Population dynamics and community structure of Anopheles mosquitoes along the China-Myanmar border

Understanding the ecology of malaria vectors such as species composition and population dynamics is essential for developing cost-effective strategies to control mosquito vector populations. Adult mosquitoes (n = 79,567) were collected in five villages along the China-Myanmar border from April 2012 to September 2014 using the CDC light trap without bait method. Mosquito community structure, Anopheles species composition and diversity were analyzed. Twenty species of Anopheles mosquitoes were identified, with An. minimus s.l. accounting for 85 % of the total collections. Mosquito densities varied from 0.05 females per trap per night (f/t/n) to 3.00 f/t/n, with strong seasonality in all sites and densities peaked from June to August. An. minimus s.l. was predominant (accounting for 54–91 % of total captures) in four villages, An. maculatus s.l. was predominant (71 %) in the high elevation village of Dao Nong, and An. culicifacies accounted for 15 % of total captures in the peri-urban area of Simsa Lawk. All 20 species have been captured in the Mung Seng Yang village, 18 and 15 species in Ja Htu Kawng and Na Bang respectively, and nine species in both Simsa Lawk and Dao Nong. Species richness peaked from April to August. Species diversity, species dominance index, and species evenness fluctuated substantially from time to time with no clear seasonality, and varied greatly amongst villages. Mosquitoes were abundant in the China-Myanmar bordering agricultural area with clear seasonality. Species composition and density were strongly affected by natural environments. The targeted intervention strategy should be developed and implemented so as to achieve cost-effectiveness for malaria control and elimination along the border areas.


Background
Mosquitoes are vectors of various human and animal infectious diseases [1]. Vector control is an important component and sometimes the only effective way (e.g., for dengue fever) to reduce the transmission of these diseases [1][2][3][4][5]. Understanding the species composition and population dynamics of the local mosquitoes is the crucial step for developing and implementing appropriate strategies to control mosquito vector populations and subsequently the mosquito-borne infectious diseases.
Malaria is one of the major mosquito-borne diseases endemic to East and Southeast Asia [6][7][8]. The Greater Mekong Subregion (GMS) is one of the most threatening foci of malaria in Southeast Asia [7][8][9][10]. The malaria burden in Myanmar is the heaviest among the GMS nations. More than half of the malaria cases and ∼ 75 % of the malaria deaths in the GMS occurred in Myanmar [11]. International border regions, such as the one shared by China and Myanmar's Kachin State, have the highest malaria incidence and the highest malariarelated mortality rate, at 7.8 deaths/1000 people [11][12][13]. A large proportion of ethnic minorities live in these remote, often hilly and forested border areas, which are prone to malaria transmission by the forest mosquitoes Anopheles minimus s.l. and Anopheles dirus s.l.. The Chinese side of the China-Myanmar border remains a significant foci of clinical malaria despite significant decline of malaria in other parts of China in the past two decades [14,15].
Effective control of disease vectors requires a good understanding of vector ecology, including vector distribution, development and population dynamics. Recent achievements in malaria control and enthusiasm from international communities have inspired a number of East and Southeast Asian countries such as Thailand and China to redirect program strategies from malaria control towards elimination. There is an increased focus on studying malaria transmission along the China-Myanmar border; however, mosquito vector ecology in this area remains almost untouched [16][17][18][19]. This lack of information hinders progress toward the goals of eliminating malaria and controlling other mosquitoborne diseases in the area.
The purpose of this study is to acquire critical epidemiological information on vector population ecology by conducting field surveys on vector species composition, species richness, diversity, and population dynamics on both sides of the China-Myanmar border. The results will help to fill important knowledge gaps and aid in developing cost-effective malaria control and elimination strategies in the area.

