Differences in the niches of keratin/chitin feeding moths (Lepidoptera: Tineidae) in bird nests in central Japan

Some moth larvae feed not on plants but on keratin and/or chitin produced by animals. These substances are polymers and are commonly found in bird nests as feathers and raptor pellets. Many qualitative studies have examined the association of keratin/chitin feeding moths with bird nests. However, few studies have quantifi ed the species composition with respect to type of nest and habitat. Hence, we have studied the degree to which the niches of these moths differ in terms of type of nest and habitat. We set-up open-top nest boxes for the Ural owl Strix uralensis (damp exposed nests from which owl chicks were fl edged successfully) and artifi cial bird nests (mesh bags fi lled with duck down to imitate dry exposed nests) in a deciduous forest and artifi cial bird nests in an urban setting in Aichi Prefecture, central Japan, and collected the contents of the nests every two months from June to December 2014. We recorded 592 individuals of fi ve keratin/chitin feeding moth species (Tineidae, Tineinae) from the contents. Using non-metric multidimensional scaling and cluster analysis of the relative abundances of individual species in each type of nest in forest and urban settings, these species were classifi ed into three groups: (1) Monopis longella and M. congestella inhabiting forest, (2) M. fl avidorsalis and Niditinea tugurialis mainly in damp exposed nests in forest and (3) N. piercella mainly in dry exposed nests in urban areas. This classifi cation was compared with fi ndings of other studies. As a result, these moths probably have different niches with respect to nest type (damp or dry), keratin/chitin as a source of food (raptor pellets or feathers), and habitat (forest or urban area). Furthermore, we suggest that the evolution of larviparity in M. congestella might be related to its preference for feathers as a source of food for the development of its larvae.


ORIGINAL ARTICLE
into a plastic mesh bag (Robinson & Nielsen, 1993). The owl nest boxes and artifi cial bird nests were categorized as damp exposed type and dry exposed type, respectively, following Robinson's (2004) classifi cation. We compared the composition of keratin/chitin feeding species of moth in the two types of nests in different habitats.

Ecology of the Ural owl
Paired Ural owls usually nest in cavities in tree trunks. Females lay two to four eggs in March to April and incubate them for 27 to 32 days (Nakamura & Nakamura, 1995). Prey for chicks are small mammals, birds, frogs, newts, lizards and centipedes (our unpublished data). Chicks leave the nest 30 to 34 days after hatching.

Collection of the organic contents of owl nest boxes
Nine open-top plastic boxes (26 cm wide, 26 cm long, 30 cm deep) were installed in trees 2 to 3 m above ground level in a secondary deciduous forest (80 to 360 m a.s.l.) extending across Toyota and Okazaki Cities, Aichi Prefecture, central Japan in May 2013 (Fig. 1). Several twigs were put in each box as nest material. In spring 2014, Ural owls bred in each box, with one or two chicks fl edged by late May. All or half the organic contents (feathers, owl pellets, faeces and other detritus) were collected from each of three or four randomly chosen boxes on 7 June, 17 August, 18 October and 6 December 2014; as a result, the organic contents in fi ve boxes were collected twice (Table 1). Feathers were present in only two boxes in June and August. The quantity of nest material collected at any one time ranged from 0.6 to 3.1 litres.

Sampling using artifi cial birds' nests
Artifi cial birds' nests were made according Robinson & Nielsen (1993) using plastic mesh bags (16 × 20 cm in size; 3 × 3 mm in mesh size) stuffed with 10 g of dry duck down bound together with 5 g of beer yeast and some water. Four artifi cial nests were hung on tree branches (several hundred meters apart; 90 to 120 m a.s.l.) at 1.5 m above the ground in a secondary deciduous forest in the Toyota City Nature Sanctuary, Aichi Prefecture on 7 June, 17 August and 18 October 2014 (hereafter referred to as forest artifi cial nests; Fig. 1). To compare the species composition of the moths in the forest and that in the urban area, three artifi cial bird nests were set-up at the Toyota Yahagi River Institute in the urban area of Toyota City on 2 June, 18 August and 16 October 2014 (hereafter referred to as urban artifi cial nests; Fig. 1); this site is 400 m from the nearest green park. Two artifi cial nests were hung on the branches of trees at this site, while the third was hung under the eaves of a building. The nests were retrieved approximately two months later on 17 August, 18 October or 6 December 2014.

