Local extinction of the Asian tiger mosquito (Aedes albopictus) following rat eradication on Palmyra Atoll

The Asian tiger mosquito, Aedes albopictus, appears to have been extirpated from Palmyra Atoll following rat eradication. Anecdotal biting reports, collection records, and regular captures in black-light traps showed the species was present before rat eradication. Since then, there have been no biting reports and no captures over 2 years of extensive trapping (black-light and scent traps). By contrast, the southern house mosquito, Culex quinquefasciatus, was abundant before and after rat eradication. We hypothesize that mammals were a substantial and preferred blood meal for Aedes, whereas Culex feeds mostly on seabirds. Therefore, after rat eradication, humans and seabirds alone could not support positive population growth or maintenance of Aedes. This seems to be the first documented accidental secondary extinction of a mosquito. Furthermore, it suggests that preferred host abundance can limit mosquito populations, opening new directions for controlling important disease vectors that depend on introduced species like rats.


Background
Introduced rats threaten native species (like seabirds), cause economic damage, and can transmit diseases to humans [1]. In response, humans invest billions of dollars in rat control. However, rat eradications have not been linked to mosquito extirpations. This is consistent with the assumption that blood-feeding success does not limit mosquitoes [2]. Under such assumptions, rat declines should increase mosquito bites on humans owing to vector switching (electronic supplementary material). However, because egg laying typically requires a blood meal, it makes sense that mosquito abundance, and perhaps persistence, would depend on blood-feeding success, which itself should increase with host density and suitability [3]. If so, removing preferred hosts such as rats could drive mosquitoes to secondary extinction, thereby reducing bites on humans [4,5] (electronic supplementary material).
Palmyra Atoll has no native mosquitoes, but its wet tropical climate is suitable for the southern house mosquito, Culex quinquefasciatus (hereafter Culex), and the Asian tiger mosquito, Aedes albopictus (hereafter Aedes). Culex was introduced during World War II [6]; it is small and bites at night, feeding mostly on birds, but also on mammals, including humans [7] (figure 1, top-left panel). Through its bites, this Culex species can vector lymphatic filariasis, West Nile fever and Japanese encephalitis. Aedes arrived sometime before 2002, when Chris Depkin collected larvae and adults (Bishop Museum accession no. 2018.003). Adult females are large, aggressive day-time biters, with conspicuous striped coloration, preferring mammals, including rats [7], but some Aedes populations will feed on birds when mammals are not available [7] (figure 1, bottom-left panel). This Aedes species can vector lymphatic filariasis, yellow fever, Rift Valley fever, dengue fever, chikungunya and Zika [8]. Although there have been no documented vectored diseases on Palmyra, both mosquitoes were a nuisance.
In June 2011, the approximately 40 000 rats on Palmyra were eradicated by applying brodifacoum [9]. After rat eradication, rat prey, like palm seedlings and crabs, increased [10]. Although mosquitoes still harassed people in the evenings, visitors found it unnecessary to apply mosquito repellent during the day and began to suspect Aedes had been extirpated.

Material and methods
Palmyra Atoll National Wildlife Refuge (5852 0 N, 162804 0 W) lacks an indigenous human population. The atoll has a saltwater lagoon encircled by two-dozen natural and created islets covered by introduced coconuts or tall native trees and shrubs. Breeding seabirds nest in the native forests, whereas shorebirds use the flats and shoreline. Although there are no native mammals and few native insects [11], humans (especially a US military occupation in World War II) have introduced rats and many plant and insect species.
After rat eradication, we (i) surveyed researchers about when and how often they were bitten by mosquitoes (see electronic supplementary material), (ii) intensified mosquito survey efforts, and (iii) modified a model [5] describing conditions for mosquito extirpation (electronic supplementary material). We completed 53 trap-nights across 25 islets in 2015 and 80 trap-nights across 24 islets in 2016 (see electronic supplementary material). We also conducted a year-long mosquito-monitoring programme near the main camp on Cooper Island. From 7 December 2015 to 7 August 2017, we hung a black-light trap in the forest, 100 m from camp. This trap was deployed overnight, once a week (during dry weather) 54 times over 20 months. We also used scent traps after rat eradication, because these are more effective than black-light traps for sampling Aedes [13]. Scent traps were hung in high human-use areas for eight continuous months We compared each mosquito species captured per trap-night before and after eradication using a general linear mixed model with island as a random effect (using the square-root transformation on mosquito count to help meet normality assumptions, though we present the untransformed means below). We also birds n = 5000-500 000 n.s. rsbl.royalsocietypublishing.org Biol. Lett. 14: 20170743 calculated the pooled proportional abundance of species trapped (i.e. Aedes/(Culex þ Aedes)), allowing us to estimate percentage of Aedes (+ 95% CI, binomial exact method) before and after rat eradication.

