Plasmodium falciparum infection rates for some Anopheles spp. from Guinea-Bissau, West Africa

Presence of Plasmodium falciparum circumsporozoite protein (CSP) was detected by enzyme linked immunosorbent assay (ELISA) in a sample of Anopheles gambiae s.s., A. melas and A. pharoensis collected in Guinea-Bissau during October and November 2009. The percentage of P. falciparum infected samples (10.2% overall; confidence interval (CI): 7.45-13.6%) was comparable to earlier studies from other sites in Guinea-Bissau (9.6-12.4%). The majority of the specimens collected were identified as A. gambiae which had an individual infection rate of 12.6 % (CI: 8.88-17.6) across collection sites. A small number of specimens of A. coluzzii, A. coluzzii x A. gambiae hybrids, A. melas and A. pharoensis were collected and had infection rates of 4.3% (CI:0.98-12.4), 4.1% (CI:0.35-14.5), 11.1% (CI:1.86-34.1) and 33.3% (CI:9.25-70.4) respectively. Despite being present in low numbers in indoor collections, the exophilic feeding behaviors of A. melas (N=18) and A. pharoensis (N=6) and high infection rates observed in this survey suggest falciparum-malaria transmission potential outside of the protection of bed nets.

Malaria is among the leading causes of childhood mortality in Guinea-Bissau, comprising 18% of mortality of children less than five years of age as of 2010 (WHO, 2010). However, the human malaria incidence rate in Guinea Bissau varies considerably from year to year with a general decrease in recent years to about 3 children (<5 yrs of age) per thousand in some locations (Ursing et al., 2014). Plasmodium falciparum predominates, causing 98% cases, followed by a few cases of Plasmodium malariae and Plasmodium ovale. Mixed infections of P. malariae, and to a lesser extent P. ovale, have been recorded but appear to be rare and highly variable in both Guinea-Bissau (Snounou et al., 1993) and neighboring Senegal (Fontenille et al., 1997a;Fontenille et al., 1997b).
Limited research has been conducted on the vectors and malaria parasite infection rates in Guinea-Bissau populations of Anopheles species in general and there is no data on comparative infection rates between A. gambiae and A. coluzzii and members of the A. gambiae complex. Variability is also high among the Anopheles spp. implicated as vectors in this region of West Africa in terms of both their temporal population dynamics as well as species composition among study sites (Carnevale et al., 2010;Fontenille et al., 1997a;Jaenson et al., 1994;Snounou et al., 1993).
Here we present much needed data on P. falciparum infection of Anopheles spp. specimens collected from inside and around associated human habitations at eight sites in Guinea-Bissau (Table 1).

Amendments from Version 1
An adjusted Wald confidence interval was calculated for overall infection rate data where needed. The data presented in the table are not replicated but a simple calculation of the percent of the total collection that were infected. We did not replicate within time or locations. Therefore there is no variation to report in the table.
Response to the reviewer (Dr. Charlwood) concerning the ethics of outdoor collection was addressed in our response to reviewer.
Regarding false positives, this is a possibility for any test of parasite detection. We specifically chose to use the P. falciparum circumsporozoite protein sensitive ELISA test on the parts of the mosquito most likely to contain the infective stages (head and thorax) to minimize false positives in the results. Unfortunately the ELISA test was destructive and these specific parts of the mosquito have long been discarded. The DNA from the remainder of each mosquito has been archived but this contains the abdomen which presents another potential source of false positives as the gut contents may contain material that may never progress to rendering the mosquito infective.
Another confounding factor associated with using this type of testing is that it needs to be conducted more rapidly than the DNA extraction from the rest of the mosquito and the ELISA tests were often performed before the mosquito identification via PCR could be conducted. Such that an unusual result such as the one observed here would not have been detected until the ELISA samples had been discarded. It would definitely be beneficial to keep this in mind for future work in this area. Revised method clarifying this and discussion of the results are added to the revised manuscript.
Other minor edits that two reviewers suggested were also made.