Rearing
Organic contents collected from each owl nest box on each occasion they were sampled were put in one to three plastic boxes (16 cm wide, 22 cm long, 8 cm deep) according to the amount. To ventilate the boxes, the centre of the lid of each box was cut away to form 4-cm square window, which was covered by a sheet of fi ne plastic mesh. Duck down from each artifi cial nest was also kept in a plastic box of the same type. These boxes were stored under semi-natural temperature and light-dark photoperiod conditions (adjusted to monthly average values in fi eld) in an incubator for two to seven months according to the collection date; materials collected in June, August, October and December were kept concealed, dry concealed, damp exposed, or dry exposed; concealed nests are those in holes in trees (including nest boxes), while exposed nests are constructed in trees and hedges. For example, the Monopis congestella-group and M. monachella-group are found in damp exposed and dry exposed nests, respectively, while the Niditinea fuscella-group are found in nests of both types. The Niditinea piercella-group occur only in concealed nests. However, Robinson (2004) does not provide quantitative data supporting these conclusions.
Since Robinson (2004), much information has been collected about keratin/chitin feeding moths in bird nests, especially in Japan (e.g., Nasu et al., 2007aNasu et al., , b, 2008aNasu et al., , b, 2012a. In these studies moths were reared on the organic contents of concealed and exposed nests of 16 species of birds in forests and urban areas and  Nasu (2012) summarizes the associations of these moths with bird nests in terms of type of nest and habitat as follows: M. longella and M. congestella occur in concealed and exposed nests in forests, while M. fl avidorsalis and N. striolella are found only in concealed nests in forests; in contrast, Tinea translucens and Tineola bisselliella only occur in concealed and exposed nests in urban areas. However, the studies by Nasu and colleagues did not quantitatively or systematically assess the degree to which these moths depend on concealed and exposed nests. Furthermore, the specimens identifi ed as N. tugurialis probably included N. piercella Bentinck, so information on the occurrence of this species in bird nests is unreliable .
Larviparity, or laying fully developed fi rst-instar larvae instead of eggs, is very rare in Lepidoptera (Hinton, 1981;Heppner, 2009). However, it is relatively common in Monopis, a keratin/chitin feeding genus with 90 species worldwide (Robinson, 2019), 12 of which are known to be larviparous (Xiao & Li, 2006). Three species of Monopis are recorded in Japan (see above; Sakai, 2013), and one, M. congestella, is larviparous (Robinson, 1988(Robinson, , 2004Nasu et al., 2008a). Robinson (2004) has hypothesized that larviparity in Monopis evolved due to (1) high mortality of eggs and neonates due to intense predation by mites in bird nests, and (2) high cost of fi nding more than one oviposition site, or bird's nest. If there is a difference in nests exploited by larviparous and oviparous Monopis moths, then there could be another reason.
This study determines the degree to which keratin/chitin feeding moths have different niches in terms of the types of nests or habitats in which they are constructed. This involved sampling moths quantitatively and systematically in a deciduous forest and an urban area using nest boxes and artifi cial bird nests. The open-top nest boxes were designed for the Ural owl Strix uralensis Pallas, while the artifi cial bird nests were made from duck-down stuffed for two, two, seven and fi ve months, respectively. Adult moths emerging during these periods were identifi ed to species and the numbers were counted. The number of cocoons of M. congestella within which a dead or alive mature larva or a pupa existed was also counted. Niditinea tugurialis and N. piercella were distinguished by their genitalia according to .
Niditinea piercella is sometimes treated as a junior synonym of N. striolella (Matsumura) (Karsholt & Razowski, 1996;Gaedike, 2000Gaedike, , 2019. However, the drawings of male and female genitalia identifi ed as those of N. pacifella (which is a junior synonym of N. striolella) in Zagulyaev (1960) differ greatly from those of N. piercella in Pelham-Clinton (1985). According to these authors, N. striolella occurs in the Russian Far East and Japan, while N. piercella is found in Britain and continental Europe. Accordingly, Robinson (2009) recommends keeping the two species separate until material from across the continent can be compared. Recently, Sakai (2013) and , based on the structure of the male and female genitalia of Japanese specimens of possible N. striolella and N. piercella, conclude that these two species are distinctly different. Therefore, this paper treats N. piercella and N. striolella as two species. . Four and three artifi cial nests (mesh bags stuffed with duck down) were located in a deciduous forest in the Toyota City Nature Sanctuary (forest artifi cial nests) and at the Toyota Yahagi River Institute in an urban area (urban artifi cial nests), respectively. Table 1. Details of owl-nest contents and when collected. Contents of nests 1, 2, 4, 6 and 7 (bold-face) were collected twice on different dates. Half the contents of these nests was sampled the fi rst time and the remaining contents the second time. Nest numbers correspond to those in Fig. 1 3 1

