The impact of climate on the abundance of Musca sorbens, the vector of trachoma

To assess the extent to which climate may affect the abundance of Musca sorbens, a putative vector of trachoma. Studies were identified by systematically searching online databases including CAB abstracts, Embase, Global Health, Medline, Web of Science and BIOS Online, references from key articles, and the websites of relevant international agencies. A systematic literature review was conducted of field and laboratory studies that reported the impact of climate factors (e.g., temperature, humidity) on the synanthropic fly Musca sorbens. Data were systematically extracted and studies assessed for quality by two readers. Study results were reported narratively. A total of 16 studies met the inclusion criteria but only three evaluated associations between climatic/abiotic factors and M. sorbens. Limited evidence indicates that M. sorbens abundance has an optimal temperature and humidity range. Thirteen studies reported seasonal patterns but no consistent pattern was found between season and the abundance of M. sorbens. The evidence base regarding the effect of climatic factors on M. sorbens is limited, so it is difficult to construct a biological model driven by climate for this species. A multivariate statistical approach based on the climate of sites where M. sorbens is found may better capture its complex relationship with climatic factors as well as aid in mapping the global range of M. sorbens.


Background
Trachoma, caused by the bacterium Chlamydia trachomatis, is the world's leading infectious cause of blindness. It is thought to be endemic in 51 countries, primarily in sub-Saharan Africa (Fig. 1) [1]. It is estimated that 232 million people living in trachomaendemic areas are at risk of the condition; of the over 1.8 million who are visually impaired due to trichiasis (in-turned eyelashes) and its complications, over 500,000 people are irreversibly blind. International efforts are underway to map affected populations and scale-up interventions to eliminate blinding trachoma as a public health problem [2][3][4].
Chlamydia trachomatis can be transmitted directly from person-to-person via contact with infected ocular or nasal secretions, or indirectly from fomites via contact with items such as washcloths and bedding. However, eye-seeking fliessuch as the 'bazaar fly' Musca sorbens (Diptera: Muscidae) in Africa and Asia, the 'bush fly' M. vetustissima in Australia, and eye gnats Hippelates spp. and Liohippelates spp. in Latin Americahave also been implicated in trachoma transmission to varying degrees [5][6][7][8][9]. Female flies feed on human ocular and nasal secretions in order to obtain the nutrients necessary for egg production, so are thought to play a substantial role in fly-related mechanical transmission of trachoma [10]. However, the relative importance of flies in trachoma transmission is likely to vary by geographic location, socio-economic development levels, and time of year. Evidence for a role of M. sorbens in the mechanical transmission of trachoma comes from both observational studies in which high fly densities on or around human faces have been associated with trachoma as well as intervention studies in which reductions in M. sorbens densities have been associated with a lower incidence of C. trachomatis infection [5,7]. For example, in The Gambia, a randomised controlled trial demonstrated that, insecticide spraying or the provision of pit latrines was associated with a lower incidence of trachoma compared to no intervention [7]. In Tanzania, another trial of insecticide spraying after mass distribution of antibiotics led to a 39 % lower prevalence of trachoma after six months spraying; although this was of borderline significance, after 12 months there was no difference in trachoma prevalence despite consistent reductions in M. sorbens abundance [11]. The relative importance of M. sorbens as a trachoma vector is likely to vary between geographic locations and change over time.
The geographical distribution of M. sorbens overlaps with areas of active transmission of trachoma, as there are areas where M. sorbens exists which are not endemic for trachoma (Fig. 1) [12]. A recent systematic literature review of climatic effects on the prevalence, distribution and severity of trachoma concluded that the current distribution of trachoma is associated with higher temperatures, and the prevalence is lower at higher altitudes [13]. This is consistent with what one would expect if flies facilitate transmission [9].
As has been shown for other insect species, climate affects several stages in the life-cycle of M. sorbens (Fig. 2) [14][15][16]. The vectorial capacity of M. sorbens is dependent on the survival of adult flies, the proportion of flies carrying infectious C. trachomatis, the number of flies that land on a face, and the proportion of these flyhuman contacts that leads to infection. The development, survival and activity of flies is likely to be associated with ambient temperature and humidity. There is also an optimum range of temperature and humidity for the fly, above and below which the fly will fail to thrive (Fig. 3) [17][18][19][20]. Climate may also influence the abundance of M. sorbens through effects on environmental mediators, such as access to faeces for egg production, breeding, and maturation. Adult females preferentially oviposit on human faeces close to the ground, and the number and size of emerging offspring are greater for human than herbivore faeces, suggesting that human faeces provides the optimal conditions for fly development [6,8,21,22]. High temperatures combined with low humidity could dry out faeces or lead to crust formation on faeces, with the potential to limit adult emergence [23]. Climatic conditions that favour the growth of dung beetle populations, such as warm and wet weather, could lead to a reduction in M. sorbens; since dung beetles can remove dung from the soil surface within several hours [24].
The purpose of this study was to review the availability and quality of the evidence for the impact of climate factors on M. sorbens abundance, and to identify gaps in the literature on this under-researched vector.

