Who’s for dinner? Bird prey diversity and choice in the great evening bat, Ia io

Abstract The mysterious predator–prey interaction between bats and nocturnally migrating birds is a very rare and incredible process in natural ecosystems. So far only three avivorous bat species, including two noctule bats (Nyctalus lasiopterus and Nyctalus aviator) and the great evening bat (Ia io), are known to regularly prey on songbirds during nocturnal avian migration. The information related to the diversity and the characteristics of the birds as prey and the hunting strategy in both species of noctule bats are already clear. However, the diversity of bird prey in the diet of I. io as confirmed by molecular identification remains unknown. Moreover, like hunting insects, it remains unclear whether the avivorous bats opportunistically prey on birds. Here, we used DNA metabarcoding to investigate the bird prey composition, diversity, and choice in diets of I. io. We found I. io consumed 22 species of seven families from Passeriformes with a body mass of 6–19 g, and preferentially selected small‐sized passerine birds for optimizing the benefit/risk trade‐off. Moreover, most of the species preyed upon were migratory birds, while four species were local resident birds, indicating that I. io may adopt both aerial‐hawking and gleaning strategies on songbirds as do the other two noctules. Further, I. io body mass did not influence in prey choice and predation richness on birds, suggesting I. io is an opportunistic avivorous predator. This study provides novel insights into the avian dietary ecology of I. io and completes the analysis of predator/prey interaction between three avivorous bats and nocturnally migrating birds. Our results also indicate bat predation on birds which occurs as an act of ecological opportunity may subject bats to intense natural selection pressure, causing them access to the new diet‐defined adaptive zones.

bats on birds (namely, avivorous bats) is a rare process in nature and represents a case of dietary niche expansion from insects to birds, it has so far been found in 13 species belonging to 6 families (see Table S1). Most avivorous bats are mainly large bats, which occasionally capture resting birds using a gleaning foraging strategy (e.g., Vehrencamp et al., 1977). Only three temperate-subtropical species are known to regularly prey on migrating birds but also prey on insects using an aerial-hawking strategy. These include the greater noctule bat (Nyctalus lasiopterus) in Italy, Spain, and Russia (Dondini & Vergari, 2000;Ibáñez et al., 2001Ibáñez et al., , 2016Smirnov & Vekhnik, 2013), the birdlike noctule bat (Nyctalus aviator) in Japan (Fukui et al., 2013;Ibáñez et al., 2020), and the great evening bat (Ia io) in India and China (Han et al., 2007;Thabah et al., 2007).
So far, bats are the only mammals that have been reported to prey on nocturnally migrating passerines (Popa-Lisseanu et al., 2007).
The predation of bats on birds has probably influenced the evolution of bird migration strategies and antipredator behavior (Ibáñez et al., 2016). Thus, acquiring integral knowledge related to birds as prey to our understanding of predator-prey interaction between bats and migrating birds is critical. Traditional morphological analyses in droppings showed that N. lasiopterus and I. io captured only several species of birds in addition to insects (Dondini & Vergari, 2000;Han et al., 2007;Ibáñez et al., 2001;Thabah et al., 2007). For example, the diet of N. lasiopterus contained three small passerine birds: the European robin (Erithacus rubecula), the blue tit (Cyanistes caeruleus), and the wood warbler (Phylloscopus sibilatrix) (Dondini & Vergari, 2000;Ibáñez et al., 2001); meanwhile, one bird species, the Tickell's leaf warbler (Phylloscopus affinis) was presumably eaten by I. io (Han et al., 2007;Thabah et al., 2007). However, 31 bird species of eight families and 14 bird species of seven families from Passeriformes were identified in diets of N. lasiopterus and N. aviator by using molecular identification, respectively (Ibáñez et al., 2016(Ibáñez et al., , 2020. Thus, N. lasiopterus and N. aviator have been confirmed to prey mainly on nocturnal migrating birds on the wing and also only occasionally on resting birds in tree hollows or nest boxes (Dondini & Vergari, 2000;Fukui et al., 2013;Ibáñez et al., 2001Ibáñez et al., , 2016Ibáñez et al., , 2020Smirnov & Vekhnik, 2013). However, the diversity of bird prey in the diet of I. io confirmed by molecular identification remains unknown.
More notably, molecular dietary analysis has shown that both N. lasiopterus and N. aviator prey seasonally on songbirds with a body mass of 5-25 g during their nocturnal migration (Ibáñez et al., 2016(Ibáñez et al., , 2020, and N. lasiopterus prefers to catch medium-sized birds (10-15 g), with bird prey to bat body mass ratio averaging 25% (Ibáñez et al., 2016). The new prey to bat size threshold widely exceeded the traditional 5% threshold for bats hunting airborne prey (Fenton, 1990), which may explain why scientists were surprised and contested the emerging reports of predation strategy in bats (Andreas, 2010;Bontadina & Arlettaz, 2003). So far, only one study based on stable isotopes analyses has confirmed that N. lasiopterus preyed on a multitude of flying passerine birds during their nocturnal migratory journeys using an aerial-hawking strategy (Popa-Lisseanu et al., 2007). Thus, in order to amply confirm that bats employ an aerial-hawking strategy to prey on birds, it is necessary to determine diversity and migratory patterns of bird prey in the diet of the all avivorous bats especially in I. io. Moreover, it is also important to clarify whether phenotypic attributes, such as body mass, influence prey choice and predation richness (PR) on birds in bats. (Figure 1a) is one of the largest and rarest species in the family Vespertilionidae and is widely distributed in Southeast Asia, F I G U R E 1 (a) A great evening bat, Ia io (Chiroptera: Vespertilionidae) captured from Xingyi City, Guizhou Province, China. (b-g) Evidence of Ia io predation on birds: (b) tail membrane carrying bird feathers; (c) residual blood on tail membrane; (d) bird feather clamped in the forearm with a mark ring; (e) bloodstain on a hind foot; (f) fecal pellets containing numerous feathers; (g) undigested muscle and bone fragments. Photos taken by Lixin Gong

