Site-Specic Variation in Bacterial Loads of Wooden Nest Boxes in the Pre-Breeding Period

Among environmental factors affecting life - history traits of birds breeding in nest boxes, an inuence of microbial communities is relatively poorly understood. In this study, nest boxes used for breeding by great tit (Parus major) and blue tit (Cyanistes caeruleus) were sampled before the start of the breeding season to assess the bacterial loads of the nest box. Samples from the entrance hole and the interior of the nest box were taken at two different study sites: an urban parkland and a natural forest. Nest boxes were sampled to check if their bacterial loads differed between habitats. The second objective of this study was to check whether the occupancy of the nest boxes during the previous season would inuence the bacterial load of the nest box. To verify this prediction, two categories of nest boxes were sampled at both study sites: nest boxes occupied by any of the two tit species in the previous season for breeding and nest boxes that had remained empty that year. The bacterial load of the nest box was signicantly higher in the forest study area in both the occupied and unoccupied nest boxes. The nest boxes used for breeding in the previous season had signicantly higher bacterial loads, but only in the forest area. Our results suggest that the bacterial load of the nest box can vary between habitats and may be positively related to the presence of the nests in the previous breeding season.


Introduction
Microorganisms are ubiquitous in a variety of habitats, and they inhabit all available spaces there. One of these habitats, avian nests, constitutes a unique environment for a variety of them 1,2,3,4 . Bacteria colonizing avian nests and spaces, which birds can potentially choose to breed, affect their avian hosts in both positive and negative ways 3,5,6 . Some of the bacteria that colonize avian nests are commensals that feed on nest components consisting mainly of organic matter, but many bacterial species, including pathogenic strains, can exert a negative impact on birds (see 6 for the review). Recent research has shown that the nest microbiota plays an important role in mediating the life histories of birds. In particular, microorganisms shape the microbiome of their hosts and play a key role in the digestion of food (see 7 for the review), pathogen defence 8, 9 , as well as in uence plumage quality 10,11 . There is also a growing body of evidence showing the importance of microbial species for traits of avian condition and, consequently, for their reproductive success 2,3,4 . However, relationships between nest bacterial load and life -history traits of birds are relatively poorly understood, mainly because the majority of studies on animalassociated microbiota are conducted on captive animals 12 . Little is still known about the complex relationships between different species of birds and their microbiota under particular environmental conditions in the wild. Given the ubiquity of bacterial communities and the complexity of host-microbiota interactions, it is important to document both the differences in abundance and diversity of microorganisms in different habitats 13,14 .
Among a wide variety of locations where different species of birds breed, nests located in cavities are unique, as cavities constitute a particularly suitable environment for different types of microorganism 1,3,15,16 . The special characteristics of nest boxes (and other cavities, including natural ones) result from maintaining relatively stable physical conditions, such as humidity and temperature inside them, which make them suitable for microbial colonization and growth 2,17 . Additionally, nest boxes are frequently used for breeding in several consecutive seasons, ensuring steady delivery of organic matter (feathers, peeled fragments of the epidermis, or excreted feces) essential for microbial growth. It should also be mentioned that outside the breeding season, several species of birds use nest boxes as roosting places during an autumn-winter period 18 . Furthermore, available data suggest that the prevalent practice of cleaning nest boxes after the breeding season does not remove all litter, leaving enough residues for bacterial growth 4 . Furthermore, it is highly probable that the bacterial load of the nest box itself depends on a variety of environmental characteristics of a particular habitat. The physical properties of habitats include several traits such as humidity, thermal conditions, chemical properties of soil, and probably many other characteristics that in uence the assemblages of bacteria within different spatial scales (see 13 for a review). Furthermore, the richness of plant species in a particular habitat can also play an important role 19 . Since the available studies on natural microbial diversity in nest boxes placed in different habitats are rather scant 20 , there is a need to focus on microorganisms colonizing nest boxes which are used not only for breeding but also as roosting sites of many avian species.
For the above reasons and given that in our previous study we had demonstrated a negative in uence of bacteria on the physiological condition in wild birds 4 , we conducted the study in two sets of wooden nest boxes that are used by the two secondary cavity nesting bird species, the great tit Parus major and the blue tit Cyanistes caeruleus in two different habitats: urban parkland and a deciduous forest. Both these tit species interchangeably use nest boxes for breeding in consecutive breeding seasons, however, in a particular year some of the nest boxes may remain unoccupied throughout the breeding season. This may potentially lead to variation in the abundance of bacteria in nest boxes. The second potential source of variation in bacterial loads in nest boxes may result from habitat differences between the two study sites. Probably, the different physical conditions prevailing at either site (temperature, humidity, soil characteristics, and similar characteristics) can shape the microbial communities in nest boxes. Therefore, we estimated the interior bacterial loads of the nest box and the entrance hole of the nest box to (i) check if there was a signi cant difference between the two habitats and (ii) test the following prediction: the nest boxes used for breeding during the preceding year had higher bacterial loads than the nest boxes that had remained empty that year.