Statistical analysis
To determine whether there are differences in the niches of the different species of moths we carried out a non-metric multidimensional scaling (NMDS) and cluster analysis using arcsinetransformed proportions of individual species in owl nest boxes, forest artifi cial nests and urban artifi cial nests. Both statistical procedures were based on a Bray-Curtis dissimilarity matrix. Niche differences were visualized using a NMDS ordination plot in 2D space. Moth species with similar niches were grouped by cluster analysis using Ward's linkage method. These procedures were performed using IBM SPSS ver. 25 (IBM 2017).

RESULTS
A total of 592 individuals of fi ve species of Tineinae, i.e., M. longella, M. congestella, M. fl avidorsalis, N. tugurialis and N. piercella, were obtained from owl nest boxes and artifi cial nests (Fig. 2). The composition of species in the owl nest boxes, forest artifi cial nests and urban artifi cial nests differed. The dominant species in owl nest boxes was M. longella (78.5%), followed by N. piercella (11.0%). The dominant species in forest artifi cial nests was also M. longella (75.4%), followed by M. congestella (24.1%), which was not found in owl nest boxes. Monopis fl avidorsalis was recorded only from owl nest boxes (6.3%). Most moths in urban artifi cial nests were N. piercella (99.5%) with the other species M. longella (0.5%). Niditinea piercella was also common in owl nest boxes (11.0%), but not in forest artifi cial nests. Niditinea tugurialis was found in both owl nest boxes (4.2%) and forest artifi cial nests (0.5%), but not in urban artifi cial nests.
The NMDS ordination plot in 2D space (Fig. 3A) had a stress value of 0.0091. This near-to-zero value indicates that the plot is a reliable representation (Clarke, 1993). The diagram appears to show that the fi ve species of Tineinae are classifi ed into three groups. Monopis longella and M. congestella belong to the same group, while M. fl avidorsalis and N. tugurialis are classed together. Niditinea piercella forms a separate group. This classifi cation was confi rmed by cluster analysis (Fig. 3B). The dendrogram and the NMDS diagram both put the fi ve species into three groups.
No adult moths emerged from the contents collected from owl nest boxes or urban artifi cial nests in December. Monopis longella adults were reared from owl nest contents collected in June, August and October (Fig. 4A), and from forest artifi cial nests in August, October and December (Fig. 4B), which indicates two peaks in occurrence. Monopis congestella and M. fl avidorsalis were recorded in forest artifi cial nests in August, October and December, and in owl nest boxes in June, August and October. Larvae of M. congestella that were collected in December grew into fi nal stage and used silk to construct a cocoon of down, but all of them subsequently died from unknown causes. In comparison to the species of Monopis, N. piercella and N. tugurialis were present for short periods from June to August and June to October, respectively.

