Introduction

Bird nests are designed to contain and protect parents, eggs and nestlings and to facilitate optimal safety and climatic conditions from egg-laying to fledging offspring (Collias and Collias 1984). They also constitute a home for a rich and diverse community of invertebrates which use nests as a foraging site, shelter, hiding place and reproduction and overwintering site (Nordberg 1936; Hicks 1959, 1962, 1971; Marshall 1981; Loye and Zuk 1991; Clayton and Moore 1997). In terms of nest fauna studies, most attention has been given to parasites, which are usually invertebrate ectoparasites on nestling or adult birds. These parasites already include a variety of taxa, like ticks, mites, fleas, flies and bugs (Nordberg 1936; Hicks 1959, 1962, 1971). Bird nests not only provide a major food source in terms of the host species itself but detritus provided by the nest material, food remains, feather and skin keratin and offspring faeces also constitute a rich food source. Thus, bird nests can host a huge variety of species that can be assigned to different ecological, particularly foraging, niches. Organisms that use nests as their favourite foraging sites or live in nests include zoophagous, schizophagous (saprophagous, detritophagous, coprophagous) and phytophagous species and parasites (Nordberg 1936). Some species, particularly nidicolous ones, are even specialized to thrive in nest environments. Besides food, nests also provide other beneficial environmental conditions including a safe hiding place as well as a warm and relatively stable climate (Nordberg 1936).

While composition and diversity of nest fauna are well described in many bird species (Nordberg 1936; Hicks 1959, 1962, 1971), almost nothing is known about the potential impact of their coexisting nest fauna, other than parasites. Depending on their abundance, the effect of nest inhabitants could range from beneficial to detrimental (Marshall 1981; Clayton and Tompkins 1995; Hebda et al. 2013).

One important aspect influencing nest quality and hence offspring fitness has been shown to be nest sanitation behaviour. Nest sanitation behaviour varies significantly between different species (Guigueno and Sealy 2012), and in particular those which are able to adjust their investment according to the needs (Gow and Wiebe 2015). In this context, European Bee-eaters (Merops apiaster) are extreme since they completely lack any nest sanitation behaviour. In particular, they do not remove faeces, dropped food items, regurgitated food pellets or even dead nestlings from the nest cavity (Glutz von Blotzheim and Bauer 1980; Brust et al. 2015). Furthermore, nestlings do not eat prey items dropped on the cavity floor (Hoi et al. 2015), which is probably due to the poor light conditions in the nest cavity at the end of up to 200-cm-long tunnels (Glutz von Blotzheim and Bauer 1980). The sand in the cavity intermingles with nestling faeces, keratin from growing feathers, and food remains and accumulates from up to seven nestlings. This may stimulate the production of gases via putrefactive fermentation, which together with reduced ventilation (Ar and Piontkewitz 1992) may lower nest oxygen concentration (Lill and Fell 2007). Hence, the fact that members of their nest fauna forage on waste that accumulates in their nests, like skin and feather particles, uneaten food, faeces and dead nestlings, might be an ideal alternative to compensate for this lack in nest sanitation behaviour. Therefore, one possible advantage could be that the decomposition of nest material provides a nest sanitation effect (Nasu et al. 2012), a potential scenario that involves a mutualistic relationship or cooperation between nest owners and members of the nest fauna. Regarding negative consequences for bird species like European Bee-eaters, which breed in deep underground burrows, a higher number of non-conspecific nest inhabitants could theoretically also create detrimental cavity conditions. This may include increased defaecation, risk of disease transmission or oxygen consumption (Eichler 1963; Marshall 1981). However, no study has yet tried to experimentally investigate the importance of non-parasitic nest fauna on the host species.

Therefore, in this study, we intended to experimentally determine the impact of one very common and abundant nest inhabitant on nestling European Bee-eaters. Their nest cavities provide specially favourable climatic conditions for a rich invertebrate fauna (Krištofík et al. 1996), which can vary, e.g., with geographic conditions (Casas-Crivillé and Valera 2005). Krištofík et al. (1996) found that the saprophagous larvae of the fly Fannia spp. were the most dominant insect species in terms of biomass (their biomass can equal one or even more nestlings). During the nestling feeding period, Fannia spp. larvae can be very abundant, with on average 200 larvae per nest. The larvae, about 1 cm long, move on the soil surface of the nest cavity, where nestling bee-eaters sit on them, as shown in Fig. 1.