Methods
A systematic literature review was conducted to evaluate the evidence for an association between climatic factors and the abundance of M. sorbens. Additional file 1: Table  S1 describes the search terms, key words, and MESH headings used. The following outcomes in relation to M. sorbens were included: 1) abundance of adult flies; 2) fitness (size, life span/longevity, basic reproductive number and parity); 3) adult fly behaviour (e.g., daily activity times, feeding, mating, and oviposition); and 4) life cycle (i.e., development and maturation of eggs, pupae, and larvae). Climate-related exposure terms included: meteorological factors (temperature, rainfall, humidity, sunlight), weather events (e.g., drought, floods), and the related abiotic exposures for laboratory based studies (e.g., temperature, humidity, radiation). As M. sorbens is largely exophilic, fly count data from catches made outdoors are reported [25].

Inclusion criteria
Peer reviewed journal articles published between 1 January 1950 and 1 August 2013; field studies (observational, crosssectional, intervention or spatial) regarding the effect of climate factors on the abundance of M. sorbens; laboratory studies reporting an association between temperature, humidity, or other abiotic factors and M. sorbens outcomes, and studies describing seasonal patterns of fly counts.

Exclusion criteria
Review articles and conference abstracts.
There were no restrictions on geographic location of study, or language of journal publication.
Structured searches were conducted in the following standardized electronic databases: CAB, abstracts, Embase, Global Health, Medline, Web of Science and BIOS Online. Reference lists of selected articles were examined to find any other relevant papers. Limited searches were conducted among the following expert sources: World Health Organization (WHO) Special Programme for Research and Training in Tropical Diseases (TDR): http://www.who.int/tdr/en/; United Nations Children's Fund: http://www.unicef.org/; The

Data extraction and analysis
The review was conducted by two readers (JB and AR). An initial search was conducted of each database for all fly species implicated in trachoma transmission (see Additional file 1: Table S1). Titles were scanned to ensure that articles related to at least some aspect of the study question. Titles and abstracts were then reviewed to exclude papers that did not meet the inclusion criteria. Full texts of papers provisionally included were then obtained and inclusion criteria checked by the second reader. Data were then extracted using a standard format to collect data on study setting, design and target population, the nature of the study, climate factors and how they were measured, fly outcomes and how they were measured, and whether there were any statistical tests of association.

Quality assessment
A modified version of a quality assessment tool developed for quantitative studies was used to assess the quality of each study [26]. Each study was assessed as providing strong, moderate or weak evidence depending on methods of sampling, duration of sample, sample size, minimization of bias, control of confounding, and exposure assessment. Studies that met the criteria were assessed as having high, moderate, or poor quality. Studies graded to be of poor quality included studies where only descriptive data on fly abundance were presented.
It was anticipated that heterogeneity in study designs and exposure measurements would preclude a metaanalysis; thus, findings were compiled and reported using a narrative synthesis.

Results
From 11,130 that were initially recovered, a total of 16 studies met the criteria for our review (Fig. 4). Of the original 11,130 studies initially recovered, 10,833 were removed as duplicates or completely non-topical. Of remaining 297 abstracts screened for full review, 252 studies were excluded because they lacked data on flies, climate variables, or both, and one article was not based on primary data.
The 16 studies included in this review were heterogeneous with respect to outcome measures and methods used. Most studies were from previous decades, and were undertaken in a wide range of locations around the world. Most were field studies, and only one laboratorybased study was identified. Table 1 describes the two studies that were assessed to be of highest quality, Table 2 details the one study assessed to be of moderate quality, while Table 3 provides information on the thirteen studies deemed to be of low quality while still     providing some possible insights related to seasonal effects on flies. Three studies examined associations between distribution of flies and climatic factors [9,22,23].