DISCUSSION
We found fi ve species of Tineinae (Tineidae) that feed on keratin/chitin in the organic contents of owl nest boxes and artifi cial nests (i.e., duck down stuffed in mesh bags) placed in a forest and urban areas. These moths were assigned to one of three groups, which include: (1) M. longella and M. Fig. 2. Relative abundance of moths of Tineinae reared from the contents of owl nest boxes, forest artifi cial nests and urban artifi cial nests. Numbers in parentheses represent the total number of moths. Nine owl nest boxes were located in a forest covering a geographic region, while forest artifi cial nests and three urban artifi cial nests were located in a small forest and at an institute in an urban area, respectively (see Fig. 1).  (3) N. piercella (Fig. 3). However, unlike M. longella, most M. congestella were found in forest artifi cial nests, which is not refl ected in the above three-group classifi cation.
Robinson (2004) suggests that the type of nest and food resource are important in determining niche separation in keratin/chitin feeding moths and defi nes four types of nests: damp concealed, dry concealed, damp exposed and dry exposed. Based on his defi nitions, owl nest boxes in the present study are classed as damp exposed, because the top is open and drainage is poor, so the organic contents in the box are wet. Hence, these nest boxes are unlike natural owl nests, which are usually dry and concealed in cavities in tree trunks. In contrast, mesh bags fi lled with down to imitate bird nests are classed as dry exposed, because the bags were hung in trees or in the eaves of a building, so the down remained dry although left outdoors for two months. Robinson (2004) describes the food sources of keratin/chitin feeding larvae as either compact or loose. The former includes carnivore droppings, raptor pellets and guano, while the latter mainly comprises hair, fur and feathers. In this study, the contents of owl nest boxes are classed as a compact resource, while those in the artifi cial bird nests are classed as a loose resource.
Based on Robinson's (2004) defi nitions together with habitats (forest and urban) three groups of Tineinae moths may be characterized as follows: (1) M. longella and M. congestella inhabiting forests, (2) M. fl avidorsalis and N. tugurialis exploiting damp exposed nests containing compact sources of food in forest, and (3) N. piercella mainly using dry exposed nests containing loose resources in urban areas. There may be doubts about the differences in niches of these moths, because the forest and urban artificial nests were in one forest and one urban area, respectively, whereas the owl nest boxes were spread across a geographical region (Fig. 1). Nevertheless, the urban site is regarded as typical because it is 400 m from the nearest green park and little affected by semi-natural environments. Furthermore, the forest where the owl nest boxes were deployed is located in a wide stretch of hills covered by secondary deciduous forest dominated by oak, Quercus serrata. Hence, even if artifi cial nests had been set at other sites in the hills as the owl nest boxes were, some species might have been added, but the dominant species might have been little changed. In fact, N. striolella is the only Tineinae species recorded from owl nests in Japan, but was not recorded in the present study; in addition, this species is recorded only infrequently (Nasu et al., 2012b). Consequently, the differences in the niches of the fi ve species of Tineinae seem acceptable, albeit insuffi ciently supported. Therefore, we compare the nature of the niches of these moths reported here with that reported in other studies.
In Japan, Nasu et al. (2007aNasu et al. ( , 2008bNasu et al. ( , 2012a report that many more M. longella than M. congestella emerge from the organic contents of Ural owl nests (and concealed and open nest boxes) and of the nest boxes of the varied tit (Sittiparus varius Temminck & Schlegel). More M. longella than M. congestella are usually recorded from nest boxes of the Japanese tit (Parus minor Temminck & Schlegel, mis-described as P. major L.) and nests of Styan's grasshopper-warbler (Locustella pleskei Taczanowski) (Nasu et al., 2007b(Nasu et al., , 2012c. Sittiparus varius and P. minor nest boxes are dry and concealed, whereas those of L. pleskei are dry and exposed. These passerines are insectivorous and their nests contain faeces, fur and feathers. All these nests were located in secondary forests. In Korea, Lee et al. (2016) reared as many M. congestella as M. longella from the contents of mesh bags containing feathers, pellets and faeces of a raptor in a forest. In addition, Nasu et al. (2007bNasu et al. ( , 2012bY. Nasu & S. Murahama, unpubl. data) reared M. longella, but not M. congestella, from pellets of the common buzzard (Buteo buteo (L.)), sparrowhawk (Accipiter nisus (L.)) and common kestrel (Falco tinnunculus L.) lying on the ground, which indicates that M. longella prefers more compact sources of keratin/chitin than M. congestella. Combining these and our results, we conclude that (1) both M. longella and M. congestella are forest inhabitants, (2) M. longella has a larger niche breadth, exploiting damp and dry concealed nests and dry exposed nests and consuming both compact and loose sources of food and (3) M. congestella has a smaller niche breadth, exploiting dry concealed and exposed nests and loose sources of food. However, this conclusion is partly inconsistent with Robinson's (2004) classifi cation as he considers M. longella to be a damp-exposed nest exploiter. This inconsistency is possibly due to small sample size.