Fig. 1
figure 1

Nestling European Bee-eater (Merops apiaster) sitting on its natural substrate with many Fannia spp. larvae

Full size image

Thus, to investigate the impact of Fannia spp. larvae on nestling Bee-eaters, we experimentally manipulated their abundance during the early nestling period and determined parameters indicative for nestling body condition and health.

We hypothesised that if Fannia spp. larvae contribute to improve nest sanitation, due to their detritivorous habits, an increased abundance of Fannia spp. larvae would result in improved chick growth (body condition) and health (reflected by, e.g. haematocrit and sedimentation rate), compared to experimental nests from which Fannia spp. larvae had been removed. The genus Fannia is known to be a vector for some pathogenic bacteria, protozoa, tapeworms and roundworms (Greenberg 1971). Thus, if an increased abundance of Fannia spp. larvae in nests resulted in a higher risk of disease transmission (see also Gregor and Rozkošný 1995) and poorer nest conditions due to lowered oxygen availability, then Bee-eater nestlings in nests faced with experimentally increased abundances of Fannia spp. larvae should have a reduced growth and suffer negative effects on blood parameters, compared to experimental nests in which Fannia spp. abundance was reduced.

Methods

The study was performed between 6 and 18 July 2013 in western Slovakia (47°48′–47°56′N, 18°20′–18°44′E). Nests for the experimental manipulation were selected from 12 localities (colonies). The number of breeding pairs ranged from 3 to 30 per colony and two nests from each locality were randomly selected for the experimental treatment. For the two types of controls (see below), we further randomly selected 18 nests (9 nests for each control group) from 10 colonies.

Study design

To examine the effect of Fannia spp. larvae abundance on nestling Bee-eaters, we produced two treatment groups in which the number of fly larvae was manipulated. Fannia spp. larvae were either added (added group, n = 13 nests) or removed (removed group n = 11 nests). In the added group, we increased the larvae number on average by 269 ± 3.3 (±SE) individuals above the original number, which is about 200 % above the average number of Fannia spp. larvae found in the population (see “Results”). To have enough Fannia spp. larvae to add, we additionally collected them from other nests not included in the experiment. In the removed group, all Fannia spp. individuals were removed, resulting in zero Fannia spp. larvae per nest. For the removed group, to facilitate a meaningful modification, we used only nests with Fannia spp. numbers higher than the population average (see below).

For the experimental groups, in the first round we extracted all nestlings from the nest cavity and removed about 2 kg of sand material with a spoon attached to the end of a 150-cm-long stick (Hoi et al. 2002, 2010). We selected nests with nestlings of similar age (approximately 8–12 days old). We measured the wing length (mm) of each nestling (n = 109) using a digital calliper to the nearest 0.1 mm and body mass (g) by means of an electronic balance to the nearest 0.1 g. We then counted the number of Carnus hemapterus on the body surface of each nestling, as these are known to be the most important blood sucking ectoparasite of nestling Bee-eaters (Hoi et al. 2010). To eliminate an effect of this parasite, nestlings were then returned into the nest chamber without their parasites.

After 7 days (13–18 July 2013), nests were re-inspected. In this second round, we again extracted the sand material and counted the number of Fannia spp. larvae and C. hemapterus on the nestlings and in the nest material, recorded the number of nestlings, and measured their wing length and body mass. Furthermore, we took a blood sample (about 50 µl) from each nestling to determine their erythrocyte sedimentation rate and haematocrit.

We also established two control groups with nests containing nestlings of similar age as in the experimental group (see above) with the aim of detecting various degrees of perturbation on the Fannia spp. larvae as well as on the nestling Bee-eaters, e.g. on nestlings extracted once or twice. In the first control group (control I), nestlings were taken twice from each nest, matching the timing of the experimental groups (for the first time: 6–11 July 2013; for the second time: 13–18 July 2013), but without touching or manipulating the sand material and the nest fauna. In the second control group (control II), we removed nestlings only once (13–18 July 2013), matching the second inspection in the experimental as well as the control I group.

Fannia spp. larvae and C. hemapterus collection

We removed sand material from each nest during the first and second round. The sand material was immediately spread in a thin layer onto a white board from which we collected all Fannia spp. larvae and C. hemapterus individuals using a tea spoon. To get a complete sample of C. hemapterus per nest, we additionally removed them from all nestlings. Depending on the experimental treatment, the sand material was returned either with or without Fannia spp. larvae (see “Study design”), but always without C. hemapterus individuals.

