INTRODUCTION

The incrimination of the vector or vector(s) that transmit EEE virus is of great importance to mosquito control workers, public health officials and the general public. Discovering this mosquito species or reducing the list of suspected vectors would help in planning control measures regarding EEE transmission. We have gathered evidence which may bring us closer to incriminating the EEE virus vector in southeastern Massachusetts. Our approach was to compare putative epidemic vector populations in EEE epidemic case sites on southeastern Massachusetts.

To ascertain the agent of transmission, we need to understand the life cycle of the pathogen being transmitted. The life cycles of vector-borne pathogens are tenuous in appearance but they are nevertheless successful methods for pathogen persistence and transmission. EEE virus transmission requires an auspicious association of the virus with its mosquito vector, avian host, and dead end vertebrate host. This interaction must manifest itself not only in terms of the life of the virus within its hosts (if indeed a virus can be said to be living) but also in terms of its relationship to the life cycles, ecology and behavior of its hosts and of these to one another. At every component of this complex interaction there is a chance for an impediment for virus survival to arise. A good vector will enable the virus to successfully pass through some of these barriers. What we are interested in, for the purposes of this study, is discovering what mosquito species is able to transmit the virus from avian hosts to humans and horses; what is, in other words, the epidemic vector. The species that were considered to be most likely candidates at the time of this study were: Cq. perturbans, Ae. canadensis, Ae. vexans, Cx. salinarius, An. punctipennis, and An. quadrimaculatus.

The potential epidemic vector or vectors need to have certain characteristics to increase the likelihood of successful virus transmission. One characteristic is vector competence. This refers to the ability of a vector to become infected, to support replication, and finally to transmit the virus from its salivary glands to a vertebrate host. A second characteristic is host preference. In the case of EEE a good epidemic vector should be catholic in its feeding preference. That is, it should feed on birds as well as humans and horses. A third characteristic is field isolations: the identification of EEE virus from pools of field caught mosquitoes. Although not a indication of competence, the isolation of mosquito pools with EEE virus does suggest the involvement of these mosquito species in transmission. A fourth characteristic is flight range. A greater flight range from forest resting sights for host seeking may enhance the chances of serving as a bridge vector between birds and humans or horses, especially when these cases occur at distances away from the forest margin. A fifth characteristic is seasonal overlap. A vector needs to be present at around the time of the season when historical transmission occurred. Most of the recent cases that we have included in our study have occurred from mid-late August. A final characteristic is site overlap. A vectoring species must to be present at the site where transmission occurred. It is these last two characteristics which our study addresses.

The history of EEE in Massachusetts began in 1938 with an epidemic that affected 35 humans and almost 300 horses. Minor outbreaks in horses have occurred periodically since then; but it was not until 1955 and '56 when humans were again affected. Horse cases climbed to 85 in these two years combined. The late 1950's and 1960's spared humans from EEE and horse cases were uncommon. Then in the early 1970s a third epidemic episode surfaced. This one affected over 130 horses and six humans. After this episode horse vaccines became available to horse owners and since then the numbers of horse cases has greatly diminished. The average number of human cases has gone down in comparison to the initial 1938 outbreak but has not varied greatly during epidemics episodes in this century. During the 1938 epidemic, most of the cases occurred in the southeast. Cases seemed to be especially concentrated to Bristol, Plymouth, Norfolk, and Middlesex and Suffolk counties. The 1955-56 epidemic was mostly focused to Plymouth, Bristol, Norfolk and Middlesex counties

All the same counties were involved in during the 1970-73 epidemic. Suburban development may have played a role in reducing risk since the 1950's in Suffolk county (a small county encompassing Boston and its suburbs). Unlike the other maps this only shows the location of confirmed cases. There were over 130 horse cases during this epidemic episode.

MATERIALS AND METHODS

For the purposes of our investigations we employed the most recent cases and virus isolation data to decide upon our study area. Since most of the historical cases have occurred in Plymouth and Bristol counties, we focused our attention to virus activity that had taken place in these two areas from 1982-1993.

Virus activity was observed as three categories: Horse cases, human cases and isolation of EEE virus from mosquito pools. As with all mosquito-borne diseases, it is difficult to know exactly where a vertebrate host acquired the virus. It is reasonable to presume that horse cases are a more reliable indications of EEE virus activity than are human cases since horses do not travel as much as humans. Virus isolations from mosquitoes confirm the presence of EEE in enzootic and putative epidemic mosquitoes. A combination of these three indicators of virus activity strengthens the evidence for a case location.