Study sites
This study was conducted at five sites along the China-Myanmar border (Fig. 1) Traps were placed in the same house for two consecutive nights each sampling occasion. A total of 36 traps were deployed each night. Traps shifted from house to house and from village to village over time so that every house in a given village and every village in the study area has been sampled once (2 nights) every two weeks or two times (4 nights) each month. Once collected, mosquitoes were morphologically separated as Anopheles, Culex, Aedes, and other subfamilies or genera.
Anopheles mosquitoes were further morphologically identified to species or species complex in the field when possible [20,21]. Subsample specimen were stored in −20°C freezer immediately after species or species complex was identified and kept frozen until further analysis. The DNA of An. minimus s.l. mosquitoes was extracted from the abdomen of each mosquito for further species confirmation [17]. Anopheles collections from Dao Nong, Na Bang, Ja Htu Kawng, and Mung Seng Yang villages between May 2012 to April 2013 have been used in previous publication [17], the previous stud only published the summary of pooled data of all villages and time frames, and it focused on the role of An. minimus as the malaria vector [17]. In addition to new collections, species diversity, evenness, similarity and their temporal changes were not determined in the previous study. Inter-village heterogeneity, e.g., differences in Anopheles densities, species richness, and diversity, etc. have not been compared.

Data analysis
Density of mosquitoes was calculated as the number of females per trap per night (f/t/n) and monthly average was calculated over all houses in a given village. Pooled Anopheles density was calculated by summing across all Anopheles species. Species richness was measured by the number of species. Species diversity of Anopheles mosquitoes was measured using the Simpson diversity index, D, which can be interpreted as the probability of interspecific encounter, and Shannon-Weiner index, H, which takes into account of individuals of each species [22][23][24][25].
The Simpson diversity index is defined as where p i is the fraction of a species which belong to the i-th species, and N is the number of species. If all species are equally distributed, then the index has its highest value of [1 − (1/N)]; if one species is dominant, then the index will approach zero. Other indices calculated included the Berger-Parker dominance index, d, evenness index, E = H/ Ln(S), where H is the Shannon-Weiner diversity index and S is the total number of species observed at a given village, and the Morista-Hoen similarity index, C [25]. Mosquito abundance data was transformed by the power function ffiffiffi ffi X 4 p , where X is the monthly density, before comparison and this transformation ensured overall data normality as required for data comparison using analysis such as analysis of variance (ANOVA) and Student t-test [25,26]. Differences in monthly average mosquito abundance (time series) among different villages were tested using one-way analysis of variance (ANOVA) with repeated measures after data transformation. Pairwise differences in mosquito abundance were compared using the ANOVA post-hoc Tukey-Kramer HSD test or Hsu's MCB test with a significance level of 5 %. Statistical analysis was carried out using JMP statistical software (JMP 9.0, SAS Institute Inc., USA).