Ia io
northeastern India, and southern China. This species also is currently the only known bat to catch birds on the wing in southern China, where it mainly feeds on both insects and a passerine (P. affinis) based on traditional morphological identification (Han et al., 2007;Thabah et al., 2007). The present study employed DNA metabarcoding to determine the composition and diversity of bird prey species in I. io. Further, we investigated migratory patterns of bird prey, prey choice, and phenotypic constraints involved in bats preying on birds.
Specifically, we tested the following hypothesis: (1)  Bats were captured using mist nets spread at cave entrances when the bats returned from foraging (between 20:00 and 07:00).
Samples were collected at intervals of 1 to 2 days in three different entrances during a 12-day period in every month to minimize interfering with the bats. Our sampling did not include pregnant individuals; no individuals were collected during the winter. The feather-containing scats were collected from 43 individuals (40% of the total captured individual bats, n = 108) of I. io from March to November, 2017. Feather fragments were present in all fecal pellets of these 43 I. io and were estimated to form more than 90% of the fecal volume. The collections were divided into three seasons based on seasonal climatic periods in Guizhou (Zhang et al., 2014): spring (n = 9, March to early May), summer (n = 2, June to August), and autumn (n = 32, September to November). Each bat was placed individually in a clean and sterilized cotton bag for 30-60 min or until they defecated (less than two hours). Fecal pellets were collected from the bags and stored in 2-ml cryo tubes filled with 100% ethanol.
After bats emptied their feces, they were then sexed, weighed using an electronic balance (ProScale LC-50, Accurate Technology, Inc., Asheville, NC, USA) to the nearest 0.01 g, and the forearm length of each individual was measured with a digital caliper (TESA-CAL IP67, Tesa Technology, Renens, Switzerland) to the nearest 0.01 mm.
In this case, body mass of bats was not affected by physiological condition (i.e., hungry status, pregnancy, hibernation). Finally, bats were tagged with a numbered split aluminum alloy bat ring (5.2 mm, Porzana Ltd., Icklesham, UK) on the right forearm for individual identification before release into the cave. Any recaptured individuals within the same season were excluded from fecal collection.
Samples were short-term refrigerated (0-4°C) until they were transported to the laboratory by dry ice, after which they were stored at -20°C until laboratory procedures. Co., Ltd., China.
Valid sequences were obtained after quality control process, then dereplicated and excluded singletons sequences using Usearch (Edgar, 2010). The remaining sequences were clustered into MOTUs at 97% similarity thresholds using Usearch (Edgar, 2010), and chimeras also were removed simultaneously. We adopted a conservative approach for prey identification. The MOTUs with sequence numbers of <1% for avian samples of the total sequences in each sample were discarded in order to remove potentially erroneous and lowabundance sequences. We used the reference database in BOLD