Nestling measurements

To examine the effect of Fannia spp. larvae on offspring development, we used offspring condition by determining the residual body mass not explained by size [from the regression between body mass (dependent) and wing length (independent variable): r = 0.81, p < 0.0001, df = 109] for all nestlings of all nests (Hoi et al. 2002), together with the two blood parameters indicative of health and condition. The erythrocyte sedimentation rate was measured in capillary tubes placed upright for 4 h at 4 °C, calculated as the distance occupied by the plasma divided by the total length of the blood column (in %). Then, the capillary tubes were centrifuged for 5 min at 12,000g and the haematocrit was measured as the relative amount of red blood cells divided by the total blood volume using a digital caliper to the nearest 0.01 mm. Erythrocyte sedimentation is a non-specific diagnostic method regularly used in medicine which increases with inflammatory and infectious diseases (Saadeh 1998; Heylen and Matthysen 2008). In contrast, haematocrit decreases with acute or chronic diseases, nutritional deficiencies or parasite and bacterial infections (Fair et al. 2007; Johnstone et al. 2015). Thus, the two blood measurements are independent indicators of health and condition (Johnstone et al. 2015). A positive effect on offspring condition with increasing larvae number would point towards a nest-cleaning effect. A negative effect, due to increased risk of diseases transmission and, e.g., hypoxia, would be more pronounced for haematocrit and blood sedimentation.

Determining nestling age

Nestling age is difficult to determine in European Bee-eaters, but wing length seems to be a reliable proxy (Lessells et al. 1994; Hoi et al. 2002). Due to the high hatching asynchrony in this species (Glutz von Blotzheim and Bauer 1980), nestling age also varies within a nest. Given that brood size did not differ between experimental groups, we used average wing length of all nestlings per nest to account for within-brood variation. Finally, we presumed that nestling age was also similar, since nestling wing length did not differ between the four groups (F = 1.43, p > 0.21, df = 3,51).

Statistical analyses

The correlation between body mass as a dependent variable and wing length as independent size variable was used to determine residual body mass (r = 0.81, p < 0.0001, df = 1,109). In European Bee-eaters, similar to other mainly areal birds like swifts (Apus apus), legs are less important and therefore their short and fleshy tarsi are not appropriate size determinants (Lessells et al. 1994).

To analyse the effect of Fannia spp. larvae on bee-eater offspring, we used two different approaches. In the first approach, we used generalised linear models (GLM) including the two experimental and the two control groups as independent factors in the analyses. Average nestling body condition expressed as residual body mass not explained by size (Hoi et al. 2002; Ardia 2005), residual sedimentation rate (not explained by total blood amount) and haematocrit per nest were entered as dependent variables, respectively, and the number of nestlings was introduced as covariate.

There are, however, several other factors that could also influence the outcome of our results, like nestling body condition, number of nestlings and Fannia spp. larvae at the beginning of the experiment, and the number of C. hemapterus present prior to the experiment. Due to the nature of the experiment, these variables could not be determined in the control groups, and this information was only available for the experimental groups. Therefore, in a second approach, we included these variables and used a generalised linear mixed models (GLMM) but restricted to the two experimental groups. To examine the importance of Fannia spp. treatment, we again used nestling condition, residual sedimentation rate and haematocrit, respectively, as dependent variables. The initial number of nestlings was included as fixed factors. This variable was included because we found a positive relationship between the number of Fannia spp. larvae and the number of nestlings (r = 0.44, p = 0.034, df = 23,109, F = 5.17) prior to manipulation. The initial number of Fannia spp. and C. hemapterus and the initial nestling condition (residual body weight not explained by size – wing length at the start of the experiment) were included as covariate and nest identity as a random effect.

Models were checked for normality and heteroscedasticity, and a non-gaussian link function was used for the GLMM.

To examine the effect of fly larvae on nestling mortality we compared nestling mortality between the two treatment groups using a t test because overall nestling mortality was very low during the experimental period.