Results
The mean, minimum and maximum ln transformed values of the bacterial loads in the entrance hole and the interior bacterial loads of the nest box in the parkland and forest, occupied for breeding and empty in the previous year, are given in Tables 1 and 2.  Table 1 The mean, minimum, and maximum values of the entrance hole (CFU/ml) in the parkland and forest study areas in both nest categories (occupied and unoccupied in the previous year The mean entrance hole bacterial load differed signi cantly between the study areas and was higher in the forest study area (Table 1, Table 3, Fig. 1). Similarly, the interior bacterial load of the nest box differed signi cantly between the nest boxes occupied and unoccupied in the previous year (which was higher in the nest boxes occupied in the previous season), but the difference was signi cant only in the forest study area (Table 2, Table 3, Fig. 2). The correlation between the entrance bacterial load and the interior bacterial load was non-signi cant in the parkland study area and in the forest study area (r = 0.41, p = 0.075 and r = -0.43, p = 0.129, respectively).

Discussion
We found that both the bacterial load of the entrance hole and the bacterial load of the nest box were signi cantly higher in the forest than the bacterial loads in the parkland, both in the nest boxes occupied and unoccupied in the previous season. We also found that the interior bacterial load of the nest box was higher in the nest boxes used for breeding in the previous season, but the difference was signi cant only at the forest site.
We revealed a clear difference in the interior bacterial load of the nest box between deciduous forest and urban parkland. This result suggests that the bacterial loads in nest boxes are strongly habitat dependent. The interior of the nest box is probably readily colonized by a variety of microorganisms shortly after placing the nest box in a particular environment. Birds play the main role of transferring microorganisms to nest boxes when visiting them, as well as to other various cavities as places potentially suitable for breeding or roosting 19,20 . Bacteria are found alike on avian beaks and toes, however, these communities in the plumage are most important in this context, since bacteria are both most numerous and most diverse in feathers 20,21 . In our study system, nest boxes were used mainly by the Blue Tit and the Great Tit. Both species forage mainly on trees leaves and twigs. Plants, as a source of bacteria, are suggested to play an important role in the transmission of these organisms to plumage 21 .
This impact of the plant host on the bacterial microbiota was shown between predators and prey in the trophic networks of blue tit 22 . Given that our study sites differ markedly oristically with deciduous forest being richer in plant species than urban parkland, it seems very likely that the forest site may maintain richer bacterial communities. Saag et al. 19 showed such a pattern of habitat-related differences in bacterial density and species richness. They revealed that while the number of phylotypes per bird was higher in coniferous habitat, the bacterial densities were higher in deciduous habitat. It is also supported by the results presented by Broughton and Gross 23 who suggested that microbial activity was positively correlated with plant productivity. Similarly, Bisson et al. 21 found that microorganisms from particular groups were more abundant in plumage sampled from American redstart (Setophaga ruticilla) in wet versus dry habitats. It is in line with our results, since rich, mature forests retain more moisture than urban parkland with its many open areas with no or little tree cover and, therefore, open to direct sunlight.
However, the richness of plant species is not the only distinction between the two sites. The Botanical Garden is a place where various agrotechnological works are conducted throughout the year (own observations). These procedures include the application of different chemicals, such as insecticides, which probably affect different organisms, including bacteria that are present not only in plants but also in soil. For example, in 2013 and 2014, a large-scale molluscicide treatment was applied in the Garden to eradicate the invasive Spanish slug (Arion vulgaris) (see 24 for details). Presumably, regular use of chemicals in parkland, contrary to the forest area, can disturb microbial communities, including the sheer number of bacteria in different ways. For the above reasons, the microbial pro le of the soil itself may play an important role in plumage bacterial acquisition 21,25,26 . It may be important in the context of this study, since the great tit relatively frequently forages on the ground 27, own obs ., which means that the birds acquire bacteria directly from the soil and then carry them into nest boxes.
In addition to basic habitat characteristics including plant species composition and soil characteristics, there are probably other factors that can contribute to bacterial loads in nest boxes. It is known that different taxa of animals visit nest boxes for a variety of purposes. Many invertebrates (snails, spiders, or insects) are opportunistic species that use nest boxes year-round as shelters, roosting places, or to build the nests and raise their offspring (i.e. wasps) 28,29, own.obs .. Invertebrate species colonizing nest boxes carry their bacteria acquired from the environment. Since the forest area is a more diverse habitat than the parkland area in terms of plant species composition, it is likely that the invertebrate assemblies are also richer in the forest. The next potential factor is the presence of particular mammal species, which visit nest boxes for two main reasons. In both our study areas, brown long-eared bats (Plecotus auritus) and noctules (Nyctalus noctula) occasionally use empty nest boxes for breeding and roosting during the summer / autumn period, and they are more frequent in the forest (own observations). The pine marten (Martes martes), in turn, is a predatory species that tries to reach the nest in the nest box and grab an adult bird or nestlings with its paws as a prey. In some breeding seasons, the predation rate was very high in the forest 30 , but not in the parkland where marten predation occurred only exceptionally (own observations). Both bats and martens carry their bacterial ora which is transferred to nest boxes; however, in the case of martens, this transfer is restricted mainly to the entrance hole size.
We also tested whether the occupancy of the nest box in the previous year had affected the bacterial load of the nest box. Our results are ambiguous on that point since this effect was signi cant in the case of the nest box interior but not in the case of the nest box entrance and only in the forest study site. It is not clear why this effect was found only at one study site. A possible explanation may be related to the intensity of using nest boxes as roosting sites during an autumn-winter period. It is well known that wintering tits spend the nights in nest boxes 18, own obs . and thus probably transfer bacteria there. In the parkland, contrary to the forest area, both tit species are more abundant during winter (own observations) and probably use nest boxes as roosting sites more intensely. If this assumption is true, it should at least partially eliminate the expected difference in bacterial load between nest boxes occupied and unoccupied for breeding. However, it seems that this effect is not strong enough to nullify the striking difference between the study sites. Yet, we do not have any quantitative data on the intensity of using nest boxes as roosting sites out of the breeding season and, therefore, we cannot make any plausible conclusion on that issue.
In conclusion, our study demonstrates that the observed difference in the bacterial load of the nest box is probably related to habitat-related differences between the two study sites. The most important factors are probably associated with the richness of plant species and the physical properties of the soil, with additional factors resulting from the speci city of the two study sites. However, taking into account the complex and yet relatively weakly understood relations between microorganisms, their hosts, and the environment, probably other factors play their role and we are still far from understanding this issue.
Furthermore, we should treat our results with care since we do not have data on the richness of bacterial species at both study sites. However, our study suggests that the bacterial load may be a potentially important factor for the life-history traits of secondary cavity nesting birds, since it may vary considerably between habitats. Therefore, since studies on cavity nesting birds with the use of arti cial wooden nest boxes are widely conducted across a geographical range and a variety of habitats, and given that bacterial load may affect host tness components, we consider our study perspective as a promising eld for future research.