Temperature
Toyama and Ikeda (1981) studied the role of climate factors on flies in a field study in Hawaii (Table 1) [23]. Summer season was significantly associated with greater abundance of all life stages of M. sorbens. The study also found that the crusts on dung pats at higher summer temperatures reduced predation, most likely by Labidura riparia, an earwig, which was unable to penetrate the pats to hunt fly larvae. In contrast, during the winter months, no such crust formed on the pats, and the earwigs depressed the population of M. sorbens.  (Table 1) [9]. Over 99 % of M. sorbens were caught outdoors and only 0.7 % of M sorbens were caught on children living above 3,000 m. At altitudes <2000 m there was less seasonal variation in the median number of M. sorbens caught per child, ranging from 7/ child in the dry, winter season to 13/child in the wet, summer season. A different seasonal pattern and range was reported at higher altitude (2200-2,500 m) with the number of M. sorbens caught on children's faces ranging from 2.5/child in the wet, summer season to 23/child in the dry, winter season. Findings were similar from M. sorbens caught in beef-baited traps. At <2000 m there Other details: Traps baited with vinegar and sugar; deployed 9 am-12 pm and 3-4 pm.
were higher catches in all four seasons with not much seasonal variation (402 in the dry winter compared with 611 in semi-dry spring) compared with higher altitude (2200-2500 m) where the lowest counts were in the wet summer season (11 M. sorbens) and highest in the semiwet autumn season (305 M. sorbens). The authors comment that at altitudes <2000 m the annual temperature ranges from 23°C in the wet season to 36°C in the dry season and from 5 to 18°C above 3,000 m but details were not published. There were no tests of association between season, altitude and M. sorbens counts.

Relative humidity
Hafez and Attia (1958) explored the impact of relative humidity and temperature on different aspects of M. sorbens life cycle (egg incubation period, egg hatching rate, duration of the larval and pupal periods, and egg structure (humidity only) in laboratory studies (Table 2) [22]. At 100 % humidity, egg hatching occurred across a wide temperature range, with the optimal temperature for hatching being 28°C. At 70 % humidity, adult emergence was maximal between 24 and 28°C, but did not occur above 40°C. When temperature was maintained at 31-32°C, eggs did not hatch at humidities below 80 %, but 100 % of eggs hatched at 100 % humidity (Fig. 5 and 6). None of these associations were assessed statistically.
The three higher quality papers demonstrate that, amongst the climatic factors studied, temperature appears to be the most strongly associated with M. sorbens abundance. Consistent with other Muscids, like M. domestica and M. autumnalis, M. sorbens have an optimum temperature for survival and behaviour, after which increasing temperatures are associated with a decrease in fly abundance [27][28][29]. Importantly, each species have different optimal conditions for their survival. The paper by Toyama et al. presents a plausible explanation why M. sorbens no longer thrive at high temperatures (e.g., high temperature leading to crust formation on dung pats, thereby hindering fly emergence) [23].
The remaining 13 studies are described in Table 3 [21,[30][31][32][33][34][35][36][37][38][39][40]. These studies are mostly descriptive in nature, and report the relative abundance of M. sorbens in relation to other fly species, by season or month. Additional information on the climate for each location was obtained to see if there was consistency across the climate factors that described the peak fly season.

