Changes in Relative Species Compositions of Biting Midges (Diptera: Ceratopogonidae) and an Outbreak of Oropouche Virus in Iquitos, Peru

Abstract

Species compositions of Culicoides paraensis (Goeldi) (Diptera: Ceratopogonidae), the major vector of Oropouche virus to humans in Central and South American urban cycles, and Culicoides insinuatus Ortiz & Leon differed along a northeast-to-southwest transect across Iquitos, Department of Loreto, Peru. The relative distributions of the species were consistent with patterns of human outbreaks along the Amazon River. We resumed collection of biting midges between May 2000 and January 2004 at three sites previously sampled (1996–1997) to determine whether the known vector was expanding its range relative to the earlier survey. C. paraensis did not replace C. insinuatus across the region surveyed. Instead, C. insinuatus dominated the more southern sites and significantly increased its relative proportion at all three sites. Apparently, microhabitat differences and not range expansion by C. paraensis were responsible for differences in species compositions across the sample sites.

disease vector, tropical Culicoides, arbovirus, Oropouche virus

Biting midges (Diptera: Ceratopogonidae) are vectors of viral and filarial pathogens to humans and other mammalian hosts (Blackwell 2004). Oropouche (ORO) virus has emerged as the most important human disease with biting midge vectors in the Western Hemisphere. ORO virus is a member of the Simbu serogroup in the family Bunyaviridae that causes a self-limited, dengue-like illness accompanied by prostration of 2- to 7-d duration (LeDuc et al. 1981, Wirth and Felippe-Bauer 1989). ORO virus is noteworthy due to its high incidence during outbreaks and in causing one or more cycles of symptoms after recovery in some patients.

After viral isolation during 1955 from a charcoal maker in Trinidad (Anderson et al. 1961), ORO virus caused epidemics in South America beginning in 1961 in Belém, Pará State (Brazil), at the mouth of the Amazon River; outbreaks recurred during 1967, 1968, 1975, and 1978 (Pinheiro et al. 1976, LeDuc and Pinheiro 1989). Additional outbreaks occurred in Mazagao, Amapo Territory (upriver, adjacent to the Amazon River delta) in 1980 and Manaus, Amazonas State (≈1,300 km upriver along the Amazon at the confluence with Rio Negro) in 1980–1981 (LeDuc and Pinheiro 1989). Subsequent epidemics occurred in Maranhão (coastal Brazil, south of Pará) and Goiás (inland, south of the Amazon) states in 1988 (Vasconcelos et al. 1989) and Iquitos (≈2,900 km from Belém), Department of Loreto, Peru, in 1992 (Watts et al. 1997). This epidemiological pattern suggests that the disease spread upstream along the Amazon River (LeDuc and Pinheiro 1989).

Culicoides paraensis (Goeldi), which is widespread from northern United States to Argentina (Wirth and Blanton 1973), has been identified as the primary vector of ORO virus to humans in Neotropical urban cycles (Dixon et al. 1981, Roberts et al. 1981, Wirth and Felippe-Bauer 1989). C. paraensis adult numbers correlated well with human infection levels at several outbreak sites (Roberts et al. 1981, LeDuc and Pinheiro 1989), adults transmitted ORO virus in laboratory studies (Pinheiro et al. 1982), and virus was isolated from field-collected adults (Vasconcelos et al. 1989).

As part of a study of the biology of putative vectors, biting midges were captured in Iquitos during 1996–1997. Collections by sentinel humans indicated that C. paraensis was the most common biting midge, representing 55.1% of all host-seeking ceratopogonids at 16 sites around Iquitos (Mercer et al. 2003). Culicoides insinuatus Ortiz & Leon was the second most common biting midge (43.8% of captured flies). Culicoides daviesi Wirth & Blanton and Culicoides fluviatilis (Lutz) made up <1% of the 16,503 host-seeking Ceratopogonidae captured during 8 mo. The two dominant species were very similar in temporal biting patterns. Similarly, immatures of both species developed in decaying platano (Musa × paradisiaca L.) stems, fruits, and flowers (which were plentiful throughout the region surveyed) and detritus-rich soil.

Nonetheless, one obvious difference was noted between the two dominant species across the study area. Collection sites northeast of Iquitos (i.e., closer to the mouth of the Amazon) featured a preponderance (i.e., >95%) of C. paraensis, whereas C. insinuates dominated sites southwest of Iquitos that were upriver or along tributaries of the Amazon but <25-km linear distance away. We envisioned two alternatives that would explain the differences in species composition observed along the transect sampled. Perhaps microhabitat conditions (natural or human-induced) across the region surveyed differed enough to favor one or the other species; as long as the microhabitat differences persisted, we expected that the species compositions would not change. Alternatively, one species (presumably C. paraensis) was expanding its range or increasing its dominance at the expense of an earlier established species (presumably C. insinuatus) and our 1996–1997 survey coincided with the edge of range expansion; evidence for this alternative would be a relative increase in C. paraensis in southwestern sites over time.