Study sites
This study was carried out in 2020 in two different habitats: an urban parkland and a deciduous forest.
The study sites are located ~ 10 km apart and are separated by the city of Łódź. The urban parkland study site (51º45'N; 19º24'E) consists of Łódź Botanical Garden, which covers a total area of ca. 67 ha. It is mainly of anthropogenic origin and has fragmented tree and bush cover with very few remnants of natural stands 31 . The forest study site (51°50' N;19°29' E) is an area of ca. 145 ha located in the interior of a rich, mature mixed deciduous forest called the Łagiewniki Forest (1250 ha in total). Oaks (Quercus robur and Q. petrea) are the dominant tree species in the forest. Both study sites were supplied with standard wooden nest boxes, each made of pinewood (200 in the parkland area and 300 in the forest area). After each breeding season, each nest box was cleaned with a wire brush, so there were no visible nest remnants.

Bacterial sampling
In the rst half of March 2020, 40 nest boxes (20 at each study site) were randomly chosen for bacterial sampling. About half of the nest boxes at each study site had been occupied by one of the two tit species (great tit or blue tit) in the previous year for breeding, while the other half had remained empty that year (5 occupied by Blue Tits, 6 by great tits and 4 by blue tits, 6 by great tits in the parkland and in the forest, respectively). Using disinfected latex gloves, an entrance hole of each nest box was swabbed with circular movement for 30 s with a sterile cotton swab previously moistened with sterile phosphatebuffered saline (PBS, pH 7.2; Adlab, Poland). Subsequently, a restricted area of ca. 2.5 cm 2 of an internal rear wall of every nest box in the middle of its height was swabbed in the same manner. A sample per entrance hole and another sample per inner wall of each nest box were taken (in a standardized manner with respect to time and area).

Lab procedures
Once in the laboratory, bacteria from the swabs were transferred to the solution in the manner used in our previous study on the bacterial load (see 4 2020 for technical details). The bacteria were cultivated and incubated on Tryptic Soy Agar medium (TSA) for 48 h at 37±1°C, then for an additional 48 h at 25±1°C and the colony forming units (CFU) were counted. The results were expressed as CFU/ml.

Statistical analyzes
The bacterial load of the nest box entrance hole and the nest box interior variables were normalized by ln transformation prior to analyses. A two-way ANOVA was used to test whether the study area and the occupancy of the nest box affected the nest box entrance hole bacterial load in the study year; the interaction term was also calculated. Interactions, both signi cant and non-signi cant were included in the models. This approach was repeated for nest box interior bacterial load. Analyses were performed using Statistica ver. 12 software 32 .

Declarations
Data availability statement The datasets generated and analysed during the current study are available in the gshare repository, https:// gshare.com/s/8cf266bd98ae89b5f394.

Figure 1
Mean (± standard errors) entrance hole bacterial load in the nest boxes occupied and unoccupied in the previous season in two study areas