Diversity and plasticity in mosquito feeding patterns: a meta-analysis of ‘universal’ DNA diet studies

Although mosquitoes can have innate preferences for particular blood-meal hosts, their realised feeding patterns on different host species can be modified under climate and land use change, with implications for disease spread. It is therefore important to understand the niche breadth of vectors and to what extent shifts in feeding patterns are predictable. We investigated global shifts in feeding patterns among different functional and taxonomic groups of host species for six prominent disease-vectoring mosquitoes.Time period: 2000–2019. Major taxa studied: six disease-vectoring mosquito species: Aedes aegypti, Ae. albopictus, Anopheles funestus, An. gambiae, Culex pipiens, and Cx. quinquefasciatus. Focusing on blood meal studies that used universal molecular methods, we compiled evidence from >15,600 blood-meals for the six mosquito species. We estimated mosquito’s host niche breadth, and we used hierarchical Dirichlet regression models to investigate shifts in feeding patterns in relation to human and livestock density, land use, and climate gradients. We estimated host ranges of 179–321 species for each of the two Culex mosquitoes and 24 - 65 species for Aedes mosquitoes, comprising considerably broader host niche breadths than previously anticipated. We found some evidence that shifts in feeding patterns among different host functional and taxonomic groups were associated with environmental conditions such as temperature and livestock density, while our results also demonstrate that, with the currently available evidence, global predictions of shifts in mosquito feeding patterns are challenged by considerable uncertainty. Our global metaanalysis afforded first insights into the shifts of feeding patterns in variable environments, suggesting that host choice is not a simple function of host availability, but contingent on other environmental drivers. Improving resolution and consistency of data gathering and reporting will improve the precision of how blood-meal studies can inform us of present and potential risks of pathogen transmission events.

Methods
We compiled blood meal data from the literature for six focal mosquito species: Aedes aegypti, Aedes albopictus, Anopheles gambiae, Anopheles funestus, Culex pipiens s.l., and Culex quinquefasciatus. These are of considerable interest as vectors of infectious diseases and feed to various extents on humans. We conducted a systematic literature search that reported counts of mosquitos feeding on different blood-meal hosts. Our initial search (August 2022) was broad and afforded us a “bird’s eye view” of the literature. For this, we used Web of Science, Pubmed, Scopus, GoogleScholar online databases with the search terms ‘bloodmeal*’, ‘blood meal’, blood-meal’, ‘feeding’, ‘habit’, ‘pattern*’, ‘preference*’, ‘interaction*’, combined with the terms ‘mosquito*’, ‘vector*’, ‘vector-host’, ‘host*’, ‘vertebrate*’, ‘animal*’. The asterisk (‘*’) was used to search for possible variations of terms. Our subsequent systematic search of the literature (July-September 2023) focused on Web of Science and involved two steps. The first used the phrase ‘bloodmeal*’ OR ‘blood meal’ OR 'blood-meal’” and the names (including synonyms) of the focal mosquito species. If the search returned too many irrelevant hits (>200), then additional ‘methods’ search terms were used (refer to Table S1 in associated publication). Another search used publications of popular PCR primer sets as a search term (refer to Table S1 in associated publication). If the search returned too many irrelevant hits, then the number of linked articles was refined using the search term ‘mosquito*’ (refer to Table S1 in associated publication). These searches resulted in 1,495 published articles that we subsequently filtered with the following criteria:

1) The arthropod vector was one of six focal species

2) PCR primers used to identify blood-meal hosts were ‘universal’ (for vertebrates/mammals/birds) and not targeted to particular species

3) Either Sanger (dideoxy) sequencing or high-throughput DNA sequencing is used at some point to confirm host identity

4) Spatial information was provided as either geographical coordinates or salient location names and sampling sites were not pooled over spatial extents >50km in diameter

5) Host species identity data were available

6) For any given mosquito species and sampling site, there must be ≥7 bloodmeals (corresponding to an ~80% chance of detecting an uncommon host present in ~20% of samples).

This resulted in 86 studies that covered 121 different study sites (some studies provided data for multiple sites, refer to Figure S1 in associated publication for location map, refer to Table S4 in associated publication for references) with data collected between 2000 and 2019. From these, we generated a database of geographical coordinates, counts of blood-meals, and the host species (resolved to smallest taxonomic unit) for each mosquito species. Counts of blood-meal origins were assigned to both hosts in the sixteen studies that recorded mosquitos feeding on multiple hosts (e.g., ‘human + cow’). All extracted scientific species names were checked and aligned to the Integrated Taxonomic Information System (ITIS) database, using the taxize 0.9.1 R package.

For studies with missing geographical coordinates, we identified coordinates for location names at smallest available administrative units using various online maps and search engines. For studies with multiple sites/coordinates from a single region, we computed the average of coordinates for use in analyses.