The latter explanation was consistent with the apparent spread of ORO from Belém upriver to Iquitos. The implications for disease transmission differed between these two alternatives: if C. paraensis was expanding its range, we might expect new outbreaks of ORO deeper into Peru at the rate of vector range expansion, especially due to the close association between C. paraensis and human environments (Hoch et al. 1990). We resumed collections of biting midges to identify which alternative was correct.

Materials and Methods

Blood-seeking biting midges were collected monthly between May 2000 and January 2004 at three sites used in the previous study: Punchana (3° 42.931′ S, 73° 14.308′ W; 130 m above sea level), Santa Clara (3° 46.928′ S, 73° 20.486′ W; 120 m above sea level), and finca de Sabino (3° 49.189′ S, 73° 19.316′ W; 132 m above sea level). Three human sentinels again aspirated all biting midges from exposed ankles between 1545 and 1745 hours. Air temperatures, relative humidities, and barometric pressures were recorded during each collection period. All flies were preserved in 85% ethanol. During August 2004, one of us (D.R.M.) identified all biting midges based upon morphological characteristics (Wirth and Felippe-Bauer 1989) and specimens from the previous study. Because new members of a C. paraensis species complex (Felippe-Bauer et al. 2003) had not been identified before the first survey, all flies that fit the original description of C. paraensis were included with that species to be consistent with the earlier survey.

The relative proportion of C. paraensis was calculated for each site and month of collection. A two-way analysis of variance (ANOVA) was used to test for significant differences in mean proportions C. paraensis among the three sites and among months; all tests involving proportions were performed using arcsine-transformed data (Sokal and Rohlf 1981). Correlations between relative proportion of C. paraensis and date and with weather conditions also were tested for significance by using Statistica (1995). A χ² test was used to test for statistical differences between observed and expected (i.e., based upon relative proportions among biting midges from the 1996–1997 survey) numbers of C. paraensis captured for each site between 2000 and 2004. Finally, the mean numbers (across years 2000–2004) of C. paraensis and C. insinuatus collected per month at each site between January and May were compared with the corresponding single value for that species reported for the same month during 1997 (Mercer et al. 2003) by using a modified two-tailed t-test (i.e., single values compared with means; Sokal and Rohlf 1981).

Results and Discussion

In total, 2,526 biting midges were captured at the three sites during the 45 mo of the survey. The same four species identified in the previous survey were again represented, including C. paraensis (32.4%), C. insinuatus (67.0%), C. fluviatilis (0.3%), and C. daviesi (0.3%). Because C. fluviatilis and C. daviesi represented <1% of biting midges collected, they were not included in calculations, again consistent with the previous study. The mean ± SD of relative proportions C. paraensis among biting midges by month were 0.839 ± 0.308 for Punchana, 0.375 ± 0.355 for Santa Clara, and 0.219 ± 0.239 for finca de Sabino. Proportions of C. paraensis by month differed among the three sites (two-way ANOVA: F_{2, 81} = 27.8; P < 10^-6, n = 117; comparisons performed on arcsine-transformed proportions; months without flies were excluded); Punchana had a significantly higher proportion of C. paraensis per month than Santa Clara or finca de Sabino (post hoc comparisons by Scheffé test). By contrast, there was no difference in proportions of C. paraensis among months of the year (F_{11, 81} = 1.55; P = 013) and the site × month interaction term was not significant (F_{22, 81} = 1.31; P = 0.19).

There was a significant positive correlation (Pearson’s correlation coefficient r = 0.255, P = 0.006) between proportion of C. paraensis collected at a site and the month during the survey (n = 135), which reflected the appearance of biting midges (mostly C. paraensis) at the Punchana site beginning in the 19th month, November 2001 (Fig. 1). Relative proportion of C. paraensis for the month was not significantly correlated with temperature (r = -0.008, P = 0.94, n = 102), relative humidity (r = -0.039, P = 0.69, n = 102), or barometric pressure (r = -0.074, P = 0.460, n = 102).

Fig. 1

Monthly collections of biting midges at three Iquitos, Peru, sites.

Hoch et al. (1990) recorded peaks in biting activity for C. paraensis that partially corresponded with the rainfall pattern in Belém, Brazil. Although precipitation occurs throughout the year in Iquitos, peak rainfall occurs October-May and low-lying areas are flooded December-May. We plotted the numbers of both species against collection date to suggest temporal biting patterns at the three sites (Fig. 1). There seemed to be peaks in biting activity (although not exclusively) between October and December for C. paraensis in Punchana during 2001, 2002, and 2003 and for all years in Santa Clara. Likewise, some peaks in biting activities for C. insinuatus occurred between October and April for Santa Clara and finca de Sabino, but patterns were more variable for this species. However, in comparing mean numbers of C. paraensis and C. insinuatus per month (for individual sites across years), none of the differences was significant (Table 1) due to high variability (e.g., only two ceratopogonids captured at Punchana between May 2000 and September 2001, 248 C. insinuatus during May 2000 at finca de Sabino; Fig. 1).