| Bird prey composition and diversity analyses
The bird prey composition in diets of I. io was quantified using percent of occurrence (POO) of bird species and prey items, as well as

| Bird prey migratory patterns analyses
Data classifying the status of prey birds as resident or migrant as well as the prey size of birds (estimated from the average body mass of pooled male and female individuals of each bird species) were obtained from The Avifauna of Guizhou (Wu et al., 1986) and Fauna Sinica: Aves: Passeriformes (Vols. 10, 12, 13, 14) 1982;1998). To investigate the migratory patterns of bird prey species, we classified them as migratory birds (M, including Ssummer visitor, W-winter visitor, and P-passing bird) and resident birds (R).

| Prey choice and predation richness analyses
To probe bird prey size selection of I. io, we grouped the average body mass of each bird species into three categories (small size, <10 g; medium size, 10-15 g; and large size, >15 g) following Ibáñez et al. (2016). We then performed a simple linear regression to test the correlation between body mass of bats and birds. Here, we used body mass rather than forearm length instead as bat body size estimate, because bats will typically hunt their prey in flight based on the bat's body mass regardless of forearm length (Fenton, 1990).
Additionally, our analysis found that no linear positive correlation existed between the forearm length of bats and body mass of birds ( Figure S1). Data on migratory patterns and body mass were not obtained for two species identified at the genus level in the feces, Phylloscopus sp. and Horornis sp.; therefore, these were excluded from the above analyses.
We also calculated PR for 43 individual bats. The PR was defined as the number of different bird species that were preyed upon (i.e., the minimum number of birds) by each bat per night. The PR is considered that way for the sake of simplicity of the presented analysis,

| On-site evidence of predation on birds
During the course of field sampling, we found some evidence of I. io predation on birds, including a tail membrane carrying bird feathers and residual blood, a bird feather caught in a mark ring, bloodstains on a bat's hind foot, fecal pellets containing numerous feathers, and undigested muscle and bone fragments found in the feces of some individuals (Figure 1b-g). Moreover, of the total captured 108 individual bats, these included 31 individuals from spring, 36 from summer, and 41 from autumn. And we found the percentage of bats with bird predation evidence per season 29% in spring, 6% in summer, and 78% in autumn (Figure 2).

| Bird prey composition and species diversity in the diet of I. io
Using bird-specific primers, we identified a total of 85 bird prey

| Migratory patterns of bird prey
All of the identified bird prey species in the diet of I. io included songbirds found during migration (spring migration: March to early May; fall migration: September to November). Moreover, two species, Phylloscopus armandii and Calliope calliope, also were found during the breeding season (summer: June to August). The majority of prey species (80%, excluded two species without data; Figure 4a) and prey items (85%, excluded three prey items without data; Figure S3) were migratory birds (M, including S, W, P, W + P). Four species (20% prey species) were considered as local resident birds (R), which contributed only twelve prey items (15% prey items; Table 1, Figure 4a, and Figure S3).

| Bird prey choice of I. io
The average body mass of the 20 prey species (excluded two species without data) was 10.53 ± 3.58 g, ranging from 6.3 g for P. proregulus to 18.5 g for C. calliope. The average body mass of the 82 identified prey items (excluded three prey items without data) was 9.97 ± 3.68 g (Table 1). In this case, an estimate of 6.3-18.5 g was considered as the bird body mass window of I. io (n = 43, mean body mass = 54.29 ± 5.84 g). Based on the choice of preference in body mass of 20 prey species and 82 prey items, we found that I. io tends to prey on small-sized (<10 g) passerine birds (55% prey species and 68% of all prey items; Figure 4b). No significant linear positive correlation in body mass was found between bats and birds (Figure 4d), suggesting that bats prey on birds during opportunistic encounters when foraging under the precondition that they have ability to catch small birds.

| Predation richness did not correlate with bat and bird prey body mass
The average PR on birds for each bird-eating bat individual per night was 1.98 birds with a range from one to seven birds (Figure 4c and Figure S4). Model selection showed the null model was best supported (Table 2). Moreover, in the averaged model, BM of bats and MBMB also had no significant effects on PR (Table 3). These results indicate that PR of I. io was not related to the body mass of each bat and the body mass of birds which the bats preyed upon. This was consistent with the result of the relationship between body mass of bats and birds.