Results
In 2009, before rat eradication, Aedes (0.03 + 0.40 s.e. per trap set) were present, but less abundant per black-light trap-night than were Culex (2.46 + 0.40 s.e.) ( p , 0.0001, figure 1, right panel), though, because Aedes are less likely to be captured by black-light traps [13], this does not imply that Aedes were less abundant than Culex. Pooling counts per mosquito species across traps suggested that Aedes composed 5.9% (2.6% -11.3%, 95% CI) of the mosquito individuals in blacklight traps before rat eradication. A simple model suggests that Aedes could persist on a dense rat population (e.g. the approximately 40 000 rats present in 2009) or a dense human population (e.g. the 2400 military personnel in World War II), but not under current conditions with no rats and 5-30 humans (figure 2).
After rat eradication, researchers were bitten less often, and almost always at night (electronic supplementary material). The increased sampling effort captured 35-fold more mosquitoes than before rat eradication (electronic supplementary material). There was no significant difference in Culex caught per trap-night before (2.7 + 5.5 s.e.) than after (10.6 + 4.7 s.e.) rat eradication ( p ¼ 0.69). However, the zero Aedes caught per black-light trap-night after (0 + 0.047 s.e.) was significantly less than before (0.15 + 0.052 s.e.) rat eradication ( p ¼ 0.0004, figure 1, bottom-right panel). In the pooled samples, Aedes composed none (0.00%-0.22%, 95% CI) of the mosquitoes in black-light traps. Pooling the more sensitive scent trap data gave us more confidence that Aedes composed none (0.00%-0.12%, 95% CI) of the mosquitoes after rat eradication.
In summary, Culex persisted after rat eradication, while Aedes went from being present even in non-targeted trapping efforts before rat eradication to undetectable after rat eradication, despite much greater, and more targeted trapping effort (table 1).

Discussion
Had rat densities simply been reduced rather than eradicated, or had human densities been higher after rat eradication, biting rates on humans could have increased as mosquitoes switched from rats to humans (see figure in electronic supplementary material). Instead, bites from Aedes ceased. Our inability to document an Aedes bite, or trap an Aedes mosquito, over 2 years of sensitive surveillance meets the World Health Organization's standards for demonstrating mosquito eradication [8]. Most mosquito eradications are fleeting, because mosquitoes can soon recolonize. For instance, cycles of Aedes eradication and reintroduction followed intensive spraying on Kwajalein Island [14]. The Palmyra eradication seems different: lack of recovery over 6 years suggests that conditions on Palmyra have become unsuitable for Aedes.  Table 1. Relative mosquito abundance, by species, before (2009) and after (2015 -2016) rat eradication (2011) using black-light traps, scent traps and biting observations. n.a., not available.

species method sensitivity
We hypothesize that Aedes was eradicated from Palmyra primarily because its persistence depended on taking blood meals from rats (figure 1). Rat eradication could have also reduced larval habitat because rats open coconut husks, creating suitable habitat for container-breeders like Aedes and Culex [15]. Larval habitat might have also declined after the 2011 rat eradication, because 2011 and 2012 were drier than average years. However, rainfall records since 2002 indicate both wetter and dryer than average years before and after rat eradication, with no prolonged droughts (see electronic supplementary material), suggesting that such dry periods would not have eradicated Aedes on their own. Therefore, we expect this Aedes eradication will last as long as rats fail to re-invade Palmyra.
Although there are few documented coextinctions [16], examples include 10 parasitic trematode species that went locally extinct after their snail host was extirpated [17], a rat tick that went extinct along with the Christmas Island rat [18], and 11 bird lice species extirpated when their islandendemic bird hosts went extinct [19]. In fact, Aedes is not the only putative secondary extinction associated with the Palmyra rat eradication. Most rats on Palmyra were parasitized with a rodent-specific stomach nematode [20], which must have also gone extinct on the atoll after rat eradication.
These changes in the Palmyra food web show how removing introduced rats can have unintended indirect effects, including eradicating an introduced disease vector.