The Three Gorges Reservoir is located at North latitude 29°16′∼31°50′, East longitude 106°20′∼110°30′ ^[1]. The area had a history of falciparum malaria and vivax malaria epidemic, while the transmission vectors were only Anopheles sinensis with the density peaking from June to September in recent years ^[2]. The collection of malaria mosquitoes landing on human ‘baits’ is considered the most direct and reliable method for determining human-biting activity since female mosquitoes are collected as they attempt to feed on human collectors ^[3]. However, the human biting catches method is labour intensive and unreliable because of variation ^[4]–^[5] in host attractiveness of collectors. For the estimation of malaria transmission intensity, it is an important prerequisite that the sampling methods used are calibrated against the human biting catches ^[6]. This is because human biting catches translates directly into human biting rates, which serves as an essential parameter in the estimation of both entomologic inoculation rate and vectorial capacity ^[7].

Many sampling methods ^[8] have been evaluated as an alternative to human biting catches with varying degrees of success. The evidence ^[9] that light-traps can provide an estimate of human-biting activity was validated by comparison with human biting catches conducted concurrently. The work of Lines ^[10] suggesting that the number of Anopheles caught by light-traps in East Africa is proportional to that by human bait, represented an important advance. The study ^[11] in Lawanda village of western Kenya also indicated that despite the clear difference in the number of mosquitoes caught by each method, both the Mbita trap and light trap catches were directly proportional to human biting catches regardless of mosquito density. Above all, a series of studies ^[12]–^[14] demonstrated that light-traps may be as a sensitive alternative to estimate human-biting activity of Anophelines.

On the contrary, Mbogo ^[15] claimed that in Kilifi, Kenya, this proportionality was not observed, and the CDC light-trap ^[16] hung close to a human sleeping under a bed net with an incandescent bulb, was not considered a reliable means for estimating malaria vector outdoor biting densities.

However, no studies about the comparative field evaluation of light-trap catches, electric motor mosquito catches and human biting catches had been carried out in China until now. Therefore, we report the results of a parallel series of conventional human biting catches and light-trap catches or electric motor mosquito catches, and our objective was to determine whether the light-trap catches or electric motor mosquito catches may be used in place of human biting catches to monitor the human-biting rate.

Overall, our study was carried out in the Three Gorges Reservoir in four representative villages for 256 nights distributed over sixteen months, with a final comparison between 256 nights for light-trap catches or 512 men-nights for electric motor mosquito catches and 512 men-nights for human biting collections. The number of mosquitoes collected by the light-trap catches, electric motor mosquito catches and the human biting catches method were 6610, 680, and 1640 Anopheles sinensis respectively. The parameter indicate that the light-trap catches, electric motor mosquito catches caught about 403% and 41% of the number of Anopheles sinensis caught in the human biting catches.

Further, scatter distribution for the relationship between light-trap catches or the electric motor mosquito catches and the human biting catches (Fig. 1) indicated that there was consistency in the sampling efficiency between the light-trap catches and the human biting catches for Anopheles sinensis (r = 0.82, P<0.01), while the correlation coefficient between the electric motor mosquito catches and the human biting catches was found to be not statically significant (r = 0.43, P>0.01).

One might wish to go further, and predict what the human biting catches would have been on the night of a given light-trap catches. Altman & Bland ^[20] have pointed out that making such predictions from a linear regression may be misleading. Instead, the ratios between two types of catches against the geometric mean of the two catches were used as a measure of their relative sampling efficiency.

There was no significant tendency for the ratio of light-trap catches to human biting catches to increase with increasing mosquito abundance (r = 0.33, P>0.01), Fig. 2 also shows that the vertical scatter of the observations (i.e., the variance of the log-ratios) shows little or no relationship to mosquito density which means that the variability of the ratio between the catches varied independently of changes in mosquito density. The mean log ratio was 0.1816 (s.e. 0.026). Taking the antilog gives the geometric mean ratio, 1.52 (95% confidence interval 1.35 to 1.71). This means that on average, the catches from the light-traps was 1.52 times that from the human biting catches, and 95% of light-trap catches are expected to lie from 0.35-fold to 0.71-fold greater than the catches with two human collectors outdoors.

Analysis of variance was used to test whether the relative sampling efficiency of the two types of catches varied according to months, or to the villages where the catches were made. No significant biases were found among the months, while significant difference existed in the villages (table 1).

On the contrary, no significant tendency for the ratio of EMC/HBC to increase with increasing mosquito abundance (r = −0.40, P>0.01), while the Pearson correlation coefficient was negative (Fig. 3) which means the sampling efficiency of EMC was below that of HBC at high density.

Both the electric motor mosquito catches and the light-trap catches were intended to provide estimates of the density of human-biting mosquitoes in the trial villages. This study demonstrated that light-trap catches can be used as an alternative to human biting catches of Anopheles sinensis in the study area. The evidence that light-trap catches can provide an estimate of human-biting activity was validated by comparisons with human biting catches. We found that light-trap caught 1.52 times the number of Anopheles sinensis as captured by human biting catches which indicate that light-traps may be a sensitive means to estimate human-biting activity of Anopheles sinensis in this area. Further, the relationship was not affected significantly by changes in the mosquito density, date of sampling the mosquitoes in the study area. Though some studies ^[21]–^[24] showed that light traps catched fewer Anopheles sinensis than human bait, our findings was in agreement with some studies ^[10], ^[25]–^[26] that they catched more mosquitoes than human biting catches. This difference of sampling efficiency may be related to some differing epidemiologic settings, such as the differences in sleeping arrangements, availability of alternative hosts, temperatures, humidity, and wind speed. There is, therefore, a need to standardize the operational conditions and sampling procedures used if valid comparisons between various studies are to be made.

However, the correlation between the electric motor mosquito catches and the human biting catches was not statistically significant, though no density-dependent sampling efficiency was noted for the comparison, their correlation coefficient indicated that the electric motor mosquito catches was less efficient than the human biting catches for sampling Anopheles sinensis at high density in the study area.

Several factors might explain this observation. First, the engorged mosquitoes may escape before being trapped indoors. Second, the malaria vectors in this area are dominated by Anopheles sinensis , a mosquito species that is usually largely exophilic but zoophagic. Third, a questionnaire administered to the study villages indicated that more than one pesticide was applied indoors in every family on hot days. In addition, some residents like to sleep outdoors without protections.

This is the first study for evaluating the performance of the light trap catches relative to the human biting catches for Anopheles sinensis in China, additional field trials are needed to determine whether the relationship between light-trap catches and human biting catches are influenced by differences in host preference, feeding behavior among different epidemiologic settings so that its flexibility and consistency can be fully explored prior to wider application. However, the human biting catches should be maintained as the standard reference method, it is important to perform limited number of human biting catches to re-check whether the light-traps can be relied upon to provide an unbiased measure of density at each new location. There is therefore a need to standardize the use of this method to enable valid comparisons of results from the various studies in different epidemiologic settings.

Nevertheless, at this stage it can be argued that the light trap catches is a promising tool for sampling malaria vector populations and may be very useful for enabling community members in collecting large numbers of samples that are representative of the overall vector population at a less cost.

Competing Interests: The authors have declared that no competing interests exist.

Funding: The authors have no support or funding to report.

Special thanks are extended to Dr. Yang Wei-zhong and Zhang Jing for their help in organizing and coordination of field survey.