We located horse and human case sites as well as sites where virus had been isolated through information we obtained from the Massachusetts State Lab and State Veterinarian files. Case sites were chosen for the study only after they were verified by horse owners, human case survivors or family members. Horse cases of our study area had occurred in Brockton, Abington, Whitman, Pembroke, Wareham, Rochester, and Middleboro. Human cases had taken place in Norton, Bridgewater, Kingston, Lakeville, and Wareham. Virus isolations from mosquitoes had occurred near many of these case sites and some cases have occurred in clusters.

Trapping for mosquitoes therefore took place in Brockton, Abington, Whitman, Pembroke, Kingston, Wareham, Rochester, Lakeville, Middleboro, Norton and Bridgewater. There were a total of 15 trap sites.

We used American Biophysics Corporation or ABC traps to sample adult mosquito populations at these sites. These traps were set with a flickering light to go on at dusk and with a CO2 flow rate of 500ml/minute coming from a 10 lb. CO2 source tank. Each of the sites was sampled during two consecutive nights per week from mid-July to mid-September. Individuals were taken back to the University of Massachusetts Cranberry Field Station and identified to species.

For the purposes of simplifying our species comparison we focused on data obtained from mid-July to Late-August.

RESULTS

In order to compare the temporal distribution of each species in the study area, we calculated the Log 10 of the mean number of mosquitoes per trap per night for each species in all 15 sites combined during 3-two week intervals, which we designated as: Mid-Late July, Early-Mid August and Mid-Late August.

The result was that (1) in Mid-Late July species abundance varied in the following order from highest to lowest: Cq. perturbans, Ae. canadensis, Cx. salinarius, Ae. vexans, An. punctipennis, and An. quadrimaculatus. (2) In Early-Mid August the same order was observed but with an increase in populations of Ae. canadensis, Cx. salinarius, and Ae. vexans. (3) Mid-Late August (the time period when most of the recent EEE cases have occurred) demonstrated a decrease in the Cq. perturbans population and relatively no change in population size of the remaining species when compared to the Early-Mid August interval.

To determine if there was a significant difference between species and within the summer season time intervals, we ran a repeated measures ANOVA in SAS. We found that there was a significant difference between the species and within the time period (summer). This means that the difference in population sizes that were observed among the 6 species was significant and that there were significant differences among the time intervals with respect to mosquito populations. The latter was expected as was the finding that there was a significant interaction between species abundance and time. Mosquito populations are known to vary with the summer season.

It was obvious by looking at the seasonal distribution patterns of the mosquito populations at each of the 15 epidemic sites that there was a site variation as well. This is the pattern observed in, for example, Bridgewater, Massachusetts. The most prevalent species in the late summer was Ae. canadensis. This was followed by Cx. salinarius and Ae. vexans.

If we compare this to another site, Norton, the most prevalent species in the late summer was Cx. salinarius (with over 100 mosquitoes for a biweekly average) followed by Ae. canadensis and Cq. perturbans.

To compare the spatial distribution of each species in the entire study area, we ranked the six putative epidemic vectors according to their abundance during each of the three seasonal intervals at each site. We then summed these ranks among all sites per seasonal interval. We then tallied up the ranks to obtain a sum of the ranks for each species throughout the study area.

We found that during the (1) Mid-Late July interval the sum of the ranks per site revealed that the most prevalent species were in order from greatest to least: Cq. perturbans, Ae. canadensis, Cx. salinarius, Ae. vexans, An. punctipennis, and An. quadrimaculatus. (2) This order remains constant during Early-Mid July. (3) During Mid-Late August (the period of greatest interest) Cq. perturbans and Ae. canadensis are equally common to the entire study area. The remaining species do not change with respect to their prevalence.

A ranking system has previously been constructed by John Edman and Rajeev Vaidyanathen that may serve to incriminate the epidemic vector in southeastern Massachusetts. This system includes most of the characteristics that must be held by a mosquito species to be considered as capable of transmitting EEE to horses and humans. (1) Laboratory demonstration of vector competence: 3 = greater than 10%, 2 = between 6-10%, 1 = between 1-5%, 0 = no transmission. (2) Diversity of host preference: 3 = broad host range, 2 = host preference influenced by host availability, 1 = narrow host range, 0 = extremely narrow host range (only ornithophagic or mammalophagic). (3) EEE virus isolates from field caught mosquitoes are ranked as 3 = common, 2 = occasional, 1 = rare, 0 = never. (4) Flight range from forest resting habitats during host seeking: 3 = greater than 1 km radius from forest edge, 2 = less than 1 km radius, 1 = close to the forest margin, 0 = does not leave the forest. (5) Seasonal overlap of host seeking and human disease incidence: 3 = complete overlap, 2 = some overlap, 1 = little overlap, 0 = no overlap.