Results
Over the study period, a total of 14,786 trap nights were conducted and 79,567 mosquitoes were collected in the five villages. Among the mosquitoes captured, the majority were Culex (n = 62,828 or 79.0 %), followed by Anopheles (n = 15,410, 19.4 %), Aedes (n = 466, 0.6 %); the remainder belonged to other subfamilies or genera ( Fig. 1). Twenty Anopheles species have been identified (Table 1), An. minimus s.l. accounted for 84.6 % of total Anopheles collections. All other species accounted less than 5 % of total collections. Thirteen of the 20 Anopheles species each accounted for less than 1 % of total Anopheles collections (Table 1). PCR results (n = 425) confirmed that all An. minimus s.l. was An. minimus A.
Anopheles mosquito density varied considerably among study sites (Table 2). Ja Htu Kawng had the highest Anopheles abundance with an average of 3.01 females per trap night (f/t/n), which was significantly higher than any other site; followed by Mung Seng Yang (0.76 f/t/n) and Na Bang (0.42 f/t/n). The high elevation site of Dao Nong (0.13 f/t/n) and the periurban site of Simsa Lawk (0.05 f/t/n) and had the lowest mosquito density.
Anopheles mosquito species composition varied considerably among study sites (Tables 1 and 2 (Table 1).
Na Bang had the highest species diversity, with a Simpson diversity index of 0.67, Shannon-Weiner index of 1.66 and 15 Anopheles species. It also had the highest evenness index (Table 2). Although 18 Anopheles species were detected, Ja Htu Kawng had the lowest species diversity (Simpson 0.18 and Shannon-Weiner 0.50) because An. minimus was so dominant (90.7 %) that other species contributed very little to the diversity, leading to the lowest evenness ( Table 2). Despite very similar species diversity in the other three villages (diversity index 0.44-0.48), all 20 Anopheles species were captured in Mung Seng Yang, whereas only nine species were collected in both Dao Nong and Simsa Lawk villages (Table 2). Interestingly, similarity analysis showed that species composition in Dao Nong was significantly dissimilar to the rest of the villages (Morista-Horn indices < 0.43) while the other four villages had very similar species compositions (Morista-Horn indices range from 0.86 to 0.98) ( Table 3).
Anopheles population densities showed clear seasonality in all villages, with peak months from April to August each year (Fig. 2a). Anopheles population dynamics differed significantly amongst the study sites (repeated measure ANOVA F 4,114 = 33.86, P < 0.0001). Ja Htu Kawng had significantly higher densities of Anopheles mosquitoes than the other villages, and differences in average monthly population densities among other villages was not statistically significant (Tukey-Kramer HSD, q* = 2.77, P < 0.05). Ja Htu Kawng had the highest density of any site in every month surveyed, and the periurban site of Simsa Lawk had more than 60-fold lower Anopheles density than Ja Htu Kawng. Species richness, measured by the number of Anopheles species trapped, also showed strong seasonality in three study sites, Na Bang, Ja Htu Kawng, and Mung Seng Yang, but varied little in the other two sites (Fig. 2b). The periurban site, Simsa Lawk, and the high elevation site, Dao Nong, had the lowest number of Anopheles species, varying little over time; the other three sites had similar number of species all over the study period (Fig. 2b).
Interestingly, different from population density and species richness, species diversity fluctuated over time but did not show any seasonality in all study sites ( Fig. 2c  and d). Mung Seng Yang was the only village where species diversity was relatively stable and high in 2013 and 2014 ( Fig. 2c and d). Similarly, dominance index and evenness indices varied substantially from time to time (Fig 2e and f ). In Ja Htu Kawng and Simsa Lawk, dominance index was sometimes very high, i.e., with one species being dominant (Fig. 2e). In Ja Htu Kawng, dominance index was high most of the time, while evenness index was low in 2013 and 2014 (Fig. 2f ).