| Evidence of I. io preying on birds
Do feathers in fecal pellets indicate that a bat has fed on birds?
This question caused a scientific polemic and it has been proposed that additional evidence is needed to clarify the issue, such as knowing whether bone fragments would appear in a bat's feces (Andreas, 2010;Bontadina & Arlettaz, 2003;Ibáñez et al., 2003).
Moreover, Bontadina and Arlettaz (2003) developed an accidental consumption hypothesis pointing out that the presence of feathers in feces could result from the accidental ingestion of free fluttering feathers in the air, or may occur because bats capture birds by concentrating on perches on marshes or wetlands during migration. So far, many studies have reported the discovery of feathers and bone fragments in bat feces (Dondini & Vergari, 2000;Fukui et al., 2013;Han et al., 2007;Ibáñez et al., 2001;Smirnov & Vekhnik, 2013;Thabah et al., 2007). Here, we also found undigested muscle in feces and direct evidence of the ingestion of birds, such as feathers and residual blood on the tail membrane of some bats. These findings strongly confirmed some I. io individuals had recently eaten the birds. Importantly, we found a mark ring on the forearm of a bat that was clamped on to a bird feather, inferring there may have been a fight between the bat and a bird and suggesting that birds displayed antipredator behavior in response to bats. Our findings further confirmed that bats can undoubtedly hunt birds rather than the occurrence of accidental consumption (Bontadina & Arlettaz, 2003) and presented empirical evidence supporting the hypothesis that nocturnally migrating bird species succumb to aerial-hawking by several bat species (Dondini & Vergari, 2005;Ibáñez et al., 2016;Popa-Lisseanu et al., 2007).
However, further studies of the foraging ecology of these mysterious bats will be required to answer questions related to how they catch their bird prey and how this predatory behavior evolved (Dondini & Vergari, 2005). Note: Migration patterns: type of resident or migrant, divided into migratory birds (M, including S-summer visitor, W-winter visitor, and P-passing bird) and resident birds (R). Body mass: estimated from the average body mass of pooled male and female individuals of each bird species.

| Dietary bird composition and its potential benefits for fitness of I. io
Molecular analysis can complement previously used traditional morphological dietary data collection and provide deeper insights into the dietary ecology of wild animals. We found that I. io preys on a diverse variety of bird prey (at least 22 species) based on molecular dietary analysis, similar to findings for N. lasiopterus (Ibáñez et al., 2016) and N. aviator (Ibáñez et al., 2020), and I. io preferred to prey on species of the Phylloscopidae. However, here P. affinis was not found in the diet, indicating the morphological method of prey identification may not be accurate. This result was consistent with previous studies showing that C. caeruleus and P. sibilatrix also were not included in the diet of N. lasiopterus (Ibáñez et al., 2016). Additionally, these birds were not found in the diet of bats, possibly because of the low number of these birds near the studied colonies, resulting in a small chance of predation by bats. In particular, we clearly found that I. io prey on many species of Passeriformes to attain their optimal diet in spring and autumn. The size and/or nutrition in prey species of Passeriformes are actually higher than those of other prey species (i.e., insects) (Popa-Lisseanu et al., 2007). The choice of food in I. io may be designed to maximize the intake of energy and protein before and after hibernation, which can be proposed as a food quality hypothesis. However, whether the food quality hypothesis is associated with the protein maximization theory (Mattson, 1980) and/ or energy maximization theory (Schoener, 1971) of animal foraging strategy still needs to be further verified.