Under this classification scheme we discern that the six mosquito species can be ranked from most to least likely vectors in this order: Cx. salinarius, An. quadrimaculatus/Ae. canadensis, Cq. perturbans/Ae. vexans, and An. punctipennis.

With our data of mosquito populations at epidemic sites, we can further resolve this ranking scheme by distinguishing species population differences with respect to seasonal overlap and site overlap. By including these two more detailed characteristics the relative importance of these species is altered. Seasonal abundance during the late summer when the most recent epidemic cases have occurred can be ranked as 3 = most abundant, 2 = abundant, 1 = less abundant, 0 = least abundant. Prevalence at epidemic sites during mid-late August can be ranked as 3 = most prevalent, 2 = prevalent, 1 = less prevalent, and 0 = least prevalent.

Under this new scheme the ranking we find that Cq. perturbans, Ae. canadensis, and Cx. salinarius may be relatively similar in relation to the biological, ecological and behavioral attributes that characterize them as epidemic vectors. And they are more likely to be vectors of EEE in southeastern Massachusetts than Ae. vexans, An. punctipennis and An. quadrimaculatus.

DISCUSSION

We have found that there is significant variation in abundance among the 6 species we investigated in the study area. This variation allowed us to narrow the list of potential vectors.

We also recognized variation within the summer season or among the summer intervals that we described. This was expected.

We saw a variation among the sites. This also allowed us to further resolve the list of potential vectors. This site variation leads us to believe that of the three most likely vectors suspected after viewing the entire study area, it may be that different species are causing transmission at different sites.

We observed that all 6 species exhibited seasonal overlap as well as epidemic site overlap. This may pose a problem which will be discussed shortly.

Our ultimate goal has been to direct the focus of mosquito control programs to a reduced number of putative EEE virus vector populations. Although further work is needed in order to understand the behavior, biology and ecology of these six mosquito species in relation to their ability to transmit EEE virus, our study indicates that Cq. perturbans, Ae. canadensis, and Cx. salinarius may be the primary epidemic vectors in southeastern Massachusetts.

There are admittedly problems with our study. One problem is that our study areas is based on the occurrence of cases in the past. We are therefore assuming that local habitat in these areas has not changed dramatically and that the population of mosquitoes has not changed dramatically. Fortunately most of our case sites were where cases occurred in 1990.

A second problem is that of case reliability. As I mentioned before, horse cases may be more reliable than human cases, it would be difficult for most people to say without any doubt where they were when they were bitten by the mosquito that exposed them to EEE virus.

A third problem is how different are epidemic sites from southeastern Massachusetts as a whole. If our intention is to incriminate the vector(s) due to their abundance at epidemic sites similar abundance patterns in all of southeastern Massachusetts are not a problem. If, however, we are interested in using abundance patterns to help us predict potential epidemic sites, then finding similar abundance patterns in the rest of southeastern Massachusetts may be problematic.

Another problem is that we are assuming that the most abundant mosquito populations are the most likely culprits. It seems reasonable to believe that the most abundant species should be encountered more often by horses and humans, but we can still not be sure of which species has served as a vector based in this assumption.

This leads us to our next problem, which is what weight should be given to each of the characteristics we have used to compare our six species? We have weighed them all equally when adding up the ranks. But is this reasonable? Vector competence, for example, should be more important than virus isolation from field caught mosquitoes, as alluded to earlier. And how should the other characteristics be ranked in relation to one another?

In any case these vector attributes do need to be part of the profile of a mosquito species to be considered a potential vector and greater resolution of these attributes needs to take place to be more confident about which species are transmitting EEE virus in southeastern Massachusetts.

ACKNOWLEDGMENTS

John D. Edman Veronique Kerguelen

Ralph Timperi

Geoff Attardo

Barbara Werner

OGIA

Mary Tobin

Meghan Moncayo

Donald Buckley

back to Table of Contents click button to return home