Monopis fl avidorsalis and N. tugurialis [= baryspilas] are recorded from the pellets of common buzzards and sparrow hawk lying on the ground (Nasu et al., 2007b(Nasu et al., , 2012b. Taking our results into consideration, these two species are likely to prefer compact sources of food regardless of nest type. In addition, N. tugurialis is not only recorded from Ural owl nests (Nasu et al., 2007a(Nasu et al., , b, 2012a but also the dry concealed nests of brown hawk-owl (Ninox scutulata (Raffl es)) (Nasu et al., 2012a), Japanese tit nest boxes (Nasu et al., 2008b;Nasu & Saito 2008), varied tit nest boxes (Nasu et al., 2008b), damp exposed nests of the Oriental stork (Ciconia boyciana Swinhoe) (Nasu et al., 2010) and dry concealed nest boxes of the Oriental dollar bird (Eurystomus orientalis L.) (Nasu et al., 2012c). Moreover, adults and larvae of N. tugurialis are recorded from guano of the greater horseshoe bat (Rhinolophus ferrumequinum (Schreber)) . These records might indicate that this species uses a variety of types of nest. However, specimens identifi ed as N. tugurialis in studies before 2012 probably include N. piercella, because prior to that date genitalia were not used to distinguish between these morphologically very similar species . This prevents a more rigorous assessment of the types of nests N. tugurialis tends to exploit. According to Robinson (2004), N. tugurialis exploits both damp-exposed and dry-exposed nests. If so, the niches of N. tugurialis and M. fl avidorsalis might be different.
Our results show that N. piercella has a very different niche from the remaining four species, depending primarily on dry exposed nests containing loose sources of food in urban areas. Nevertheless, it is puzzling that no N. piercella adults emerged from forest artifi cial nests despite the fact that this species inhabits forests and exploits owl nest boxes (Fig. 2). This might be due to the artifi cial nests only sampling a small area of forest. Contrary to our results, Robinson (2004Robinson ( , 2009 considers N. piercella to be an exploiter of concealed-nests based on his and other studies (Jalava, 1980;Pelham-Clinton, 1985;Robinson & Nielsen, 1993;Robinson, 2004). Results of recent studies of tawny owl (Strix aluco L.) nests in Poland (Jaworski et al., 2012) and concealed nests of passerines in mid-Wales, UK (Boyes & Lewis, 2019) support his proposal. However, specimens identifi ed as N. striolella are probably N. piercella as mentioned above. In Japan, N. striolella is neither frequent nor dominant in owl nests (Nasu et al., 2012a), but is recorded occasionally in Oriental dollar bird nests (Nasu et al., 2012c). Another problem is that N. piercella was once confused with N. tugurialis in Japan as mentioned above . To properly defi ne the niche of N. piercella we must collect further data and identify specimens more carefully.
Based on larval occurrence over more than three months, the fi ve species of Tineinae in the present study are likely to be multivoltine. For M. longella, the two peaks in the year in larval occurrence (Fig. 4) probably indicate two generations. Monopis longella and M. congestella are believed to overwinter in the larval stage, because the larvae recorded in owl nest boxes and artifi cial bird nests in October and/or December developed into mature larvae or adults by May of the next year under semi-natural conditions. Overwintering of M. longella, M. congestella and M. fl avidorsalis in the larval stage is reported by Nasu et al. (2007bNasu et al. ( , 2012b. Nevertheless, no adults emerged from organic contents collected from owl nest boxes in December (Fig. 4). One possible reason is that keratin/chitin resources had been mostly consumed or decomposed by that time. If so, the result might indicate that there is intense competition for food among larvae in owl nests. Although we did not record the overwintering stage of N. piercella and N. tugurialis, other studies report overwintering in the larval stage (Nasu et al., 2012c;Sakai, 2013).
Monopis congestella is known to be larviparous (Robinson, 1988(Robinson, , 2004Nasu et al., 2008a). Females of M. congestella lay fully-developed fi rst instar larvae, instead of laying eggs. Although larviparity is rare in Lepidoptera (Hinton, 1981;Heppner, 2009), it often occurs in Monopis; 12 of 90 species of Monopis are larviparous (Xiao & Li, 2006). Robinson (2004) presents two hypotheses for the evolution of larviparity in Monopis: (1) high mortality of eggs and newly-hatched larvae in bird nests due to predation by mites, and (2) females have diffi culty fi nding more than one oviposition site or bird's nest. However, these hypotheses do not explain why the majority of Monopis species are oviparous. In M. congestella, the evolution of larviparity might be related to the main food source being feathers. The quantity of feathers in bird nests is usually small, so larvae that can pre-empt an eventual shortage of food have an advantage. To corroborate this hypothesis, we need to compare the feeding habits of larviparous and oviparous Monopis.
In conclusion, this study indicates that Tineinae moths associated with bird nests differ in their niches in terms of type of nest (concealed or exposed), source of keratin/ chitin (raptor pellets or feathers) and habitat (forest or urban area). However, some problems remain unresolved. First, we cannot explain why N. piercella did not exploit the exposed nests in a forest where it lives even though it exploits such nests. Second, the degree to which Tineinae larvae compete for sources of keratin/chitin in owl nests is unclear. Third, the factors contributing to the evolution of larviparity in Monopis are unclear. We will address these problems in the near future.
ACKNOWLEDGEMENTS. We thank T. Mano for setting up and retrieving artifi cial bird nests at the Toyota Yahagi River Institute and T. Sugiyama for allowing us to collect the organic contents from owl nest boxes. Toyota City Nature Sanctuary permitted us to collect moths in the forest. The Research Institute of the Envi-