| Migratory patterns of bird prey and its implication for hunting strategy
In addition to mainly preying on birds mostly during migration, I. io also preys on a small number of local resident birds, as well as two migratory bird species (P. armandii and C. calliope) during the breeding period. This finding was also consistent with previous studies of N. lasiopterus and N. aviator showing that their diet included both migratory and sedentary birds (Ibáñez et al., 2016(Ibáñez et al., , 2020. Many species are migratory, but do not necessarily migrate directly through the study area; these birds may also be preyed upon when they could be reproducing or wintering in the area. These sedentary birds may be preyed upon as a result of their nocturnal (dispersal) movements (Mukhin et al., 2009;Ward et al., 2014;Zheng, 1995). Some diurnal songbirds or resident birds will perform nocturnal activity patterns or nondirectional and short-distance migration according to habitat, climate, and seasonal changes. For example, nocturnally migrating Eurasian reed warbler (Acrocephalus scirpaceus) use nocturnal movements at other times of year, including during the breeding season (Mukhin et al., 2009). Thus, I. io may behave like two other noctules that may hunt birds either mainly by using an aerial hunting strategy during the nocturnal migration of birds or occasionally by using a gleaning strategy while the birds are resting at night (Dondini & Vergari, 2000;Ibáñez et al., 2016Ibáñez et al., , 2020. However, future behavioral studies should be carried out to clarify this issue.

| Prey choice and an opportunistic avivorous predator
We found I. io preferentially selected smaller species of the family Phylloscopidae and small-sized (< 10 g) passerine birds. This was inconsistent with N. lasiopterus, which selected medium-sized bird species (10-15 g) (Ibáñez et al., 2016), thereby presumably to more optimize the benefit trade-off between energy intake and predation risk. This may not be explained by differences in body size of bats because the average body mass of I. io (54.3 g) was slightly bigger than that of N. lasiopterus (52.0 g) (Ibáñez et al., 2020). Moreover, here the nonsignificant relationship between body mass of I. io and body mass of birds may also support the view. Thus, the disparity in the bird prey size choice between I. io and N. lasiopterus probably occurs because of the difference in species types and size of the nocturnal passerine migrants in different regions. Ia io has an average bird prey to bat body mass ratio of 18.4% (ranged from 11.6%-34.1%), conforming to a new threshold (<25%) (Ibáñez et al., 2016), but widely exceeded the 5% threshold in respect of bats hunting airborne prey (Fenton, 1990). Ibáñez et al. (2003) suggested that the 5% threshold rule remains valid for bats hunting for flying insects, but not for these bats that hunt for birds at high altitudes. The avivorous bats have a relatively large body  Table S2) size, low frequency echolocation calls, and high wing loading, as well as being adapted for rapid flight and for detecting and catching relatively large prey in open spaces (Fukui et al., 2004(Fukui et al., , 2011Ibáñez et al., 2001;Thabah et al., 2007). Moreover, the large body size, strong skull, and high bite force (Shi et al., 2020) combined with a long-distance echolocation system of I. io, could allow it to exploit a recently found feeding niche (nocturnally migrating songbirds). Birds may be easier to detect than insects for these bats at a greater spatial range. This supports the hypothesis stating that detecting small passerine birds would be similar to locating large moths from an acoustic perspective (Ibáñez et al., 2003). However, obviously only further behavioral experiments will help in answering how bats prey on birds by an aerial-hawking strategy.
Optimal foraging theory predicts that prey size of a predator depends on the predator's own size (Stephens & Krebs, 1986). That is, in any species of a given species pool (terrestrial, aquatic, or marine), prey body size and feeding range increase with an increase in the body size of the predators (Gravel et al., 2013). However, our study found that no significant positive correlation exists in body mass between I. io and bird prey, suggesting that I. io is an opportunistic  avivorous predator under the precondition that they have ability to catch small birds. This may explain why some individual bats could catch more than two and up to seven bird species in one night, while the PR was also not affected by a bat's own body mass and captured bird body mass. However, for gleaning foragers, the bats could capture even more prey and actually eat them on the substrate or ground because they experience less risk of predation and injury. For example, the frog-eating bat (Trachops cirrhosis) consumed as many as six small frogs per hour (Ryan et al., 1981). Thus, to be a generalist may be beneficial for avivorous bats because of the nature of high energy cost for hunting migrating birds. The future studies should determine that whether a predation threshold of body size exists for bats hunting for birds in flight.

ACK N OWLED G M ENTS
We are grateful to Liying Zhai and Ersheng He for their help during sample collection. We thank LetPub (www.letpub.com) for its linguistic assistance during the preparation of this manuscript. This

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

AUTH O R CO NTR I B UTI O N S
Lixin Gong: Conceptualization (equal); Data curation (equal); Formal analysis (lead); Investigation (equal); Methodology

TA B L E 3
Model-averaged parameter estimates of the best-supported (before and including the null model) generalized linear models describing variation in predation richness with independent variables in Ia io