Results

Prior to manipulation, we found an average of 99.7 ± 27.3 (±SE, n = 24) Fannia spp. larvae per nest, ranging from 0 to 486 individuals per nest. The larvae number prior to the experiment did not differ between the added (92.8 ± 15.8 individuals/nest) and removed groups (116.5 ± 21.7 individuals/nest; t test: t = −1.2, p > 0.2). Also, average nestling number at the beginning of the experiment did not differ between the experimental groups (added group: 5.08 ± 0.21; removed group: 5.0 ± 0.31; t = 0.2, p > 0.8).

At the second nest inspection (moment of termination of the experiment), significantly more Fannia spp. larvae were found in nests where the larvae number was originally increased (added group, mean: 132.8 ± 27.1, n = 13; removed group: 23.2 ± 9.8, n = 11; F = 3.2, p < 0.003), which suggests that our manipulation worked well.

An influence of C. hemapterus on the experiment can be neglected (1) because their numbers did not differ between experimental groups, neither prior to the experiment (mean number of C. hemapterus in the group in which Fannia spp. was added: 74.08 ± 20.5; or removed: 62.3 ± 18.1; t = −0.5, p > 0.6), nor at the end of experiment (mean number of C. hemapterus in the group in which Fannia spp. was added: 44.0 ± 13.6; or removed: 22.6 ± 8.7, t = −1.3, p > 0.2), (2) because these parasites were removed at the beginning of the experiment, they were significantly lower at the end of the experiment than prior to manipulation (t = −2.8, p = 0.016), and (3) Fannia spp. and C. hemapterus abundances were not interrelated (prior to experiment: r = 0.14, p > 0.2, n = 24; end of experiment: r = 0.09, p > 0.8, n = 24).

As shown in Fig. 2, a comparison of the different groups revealed a significant difference in average nestling body condition (F = 3.7, p = 0.041, df = 3,49). This difference seems to be mainly due to the two experimental groups, whereas the two control groups are inbetween and do not seem to differ from each other. In contrast, no difference was found in the average residual nestling sedimentation rate per nest (GLM, F = 1.67, p > 0.186, df = 3,49) and the average nestling haematocrit per nest (F = 1.24, p > 0.3, df = 3,49).

Fig. 2
figure 2

Nestling body condition in relation to the experimental manipulation of Fannia spp. larvae. Nestling condition is expressed as residual body mass (g), sedimentation rate (%), haematocrit (%) for the different experimental and control groups (for details, see “Methods”) as means ± SE

Full size image

Concentrating on the two experimental groups only and including additional influential factors (see “Methods”), again offspring condition turned out to be important and was significantly better in nests where Fannia spp. larvae had been added (mean residual body mass: 1.78 g ± 0.9), than in nests where they have been removed (mean residual body mass: −2.5 g ± 1.07; Table 1). Our results further suggest that nestling body condition depends on nestling number and decreases with increasing number of nestlings (Fig. 3; Table 1). Likewise sedimentation rate and haematocrit did not turn into being significant when considering these other variables (see Table 1).

Table 1 The relationship between experimental manipulation of Fannia spp. (fly larvae) on health and condition of nestling European Bee-eaters (Merops apiaster)
Full size table
Fig. 3
figure 3

Nestling body condition in relation to brood size. Residual body mass (g) as mean ± SE

Full size image

Finally, the number of nestlings that died, in total 9 nestlings, during the experiment did not differ between the experimental groups (t = 0.1, p > 0.75).

Discussion

Our results suggest that the nest fauna can influence nestling development. The existence of a mutualistic relationship between nest owners and nest fauna was suggested by Nasu et al. (2012). However, our results are the first to experimentally demonstrate such a positive effect on nestling development. Faced with an increased number of Fannia spp. larvae, nestling Bee-eaters exhibited a significantly better condition in comparison to the controls (Fig. 2), and nestling condition was reduced in nests from which larvae had been removed. All this suggests that the positive effect on nestling body condition is very likely directly due to Fannia spp. larvae eating the detritus and in this way cleaning the nest chamber and improving cavity conditions. Such a sanitation effect is very obvious when, for instance, keeping the nest contents except the nestlings together with Fannia spp. larvae in a plastic bag. In this case, our observations revealed that fly larvae collected from a nest can consume large insects like a dragonfly, butterfly, or bumble bee in a few hours, to such an extent that there are no more visible remains (Hoi et al. 2015). However, further analyses would be necessary to understand causal relationships, in particular which factors really determine the change in nestling condition.