Method
Mosquitoes were collected by mouth aspiration from both the island and inland areas of Guinea-Bissau ( Figure 1) in 2009 between October and November, which corresponds with the time of year previously observed to have the highest infection rate in Anopheles species (Jaenson et al., 1994). All mosquitoes were collected indoors with a single exception; samples in Ponta Anabaca were opportunistically collected outside while host-seeking at about 18:00hr. Each mosquito was dissected and the head and thorax were preserved in 100% ethanol for subsequent ELISA analysis. The remainder of each mosquito was preserved in 70% ethanol for genomic DNA extraction using the DNeasy extraction kit (Qiagen). Species determination of mosquitoes from the A. gambiae complex was made with the combination of species diagnostic assays (Fanello et al., 2002;Favia et al., 2001;Santolamazza et al., 2008;Scott et al., 1993) and a divergence island SNP (DIS) genotyping assay (Lee et al., 2014a) while other species were identified by morphology.
The results of the ELISA were analyzed for both CSP concentration, adjusted for plate-to-plate variation, with an analysis of variance and for a binary outcome using a χ 2 test implemented in SPSS 16.0 (SPSS, 2007). The data were analyzed for differences between species and among collection sites, using a G-test implemented in Deducer library under R software (http://www.r-project.org/). A Confidence interval (CI) was calculated using adjusted Wald confidence intervals using an online calculator (www.measuringu.com/ wald.htm). Mosquito species. P. falciparum infection state and CSP concentration for each individual are provided in Dataset 1.   et al., 1989;Snow et al., 1988). All species were collected at multiple sites except A. pharoensis, which was only collected at the more inland site of Bambadinca. A. pharoensis is not generally considered a significant vector in West Africa but the distribution observed in this study matches the previously observed pattern in Senegal (Carrara et al., 1990). Anopheles arabiensis was absent from collections.
No significant differences were observed for CSP concentration or in the analysis of positive samples with χ 2 . This is probably due to the variation in the distribution of vector species and P. falciparum in the environment at the time of sampling. In this study, a few A. pharoensis (N=6) were collected, half of which were CSP positive. Other studies in this region of West Africa have found that A. funestus and A. arabiensis may also be important vector species at different times in nearby Senegal (Fontenille et al., 1997a;Fontenille et al., 1997b). A. arabiensis was not collected in our study while a small number (N<10) of A. funestus were observed but not collected.
Recent studies on the prevalence of malaria parasites in humans have suggested that infection rates in Guinea-Bissau may be in decline due to widespread use of effective treatment and insecticide treated bed nets (ITNs and long lasting insecticide treated bed nets, LLINs) by the most high-risk groups (Rodrigues et al., 2008;Ursing et al., 2014). The malaria parasite life cycle is complicated and may not directly relate to the prevalence of human cases but it is possible that the lack of data during periods of political unrest has concealed a more stochastic pattern than was previously observed in Guinea-Bissau (Ursing et al., 2014).
Outdoor mosquito collection was not the focus of this survey and was only made at Ponta Anabaca Hotel grounds when we fortuitously noted mosquitoes biting. Consequently no general comments about the degree of exophily of A. gambiae in Guinea-Bissau can be made. However, evidence of exophily by the major malaria vector A. gambiae in this study and by others in West Africa (Reddy et al., 2011;Tchouassi et al., 2012) raises the concern of the long term effectiveness of Indoor Residual Spraying (IRS) and Long lasting Insecticide-treated Nets (LLINs) in reducing outdoor transmission of malaria especially before bedtime and by people sleeping outdoors. The relatively high infection rate of 11.1% of A. melas in Guinea-Bissau together with its tendencies to be both endophilic and exophilic and have a high human blood index (Sharp et al., 2007;Tuno et al., 2010) make the species a significant vector, which may also be hard to control by reliance on ITNs and LLINs.
The high CSP rate of 33.3% in the 6 indoor collected A. pharoensis might implicate a significant role in malaria transmission in drier inland Guinea Bissau, however this should be viewed with caution due the small sample size. This contrasts the recent finding in Mozambique where none of the 4390 A. pharoensis samples were positive for CSP-ELISA (Charlwood et al., 2013). Very low infection rates and absence of malarial parasites, traditionally found in West and Central African populations of A. pharoensis has always led to the conclusion that this mosquito plays little role in malaria transmission despite its anthropophilic habits and that it can be easily experimentally infected (De Meillon, 1947;Ndiath et al., 2012;Tchouassi et al., 2012). In drier Sahel regions of Africa where the major vectors of malaria are absent or very rare and irrigated rice and other crop lands are increasing, A. pharoensis is considered more important at maintaining low levels of malaria (Kerah-Hinzoumbé et al., 2009;Kibret et al., 2010). More rigorous sampling effort and infection confirmation using multiple approaches (ELSA + SNP genotyping assays (Lee et al., 2014b) will be required to confirm the definitive role of this species in malaria transmission in this region.
Author contributions YL, GCL and AJC conceived the study, designed experiments and conducted field collections. JD conducted field collection. AR provided logistical support and coordination for field collection in Guinea-Bissau. CCN, CDM, AMW and SH conducted DNA extraction, ELISA and PCR. MRS performed data analysis and wrote manuscript. All authors were involved in the revision of the draft manuscript and have agreed to the final content. This "observation article" is very well written in a format that is accessible to both general and specialist audience. It describes some novel observations about malaria infection rates in different vector species in Guinea-Bissau. In particular it reports the some how unexpected high infection rates in Anopheles melas and , thus suggesting -malaria transmission potential outside of A. pharoensis Plasmodium falciparum protection (i.e. bed nets).