Table 1

Comparisons of mean numbers of biting midges captured per month at three Iquitos, Peru, sites between May 2000 and January 2004

C. paraensis and C. insinuatus, respectively, dominated Punchana and finca de Sabino as each species had in the earlier survey. However, for Santa Clara, C. insinuatus replaced C. paraensis as the dominant species (Table 2). We used χ² tests to determine whether proportions of C. paraensis and C. insinuatus had changed for each of the three sites compared with the earlier survey. For all three sites, proportions of C. paraensis were significantly lower than expected based upon the earlier survey, reflecting an overall increase in the relative proportions of C. insinuatus at the three survey sites (Table 2).

Table 2

Numbers and relative proportions of C. paraensis among biting midges collected by human sentinels in Iquitos, Peru, between May 2000 and January 2004 and in comparison with collections made between October 1996 and May 1997

Overall, there were significantly fewer biting midges collected during the second survey compared with corresponding months of the earlier survey. For 2000–2004, there were significantly fewer C. paraensis collected during every month at Punchana, during January, February, April, and May for Santa Clara, and during February, March, and May at finca de Sabino compared with those months in 1997 (modified t-test, Table 3). Numbers of C. insinuatus were more similar except for fewer flies collected during March 2000 -2004 at finca de Sabino (Table 3).

Table 3

Comparisons of numbers of biting midges captured by month during 1997 with mean numbers during 2000–2004 in three sites in Iquitos, Peru

These three sites were selected from the 16 included in the earlier survey because they would have clearly demonstrated replacement of C. insinuatus by C. paraensis if that had occurred since the earlier survey. Instead, C. insinuatus increased for the three Iquitos sites between October 1996 and January 2004. Although our entomological data during the initial survey seemed to coincide with human epidemiological data for ORO outbreaks, we did not find an expansion of its major known vector and cannot make predictions about incidence of ORO because the vector status of C. insinuatus is unknown. Movement of the main vector species is not the only possible explanation for the 1992 epidemic of ORO virus in Iquitos. Indeed, C. paraensis was reported from the Department of Loreto as early as 1955 (Wirth and Blanton 1973).

Based upon the results presented here, the difference in species composition from northeast to southwest Iquitos is likely due to microhabitat differences. However, air temperatures at the time of collections did not differ among the three sites (F_{2, 116} = 0.157; P = 0.85; n = 119). Similarly, relative humidities (F_{2, 124} = 0.414; P = 0.66; n = 127) and barometric pressures (F_{2, 114} = 0.021; P = 0.98; n = 117) were equal across sites and the three sites are similar in elevation above sea level. However, the Punchana and Santa Clara sites are within meters of the Amazon and Nanay rivers, respectively, whereas finca de Sabino is further from any permanent stream but near standing pools. It is unclear what microhabitat differences might account for differences in species present because immature developmental substrates and sources of blood for adults are common throughout the region surveyed.

The wide distribution of C. paraensis has led several authors to speculate that it is composed of a species complex. Correspondingly, researchers in North America first identified genetic variation within populations of Culicoides variipennis (Coquillett) sensu lato (Tabachnick 1990; Tabachnick 1992a, b; Holbrook et al. 1996; Schmidtmann et al. 1998) before they recognized species with differences in vector competences for bluetongue virus (Holbrook et al. 2000). Members of this complex may be sympatric as adults (Holbrook et al. 2000) but seem to exploit different aquatic microhabitats during development (Schmidtmann et al. 2000). In South America, Felippe-Bauer et al. (2003) recently described two new morphological species in the C. paraensis complex from the Departments of Amazonas and Loreto, Peru. Descriptions of these species differ from that of C. paraensis in terms of relative body sizes, the absence of a sensory pit and spermathecal morphologies. Although these new species are known to bite humans, nothing is known about their genetic similarities to C. paraensis, developmental habits or efficacies as vectors of ORO. It will be necessary to investigate relative distribution, biting pattern, affiliation with human settlements, and vector status for each of these species and C. insinuatus to understand ORO virus transmission in Iquitos.

Acknowledgements

We thank the collectors and residents of Iquitos for participation in this study. This research was partially supported by funds generously provided by the Department of Biology and the College of Natural Sciences, University of Northern Iowa. Lane Foil and two anonymous reviewers graciously provided valuable comments on an earlier version of the manuscript.