Discussion
Aside from Africa, Southeast Asia is the key focusing area of malaria transmission because of its intense P. vivax transmission and mixed infections. In addition, mosquito vectors show greater species diversity in Southeast Asia than in Africa. For example, in Africa, three Anopheles species, An. gambiae s.s., An. funestus s.l., and An. arabiensis are responsible for almost all malaria transmissions [27][28][29]. Conversely, there are more than 20 major malaria vector species in Southeast Asia, which makes vector control extremely difficult because of the complex biology and ecology of the different vectors [29][30][31][32][33][34][35][36][37][38]. Among the 20 identified Anopheles species in this study, l4 of them have been implicated in malaria transmission in East and Southeast Asia [29][30][31][32][33][34][35][36][37][38]. Furthermore, the top four most abundant Anopheles mosquitoes found in this study are all malaria vectors, including one of the key malaria vectors in Southeast Asia, An. minimus s.l., which was also the most abundant Anopheles species in the study area. This species is not exclusively anthropophilic or zoophilic, not exclusively endophagic or exophagic, and not exclusively exophilic or endophilic, therefore, it is extremely difficult to control it with currently available tools [17,39,40]. While other countries have implemented intensive vector-borne disease control efforts and have seen a decline in the prevalence of malaria, the malaria situation in Myanmar is largely unknown [6][7][8]. It is believed that Myanmar is and will remain the key malaria parasite reservoir in Southeast Asia, significantly offsetting the enormous regional efforts for malaria elimination [7][8][9]. Studies found that the majority of the P. falciparum malaria cases in China were imported from Myanmar [14,41,42]. To effectively control malaria in Myanmar and to eliminate malaria in China, it is essential to understand the community structure and species composition, the temporal dynamics, and the spatial distribution of mosquito vectors in this area.
The seasonal fluctuations in vector density and species richness is likely due to the seasonal rainfalls and temperature changes [43,44]. Although there is no clear dry season in the study area, summer time (May to August) is usually the rainy season, when temperatures can reach 25-30°C, which is perfect for malaria   [44,45]. Winter is relatively dry and temperatures drop to an average of around 10-15°C, which is unfavorable to the development of both immature and adult mosquitoes. This climatic-vector interaction is common in Culex, Aedes, and Anopheles mosquitoes, for example, vector population dynamics are strongly associated with climatic conditions, winter diapause or refuge, and water availability affecting mosquito development [46][47][48][49]. This seasonality in vector abundance leads to seasonality of vector-borne infectious diseases [41,42,45,[49][50][51]. Differences in vector abundance, community structure, and Anopheles species composition at different sites may be explained by the contextual determinants, mainly environmental factors. There are many water ponds in Na Bang, Ja Htu Kawng, and Mung Seng Yang villages. Residents in these villages raise a lot of livestock such as pigs and bovines (mainly domestic cattle and water buffalo), providing a perfect environment for Anopheles mosquitoes, especially the zoo-anthropophilic species An. minimus. Larvae of An. sinensis (in Na Bang) and An. minimus s.l. (in Ja Htu Kawng) were frequently found in these villages (GZ unpublished data). On the other hand, in both Dao Nong and Simsa Lawk villages, abundance of Anopheles was very low. Both villages have heavy forest coverage, although with totally different settingsrural in Dao Nong vs. urban in Simsa Lawk, and both have fewer livestock than the other three villages. An. minimus s.l. was dominant in Ja Htu Kawng, but both Na Bang and Mung Seng Yang had a diverse Anopheles composition. These differences in Anopheles abundance and species composition suggest that micro-environmental factors may play an important role in influencing the occurrence and abundance of these species. It is worth mentioning that the result from this study is different from the result of the study conducted in northern Thailand [37], which found that An. minimus s.l. occurs in less disturbed forested area and that deforestation has negative effects on An. minimus s.l. abundance [38]. In our study, An. minimus s.l. was most abundant in deforested areas such as Ja Htu Kawng and Mung Seng Yang villages and it was less abundant in the forested areas such as Na Bang and Dao Nong. In the Thailand study, breeding habitats were mainly found along streams outside the villages, whereas in this study, habitats were mainly standing water ponds within the village (GZ personal observations), we suspect that the species found in the two studies are different. Unfortunately, the Thailand study did not mention details about the An. minimus s.l. species. Most of the Anopheles mosquitoes caught (97 %) were known malaria vectors, including the key malaria vectors in Southeast Asia An. minimus s.l.. Anopheles sinensis is considered the most common malaria vector in China [32], but it is not common in the study area, perhaps because An. sinensis prefers rice paddies and does not like the forest environment. In addition to inter-villages variations in Anopheles densities and species diversity, there is the potential of intravillage heterogeneity, such as house-to-house variations in Anopheles densities and possible spatial clustering, which is open for further investigation.

Conclusion
In conclusion, this study provides fundamental information on mosquito abundance and population dynamics in the China-Myanmar border area. The results indicate that both vector abundance and species diversity vary greatly from village to village and from season to season. The knowledge gained from this study will be useful for designing targeted intervention strategies for malaria control in Myanmar and malaria elimination in China.