Our results did not reveal a difference in residual sedimentation rate and haematocrit between experimental and control groups (see Fig. 2). Therefore, a changing risk of infectious diseases is unlikely to be a consequence of fly larvae abundance but cannot be ruled out completely. Depending on the incubation time of an infection, disease symptoms may have become visible only after our experiment was finished or specific infections may be of concern only at later stages of juvenile development.

Chronic hypoxic conditions are common for subterranean mammals and birds (Widmer et al. 1997), but we could not find an experimental effect of fly larvae manipulation on haematocrit levels, assuming that our experimental manipulation would have changed nest cavity oxygen concentration (Table 1). There are several potential options to explain the lack of a detectable effect on haematocrit. Either the changes in cavity oxygen concentrations due to the nest sanitation effect are minimal or nestlings adapt to hypoxic conditions, e.g. increase their breathing rate (Widmer et al. 1997). For many subterranean species, physiological and structural adaptations to chronic hypoxic conditions have been demonstrated (Widmer et al. 1997). Thus, to better understand the functional relationships between nest sanitation and oxygen concentration, it would be necessary to measure it directly in the nest cavity in relation to Fannia spp. larvae activity.

Besides Fannia larvae, a variety of other species including parasites were identified in the Bee-eater cavities in our study area (Krištofík et al. 1996), and it is likely that interactions among species occur. The question arises whether species interactions could be an alternative pathway to explaining the effect induced by our treatment. Adding Fannia spp. larvae, for example, might reduce the growth of C. hemapterus, the most important ectoparasite of nestling European Bee-eaters (Krištofík et al. 1996; Hoi et al. 2010) and, as a result, indirectly favour nestlings. However, a general and strong role of C. hemapterus can be neglected for our study, since they were removed at the beginning of the experiment. On the other hand, they resettled, but only to some extent, resulting in a still significantly reduced abundance at the end of the experiment (see “Results”). It has been shown that C. hemapterus is able to spread to neighbouring nest cavities (Valera et al. 2003). For several reasons, an interaction is still unlikely. Firstly, according to our results, the abundance of C. hemapterus and Fannia spp. larvae seem to be independent. Secondly, the number of C. hemapterus resettling in the nests after removal (counted at the end of the experiment) did not differ depending on whether the number of Fannia spp. larvae had been experimentally removed or increased. Finally, most work cariied out on this parasite suggests that is not very harmful with only a weak damaging potential (Kirkpatrick and Colvin 1989; Dawson and Bortolotti 1997; Liker et al. 2001).

Thus, we propose that nest cleaning by a rich nest fauna constitutes an alternative factor, compensating for the lack of nest sanitation behaviour observed in some bird species, including the European Bee-eater (Guigeno and Sealy 2012). In this context, their contribution might be particularly important in species for which nest sanitation by the birds themselves might be complicated or impossible (e.g. dark cavities). A rich nest fauna might, for instance, be important in other birds nesting in cavities or in deep burrows like Sand Martin (Riparia riparia) or Kingfisher (Alcedo atthis). However, the nest fauna could also be expected to play an important role in nests in which organic matter accumulates, e.g. in nests which are repeatedly used, or in nests wihch are built on traditional sites, e.g. by many raptor species or storks.

The importance of the nest fauna may also depend on the necessity of nest sanitation. Nest sanitation behaviour has been found in many bird species but also shows intra- as well as interspecific variation (Giugueno and Sealy 2012). The allocation of investment into nest sanitation has been shown to depend on its importance for offspring fitness (Gow et al. 2015). The diversity of nests constructions, e.g. simple nests without real nest construction, to complicated nest structures, such as cup, domed, or compound nests, could also contribute to explain variation in nest sanitation behaviour.

Another potential indirect beneficial effect related to the nest sanitation by the fly larvae could be that the removal of detritus reduces the detection of nests by olfactory predators. However, this is less likely in European Bee-eaters because their nests are in cavities at the end of about 1.5- to 2-m-long tunnels.

In our study, we investigated the effect of one specific species. However, 108 species, including parasites, have in fact been identified in Bee-eater cavities in our study area (Krištofík et al. 1996). Thus, it is not unlikely that many more interactions may exist both between lodgers and between nest owners and lodgers. Our results offer only a glimpse of the complex relationships that exist between nest owners and nest fauna. Future studies using bird nest fauna as a model system for ecosystem analyses will almost certainly reveal new insights into ecosystem functioning.