectancy of these groups of flies at the time of engorgement is estimated non-parametrically (from mean survival times and the daily proportion of simuliids surviving from the start of the experiment), and parametrically (from an expression in which fly survival is an explicit function of time after the bloodmeal and mean microfilarial load). In this way, we investigate several Onchocerca-Simulium combinations that include important onchocerciasis vectors with and without well-developed cibarial armatures. In the former category is the main vector in Guatemala, S. ochraceum s.L, whilst in the latter are S. damnosum s.l. from West Africa, and S. guianense from South Venezuela. Analyses are carried out upon published results and original data sets. The new data have been collected during recent field studies in Latin America.

MATERIALS AND METHODS

Sources of data As in Basafiez et al. (1994, 1995), analyses are presented on 5. ochraceum s.l. from Guatemala (data of De Leon & Duke, 1966 and Collins et al. 1977); S. damnosum s.l. from West Africa (data of Duke, 1962a from western Cameroon), and S. guianense from South Venezuela (this work). However, detailed published data on simuliid survival according to different microfilarial intakes were available only for the homologous forest combination O. volvulus-S. damnosum s.l. (S. mengense and 5. squamosum according to WHO, 1987). Data on the mortality of flies infected with heterologous, incompatible forms of the nematode, i.e. forest S. damnosum s.l. with savanna O. volvulus from West Africa (Duke, 1966), and Guatemalan S. ochraceum s.l. with either West African forest or savanna

parasites (De Leon & Duke, 1966), are not considered here. Consequently, survival analyses of S. damnosum s.s./S. sirbanum have still to be undertaken. Parasitological and entomological procedures All the results reported in this paper are based on feeding experiments, in which samples of wild flies of the main local vector species were fully fed on human carriers presenting with different microfilarial skin burdens and who, consequently, infected the flies engorged upon them with different average parasite loads of the local population of O. volvulus. Controls were obtained by engorging flies on uninfected volunteers (S. ochraceum s.l., S. damnosum s.l.), or very lightly infected subjects (S. guianense). Since all the studies here analysed have relied unavoidably on using wild simuliids, the sampled populations constitute a mixture of insects of uncertain and variable age. Hence, observed patterns of survival and mortality are explored in relation to time post-feeding, a variable which certainly representsflyageing but which is applied to a blackfly sample of unknown age-structure at the time of engorgement. Another source of heterogeneity in this type of study is that fly populations caught on different days, seasons, or places, could have very different age-compositions and, consequently, different survival potentials. To reduce this variation, we focus on the results of groups of experiments in which the participants were exposed simultaneously to simuliid bites at the same places and times of the day. Here, these experiments have been analysed together for each blackfly species. Assessment of the mean microfilarial intake perfly.In the different studies, in an attempt to reduce heterogeneity of parasite intakes, samples of wild flies were fed on selected body regions of the various volunteers whose local microfilardermic levels had been previously determined by skin-snipping. A subsample of the flies fed on each subject was dissected within 12 h post-engorgement (p.e.) in order to score the total number of ingested microfilariae (mff). The methods for assessment of skin microfilarial load, fly collection and dissection, as well as the reasons for the timing of the dissections to record microfilarial intake have been described in detail by Basafiez et al. (1994). Results comprise mean values obtained either from different individual carriers or from local microfilarial densities along the body of those carriers. Fly maintenance and assessment of survivorship. The remaining specimens collected from each participant were maintained in captivity according to protocols similar to those described by Basafiez et al. (1988). Flies were kept at high relative humidity and

334

M. G. Basanez and others

temperatures of 22-27 °C (De Leon & Duke, 1966), 20-26 °C (Collins et al. 1977), 23-27 °C (Duke, 1962a), and 22-28 °C (this work); they were checked every 12 or 24 h p.e., and dead insects removed and counted. The number of flies remaining alive at the start of each day p.e., N(t), and the daily fraction of flies surviving from the moment of blood-feeding, S(t), were calculated. In some studies the experiments were interrupted by killing those flies alive at the beginning of the day marking the completion of the extrinsic incubation period of the parasite (7-9 days p.e.) in order to assess the mean infective larval output per fly. These flies, lost to follow-up, are said to constitute censored observations. When the simuliids were not killed (uncensored), corresponding average loads of successful larvae per fly were estimated from simultaneous feeding experiments performed with the same carriers and conditions (Duke, 19626; Basanez et al. 1995). In experiments with censoring, the numbers of insects exposed to the 'risk factor' at the time flies were withdrawn for dissection were the numbers alive at the beginning of that time interval minus half the number lost to follow-up. Otherwise, the numbers at risk were equal to the numbers entering each day p.e. Analysis of data Measures of the mean numbers of microfilariae ingested and of successful larvae per engorged fly. The various data sources examined here differed in their methods of calculating average numbers of parasites per fly in the survival experiments, with Duke (1962