Open Peer Review
Details about the circumstances of the finding and evidence of the observation are properly provided. The manuscript appropriately cites relevant bibliography in the field. Figures and tables are informative and helpful. Methods section is well organized and nicely descriptive.
As observational article it looks perfectly adequate to the journal purpose.
I have only a very minor concern: in the Introduction (line 8) should be re-written as Plasmodium malaria . Plasmodium malariae I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
No competing interests were disclosed. This paper provides information on sporozoite rates from a relatively unstudied area the mainland and islands of Guinea Bissau. It gives the impression that it is a spin off from another study that perhaps aimed at characterizing the genetics of populations of from the mainland and the islands, Anopheles gambiae perhaps for future genetic control efforts. Given the widespread use of mosquito bednets rates are exceptionally high and not, apparently different from rates recorded earlier. The authors do make the comment that this might be the result of civil strife in Guinea-Bissau but whatever the cause this is disquieting and implies that the gains in reduction of malaria are going to be at best temporary.
The data are presented without confidence intervals but these should be added. Given the relatively small numbers involved either adjusted Wald confidence intervals (that can easily be calculated using the site www.measuringu.com/ .htm or a routine in R) can, I think, be used. (But since I am signing this review wald everyone should know that my statistical abilities are limited! The kind of collection undertaken needs to be explained in more detail. Were the mosquitoes collected resting or were they landing collections? I do not really want to be the person raising this issue but something on ethics should be included somewhere. (My own thoughts on ethics in general is that if the rule of 'first do no harm and second maybe do some good' is followed then a study -that may include even landing collections -is not unethical.) This is especially important if the collections were ad hoc landing collections.
To avoid possible misunderstanding, the sentence 'Four species were collected during sampling; A. , , and .
x There are a number of small errors in the paper that need to be rectified. For example in the last paragraph they state '33% of the 4 collected indoors when they either mean 33% Anopheles pharoensis of the six Anopheles pharoensis collected or 50% of the four collected indoors. With regard to this species it may be worth pointing out that in Mozambique none of the 4390 tested were positive for sporozoites ( ) but at the same time in Ghana, ( ) found that two of Charlwood 2013 et al., Dzodzomenyo , 1999 et al. three specimens of examined were infected (with Bancrotian filariasis) and one of these An. pharoensis was infectious. Given the possibility of false positives among primarily zoophilic anophelines (that may also include ) and given that the authors have access to a sophisticated laboratory it is a shame An. melas that they did not run a PCR on the sporozoite positive specimens to ensure that they were indeed human malarias.

I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
No competing interests were disclosed.

Yoosook Lee
Thank you very much for your review. Revisions to the manuscript will be made as suggested.
Thank you very much for your review. Revisions to the manuscript will be made as suggested.
WIth respect to you comment about CI, the data presented in the table are not replicated but a simple calculation of the percent of the total collection that were infected. We did not replicate within time or locations. Therefore there is no variation to report in the table.
About the concern related to outdoor mosquito collection, our only outdoor collection is from Ponta Anabaca where the hotel was located where we stayed. While we were processing our sample collections in the early evening at the hotel, we (Drs. Cornel, Lee and Lanzaro) were harassed by mosquitoes, which we identified as Anophelines by morphology. Although this was unplanned ad landing collections, we were all taking anti-malaria prophylaxis at the time. hoc About the concern you raised about false positive of ELISA, this is a possibility for any test of parasite detection. We specifically chose to use the . circumsporozoite protein P falciparum sensitive ELISA test on the parts of the mosquito most likely to contain the infective stages (head and thorax) to minimize false positives in the results. Unfortunately the ELISA test was destructive and these specific parts of the mosquito have long been discarded. The DNA from the remainder of each mosquito has been archived but this contains the abdomen which presents another potential source of false positives as the gut contents may contain material that may never progress to rendering the mosquito infective.
Another confounding factor associated with using this type of testing is that it needs to be conducted more rapidly than the DNA extraction from the rest of the mosquito and the ELISA tests were often performed before the mosquito identification via PCR could be conducted. Such that an unusual result such as the one observed here would not have been detected until the ELISA samples had been discarded. It would definitely be beneficial to keep this in mind for future work in this area.
No competing interests were disclosed. Competing Interests:

Yoosook Lee
The infection rate is conservative (using 99% CI for calling uninfected samples). The infection rate estimates are higher with 95% CI.
I am an author of this paper. Competing Interests: