Taxonomic Shifts in Philornis Larval Behaviour and Rapid Changes in Philornis downsi Dodge & Aitken (Diptera: Muscidae): An Invasive Avian Parasite on the Galápagos Islands

The parasitic larvae of Philornis downsi Dodge & Aitken (Diptera: Muscidae) were first discovered in Darwin’s finch nests on the Galápagos Islands in 1997. Larvae of P. downsi consume the blood and tissue of developing birds, causing high in-nest mortality in their Galápagos hosts. The fly has been spreading across the archipelago and is considered the biggest threat to the survival of Galápagos land birds. Here, we review (1) Philornis systematics and taxonomy, (2) discuss shifts in feeding habits across Philornis species comparing basal to more recently evolved groups, (3) report on differences in the ontogeny of wild and captive P. downsi larvae, (4) describe what is known about adult P. downsi behaviour, and (5) discuss changes in P. downsi behaviour since its discovery on the Galápagos Islands. From 1997 to 2010, P. downsi larvae have been rarely detected in Darwin’s finch nests with eggs. Since 2012, P. downsi larvae have regularly been found in the nests of incubating Darwin’s finches. Exploring P. downsi ontogeny and behaviour in the larger context of taxonomic relationships provides clues about the breadth of behavioural flexibility that may facilitate successful colonisation.


Introduction
Three genera of flies within the order Diptera have larvae that parasitise avian hosts: Protocalliphora Hough (Calliphoridae), as well as Passeromyia Rodhain & Villeneuve (Muscidae) and Philornis Meinert (Muscidae). The adult flies in these genera are free-living and do not parasitise birds, but their larvae develop in the nests of altricial birds, feed on their avian hosts, and exhibit feeding behaviours Table 1. Philornis species ordered according to taxonomy, from the most basal 'aitkeni-group' to the most recently evolved 'angustifrons-group' (groups from [33]). Larval feeding habits are shown where known and abbreviated as follows: free-living coprophagous larvae (FLC), free-living semi-hematophagous larvae (FLSH), subcutaneous hematophagous larvae (SubH). The following nine species are not included in the list as they are currently not assigned to a taxonomic group [33] given insufficient information: P. molesta, P. nielseni, P. blanchardi, P. cinnamomina, P. convexus, P. mima, P. obscurus, P. steini, P. umanani. based on larvae with distinctive posterior spiracles found parasitising nestlings. At this time, Philornis was suggested to be a synonym for Protocalliphora and assigned to the family Calliphoridae [26]. In 1921, Malloch [27] proposed the genus Neomusca based on adult specimens, whereas the genus Philornis was based on larval characters. Aldrich [28] revised this group and synonymized Neomusca with Philornis as independent genera, assigning both within the family Muscidae (Anthomyiidae at the time). This revision was supported by further work on Philornis species, as new and previously described species were transferred from other genera including Hylemyia, Mesembrina, Neomusca and Mydaea [9, [28][29][30][31]. Philornis adults are distinguished from other muscid genera by the presence of hair on the anepimeron and the postalar wall [1,32].
Using morphological and ecological data, Philornis can be divided into three phylogenetic groups: the 'aitkeni-group', the 'falsificus-group' and the 'angustifrons-group' [33]. Male characters (given few female specimens) generally define the most basal lineage of Philornis, the 'aitkeni-group', including enlarged upper eye facets in holotypic males [29,33]. The members of this group display adult character states that are considered primitive among muscids (i.e., enlarged upper eye facets and presence of cilia on the surface of the wing vein R 4+5 ) [33]. This group includes P. aitkeni (Dodge), P. rufoscutellaris (Couri), and P. fasciventris (Wulp). The phylogeny of the aitkeni-group is not completely resolved because of missing information about life history and morphology, as female and larval specimens are not available for many species. The second group, the falsificus-group, is defined primarily by P. falsificus (Dodge and Aitken), whose larvae are free-living [9]. Common morphological characters include five scutellar marginal setae that also place P. fumicosta (Dodge), P. univittatus (Dodge), P. grandis (Couri) and P. sabroskyi (Albuquerque) within this group [33]; however, data on the ecology of these species are missing. More information on larval life history is necessary to confirm whether species other than P. falsificus belong in this lineage. Despite a similar life history to P. falsificus, P. downsi is not within the falsificus-group [1, 9, 33], but forms a sister-group to all species within the angustifrons-group for which larval habits have mostly been documented (Table 1). The angustifrons-group is the most recently evolved and largest of the three Philornis lineages and contains species with subcutaneous hematophagous larvae as well as P. downsi with semi-hematophagous larvae.
Comparative taxonomic analyses of Philornis species have been hampered by a lack of specimens and information [9]. For several species of Philornis, their morphological descriptions are based solely on one sex, generally males. In others, the holotype is missing or destroyed, and so other traits and ecological information are missing. Philornis blanchardi (Garcia) has been originally identified and described in Argentina from a single female specimen, which has since been lost [34]. This specimen may belong to a previously described species as it has not been captured and identified since, however the original description is considered sufficiently unique that it may be a separate species [34]. The single male holotype used to describe P. umanani (Garcia) has also been lost and due to the lack of detail provided in the original description, this species is deemed unrecognisable and is now considered a nomen dubium [34]. Evidence of a Philornis species complex within specimens of P. seguyi (Garcia) and P. torquans (Nielsen) in Argentina throws further doubt on the original taxonomic characterisation of many Philornis species [35]. These issues highlight the need for more extensive molecular and morphological analysis of currently recognised Philornis species to confirm species classifications and their evolutionary relationships.

Philornis larval behaviour
Philornis species differ in their larval feeding habits, which include coprophagous and hematophagous diets (Table 1). Larval habits have been documented for 30 out of 52 described species (Table 1). The most basal group in the Philornis phylogeny (aikteni) have free-living coprophagous larvae [33]. These larvae parasitise cavity nesting host species that do not remove waste, such as the rufous-tailed jacamar (Galbula ruficauda) Cuvier (Piciformes: Galbulidae) and appear to be specific to this type of nest [2,5,30,36,37]. Free-living saprophagous larvae in the nest are regarded as the ancestral trait, evolving into coprophagous larvae, semi-hematophagous larvae and then subcutaneous larvae [9,33]. This transition is also supported in Passeromyia where species show a similar order of descent [4,10,33]. Two documented species, P. downsi (angustifrons-group) and P. falsificus (falsificus-group), have free-living and semi-hematophagous larvae, although other undescribed species within the falsificus-group may also have free-living larvae [1, 30,33].
Most Philornis species (83%) have larvae with subcutaneous hematophagous feeding habits, which is also the primary larval strategy in the angustrifrons-group. Within this group, only P. downsi has non-subcutaneous larvae. The semi-hematophagous P. downsi larvae may be similar to P. falsificus (falsificus-group), which is also suspected of having free-living semi-hematophagous larvae [33]-but not enough is known about the biology of the falsificus-group. While P. falsificus is considered a free-living ectoparasite [30], this assessment is limited by the observations to date of later instars and puparia [38,39]. On the other hand, in two species with subcutaneous feeding habits in the angustifrons-group, a few Philornis larvae have been also observed in avian nares. Specifically, P. mimicola larvae have been found in the nares of ferruginous pygmy-owl nestlings (Glaucidium brasilianum) Gmelin (Strigiformes: Strigidae), but most larvae occurred subcutaneously [40]. Larvae of P. porteri (Dodge) have been found in the nares and ear canals of some nestlings [41,42], and 3rd instar larvae observed to feed externally on the abdomen and wings of their hosts [41,43]. In the semi-hematophagous P. downsi larvae, 1st instars regularly reside within the avian nares [44][45][46] and later instars move to the base of the nest where they emerge at dusk and dawn to feed externally on the blood and tissue of the developing birds [45,46]. Lineages with free-living larvae have been far less studied than lineages with subcutaneous larvae (Table 1). Free-living larvae move freely within the host nest, detach from the host at various times and reside in the nest base during the day, making them less conspicuous to human observers [45,46]. In contrast, subcutaneous larvae reside under the skin of the host and hence can be detected when nestlings are examined.

Philornis downsi larval instars
Philornis downsi larval development is split into three instar development stages. 1st instar larvae generally reside in the naris and ear canals of developing nestlings, but some have also been found moving freely within the nesting material [21,52,53]. First instars are commonly collected from 2 to 3 day old nestlings [43]. Late 2nd and 3rd instar larvae are generally free-living, residing within the base of the nest and feeding externally on nestlings at night [14, 45,46]. These later instar larvae feed on the blood and fluids of their host by penetrating the skin of the nestlings [2, 30]. Larval instar morphological descriptions are given by Fessl et al. [44]. The most distinct character between the instars is the posterior spiracles, which change in colour, shape and number of spiracular slits present throughout larval development [44].
Figure 1(1A) shows the posterior spiracles of a 1st instar P. downsi larva, characterised by their light pigmentation and two oval slits present [44]. The spiracles of a  1st instar larva are separated by slightly more than their diameter. First instar larvae lack anterior spiracles (Figure 1(1B)). The posterior spiracles of a 2nd instar larva are similarly round with two oval slits; however, the distance between them is two to three times of their diameter (Figure 1(2D); [44]). Anterior spiracles are present during the 2nd instar, and semicircular in shape, lightly pigmented and visible in Figure 1(2E)). 3rd instar posterior spiracular plates are darkly pigmented and round in shape, distinct C-shaped spiracular slits radiate from median ecdysial scar (Figure 1(3G)). Pigmentation of the median ecdysial scar is light in early 3rd instar larvae and becoming darkly pigmented later in the stage. Semi-circular anterior spiracles are retained in 3rd instar larvae (Figure 1(3H)). Cephaloskeleton morphology differs between instars as outlined in Fessl et al. [44]. Recent studies report a decrease in P. downsi puparia size across 2004-2014 [54]. Common et al. (unpublished data), and hence body size is certainly not a useful method to classify instars. In general, it is recommended to use a suite of morphological characters, including anterior and posterior spiracular morphology, to determine the larval instar.

Larval development
The developmental period of Philornis larvae is associated with the species' larval feeding habit. For example, time to pupation in coprophagous species takes up to 29 days, but only 4-8 days in subcutaneous species [2,55]. Larval development periods in free-living species such as P. downsi are difficult to determine in the wild as the host nest needs to be dismantled to observe the larvae. Early studies of abandoned or failed nests found 1st instar larvae in nests with 1-3 day old nestlings, 2nd instars in nests with 3-6 day old nestlings and 3rd instars in nests with 3-10 day old nestlings [44]. Larval collections following the cessation of activity at host nests suggest that the minimum time for pupation in P. downsi is 4-7 days [54].
Compared with larval development times in the wild, larval development times under laboratory conditions are longer. First attempts to rear P. downsi larvae in the absence of a living host had a low success rate, with only three larvae out of 477 reaching the adult stage after a 36 day development time (mean 18 day larval development, 12 day pupation) [56]. As the diet for rearing larvae in captivity was refined, the success rate increased to 10% and larval development time decreased [57]. Development time in the laboratory ranged from 9 to 10 days from larva to pupa [57] with even faster development times occurring as the rearing conditions have improved [pers. comm. P. Lahuatte]. Egg hatch rates in captivity have been high (96%), with most mortality in 1st instar larvae (77%) [57]. Laboratory-based diets that have been developed in the absence of a bird host are primarily based on chicken blood, with more successful diets including hydrolysed protein and vitamin fortification [57]. The lack of keratin in the diet may be causing elevated 1st instar mortality, as 1st instars consume the keratin of the beak in which they reside [44], however the true cause is unknown.

Philornis downsi adult behaviour
The behaviour of adult P. downsi is much less understood than that of the larvae. The adult fly is vegetarian, feeding on decaying fruits and flowers, including the invasive blackberry (Rubus niveus) Thunb (Rosales: Rosaceae) [9, 15, 31]. Philornis downsi is commonly attracted to a mix of blended papaya and sugar, which is used to trap adult flies (developed by P. Lincango and C. Causton; used by [58], Causton et al. in review). This mix is particularly attractive to adult flies due to the presence of yeast and fermentation products such as ethanol and acetic acid [59].
A one-year study on Floreana Island found that male and female P. downsi display dimorphic flight patterns, with females more likely to be caught in high and low traps (2 m, most common at 6-7 m), and males more likely to be caught in traps of intermediate height (4-5 m) [58]. As the pattern of male and female abundance are quadratic opposites, this has tentatively been suggested to be an advantage for females to avoid male flies, as frequent mating in other Dipterans has been found to decrease female reproductive success and lifespan [60,61]. This flight pattern may also explain why certain host species experience higher parasite intensities, such as the medium tree finch (Camarhynchus pauper) Ridgway (Passeriformes: Thraupidae) that has an average nest height of 6 m, thus making it more susceptible to being encountered by female P. downsi [58,62,63]. However, the factors that cause bird species to experience differing intensities of P. downsi are complicated and vary between years. Comparison of flight height in P. downsi on different islands is needed to test the generality of this pattern, which may be influenced by average tree height and/or other ecological variables.

Mating behaviour
The mating behaviour of Philornis in general is not well understood, though there are some insights into P. downsi mating patterns. While mating has not been observed at or inside the nest, multiple P. downsi flies have been video recorded to enter host nests concurrently [45,64]. Analysis of offspring genetic relatedness has provided estimates of the re-mating frequency of P. downsi [65]. Evidence for multiple mating by females has been frequently detected, and each larval infrapopulation (i.e., within nests) is sired by 1-5 males (average ~1.9 males per female) [65]. How P. downsi adults find each other to initiate mating is unknown. Pheromones for attraction and aggregation in muscid flies have been identified and studied [66][67][68]. Cuticular compounds show promise for determining if P. downsi produces pheromones, as females and mature males showed distinct cuticular profiles and females respond to chemicals produced by males [69][70][71]. Cuticular profiles could be developed as an attractant to capture flies in the field [20,72].

Oviposition behaviour
Studies into oviposition in the genus Philornis have revealed that species spanning diverse larval feeding habits are oviparous [1,9,31,73,74]. This current view has previously been hotly debated, in part because the majority of species remain unstudied. Laboratory rearing and field observation have confirmed that P. downsi is oviparous [45,56,57,75]. Philornis flies enter and oviposit in nests regardless of nesting phase or nestling age but have not been observed to enter nests abandoned by the parent birds during the incubation phase [45,47]. From in-nest video recordings, P. downsi flies have been observed entering nests throughout the day, but generally during dusk between 1500 and 1800, with visiting rates peaking around 1700 [45,64]. Visit length averaged 1.3-1.5 min and occurred most commonly when the adult host is away from the nest and completed once the adult host returned [45,64]. Eggs have been generally deposited on nesting material and the base of the nest [45,57], however on one occasion, eggs have been also laid directly by the naris of a nestling [45]. A genetic study of P. downsi larvae estimated that 1-6 adult females (average ~3 females) oviposit within a single nest, supporting previous observations of different sized larval groups within nests and suggesting repeated nest infestations throughout the nestling period [7, 65].

Effects of host species on Philornis behaviour and microbiome
Philornis downsi is one of the most generalist species within the genus, known to infest 38 host species across avian taxa [5, 6, 76]. However, this high host number may reflect the large number of studies focused on P. downsi due to its invasive status on the Galápagos Islands [15,16].
It is currently unclear how Philornis species in general or P. downsi in particular find their hosts. Preliminary studies into the role of semiochemicals and volatiles in host nests as an attractant for P. downsi have produced inconclusive results [70]. Long-term ornithological field studies have provided some hints that the intensity of host cues may be relevant for P. downsi search behaviour, or alternatively that the density of host nests influences P. downsi oviposition behaviour. Aggregated host nests may attract P. downsi females due to an increase in olfactory or visual cues. These aggregated nests also provide a greater opportunity for P. downsi females to infest multiple nests. Indeed, small tree finch nests (Camarhynchus parvulus) Gould (Passeriformes: Thraupidae) with close neighbours contained more P. downsi larvae compared to solitary, more isolated nests [16]. Nests in areas of lower nesting density (i.e., lowlands) have been more likely to contain the offspring of a single P. downsi female than nests in areas of higher nesting density (i.e., highlands) that are more likely to contain the offspring of many P. downsi females A metagenomic study into P. downsi larval microbiome sampled from different host species found an effect of host diet on the gut bacterial community of P. downsi larvae [77]. Larvae retrieved from strictly insectivorous warbler finch (Certhidea olivacea) Gould (Passeriformes: Thraupidae) nests have a different microbiome structure compared with larvae parasitising hosts with broader dietary preferences (ground and tree finches, Geospiza and Camarhynchus sp., respectively) [77]. The gut microbiome also differed between P. downsi larvae (blood diet) and adults (plant diet), supporting the hypothesis that P. downsi microbiome changes during development and according to diet [77]. Further behavioural, biochemical and genetic studies are needed to understand P. downsi oviposition across host species, host locating behaviour and host specificity.

Age of larval cohort in host nests
There is evidence that the oviposition behaviour of female P. downsi has changed since its discovery on the Galápagos archipelago. Philornis downsi flies are now known to oviposit during any stage of the nesting cycle [45]. In the first decades following initial discovery of P. downsi in Darwin's finch nests, changes in the proportions of instar classes among P. downsi have been observed, with evidence that oviposition occurred earlier and more synchronously in the nesting phase in the later years of the study [54]. Synchronisation in oviposition date may lead to an increase in larval competition for host resources, and as a consequence result in increased virulence for nestlings that must contend with a greater number of large, mature larvae at a younger age [16]. The fitness consequences of female oviposition behaviour are further supported by observations in other Philornis systems. Host nests that are infested later in the nesting cycle are more likely to have higher fledging success than nests parasitized early in the nesting cycle [50,78].

Larval feeding on adult birds
Philornis larvae are generally exclusive parasites of developing nestlings, whether they be subcutaneous or free-living semi-hematophagous species. Infestation of host nests can happen quickly and is often observed within 24 h of the first nestling hatching [41,43,50,79]. Many studies on Philornis species in their native range found no evidence of larvae present during incubation [47,48,80,81]. There have been a few cases of larvae feeding on adults in subcutaneous species [82][83][84], however these reports are rare, with generally only a few larvae per adult. For this reason, larval feeding on adults is generally regarded as opportunistic [2]. More data are needed to examine the oviposition behaviour of Philornis species to determine whether larvae are present during the incubation phase.
On the Galápagos Islands between 1998 and 2005, there have been no reported cases of P. downsi larvae present in host nests with eggs that would suggest that larvae also feed on incubating females. Two studies during this time period specifically stated that no P. downsi larvae have been found during host incubation (Table 2) [21,85]. On Santa Cruz Island during 1998-2010, published studies report findings for 38 nests with eggs that have been inspected for the presence of P. downsi and found no larvae (Table 2) [21,85]. In 2012, Cimadom and colleagues first observed P. downsi larvae in host nests during incubation where larvae have been found present in 17 of the 26 nests inspected [85]. Since this initial observation, the prevalence of P. downsi in host nests with eggs has increased to 80% in some species and years on Santa Cruz Island, with larvae and puparia found in 70 of 177 nests inspected with eggs [86]. Concurrently across this time period, brooding Darwin's finch females have P. downsi antibodies that are associated with decreased P. downsi intensity, but not increased fledging success [87,88]. This suggests that P. downsi larvae on the Galápagos Islands may have switched to feed on adult finches at some stage [87]. On Floreana Island, inspection of nests that failed during incubation during 2006 and 2016 found P. downsi larvae in 4 of 72 (5.6%) nests with host eggs ( Table 2). In 2006, three medium ground finch (G. fortis) nests with eggs in the arid lowlands have P. downsi larvae and puparia, and in 2010 one highland small tree finch (C. parvulus) has P. downsi larvae during the egg stage. During a period of intense drought from 2003 until 2006 with less than 300 mm of rain per year in the lowlands, there were very few active host nests available for oviposition, which may be an explanation for a shift in P. downsi female oviposition and larval feeding on incubating females at the end of the drought during 2006. Notably, smaller larvae and eggs are not easily visible in nests and it is possible that P. downsi is present, but not detected during incubation in the early years of study.
In laboratory trials, P. downsi hatching success is found to be the same in nests with host eggs and nests with finch hatchlings (Lonchura striata domestica) Linnaeus (Passeriformes: Estrildidae) [89]. In these trials, there is even a fitness benefit for P. downsi that hatched during incubation and hence earlier during the host cycle, as they survived for longer [89]. Other than P. downsi, there is one report of an unidentified Philornis species parasitising adults in the pearly-eyed thrasher (Margarops fuscatus) Vieillot (Passeriformes: Mimidae) studied in Puerto Rico [49]. About 46% of incubating and brooding females and 13% of attending adult males sustained subcutaneous Philornis [49]. It has been suggested that this Philornis species may have invaded Puerto Rico, as the patterns of prevalence and host   48,49]. Philornis consumption of attending adult hosts may be an oviposition tactic that is more prevalent under conditions of resource limitation. Resource limitation could be influenced by resource termination such as early host death, resource availability when there is a limited supply of host nests (e.g., during drought), and resource accessibility, for example when competition within and between fly cohorts changes [54].

Conclusions
As one of three avian nest parasitic genera in Diptera, the genus Philornis provides a useful system to explore shifts in larval feeding behaviour in native and invasive species. Philornis downsi has been accidentally introduced to the Galápagos Islands and first observed in the nests of Galápagos land birds in 1997. In this chapter, we explored similarities and differences between P. downsi larval development and behaviour with what is known from the other 52 Philornis species. More basal Philornis (aitkeni-group) species have free-living coprophagous larvae and more recently evolved Philornis (angustifrons-group) tend to have subcutaneous hematophagous larvae with the exception of P. downsi that has free-living semi-hematophagous larvae. Since its introduction to the Galápagos Islands, there have been documented changes in the behaviour of P. downsi. During the early years after initial discovery of P. downsi on the Galápagos Islands, oviposition behaviour was asynchronous across the nesting cycle and larvae appeared to have fed exclusively on developing nestlings until 2005. In later years, P. downsi oviposition behaviour was earlier in the nesting cycle and more synchronous, and since 2006, larvae have also been recorded to feed on incubating females. The first records of P. downsi larvae in host nests with eggs rather than hatchlings occurred at the end of a four-year drought on the Galápagos in 2006. Since 2012, up to 80% of host nests with eggs may contain P. downsi larvae on Santa Cruz Island. Larval feeding by P. downsi on adult birds has been observed in laboratory finches and in one Philornis system (species unknown) in Puerto Rico. In light of changes in P. downsi larval feeding behaviour, we provided a description and photos of the larval instars for use in field identification. We compiled the observations to date of Philornis behaviour and ontogeny within a broad taxonomic framework and summarised patterns of change in the oviposition behaviour of P. downsi in its (presumably) novel habitat on the Galápagos Islands. By examining P. downsi in relation to other Philornis species, we provided a broad phylogenetic context for the potential behavioural repertoire of an invasive species under conditions of intense natural selection in a novel environment.