Discussion and conclusion
This review of potential associations between climatic factors and M. sorbens was conducted in order to provide evidence on the climatic boundaries beyond which M. sorbens is unlikely to be a vector of trachoma. Outside these climatic constraints, transmission of C. trachomatis is more likely to be through person-to-person contact rather than via flies. Consequently, in such conditions, interventions should focus on facial cleanliness, other behaviours, and reductions in overcrowding. It was envisioned that insights gained could also assist in planning future trachoma control strategies in the face of possible global climate change. However, this systematic literature review indicates that evidence for the effect of climate factors on the abundance and seasonal activity of M. sorbens is very limited.
One reason for the lack of evidence regarding climatic factors which affect M. sorbens is that the fly has only recently been implicated as an important route of transmission in trachoma [7]. In contrast, the vectors of many other infectious diseases (e.g., malaria, dengue, leishmaniasis and trypanosomiasis) have been studied for years in order to improve vector control methods. Maps of disease vectors generally lag behind those of the diseases themselves, although the recently-produced Malaria Atlas provides   [21,22]. 3. Pupal phase duration was independent of humidity at humidities studied by Hafez & Attia 1958 [21, 22] a comprehensive, well-funded project mapping the distribution of the forty one dominant malaria vector species [41]. The effect of climatic factors on the biology of malaria vectors has also been studied, allowing for the development of models that forecast possible effects of climate change on the distribution of malaria vectors and hence vectorial capacity [42,43]. Whilst the Trachoma Atlas is currently being compiled to map the distribution and endemicity of trachoma, at present there is no equivalent map of the distribution of M. sorbens and the effects of climatic factors on the abundance and distribution of this fly remain poorly understood in comparison to the state of knowledge on malaria vectors [3].
Although there is still some uncertainty about the relative importance of transmission by M. sorbens, results of this review support the implementation of the SAFE (S = surgery, A = mass antibiotic distribution, F = facial cleanliness to reduce ocular secretions and, E = environmental hygiene) strategy, which is being led and scaled-up by the WHO Alliance for the Global Elimination of Blinding Trachoma (GET 2020) with the aim of eliminating blinding trachoma by the year 2020 [32]. In particular, the aim of the 'E' component of the strategy is to reduce C. trachomatis transmission by means of environmental improvements such as water and sanitation; increased access to water could improve hygiene, while reductions in open defecationwhich provides ample, open breeding sites for M. sorbenscould result in reductions in trachoma [6,44].
This systematic review had several limitations. First, given the dearth of moderate and high quality evidence related to the effects of climate factors on M. sorbens, it was difficult to interpret the results of the very limited evidence base. The sixteen studies included in this review employed a wide variety of designs, methods, and analytical tools. Only one study provided evidence from a laboratory setting, and this study was conducted nearly 60 years ago, when methods and analytical techniques were less advanced than at present [22]. The highest quality papers were both conducted within the last 30 years and used statistical methods to provide evidence of the strength of associations [9,23]. Unfortunately most studies reviewed did not provide statistical evidence. None of the studies used robust designs, and none provided details on all of the entomological methods used (e.g., trap designs, speciation methods, etc.). These limitations mean that most studies were categorized as being of poor quality and the results were difficult to interpret.
There is a need for more evidence on the relationship between climatic factors and M. sorbens distribution and abundance which could be used to guide whether the F and/or E elements of the SAFE strategy need to be prioritized. For example, interventions in locations where climatic conditions promote M. sorbens abundance all the year round need to focus on all elements of the SAFE strategy. In locations where climatic factors promote seasonal variation in the abundance of M sorbens targeted vector control could be considered. In locations where climatic factors are not conducive to M. sorbens the F elements need to be promoted to reduce transmission via fomites.
This review found a dearth of data linking climate variables with important life-history traits of M. sorbens. Such data are typically used to construct mathematical models that allow investigators to explore the impact of climate change on the risk of vector-borne diseases. One alternative approach that is used commonly by biologists where detailed data on a species is lacking is to produce maps based on the climate at sites where a species was recorded i.e., presence only data [45]. Machine-learning techniques and community models, that are well suited to sparse occurrence data, are then used to map a species' distribution based on climate factors. The statistical relationship that best captures the climate envelope for the fly can be used to explore future distributions when it is linked to future climate change scenarios [46]. Such an approach would require collating data on the distribution of this species from museum collections, record lists and active surveys.
An alternative approach is to build mathematical models that describe the relationship between the life stages of M. sorbens and climate based on what is known for other Muscids. For example the relationship between climate and the bionomics of the house fly, M. domestica, and the face fly, M. autumnalis, have been described in more detail [27][28][29]. These models could be used to help produce models for M. sorbens where there are few data.
Overall, M. sorbens is adapted to hot tropical climates and can withstand highly desiccating environments. It breeds most actively in hot humid conditions, but this relationship is likely to be non-linear with extremely high temperatures reducing fly survival. Indirect effects of climate on fly abundance are difficult to predict, but higher temperatures and low humidity may create a hard crust on stools harbouring immature flies, restricting the emergence of the young flies and preventing predation of the immature stages. During the wet season dung beetles may remove surface faeces so quickly that flies may have to compete for a severely limited resource resulting in higher numbers of flies per stool with density-dependent effects coming into play and further reducing the number of flies emerging. Perhaps most importantly we are uncertain whether flies carry Chlamydia on the surface of the fly or internally. If the pathogen is found largely on the body of a fly it will be more vulnerable to the effects of desiccation, whilst if carried internally it will be most sensitive to changes in temperature. Whilst many of these suggestions are conjectures, they do illustrate the potential complexity of the system and how difficult it is to predict how climate factors influence the risk of trachoma transmission.
More data on how climatic variables influence the current abundance of M. sorbens will allow forecasts to be made concerning the potential impact of future climate change, indicating areas where the role of M. sorbens as a vector of trachoma may vary in importance over time.

Additional file
Additional file 1: Table S1. Search Terms for Systematic Literature Review. (DOCX 13 kb)